Testing of Concrete in Structures: Fourth Edition

Also available from Taylor & Francis∗∗ Handbook on Nondestructive Testing ofConcrete, 2nd edition ∗∗N.J. Carino, V.M. MalhotraHb: 0-849-31485-2Spon Press∗∗ Concrete Structures ∗∗A. Ghali et al.Spon PressHb: 0-415-24721-7∗∗ Corrosion in Concrete Structures ∗∗Edited by: H. BohniSpon PressHb: 0-84932583-8∗∗ Durability of Concrete Structures andConstructions ∗∗L.M. PoukhontoSpon PressHb: 9-058-09229-1Information and ordering detailsFor price availability and ordering visit our website www.tandf.co.uk/builtenvironmentAlternatively our books are available from all good bookshops.

Testing of Concrete inStructuresFourth editionJ.H. BungeyEmeritus Professor of Civil EngineeringUniversity of LiverpoolS.G. MillardReader in Civil EngineeringUniversity of LiverpoolM.G. GranthamConsultant – M.G. Associates Construction Consultancy LtdDirector – G.R. Technologie Ltd

First published 1982by Surrey University PressSecond edition published 1989by Spon PressThird edition published 1995by Spon PressFourth edition published 2006by Taylor & Francis2 Park Square, Milton Park, Abingdon, Oxon OX14 4RNSimultaneously published in the USA and Canadaby Taylor & Francis270 Madison Ave, New York, NY 10016, USATaylor & Francis is an imprint of the Taylor & Francis Group© 1982, 1989 and 1995 John H. Bungey© 2006 John H. Bungey, Steve G. Millard, Michael G. GranthamThis edition published in the Taylor & Francis e-Library, 2006.“To purchase your own copy of this or any of Taylor & Francis or Routledge’scollection of thousands of eBooks please go to www.eBookstore.tandf.co.uk.”All rights reserved. No part of this book may be reprinted orreproduced or utilised in any form or by any electronic, mechanical, orother means, now known or hereafter invented, including photocopyingand recording, or in any information storage or retrieval system, withoutpermission in writing from the publishers.The publisher makes no representation, express or implied, with regardto the accuracy of the information contained in this book and cannotaccept any legal responsibility or liability for any errors oromissions that may be made.British Library Cataloguing in Publication DataA catalogue record for this book is available from the British LibraryLibrary of Congress Cataloging in Publication DataBungey, J.H.Testing of concrete in structures / John H. Bungey,Steve G. Millard, Michael G. Grantham. — 4th ed.p. cm.ISBN 0–415–26301–8 (hardback : alk. paper)1. Concrete—Testing. 2. Concrete construction—Testing.I. Millard, S.G. II. Grantham, Mike. III. Title.TA440.B79 20066241 ′ 834 ′ 0287—dc222005022828ISBN10: 0–415–26301–8ISBN13: 978–0–415–26301–6

ContentsPrefacex1 Planning and interpretation of in-situ testing 11.1 Aims of in-situ testing 11.1.1 Compliance with specification 21.1.2 Assessment of in-situ quality and integrity 31.2 Guidance available from ‘standards’ and otherdocuments 51.3 Test methods available 61.4 Test programme planning 81.4.1 General sequential approach 81.4.2 Visual inspection 81.4.3 Test selection 121.4.4 Number and location of tests 151.5 In-situ concrete variability 171.5.1 Within-member variability 181.5.2 In-situ strength relative to standard specimens 211.6 Interpretation 231.6.1 Computation of test results 231.6.2 Examination of variability 231.6.3 Calibration and application of test results 261.7 Test combinations 321.7.1 Increasing confidence level of results 321.7.2 Improvement of calibration accuracy 321.7.3 Use of one method as preliminary to another 331.7.4 Test calibration 331.7.5 Diagnosis of causes of deterioration 331.8 Documentation by standards 352 Surface hardness methods 362.1 Rebound test equipment and operation 362.2 Procedure 39

viContents2.3 Theory, calibration and interpretation 402.3.1 Factors influencing test results 402.3.2 Calibration 452.3.3 Interpretation 462.3.4 Applications and limitations 483 Ultrasonic pulse velocity methods 513.1 Theory of pulse propagation through concrete 523.2 Pulse velocity equipment and use 533.2.1 Equipment 533.2.2 Use 553.3 Test calibration and interpretation of results 603.3.1 Strength calibration 613.3.2 Practical factors influencing measuredresults 633.4 Applications 723.4.1 Laboratory applications 733.4.2 In-situ applications 733.5 Reliability and limitations 814 Partially destructive strength tests 824.1 Penetration resistance testing 824.1.1 Windsor probe 834.1.2 Pin penetration test 934.2 Pull-out testing 934.2.1 Cast-in methods 944.2.2 Drilled-hole methods 1014.3 Pull-off methods 1114.4 Break-off methods 1164.4.1 Norwegian method 1164.4.2 Stoll tork test 1184.4.3 Shearing-rib method 1195 Cores 1205.1 General procedures for core cutting andtesting 1205.1.1 Core location and size 1205.1.2 Drilling 1225.1.3 Testing 1245.2 Interpretation of results 1285.2.1 Factors influencing measured core compressivestrength 1285.2.2 Estimation of cube strength 1315.2.3 Reliability, limitations and applications 133

Contentsvii5.3 Small cores 1355.3.1 Influence of specimen size 1365.3.2 Reliability, limitations and applications 1386 Load testing and monitoring 1406.1 In-situ load testing 1416.1.1 Testing procedures 1416.1.2 Load application techniques 1446.1.3 Measurement and interpretation 1496.1.4 Reliability, limitations and applications 1556.2 Monitoring 1576.2.1 Monitoring during construction 1576.2.2 Long-term monitoring 1576.3 Strain measurement techniques 1636.3.1 Methods available 1646.3.2 Selection of methods 1706.4 Ultimate load testing 1706.4.1 Testing procedures and measurementtechniques 1716.4.2 Reliability, interpretation and applications 1747 Durability tests 1767.1 Corrosion of reinforcement and prestressing steel 1767.1.1 Electromagnetic cover measurement 1797.1.2 Half-cell or rest-potential measurement 1857.1.3 Resistivity measurements 1917.1.4 Direct measurement of corrosion rate 1957.2 Moisture measurement 2017.2.1 Simple methods 2017.2.2 Neutron moisture gauges 2027.2.3 Electrical methods 2027.2.4 Microwave absorption 2047.3 Absorption and permeability tests 2067.3.1 Initial surface absorption test 2087.3.2 Figg air and water permeability tests 2127.3.3 Combined ISAT and Figg methods 2157.3.4 Germann Gas permeability test 2157.3.5 ‘Autoclam’ permeability system 2157.3.6 Other non-intrusive water and air methods 2167.3.7 Flow tests 2177.3.8 BS absorption test 2187.3.9 ‘Sorptivity’ test 2197.3.10 Capillary rise test 2207.4 Tests for alkali–aggregate reaction 220

viiiContents7.5 Tests for freeze–thaw resistance 2217.6 Abrasion resistance testing 2218 Performance and integrity tests 2238.1 Infrared thermography 2238.2 Radar 2258.2.1 Radar systems 2268.2.2 Structural applications and limitations 2298.3 Dynamic response testing 2368.3.1 Simple ‘non-instrumented’ approaches 2368.3.2 Pulse-echo techniques 2378.3.3 Analysis of surface waves 2438.3.4 Testing large-scale structures 2448.4 Radiography and radiometry 2448.4.1 X-ray radiography 2458.4.2 Gamma radiography 2468.4.3 Gamma radiometry 2478.5 Holographic and acoustic emission techniques 2498.5.1 Holographic techniques 2498.5.2 Acoustic emission 2508.6 Photoelastic methods 2538.7 Maturity and temperature-matched curing 2538.7.1 Maturity measurements 2548.7.2 Temperature-matched curing 2578.8 Screed soundness tester 2588.9 Tests for fire damage 2589 Chemical testing and allied techniques 2619.1 Sampling and reporting 2629.1.1 Sampling 2629.1.2 Reporting 2639.2 Cement content and aggregate/cement ratio 2649.2.1 Theory 2649.2.2 Procedures 2649.2.3 Reliability and interpretation ofresults 2689.3 Original water content 2699.3.1 Theory 2699.3.2 Procedure 2709.3.3 Reliability and interpretation of results 2719.4 Cement type and cement replacements 2729.4.1 Theory 2729.4.2 Procedures 2729.4.3 Reliability and interpretation of results 274

PrefaceInterest in testing of hardened concrete in-situ has increased considerablysince the 1960s, and significant advances have been made in techniques,equipment and methods of application since publication of the first editionof this book in 1982. This has largely been a result of the growing numberof concrete structures, especially those of relatively recent origin, that havebeen showing signs of deterioration. Changes in cement manufacture andspecifications, increased use of cement replacements and admixtures, anda decline in standards of workmanship and construction supervision haveall been blamed as has inadequate international standards, especially whereexposure to chlorides is concerned. Particular attention has thus been paidto development of test methods which are related to durability performanceand integrity. There has also been increasing awareness of the shortcomingsof control or compliance tests which require a 28-day wait before results areavailable and even then reflect only the adequacy of the material suppliedrather than overall construction standards. Recognition is slowly growingthat in-situ tests potentially have much to offer in this situation.In each case the need for in-situ measurements is clear, but to many engineersthe features, and especially the limitations, of available test methodsare unknown and consequently left to ‘experts’. Although it is essentialthat the tests should be performed and interpreted by experienced specialists,many difficulties arise both at the planning and interpretation stagesbecause of a lack of common understanding. A great deal of time, effortand money can be wasted on unsuitable or badly planned testing, leadingto inconclusive results which then become the subject of heated debate.The principal aim of this book is to provide an overview of the subjectfor non-specialist engineers who are responsible for the planning orcommissioning of test programmes. The scope is wide in order to covercomprehensively as many aspects as possible of the testing of hardenedconcrete in structures. The tests, however, are treated in sufficient depthto create a detailed awareness of procedures, scope and limitations, andto enable meaningful discussions with specialists about specific methods.Carefully selected references are also included for the benefit of those who

Prefacexiwish to study particular methods in greater detail. The information anddata contained in the book have been gathered from a wide variety of internationalsources. In addition to established methods, new techniques whichshow potential for future development are outlined, although in many casesthe application of these to concrete is still at an early stage and of limitedpractical value at present. Emphasis has been placed on the reliability andlimitations of the various techniques described, and the interpretation ofresults is discussed from the point of view both of specification complianceand application to design calculations. A number of illustrative exampleshave been included with this in mind.In preparing this fourth edition the original author and his colleagueDr Steve Millard have been joined by Mr Mike Grantham who has a wealthof practical industrial experience with many of the techniques described. Hehas also previously assisted with material on chemical analysis in previouseditions. The opportunity has been taken to reflect trends in equipmentand procedures which have developed over the past nine years. A substantialamount of new research has been published on a worldwide basisincluding developments of understanding, procedures, interpretation andapparatus. An important feature has been the continuation of general movesto automate test methods, particularly in terms of data collection, storageand presentation. It is important to recognize that this does not necessarilyimply increased accuracy, although developments of digital technologyhave led to significant enhancements of capabilities in many cases. Interestin the application of statistical methods to interpretation of strength testresults has continued to grow, especially in the USA. A key feature on theUK/European scene is the introduction of new ‘Euronorm’ Standards (EN)which have recently replaced some of the well-established British Standards.These, unfortunately, often provide less detailed guidance on interpretationand application of results. This process is not yet complete and more willfollow. Several other important ‘guidance’ documents have been publishedby industrial bodies and details have been incorporated into Chapter 1 withappropriate referencing elsewhere.The growing importance of performance monitoring, including the propertiesof materials in the surface zone, is reflected in Chapter 6, whilstChapter 7 which deals with durability has a good deal of new informationrelating to corrosion assessment. Information on localized dynamicresponse tests has also been enhanced in Chapter 8. These areas haveseen significant developments of apparatus and procedures since the thirdedition was prepared. The coverage of sub-surface radar, which has nowbecome an established technique, is similarly further increased in Chapter 8,whilst many other developments have also been incorporated throughoutthe book. References to Standards have been updated and a significantnumber of recent new references have been added, many replacing those

xiiPrefacewhich have become outdated. Photographs of representative commerciallyavailable equipment have similarly been updated and extended.The basic testing techniques will be similar in all parts of the world,although national Standards may introduce minor procedural variationsand units will of course differ. This book has been based on the SI unitscurrently in use in Britain, and where reference to Codes of Practice has beennecessary, emphasis is placed on the current recommendations of British orEuropean Standards. The most recent versions of Standards should alwaysbe consulted, since recommendations will inevitably be modified from timeto time.We are very grateful to many engineers worldwide for discussions inwhich they have provided valuable advice and guidance. Particular thanksare due to members of the former BSI Subcommittee CAB/4/2 (NondestructiveTesting of Concrete) for the stimulation provided for early editionsby their contributions to meetings of that subcommittee, and to ourcolleague at Liverpool, Mr R.G. Tickell, for his assistance with statisticalmaterial. Photographs have been kindly provided by many individuals andcompanies as indicated in the relevant captions and their contributions aregratefully acknowledged. Thanks are also due to Ms M.A. Revell for typingthe original manuscript, to Ms A. Ventress for typing new material associatedwith the third edition, Mrs Grace Martin for secretarial assistance withthis edition and Mrs B. Cotgreave for preparation of the original diagrams.J.H.B.S.G.M.M.G.G.

Chapter 1Planning and interpretation ofin-situ testingA great deal of time, effort and expense can be wasted on in-situ testingunless the aims of the investigation are clearly established at the outset.These will affect the choice of test method, the extent and location ofthe tests and the way in which the results are handled – inappropriate ormisleading test results are often obtained as a result of a genuine lack ofknowledge or understanding of the procedures involved. If future disputesover results are to be avoided, liaison of all parties involved is essential at anearly stage in the formulation of a test programme. Engineering judgementis inevitably required when interpreting results, but the uncertainties canoften be minimized by careful planning of the test programme.A full awareness of the range of tests available, and in particular theirlimitations and the accuracies that can be achieved, is important if disappointmentand disillusion is to be avoided. Some methods appear to bevery simple, but all are subject to complex influences and the use of skilledoperators and an appropriately experienced engineer is vital.In-situ testing of existing structures is seldom cheap, since complex accessarrangements are often necessary and procedures may be time-consuming.Ideally a programme should evolve sequentially, in the light of resultsobtained, to provide the maximum amount of worthwhile information withminimum cost and disruption. This approach, which requires ongoing interpretation,will also facilitate changes of objectives which may arise duringthe course of an investigation.1.1 Aims of in-situ testingThree basic categories of concrete testing may be identified.(i) Control testing is normally carried out by the contractor or concreteproducer to indicate adjustments necessary to ensure an acceptablesupplied material.(ii) Compliance testing is performed by, or for, the engineer according toan agreed plan, to judge compliance with the specification.

2 Planning and interpretation of in-situ testing(iii) Secondary testing is carried out on hardened concrete in, or extractedfrom, the structure. This may be required in situations where there isdoubt about the reliability of control and compliance results or they areunavailable or inappropriate, as in an old, damaged or deterioratingstructures. All testing which is not planned before construction will bein this category, although long-term monitoring is also included.Control and compliance tests have traditionally been performed on ‘standard’hardened specimens made from samples of the same concrete as usedin a structure; it is less common to test fresh concrete. There are alsoinstances in which in-situ tests on the hardened concrete may be used forthis purpose. This is most common in the precasting industry for checkingthe quality of standardized units, and the results can be used to monitorthe uniformity of units produced as well as their relationship to some preestablishedminimum acceptable value. There is, generally, an increasingawareness amongst engineers that ‘standard’ specimens, although notionallyof the same material, may misrepresent the true quality of concrete actuallyin a structure. This is due to a variety of causes, including non-uniformsupply of material and differences of compaction, curing and general workmanship,which may have a significant effect on future durability. As aresult, a trend towards in-situ compliance testing, using methods whichare either non-destructive or cause only very limited damage, is emerging,particularly in North America and Scandinavia. Such tests are most commonlyused as a back-up for conventional testing, although there are notableinstances such as the Storebaelt project where they have played a major role(1). They offer the advantage of early warning of suspect strength, as well asthe detection of defects such as inadequate cover, high surface permeability,voids, honeycombing or use of incorrect materials which may otherwise beunknown but lead to long-term durability problems. Testing of the integrityof repairs is another important and growing area of application.The principal usage of in-situ tests is nevertheless as secondary testing,which may be necessary for a wide variety of reasons. These fall into twobasic categories.1.1.1 Compliance with specificationRecent changes to the standards system in the United Kingdom have seenthe introduction of a new European Standard on concrete production andcompliance (2) coupled with complementary British Standards on concretematerials. For the first time these have placed the burden of demonstratingcompliance of the concrete with the specification on the supplier. To thisend the UK Quality Scheme for Ready Mixed Concrete has set stringentstandards for producers to meet to demonstrate compliance of the concretethey supply. There is still the option for clients to carry out so-called ‘identity

Planning and interpretation of in-situ testing 3tests’ to confirm that the concrete supplied is the correct concrete and it islikely that identity testing will continue to be used at least to some extentby many clients.Where doubt exists that the concrete in the structure meets the specifications,secondary testing may be called for. Retrospective testing may alsobe required following deterioration of the structure. The most commonexample where additional evidence is required is in contractual disputesfollowing non-compliance of standard specimens, or where doubt existsfollowing identity tests. Other instances involve retrospective checking followingdeterioration of the structure, and will generally then be related toapportionment of blame in legal actions. Strength requirements form animportant part of most specifications, and the engineer must select the mostappropriate methods of assessing the in-situ strength on a representativebasis, with full knowledge of the likely variations to be expected withinvarious structural members (as discussed in Section 1.5). The results shouldbe interpreted to determine in-situ variability as well as strength, but amajor difficulty arises in relating measured in-situ strength to anticipatedcorresponding ‘standard’ specimen strength at a specific but different age.Borderline cases may thus be difficult to prove conclusively. This problemis examined in detail in Section 1.5.2.Minimum cement content will usually be specified to satisfy durabilityrequirements, and chemical or petrographic tests may be necessary toconfirm compliance. Similar tests may also be required to check for the presenceof forbidden admixtures, contamination of concrete constituents (e.g.chlorides in sea-dredged aggregates) or entrained air, and to verify cementcontent following deterioration. Poor workmanship is often the principalcause of durability problems, and tests may also be aimed at demonstratinginadequate cover or compaction, incorrect reinforcement quantities orlocation, or poor quality of curing or specialist processes such as groutingof post-tensioned construction.1.1.2 Assessment of in-situ quality and integrityThis is primarily concerned with the current adequacy of the existing structureand its future performance. Routine maintenance needs of concretestructures are now well established, and increasingly utilize in-situ testingto assist ‘lifetime predictions’ (3,4). It is important to distinguish betweenthe need to assess the properties of the material, and the performance of astructural member as a whole. The need for testing may arise from a varietyof causes, which include(i) Proposed change of usage or extension of a structure(ii) Acceptability of a structure for purchase or insurance

4 Planning and interpretation of in-situ testing(iii) Assessment of structural integrity or safety following material deterioration,or structural damage such as caused by fire, blast, fatigue oroverload(iv) Serviceability or adequacy of members known or suspected to containmaterial which does not meet specifications, or with design faults(v) Assessment of cause and extent of deterioration as a preliminary tothe design of repair or remedial schemes(vi) Assessment of the quality or integrity of applied repairs(vii) Monitoring of strength development in relation to formwork stripping,curing, prestressing or load application(viii) Monitoring long-term changes in materials properties and structuralperformance.Although in specialized structures, features such as density or permeabilitymay be relevant, generally it is either the in-situ strength or durabilityperformance that is regarded as the most important criterion. Where repairsare to be applied using a different material from the ‘parent’ concrete, it maybe desirable to measure the elastic modulus to determine if strain incompatibilitiesunder subsequent loading may lead to a premature failure of therepair. A knowledge of elastic modulus may also be useful when interpretingthe results of load tests. For strength monitoring during construction, it willnormally only be necessary to compare test results with limits establishedby trials at the start of the contract, but in other situations a predictionof actual concrete strength is required to incorporate into calculations ofmember strength. Where calculations are to be based on measured in-situstrength, careful attention must be paid to the numbers and location of testsand the validity of the safety factors adopted, and this problem is discussedin Section 1.6.Durability assessments will concentrate upon identifying the presenceof internal voids or cracking, materials likely to cause disruptions of theconcrete (e.g. sulfates or alkali-reactive aggregates), and the extent or riskof reinforcement corrosion. Carbonation depths, chloride concentrations,cover thicknesses, and surface zone resistivity and permeability will bekey factors relating to corrosion. Electrochemical activity associated withcorrosion can also be measured to assess levels of risk, using passive orperturbative test methods.Difficulties in obtaining an accurate quantitative estimate of in-situ concreteproperties can be considerable: wherever possible the aim of testingshould be to compare suspect concrete with similar concrete in other partsof the structure which is known to be satisfactory or of proven quality.Investigation of the overall structural performance of a member is frequentlythe principal aim of in-situ testing, and it should be recognized thatin many situations this would be most convincingly demonstrated directlyby means of a load test. The confidence attached to the findings of the

Planning and interpretation of in-situ testing 5investigation may then be considerably greater than if member strengthpredictions are derived indirectly from strength estimates based on in-situmaterials tests. Load testing may however be prohibitively expensive orsimply not a practical proposition.The authors have frequently demonstrated the usefulness of SchmidtHammer and UPV methods applied to cubes at the time of testing whichcan then allow simple correlation with concrete in the structure where aproblem is suspected.1.2 Guidance available from ‘standards’ and otherdocumentsNational standards are available in a number of countries, notably the UK,USA and Scandinavia, detailing procedures for the most firmly establishedtesting methods. Principal British, European and ASTM standards are listedat the end of this chapter and specific references are also included in thetext. Details of all methods are otherwise contained in an extensive body ofpublished specialist research papers, journals, conference proceedings andtechnical reports. References to a key selection of these are provided asappropriate.General guidance concerning the philosophy of maintenance inspection ofexisting structures is provided by FIP (5) and also by the Institution of StructuralEngineers (6), who consider appraisal processes and methods as wellas testing requirements. Advice is also offered on sources of information,reporting and identification of defects with their possible causes. Specificguidance on damage classification is proposed by RILEM (7) whilst ACIcommittee 364 have produced a guide for evaluation of concrete structuresprior to rehabilitation (8). Guidance relating to assessment approaches tospecialized situations such as high alumina cement concrete (9), fire (10) andbomb-damaged structures (11) is also available. BS 1881: Part 201, Guideto the use of non-destructive methods of test for hardened concrete (12),provides outline descriptions of 23 wide-ranging methods, together withguidance on test selection and planning, whilst BS 6089 (13) relates specificallyto in-situ strength assessment. Both these latter standards currentlyhave no European equivalent although one is currently under developmentrelating to in-situ compressive strength. Methods and apparatus which arecommercially available are constantly changing and developing, but CIRIATechnical Note 143 (14) reviewed those existing in the UK in 1992 whilstSchickert has outlined the situation in Germany in 1994 (15). A Germancompendium of test methods is available on the Internet (16). Carino hasalso reviewed the worldwide historical development of non-destructive testingof concrete from the North American perspective and has identifiedfuture prospects (17) whilst current practice in other parts of the world hasbeen reviewed by several authors (18,19,20,21). As newer methods become

6 Planning and interpretation of in-situ testingestablished it is likely that further standards and reports will appear. ACICommittee 228 has produced two major reports on NDT of concrete structures(22) and in-situ strength testing (23) which are regularily updated.RILEM Committees have recently considered in-place strength testing andnear-surface durability testing (NEC) whilst current committees are lookingat interpretation of NDT results (INR) and Acoustic Emission (ACD). In theUK, the Concrete Society has prepared recent Technical Reports on reinforcementcorrosion assessment (24) and subsurface radar methods (25),and the Highways Agency have published Advice Notes on NDT of highwaystructures embodying recent research with developing techniques (26).1.3 Test methods availableDetails of individual methods are given in subsequent chapters and maybe classified in a variety of ways. Table 1.1 lists the principal tests interms of the property under investigation. The range of available tests islarge, and there are others which are not included in the table but aredescribed in this book. Visual inspection, assisted where necessary by opticaldevices, is a valuable assessment technique which must be included in anyinvestigation. There will of course be overlap of usage of some tests betweenthe applications listed (see Section 1.4.3), and where a number of optionsare available considerations of access, damage, cost, time and reliability willbe important.The test methods may also be classified as follows:Non-destructive methods. Non-destructive testing is generally definedas not impairing the intended performance of the element or memberunder test, and when applied to concrete is taken to includemethods which cause localized surface zone damage. Such tests arecommonly described as partially destructive and many of those listedin Table 1.1 are of this type. All non-destructive methods can be performeddirectly on the in-situ concrete without removal of a sample,although removal of surface finishes is likely to be necessary.Methods requiring sample extraction. Samples are most commonlytaken in the form of cores drilled from the concrete, which may beused in the laboratory for strength and other physical tests as wellas visual, petrographic and chemical analysis. Some chemical testsmay be performed on smaller drilled powdered samples taken directlyfrom the structure, thus causing substantially less damage, but the riskof sample contamination is increased and precision may be reduced.However the authors have seen results taken from a series of fourdrilled holes around core samples which showed superior precisionand accuracy when tested for cement content. Making good the samplingdamage will be necessary, as with partially destructive methods.

Planning and interpretation of in-situ testing 7Table 1.1 Principal test methodsProperty underinvestigationTestEquipment typeCorrosion of Half-cell potentialElectrochemicalembedded ResistivityElectricalsteelLinear polarization resistance ElectrochemicalAC ImpedanceElectrochemicalCover depthElectromagneticCarbonation depthChemical/microscopicChloride concentration Chemical/electricalConcrete quality, Surface hardnessMechanicaldurability and Ultrasonic pulse velocity Electromechanicaldeterioration RadiographyRadioactiveRadiometryRadioactiveNeutron absorptionRadioactiveRelative humidityChemical/electronicPermeabilityHydraulicAbsorptionHydraulicPetrographicMicroscopicSulfate contentChemicalExpansionMechanicalAir contentMicroscopicCement type and content Chemical/microscopicAbrasion resistanceMechanicalConcrete strength Cores MechanicalPull-outMechanicalPull-offMechanicalBreak-offMechanicalInternal fractureMechanicalPenetration resistance MechanicalMaturityChemical/electricalTemperature-matched curing Electrical/electronicIntegrity and TappingMechanicalperformance Pulse-echoMechanical/electronicDynamic responseMechanical/electronicAcoustic emissionElectronicThermoluminescence ChemicalThermographyInfraredRadarElectromagneticReinforcement location ElectromagneticStrain or crack measurement Optical/mechanical/electricalLoad testMechanical/electronic/ electricalThe nature of the testing equipment ranges from simple inexpensivehand-held devices to complex, expensive, highly specialized items, possiblyrequiring extensive preparation or safety precautions, which will be usedonly where no simple alternative exists. Few of the methods give direct quantitativemeasurement of the desired property, and correlations will often be

8 Planning and interpretation of in-situ testingnecessary. Practical limitations, reliability and accuracy vary widely and arediscussed in the sections of this book dealing with the various individualmethods. Selection of the most appropriate tests within the categories ofTable 1.1 is discussed in Section 1.4.3 of this chapter.1.4 Test programme planningThis involves consideration of the most appropriate tests to meet the establishedaims of the investigation, the extent or number of tests requiredto reflect the true state of the concrete, and the location of these tests.Investigations have been made into the use of Expert Systems to assist thisprocess but at the present time it seems likely that their application willbe largely confined to a training role (27). Detailed guidance regardingspecification and pricing is given by the UK Concrete Bridge DevelopmentGroup (19) relating to durability assessment. Visual inspection is an essentialfeature whatever the aims of the test programme, and will enable themost worthwhile application of the tests which have been summarized inSection 1.3. Some typical illustrative examples of test programmes to meetspecific requirements are given in Appendix A. Engineers should rememberthat sampling of both good and bad areas is very useful. Comparing thetwo can often reveal the cause of problems. Simply sampling only bad areascan make judgement more difficult.1.4.1 General sequential approachA properly structured programme is essential, with interpretation as anongoing activity, whatever the cause or nature of an investigation. Figure 1.1illustrates the stages typically involved, which will generally require increasingcost commitment, and the investigation will proceed only as far as isnecessary to reach firm relevant conclusions.1.4.2 Visual inspectionThis can often provide valuable information to the well-trained eye. Visualfeatures may be related to workmanship, structural serviceability and materialdeterioration, and it is particularly important that the engineer be ableto differentiate between the various types of cracking which may be encountered.Figure 1.2 illustrates a few of these in their typical forms.Segregation or excessive bleeding at shutter joints may reflect problemswith the concrete mix, as might plastic shrinkage cracking, whereas honeycombingmay be an indication of low standards of construction workmanship.Lack of structural adequacy may show itself by excessive deflectionor flexural cracking, and this may frequently be the reason for an in-situassessment of a structure. Long-term creep deflections, thermal movements

Planning and interpretation of in-situ testing 9Establish aims andinformation requiredDocumentation surveySTAGE 1PlanningPreliminary site visit(access and safety)Agree interpretationcriteriaSTAGE 2NondestructivetestingSTAGE 3FurthertestingSystematic visual inspection,initial test selection & costingsComparative surveyCalibrated assessmentLocalised investigation(cores, break-out, etc.)Load testingAnalysis, interpretation and reportingConclusionsActionsDocumentationof resultsFigure 1.1 Typical stages of test programme.or structural movements may cause distortion of door frames, cracking ofwindows, or cracking of a structure or its finishes. Visual comparison ofsimilar members is particularly valuable as a preliminary to testing to determinethe extent of the problem in such cases.Material deterioration is often indicated by surface cracking and spallingof the concrete, and examination of crack patterns may provide a preliminaryindication of the cause. Considerable caution must however be

10 Planning and interpretation of in-situ testing(a)(b)(c)(d)Figure 1.2 Some typical crack types: (a) reinforcement corrosion; (b) plastic shrinkage;(c) sulfate attack; (d) alkali/aggregate reaction.exercised when attempting to judge the cause of damage by visual appearancealone. Both in-situ and laboratory testing are likely to be needed toconfirm the cause of damage. The most common causes are reinforcementcorrosion due to inadequate cover or high chloride concentrations, and concretedisruption due to sulfate attack, frost action or alkali–aggregate reactions.As shown in Figure 1.2, reinforcement corrosion is usually indicatedby splitting and spalling along the line of bars possibly with rust staining,whereas sulfate attack may produce a random pattern accompanied by awhite deposit leached on the surface. Alkali–aggregate reaction is sometimes(but not necessarily) characterized by a star-shaped crack pattern, andfrost attack may give patchy surface spalling and scabbing. Some furtherexamples with illustrative photographs are given by the Concrete BridgeDevelopment Group (19). Because of similarities it will often be impossibleto determine causes by visual inspection alone, but the most appropriateidentification tests can be selected on this basis. Careful field documentationis important (28) and Pollock, Kay and Fookes (29) suggest that systematic‘crack mapping’ is a valuable diagnostic exercise when determining the

Planning and interpretation of in-situ testing 11causes and progression of deterioration, and they give detailed guidanceabout the recognition of crack types. Non-structural cracking is describedin detail by Concrete Society Technical Report 22 (30), and the symptomsrelating to the most common sources of deterioration are summarized inTable 1.2, which is based on the suggestions of Higgins (31). Observationof concrete surface texture and colour variations may be a useful guide touniformity, and colour change is a widely recognized indicator of the extentof fire damage.Visual inspection is not confined to the surface, but may also includeexamination of bearings, expansion joints, drainage channels, posttensioningducts and similar features of a structure. Binoculars, telescopesand borescopes may be useful where access is difficult and portable ultravioletinspection systems may be useful in identifying alkali–aggregate reactions(see Section 9.11.1). Recently there has been an increasing acceptance of‘unconventional’ methods such as abseiling and robotics to provide costeffectiveinspection and remediation access (32). For existing structures,the existence of some features requiring further investigation is generallyinitially indicated by visual inspection, and it must be considered the singlemost important component of routine maintenance. Recent RILEM(7) proposals attempt to provide a numerical classification system to permitthe quantification of visual features to assist planning and prioritization.Visual inspection will also provide the basis of judgements relatingto access and safety requirements (32) when selecting test methods andtest locations. The authors have seen some frightening examples wherepublic safety has been put at risk due to a lack of simple regular visualinspections.Table 1.2 Diagnosis of defects and deteriorationCauseSymptomsAge of appearanceCracking Spalling Erosion Early Long-termStructural deficiency × × × ×Reinforcement corrosion × × ×Chemical attack × × × ×Frost damage × × × ×Fire damage × × ×Freeze–thaw × × ×Internal reactions × × ×Thermal effects × × × ×Shrinkage × × ×Creep × × ×Rapid drying × ×Plastic settlement × ×Physical damage × × × × ×

12 Planning and interpretation of in-situ testing1.4.3 Test selectionTest selection for a particular situation will be based on a combination offactors such as access, damage, cost, speed and reliability, but the basicfeatures of visual inspection followed by a sequence of tests according toconvenience and suitability will generally apply. The use of combinationsof test methods is discussed in Section 1.7.Testing for durability including causes and extent of deterioration. Relativefeatures of various test methods are summarized in Table 1.3, whilstmore extensive tables of test methods and their selection are given by theConcrete Bridge Development Group (19). Corrosion risk of embeddedreinforcement is related to the loss of passivity which is provided by thealkaline concrete environment. This is usually as a result of carbonationor chlorides. Simple initial tests will thus involve localized measurementsof reinforcement cover, carbonation depths and chloride concentrations.These may be followed by more complex half-cell potential and resistivitytesting to provide a more comprehensive survey of large areas. If excessivecarbonation is found to be the cause of deterioration, then chemical orpetrographic analysis and absorption tests may follow if it is necessary toidentify the reasons for this. Direct measurement of the rate of corrosionof reinforcing steel is slowly gaining acceptance as an effective means ofassessing the severity of ongoing durability damage and has the potentialfor use to predict the remaining service lifetime of a corrosion-afflictedstructure.Table 1.3 Durability tests – relative featuresMethod Cost Speed oftestDamageApplications⎫Cover measurement Low Fast None⎬Corrosion riskCarbonation depth Low Fast Minor⎭ and causeChloride content Low Fast Minor}Half-cell potential Moderate Fast MinorCorrosion riskResistivity Moderate Fast Minor/none⎫Linear polarization Moderate/high Moderate Minor⎪⎬resistanceCorrosion rateAC impedance Moderate/high Slow Minor⎪⎭evaluationGalvanostatic pulse Moderate/high Fast Minor⎫Absorption Moderate Slow Moderate/minorPermeability Moderate Slow Moderate/minor Cause and risk⎪⎬Moisture content Moderate Slow MinorChemical Moderate/high Slow ModeratePetrographic High Slow ModerateExpansion High Slow ModerateRadiography High Slow None⎪⎭of corrosionand concretedeterioration

Planning and interpretation of in-situ testing 13Surface absorption and permeability tests are important in relation tocorrosion since both oxygen and water are required to fuel the process, andcarbonation rates are also governed by moisture conditions and the abilityof atmospheric carbon dioxide to pass through the concrete surface zone.Most other forms of deterioration are also related to moisture which isneeded to transport aggressive chemicals and to fuel reactions, thus moisturecontent, absorption and permeability measurements may again be relevant.Expansion tests on samples of concrete may indicate future performance,and chemical and petrographic testing to assess mix components may berequired to identify the causes of disruption of concrete (33).Testing for concrete strength. Relative features of various concretestrength test methods are summarized in Table 1.4.In the common situation where an assessment of material strength isrequired, it is unfortunate that the complexity of correlation tends to begreatest for the test methods which cause the least damage. Although surfacehardness and pulse velocity tests cause little damage, are cheap and quick,and are ideal for comparative and uniformity assessments, their correlationfor absolute strength prediction poses many problems unless calibrateddirectly on the concrete in question. Core tests provide the most reliable insitustrength assessment but also cause the most damage and are slow andexpensive. They will often be regarded as essential, and their value may beenhanced if they are used to form a basis for calibration of non-destructiveor partially destructive methods which may then be adopted more widely.Table 1.4 Strength tests – relative meritsTest method Cost Speed oftestDamage Representativeness Reliability ofabsolutestrengthcorrelationsGeneral applicationsCores ⎫ High Slow Moderate Moderate GoodPull-out ⎬Penetration Moderate Fast Minor Near surface only Moderateresistance ⎭}Pull-offBreak-offModerate Moderate Minor Near surface onlyModerateInternal fracture Low Fast Minor Near surface only ModerateComparative assessmentUltrasonic pulse Low Fast None Good PoorvelocitySurface hardness Very low Fast Unlikely Surface only PoorStrength development monitoringMaturity Moderate Continuous Very minor Good ModerateTemperaturematchedcuringHigh Continuous Very minor Good Good

14 Planning and interpretation of in-situ testingEngineers must realize, however, that core test results may not relate directlyto results from cube tests made at the time of construction (see Tables 1.6and 1.7 and Section 1.5.2). This issue will also be considered by a newEuropean Standard currently under development. Whilst most test methodscan be successfully applied to concretes made with lightweight aggregates,strength correlations will always be different from those relating to concreteswith normal aggregates (34). Partially destructive methods generallyrequire less-detailed calibration for strength but cause some surface damage,test only the surface zone, and may suffer from high variability. Theavailability and reliability of strength correlations and the accuracy requiredfrom the strength predictions may be important factors in selecting the mostappropriate methods to use. This must be coupled with the acceptability ofmaking good any damaged areas for appearance and structural integrity.When comparison with concrete of similar quality is all that is necessary,the choice of test will be dominated by the practical limitations of the variousmethods. The least destructive suitable method will be used initially,possibly with back-up tests using another method in critical regions. Forexample, surface hardness methods may be used for new concrete, or ultrasonics,where two opposite surfaces are accessible. When there is only oneexposed face, penetration resistance testing is quick and suitable for largemembers such as slabs, but pull-out or pull-off tests may be more suitablefor smaller members. Pull-out testing is particularly useful for directin-situ measurements of early age strength development, while maturityand temperature-matched curing techniques are based on measurements ofwithin-pour temperatures. Increasing emphasis on fast-track constructionhas led to a significant growth of interest in these techniques, and this isconsidered more fully in Chapters 4 and 8.Testing for comparative concrete quality and localized integrity. Comparativetesting is the most reliable application of a number of methodsfor which calibration to give absolute values of a well-defined physicalparameter is not easy. In general, these methods cause little or no surfacedamage, and most are quick to use, enabling large areas to be surveyedsystematically. Some do, however, require relatively complex and expensiveequipment.The most widely used methods are surface hardness, ultrasonic pulsevelocity and chain dragging or surface tapping. The latter is particularlyuseful in locating delamination near to the surface and has been developedwith more complex impact-echo techniques. Surface-scanning radar andinfrared thermography are both sophisticated methods of locating hiddenvoids, moisture and similar features which have recently grown in popularity;radiography and radiometry may also be used. The use of tomographyto identify and locate subsurface features is receiving increasing attention(26,35). This involves taking a series of measurements on the member underinvestigation between different faces to provide a pattern of intersecting ray

Planning and interpretation of in-situ testing 15paths from which a two- or three-dimensional reconstruction can be madeusing appropriate computational software. Wear tests, surface hardnessmeasurements or surface absorption methods may be used to assess surfaceabrasion resistance, and thermoluminescence is a specialized technique toassess fire damage.Testing for structural performance. Large-scale dynamic response testingis available to monitor structural performance, but large-scale staticload tests, possibly in conjunction with monitoring of cracking by acousticemission, may be more appropriate despite the cost and disruption.Static load tests usually incorporate measurement of deflections andcracking, but problems of isolating individual members can be substantial.Where large numbers of similar elements (such as precast beams) areinvolved, it may be better to remove a small number of typical elementsfor laboratory load testing and to use non-destructive methods to comparethese elements with those remaining in the structure.It is essential that the test programme relates the costs of the varioustest methods to the value of the project involved, the costs of delays toconstruction, and the cost of possible remedial works. Accessibility of thesuspect concrete and the handling of test equipment must be considered,together with the safety of site personnel and the general public, duringtesting operations. Detailed risk assessments are required, which must beadhered to by all parties.Typical examples of test programmes suggested for particular situationsare included in Appendix A.1.4.4 Number and location of testsEstablishing the most appropriate number of tests is a compromise betweenaccuracy, effort, cost and damage. Test results will relate only to the specificlocations at which the readings or samples were obtained. Engineeringjudgement is thus required to determine the number and location of tests,and the relevance of the results to the element or member as a whole. Theimportance of integration of planning with interpretation is thus critical.A full understanding of concrete variability (as discussed in Section 1.5)is essential, as well as a knowledge of the reliability of the test methodused. This is discussed here with particular reference to concrete strength,since many other properties are strength-related. This should provide auseful general basis for judgements, and further guidance is contained inthe chapters dealing with the various test methods. If aspects of durabilityare involved, care should be taken to allow for variations in environmentalexposure and test conditions. Corrosion activity may vary significantly withambient fluctuations in temperature and rainfall. Care should be taken whenestimating mean annual behaviour on the basis of measurements taken ona single occasion. Test positions must also take into account the possible

16 Planning and interpretation of in-situ testingTable 1.5 Relative numbers of readings recommended for varioustest methodsTest methodNo. of individual readingsrecommended at a location‘Standard’ cores 3Small cores 9Schmidt hammer 12Ultrasonic pulse velocity 1Internal fracture 6Windsor probe 3Pull-out 4Pull-off 6Break-off 5effects of reinforcement upon results, as well as any physical restrictionsrelating to the method in use.Table 1.5 lists the number of tests which may be considered equivalentto a single result. The accuracy of strength prediction will depend in mostcases on the reliability of the correlation used, but for ‘standard’ cores 95%confidence limits may be taken as ±12/ √ n% where n is the number ofcores from the particular location. Statistical methods taking account of thenumber of tests, test variability and material variability have been developedand are considered more fully in Section 1.6.3. Where cores are being usedto provide a direct indication of strength or as a basis of calibration forother methods, it is important that a sufficient number are taken to providean adequate overall accuracy. It is also essential to remember that theresults will relate only to the particular location tested, thus the number oflocations to be assessed will be a further factor requiring consideration.For comparative purposes the truly non-destructive methods are the mostefficient, since their speed permits a large number of locations to be easilytested. For a survey of concrete within an individual member, at least 40locations are suggested, spread on a regular grid over the member, whereasfor comparison of similar members a smaller number of points on eachmember, but at comparable positions, should be examined. Where it isnecessary to resort to other methods such as internal fracture or Windsorprobe tests, practicalities are more likely to restrict the number of locationsexamined, and the survey may be less comprehensive.In-situ strength estimates determining structural adequacy should ideallybe obtained for critically stressed locations, in the light of anticipatedstrength distributions within members (described in Section 1.5.1). Attentionwill thus often be concentrated on the upper zones of members, unlessparticular regions are suspect.Tests for material specification compliance must be made on typicalconcrete, and hence the weaker top zones of members should be avoided.

Planning and interpretation of in-situ testing 17Testing at around mid-height is recommended for beams, columns andwalls, and surface zone tests on slabs must be restricted to soffits unless thetop layer is first removed. Care must similarly be taken to discard materialfrom the top 20% (or at least 50 mm) of slabs when testing cores.Where specification compliance is being investigated, it is recommendedthat no fewer than four cores be taken from the suspect batch of concrete.Where small cores are used, a larger number will be required to give acomparable accuracy, due to greater test variability, and probably at least12 results are required. With other test methods, a minimum number ofreadings is less clearly defined but should reflect the values given in Table 1.5coupled with the calibration reliability. Likely maximum accuracies aresummarized in Section 1.6. It is inevitable that a considerable ‘grey’ or‘not proven’ area will exist when comparing strength estimates from in-situtesting with specified cube or cylinder strengths, and a best possible accuracyof ±15% has been suggested for a group of four cores (36). This valuemay increase when dealing with old concrete, due to uncertainties aboutage effects on strength development. This needs to be added to the likelydifference in actual values obtained from cores and cubes which furthercomplicates the issue. Tests may, however, sometimes be necessary on areaswhich show signs of poor compaction or workmanship for comparisonwith other aspects of specifications.The number of load tests that can be undertaken on a structure willbe limited, and these should be concentrated on critical or suspect areas.Scaffolding support in the area to be tested is always required in case ofcollapse.Visual inspection and non-destructive tests may be valuable in locatingsuch regions. Where individual members are to be tested destructively toprovide a calibration for non-destructive methods, they should preferablybe selected to cover as wide a range of concrete quality as possible.1.5 In-situ concrete variabilityIt is well established that the properties of in-situ concrete will vary within amember, due to differences of compaction and curing as well as non-uniformsupply of material. Supply variations will be assumed to be random, butcompaction and curing variations follow well-defined patterns according tomember type. A detailed appreciation of these variations is essential to planningany in-situ test programme and also to permit sensible interpretationof results.The average in-situ strength of a member, expressed as the strength ofan equivalent cube, will almost invariably be less than that of a standardcube of the same concrete which has been properly compacted and moistcuredfor 28 days. The extent of the difference will depend upon materialscharacteristics, construction techniques, workmanship and exposure, but

18 Planning and interpretation of in-situ testinggeneral patterns can be defined according to member type. This aspect,which is particularly important for interpretation of test results, is discussedin detail in Section 1.5.2. and in Chapter 5 dealing with core testing.1.5.1 Within-member variabilityVariations in concrete supply will be due to differences in materials, batching,transport and handling techniques. These will reflect the degree ofcontrol over production and will normally be indicated by control and compliancetest specimens in which other factors are all standardized. In-situmeasurement of these variations is difficult because of the problem of isolatingthem from compaction and curing effects. They may however be roughlyassessed by consideration of the coefficient of variation of tests taken at anumber of comparable locations within a member or structure. Compactionand curing effects will depend partially upon construction techniques butare also closely related to member types and location within the member.Reinforcement may hinder compaction but there will be a tendency formoisture to rise and aggregate to settle during construction. Lower levelsof members will further be compacted due to hydrostatic effects, related tomember depth, with the result that the general tendency will be for strengthsto be highest near the base of pours and lowest in the upper regions. Thebasic aim of curing is to ensure that sufficient water is present to enablehydration to proceed. For low water: cement ratio mixes, self-desiccationmust be avoided by allowing water ingress, and for other mixes, drying outmust be prevented. Incomplete hydration resulting from poor curing maycause variations of strength between interior and surface zones of members.A figure of about 5–10% has been suggested for this effect in gravel concretes(37); higher values may apply to lightweight concretes (38). Temperaturerises due to cement hydration may cause further strength differencesbetween the interior and outer regions, especially at early ages. Differentialcuring across members may serve to further increase the variations fromcompactional factors.Typical relative strength variations for normal concretes according to membertype are illustrated in Figure 1.3. These results have been derived fromnumerous reports of non-destructive testing including that by Maynard andDavis (39) and can only be regarded as indicating general trends which may beexpected, since individual construction circumstances may vary widely. Forbeams and walls the strength gradients will be reasonably uniform, althoughvariations in compaction and supply may cause the type of variability indicatedby the relative strength contours of Figures 1.4 and 1.5. Few data areavailable for slabs, but it has been suggested that the reduced differentialof about 25% across the depths may be concentrated in the top 50 mm inthin slabs (37). Thicker slabs will be more similar to beams. Variations inplan may, however, be expected to be random due to compaction and supplyinconsistencies although the authors have found a tendency to reduced

Planning and interpretation of in-situ testing 19Figure 1.3 Within-member variations.Figure 1.4 Typical relative percentage strength contours for a beam.strength in corner regions. Columns may be expected to be reasonably uniformexcept for a weaker zone in the top 300 mm or 20% of their depth (40).Further relevant data has been provided by Bartlett and MacGregor (41).It is important to recognize that non-standard concretes may be expectedto behave in a manner different from those described above. In particular,Miao et al. (42) have demonstrated that high-strength concretes (up to120 N/mm 2 cylinder strength) exhibit significantly smaller strength reductionsover the height of 1 m 2 columns than a 35 N/mm 2 concrete, which

20 Planning and interpretation of in-situ testingFigure 1.5 Typical relative percentage strength contours for a wall.was shown to be reasonably consistent with Figure 1.3. General in-situvariability at a particular height was also found to be smaller at highstrengths. Lightweight aggregate concretes have also been shown to havesmaller within-depth variations in beams than gravel concrete accordingto aggregate type and the nature of fine material used (38). This is illustratedin Figure 1.6, which also incorporates differences in in-situ strengthTop3 4LecaAll-LytagLytagPelliteMiddle1 4GravelBottom0.6 0.7 0.8 0.9 1.0 1.1Estimated in-situ cube strengthStandard cube strengthFigure 1.6 Average relative strength distributions within beams of different concrete types(based on ref. 34).

Planning and interpretation of in-situ testing 21relative to ‘standard’ cube strength (34), discussed in Section 1.5.2 below.The most significant reduction in variation can be seen to have occurredwhen lightweight fines were used, and general within-member variabilitywas also reduced in this case.1.5.2 In-situ strength relative to standard specimensLikely strength variations within members have been described inSection 1.5.1. If measured in-situ values are expressed as equivalent cubestrengths, it will usually be found that they are less than the strengths ofcubes made of concrete from the same mix which are compacted and curedin a ‘standard’ way. In-situ compaction and curing will vary widely, andother factors such as mixing, bleeding and susceptibility to impurities aredifficult to predict. Nevertheless a general trend according to member typecan be identified and the values given in Table 1.6 may be regarded as typical.Although these are generally accepted (13), and appear to be generallysupported by recent reports (43,44), cases have been reported where in-situstrengths were found to be closer to that of standard specimens (45) andthis is also likely for lightweight aggregate concretes (see Figure 1.6). Thelikely relationships between standard specimen strength and in-situ strengthare also illustrated in Figure 1.7 for a typical structural concrete mix usingnatural aggregates.A ‘standard’ cube is tested whilst saturated, and for ease of comparisonthe values of Table 1.6 are presented on this basis also. Dry cubes generallyyield strengths which are approximately 10–15% higher, and this must beappreciated when interpreting in-situ strength test results. Cores will betested while saturated under normal circumstances, and the above relationshipswill apply, but if the in-situ concrete is dry the figures for likely in-situstrength must be increased accordingly. Where non-destructive or partiallydestructive methods are used in conjunction with a strength calibration, it isessential to know whether this calibration is based on wet or dry specimens.Another feature of such calibrations is the size of cube upon which they areTable 1.6 Comparison of in-situ and ‘standard’ cube strengthsMember typeTypical 28-day in-situ equivalent wet cubestrength as % of ‘standard’ cube strengthAverageLikely rangeColumn 65 55–75Wall 65 45–95Beam 75 60–100Slab 50 40–60

22 Planning and interpretation of in-situ testingfresh concretespecified‘standard’cube‘standard’cylindergrade35mean40–43N/mm 2ormean32–35N/mm 2top22–27 18–22 24–29 16–22N/mm 2 N/mm 2 N/mm 2 N/mm 2bottomstructuralmember30–36 38–46 40–48 24–28N/mm 2 N/mm 2 N/mm 2 N/mm 2columnwall beam slabTypical in-situ equivalent 28-day cube strengthFigure 1.7 Typical relationship between standard specimen and in-situ strength.based. Design and specification are usually based on a 150 mm cube, butlaboratory calibrations may sometimes be related to a 100 mm cube whichmay be up to 4% stronger.The age at which the concrete is tested is a further cause of differencesbetween in-situ and ‘standard’ values. Although ‘age correction’ factorsare given in some Codes of Practice, care is needed when attempting toadjust in-situ measurements to an equivalent 28-day value. Developmentsin cement manufacture over the years have tended towards yielding a highearly strength with reduced long-term increases, and strength developmentalso is largely dependent on curing. If concrete is naturally wet the strengthmay increase, but often concrete is dry in service and unlikely to makesignificant gains after 28 days.The incorporation of cement replacements such as pulverized fuel ash orground granulated blast-furnace slag (GGBS) into the mix will also influencelonger-term strength development characteristics, and age adjustmentsshould be treated with caution in such cases.

1.6 InterpretationPlanning and interpretation of in-situ testing 23Interpretation of in-situ test results may be considered in three distinctphases leading to the development of conclusions:(i) Computation(ii) Examination of variability(iii) Calibration and/or application.The emphasis will vary according to circumstances (detailed interpretativeinformation is given in other chapters) but the principles will be similarwhatever procedures are used, and these are outlined below. The examplesof Appendix A further illustrate the application of those procedures to anumber of simple commonly occurring situations.The need for comprehensive and detailed recording and reportingof results is of considerable significance, no matter how small orstraightforward the investigation may at first appear to be. In the eventof subsequent dispute or litigation, the smallest detail may be crucial anddocumentation should always be kept with this in mind. Comprehensivephotographs are often of particular value for future reference. In-situ testresults are also increasingly being incorporated into computer databases,associated with prioritization and management of maintenance and repairstrategies (27).1.6.1 Computation of test resultsThe amount of computation required to provide the appropriate parameterat a test location will vary according to the test method but will followwell-defined procedures. For example, cores must be corrected for length,orientation and reinforcement to yield an equivalent cube strength.Pulse velocities must be calculated making due allowance for reinforcementand pull-out, penetration resistance and surface hardness tests mustbe averaged to give a mean value. Attempts should not be made at this stageto invoke correlations with a property, other than that measured directly.Chemical or similar tests will be evaluated to yield the appropriate parametersuch as cement content or mix proportions. Load tests will usually besummarized in the form of load/deflection curves with moments evaluatedfor critical conditions, and creep and recovery indicated as described inChapter 6.1.6.2 Examination of variabilityWhenever more than one test is carried out, a comparison of the variabilityof results can provide valuable information. Even where few results

24 Planning and interpretation of in-situ testingare available (e.g. in load tests), these provide an indication of the uniformityof the construction and hence the significance of the results. In caseswhere more numerous results are available, as in non-destructive surveys,a study of variability can be used to define areas of differing quality. Thiscan be coupled with a knowledge of test variability associated with themethod to provide a measure of the construction standards and controlused.Tomsett (46) has reported the development of an analysis procedure foruse on large-scale integrity assessment projects involving a coefficient ofvariation ratio relating local variability to expected values, an area factorrelating the area of the assessed problem to the total area and a comparativedamage factor. Interpretation is facilitated by the use of interaction diagramsincorporating these three parameters. Some test methods such as radar andimpact-echo rely on recognition of characteristic patterns of test results,and the possibilities for application of neural networks to such cases arecurrently being studied.1.6.2.1 Graphical methods‘Contour’ plots showing, for example, zones of equal strength (Figures 1.4and 1.5) are valuable in locating areas of concrete which are abnormallyhigh or low in strength relative to the remainder of the member. Such contoursshould be plotted directly on the basis of the parameter measured(e.g. pulse velocity) rather than after conversion to strength. Under normalcircumstances the contours will follow well-defined patterns, and anydeparture from this pattern will indicate an area of concern. ‘Contour’ plotsare also valuable in showing the range of relative strengths within a memberand may assist the location of further testing which may be of a morecostly or damaging nature. The use of contours is not restricted to strengthassessment and they are commonly used for reinforcement corrosion andintegrity surveys.Concrete variability can also be usefully expressed as histograms, especiallywhere a large number of results are available, as when large membersare under test or where many similar members are being compared.Figure 1.8(a) shows a typical plot for well-constructed members using a uniformconcrete supply. The parameter measured should be plotted directly,and although the spread will reflect member type and distribution of testlocations as well as construction features, a single peak should emerge withan approximately normal distribution. A long ‘tail’ as in Figure 1.8(b) suggestspoor construction procedures, and twin peaks, Figure 1.8(c), indicatetwo distinct qualities of concrete supply. Simply studying such data as atable of values can be very difficult to interpret. Histograms provide a usefulvisual method of appraisal.

Planning and interpretation of in-situ testing 25(a)(b)No. of results No. of resultsStrength(c)StrengthNo. of resultsStrengthFigure 1.8 Typical histogram plots of in-situ test results: (a) uniform supply; (b) poorconstruction; (c) two sources.1.6.2.2 Numerical methodsCalculation of the coefficient of variation (equal to the standard deviation ×100/mean) of test results may provide valuable information about the constructionstandards employed. Table 1.7 contains typical values of coefficientsof variation relating to the principal test methods which may beexpected for a single site-made unit constructed from a number of batches.This information is based on the work of Tomsett (47), the authors (37),Concrete Society Report 11 (36) and other sources. Results for concretefrom one batch would be expected to be correspondingly lower, whereasif a number of different member types are involved, the values may beexpected to be higher. The values in Table 1.7 offer only a very approximateguide, but they should be sufficient to detect the presence of abnormalcircumstances.

26 Planning and interpretation of in-situ testingTable 1.7 Typical coefficients of variation (COV) of test results and maximumaccuracies of in-situ strength prediction for principal methodsTest methodTypical COV for individualmember of good qualityconstruction (in %)Best 95% confidencelimits on strengthestimates (in %)Cores – standard 10 ±10 (3 specimens)– small 15 ±15 (9 specimens)Pull-out 8 ±20 (4 tests)Internal fracture 16 ±28 (6 tests)Pull-off 8 ±15 (6 tests)Break-off 9 ±20 (5 tests)Windsor probe 4 ±20 (3 tests)Ultrasonic pulse velocity 2.5 ±20 (1 test)Rebound hammer 4 ±25 (12 tests)The coefficient of variation of concrete strength is not constant with varyingstrength for a given level of control because it is calculated using the averagestrength. Leshchinsky et al. (48) have also confirmed that the distributionof within-test coefficient of variation is asymmetrical. Hence general relationshipsbetween coefficient of variation of measured concrete strength andlevel of construction quality should not be used. Figure 1.9 illustrates typicalrelationships for ‘standard’ control cubes and in-situ strengths based ona variety of European and North American sources covering variations inthe supply as well as those within the structure. From these values, anticipatedstandard deviations can be deduced (for example at 30 N/mm 2 meanin-situ strength, a standard deviation of 02 ×30 = 6N/mm 2 is likely for normalquality construction) and hence confidence limits can be placed on theresults obtained. Values such as those of estimated in-situ standard deviationin Table 1.8 can be derived in this way, and in-situ strength accuracy predictionsmust make allowance for this as well as the accuracy of the test method.1.6.3 Calibration and application of test resultsThe likely accuracies of calibration between measured test results anddesired concrete properties are discussed in detail in the sections of thisbook dealing with each specific test. It is essential that the application ofthe results of in-situ testing takes account of such factors to determine theirsignificance.Particular attention must be paid to the differences between laboratoryconditions (for which calibration curves will normally be produced) and siteconditions. Differences in maturity and moisture conditions are especiallyrelevant in this respect. Concrete quality will vary throughout members andmay not necessarily be identical in composition or condition to laboratory

Planning and interpretation of in-situ testing 27Figure 1.9 Coefficient of variation of test results related to concrete strength.Table 1.8 Typical values of standard deviation of control cubes and in-situ concreteMaterial controland constructionAssumed std. dev. ofcontrol cube sN/mm 2 Estimated std. dev. of in-situconcrete s ′ N/mm 2 Very good 3.0 3.5Normal 5.0 6.0Low 7.0 8.5specimens. Also, the tests may not be so easy to perform or control due toadverse weather conditions, difficulties of access or lack of experience ofoperatives. Calibration of non-destructive and partially destructive strengthtests by means of cores from the in-situ concrete may often be possible andwill reduce some of these differences.Interpretation of strength results requires the use of statistical proceduressince it is not sufficient simply to average the values of the in-situ testresults and then compute the equivalent compressive strength by means of

28 Planning and interpretation of in-situ testingthe previously established relationship. Efforts have been made to establishlower confidence limits for the correlation relationship (1,43) based onstatistical tolerance factors, and the procedures outlined in the followingsections are based on this relatively simple approach. These methods failhowever to take account of measurement errors in the in-situ test result,as demonstrated by Stone et al. (49). These issues and a range of possibleanalysis procedures are considered fully in a report by ACI 228 (23) in2003 which incorporates work by Carino (50).Difficulties regarding a consensus-based statistical procedure may be abarrier to more widespread use of in-place testing for compliance purposes.Leshchinsky (51) reviewed provisions of national standards in 1992.Table 1.7 summarizes the maximum accuracies of in-situ strength predictionthat can realistically be hoped for under ideal conditions, with specificcorrelations for the particular concrete mix in each case. If any factor variesfrom this ideal, the accuracies of prediction will be reduced, although atpresent there is little available information to permit this to be quantified.Wherever possible, test methods should be used which directly measurethe required property, thereby reducing the uncertainties involved. Even inthese situations, however, care must be taken to make a realistic assessmentof the accuracy of the values emerging when formulating conclusions.1.6.3.1 Application to specificationsIt is essential that the concrete tested is representative of the materialunder examination and this will influence the number and location of tests(Section 1.4.4). Where some clearly defined property, such as cover orcement content, is being measured, it will generally be sufficient to comparemeasured results with the minimum specified value bearing in mindthe likely accuracy of the test. A small proportion of results marginallybelow the specified value may be acceptable, but the average for a numberof locations should exceed the minimum limit. If the test has a loworder of accuracy (e.g. cement content determination is unlikely to be betterthan ±40 kg/m 3 ) the area of doubt concerning marginal results maybe considerable. This is an unfortunate fact of life, although engineeringjudgements may perhaps be assisted by corroborative measurements of adifferent property.Strength is the most common criterion for the judgement of compliancewith specifications, and unfortunately the most difficult to resolve fromin-situ testing because of the basic differences between in-situ concrete andthe ‘standard’ test specimens upon which most specifications are based(Section 1.5.2). The number of in-situ test results will seldom be sufficientto permit a full statistical assessment of the appropriate confidence limits(usually 95%), hence it is better to compare mean in-situ strength estimateswith the expected mean ‘standard’ test specimen result. This requires an

Planning and interpretation of in-situ testing 29estimate to be made of the likely standard deviation of standard specimensunless the value of target mean strength for the mix is known. Themean ‘standard’ cube strength using British ‘limit state’ design proceduresis given byf mean = f cu + 164s (1.1)where f cu = characteristic strength of control cubess = standard deviation of control cubes.The accuracy of this calculation will increase with the number of resultsavailable; 50 readings could be regarded as the minimum necessary to obtaina sufficiently accurate estimate of the actual standard deviation. If sufficientinformation is not available the values given in Table 1.8 may be used as aguide.In theory it is possible to estimate the in-situ characteristic strength fcu′from the measured in-situ values of the mean fmean ′ and standard deviations ′ . The values of s ′ given in Table 1.8 may be used in the absence of morespecific data, but cannot be considered very reliable in view of withinmembervariations and the many variable constructional factors.In most cases the number of readings available from in-situ results will besignificantly less than 50, in which case the coefficient of 1.64 used in Equation(1.1) will increase. Equation (1.2) for the 95% confidence limit willthus apply, with k given by Table 1.9 according to the number of results n.f ′cu = f ′ mean − ks′ (1.2)This equation assumes a ‘normal’ distribution of concrete strength results(as in Equation (1.1)) but where concrete variability is high, as for poorTable 1.9 Suggested 95% confidence limit factorrelated to number of testsNumber of tests nConfidence factor k3 10.314 4.005 3.006 2.578 2.2310 2.0712 1.9815 1.9020 1.82 1.64

30 Planning and interpretation of in-situ testingquality control, a ‘log-normal’ distribution is considered to be more realistic.In this caselog f ′cu = mean value of log f ′ −k×standard deviation of log f ′ (1.3)where f ′ is an individual in-situ strength result.These relationships can conveniently be represented in graphical formas in Figure 1.10, which can be used to evaluate the characteristic valueas a proportion of the mean for a particular coefficient of variation ofresults. In this figure ‘normal’ and ‘log-normal’ distributions are compareddirectly for a coefficient of variation of 15% and the less demanding natureof the ‘log-normal’ distribution is demonstrated. This effect increases withincreasing coefficient of variation. The combined effects of variability ofresults and number of tests can also be clearly seen and the importanceof having at least four results is apparent. Bartlett and MacGregor haveapplied this approach to the evaluation of equivalent in-place characteristicstrength from core test data (52).Where some indications of the expected mean and material variabilityare available, a preliminary calculation can be made to obtain the desiredcharacteristic strength as a proportion of the mean, and hence the minimumFigure 1.10 Characteristic strength (95% confidence limit) as a function of coefficientof variation and number of tests (based on ref. 14).

Planning and interpretation of in-situ testing 31number of tests required to confirm the desired acceptability can be evaluated(14). Similar plots can be produced for different confidence limits anddistributions and it should be noted that less demanding 90% confidencelimits are adopted in some countries. The choice of distribution type andconfidence limits for use in particular circumstances is thus a matter ofjudgement.If an in-situ characteristic strength is estimated it can be compared withthe specified value, but this approach is not recommended unless numerousin-situ results are available.Whichever approach is adopted, the comparison between in-situ andstandard specimen strengths must allow for the type of differences indicatedin Table 1.6 and Figure 1.7 and this is illustrated in the examples ofAppendix A.1.6.3.2 Application to design calculationsMeasured in-situ values can be incorporated into calculations to assessstructural adequacy. Although this may occasionally relate to reinforcementquantities and location, or concrete properties such as permeability, in mostinstances it will be the concrete strength which is relevant. It is essentialthat the measured values relate to critical regions of the member underexamination and tests must be planned with this in mind (Section 1.4.4).Calculations are generally based on minimum likely, or characteristic,‘standard specimen’ values being modified by an appropriate factor of safetyto give a minimum in-situ design value. In-situ measurements will yielddirectly an in-situ strength of the concrete tested and this must be related toa similar specimen type and size to the ‘standard’ used in the calculations. Ifthis concrete is from a critical location, it could be argued that the minimummeasured value can be used directly as the design concrete strength with nofurther factors of safety applied. In practice, however, it is more appropriateto use the mean value from a number of test readings at critical locations,and to apply a factor of safety to this to account for test variability, possiblelack of concrete homogeneity and future deterioration. The accuracy ofstrength prediction will vary according to the method used, but a factor ofsafety of 1.2 is recommended by BS 6089 (13) for general use. Providing therecommendations of Section 1.4.4 have been followed when determiningthe number of readings, this value should be adequate. The application ofthis approach is illustrated in detail by the examples of Appendix A. Ifthere is particular doubt about the reliability of the test results, or if theconcrete tested is not from the critical location considered, then it may benecessary for the engineer to adopt a higher value for the factor of safetyguided by the information contained in Sections 1.5.1, 1.5.2 and 1.6.3.1.Alternatively, other features discussed in Section 1.5.2, including moisturecondition and age, may possibly be used to justify a lower value for the

32 Planning and interpretation of in-situ testingfactor of safety. The in-situ stress state and rate of loading may also betaken into account in critical circumstances.1.7 Test combinationsAll the test methods which are available for in-situ concrete assessmentsuffer from limitations, and reliability is often open to question. Combiningmethods may help to overcome some of these difficulties and is to berecommended, and some examples of typical combinations are outlinedbelow.1.7.1 Increasing confidence level of resultsConsiderably greater weight can be placed on results if corroborative conclusionscan be obtained from separate methods. Expense will usuallyrestrict large-scale duplication, but if different properties are measured,confidence will be much increased by the emergence of similar patterns ofresults. This will generally be restricted to tests which are quick, cheap andnon-destructive, such as combinations of surface hardness and ultrasonicpulse velocity measurements on recently cast concrete. In other circumstances,radiometry, pulse-echo, radar, thermography, or slower near-tosurfacestrength methods may be invaluable.If small volumes are involved and a specific property (e.g. strength) isrequired, it may sometimes be worthwhile to compare absolute estimatesachieved by different methods.1.7.2 Improvement of calibration accuracyIt may, in some cases, be possible to produce correlations of combinationsof measured values with desired properties to a greater accuracy than ispossible for either individual method. This has been most widely developedin relation to strength assessment using ultrasonic pulse velocities in conjunctionwith density (53) or rebound hammer readings (which are relatedto surface density).In the latter case, appropriate strength correlations must be producedfor both methods enabling multiple regression equations to be developedwith compressive strength as the dependent variable (54). This approach islikely to be of greatest value in quality control situations but is not widelyused (although in the authors’ view perhaps it should be!). A more complexversion of the technique has been encompassed in the SONREB method as adraft RILEM recommendation (55). This is based largely on work in EasternEurope and involves the principle that correlation graphs may be producedinvolving coefficients relating to various properties of the mix constituents.The increased accuracy is attributed to the opposing influences of some of

Planning and interpretation of in-situ testing 33the many variables for each of the methods, and strength predictions to anaccuracy of ±10% are claimed under ideal conditions. Recent work hasalso been reported from Argentina which includes application to lightweightconcrete (56), and from Poland using neural networks to help interpretresults (57).The more common in-situ tests may certainly be combined in a varietyof other ways but although valuable corroborative evidence may be gainedit is unlikely that the accuracy of absolute strength predictions will besignificantly improved.1.7.3 Use of one method as preliminary to anotherCombinations of methods are widely used in situations where one methodis regarded as a preliminary to the other. Common examples include thelocation of reinforcement prior to other forms of testing, and the use ofsimple non-destructive methods for comparative surveys to assist the mostworthwhile location of more expensive or damaging tests (see Figure 1.1).Tomsett has reported the successful combination of thermography andultrasonic pulse velocity measurements used in this way (47).Where monitoring strength development is important, maturity measurementsmay provide useful preliminary information, for confirmation byother strength assessment methods. A further case is the use of half-cellpotential measurements to indicate the level of possibility of corrosionoccurring, when subsequent resistivity measurements on zones shown to beat risk will identify the likelihood of corrosion actually occurring. A combinationof these methods, used correctly, can map areas for remedial workon car parks, bridge decks and other vulnerable concrete structures.1.7.4 Test calibrationThe most frequently occurring examples of calibration involving test combinationswill be the use of cores or destructive load tests to establish correlationsfor non-destructive or partially destructive methods which relatedirectly to the concrete under investigation. Coring or drilling may alsobe required to calibrate or validate the results of radar surveys, half-cellpotential and similar methods.1.7.5 Diagnosis of causes of deteriorationIt is most likely that more than one type of testing will be required to identifythe nature and cause of deterioration, and to assess future durability.Cover measurements will be included if reinforcement corrosion is involved,together with a possible range of chemical, petrographic and absorptiontests. Where deterioration is due to disruption of the concrete, a variety

Table 1.10 Relevant standardsBritish StandardsBS 1881: Testing concretePart 5Part 122Part 124Part 130Part 201Part 204Part 205Part 206Part 207Part 208BS 6089BS 8110BS 8204European StandardsBS EN 1542Methods of testing concrete for other than strengthMethod for the determination of water absorptionChemical analysis of hardened concreteTemperature matched curing of concretespecimensGuide to the use of NDT for hardened concreteThe use of electromagnetic covermetersRadiography of concreteDetermination of strain in concreteNear to surface test methods for strengthInitial surface absorption testAssessment of concrete strength in existingstructuresStructural use of concreteScreeds, bases and insitu flooringsProducts and systems for the protection andrepair of concrete structures. Test Methods:Measurement of bond strength by pull-offBS EN 12504 Testing Concrete in StructuresPart 1Cored specimens – Taking, examining and testing incompressionPart 2Non-destructive testing – determination ofrebound numberPart 3Determination of pull-out forcePart 4Determination of ultrasonic pulse velocityBS EN 13554 Non-destructive testing – Acoustic emission –General principlesBS EN 13894-4 Methods of test for screed materials –Determination of wear resistance – BCA∗ BS prEN 13791 ∗ in preparation Assessment of concrete compressive strength instructures or in structural elementsAmerican Standards ASTMC42Standard method of obtaining and testing drilledcores and sawed beams of concreteC85Cement content of hardened Portland cementconcreteC457Air void content in hardened concreteC597Standard test method for pulse velocity throughconcreteC779Abrasion resistance of horizontal concrete surfacesC803Penetration resistance of hardened concreteC805Rebound number of hardened concreteC823Examining and sampling of hardened concrete inconstructions

Planning and interpretation of in-situ testing 35Table 1.10 (Continued)C856C876C900C918C944C1040C1074C1084C1383C1583D4580D4748D4788D6087Petrographic examination of hardened concreteHalf-cell potential of uncoated reinforcing steel in concretePull-out strength of hardened concreteMeasurement of early-age compressive strength andprojecting later age strengthAbrasion resistance of concrete or mortar surfaces by therotating cutter methodDensity of unhardened and hardened concrete in place bynuclear methodsEstimating concrete strength by the maturity methodPortland cement content of hardened hydraulic concreteMeasuring the P-wave speed and thickness of concrete platesusing the IMPACT-Echo methodTensile strength of concrete surfaces and the bond strengthor tensile strength of concrete repair and overlay materials bydirect tension (Pull-off method)Measuring delaminations in concrete bridge decks by soundingDetermining the thickness of bound pavement layers usingshort-pulse radarDetecting delaminations in bridge decks using infraredthermographyEvaluating asphalt covered concrete bridge decks using groundpenetrating radarof tests on samples removed from the concrete is likely to be required,as discussed in Section 1.4.3. The most effective approach to diagnosis isto plot data, including visual appearance, in a spreadsheet array. Carefulexamination of such data can often reveal patterns of consistent behaviourbetween deteriorated areas, for example low cover, high half-cell potentialand high chloride levels, as compared to non-deteriorated areas.1.8 Documentation by standardsMany British, European and American Standards now available are applicableto in-situ concrete testing. A selection of those which are most relevant islisted in Table 1.10, and these are fully referenced in the appropriate parts ofthe text elsewhere in this book. Many other countries (e.g. Japan) have alsodeveloped relevant standards. Although requirements are generally similarthe amount of guidance provided, especially relating to applications andinterpretation of results, varies considerably.

Chapter 2Surface hardness methodsOne of many factors connected with the quality of concrete is its hardness.Efforts to measure the surface hardness of a mass of concrete were firstrecorded in the 1930s; tests were based on impacting the concrete surfacewith a specified mass activated by a standard amount of energy. Earlymethods involved measurements of the size of indentation caused by asteel ball either fixed to a pendulum or spring hammer, or fired from astandardized testing pistol. Later, however, the height of rebound of themass from the surface was measured. Although it is difficult to justify atheoretical relationship between the measured values from any of thesemethods and the strength of a concrete, their value lies in the ability toestablish empirical relationships between test results and quality of thesurface layer. Unfortunately these are subject to many specific restrictionsincluding concrete and member details, as well as equipment reliability andoperator technique.Indentation testing has received attention in Germany and in former statesof the USSR as well as the United Kingdom, but has never become verypopular. Pin penetration tests have, however, received attention in the USAand Japan (see Section 4.1.2). The rebound principle, on the other hand, ismore widely accepted: the most popular equipment, the Schmidt ReboundHammer, has been in use worldwide for many years. Recommendations forthe use of the rebound method are given in BS EN 12504-2 (58) and ASTMC805 (59).2.1 Rebound test equipment and operationThe Swiss engineer Ernst Schmidt first developed a practicable rebound testhammer in the late 1940s, and modern versions are based on this. Figure 2.1shows the basic features of a typical type N hammer, which weighs lessthan 2 kg, and has an impact energy of approximately 2.2 Nm.The spring-controlled hammer mass slides on a plunger within a tubularhousing. The plunger retracts against a spring when pressed against the

Surface hardness methods 37Figure 2.1 Typical rebound hammer.concrete surface and this spring is automatically released when fully tensioned,causing the hammer mass to impact against the concrete throughthe plunger. When the spring-controlled mass rebounds, it takes with it arider which slides along a scale and is visible through a small window in theside of the casing. The rider can be held in position on the scale by depressingthe locking button. The equipment is very simple to use (Figure 2.2),and may be operated either horizontally or vertically, either upwards ordownwards.The plunger is pressed strongly and steadily against the concrete at rightangles to its surface, until the spring-loaded mass is triggered from itslocked position. After the impact, the scale index is read while the hammeris still in the test position. Alternatively, the locking button may bepressed to enable the reading to be retained, or results can be recordedautomatically by an attached paper recorder. The scale reading is knownas the rebound number, and is an arbitrary measure since it depends onthe energy stored in the given spring and on the mass used. This versionof the equipment is most commonly used, and is most suitable forconcretes in the 20–60 N/mm 2 strength range. Electronic digital readingequipment with automatic data storage and processing facilities is alsowidely available (Figure 2.3). Other specialized versions are available forimpact sensitive zones and for mass concrete. For low strength concrete inthe 5–25 N/mm 2 strength range it is recommended that a pendulum type

Figure 2.2 Schmidt hammer in use.Figure 2.3 Digi-Schmidt (photograph by courtesy of Proceq).

Surface hardness methods 39Figure 2.4 Pendulum hammer.rebound hammer as shown in Figure 2.4 is used which has an enlargedhammer head (Type P).2.2 ProcedureThe reading is very sensitive to local variations in the concrete, especiallyto aggregate particles near to the surface. It is therefore necessary to takeseveral readings at each test location, and to find their average. Standardsvary in their precise requirements, but BS EN 12504-2 (58) recommends notless than nine readings taken over an area not exceeding 300 mm square,with the impact points no less than 25 mm from each other or from an edge.The use of a grid to locate these points reduces operator bias. Prior to testing,the equipment should be operated at least three times to ensure properfunctioning and checked on the steel reference anvil with adjustment asnecessary. Temperature should be in the range 10–35 C. Any measurementswhere the surface has crushed or broken through a near surface void shouldbe discounted, whilst if more than 20% of results are more than 6 units fromthe median the whole set should be discarded. ASTM C805 (59) requiresten readings to be taken. The surface must be smooth, clean and dry, andshould preferably be formed, but if trowelled surfaces are unavoidable theyshould be rubbed smooth with the Carborundum stone usually provided

40 Surface hardness methodswith the equipment. Loose material can be ground off, but areas whichare rough from poor compaction, grout loss, spalling or tooling must beavoided since the results will be unreliable.2.3 Theory, calibration and interpretationThe test is based on the principle that the rebound of an elastic mass dependson the hardness of the surface upon which it impinges, and in this casewill provide information about a surface layer of the concrete defined as nomore than 30 mm deep. The results give a measure of the relative hardnessof this zone, and this cannot be directly related to any other property of theconcrete. Energy is lost on impact due to localized crushing of the concreteand internal friction within the body of the concrete, and it is the latter,which is a function of the elastic properties of the concrete constituents, thatmakes theoretical evaluation of test results extremely difficult (60). Manyfactors influence results but all must be considered if rebound number is tobe empirically related to strength.2.3.1 Factors influencing test resultsResults are significantly influenced by all of the following factors:1. Mix characteristics(i) Cement type(ii) Cement content(iii) Coarse aggregate type2. Member characteristics(i) Mass(ii) Compaction(iii) Surface type(iv) Age, rate of hardening and curing type(v) Surface carbonation(vi) Moisture condition(vii) Stress state and temperature.Since each of these factors may affect the readings obtained, any attemptsto compare or estimate concrete strength will be valid only if they are allstandardized for the concrete under test and for the correlation specimens.These influences have different magnitudes. Hammer orientation will also

Surface hardness methods 41influence measured values (Section 2.3.2) although correction factors canbe used to allow for this effect.2.3.1.1 Mix characteristicsThe three mix characteristics listed above are now examined in more detail.(i) Cement type. Variations in fineness of Portland cement are unlikely tobe significant – their influence on strength correlation is less than 10%.Super-sulfated cement, however, can be expected to yield strengths 50%lower than suggested by a Portland cement correlation, whereas highalumina cement concrete may be up to 100% stronger.(ii) Cement content. Changes in cement content do not result in correspondingchanges in surface hardness. The combined influence of strength,workability and aggregate/cement proportions leads to a reduction ofhardness relative to strength as the cement content increases (61). Theerror in estimated strength, however, is unlikely to exceed 10% fromthis cause for most mixes.(iii) Coarse aggregate. The influence of aggregate type and proportionscan be considerable, since strength is governed by both paste andaggregate characteristics. The rebound number will be influenced moreby the hardened paste. For example, crushed limestone may yield arebound number significantly lower than for a gravel concrete of similarstrength which may typically be equivalent to a strength differenceof 6–7 N/mm 2 . A particular aggregate type may also yield differentrebound number/strength correlations depending on the source andnature, and Figure 2.5 compares typical curves for hard and soft gravels.These have measured hardness expressed in terms of the Mohs’number (see Section 4.1.1.2) of 7 and 3 respectively.Lightweight aggregates may be expected to yield results significantlydifferent from those for concrete made with dense aggregates, and considerablevariations have also been found between types of lightweightaggregates (34). Correlations can, however, be obtained for specificlightweight aggregates, although the amount of natural sand used will affectresults.The extent of these differences is illustrated by Figure 2.6 which comparesstrength correlations obtained by varying age of otherwise ‘identical’ drycuredlaboratory specimens containing different lightweight coarse aggregates.Mix 5 included lightweight fine materials whilst all others containednatural sand and the effects of this can be seen by comparing results formixes 4 and 5 which are otherwise similar.

42 Surface hardness methodsFigure 2.5 Comparison of hard and soft gravels – vertical hammer.Figure 2.6 Comparison of lightweight aggregates (based on ref. 34).2.3.1.2 Member characteristicsThe member characteristics listed above are also to be discussed in detail.(i) Mass. The effective mass of the concrete specimen or member undertest must be sufficiently large to prevent vibration or movement causedby the hammer impact. Any such movement will result in a reducedrebound number. For some structural members the slenderness ormass may be such that this criterion is not fully satisfied, and in such

Surface hardness methods 43cases absolute strength prediction may be difficult. BS EN 12504-2 (58) requires that a member is at least 100 mm thick and fixedwithin a structure. Strength comparisons between or within individualmembers must also take account of this factor. The mass ofcorrelation specimens may be effectively increased by clamping themfirmly in a heavy testing machine, and this is discussed more fully inSection 2.3.2.(ii) Compaction. Since a smooth, well-compacted surface is required forthe test, variations of strength due to internal compaction differencescannot be detected with any reliability. All calibrations must assumefull compaction.(iii) Surface type. Hardness methods are not suitable for open-texturedor exposed aggregate surfaces. Trowelled or floated surfaces may beharder than moulded surfaces, and will certainly be more irregular.Although they may be smoothed by grinding, this is laborious and itis best to avoid trowelled surfaces in view of the likely overestimationof strength from hardness readings. The absorption and smoothnessof the mould surface will also have a considerable effect. Calibrationspecimens will normally be cast in steel moulds which are smoothand non-absorbent, but more absorbent shuttering may well producea harder surface, and hence internal strength may be overestimated.Although moulded surfaces are preferred for on-site testing, care mustbe taken to ensure that strength correlations are based on similarsurfaces, since considerable errors can result from this cause.(iv) Age, rate of hardening and curing type. The relationship betweenhardness and strength has been shown to vary as a function oftime (61), and variations in initial rate of hardening, subsequent curingand exposure conditions will further influence this relationship.Where heat treatment or some other form of accelerated curing hasbeen used, a specific calibration will be necessary. The moisture statemay also be influenced by the method of curing. For practical purposesthe influence of time may be regarded as unimportant up tothe age of three months, but for older concretes it may be possibleto develop reduction factors which take account of the concrete’shistory.(v) Surface carbonation. Concrete exposed to the atmosphere will normallyform a hard carbonated skin, whose thickness will depend uponthe exposure conditions and age. It may exceed 20 mm for old concretealthough it is unlikely to be significant at ages of less than three months.The depth of carbonation can easily be determined as described inChapter 9. Examination of gravel concrete specimens which had beenexposed to an outdoor ‘city-centre’ atmosphere for six months showeda carbonated depth of only 4 mm. This was not sufficient to influencethe rebound number/strength relationship in comparison with similar

44 Surface hardness methodsspecimens stored in a laboratory atmosphere although for these specimensno measurable skin was detected. In extreme cases, however, it isknown that the overestimate of strength from this cause may be up to50%, and is thus of great importance. When significant carbonationis known to exist, the surface layer ceases to be representative of theconcrete within an element.(vi) Moisture condition. The hardness of a concrete surface is lower whenwet than when dry, and the rebound/strength relationship will beinfluenced accordingly. This effect is illustrated by Figure 2.7, based onearly work by the US army (62), from which it will be seen that a wetsurface test may lead to an underestimate of strength of up to 20%.Field tests and strength calibrations should normally be based on drysurface conditions, but the effect of internal moisture on the strengthof control specimens must not be overlooked. This is considered inmore detail in Section 2.3.2.(vii) Stress state and temperature. Both these factors may influence hardnessreadings, although in normal practical situations this is likely to besmall in comparison with the many other variables. Particular attentionshould, however, be paid to the functioning of the test hammer if it isto be used under extremes of temperature, noting the limits of 10 to35 C in BS EN 12504-2 (58).Figure 2.7 Influence of surface moisture condition – horizontal hammer (based onref. 62).

2.3.2 CalibrationSurface hardness methods 45Clearly, the influences of the variables described above are so great that itis very unlikely that a general calibration curve relating rebound numberto strength, as provided by the equipment manufacturers, will be of anypractical value. The same applies to the use of computer data processingto give strength predictions based on results from the electronic reboundhammer shown in Figure 2.3 unless the conversions are based on casespecificdata. Strength calibration must be based on the particular mixunder investigation, and the mould surface, curing and age of laboratoryspecimens should correspond as closely as possible to the in-place concrete.It is essential that correct functioning of the rebound hammer is checkedregularly using a standard steel anvil of known mass. This is necessarybecause wear may change the spring and internal friction characteristicsof the equipment. Calibrations prepared for one hammer will also notnecessarily apply to another. It is probable that very few rebound hammersused for in-situ testing are in fact regularly checked against a standard anvil,and the reliability of results may suffer as a consequence.The importance of specimen mass has been discussed above; it is essentialthat test specimens are either securely clamped in a heavy testing machine orsupported upon an even solid floor. Cubes or cylinders of at least 150 mmshould be used, and a minimum restraining load of 15% of the specimenstrength has been suggested for cylinders (63), and not less than 7 N/mm 2is recommended for cubes tested with a type N hammer. Some typicalrelationships between rebound number and restraining load are given inFigure 2.8, which shows that once a sufficient load has been reached therebound number remains reasonably constant.It is well established that the crushing strength of a cube tested wet islikely to be about 10% lower than the strength of a corresponding cubetested dry. Since rebound measurements should be taken on a dry surface, itis recommended that wet cured cubes be dried in the laboratory atmospherefor 24 hours before test, and it is therefore to be expected that they willyield higher strengths than if tested wet in the standard manner. Dependingupon the purpose of the test programme it may be necessary to confirmthis relationship, and the relative moisture conditions of the correlationspecimens and in-place concrete must also be considered when interpretingthe field results. The use of cores cut following in-situ hardness tests mayhelp to overcome these difficulties in developing calibrations.If cubes are used, readings should be taken on at least two vertical faces ofthe specimen as cast, as described in Section 2.2, and the hammer orientationmust be similar to that to be used for the in-place tests. The influenceof gravity on the mass will depend on whether it is moving vertically upor down, horizontally or on an inclined plane. The effect on the reboundnumber will be considerable, although the relative values suggested by the

46 Surface hardness methodsFigure 2.8 Effect of restraining load on calibration specimen (incorporating data fromref. 63).manufacturer are likely to be reliable in this instance because this is purelya function of the equipment.2.3.3 InterpretationThe interpretation of surface hardness readings relies upon a knowledgeof the extent to which the factors described in Section 2.3.1 have beenstandardized between readings being compared. This applies whether theresults are being used to assess relative quality or to estimate strength.It will be apparent from Figure 2.9, which shows a typical strength calibrationchart produced under ‘ideal’ laboratory conditions, that the scatterof results is considerable, and the strength range corresponding to agiven rebound number is about ±15% even for ‘identical’ concrete. In

Surface hardness methods 47Figure 2.9 Typical rebound number/compressive strength calibration chart.a practical situation it is very unlikely that a strength prediction can bemade to an accuracy better than ±25% (63). The scatter also suggests thateven if a strength prediction is not required, a considerable variation ofrebound number can be expected for ‘identical’ concrete, and acceptablelimits must be determined in conjunction with some other form of testing.It is suggested (13) that where the total number of readings n taken ata location is not less than ten, the accuracy of the mean rebound numberis likely to be within ±15/ √ n% with 95% confidence. The results mayusefully be presented in graphical form as described in Section 1.6.2.1, andcalculation of the coefficient of variation may yield an indication of concreteuniformity, as described in Section 1.6.2.2, when sufficient results areavailable.The test location within the member is important when interpretingresults (Chapter 1) but it should be noted that the test yields informationabout a thin surface layer only. Results are unrelated to the properties ofthe interior, and furthermore are not regarded as reliable on concrete morethan three months old unless special steps are taken to allow for age effectsand surface carbonation, as described above.Although it is generally the relationship between rebound number andcompressive strength that is of interest, similar relationships can be establishedwith flexural strength although with an even greater scatter. It appearsthat no general relationship between rebound number and elastic modulus

48 Surface hardness methodsexists although it may be possible to produce such a calibration for a specificmix.2.3.4 Applications and limitationsThe useful applications of surface hardness measurements can be dividedinto four categories:(i) Checking the uniformity of concrete quality(ii) Comparing a given concrete with a specified requirement(iii) Approximate estimation of strength(iv) Abrasion resistance classification.Whatever the application, it is essential that the factors influencing testresults are standardized or allowed for, and it should be remembered thatresults relate only to the surface zone of the concrete under test. A furtheroverriding limitation relates to testing at early ages or low strengths, becausethe rebound numbers may be too low for accurate reading and the impactmay also cause damage to the surface (Figure 2.10). It is therefore notrecommended that the method is used for concrete which has a cube strengthof less than 10 N/mm 2 or which is less than 7 days old, unless of highstrength.Figure 2.10 Surface damage on green concrete.

Surface hardness methods 49(i) Concrete uniformity checking. The most important and reliable applicationsof surface hardness testing are where it is not necessary to attempt toconvert the results to some other property of the concrete. It is claimed (61)that surface hardness measurements give more consistently reproducibleresults than any other method of testing concrete. Although they do notdetect poor internal compaction, results are sensitive to variations of qualitybetween batches, or due to inadequate mixing or segregation. The valueas a control test is further enhanced by the ability to monitor the concretein members cheaply and more comprehensively than is possible by a smallnumber of control specimens. For such comparisons to be valid for a givenmix it is only necessary to standardize age, maturity, surface moisture conditions(which should preferably be dry), and location on the structure orunit.This approach has been extensively used to control uniformity of precastconcrete units, and may also prove valuable for the comparison of suspectin-situ elements with similar elements which are known to be sound. Afurther valuable use for such comparative tests may be to establish therepresentation of other forms of testing, possibly destructive, which mayyield more specific but localized indications of quality.(ii) Comparison with a specific requirement. This application is also popularin the precasting industry, where a minimum hardness reading may becalibrated against some specific requirement of the concrete. For instance,the readiness of precast units for transport may be checked, with calibrationbased on proof load tests. The approach may also be used as an acceptancecriterion, in relation to the removal of temporary supports from structuralmembers, or commencement of stress transfer in prestressed concreteconstruction.(iii) Approximate strength estimation. This represents the least reliableapplication and (unfortunately, since a strength estimate is frequentlyrequired by engineers) is where misuse is most common. The accuracydepends entirely upon the elimination of influences which are not takeninto account in the calibration. For laboratory specimens cast, cured andtested under conditions identical to those used for calibration, it is unlikelythat a strength estimate better than ±15% can be achieved for concreteup to three months old. Although it may be possible to correct for one ortwo variables which may not be identical on site, the accuracy of absolutestrength prediction will decline as a consequence and is unlikely to be betterthan ±25%. The use of the rebound hammer for strength estimationof in-place concrete must never be attempted unless specific calibrationcharts are available, and even then, the use of this method alone is notrecommended, although the value of results may be improved if used inconjunction with other forms of testing as described in Chapter 1.

50 Surface hardness methods(iv) Abrasion resistance classification. Abrasion resistance is generallyaffected by the same influences as surface hardness, and Chaplin (64) hassuggested that the rebound hammer may be used to classify this property.This is discussed in Chapter 7. It is also reasonable to suppose thatother durability characteristics that are related to a dense, well cured, outersurface zone may similarly be classified.

Chapter 3Ultrasonic pulse velocitymethodsThe first reports of the measurement of the velocity of mechanically generatedpulses through concrete appeared in the USA in the mid-1940s. Itwas found that the velocity depended primarily upon the elastic propertiesof the material and was almost independent of geometry. The potentialvalue of this approach was apparent, but measurement problems wereconsiderable, and led to the development in France, a few years later,of repetitive mechanical pulse equipment. At about the same time, workwas undertaken in Canada and the United Kingdom using electro-acoustictransducers, which were found to offer greater control on the type andfrequency of pulses generated. This form of testing has been developedinto the modern ultrasonic method, employing pulses in the frequencyrange of 20–150 kHz, generated and recorded by electronic circuits. Ultrasonictesting of metals commonly uses a reflective pulse technique withmuch higher frequencies, but this cannot readily be applied to concretebecause of the high scattering which occurs at matrix/aggregate interfacesand microcracks. Concrete testing is thus at present based largely on pulsevelocity measurements using through-transmission techniques. The methodhas become widely accepted around the world, and commercially producedrobust lightweight equipment suitable for site as well as laboratory use isreadily available.Nogueira and Willam (65) found that UPV methods, where the amplitudeof the signal was studied, could be used to estimate microcrack growth inconcrete and hence to study mechanical damage, whilst Pavlakovic et al.(66) have used a guided wave technique to study damage in post-tensionedtendons in bridges. Krause et al. (67) have studied ultrasonic imaging withan array system to examine defects behind dense steel reinforcement, includingcover to pipe ducts and ungrouted tendon ducts. Koehler (68) has furtherexamined the use of specialized Synthetic Aperture Focussing Techniques(SAFT) to provide 3D visualization of defects in concrete structures, suchas gravel pockets, and to locate tendon ducts. Krause and Wiggenhauser(69) also successfully used ultrasonic 2D and 3D methods to establish theposition of tendon ducts in a bridge deck, and Popovics (70) has recently

52 Ultrasonic pulse velocity methodsreviewed some of these techniques together with tomography. Andrews(71) has suggested that there is much scope for new applications with thedevelopment of improved fidelity transducers and computer interpretation.Study of pulse attenuation characteristics has been shown by the authors toprovide useful data relating to deterioration of concrete due to alkali–silicareaction (72) although there are practical problems of achieving consistentcoupling on site. Hillger (73) and Kroggel (74) have both described thedevelopment of pulse-echo techniques to permit detection of defects andcracks from tests on one surface as well as the use of a vacuum couplingsystem, and the application of signal processing techniques to yield informationabout internal defects and features is the subject of current researchas noted above. Another interesting development, described by Sack andOlson (75), involves the use of rolling transmitter and receiver scanners,which do not need any coupling medium, with a computer data acquisitionsystem that permits straight line scans of up to 9 m to be made within atimescale of less than 30 seconds.Although it is likely that many of these developments will expand intocommercial use in the future, the remainder of this chapter will concentrateupon conventional pulse velocity techniques.If the method is properly used by an experienced operator, a considerableamount of information about the interior of a concrete member can beobtained. However, since the range of pulse velocities relating to practicalconcrete qualities is relatively small (3.5–4.8 km/s), great care is necessary,especially for site usage. Furthermore, since it is the elastic properties of theconcrete which affect pulse velocity, it is often necessary to consider in detailthe relationship between elastic modulus and strength when interpretingresults. Recommendations for the use of this method are given in BS EN12504-4 (76) and also in ASTM C597 (77).3.1 Theory of pulse propagation through concreteThree types of waves are generated by an impulse applied to a solid mass.Surface waves having an elliptical particle displacement are the slowest,whereas shear or transverse waves with particle displacement at right anglesto the direction of travel are faster. Longitudinal waves with particle displacementin the direction of travel (sometimes known as compressionwaves) are the most important since these are the fastest and generallyprovide more useful information. Electro-acoustical transducers primarilyproduce waves of this type; other types generally cause little interferencebecause of their lower speed.The wave velocity depends upon the elastic properties and mass of themedium, and hence if the mass and velocity of wave propagation are known

Ultrasonic pulse velocity methods 53it is possible to assess the elastic properties. For an infinite, homogeneous,isotropic elastic medium, the compression wave velocity is given by:√KEV = dkm/s (3.1)where E d = Dynamic modulus of elasticity N/mm 2 = density kg/m 3 K =1−v1+v1−2vand v = dynamic Poisson’s ratioIn this expression the value of K is relatively insensitive to variations of thedynamic Poisson’s ratio , and hence, provided that a reasonable estimateof this value and the density can be made, it is possible to compute E d usinga measured value of wave velocity V . Since and will vary little for mixeswith natural aggregates, the relationship between velocity and dynamicelastic modulus may be expected to be reasonably consistent despite the factthat concrete is not necessarily the ‘ideal’ medium to which the mathematicalrelationship applies, as indicated in Section 3.3.3.2 Pulse velocity equipment and use3.2.1 EquipmentThe test equipment must provide a means of generating a pulse, transmittingthis to the concrete, receiving and amplifying the pulse and measuring anddisplaying the time taken. The basic circuitry requirements are shown inFigure 3.1.Repetitive voltage pulses are generated electronically and transformedinto wave bursts of mechanical energy by the transmitting transducer, whichmust be coupled to the concrete surface through a suitable medium (seeSection 3.2.2). A similar receiving transducer is also coupled to the concreteat a known distance from the transmitter, and the mechanical energyconverted back to electrical pulses of the same frequency. The electronictiming device measures the interval between the onset and reception of thepulse and this is displayed either on an oscilloscope or as a digital readout.The equipment must be able to measure the transit time to an accuracy of±1%. To ensure a sharp pulse onset, the electronic pulse to the transmittermust have a rise time of less than one-quarter of its natural period. Therepetition frequency of the pulse must be low enough to avoid interferencebetween consecutive pulses, and the performance must be maintained overa reasonable range of climatic and operating conditions.Transducers with natural frequencies between 20 and 150 kHz are themost suitable for use with concrete, and these may be of any type, although

54 Ultrasonic pulse velocity methodsFigure 3.1 Typical UPV testing equipment.the piezo-electric crystal is most popular. Time measurement is based ondetection of the compressive wave pulse, the first part of which may haveonly a very small amplitude. If an oscilloscope is used, the received pulse isamplified and the onset taken as the tangent point between the signal curveand the horizontal time-base line, whereas for digital instruments the pulseis amplified and shaped to trigger the timer from a point on the leadingedge of a pulse.A number of commercially produced instruments have become availablein recent years which satisfy these requirements. The most popular of theseare the V-meter produced in the USA (78) and the PUNDIT (Portable UltrasonicNon-destructive Digital Indicating Tester) (79) produced in the UnitedKingdom. These have many similarities: both measure 180×110×160 mm,weigh 3 kg and have a digital display. Nickel–cadmium rechargeable batteriesallow over 9 hours continuous operation. Both also incorporate constantcurrent charges to enable recharging from an AC mains supply, and mayalso be operated directly from the mains through a mains supply unit. Foruse in the laboratory an analogue unit can be added and this in turn can beconnected to a recorder for continual experimental monitoring. The PUN-DIT PLUS is also more recently available, featuring a large LCD display,NiMH batteries and a battery life of up to 8 hours, depending on use of

Ultrasonic pulse velocity methods 55the backlight for the display. It also has a memory store for data, an RS232computer connection and special transducers with attached microswitchesto simplify the recording of data. Other available equipment incorporatesan oscilloscope and permits amplitude monitoring and study of attenuation.Figure 3.2 shows the PUNDIT PLUS set up in the laboratory to measurethe properties of a concrete sample. The probes are normally first calibratedusing a special steel reference bar, which has known characteristicsand is used to set the calibration of the instrument by means of a variabledelay control unit each time it is used but PUNDIT PLUS is factorycalibrated and simply requires zeroing. The display gives a direct transittime reading in microseconds. A wide range of transducers between 24 and200 kHz are available, although the 54 and 82 kHz versions will normallybe used for site or laboratory testing of concrete. Waterproof and evendeep-sea versions of these transducers are available (Figure 3.3). An alternativeform is the exponential probe transducer which makes a point contact,and offers operating advantages over flat transducers on rough or curvedsurfaces (see Section 3.2.2.2). The equipment is generally robust and providedwith a carrying case for site use. Signal amplifiers are also availablewhere long path lengths are involved on site, and the range of acceptableambient temperatures of 0–45 C should cover most practical situations.Equipment is also commercially available and is in use in the UK whichutilizes an array of transducers applied to one surface and operates usingshear waves.3.2.2 UseOperation is relatively straightforward but requires great care if reliableresults are to be obtained. One essential is good acoustical coupling betweenFigure 3.2 PUNDIT PLUS in laboratory (photograph courtesy of CNS Farnell Ltd).

56 Ultrasonic pulse velocity methodsFigure 3.3 PUNDIT in use underwater, with waterproof probes (photograph courtesyof CNS Farnell Ltd).the concrete surface and the face of the transducer, and this is provided by amedium such as petroleum jelly, liquid soap or grease. Air pockets must beeliminated, and it is important that only a thin separating layer exists – anysurplus must be squeezed out. A light medium, such as petroleum jelly orliquid soap, has been found to be the best for smooth surfaces, but a thickergrease is recommended for rougher surfaces which have not been castagainst smooth shutters. If the surface is very rough or uneven, grinding orpreparation with plaster of Paris or quick-setting mortar may be necessaryto provide a smooth surface for transducer application. It is also importantthat readings are repeated by complete removal and re-application oftransducers to obtain a minimum value for the transit time. Although the

Ultrasonic pulse velocity methods 57measuring equipment is claimed to be accurate to ±01 microseconds, if atransit time accuracy of ±1% is to be achieved it may typically be necessaryto obtain a reading to ±07 s over a 300 mm path length. This can only beachieved with careful attention to measurement technique, and any dubiousreadings should be repeated as necessary, with special attention to theelimination of any other source of vibration, however slight, during the test.The authors’ site experience has confirmed the necessity for an adequatelyprepared surface, with misleading readings obtained if this is not done.The path length must also be measured to an accuracy of ±1%. Thisshould present little difficulty with paths over about 500 mm, but for shorterpaths it is recommended that calipers be used. The nominal member dimensionsshown on drawings will seldom be adequate.3.2.2.1 Transducer arrangementThere are three basic ways in which the transducers may be arranged, asshown in Figure 3.4. These are:(i) Opposite faces (direct transmission)(ii) Adjacent faces (semi-direct transmission)(iii) Same face (indirect transmission).Figure 3.4 Types of reading: (a) Direct; (b) semi-direct; (c) indirect.

58 Ultrasonic pulse velocity methodsSince the maximum pulse energy is transmitted at right angles to the faceof the transmitter, the direct method is the most reliable from the point ofview of transit time measurement. Also, the path is clearly defined and canbe measured accurately, and this approach should be used wherever possiblefor assessing concrete quality. The semi-direct method can sometimesbe used satisfactorily if the angle between the transducers is not too great,and if the path length is not too large. The sensitivity will be smaller, andif these requirements are not met it is possible that no clear signal will bereceived because of attenuation of the transmitted pulse. The path lengthis also less clearly defined due to the finite transducer size, but it is generallyregarded as adequate to take this from centre to centre of transducerfaces.The indirect method is definitely the least satisfactory, since the receivedsignal amplitude may be less than 3% of that for a comparable directtransmission. The received signal is dependent upon scattering of the pulseby discontinuities and is thus highly subject to errors. The pulse velocitywill be predominantly influenced by the surface zone concrete, which maynot be representative of the body, and the exact path length is uncertain.A special procedure is necessary to account for this lack of precision ofpath length, requiring a series of readings with the transmitter fixed andthe receiver located at a series of fixed incremental points along a chosenradial line (Figure 3.5). The results are plotted (Figure 3.6) and the meanpulse velocity is given by the slope of the best straight line. If there isa discontinuity in this plot it is likely that either surface cracking or aninferior surface layer is present (see Section 3.4). Unless measurements arebeing taken to detect such features, this method should be avoided if at allpossible and only used where just one surface is available.Figure 3.5 Indirect reading–transducer arrangement.

Ultrasonic pulse velocity methods 59Figure 3.6 Indirect reading results plot.3.2.2.2 Transducer selectionThe most commonly used transducers have a natural frequency of 54 kHz.They have a flat surface of 50 mm diameter, and thus good contact must beensured over a considerable area. However, the use of a probe transducermaking only point contact and normally requiring no surface treatment orcouplant offers advantages. Time savings may be considerable and pathlength accuracy for indirect readings may be increased, but this type oftransducer is unfortunately more sensitive to operator pressure. Receivershave been found to operate satisfactorily in the field, but the signal poweravailable from a transmitting transducer of this type is so low that its useis not normally practicable for site testing. The exponential probe receiver,which has a tip diameter of 6 mm, may also be useful on very rough surfaceswhere preparatory work might otherwise be necessary.The only important factors which are likely to require the selection ofan alternative transducer frequency relate to the dimensions of the memberunder test. Difficulties arise with small members as the medium under testcannot be considered as effectively infinite. This will occur when the pathwidth is less than the wavelength . Since = pulse velocity/frequency ofvibration, it follows that the least lateral dimensions given in Table 3.1

60 Ultrasonic pulse velocity methodsTable 3.1 Minimum lateral path and maximum aggregate dimensionsTransducerfrequency (kHz)Minimum lateral path dimension or maximum aggregate size (mm)V c = 38km/sV c = 46km/s54 70 8582 46 56150 25 30should be satisfied. Aggregate size should similarly be less than to avoidreduction of wave energy and possible loss of signal at the receiver, althoughthis will not normally be a problem. Although use of higher frequenciesmay reduce the maximum acceptable path length (10 m for 54 kHz to 3 mfor 82 kHz), due to the lower energy output associated with the higherfrequency, this problem can easily be overcome by the use of an inexpensivesignal amplifier.3.2.2.3 Equipment calibrationThe time delay adjustment must be used to set the zero reading for theequipment before use, and this should also be regularly checked duringand at the end of each period of use. Individual transducer and connectinglead characteristics will affect this adjustment, which is performed with theaid of a calibrated steel reference bar that has a transit time of around25 s. A reading through this bar is taken in the normal way ensuringthat only a very thin layer of couplant separates the bar and transducers.It is also recommended that the accuracy of transit time measurement ofthe equipment is checked by measurement on a second reference specimen,preferably with a transit time of around 100 s.3.3 Test calibration and interpretation of resultsThe basic problem is that the material under test consists of two separateconstituents, matrix and aggregate, which have different elastic andstrength properties. The relationship between pulse velocity and dynamicelastic modulus of the composite material measured by resonance tests onprisms is fairly reliable, as shown in Figure 3.7. Although this relationshipis influenced by the value of dynamic Poisson’s ratio, for most practicalconcretes made with natural aggregates the estimate of modulus of elasticityshould be accurate within 10%.

Ultrasonic pulse velocity methods 61Figure 3.7 Pulse velocity vs. dynamic elastic modulus.3.3.1 Strength calibrationThe relationship between elastic modulus and strength of the compositematerial cannot be defined simply by consideration of the properties andproportions of individual constituents. This is because of the influence ofaggregate particle shape, efficiency of the aggregate/matrix interface andvariability of particle distribution, coupled with changes of matrix propertieswith age. Although some attempts have been made to represent this theoretically,the complexity of the interrelationships is such that experimentalcalibration for elastic modulus and pulse velocity/strength relationships isnormally necessary. Aggregate may vary in type, shape, size and quantity,and the cement type, sand type, water/cement ratio and maturity are allimportant factors which influence the matrix properties and hence strengthcorrelations. A pulse velocity/strength curve obtained with maturity as theonly variable, for example, will differ from that obtained by varying thewater/cement ratio for otherwise similar mixes, but testing at comparablematurities (Figure 3.8). Similarly, separate correlations will exist for varyingaggregate types and proportions as well as for cement characteristics. Thiswill include lightweight concretes (34) and special cements (80).Strength calibration for a particular mix should normally be undertakenin the laboratory with due attention to the factors listed above. Pulse velocityreadings are taken between both pairs of opposite cast faces of cubes of

62 Ultrasonic pulse velocity methods48004600UPV Development for Different w/c’sUPV of concrete (m/s)44004200400038003600340032000 10 20 30 40Age (Days)w/c 0.3w/c 0.4w/c 0.5w/c 0.6w/c 0.7Figure 3.8 Effect of water/cement ratio for concretes at different ages (basedon ref. 81).known moisture condition, which are then crushed in the usual way. Ideally,at least ten sets of three specimens should be used, covering as wide arange of strengths as possible, with the results of each group averaged. Aminimum of three pulse velocity measurements should be taken for eachcube, and each individual reading should be within 5% of the mean for thatcube. Where this is not possible, cores cut from the hardened concrete maysometimes be used for calibration, although there is a danger that drillingdamage may affect pulse velocity readings. Wherever possible, readingsshould be taken at core locations prior to cutting. Provided that cores aregreater than 100 mm in diameter, and that the ends are suitably preparedprior to test, it should be possible to obtain a good calibration, althoughthis will usually cover only a restricted strength range. If it is necessaryto use smaller diameter cores, high frequency transducers (Section 3.2.2.2)may have to be used, and the accuracy of crushing strength will also bereduced (see Section 5.3).Lin et al. (81) studied the prediction of pulse velocity in concrete basedon the mix proportions as shown in Figure 3.8, with very good results inthe laboratory. They suggested that the method could be used to accuratelypredict strength in structures. Their method used a mathematical modelbased on the UPV behaviour of the individual components of the concreteweighted for the amount of each in the mixture.

Ultrasonic pulse velocity methods 63Although the precise relationship is affected by many variables, the curvemay be expected to be of the general formf c = Ae BVwhere f c = equivalent cube strengthe = base of natural logarithmsV = pulse velocityand A and B are constants.Hence a plot of log cube strength against pulse velocity is linear for aparticular concrete. It is therefore possible to use a curve derived fromreference specimens to extrapolate from a limited range of results fromcores. Concrete made with lightweight aggregates is likely to give a lowerpulse velocity at a given strength level. This is demonstrated in Figure 3.9,in which the effects of lightweight fines (All-Lytag) can also be seen. Itshould also be noted that for most lightweight aggregates there is likely tobe reduced variability of measured values (34).3.3.2 Practical factors influencing measured resultsThere are many factors relating to measurements made on in-situ concretewhich may further influence results.3.3.2.1 TemperatureThe operating temperature ranges to be expected in temperate climatesare unlikely to have an important influence on pulse velocities, but ifCube Compressive Strength (N/mm 2 )604020LecaAll-LytagLytag Pellite Gravel03.00 3.40 3.80 4.20 4.60Ultrasonic Pulse Velocity (km/s)Figure 3.9 Comparison of lightweight and gravel aggregates (based on ref. 38, withpermission of Elsevier).

64 Ultrasonic pulse velocity methodsFigure 3.10 Effect of temperature (based on ref. 82).extreme temperatures are encountered, their effect can be estimated fromFigure 3.10. These factors are based on work by Jones and Facaoaru (82)and reflect possible internal microcracking at high temperatures and theeffects of water freezing within the concrete at very low temperatures. Similarvalues are proposed by BS EN 12504-4 (76).3.3.2.2 Stress historyIt has been generally accepted that the pulse velocity of laboratory cubes isnot significantly affected until a stress of approximately 50% of the crushingstrength is reached. This has been confirmed by the authors (83) and others(65) who have also shown from tests on beams that concrete subjectedto flexural stress shows similar characteristics. At higher stress levels, anapparent reduction in pulse velocity is observed due to the formation ofinternal microcracks which will influence both path length and width.It has been clearly shown that, under service conditions in which stresseswould not normally exceed one-third cube strength, the influence of compressivestress on pulse velocity is insignificant, and that pulse velocitiesfor prestressed concrete members may be used with confidence. Only if amember has been seriously overstressed will pulse velocities be affected.Tensile stresses have been found to have a similarly insignificant effect,

Ultrasonic pulse velocity methods 65but potentially cracked regions should be treated with caution, even whenmeasurements are parallel to cracks, since these may introduce path widthsbelow acceptable limits.3.3.2.3 Path lengthPulse velocities are not generally influenced by path length provided thatthis is not excessively small, in which case the heterogeneous nature of theconcrete may become important. Physical limitations of the time-measuringequipment may also introduce errors where short path lengths are involved.These effects are shown in Figure 3.11, in which a laboratory specimenhas been incrementally reduced in length by sawing. BS EN 12504-4 (76)recommends minimum path lengths of 100 and 150 mm for concrete withmaximum aggregate sizes of 20 and 40 mm respectively. For unmouldedsurfaces, a minimum length of 150 mm should be adopted for direct, or400 mm for indirect, readings.There is evidence (63) that the measured velocity will decrease withincreasing path length, and a typical reduction of 5% for a path lengthincrease from approximately 3 to 6 m is reported. This is because attenuationof the higher frequency pulse components results in a less clearly definedpulse onset. The characteristics of the measuring equipment are thereforean important factor. If there is any doubt about this, it is recommended thatFigure 3.11 Effect of short path length (based on ref. 83).

66 Ultrasonic pulse velocity methodssome verification tests be performed, although in most practical situationspath length is unlikely to present a serious problem.3.3.2.4 Moisture conditionsThe pulse velocity through saturated concrete may be up to 5% higher thanthrough the same concrete in a dry condition, although the influence will beless for high-strength than for low-strength concretes. The effect of moisturecondition on both pulse velocity and concrete strength is thus a furtherfactor contributing to calibration difficulties, since the moisture contentof concrete will generally decrease with age. A moist specimen shows ahigher pulse velocity, but lower measured strength than a comparable dryspecimen, so that drying out results in a decrease in measured pulse velocityrelative to strength. The effect is well illustrated by the results in Figure 3.12which relate to otherwise identical laboratory specimens, and demonstratesthe need to correlate test cube moisture and structure moisture duringstrength calibration. It is thus apparent that strength correlation curves areof limited value for application to in-place concrete unless based on theappropriate moisture conditions.Figure 3.12 Effect of moisture conditions (based on ref. 83).

Ultrasonic pulse velocity methods 67Tomsett (47) has presented an approach which permits calibration for‘actual’ in-situ concrete strength to be obtained from a correlation basedon standard control specimens. The relationship between specimens curedunder different conditions is given aslog ef 1f 2= kf 1 V 1 − V 2 where f 1 is the strength of a ‘standard’ saturated specimenf 2 is the ‘actual’ strength of the in-situ concreteV 1 is the pulse velocity of the ‘standard’ saturated specimenV 2 is the pulse velocity of the in-situ concreteand k is a constant reflecting compaction control (a value of 0.015 is suggestedfor normal structural concrete, or 0.025 if poorly compacted). Thiseffect is illustrated by Figure 3.13, which is based on Tomsett’s work. Forany given curing conditions, it is possible to draw up a strength/pulse velocityrelationship in this way, and similar members in a structure can becompared from a single correlation, which may be assumed to have theFigure 3.13 Desiccation line method (based on ref. 47).

68 Ultrasonic pulse velocity methodssame slope as the ‘standard’ saturated specimen relationship. This simpleapproach allows for both strength and moisture differences between in-situconcrete and control specimens. Swamy and Al-Hamed have also recommendeda set of k values in a similar range, based on mix characteristics,and claim that these should enable in-situ strength estimation to within±10% (84). However, a direct strength assessment of a typical referencespecimen of in-situ concrete is still preferred if the relationship is to be usedfor other than comparative applications.3.3.2.5 ReinforcementReinforcement, if present, should be avoided if at all possible, since considerableuncertainty is introduced by the higher velocity of pulses in steel coupledwith possible compaction shortcomings in heavily reinforced regions.There will, however, often be circumstances in which it is impossible toavoid reinforcing steel close to the pulse path, and corrections to the measuredvalue will then be necessary. Corrections are not easy to establish,and the influence of the steel may dominate over the properties of theconcrete so that confidence in estimated concrete pulse velocities will bereduced.The pulse velocity in an infinite steel medium is close to 5.9 km/s, butthis has been shown to reduce with bar diameter to as little as 5.1 km/salong the length of a 10 mm reinforcing bar in air (83). The velocity alonga bar embedded in concrete is further affected by the velocity of pulses inthe concrete and the condition of the bond between steel and concrete.The apparent increase in pulse velocity through a concrete memberdepends upon the proximity of measurements to reinforcing bars, the diameterand number of bars and their orientation with respect to the propagationpath. An increase will occur if the first pulse to arrive at the receivingtransducer travels partly in concrete and partly in steel. Correction factorsoriginally suggested by RILEM (85) assumed an average constant valueof pulse velocity in steel and gave the maximum possible influence of thesteel. The procedure adopted by the now obsolete British Standard (BS1881: Part 203) and described below was based on extensive experimentalwork by the authors (86) and takes bar diameter into account, yieldingsmaller corrections (see Figure 3.17). The new European Standard BS EN12504-4 (76) provides no specific guidance other than to avoid steel parallelto the pulse path. For practical purposes, with concrete pulse velocitiesof 4.0 km/s or above, 20 mm diameter bars running transversely to thepulse path will have no significant influence upon measured values but barslarger than 6 mm diameter running along the path may have a significanteffect.

Ultrasonic pulse velocity methods 69Figure 3.14 Reinforcement parallel to pulse path.There are two principal cases to be considered:(i) Axis of bars parallel to pulse pathAs shown in Figure 3.14, if a bar is sufficiently close to the path, the firstwave to be received may have travelled along the bar for part of its journey.It is suggested that a relationship ofV c =2aV s√4a2 + TV s − L 2 when V s ≥ V c (3.2)is appropriate, where V s = pulse velocity in steel barand V c = pulse velocity in concreteand that this effect disappears when√aL > 1 V s − V c2 V s + V cHence steel effects may be significant when a/L < 015 in high-qualityconcrete or

70 Ultrasonic pulse velocity methodswith = V cV sThe value of may be obtained from Figure 3.15 which has been plotted fora range of commonly occurring values of V c and bar diameter, for a 54 kHzfrequency. This may be substituted in Equation (3.4) (or Figure 3.16),to obtain a value of correction factor k to use in Equation (3.3). Theseequations are only valid where the offset a is greater than about twice theend cover to the bar b. Otherwise, pulses are likely to pass through the fulllength of the bar and(√ )a2 + b 2 − bk = + 2L(3.5)If the bar is directly in line with the transducers, a = 0 and the correctionfactor is given byk = 1 − L s1 − (3.6)Lwhere L s is the length of the bar (mm).Figure 3.15 Relationship between bar diameter and velocity ratio for bars parallelto pulse path (based on ref. 86).

Ultrasonic pulse velocity methods 71Figure 3.16 Correction factors for bars parallel to pulse path a > 2b (based onref. 86).An iterative procedure may be necessary to obtain a reliable estimate ofV c , and this is illustrated by an example in Appendix B. Estimates are likelyto be accurate to within ±30% if there is good bond and no cracking ofconcrete in the test zone. Correction factors relating to a typical case of abar in line with the transducers are shown in Figure 3.17, and comparedwith RILEM values which significantly overestimate steel effects for thesmaller bar sizes.Corrections must be treated with caution, especially since it is essentiallythe pulse through the concrete surrounding the bar that is being measured,rather than the body of the material. Complex bar configurations close tothe test location will increase uncertainty.(ii) Axis of bars perpendicular to the pulse pathFor the situation shown in Figure 3.18(a), if the total path length throughsteel across the bar diameters is L s the maximum possible steel effect isgiven by Figure 3.18(b) for varying bar diameters and concrete qualities,where V c is the true velocity in the concrete.In this case, the value of is used in Equation (3.6) to obtain the correctionfactor k. The effect on the bars on the pulse is complex, and theeffective velocity in the steel is less than that along the axis of bars of similarsize. Results for a typical case are shown in Figure 3.17, and the calculationprocedure is illustrated in Appendix B.

72 Ultrasonic pulse velocity methodsFigure 3.17 Typical correction factors.Figure 3.18 Reinforcement transverse to pulse path: (a) Path through transversereinforcement; (b) Bar diameter/velocity ratio (based on ref. 86).3.4 ApplicationsThe applications of pulse velocity measurements are so wide-ranging that itwould be impossible to list or describe them all. The principal applicationsare outlined below – the method can be used both in the laboratory and onsite with equal success.

Ultrasonic pulse velocity methods 733.4.1 Laboratory applicationsThe principal laboratory applications lie in the monitoring of experimentsthat may be concerned either with material or structural behaviour. Theseinclude strength development or deterioration in specimens subjected tovarying curing conditions, or to aggressive environments. The detectionof the onset of micro-cracking may also be valuable during loading testson structural members, although the method is relatively insensitive tovery early cracking. For applications of this nature, the equipment is mosteffective if connected to a continuous recording device with the transducersclamped to the surface, thus removing the need for repeated applicationand associated operating errors.3.4.2 In-situ applicationsThe wide-ranging and varied applications do not necessarily fall into distinctcategories, but are grouped below according to practical aims andrequirements.3.4.2.1 Measurement of concrete uniformityThis is probably the most valuable and reliable application of the methodin the field. There are many published reports of the use of ultrasonicpulse velocity surveys to examine the strength variations within members asdiscussed in Chapter 1. The statistical analysis of results, coupled with theproduction of pulse velocity contours for a structural member, may oftenalso yield valuable information concerning variability of both material andconstruction standards. Readings should be taken on a regular grid overthe member. A spacing of 1 m may be suitable for large uniform areas, butthis should be reduced for small or variable units. Typical pulse velocitycontours for a beam constructed from a number of batches are shown inFigure 3.19.Figure 3.19 Typical pulse velocity beam contours (km/s).

74 Ultrasonic pulse velocity methodsTomsett (47) has suggested that for a single site-made unit constructedfrom a single load of concrete, a pulse velocity coefficient of variation of1.5% would represent good construction standards, rising to 2.5% whereseveral loads or a number of small units are involved. A correspondingtypical value of 6–9% is also suggested for similar concrete throughout awhole structure. An analysis of this type may therefore be used as a measureof construction quality, and the location of substandard areas can beobtained from the ‘contour’ plot. The plotting of pulse velocity readingsin histogram form may also prove valuable, since concrete of good qualitywill provide one clearly defined peak in the distribution, with poor qualityconcrete or two different qualities of concrete being clearly apparent (seeSection 1.6.2.1). Used in this way, ultrasonic pulse velocity testing couldbe regarded as a form of control testing, although the majority of practicalcases in which this method has been used are related to suspected constructionmalpractice or deficiency of concrete supply. A survey of an existingstructure will reveal and locate such features, which may not otherwise bedetected. Although it is preferable to perform such surveys by means ofdirect readings across opposite faces of the member, indirect readings canbe used successfully; for example, for comparison and determination ofsubstandard areas of floor slabs.Decisions concerning the seriousness of defects suggested by surveys ofthis type will normally require an estimate of concrete strength. As indicatedin Section 3.4.2.3, a reliable estimate of absolute strength is not possibleunless a calibration is available. If the mean strength of the supply is known,the relationship f c = kV 4 has been found satisfactory for estimating relativevalues over small ranges (47). Failing this, it will be necessary to resort toa more positive partially destructive method, or core sampling, to obtainstrength values, with the locations determined on the basis of the ultrasoniccontour plot.3.4.2.2 Detection of cracking and honeycombingA valuable application of the ultrasonic pulse velocity techniques which doesnot require detailed correlation of pulse velocity with any other propertyof the material is in the detection of honeycombing and cracking. Sincethe pulse cannot travel through air, the presence of a crack or void on thepath will increase the path length (as it goes around the flaw) and increaseattenuation so that a longer transit time will be recorded. The apparentpulse velocity thus obtained will be lower than for the sound material.Since compression waves will travel through water, it follows that thisphilosophy will apply only to cracks or voids which are not water-filled.Tomsett (47) has examined this in detail and concluded that although waterfilledcracks cannot be detected, water-filled voids will show a lower velocitythan the surrounding concrete. Voids containing honeycombed concrete of

Ultrasonic pulse velocity methods 75low pulse velocity will behave similarly. The variation in pulse velocity dueto experimental error is likely to be at least 2%, notwithstanding variationsin concrete properties; hence the size of a void must be sufficient to causean increase in path length greater than 2% if it is to be detected. A givenvoid is thus more difficult to detect as the path length increases, but theabsolute minimum size of detectable defect will be set by the diameter ofthe transducer used.In crack detection and measurement, even micro-cracking of concretewill be sufficient to disrupt the path taken by the pulses, and the authors(83) have shown that at compressive stresses in excess of 50% of the cubecrushing strength, the measured pulse velocity may be expected to dropdue to disruption of both path length and width. If the velocity for thesound concrete is known it is therefore possible to detect overstressing, orthe onset of cracking may be detected by continual monitoring during loadincrease.An estimate of crack depths may be obtained by the use of indirectsurface readings as shown in Figure 3.20. In this case, where the transducersare equidistant from a known crack, if the pulse velocity through soundconcrete is V km/s, then:Path length without crack = 2x√Path length around crack = 2 x 2 + h 2Surface travel time without crack = 2xV = T sTravel time around crack = 2√ x 2 + h 2= TVcand it can be shown that√ (T )2crack depthh= x c− 1Ts2Figure 3.20 Crack depth measurement.

76 Ultrasonic pulse velocity methodsAn accuracy of ±15% may be possible, but difficulties may be caused by thetapering nature of flexural cracks and the presence of dust or debris in thecrack. In in-situ concrete, reinforcing bars which cross the crack may alsoinfluence results (87) and are difficult to allow for, whilst closely spacedcracks may cause testing problems in terms of transducer locations. Thisapproach may be modified for applications to other situations as necessaryand several alternative transducer arrangements are possible, includingstepping of transmitter and/or receiver locations as described in the formerBritish Standard (BS 1881: Part 203).Hashimoto et al. (88) have recently used the approach to examine thedepth and frequency of settlement cracks in a fly-ash modified concrete.There have also been many reports of application to the monitoring ofrepairs to concrete, based on the principle that poor bond or compactionwill hinder the passage of pulses.The location of honeycombing is best determined by the use of directmeasurements through the suspect member, with readings taken on a regulargrid. If the member is of constant thickness, a ‘contour map’ of transittimes will readily show the location and extent of areas of poor compaction.Good surface contact is critical here, however, to avoid areas of‘apparent’ poor quality, where low readings have been obtained by poorcontact. An adequately smooth surface, sufficient couplant and transducersdiametrically opposite each other, to avoid path length errors, are allcritical here.3.4.2.3 Strength estimationUnless a suitable correlation curve can be obtained, it is virtually impossibleto predict the absolute strength of a body of in-situ concrete by pulsevelocity measurements. Although it is possible to obtain reasonable correlationswith both compressive and flexural strength in the laboratory,enabling the strength of comparable specimens to be estimated to ±10%,the problems of relating these to in-situ concrete are considerable. If it isto be attempted, then the most reliable method is probably the use of coresto establish the calibration curve coupled with Tomsett’s moisture correction.The authors (83) have suggested that if a reliable correlation chartis available, together with good testing conditions, it may be possible toachieve 95% confidence limits on a strength prediction of ±20% relatingto a localized area of interest. Expected within-member variations are likelyto reduce the corresponding accuracy of overall strength prediction of amember to the order of ±10 N/mm 2 at the 30 N/mm 2 mean level. Accuracydecreases at higher strength levels, and estimates above 40 N/mm 2 shouldbe treated with great caution. Chhabra (89) has used UPV concrete strength

Ultrasonic pulse velocity methods 77estimates in decision-making on retrofitting of a series of buildings withcarbon-fibre wrapping.Although not perfect, there may be situations in which this approach mayprovide the only feasible method of in-situ strength estimation, and if this isnecessary it is particularly important that especial attention is given to therelative moisture conditions of the calibration samples and the in-situ concrete.Failure to take account of this is most likely to cause an underestimateof in-place strength, and this underestimate may be substantial.It is claimed (55) that significant improvements in accuracy can beobtained by combination with other techniques such as rebound hammertests as described in Chapter 1, but this approach has never achieved popularityin the UK or USA. In the author’s view, there may well be someadvantage in using a combination of rebound hammer and UPV measurementsto assess whether an area of suspect concrete is actually defective, bycomparing with other, similar, acceptable members. In this way, the effectsof moisture content are likely to be minimized and calibration with strengthis not an issue as only comparative UPV and rebound hammer values arebeing used. This could be considerably less disruptive than taking core samples,with the inevitable arguments about the accuracy of correlation withcube strength.3.4.2.4 Assessment of concrete deteriorationUltrasonic methods are commonly used in attempting to define the extentand magnitude of deterioration resulting from fire, mechanical, frost orchemical attack. A general survey of the type described in Section 3.4.2.1will easily locate suspect areas, whilst a simple method for assessing thedepth of fire or surface chemical attack has been suggested by Tomsett (47).In this approach it is assumed that the pulse velocity for the sound interiorregions of the concrete can be obtained from unaffected areas, and that thedamaged surface velocity is zero. A linear increase is assumed between thesurface and interior to enable the depth to sound concrete to be calculatedfrom a transit time measured across the damaged zone. For example, if atime T is obtained for a path length L including one damaged surface zoneof thickness t, and the pulse velocity for sound concrete is V c it can beshown that the thickness is given byt = TV c − LAlthough this provides only a very rough estimate of damage depth, it isreported that the method has been found to give reasonable results in anumber of fire damage investigations. Benedetti (90) has also proposed amore complex approach based on indirect measurements.

78 Ultrasonic pulse velocity methodsWhere deterioration of the member is more general, it is possible thatpulse velocities may reflect relative strengths either within or between members.There is a danger that elastic modulus, and hence pulse velocity, maynot be affected to the same degree as strength and caution should thereforebe exercised when using pulse velocities in this way.Although it may be possible to develop laboratory calibrations for a concretesubjected to a specific form of attack or deterioration, as was attemptedwhen evaluating high alumina cement decomposition in the United Kingdom(91), absolute strength predictions of in-situ deteriorated concrete mustbe regarded as unreliable. In-situ comparison of similar members to identifythose which are suspect, for subsequent load testing, has however been carriedout successfully in the course of a number of HAC investigations, andpulse velocities have been shown to be sensitive to the initiation and developmentof alkali–silica reaction (72,92). This provides a relatively quickand cheap approach where a large number of precast units, for example, areinvolved. Conducting repetitive tests on the same element can also monitorlong-term performance of concrete very successfully.3.4.2.5 Measurement of layer thicknessThis is essentially a development of the indirect reading method, which isbased on the fact that as the path length increases, the pulse will naturallytend to travel through concrete at an increasing depth below the surface.This is particularly appropriate for application to slabs in which a surfacelayer of different quality exists due to construction, weathering, or otherdamage such as fire. The procedure is exactly as described for obtainingan indirect measurement (Section 3.2.2.1). When the transducers are closetogether the pulse will travel in the surface layer only, but at greater spacings,the path will include the lower layer. This effect will be shown bya discontinuity in the plot of transit time vs. transducer spacing, with thepulse velocities through the two layers having different slopes, as shown inFigure 3.21. The thickness t of the upper layer is related to the velocitiesV 1 and V 2 , and the spacing x at which the discontinuity is observed, by theexpressiont = x 2√V 2 − V 1 V 2 + V 1 Although this is most suitable for a distinct layer of uniform thickness, thevalue obtained can be at best only an estimate, and it must be borne in mindthat there will be a maximum thickness of layer that can be detected. Littleinformation is available concerning the depth of penetration of indirect

Ultrasonic pulse velocity methods 79Figure 3.21 Layer thickness measurement.readings, and in view of the weakness of signal received using this methodthe results must be treated with care.3.4.2.6 Measurement of elastic modulusThis is the property that can be measured with the greatest numericalaccuracy. Values of pulse modulus can be calculated theoretically usingan assumed value of Poisson’s ratio to yield a value within ±10%, ormore commonly an estimate of dynamic modulus can be obtained from thereliable correlation with resonant frequency values (see Figure 3.7). Whereassuch measurements may be valuable in the laboratory when undertakingmodel testing, their usefulness on site is limited, although they may be usedto provide an estimated static elastic modulus value for use in calculationsrelating to load tests.3.4.2.7 Strength development monitoringIt has been well established that pulse velocity measurements will accuratelymonitor changes in the quality of the paste with time, and this maybe usefully applied to the control of demoulding or stressing operationsboth in precasting works and on site. In this situation a specific pulsevelocity/strength relationship for the mix, subject to the appropriate curingconditions, can be obtained and a safe acceptance level of pulse velocity

80 Ultrasonic pulse velocity methodsestablished. In the same way, quality control of similar precast units mayeasily be undertaken and automated techniques incorporating amplitudeassessment have been used. Popovics has also outlined a laboratory techniqueusing pulse echo techniques (70) which is still being developed.3.4.2.8 Ultrasonic ImagingAs indicated above, use of ultrasonic tomography for imaging of defects andinclusions in concrete has been increasing. The ultrasonic echo methods usethe principle of synthetic aperture. The ultrasonic data are taken along linesor in a defined surface range. From such data fields a reconstruction calculationis performed (SAFT). For the ultrasonic 2D-method many A-scans(ultrasonic intensity vs. time) are measured point-by-point using a broadbandtransducer (nominal frequency f = 200 kHz) (69). The transducer actsas transmitter and receiver. The L-SAFT (Linear-SAFT) reconstruction iscalculated and indicates the reflectors and scatterers below the scanningline (ultrasonic B-scan : ultrasonic intensity along the axis vs. depth). Thismethod is relatively fast; a line of 80 cm length can be measured and reconstructedin about 90 minutes. For the ultrasonic 3D-method an array of tenbroadband transducers is used, excited with programmed amplified pulsesin the frequency range of typically 80 or 150 kHz (67). The transducersare positioned on the concrete surface using a template. They are operatedin transmit–receive configuration using an electronic multiplexer. The templateis moved in steps of typically 2 cm across the surface, so that severalthousand A-scans are recorded along an investigated object, for examplea tendon duct. A typical B-Scan constructed from a 3D SAFT is shown inFigure 3.22.Front of ductindicatedLocation ofair voids00 500 1000 1500X (mm)5000 60 120 180 255Y (mm)Back of ductindicatedSignal intensityFigure 3.22 3D B-Scan Ultrasonic echo imaging of post-tensioned duct in a concretespecimen with air voids (based on ref. 67).

3.5 Reliability and limitationsUltrasonic pulse velocity methods 81Ultrasonic pulse velocity measurement has been found to be a valuable andreliable method of examining the interior of a body of concrete in a trulynon-destructive manner. Modern equipment is robust, reasonably cheapand easy to operate, and reliable even under site conditions; however, itcannot be overemphasized that operators must be well trained and awareof the factors affecting the readings. It is similarly essential that results areproperly evaluated and interpreted by experienced engineers who are familiarwith the technique. It should be noted that allowances for toleranceson measurement of transit time and path length combine to mean that achange in calculated pulse velocity of at least 2% will be needed to reflect asignificant change in properties. For comparative purposes the method hasfew limitations, other than when two opposite faces of a member are notavailable. The method provides the only readily available method of determiningthe extent of cracking within concrete; however, the use for detectionof flaws within the concrete is not reliable when the concrete is wet.Unfortunately, the least reliable application is for strength estimationof concrete. The factors influencing calibrations are so many that evenunder ideal conditions, with a specific calibration, it is unlikely that 95%confidence limits of better than ±20% can be achieved for an absolutestrength prediction for in-place concrete. Although it is recognized thatthere may be some circumstances in which attempts must be made to usethe method for strength prediction, this is not recommended. It is far betterthat attention is concentrated upon the use of the method for comparisonof supposedly similar concrete, possibly in conjunction with some otherform of testing rather than attempt applications which are recognized asunreliable and which will therefore be regarded with skepticism (see alsoSection 3.4.2.3).

Chapter 4Partially destructive strengthtestsConsiderable developments have taken place in recent years in methodswhich are intended to assess in-situ concrete strength, but cause some localizeddamage. This damage is sufficiently small to cause no loss in structuralperformance. All are surface zone tests which require access to only oneexposed concrete face. Methods incorporate variations of the concepts ofpenetration resistance, pull-out, pull-off and break-off techniques whichhave been proposed over many years. Estimation of strength is by means ofcorrelation charts which, in general, are not sensitive to as many variablesas are rebound hammer or pulse-velocity testing. There are drawbacks inapplication and accuracy, which vary according to the method, but thereare many circumstances in which these methods have been shown to be ofconsiderable value. A key feature is that an estimate of strength is immediatelyavailable, compared with delays of several days for core testing,and although accuracy may not be as good, the testing is considerably lessdisruptive and damaging.The best established of these methods are covered by American and othernational standards, and are incorporated in BS 1881: Part 207 (93) and areport by ACI Committee 228 (23). The choice of method for particularcircumstances will depend largely upon whether the testing is preplannedbefore casting, together with practical factors such as access, cost, speedand prior knowledge of the concrete involved. The tests can be used onlywhere making good of surfaces is acceptable.4.1 Penetration resistance testingThe technique of firing steel nails or bolts into a concrete surface to providefixings is well established, and it is known that the depth of penetration isinfluenced by the strength of the concrete. A strength determination methodbased on this approach, using a specially designed bolt and standardizedexplosive cartridge, was developed in the USA during the mid-1960s andis known as the Windsor probe test (94). It has gained popularity in theUSA and Canada, especially for monitoring strength development on site,

Partially destructive strength tests 83and is the subject of ASTM C803 (95). Many authorities in North Americaregard it as equivalent to site cores, and in some cases it is accepted inlieu of control cylinders for compliance testing. Use outside North Americahas been limited, but the equipment is readily available and the method isincluded in BS 1881: Part 207 (93).Although it is difficult to relate theoretically the depth of penetration ofthe bolt to the concrete strength, consistent empirical relationships can befound that are virtually unaffected by operator technique. The method isa form of hardness testing and the measurements will relate only to thequality of concrete near the surface, but it is claimed that it is the zonebetween approximately 25 and 75 mm below the surface which influencesthe penetration. The depth is considerably greater than for rebound or anyother established ‘surface zone’ tests.A smaller-scale method has also been proposed (96) in which a springloadedhammer drives a small pin into the concrete surface to a depth ofbetween 4 and 8 mm. This pin penetration test is primarily intended fordetermination of in-situ concrete strength to permit formwork stripping.4.1.1 Windsor probe4.1.1.1 Test equipment and operationThe bolt or probe which is fired into the concrete (Figure 4.1) is of ahardened steel alloy. The principal features are a blunt conical end topunch through the matrix and aggregate near the surface, and a shoulder toimprove adhesion to the compressed concrete and ensure a firm embedment.The probes are generally 6.35 mm in diameter and 79.5 mm in length,but larger-diameter bolts (7.94 mm) are available for testing lightweightconcretes. Probes are also available for use with high strength concretes upto 110 N/mm 2 . A steel firing head is screwed on to the threaded end ofthe bolt and the plastic guide locates the probe within the muzzle of thedriver from which it is fired. The driver, which is shown in operation inFigure 4.2, utilizes a carefully standardized powder cartridge. This impartsa constant amount of energy to the probe irrespective of firing orientation,and produces a velocity of 183 m/s which does not vary by more than ±1%.The power level can be reduced when dealing with low strength concretessimply by locating the probe at a fixed position within the driver barrel.The driver is pressed firmly against a steel locating plate held on the surfaceof the concrete which releases a safety catch and permits firing when thetrigger is pulled. After firing, the driver head and locating plate are removedand any surface debris around the probe is scraped or brushed away togive a level surface. A flat steel plate is placed on this surface, and a steel

84 Partially destructive strength testsFigure 4.1 Penetration resistance test probe.cap screwed onto the probe to enable the exposed length to be measuredto the nearest 0.5 mm with a spring-loaded calibrated depth gauge, as inFigure 4.3. An electronic measuring device is now also available for thispurpose.Probe penetrations may be measured individually as described, or alternativelythe probes may be measured in groups of three using a triangulartemplate with the probes at 177 mm centres. In this case, a system of triangularmeasuring plates is used which will provide one averaged readingof exposed length for the group of probes. This approach may mask inconsistenciesbetween individual probes, and it is preferable to measure eachprobe individually. The measured average value of exposed probe lengthmay then be directly related to the concrete strength by means of appropriatecalibration tables or charts.It is important to recognize that in the UK it is necessary to comply withthe requirements of BS 4078: Part 1 (97) concerning the use of powderactuateddriving units, as well as a range of Health and Safety Acts whichare listed in BS 1881: Part 207 (93). These restrictions may limit the use ofthe technique in some situations.

Partially destructive strength tests 85Figure 4.2 Driver in use.4.1.1.2 ProcedureIndividual probes may be affected by particularly strong aggregate particlesnear the surface, and it is thus recommended that at least three tests aremade and averaged to provide a result. If the range of a group of threetests exceeds 5 mm, a further test should be made and the extreme valuediscarded. Although slight surface roughness is not important, surfacescoarser than a broom finish should be ground smooth prior to test, and theprobe must always be driven perpendicular to the surface.Where the expected cube strength of the concrete is less than 26 N/mm 2 ,the ‘low power’ setting should be used, but for higher strengths, this penetrationmay not be sufficient to ensure firm embedment of the probe andthe ‘standard power’ setting is necessary. If probes will not remain fixed invery high strength concrete it may be possible to measure directly the depthof hole formed, after cleaning, and subtract this from the probe length. Itshould be noted, however, that this does not conform to BS 1881: Part 207requirements. The manufacturers of the system recommend that a minimumedge distance of 100 mm should be maintained (75 mm for low power) butthe authors’ experience suggests that these values may not always be sufficientto prevent splitting. Probes should also be at least 175 mm apart toavoid overlapping of zones of influence.

86 Partially destructive strength testsFigure 4.3 Height measurement.BS 1881: Part 207 recommends a 150 mm edge distance and 200 mmminimum spacing, with the added restriction that a test should not belocated within 50 mm of a reinforcing bar. A minimum concrete elementthickness of 150 mm is also recommended.Aggregate hardness is an important factor in relating penetration tostrength, and it may therefore be necessary to determine its value. This isassessed on the basis of the Mohs’ hardness scale, which is a system forclassifying minerals in terms of hardness into ten groups. Group 10 is thehardest and Group 1 the softest, thus any mineral will scratch anotherfrom a group lower than itself. Testing consists of scratching the surface ofa typical aggregate particle with minerals of known hardness from a test

Partially destructive strength tests 87kit; the hardest is used first, then the others in order of decreasing hardnessuntil the scratch mark will wipe off. The first scratch that can be wiped offrepresents the Mohs’ classification for the aggregate.4.1.1.3 Theory, calibration and interpretationA convincing theoretical description of the penetration of a concrete massby a probe is not available, since there is little doubt that a complex combinationof compressive, tensile, shear and friction forces must exist. Themanufacturers of the Windsor probe equipment have suggested that penetrationis resisted by a subsurface compressive compaction bulb as shownin Figure 4.4. The surface concrete will crush under the tip of the probe,and the shock waves associated with the impact will cause fracture lines,and hence surface spalling, adjacent to the probe as it penetrates the bodyof concrete. The energy required to cause this spalling, or to break pieces ofaggregate, is a low percentage of the total energy of a driven probe, and willtherefore have a small effect upon the depth of penetration. Penetration willcontinue, with cracks not necessarily reaching the surface and eventuallyceasing to form as the stress drops. Energy is absorbed by the continuouscrushing at the point, by surface friction and by compression of the bulbof contained concrete. It is this latter effect which prevents rebound of theprobe, and it is claimed that the bulb, and depth of penetration, will beinversely proportional to the compressive strength. Data are not currentlyavailable to support these proposals, which must be regarded as rather simplistic,although the concept of the measured property relating to concretebelow rather than at the surface seems reasonable.Figure 4.4 Compaction bulb.

88 Partially destructive strength testsAlthough it may theoretically be possible to undertake calculations basedon the absorption of the kinetic energy of the probe, this would be difficultand it is very much easier to establish empirical relationships betweenpenetration and strength.Calibration is hampered by the minimum edge distance requirementwhich prevents splitting. Although it may be possible to use standard150 mm cubes or cylinders for tests at low power, the specimen must besecurely held during the test. A holding jig for cylinders is available from theWindsor probe manufacturers, and cubes are most conveniently clampedin a compression-testing machine, although no data concerning the influenceof applied compressive strength are available. It is recommended byMalhotra (63) that groups of at least six specimens from the same batchare used, with three tested in compression and three each with one probetest, and the results averaged to produce one point on the calibrationgraph. Malhotra has also shown that the reduction in measured compressivestrength of cylinders which have been previously probed may be up to17.5%, and such specimens cannot therefore be tested in compression forcalibration purposes.Where the cube strength of the concrete is greater than 26 N/mm 2 it isnecessary to use a combination of cubes or cylinders for compression testingand larger slab or beam specimens from the same batch for probing. The sizeof such specimens is unimportant provided that they are large enoughto accommodate at least three probes which satisfy the minimum edgedistance and spacing requirements. These test specimens must however besimilarly compacted, and all should be cured together. In such situations, theuse of ultrasonic pulse velocity measurements to compare concrete qualitybetween specimens would be valuable. This approach has been used by theauthors (98) in an investigation in which 1000×250×150 mm beams wereused for probing, and 100 mm cubes for compression testing, and it wasfound that the beam concrete was between 10 and 20% lower in strengththan the concrete in the cubes. Since calibrations will normally relate toactual concrete strength it is also important that the moisture conditionsof the specimens are similar. Figure 4.5 shows a typical calibration chartobtained in this way with a strength range obtained by water/cement ratioand age variations. Relationships between penetration and strength for thetwo different power levels are not easily related, and it is therefore necessaryto produce calibration charts for each experimentally.The manufacturers of the test equipment provide calibration tables inwhich aggregate hardness is taken as the only variable influencing the penetration/strengthrelationship. It is clear from the authors’ work and fromreported experience in the USA (99) that this is not the case, and thataggregate type can also have a large influence. It is understood that themanufacturer’s tables are based on crushed rock, but for rounded gravelsthe crushing strength may be lower than suggested by probe results. It is

Partially destructive strength tests 89Figure 4.5 Typical low power strength calibration (based on ref. 98).to be expected that bond differences at the aggregate/matrix interface dueto aggregate surface characteristics may affect penetration resistance andcrushing strength. Nevertheless, the extent of the calibration discrepancywhich may be attributed to this, as indicated in Figure 4.6, is disturbing.Calibrations from a number of sources are compared in Figure 4.7. Itappears that moisture condition, aggregate size (up to 50 mm) and aggregateproportions all have effects which are small in relation to aggregate hardnessand type. Swamy and Al-Hamed (100) have also suggested that curingconditions and age are important, with differing penetration/strength relationshipsfor old and new concrete. It is essential therefore that appropriatecalibration charts should be developed for the particular aggregate typeinvolved in any practical application of the method, and this requirementhas also been confirmed for lightweight concretes (34). Al-Manaseer andAquino (101) have demonstrated that standard probes are liable to fracturewhere concrete cylinder strengths are above about 26 N/mm 2 . They reporttrials with modified probes which are shown to produce reliable resultswith granite aggregate concrete at cylinder strengths up to 120 N/mm 2 fora range of mixes, some of which include silica fume and fly ash.

Figure 4.6 Comparison of calibrations (based on refs 94 and 98).Figure 4.7 Influence of aggregate type and proportions (based on ref. 63).

4.1.1.4 Reliability, limitations and applicationsPartially destructive strength tests 91The test is not greatly affected by operator technique, although verticalityof the bolt relative to the surface is obviously important and a safety devicein the driver prevents firing if alignment is poor. It is claimed that anaverage coefficient of variation for a series of groups of three readings onsimilar concrete of the order of 5% may be expected, and that a correlationcoefficient of greater than 0.98 can be achieved for a linear calibrationrelationship for a single mix. Field tests by the authors on motorway deckslabs have also yielded a similar coefficient of variation of probe resultsover areas involving several truck loads of concrete. It is also apparentfrom Figure 4.5 that 95% limits of about ±20% on predicted strengthsmay be possible for a single set of three probes, given adequate calibrationcharts. Difficulty may be encountered in predicting strengths in the range25–50 N/mm 2 at ages greater than one year (100), and in the authors’experience the method cannot be reliably used for strengths below about10 N/mm 2 . Results for lightweight concrete (34) suggest that accuracy levelsmay be reduced when lightweight fines are present. It is to be expected thataggregate size will influence the scatter of individual probe readings, but atpresent insufficient data are available to assess the effect of this on strengthprediction accuracies, although a 50 mm maximum size is recommended.Similarly, the effect of reinforcement adjacent to the probe is uncertain,and a minimum clearance of 50 mm should be allowed between probes andreinforcing bars.The principal physical limitation of this method is caused by the needfor adequate edge distances and probe spacings together with a memberthickness of at least twice the anticipated penetration. After measurementthe probe can be extracted, leaving a conical damage zone (Figure 4.8)which must be made good. There is the additional danger of splitting of themember if it does not comply with the minimum recommended dimensions.Expense is a further consideration with relatively high cost equipment andrecurrent costs and the safety aspects outlined previously cannot be ignored.The limitations outlined above mean that although probe measurementtakes place at a greater depth within the concrete than rebound hammermeasurements, penetration tests are unlikely to replace rebound tests exceptwhere the latter are clearly unsatisfactory. Probes cannot examine the interiorof a member in the same way as ultrasonics and the method causesdamage that must be repaired. However, probes do offer the advantageof requiring only one surface and fewer calibration variables. In relationto cores, however, probes provide easier testing methods, speedy resultsand accuracy of strength estimation comparable to small-diameter specimens.Although the accuracies of large-diameter cores cannot be matched,it is likely that probing may be used as an alternative to cores in somecircumstances.

92 Partially destructive strength testsFigure 4.8 Surface damage caused by probe removal.In many parts of the world there is a trend towards in-place compliancetesting, especially in relation to post-tensioning. Since the most reliableapplication of the penetration method lies in comparison of similar concretewhere specific calibration charts can be obtained, a number of applicationsof this nature have been reported. It appears that the advantages of speedand simplicity, together with the ability to drive probes through timber oreven thin steel formwork without influence, outweigh the cost. Details ofacceptable thicknesses are unfortunately not available, but in such circumstancesdecisions would normally be based on previously established ‘go/nogo’ limits for measured penetration.Other applications include the detection of substandard members or areasof mature concrete, and this method is particularly appropriate for largewalls or slabs having only one exposed surface free of finishes. Investigationsof this type have been successfully performed by the authors on highwaybridge deck slabs. Probing was carried out on the deck soffit from a smallmobile hydraulic platform while the road above was in normal use, andin one such investigation a total of 18 sets of probes were placed by oneoperator in a period of six hours. The speed of operation, together with theimmediate availability of results, means that many more tests can be madethan if cores were being taken, and test locations can be determined in the

Partially destructive strength tests 93light of the results obtained. This is particularly valuable when attemptingto define the location and extent of substandard concrete.Whether or not the test will be of significant value in the strength assessmentof ‘unknown’ concrete is uncertain, but it is clear that results basedsolely on aggregate hardness are inadequate. It may be, however, that asmore results are made available it will be possible to increase the confidencewith which the method may be extended beyond comparative situations.4.1.2 Pin penetration testNasser and Al-Manaseer (96) have developed this more recent and smallerscale method, also covered by ASTM C803, aimed at determining formworkstripping times. The apparatus consists of a spring-loaded hammerwhich can grip a pin of 30.5 mm length and 3.56 mm diameter with the tipmachined at an angle of 225 . The spring is compressed by pressing thehammer against the concrete surface, and is released by a trigger causingthe pin and the attached shaft and hammer to impact the concrete surfacewith an energy of about 108 Nm. The depth of the hole created is measuredwith a dial gauge device after cleaning with an air blower.Calibration testing with gravel and lightweight concretes with cylinderstrengths between 3.1 and 241N/mm 2 has shown linear relationshipsbetween penetration and compressive strength which have good correlationcoefficients and are very close to each other (96). It is suggested that for practicalpurposes these can be combined. The same authors have subsequentlycompared the performance with a range of other test methods (99) andshown that the method compared well in terms of correlation accuracy beingthe only method not requiring separate calibration for lightweight concrete.It is suggested that a reading should be taken as the average of the bestfive of a group of seven tests to allow for local influences. The principaladvantages of the method seem to be its speed, simplicity, low cost andlow level of damage. The depth of penetration is unlikely to exceed 8 mm,hence reinforcement poses no problem. Results are limited, both in terms ofstrength range and aggregate and mix types. Features such as carbonationand temperature have yet to be examined in detail, but Shoya et al. (102)have provided some data suggesting coefficients of variation up to 18% andindicating difficulties of strength prediction with deteriorated or carbonatedsurfaces. The method seems to offer potential, however, worthy of furtherinvestigation.4.2 Pull-out testingThe concept of measuring the force needed to pull a bolt or some similardevice from a concrete surface has been under examination for many years.Proposed tests fall into two basic categories: those which involve an insert

94 Partially destructive strength testswhich is cast into the concrete, and those which offer the greater flexibilityof an insert fixed into a hole drilled into the hardened concrete. Castinmethods must be preplanned and will thus be of value only in testingfor specification compliance, whereas drilled-hole methods will be moreappropriate for field surveys of mature concrete. In both cases, the valueof the test depends upon the ability to relate pull-out forces to concretestrengths and a particularly valuable feature is that this relationship isrelatively unaffected by mix characteristics and curing history. Althoughthe results will relate to the surface zone only, the approach offers theadvantage of providing a more direct measure of strength and at a greaterdepth than surface hardness testing by rebound methods, but still requiresonly one exposed surface. Procedures have recently been reviewed in detailby Carino (103).4.2.1 Cast-in methodsReports were first published in the USA and USSR in the late 1930s describingtests in which a cast-in bolt is pulled from the concrete. These methodsdo not appear to have become popular, and it was not until 30 years laterthat practically feasible tests were developed. Two basic methods, bothof which require a threaded insert which is fixed to the shuttering priorto concreting, have emerged. A bolt is then screwed into the insert andpulled hydraulically against a circular reaction ring. The principal differencebetween the two systems, developed in Denmark and Canada respectively,lies in the shape of insert and loading technique. In both cases acone of concrete is ‘pulled out’ with the bolt, and the force required toachieve this is translated to compressive strength by the use of an empiricalcalibration.4.2.1.1 The Lok-testThis approach, developed at the Danish Technical University in the late1960s, has gained popularity in Scandinavia and is accepted by a numberof public agencies in Denmark as equivalent to cylinders for acceptancetesting (1). It has subsequently gained wide international acceptance as amethod for demonstrating adequate strength for early formwork stripping.Time savings of 30% and labour savings of 45% have been claimed withstripping taking place in as little as 19 hours. The potential for use infast-track construction has been recognized in the UK by a Best PracticeGuide (104).The insert (Figure 4.9) consists of a steel sleeve which is attached to a25 mm diameter, 8 mm thick anchor plate located at a depth of 25 mmbelow the concrete surface (105). The sleeve is normally screwed to the

Partially destructive strength tests 9525 mmFormFailure conecut-off25 mm55 mmRemovable stem8.5 mmanchor plateReaction ringFigure 4.9 Lok-test insert.shuttering, or fixed to a plastic buoyancy cup where slabs are to be tested.This is later removed and replaced by a rod of 7.2 mm diameter which isscrewed into the anchor plate and coupled to a tension jack. The wholeassembly is pre-coated to prevent bonding to the concrete, and rotation ofthe plate is prevented by the ‘cut-off’. A special extension device is alsoavailable to permit tests at greater depth if required. Load is applied tothe pull-bolt by means of a portable hand-operated hydraulic jack with areaction ring of 55 mm diameter. This equipment (Figure 4.10) is compact,with a weight of less than 5 kg.The loading equipment can determine the force required to cause failureby pulling the disc, and a range of jacks are available to cover allpractical concrete strengths including ‘high-strength’ concretes. The load ismeasured with an accuracy of ±2% over normal operating temperatures,and a precision valve system combined with a friction coupling ensures aconstant loading rate of 30 ± 10 kN/min. This system complies with therecently introduced BS EN 12504-3 (106). Electronic digital reading apparatuswith data storage facilities is also available. Load is released as soon asa peak is reached, leaving only a fine circular crack on the concrete surface.Calibration charts as those provided by Petersen (1,107) (Figure 4.11) orspecifically developed by the user are then used to estimate the compressivestrength of the concrete.BS 1881: Part 207 (93) recommends that the centres of test positionsshould be separated by at least eight times the insert head diameter, and

96 Partially destructive strength testsFigure 4.10 Lok-test equipment (photograph by courtesy of Germann Instruments).that minimum edge distances should be four diameters. An element thicknessof at least four insert head diameters is needed and tests should belocated so that there is no reinforcing steel within one bar diameter (ormaximum aggregate size if greater) of the expected conic fracture surface.BS EN 12504-3 has similar requirements. A minimum of four tests is recommendedto provide a result for a given location.The geometric configuration indicated in Figure 4.9 ensures that thefailure surface is conical and at an angle of approximately 31 to the axisof applied tensile force. This is close to the angle of friction of concrete,which is generally assumed to be 37 , and extensive theoretical work hasshown that this produces the most reliable measure of compressive strength.Plasticity theory for concrete using a modified Coulomb’s failure criterionindicates that where the failure angle and friction angle are equal, the

Partially destructive strength tests 9780150 mm cube compressive strength (N/mm 2 )60504020Mean Lok-force= 0.63 × cube comp. strength + 6.0 N/mm 2Light weightaggregates(ref. 34)Natural aggregates:maximum size upto 38 mmMean Lok-force= 0.71 × cube comp. strength + 2.0 N/mm 2(ref. 107)00 20 40 60LOK FORCE (kN)Figure 4.11 Typical Lok-test calibration chart (based on refs 34 and 107).pull-out force is proportional to compressive strength. Finite element analysesof the failure mechanism (108) have indicated that failure is initiated bycrushing, rather than cracking, of the concrete. It is suggested that a narrowsymmetrical band of compressive forces runs between the cast-in disc andthe reaction tube on the surface. Further theoretical (109) and experimental(110) research effort has been devoted to attempts to explain the failuremechanisms, and differing views remain. These primarily concern the relativeimportance of compressive crushing and aggregate interlock effectsfollowing initial circumferential cracking which is generally agreed to befully developed at about 65% of the final pull-out load.Stone and Giza (111) have examined in detail the effect of changes ingeometry and the test assembly and the effect of concrete aggregate propertieson the reliability of the pull-out test. For concrete with cylinderstrengths in the 14–17 N/mm 2 range they have concluded that pull-outforce decreases with increasing apex angle, but that there is no change invariability for apex angles between 54 and 86 , although scatter increasesrapidly for lower angles. As would be expected, pull-out load increases withdepth of embedment and it is confirmed that it is not affected by aggregate

98 Partially destructive strength teststype or size. Variability was, however, shown to be greater for 19 mm aggregatethan for smaller sizes, and mortar specimens showed less variationand lower failure loads than corresponding specimens containing naturalaggregate. Low variability was similarly found for lightweight aggregateconcrete. Carino (103) has also examined a large number of reported resultson a statistical basis.The reliability of the method is reported to be good, with correlationcoefficients for laboratory calibrations of about 0.96 on straight linerelationships, and a corresponding coefficient of variation of about 7%.Comparison with rebound hammer and ultrasonic pulse velocity strengthcalibrations shows that the slope is much steeper, hence this test is muchmore sensitive to strength variations. An important feature of this approachis the independence of the calibration of features such as water/cementratio, curing, cement type and natural aggregate properties (up to 38 mmmaximum size) although Carino (103) has indicated that coarse aggregatefeatures may affect variability of results. Strength calibration is thus moredependable than for most other non-destructive or partially destructivemethods and generalized correlations may be acceptable with predictionaccuracies of the order of ±20%. However, for large projects, it is recommendedthat a specific calibration is developed for the concrete actually tobe used in which case 95% confidence limits of ±10% may be possible.It should also be noted that artificial lightweight aggregates are likely torequire specific calibration as illustrated in Figure 4.11 (34) which showsa reduced value of pull-out force for a particular compressive strengthlevel. The two principal limitations are preplanned usage (although theCapo test, Section 4.2.2.3, overcomes this), and the surface zone natureof the test. The test equipment can be obtained in a convenient briefcasekit form containing all the necessary ancillary items, although the cost isrelatively high.Bickley (112) indicated some time ago that the use of this approachwas also growing in North America, especially for determination of formstripping times, and provides illustrative examples of statistical analysis inrelation to specification criteria. There seem to be few practical problemsassociated with in-situ usage, and an arrangement such as that shown inFigure 4.12 may be convenient in this situation. The technique has beenshown to be particularly suitable for testing at very low concrete strengthsand at early ages as illustrated by the authors’ results in Figure 4.13 andby the Best Practice Guide mentioned above (104) based on the EuropeanConcrete Building Project (113). Other applications include determinationof stressing time in post-tensioned construction, whilst in Denmark theapproach is accepted as a standard in-situ strength determination methodand may form the basis of specification compliance assessment (1). Its usein many parts of the world for in-situ strength monitoring has been considerable,and this is likely to spread in the future.

Partially destructive strength tests 99Figure 4.12 Arrangement for formwork stripping time tests.4.2.1.2 North American pull-out methodsIn the early 1970s Richards published data from tests made using equipmentof his design (103), the basic form of which is shown in Figure 4.14.During subsequent years a number of test programmes were reported inthe United States and Canada using this approach and other comparabletest assemblies. These were sufficient to confirm the potential value of themethod, and an American (114) standard has subsequently been developed.ASTM C900 allows considerable latitude in the details of the test assemblywhile specifying ranges of basic relative dimensions. It is intended that ahydraulic ram is used for load application, which should be at a uniformrate over a period of approximately two minutes. The depth of test may begreater than that of the Lok-test (Section 4.2.1.1) although this equipmentdoes satisfy the requirements of both American and Canadian standards.Indeed, recent reports suggest that use of the commercially available Loktestsystem dominates in these countries in preference to other versions ofthe method.The failure surface will be less precisely defined than with the Lok-testbecause of the range of allowable dimensions, and although little theoretical

100 Partially destructive strength testsFigure 4.13 Low strength Lok-test correlation.Figure 4.14 ‘American’ insert.work has been published relating to this type of insert it is likely that mechanismswill occur which are similar to those for the Lok-test. Presentationof results, however, according to ASTM C900 (114) should be in the formof a pull-out strength f p calculated from the ratio of pull-out force to the

Partially destructive strength tests 101failure surface area. A similar approach is also suggested in an ‘informativeannex’ to the European Standard (106).f p = F Awhere F = force on ramand A = failure surface area.A may be calculated fromA = 4 d 3 + d 2 (4h 2 + d 3 − d 2 2) 1/2where d 2 = diameter of pull-out insert headd 3 = inside diameter of reaction ringh = distance from insert head to the surface.The published numerical data relating to this particular insert type are notextensive, but variability of testing and correlation with strength are likelyto be different according to dimensions used.Applications are obviously limited to preplanned situations and will be similarto those discussed for the Lok-test; similar limitations will also apply.4.2.2 Drilled-hole methodsThese offer the great advantage that use need not be preplanned. Earlyproposals from the USSR involved bolts grouted into the holes, but morerecently two alternative methods have been developed and both have commandedinterest. In 1977 the use of expanding wedge anchor bolts wasproposed by Chabowski and Bryden-Smith (115), working for the BuildingResearch Establishment. Their technique was initially developed for usewith pretensioned high alumina cement concrete beams and is known asthe internal fracture test. The authors have continued to use this test as arapid means of assessing the residual capability of High Alumina Cement(HAC) concrete in structures. This work has subsequently been extendedto Portland cement concretes (116), and the authors have suggested that analternative loading technique offers greater reliability (117). In Denmark,work on the Lok-test (see Section 4.2.1.1) has been extended (1) to producethe Capo test (cut and pull-out) in which an expanding ring is fixed intoan under-reamed groove, producing a similar pull-out configuration to thatused for the Lok-test.Research in Canada and elsewhere has also considered drilled-hole methodsincorporating split sleeve assemblies, as well as reviving the conceptof bolts set into hardened concrete using epoxy. This suggests that, despite

102 Partially destructive strength testspractical problems and high test variability, both of these approaches areworthy of future development. An expanding sleeve device has also beenproposed in the UK (118) called ESCOT. There is little doubt that if areliable drilled-hole pull-out approach could be established, it would beextremely valuable for in-situ concrete strength assessment, especially whenthe concrete mix details are unknown.4.2.2.1 Internal fracture testsThe basic procedures for this method are as follows. A hole is drilled30–35 mm deep into the concrete using a roto-hammer drill with a nominal6 mm bit. The hole is then cleared of dust with an air blower and a 6 mmwedge anchor bolt with expanding sleeve is tapped lightly into the holeuntil the sleeve is 20 mm below the surface (Figure 4.15). Verticality of boltalignment relative to the surface can be checked using a simple slotted template.BS 1881: Part 207 (93) requires a minimum centre-to-centre spacingof 150 mm, and 75 mm edge distance.The bolt is loaded at a standardized rate against a tripod reaction ringof 80 mm diameter with three feet, each 5 mm wide and 25 mm long. Ifnecessary shims may be used to correct for minor bolt misalignments. Afterapplying an initial load to cause the sleeve to expand, the force requiredto produce failure by internal fracture of the concrete is measured. Thiswill be the peak load indicated by the typical load/movement pattern inFigure 4.16. If the load is reduced once this peak has been reached thereis likely to be no visible surface damage and it has been suggested that thebolt can be sawn off. If load application continues beyond the peak, a coneFigure 4.15 Internal fracture test.

Partially destructive strength tests 103Figure 4.16 Typical loading curve.of concrete will be pulled from the surface, often intact, and considerablemaking good may be necessary. It has been found by the authors (117)that the load application method greatly influences the value of pull-outforce required. The rate of load application affects not only the magnitudebut also the variability of the results, and continuous methods yield moreconsistent results than if pauses are involved. Whatever loading method isadopted, it is essential that any calibration curves which are used relatespecifically to the procedures followed. The importance of this is illustratedby comparison of the curves for two specific methods shown in Figure 4.20.In the BRE loading method it is recommended that load is applied througha nut on the greased bolt thread by means of a torquemeter, which is rotatedone half turn in 10 seconds and released before reading, the procedurebeing repeated until a peak is passed. The tripod assembly (Figure 4.17)incorporates a ball race and a facility for automatic alignment with the axisof the anchor bolt to ensure that an axial load is applied with no bendingeffects. Early tests also required a load cell, but subsequently the methodwas developed on the basis of calibrations between measured torque andcompressive strength.Although this loading method is simple to use on both horizontal andvertical surfaces it suffers from two main disadvantages. First, some torqueis inevitably applied to the bolt, depending to some extent on the amountof grease on the thread, and this may reduce the failure load and increasethe scatter obtained from individual results. Second, the torquemeter isrelatively insensitive, and determination of the peak load is hindered by theuse of settling pauses in the loading procedure.

104 Partially destructive strength testsFigure 4.17 Torquemeter loading method.An alternative mechanical loading method has been developed by theauthors (117), which has the advantage of providing a direct pull free oftwisting action. This equipment is shown in Figure 4.18. The use of aproving ring for load measurement is sensitive, and provides a continuousrather than a settled reading, with the result that the variabilities due toload application and measurement are reduced. Loading is provided at asteady rate, without pauses, by rotating the loading handle at the rate ofone revolution every 20 seconds. Calibration charts have been produced forthis loading procedure, which relate compressive strength to direct force,and the variability due to testing using this approach has been shown to belower than for the BRE method.The load transfer mechanisms in these methods are complex, due to theconcentrated localized actions of the expanding sleeve. The location of largeaggregate particles relative to the sleeve will further complicate mattersand affect the distribution of internal stresses. This is partially responsiblefor the high test variability found for the internal fracture test. Thebasic test-assembly dimensions have been determined largely from practicalconsiderations of suitable magnitude of force, and obtaining a depth of

Partially destructive strength tests 105Figure 4.18 Proving ring loading method.test generally to avoid surface carbonation effects while minimizing likelyreinforcement interference. As the name of the method implies, failure isthought to be initiated by internal cracking. Attempts have been madeto represent this theoretically on the basis of an observed average failuredepth of 17 mm which corresponds to a failure half angle of 78 . This isconsiderably greater than the likely angle of friction for the concrete of37 , and application of the modified Coulomb failure criterion (as for theLok-test, Section 4.2.1.1) indicates failure by a combination of sliding and

106 Partially destructive strength testsseparation. This confirms the dependence of the pull-out force upon thetensile strength of the concrete, but in practice the test method at its presentstage of development relies upon empirical calibrations.Tests on cubes which were subsequently crushed have been described fora variety of mixes by both Chabowski (116) and the authors (117). Bothreports indicate a reduction in crushing strength of 150 mm cubes of theorder of 5% as a result of previous internal fracture tests on the cube. Thismust be taken into account when developing a calibration, unless undamagedspecimens are available for comparison. There is also agreement thatfor practical purposes, mix characteristics (cement type, aggregate type,size and proportions) will not affect the pull-out/compressive strength relationshipfor natural aggregates. An upper limit of 20 mm on maximumaggregate size is suggested in view of the small test depth. The authors havealso shown that the variability of results increases with aggregate size, andthat moisture condition and maturity have negligible effects. These featuresrepresent the chief advantage of this approach compared with other nondestructiveor partially destructive methods, although the scatter of resultsis high, as illustrated by Figure 4.19 which represents the averages of sixtests on a cube. This means that a considerable number of specimens arerequired to produce a calibration curve.The effects of precompression, as may be experienced in columns orprestressed construction, have also been examined by Chabowski (116),Figure 4.19 Typical compressive strength/torque calibration (based on ref. 116).

Partially destructive strength tests 107who concludes that there is no clearly defined influence. Although a trendtowards an increase in pull-out force with increasing lateral compressionis indicated, it is suggested that (provided zones of low stress are selected)this effect can be ignored in practice. The authors (117) have reported testson beams in flexure which demonstrate a similar conclusion, although thevariability of results appears to increase with increasing lateral bendingcompressive stress. The presence of direct lateral tensile stress will have asimilar effect, and tests must not be made adjacent to visible cracks. Surfacecarbonation is another effect which both investigators conclude can beneglected in most circumstances. Only in very old concrete where the depthof carbonation approaches the depth of the test will this effect have anyinfluence. The shallowness of test also offers the advantage that reinforcementis unlikely to affect results, but BS 1881: Part 207 (93) requires thatit must be at least one bar diameter, or maximum aggregate size, outsidethe expected conic fracture surface.The influence of the loading method has been indicated above. Resultsfor the torquemeter loading method are generally expressed in the form of acompressive strength/torque relationship, but an average force/torque ratioof 1.15 is reported by Chabowski (116). Comparative tests by the authors(117) between the direct pull and torquemeter methods suggest a correspondingratio of 1.4 which reflects the differences in loading technique.The average relationships for the techniques are compared in Figure 4.20.It must also be pointed out that a calibration obtained by the author usingthe torquemeter suggests compressive strengths up to 20% lower than theBRE calibration. A similar feature has also been indicated by Keiller (119)and Long (120), which cannot be ignored.The variability of test results is high for a variety of reasons. These includethe localized nature of the test, the imprecise load transfer mechanisms andvariations due to drilling. 95% confidence limits on estimated strength of±30% based on the mean of six test results are accepted for the torquemeterload method (93), provided that individual results causing a coefficientof variation of greater than 16% are discarded. The authors (117) haveclaimed a corresponding range of ±20% based on four results for 10 mmaggregates using the direct pull equipment. The average coefficients of variationobserved for cubes of 20 mm maximum aggregate size were 16.5%for torque and 7.0% for direct force, with values 20% lower for 10 mmaggregate.The method can be applied to lightweight concretes (34) although difficultiesmay be encountered with very soft aggregate types. Typical correlationsusing the torquemeter method are compared in Figure 4.21 fromwhich the effects of aggregate type can be clearly seen. The measuredtorque corresponding to a given compressive strength is reduced in comparisonto natural aggregates, but is also significantly affected by the type oflightweight aggregate present. It is also interesting to note that a direct–pull

Figure 4.20 Comparison of calibration curves for natural aggregates (based on refs116 and 117).Figure 4.21 Comparison of calibration curves for lightweight aggregates(torquemeter method) (based on ref. 34).

Partially destructive strength tests 109load method gave much closer agreement between correlations for differentaggregate types (34).The chief advantage of the internal fracture test lies in the ability to usea general strength calibration curve for natural aggregates relating only tothe loading method. Despite the variability, localized surface nature anddamage caused, this may be of particular value in situations where a strengthestimate of in-situ concrete of unknown age or composition is required.This is especially true for slender members with only one exposed surfacewhere cores or other direct techniques are not possible. The accuracy ofstrength estimate will be similar to that obtained by small cores but withconsiderable savings of time, expense and disruption.4.2.2.2 ESCOTThis test was proposed by Domone and Castro (118) in1987. Designed as adrilled-hole method, the test works on an expanding sleeve principle whichcauses internal fracture of the concrete at a depth of just under 20 mmbelow the surface, with a conical failure zone of between 100 and 200 mmdiameter. The test is much simpler than the Capo test, and although loadis applied by a torquemeter, no bearing ring is required as in the internalfracture test with load applied internally in a less concentrated manner.Laboratory correlations with compressive strength were shown to besimilar in nature but better than for the internal fracture test, and comparablein accuracy to the modified ASTM pull-out approach described inSection 4.2.1.2. Despite its potential advantages over the internal fracturetest, the method has not been subsequently developed.4.2.2.3 The Capo testThis has been developed in Denmark (1) as an equivalent to the Lok-testfor situations where use cannot be preplanned. The basic geometry of theLok-test described in Section 4.2.1.1 has been maintained, although thepull-out insert consists of an expanding ring inserted into an undercutgroove. The name is based on the expression ‘cut and pull-out’, and theprocedure consists of drilling a 45 mm deep, 18 mm diameter hole, afterwhich a 25 mm groove is cut at a depth of 25 mm using a portable millingmachine illustrated in Figure 4.22. The expanding ring insert is then placedand expanded in the groove, as shown in Figure 4.23, and conventionalLok-test pulling equipment can be used as described previously. Testingmust continue to pull out the plug of concrete, and the ring may be recovered,recompressed and re-used up to three or four times.Extensive laboratory testing programmes (107) have shown that thebehaviour of this test is effectively identical to the Lok-test and that thestrength calibration and reliability may be regarded as the same. It is claimed

110 Partially destructive strength testsFigure 4.22 Capo test equipment.Figure 4.23 Capo test configuration.that the entire testing operation, including drilling, may be completed inabout ten minutes, and the equipment is available in the form of a comprehensivekit. In Denmark this method has been accepted as equivalent tothe Lok-test and has been used on a number of projects for in-situ strengthdetermination in critical zones (1). The method is also covered by BS 1881:Part 207 and the new BS EN 12504-3. The potential areas of application are

Partially destructive strength tests 111wide and although surface zone effects must be considered, the approachappears to offer the most reliable available indication of in-situ strengthapart from cores. Although equipment costs are high, the damage, time andcost of testing will be considerably less than for cores. Problems may arisefrom the presence of reinforcement within the test zone, and bars must beavoided, but the value of this test is considerable in situations where mixdetails are not known.4.2.2.4 Wood-screw methodA simple pull-out technique utilizing wood-screws has been described byJaegermann (121). This is intended for use to monitor strength developmentin the 5–15 N/mm 2 strength range for formwork stripping purposes inindustrialized buildings.A nail is driven into the surface of the fresh concrete to push aside aggregateparticles, and the screw with a plastic stabilizing ring attached at theappropriate height is inserted until the ring touches the concrete surface. Theunthreaded upper part of the screw is painted to prevent bonding and testscan be made at different depths by using screws of different lengths. A loadis applied to the screw head by means of a proving ring or hydraulic jackingdevice. The principal assumption is that the force required to pull the screwfrom the concrete is dominated by the fine mortar surrounding the screwthreads, and laboratory trials suggest good strength correlations with reasonablerepeatability but further development is needed to facilitate field usage.4.3 Pull-off methodsThis approach has been developed to measure the in-situ tensile strengthof concrete by applying a direct tensile force. The method may also beuseful for measuring bonding of surface repairs (122) and a wide selectionof equipment is commercially available (123) with disk diameters typically50 or 75 mm. Procedures are covered by BS 1881: Part 207 and itshould be noted that the fracture surface will be below the concrete surfaceand will thus leave some surface damage that must be made good. ASTMC1583 (124) also covers this test method for in-situ applications whilst BSEN 1542 (125) uses the technique on laboratory specimens to assess thebond properties of repair materials.Pull-off tests have been described (120), which were developed initiallyin the early 1970s for suspect high alumina concrete beams. A disk is gluedto the concrete surface with an epoxy resin and jacked off to measure theforce necessary to pull a piece of concrete away from the surface. The directtension failure is illustrated in Figure 4.24, and if surface carbonation orskin effects are present these can be avoided by the use of partial coring toan appropriate depth. ‘Limpet’ loading equipment with a 10 kN capacity

112 Partially destructive strength testsFigure 4.24 Pull-off method – surface and partially cored.is commercially available to apply a tensile force through a rod screwedaxially into a 50 mm diameter disk. This equipment (Figure 4.25) bearson the concrete surface adjacent to the test zone and is operated manuallyby steady turning of the handle, with the load presented digitally. Anothercommon type of loading system is by means of a tripod apparatus, with theload applied mechanically (as in Figure 4.26) or hydraulically. Despite widevariations in loading rates and reaction configurations between differentsystems, the authors (126) have concluded that results are unlikely to beaffected provided there is adequate clearance between the disk and reactionpoints. Considerable care is needed in surface preparation of the concreteby sanding and degreasing to ensure good bonding of the adhesive, whichmay need curing for between 1.5 and 24 hours according to material andcircumstances. Difficulties may possibly be encountered with damp surfaces.BS 1881: Part 207 requires that the mean of six valid tests should beused, and that these should be centred at least two disk diameters apart.The stiffness of the disk has been shown to be an important parameterand the limiting thickness/diameter ratio will depend upon the materialused (126). This is illustrated in Figure 4.27 from which it can be seen thatto ensure a uniform stress distribution, and hence maximum failure load,steel disk thickness must be 40% of the diameter whilst for aluminium thisrises to 60%. These experimental findings have been supported by finiteelement analyses.A nominal tensile strength for the concrete is calculated on the basisof the disk diameter, and this may be converted to compressive strengthusing a calibration chart appropriate to the concrete. This calibration will

Partially destructive strength tests 113Figure 4.25 ‘Limpet’ equipment.differ according to whether coring is used or not (119), with cored testsgenerally requiring a lower pull-off force. Partial coring will transfer thefailure surface lower into the body of the concrete, but the depth of coringmay also be critical, as illustrated by Figure 4.28, and should always exceed20 mm. Reinforcing steel clearly must be avoided when partial coring isused. A test coefficient of variation of 7.9% with a range of predicted/actualstrength between 0.85 and 1.25 related to 150 mm Portland cement cubeshas been reported by Long and Murray (120) using the mean of three testresults. A typical calibration curve is illustrated in Figure 4.29, and it isclaimed that factors such as age, aggregate type and size, air entrainment,compressive stress and curing have only marginal influences upon this.Extensive field tests during the construction of a multistorey car park havealso been successfully undertaken (127).BS 1881: Part 207 recommends that a strength correlation should beestablished for the concrete under investigation and that site results fromone location are likely to yield a coefficient of variation of about 10%.Accuracies of strength predictions under laboratory conditions of about

Figure 4.26 ‘Hydrajaws’ tripod equipment.Figure 4.27 Effects of disk type and thickness (based on ref. 126).

Partially destructive strength tests 115Figure 4.28 Effects of partial coring (based on ref. 126).Cube compressive strength (N/mm 2 )40302010specific mix with varying agemeanlower 95%confidence limit00 1.0 2.0 3.0 4.0Puff-off tensile strength (N/mm 2 )Figure 4.29 Typical pull-off/strength correlation for natural aggregate (based onref. 127).±15% (95% confidence limits) are likely. The authors (126) have alsodemonstrated that separate correlations are required for different types oflightweight aggregates, as illustrated in Figure 4.30, and that these aredifferent to those for natural aggregates due to different tensile/compressive

116 Partially destructive strength testsFigure 4.30 Typical strength correlations for lightweight aggregates ( based on ref. 34).strength relationships. It can be noted that pull-off values for lightweightaggregates are higher than for natural aggregates at a given strength level.This test is aimed primarily at unplanned in-situ strength determination.The method is particularly suitable for small-section members, and longtermmonitoring procedures could also be developed involving proof loadtests at intervals on a series of permanent probes. It is also particularlysuited, with the use of partial coring into the base material, for assessmentof bonding strength of repairs as indicated above.This is an area receiving considerable industrial interest and many repairspecifications now require pull-off testing as part of quality control procedures(see reference to US and European Standards above). In such cases itis usual to specify a minimum pull-off stress and it is thus vital that the testprocedures are carefully specified or standardized if this is to be meaningful.A novel friction transfer device has recently been reported (128) in whicha partial core is physically gripped to avoid disk adhesion problems. A torsionalload is then applied by torquemeter to cause a shear failure withinthe core or at an interface between the substrate and an applied repairmaterial.4.4 Break-off methods4.4.1 Norwegian methodJohansen (108) has reported the use of a break-off technique developed inNorway. This is intended primarily as a quality control test, and makes a

Partially destructive strength tests 117direct determination of flexural strength in a plane parallel and at a certaindistance from the concrete surface. A tubular disposable form is insertedinto the fresh concrete, or alternatively a shaped hole can be drilled, toform a slot of the type shown in Figure 4.31. The core left after the removalof the insert is broken off by a transverse force applied at the top surfaceas shown. This force is provided hydraulically using specially developedportable equipment available under the name ‘TNS-Tester’.The ‘break-off’ strength calculated from the results has been shown togive a linear correlation with the modulus of rupture measured on prismspecimens, although values were 30% higher on average. Christiansen et al.(129) have also examined relationships between break-off values and bendingtensile strengths and have shown water/cement ratio, age, curing andcement type to be significant. Values obtained (130) for an airfield pavementcontract have suggested coefficients of variation of 6.4% for laboratorysamples and 12.6% in-situ. Comparable values have also been found onother construction sites (131). It is suggested that the mean of five test resultsshould be used in view of high within-test variation. BS 1881: Part 207 (93)requires a concrete element thickness of at least 100 mm, with a minimumclear spacing or edge distance from the outer face of the groove of four timesthe maximum aggregate size (≮50 mm). Reinforcing bars must obviouslyFigure 4.31 Break-off method.

118 Partially destructive strength testsbe avoided and particular care is needed to ensure that compaction andcuring at prepared test positions are representative of the surrounding bodyof concrete.It is claimed that the method is quick and uncomplicated, taking less thantwo minutes per test. Results are not significantly affected by the surfacecondition or local shrinkage and temperature effects. A correlation withcompressive strength has been developed which covers a wide range of concrete,but this is likely to be less reliable than a tensile strength correlation inview of the factors influencing the tensile/compressive strength relationship.Compressive strength estimates to within ±20% should be possible withthe aid of appropriate calibrations. The method is regarded as especiallysuitable for very young concrete, and although leaving a sizeable damagezone, may gain acceptance as an in-situ quality control test where tensilestrength is important. Although quicker than compression testing of cores,the use of results for strength estimation of old concrete may be unreliableunless a specific calibration relationship is available. Field experience in avariety of situations has been reported by Carlsson et al. (132). Naik (133)has also reviewed field experience and indicated that crushed aggregatesmay give strengths about 10% higher than those of rounded aggregates,and that drilled tests give results 9% higher than when a sleeve is insertedinto the fresh concrete. Nevertheless recent usage worldwide has been verylimited and the relevant ASTM Standard (C1150) was withdrawn in 2002.4.4.2 Stoll tork testThis approach (134) proposed in 1985 was intended to improve upon thevariability encountered with other methods and to permit tests at greaterdepths below the surface than pull-out, pull-off or penetration resistancemethods. A cylindrical cleated spindle of 18 mm thickness and 35 mmdiameter is removably attached to a 19 mm torque bolt at least 51 mm longand cast into the concrete at the required depth. A compressible tape isattached to the periphery except for two radially extending symmetricallyopposed cleat surfaces which bear directly against the concrete mortar. Asmall grating prevents intrusion of large aggregate particles into the mortarcusp which is fractured by application of torque to the bolt by a conventionaltorque wrench.The maximum load is correlated to the compressive strength. The mortarcusp is subjected to a semi-confined compressive stress leading to a compressive/shearfailure. Limited data available at that time indicated reliablelinear relationships with cylinder strength in the range 67–34 N/mm 2 , butaffected by aggregate type, cement replacements and admixtures. Variabilityand accuracy compare favourably with other methods discussed in thischapter, with results based on the average of at least three tests. The principalvalue of the method appears to lie in preplanned monitoring of internal

Partially destructive strength tests 119in-situ strength development. Further investigation of factors influencingstrength correlations is clearly required before the method is used commercially,but there is no readily available evidence of this having happened.4.4.3 Shearing-rib methodThis is a long-established test for precast concrete quality control purposesin the former Soviet Union, which has been described by Leshchinskyet al. (48). A hand-operated hydraulic jack is clamped to a linear elementwhich is at least 170 mm thick and is used to shear-off a corner of theelement. The localized load is applied over a 30 mm width at an angle of18 to the surface and at a distance of 20 mm from the edge. Reinforcingsteel set at normal covers will thus not influence results and a highly stablestrength correlation is claimed. At present, this technique has not becomeestablished elsewhere in the world.

Chapter 5CoresThe examination and compression testing of cores cut from hardenedconcrete is a well-established method, enabling visual inspection of the interiorregions of a member to be coupled with strength estimation. Otherphysical properties which can be measured include density, water absorption,indirect tensile strength and movement characteristics including expansiondue to alkali–aggregate reactions. Cores are also frequently used assamples for chemical analysis following strength testing. In most countriesstandards are available which recommend procedures for cutting, testingand interpretation of results; BS EN 12504-1(135) in the UK, whilst ASTMC42 (136) and ACI 318 (137) are used in the USA. It must be noted howeverthat the above new European Standard offers no guidance on planningor interpretation, although a further document dealing with this is in preparation.Extremely valuable and detailed supplementary information andguidance is also given by Concrete Society Technical Report 11 (36) and itsaddendum, which are related to the former British Standard (BS 1881: Part201 – now withdrawn). A UK National Annex to BS EN 12504-1 is also inpreparation dealing with allowances for voidage, reinforcement, maturityand direction of drilling, and this is likely to reflect the Concrete Societyguidance. The Concrete Society have also published the results of extensivefield experiments aimed at enhancing interpretation in terms of estimatedcube strengths for different cement types, member types and constructionconditions (138). Interpretation is a potentially complex process and Neville(139) has recently reviewed many of the issues involved including samplingand testing planning.5.1 General procedures for core cuttingand testing5.1.1 Core location and sizeCore location will be governed primarily by the basic purpose of the testing,bearing in mind the likely strength distributions within the member,

Cores 121discussed in Chapter 1, related to the expected stress distributions. Whereserviceability assessment is the principal aim, tests should normally be takenat points of likely minimum strength, for example from the top surface atnear midspan for simple beams and slabs, or from any face near the topof lifts for columns or walls. If the member is slender, however, and corecutting may impair future performance, cores should be taken at the nearestnon-critical locations. Aesthetic considerations concerning the appearanceafter coring may also sometimes influence the choice of locations.Alternatively, areas of suspect concrete may have been located by othermethods.If specification compliance determination is the principal aim, the coresshould be located to avoid unrepresentative concrete, and for columns,walls or deep beams will normally be taken horizontally at least 300 mmbelow the top of the lift. If it is necessary to drill vertically downwards, asin slabs, the core must be sufficiently long to pass through unrepresentativeconcrete which may occupy the top 20% of the thickness. In such casesdrilling upwards from the soffit, if this is feasible, may considerably reducethe extent of drilling, but the operation may be more difficult and mayintroduce additional uncertainties relating to the effects of possible tensilecracking. Reinforcement bars passing through a core will increase theuncertainty of strength testing, and should be avoided wherever possible.The use of a covermeter to locate reinforcement prior to cutting is thereforerecommended.Where the core is to be used for compression testing, British and AmericanStandards require that the diameter is at least three times the nominalmaximum aggregate size. In many countries, including the UK, a minimumdiameter of 100 mm is used, with 150 mm preferred, although inAustralia 75 mm is considered to be generally acceptable. In general, theaccuracy decreases as the ratio of aggregate size to core diameter increasesand 100 mm diameter cores should not be used if the maximum aggregatesize exceeds 25 mm, and this should preferably be less than 20 mmfor 75 mm cores. In some circumstances smaller diameters are used, especiallyin small-sized members where large holes would be unacceptable,but the interpretation of results for small cores becomes more complexand is considered separately in Section 5.3. The choice of core diameterwill also be influenced by the length of specimen which is possible.It is generally accepted that cores for compression testing should havea length/diameter ratio of between 1.0 and 2.0, but opinions vary concerningthe optimum value. BS EN 12504-1 (135) recommends a ratioof 2.0 if results are to be related to cylinder strengths or 1.0 for cubestrengths.The Concrete Society (36) suggest that cores should be kept as shortas possible l/d = 10 → 12 for reasons of drilling costs, damage, variabilityalong length, and geometric influences on testing. Although these

122 Corespoints are valid, procedures for relating core strength to cylinder or cubestrength usually involve correction to an equivalent standard cylinder withl/d = 20, and it can be argued that uncertainties of correction factorsare minimized if the core length/diameter ratio is close to 2.0 (140) (seeSection 5.2.2) and this view is supported both by ASTM C42 (136) andNeville (139).The number of cores required will depend upon the reasons for testingand the volume of concrete involved. The likely accuracies of estimatedstrength are discussed in Section 5.2.3, but the number of cores must besufficient to be representative of the concrete under examination as well asprovide a strength estimate of acceptable accuracy as discussed in Chapter 1.ACI 318 (137) requires that at least three cores are always used.5.1.2 DrillingA core is usually cut by means of a rotary cutting tool with diamondbits, as shown in Figure 5.1. The equipment is portable, but it is heavyand must be firmly supported and braced against the concrete to preventrelative movement which will result in a distorted or broken core, anda water supply is also necessary to lubricate the cutter. Vacuum-assistedequipment can be used to obtain a firm attachment for the drilling rigwithout resorting to expansion bolts or cumbersome bracing. Uniformityof pressure is important, so it is essential that drilling is performed by askilled operator. Hand-held equipment is available for cores up to 75 mmdiameter. A cylindrical specimen is obtained, which may contain embeddedreinforcement, and which will usually be removed by breaking off byinsertion of a cold chisel down the side of the core, once a sufficient depthhas been drilled. The core, which will have a rough inner end, may thenbe removed using the drill or tongs, and the hole made good. This is bestachieved either by ramming a dry, low shrinkage concrete into the hole, orby wedging a cast cylinder of suitable size into the hole with cement groutor epoxy resin. It is important that each core is examined at this stage,since if there is insufficient length for testing, or excessive reinforcement orvoids, extra cores must be drilled from adjacent locations. Each core mustbe clearly labelled for identification, with the drilled surface shown, andcross-referenced to a simple sketch of the element drilled. Photographs ofcores are valuable for future reference, especially as confirmation of featuresnoted during visual inspection, and these should be taken as soonas possible after cutting. A typical photograph of this type is shown inFigure 5.2. Cores should be securely wrapped in several layers of ‘clingfilm’and then placed in a labelled polythene bag for return to the testinglaboratory.

Figure 5.1 Core cutting drill.

124 CoresFigure 5.2 Typical core.5.1.3 TestingEach core must be trimmed and the ends either ground or capped beforevisual examination, assessment of voidage, and density determinations.5.1.3.1 Visual examinationAggregate type, size and characteristics should be assessed together withgrading. These are usually most easily seen on a wet surface, but for otherfeatures to be noted, such as aggregate distribution, honeycombing, cracks,defects and drilling damage, a dry surface is preferable. Precise details

Cores 125of the location and size of reinforcement passing through the core mustalso be recorded. The voids should be classified in terms of the excessvoidage by comparison with ‘standard’ photographs of known voidageprovided by Concrete Society Technical Report 11 (36). These referencephotographs are based on the assumption of a fully compacted ‘potential’voidage of 0.5%. This estimated value of excess voidage will be requiredwhen attempting to calculate the potential strength (see Section 5.2.2). If amore detailed description of the voids is required, this should refer to smallvoids (0.5–3 mm), medium voids (3–6 mm) and large voids >6mm withthe term ‘honeycombing’ being used if these are interconnected. It is alsohelpful to describe whether voids are empty, or the nature of their contents,for example white gel from ASR.5.1.3.2 TrimmingTrimming, preferably with a masonry or water-lubricated diamond saw,should give a core of a suitable length with parallel ends which are normalto the axis of the core. If possible, reinforcement and unrepresentativeconcrete should be removed.5.1.3.3 CappingUnless their ends are prepared by grinding, cores should be capped withhigh alumina cement mortar or sulfur–sand mixture to provide parallel endsurfaces normal to the axis of the core. (Other materials should not be usedas they have been shown to give unreliable results.) Caps should be kept asthin as possible, but if the core is hand trimmed they may be up to aboutthe maximum aggregate size at the thickest points.5.1.3.4 Density determinationThis is recommended in all cases, and is best measured by the followingprocedure (36):(i) Measure volume V u of trimmed core by water displacement(ii) Establish density of capping materials D c (iii) Before compressive testing, weigh soaked/surface-dry capped core in airand water to determine gross weight W t and volume V t(iv) If reinforcement is present this should be removed from the concreteafter compression testing, and the weight W s and volume V s determined(v) Calculate saturated density of concrete in the uncapped core fromD a = W t − D c V t − V u − W sV u − V s

126 CoresIf no steel is present, W s and V s are both zero.The value thus obtained may be used, if required, to assess the excessvoidage of the concrete using the relationshipestimated excess voidage = D p − D aD p − 500 × 100%where D p = the potential density based on available values for 28-day-oldcubes of the same mix. And D a is the actual density.5.1.3.5 Compression testingThe standard procedure in the United Kingdom is to test cores in a saturatedcondition, although in the USA (137) dry testing is used if the in-situconcrete is in a dry state. If the core is to be saturated, testing should be notless than two days after capping and immersion in water. The mean diametermust be measured to the nearest 1 mm by caliper, with measurements ontwo axes at quarter- and mid-points along the length of the core, and thecore length also measured to the nearest 1 mm.Compression testing will be carried out at a rate within the range12–24 N/mm 2 min in a suitable testing machine and the mode of failurenoted. If there is cracking of the caps, or separation of cap and core, theresult should be considered as being of doubtful accuracy. Ideally crackingshould be similar all round the circumference of the core, but a diagonalshear crack is considered satisfactory, except in short cores or wherereinforcement or honeycombing is present.5.1.3.6 Other strength tests on coresAlthough compression testing as described above is by far the most commonmethod of testing cores for strength, recent research has indicated thepotential of other methods which are outlined below. Two of these measurethe tensile strength, although neither method is yet fully established. Tensilestrength may also be measured by ‘Brazilian’ splitting tests on cores accordingto ASTM C42 (136). Tests for other properties of the concrete, such aspermeability, alkali–aggregate expansion or air content (Chapters 7, 8 and9) may also be performed on suitably prepared specimens obtained fromcores.Robins (141) has shown that the point load test, which is an acceptedmethod of rock strength classification, may usefully be applied to concretecores. A compressive load is applied across the diameter (Figure 5.3)by means of a manually operated hydraulic jack, with the specimen heldbetween spherically truncated conical platens with a point of 5 mm radius.It has been found that the point load strength index is indirectly related to

Cores 127Figure 5.3 Point load test.the concrete compressive strength, although core size and aggregate typeaffect the relationship. For a given aggregate and core size, the index varieslinearly with cube strength for strengths greater than 20 N/mm 2 . Robins(141) also claims that the testing variability is comparable to that expectedfor conventional core testing. The advantages of this approach are thattrimming and capping are not required and that the testing forces are lower,thus permitting the use of small portable equipment on site at a reducedunit cost.The point load test is essentially a tensile test, and Robins has alsoconfirmed that a simple linear relationship exists between point load indexand flexural strength. This test may thus be particularly useful for sprayedand fibrous concretes (142).In the gas pressure tension test, Clayton (143) demonstrated that appliedgas pressure may be used on cylinders to simulate the effects of uniaxialtensile tests, and that cores may also be used for this purpose. The specimenis inserted into a cylindrical steel jacket with seals at each end, and gaspressure is applied to the bare curved surface. Nitrogen has been found to besafe and convenient. The flow is controlled by a single-stage regulator, anda pressure gauge is used for measurement. Pressure is increased manuallyat a specified rate, until failure occurs by the formation of a single cleavageplane transverse to the axis of the specimen. The two sections are forcedviolently apart and safety precautions are necessary to prevent ejection ofthe fragments from the testing jacket.The method has been developed using 100 mm cylinders, but has beensuccessfully applied to 75 mm cores of high alumina cement concrete whichin some cases had length/diameter ratios of less than 1.0. Although preliminaryevidence suggests that this may provide a reliable method of

128 Coresdetermining in-situ tensile strength, further research is necessary beforeresults can be regarded with confidence. Unfortunately there is no evidenceof recent activity in this area.A third type of ‘strength’ test on cores which has been developed by Chrispet al. (144) utilizes low strain rate compressive load cycling on 72 mm diametercores with a length/diameter ratio of 2.5 to quantify damage in caseswhere deterioration has occurred. Strain data are recorded using a sensitive‘compressometer’, and processed automatically on a microcomputer toyield hysteresis and stiffness characteristics. These can be used to establisha series of parameters of damage caused by deterioration, and cores arenot significantly further damaged by the test, thus allowing them to be subjectedto further testing. Good results have been achieved with this ‘stiffnessdamage’ test applied to concrete affected by alkali–aggregate reactions andthe approach is likely to be extended to other damage mechanisms.5.2 Interpretation of results5.2.1 Factors influencing measured core compressivestrengthThese may be divided into two basic categories according to whether theyare related to concrete characteristics or testing variables.5.2.1.1 Concrete characteristicsThe moisture condition of the core will influence the measured strength –a saturated specimen has a value 10–15% lower than a comparable dryspecimen. It is thus very important that the relative moisture conditions ofcore and in-situ concrete are taken into account in determining actual in-situconcrete strengths. If the core is tested while saturated, comparison withstandard control specimens which are also tested saturated will be morestraightforward but there is evidence (145) that moisture gradients within acore specimen will also tend to influence measured strength. This introducesadditional uncertainties when procedures involving only a few days of eithersoaking or air drying are used since the effects of this conditioning are likelyto penetrate only a small distance below the surface.The curing regime, and hence strength development, of a core and of theparent concrete will be different from the time of cutting. This effect is verydifficult to assess, and in mature concrete may be ignored, but should beconsidered for concrete of less than 28 days old.Voids in the core will reduce the measured strength, and this effect canbe allowed for by measurement of the excess voidage when comparing coreresults with standard control specimens from the point of view of materialspecification compliance. Figure 5.4, based on reference (36), shows the

Cores 129Figure 5.4 Excess voidage corrections (based on ref. 36).influence of this effect. Under normal circumstances an excess voidageof 0.5–1.0% would be expected. Higher values imply increasingly poorercompaction and should certainly be less than 2.5%.5.2.1.2 Testing variablesThese are numerous, and in many cases will have a significant influenceupon measured strength. The most significant factors are outlined below.(i) Length/diameter ratio of core. As the ratio increases, the measuredstrength will decrease due to the effect of specimen shape on stress distributionswhilst under test. Since the standard cylinder used in many partsof the world has a length/diameter ratio of 2.0, this is normally regardedas the datum for computation of results, and the relationship betweenthis and a standard cube is established. Monday and Dhir (140) haveindicated the influence of strength on length/diameter effects and this isconfirmed by Bartlett and MacGregor (146) who also indicate the influenceof moisture conditions. It is claimed that correction factors to an equivalentlength/diameter ratio of 2.0 will move towards 1.0 for soaked coresand as concrete strength increases. The authors have also demonstrated theinfluence of aggregate type when lightweight aggregates are present (34).This issue is widely recognized to be subject to many uncertainties, butthe average values shown in Figure 5.5 are based on the Concrete Societyrecommendations (36). These differ from ASTM (136) suggestions whichrecognize, but do not allow for, strength effects and are also limited tocylinder strengths in the range 13– 41 N/mm 2 .

130 Cores(ii) Diameter of core. The diameter of core may influence the measuredstrength and variability (see Section 5.1.1). Measured concrete strength willgenerally decrease as the specimen size increases; for sizes above 100 mmthis effect will be small, but for smaller sizes this effect may become significant.However, as the diameter decreases, the ratio of cut surface areato volume increases, and hence the possibility of strength reduction due tocutting damage will increase. It is generally accepted that a minimum diameter/maximumaggregate size ratio of 3 is required to make test variabilityacceptable.(iii) Direction of drilling. As a result of layering effects, the measuredstrength of specimen drilled vertically relative to the direction of casting islikely to be greater than that for a horizontally drilled specimen from thesame concrete. Published data on this effect are variable, but an averagedifference of 8% is suggested (36) although there is evidence that this effectmay be influenced by concrete workability (147) and is not found withlightweight aggregate concretes (34). Whereas standard cylinders are testedvertically, cubes will normally be tested at right angles to the plane ofcasting and hence can be related directly to horizontally drilled cores.(iv) Method of capping. Provided that the materials recommended inSection 5.1.3.3 have been used, their strength is greater than that of thecore, and the caps are sound, flat, perpendicular to the axis of the coreand not excessively thick, the influence of capping will be of no practicalsignificance.(v) Reinforcement. Published research results indicate that the reductionin measured strength due to reinforcement may be less than 10%, but thevariables of size, location and bond make it virtually impossible to allowaccurately for this effect. Reinforcement must therefore be avoided whereverpossible, but in cases where it is present the measured core strengthmay be corrected but treated with caution. Recent developments in coringtechnology in Germany (15) have resulted in a drilling machine with anautomatic detection and stop facility before reinforcement is cut. Experienceddrillers will also look at the colour of the cutting fluid. A suddendarkening is often an indication of reinforcement.It is suggested (36) that for a core containing a bar perpendicular tothe axis of the core the following correction factor may be applied to themeasured core strength although it is sometimes recommended that thecore should be disregarded if the correction is greater than 10%:[ (rcorrected strength = measured strength × 10 + 15 · h )] c l

Cores 131where r = bar diameter c = core diameterh = distance of bar axis from nearer end of corel = core length (uncapped).Multiple bars within a core can similarly be allowed for by the expression[corrected strength = measured strength × 10 + 15 ]r · h c · lIf the spacing of two bars is less than the diameter of the larger bar, onlythe bar with the higher value of r · h should be considered.5.2.2 Estimation of cube strengthEstimation of an equivalent cube strength corresponding to a particularcore result must initially account for two main factors. These are(i) The effect of the length/diameter ratio, which requires a correctionfactor, illustrated by Figure 5.5, to be applied to convert the corestrength to an equivalent standard cylinder strength.(ii) Conversion to an equivalent cube strength using an appropriate relationshipbetween the strength of cylinders and cubes.1.0Correction factor0.9ASTMConcrete Societyand BS 18810.81.01.2 1.4 1.6 1.8 2.0Length/diameter (λ)Figure 5.5 Length/diameter ratio influence (based on refs 36 and 136).

132 CoresCorrections for the length/diameter ratio of the core have been discussed inSection 5.2.1.2. Subsequent conversion to a cube strength is usually basedon the generally accepted average relationship that cube strength = 125cylinder strength (for l/d = 20). Monday and Dhir (140) have shown thatthis relationship is a simplification, and that a more reliable conversion canbe obtained fromcube strength = Af cy − Bf 2cywhere f cy is the strength of a core with l/d = 20, and constants A = 15and B = 0007 tentatively.Within the range of 20–50 N/mm 2 cylinder strengths this produces valueswithin 10% of those given by the average factor 1.25, but the discrepanciesincrease for lower- and higher-strength concretes. Such discrepancieswill, however, be partially offset for cores with l/d close to 1.0 by theerrors resulting from the use of l/d ratio correction factors which are notstrength-related. Particular care will be needed to take account of this issuewhen dealing with cores of high-strength concrete, which is increasinglybeing used worldwide.The Concrete Society (36) recommends a procedure incorporating thecorrection factors of Figure 5.5, coupled with an allowance of 6% strengthdifferential between a core with a cut surface relative to a cast cylinder.A strength reduction of 15% is also incorporated to allow for the weakertop surface zone of a corresponding cast cylinder, before conversion to anequivalent cube strength by the multiplication factor of 1.25. An 8% differencebetween vertical and horizontally drilled cores is also incorporatedwith the resulting expressions emerging.Horizontally drilled core:estimated in-situ cube strength =Vertically drilled core:estimated in-situ cube strength =25f 15 + 1/23f 15 + 1/where f is the measured strength of a core with length/diameter = .It is interesting to note that, using these expressions, the strength of a horizontallydrilled core of length/diameter = 1 will be the same as theestimated cube strength and is consistent with the discussion in Section 5.1.2above. The cube strengths evaluated in this way will be estimates of theactual in-situ strength of the concrete in a wet condition and may underestimatethe strength of the dry concrete by 10–15%.

Cores 133The strength differences between in-situ concrete and standard specimenshave been fully discussed in Chapter 1. An average recommended relationshipis that the ‘potential’ strength of a standard specimen made from aparticular mix is about 30% higher than the actual ‘fully compacted’ insitustrength (36). If this value is used to estimate a potential strength forcomparison with specifications, the uncertainty of the relationship must beremembered. Appendix 3 of the Concrete Society Report offers detailedguidance relating to curing history and their recent experiments seek to considerthe effects of cement and member type (138), but potential strengthestimations are increasingly unpopular due to the difficulties of accountingfor all variable factors.The expressions for cube strength will change as follows:Horizontally drilled core:estimated potential cube strength =Vertically drilled core:estimated potential cube strength =325f 15 + 1/30f 15 + 1/A worked example of evaluation of core results using the Concrete Societyrecommendations is given in Appendix C of this book.ACI 318 (137) suggests that an average in-situ strength of at least 85%the minimum specified value is adequate, and that cores may be tested afterair-drying for 7 days if the structure is to be dry. This is based on equivalentcylinder strengths derived from ASTM C42 (136) factors.The effect of the calculation method can be considerable, as illustratedin Figure 5.6, and this emphasizes the importance of agreement between allparties of the method to be used in advance of the testing.5.2.3 Reliability, limitations and applicationsThe likely coefficient of variation due to testing is about 6% for carefullycut and tested cores, which can be compared with a corresponding valueof 3% for cubes. The difference is largely caused by the effects of cutting,especially since cut aggregate particles are only partially embedded in thecore and may not make a full contribution during testing. It is claimed thatthe likely 95% confidence limits on actual strength prediction for a singlecore are ±12% when the Concrete Society calculation procedures (36) areadopted. It follows that for a group of n cores, the 95% confidence limitson estimated actual in-situ strengths are ±12/ √ n% (see also Section 1.6.3).Where the ‘potential’ strength of the concrete is to be assessed, a minimum

134 Cores1.6Estimated cube strength/measured core strength1.51.41.31.21.11.0Potential cube strength – horizontal (36)Potential cube strength – vertical (36)ASTM × 1.25 (136)In-situ cube strength – horizontal (36)In-situ cube strength – vertical (36)0.91.0 1.2 1.41.6 1.8 2.0Length/diameter (λ)Figure 5.6 Effect of calculation method (based on refs 36 and 136).of four cores is required and an accuracy of better than ±15% cannot beexpected. This can only be achieved if great care is taken to ensure that theconcrete tested is representative, by careful location and preparation of thespecimens. Uncertainties caused by reinforcement, compaction or curingmay lead to an accuracy as low as ±30%.Examination of Figure 5.6 shows the differences between the results forin-situ and potential strength computed by the methods currently in use inthe UK. Cube strengths derived from ASTM C42 (136) procedures coupledwith an average cube/cylinder factor of 1.25 are also indicated and it willbe clear that results computed in this way are liable to overestimate theactual strength by up to 16%. The Concrete Society method makes detailedallowance for the many variable factors influencing core results, and willprovide the more reliable estimates of equivalent cube strengths.Damage caused by drilling may be particularly significant for old brittleconcretes, where internal cracking of the core may be aggravated by theloss of the confining effect of the surrounding body of concrete. Difficultiesassociated with core testing of concrete which has been damaged by alkali–aggregate reactions, have similarly been identified (148). Tests on cores

Cores 135from flexurally cracked tensile zones must be regarded as unreliable, whilstYip (149) has demonstrated that compressive load history of the concretebefore coring may lead to reductions in measured strength of up to 30% dueto internal microcracking. This latter effect may be quite significant evenat relatively low stress levels and adds to interpretation uncertainties. Theestimated cube strengths obtained from core compression tests may tendto underestimate the true in-situ capacity in all these situations. Strengthchanges with age may also be considered when interpreting core results, butany allowances must be carefully considered as discussed in Section 1.5.2.The principal limitations of core testing are those of cost, inconvenienceand damage, and the localized nature of the results. It is strongly recommendedthat core testing is used in conjunction with some other form oftesting which is less tedious and less destructive. The aim is to providedata on relative strengths within the body of the concrete under test. Thesize of core needed for reliable strength testing can pose a serious practicalproblem; ‘small’ cores may be worthy of consideration with slendermembers. It may also be appropriate to consider using a larger numberof ‘small’-diameter cores to obtain an improved spread of test locationswhere large volumes of concrete are involved. The cutting effort for three50 mm cores may be as low as one-third of that for one 150 mm specimen.A comparable overall strength accuracy may be expected (see Section 5.3.2)provided that maximum aggregate size is less than 17 mm. Where cores areused for other purposes, it will often be possible to use a ‘small’ diameterwith considerable savings of cost, inconvenience and damage.Apart from physical testing, cores often provide the simplest method ofobtaining a sample of the in-situ concrete for a variety of purposes, butcare must be taken that the effects of drilling, including heat generated byfriction, or the presence of water, do not distort the subsequent results. Asample taken from the centre of a core may conveniently overcome thisproblem. Chemical analysis can often be performed on the remains of acrushed core, or specimens may be taken specifically for that purpose.Visual inspection of the interior of the concrete may be extremely valuableboth for the assessment of compaction and workmanship, and for obtainingbasic data about concrete for which no records are available. In cases wherestructural assessments of old structures are required, cores may also provevaluable in confirming covermeter results concerning the location and sizeof reinforcement.5.3 Small coresAlthough standards normally require cores to have a minimum diameterof 100 mm for compressive strength testing, cores of smaller diameteroffer considerable advantages in terms of reduced cutting effort, time anddamage. For applications such as visual inspection, density or voidage

136 Coresdetermination, reinforcement location or chemical testing, these savingsmay be valuable. However, the reliability of small diameter cores for compressiontesting is lower than for ‘normal’ specimens. The many factorswhich affect normal core results may also be expected to influence smallcores, but the extent of these factors may vary and other effects which arenormally unimportant may become significant.5.3.1 Influence of specimen sizeIt is well established that measured concrete strength usually increases asthe size of the test specimen decreases, and that results tend to be morevariable. This latter effect has been shown to be particularly true for corespecimens since the ratio of cut surface area to volume increases as diameterdecreases and hence the potential influence of drilling damage is increased.Also, the ratio of aggregate size to core diameter is increased and maypossibly exceed the generally recognized acceptable limit of 1:3. It is alsowell established that concrete strength is a further factor that may influencethe behaviour of a core. These various factors are interrelated and difficultto isolate. For example, increased strength due to small specimen size maybe offset by a reduction due to cutting effects. Ahmed (150) has suggestedthat measured strength reduces with diameter in the range 150–175 mm.The most common diameter for small cores is 40–50 mm. The authorshave reported extensive laboratory tests to investigate the behaviour of44 mm specimens (151), in which a total of 23 mixes were used, rangingfrom 10 to 82 N/mm 2 with 10 and 20 mm gravel aggregates, and cores werecut from 500 × 100 × 100 mm laboratory cast prism specimens to providea range of length/diameter ratios. Some key findings are considered below.5.3.1.1 Length/diameter ratioThe average relationship for length/diameter effects on 44 mm cores (151)is shown in Figure 5.7, which compares the relationships discussed inSection 5.2.1.2 for normal cores. This relationship has found to be independentof drilling orientation, aggregate size and cement type, for practicalpurposes, although the scatter of results is high, since each point inFigure 5.7 represents the average of four similar cores. It will be seen that thecorrection factor for length/diameter ratio is reasonably close to the ConcreteSociety recommendation (36) for larger cores. Lightweight concretesare likely to have values closer to 1.0 (34).5.3.1.2 Variability of resultsNo significant change of variability was found between the extremes oflength/diameter ratio for either aggregate size, and the average coefficient

Cores 1371.0Ratio of measured core strengths (K)0.90.8ASTM (136)Conc. Soc. (36)K = 0.54 + 0.23λ0.71.01.2 1.4 1.6 1.8 2.0Length/diameter (λ)Figure 5.7 Length/diameter ratio for small cores (based on refs 36, 136 and 151).of variation of 8% was also independent of orientation. However, takingaccount of concrete variability as indicated by control cubes, it is clear that20 mm aggregate cores show a higher variability due to cutting and testingthan 10 mm aggregates (151). The range of coefficients of variation ofstrength for groups of similar cores was large, and made identification of theeffects of other variables impossible to assess. Bowman (152) has reporteda coefficient of variation of 28.9% for 50 mm cores from in-situ concreteon a site in Hong Kong compared with a value of 19.5% for corresponding150 mm cores from the same concrete. Swamy and Al-Hamed havealso suggested that variability reduces as strength increases (153), whilstlightweight aggregate concretes may also be less variable (34).5.3.1.3 Measured strengthBased on the authors’ tests (127) the factors required to convert the measuredcore strength (after correction to l/d = 20) to an equivalent 100 mmcube strength are given in Table 5.1. If an equivalent 150 mm cube strengthis required, these values may be reduced by 4%.

138 CoresTable 5.1 Cube/corrected core conversion factors for 44 mm cores with = 20 (basedon ref. 151)Core orientationMaximum aggregate sizeVerticalHorizontalConversion factor to100 mm cube95% confidence limits onpredicted cube strength(4 cores)Conversion factor to100 mm cube95% confidence limits onpredicted cube strength(4 cores)10 mm 20 mm Combined1.05 1.25 1.15±17% ±23% ±23%1.14 1.22 1.17±15% ±17% ±17%It can be seen that for 10 mm aggregates, the vertically drilled cores areapproximately 8% stronger than comparable horizontally drilled specimensrelative to cubes. This is as anticipated for larger specimens, but the measuredstrengths are approximately 10% stronger than expected from theConcrete Society recommendations (36), resulting in a lower correctionfactor to obtain an equivalent cube strength.With 20 mm maximum aggregate, however, the cores were considerablyweaker relative to cubes, confirming the influence of the aggregate size/corediameter ratio discussed above. In this case the orientation effect could not bedetected. It is suggested that 10 mm and 20 mm aggregate concrete should betreated separately when converting 44 mm cores to equivalent cube strength.If this is done, the 95% confidence limits on the average of the results of groupsof four cores of this size under laboratory conditions are unlikely to be betterthan the values given in Table 5.1. These may be approximated by ±36/ √ n%when n is the number of cores in the group. Bowman’s reported results (152)also show a 7% higher strength for 50 mm cores when compared with 150 mmcores, but the aggregate size is not indicated. The Concrete Society (36), however,suggests that strength differences between ‘large’ and ‘small’ cores arenegligible and recommend the use of the formulae in Section 5.2.2 for cores of50 mm diameter and greater.5.3.2 Reliability, limitations and applicationsThe reliability of compressive tests on small diameter cores is known to beless than for ‘normal’ specimens, and the authors have suggested a factor of

Cores 1393 applied to the 95% confidence limits of predicted actual cube strengthsunder laboratory conditions. This gives a value of ±36/ √ n% for n coreswith an aggregate size/diameter ratio of less than 1:3. But if the ratio ofaggregate size/diameter is greater than 1:3, this accuracy is likely to decreaseand may be as low as ±50/ √ n%. Site cutting difficulties may further reduceaccuracy. All the procedures described in Section 5.1 concerning location,drilling and testing must be followed, just as for larger cores, and the effectsof excess voidage and moisture accounted for as described in Section 5.2.Small cores containing reinforcement should not be tested. Particular caremust be taken to ensure that the core is representative of the mass ofthe concrete, and this is particularly critical in slabs drilled from the topsurface in view of the reduced drilling depth required for a small core. BSEN 12504: Part 1 (135) does not refer to particular concrete densities, butASTM C42 (136) specifically includes lightweight concrete in the range1600–1920 kg/m 3 , as well as normal-weight concrete.There is no doubt that for applications other than compressive strengthtesting, small cores offer many economical and practical advantages comparedwith larger specimens. These applications include visual examination(including materials and mix details, compaction, reinforcement locationand sizing); density determination; other physical tests, including point loador gas pressure tests; and chemical testing. For compressive strength testing,the chief limitation is variability of results and consequent lack of accuracyof strength prediction, unless many more specimens are taken than wouldnormally be necessary. At least three times the number of ‘standard’ cores isrequired to give comparable accuracy, but it can be argued that this wouldstill require less drilling in many instances and permits a wider spread ofsample location. It is clear that considerable differences of predicted cubestrength will arise from use of the various calculation methods, and as forlarger cores it is essential that agreement is reached between all parties,before testing, about the method to be used.Bowman (152) has described a successful approach in which 50 mm coreswere used for strength tests on cast in-place piles because of their cheapnessand ease of cutting, but were backed up by 150 mm cores where results wereon the borderline of the specification. Another common situation in whichsmall cores may be necessary for strength testing is when the slendernessof the member does not permit a larger diameter from the point of viewof continued serviceability or adequate length/diameter ratio >10. Thiswill apply especially to prestressed concrete members. Although in suchcases small diameters are inevitable it is essential that the engineer fullyappreciates the limitations of accuracy that he may expect. It may be thatsome other non-destructive approach will yield comparable accuracies ofstrength prediction, according to the availability of calibrations, with lessexpense, time and damage.

Chapter 6Load testing and monitoringWhere member strength cannot be adequately determined from the resultsof in-situ materials tests, load testing may be necessary. The expense anddisruption of this operation may be offset by the psychological benefits of apositive demonstration of structural capacity which may be more convincingto clients than detailed calculations. In most cases where load tests are used,the main purpose will be proof of structural adequacy, and so tests will beconcentrated on suspect or critical locations. Static tests are most commonbut where variable loading dominates, dynamic testing may be necessary.Load testing may be divided into two main categories:(i) In-situ testing, generally non-destructive(ii) Tests on members removed from a structure, which will generally bedestructive.The choice of method will depend on circumstances, but members willnormally only be removed from a structure if in-situ testing is impracticable,or if a demonstration of ultimate strength rather than serviceability isrequired. Ultimate strength capacity may sometimes be used as a calibrationfor other forms of testing if large numbers of similar members are inquestion.Monitoring of structural behaviour under service conditions is an importantaspect of testing which has received increased attention in recent yearsdue to the growing number of older structures which are causing concernas a result of deterioration. This is considered in Section 6.2 and many ofthe measurement techniques used for in-situ load testing and monitoringmay also be useful for ultimate load test monitoring. There is also a growingtrend towards monitoring the performance of older structures such asbridges during demolition to increase understanding of structural behaviourand the effects of deterioration (154,155). Strain measurement techniquesare described in Section 6.3, and more specialized methods such as dynamicresponse and acoustic emission are included in Chapter 8.

6.1 In-situ load testingLoad testing and monitoring 141The principal aim will generally be to demonstrate satisfactory performanceunder an overload above the design working value. This is usually judgedby measurement of deflections under this load, which may be sustainedfor a specified period. The need may arise from doubts about the qualityof construction or design, or where some damage has occurred, and theapproach is particularly valuable where public confidence is involved. InSwitzerland, for example, static load tests on bridges are an established componentof acceptance criteria and useful information concerning testing anddeflection measurement procedures has been given by Ladner (156,157).In other circumstances the test may be intended to establish the behaviourof a structure whose analysis is impossible for a variety of reasons. In thiscase strain measurements will also be necessary to establish load paths incomplex structures. Views on detailed test procedures and requirementsvary widely, but some commonly adopted methods are described in the followingsections, together with suitable loading and monitoring techniques.Further guidance on general principles and basic procedures is provided bythe Institution of Structural Engineers (6), whilst simple guidelines for staticload tests on building structures have been given by Moss and Currie (158).Jones and Oliver (159) have also discussed some practical aspects of loadtesting, and Garas et al. (160) describe a number of more complex investigations.Issues concerning the load testing of bridges have been illustratedby Lindsell (161) whilst the philosophy of instrumentation of structureshas also been considered by Menzies et al. (162). Practical difficulties ofaccess and restraint will influence the preparatory work required, but in allcircumstances it is essential to provide adequate safety measures to cater forthe possible collapse of the member under test. The test loads will normallybe applied twice, with the first cycle used for ‘bedding-in’ purposes.6.1.1 Testing proceduresIn-situ load tests should not be performed before the concrete is 28 days oldunless there is evidence that the characteristic strength has been reached.ACI 318 (137) requires a minimum age of 56 days. Preliminary work isalways necessary and must ensure safety in the event of a collapse undertest, and that the full calculated load is carried by the members actuallyunder test.The selection of specific members or portions of a structure to be testedwill depend upon general features of convenience, as well as the relativeimportance of strength and expected load effects at various locations. Attentionmust also be given to the parts of the structure supporting the testmember. Selection of members may often be assisted by non-destructive

142 Load testing and monitoringmethods coupled with visual inspection to locate the weakest zones or elements.Dynamic response approaches described in Chapter 8 may be usefulin this respect.6.1.1.1 Preliminary workScaffolding must be provided to support at least twice the total load fromany members liable to collapse together with the test load. This should beset to ‘catch’ falling members after a minimum drop but at the same timeshould not interfere with expected deflections. Especial care must be takento ensure that parts of the structure supporting such scaffolding are notoverloaded in the event of a collapse under test, and that safeguards forunexpected failure modes (such as shear at supports) are provided.The problem of ensuring that members under test are actually subjected tothe assumed test load is often difficult, due to load-sharing effects. This maybe a particular problem with floors or roofs supported by beams which spanin one direction only. Even non-structural elements such as roofing boardsmay distribute loads between otherwise independent members, and in compositeconstruction the effect becomes even greater. Whenever possible themember under test should be isolated from the surrounding structure. Thismay be achieved by saw cutting, although this is an expensive, tedious andmessy operation. There will be many situations where this is not feasiblefrom the point of view of reinstatement of the structure after test, or due topracticalities of load application. In such cases, test loads must be appliedover a sufficiently large part of the structure to ensure that the critical memberscarry the required load. Load sharing characteristics are very difficultto assess in practice, but it is recommended (163) that in the case of beamswith infill blocks and screed (Figure 6.1) the loaded width must be equalto at least the span to ensure that the central member is correctly loaded. Itwill be clear that this may lead to the need to provide very large test loads.Figure 6.1 Beam and pot construction.

Load testing and monitoring 143It will frequently be more convenient to concentrate loading above themember or group of members under examination, and to monitor relativedeflections between this and all other adjacent members within acorresponding width. This will enable the proportion of load transferredaway from the test member to be estimated, and the applied load canthen be increased accordingly. It must be recognized that this increase maybe between two and four times, and the shear capacity must be carefullychecked.Moss (164) has provided useful data for beam and block floors whichincludes thermal loading effects together with the influence of groutingand different screed types. Guidelines for assessing load-sharing effects,calculation of increased test loads to compensate and reductions of maximumallowable deflection criteria are all provided. Precautions must also betaken to ensure that members under test are not inadvertently supported bynon-structural elements, such as partitions or services, although permanentfinishes on the member need not be removed. Provision of a constant datumfor deflection measurements is essential and must also be considered whencarrying out preliminary work.6.1.1.2 Test loadsViews on test loads, which should always be added and removed incrementally,vary considerably. BS 8110 (165) requires that the total loadcarried should not be less than the sum of the characteristic design deadand imposed loads, but should normally be the greater of:design dead load + 125 design imposed loador 1125 design dead load + design imposed load(If any final dead load is not in position, compensating loads should beadded.)The test load is applied at least twice, with at least one hour betweentests and with 5-minute settling time after the application of each incrementbefore recording measurements. A third loading, sustained for 24 hours,may be useful in some cases. Performance is based initially on the acceptabilityof measured deflection and cracking in terms of the design requirementscoupled with examination for unexpected defects. If significant deflectionsoccur, the deflection recovery rates after removal of loading should also beexamined. This deflection limit is not specified, but a value of 40l 2 /h mmis sometimes used where l is the effective span in metres and h the overalldepth in mm. Percentage recovery after the second test should not be lessthan that after the first load cycle, nor less than 75% for reinforced orpartially prestressed concrete. Class 1 and 2 prestressed concrete membersmust satisfy a corresponding recovery limit of 85%.

144 Load testing and monitoringACI 318 (137) has similar provisions but with a test load sustained for24 hours such thattotal load = 085 14 dead load + 17 imposed loadAny shortfall of dead load should be made up 48 hours before the teststarts, and the maximum acceptable deflection under test loads is given by:deflection limit =effective span 220 000 × member depth inchesIf this limit is exceeded, recovery must be checked.The recovery limit for prestressed concrete is 80% but this may notbe retested, whilst reinforced concrete failing to meet the 75% recoverycriterion may be retested after 72 hours unloaded, but must achieve 80%recovery of deflections caused by the second test load.The ACI requirements are 15–20% more stringent than BS 8110 interms of test load, and it is felt by many engineers that a total overload of only12.5% is inadequate when certifying the long-term safety of a structure. TheInstitution of Structural Engineers (163) recognize this in recommendingtotal load = 125 dead load + imposed loadwhen testing high alumina cement concrete structures. In such situations,where future deterioration is predicted, an even higher load may be justified.Lee (166) has proposed that a load of 1.5 (dead load + imposed load)should be adopted in all cases, but with greater emphasis on recording andanalysing the members’ response by means of load/deflection plots. Figure 6.2shows a typical plot for an under-reinforced beam; experience would berequired to recognize impending failure in order to stop load application.6.1.2 Load application techniquesThese are governed almost entirely by the practicalities of providing anadequate load as cheaply as possible at locations which are often difficultto access. The rate of application and distribution of the load must becontrolled, and the magnitude must be easily assessed.Water, bricks, bags of cement, sandbags and steel weights are amongstthe materials which may be used and the choice will depend upon the natureand magnitude of load required as well as the availability of materials andease of access. Care must be taken to avoid arching of the load as deflectionsincrease, and also to avoid unintended loads, such as rainwater, or thosedue to moisture changes of the loading material. In most cases a load whichis uniform along the member length is required, but frequently this must be

Load testing and monitoring 145Figure 6.2 Typical load deflection curve for under-reinforced beam.concentrated over a relatively narrow strip above the member under test.Steel weights, bricks or bags of known weight are best in this situation, andif the test member has been isolated it will probably be necessary to providea platform clear of the adjacent structure (Figure 6.3). Figure 6.4 shows anFigure 6.3 Test load concentrated on beam.

146 Load testing and monitoringFigure 6.4 Test load arrangement for purlins.alternative arrangement which may sometimes be more convenient for lightroof purlings.When loading is to be spread over a larger area, water may be the mostappropriate method of providing the load. Slabs may be ponded by providingsuitable containing walls and waterproofing, although care must betaken to allow for cambers or sags in calculating the loads. The effect maybe reduced by baffling to create separate pools, but the likelihood of damageto finishes by leakage is high whenever ponding is used. An alternative toponding is to provide containers such as plastic bins appropriately located,which can then be filled by hose to predetermined depths. Water is particularlyuseful in locations with limited space or difficult access, becausestorage and labour requirements will be reduced. Figures 6.5–6.8 showsome typical test loads applied to slabs and beams in ‘building’ structures.Test loads for bridges may often be conveniently provided by a suitabledistribution of loaded wagons of known weight, such as water-filled truckmixers.For safety reasons, personnel working in a test load area must be restrictedto those essential for load application and taking of measurements. Loadswill always be applied in predetermined increments and in a way whichwill cause as little lack of symmetry or uniformity as possible. Similar precautionsshould be taken during unloading, and particular care is necessary

Figure 6.5 Test load on roof slab using bricks (photograph by courtesy of TysonsContractors Ltd).Figure 6.6 Test load on floor slab using steel weights (photograph by courtesy ofG.B.G. Structural Services).

Figure 6.7 Test load on isolated beam using bricks (photograph by courtesy ofProfessor F. Sawko).Figure 6.8 Test load on roof slab using ponded water (photograph by courtesy ofG.B.G. Structural Services).

Load testing and monitoring 149to ensure that test load storage areas are not inadvertently overloaded.Deflection gauges must be carefully observed throughout the loading cycle,and if there are signs of deflections increasing with time under constantload, further loading should be stopped and the load reduced as quicklyas possible. The potential speed of load removal is thus an importantsafety consideration, and ‘bulk’ loads which rely heavily on either manualor mechanical labour suffer the disadvantage of being relatively slowto handle. Water may be dispersed quickly if ponded, by the provision of‘knock-out’ areas in the containing dyke, but the resultant damage to finishesmay be considerable. Non-gravity loading offers advantages of greatercontrol, which can also be effected at a distance from the immediate testarea, but is usually more expensive, and is restricted to use with load testsof a specialized or complex nature. Hydraulic systems may employ soilanchors, ballast or shoring to other parts of the structure to provide areaction for jacking. The loading may be rapidly manipulated, and offersthe advantage of simple cycling if required. In cases where a horizontaltest load is required this approach may also be useful. Another apparentlysuccessful technique, described by Guedelhoefer (167), involves theapplication of a vacuum. Polythene-lined partition walls and seals mustbe constructed under the test area so that a vacuum can be drawn thereby suction pump. This would be particularly suitable for slabs which cannotbe loaded from above, or when a normal load is required on curvedor sloping test surfaces. A maximum field-test pressure of 192kN/m 2 isclaimed.6.1.3 Measurement and interpretationThe measurement techniques associated with simple in-situ load tests areusually very straightforward, and are restricted to determination of deflectionsand possibly crack widths. Occasionally, more detailed results concerningstrain and stress distributions will be required from a load test, orit may be necessary to monitor long-term behaviour of a structure underworking conditions. The measurement techniques used here will be morecomplex, and are described separately below.Basic in-situ load tests are based on deflection measurements, and thesewill normally be made by mechanical dial gauges which must be clampedto an independent rigid support. If scaffolding is used for this purpose, caremust be taken to ensure that readings are not disturbed when the weightof the person taking the readings comes on to the scaffold. A system usingmeasurements on weights suspended from the test element is shown inFigure 6.9. Dial gauges are often preferred to electronic or electric displacementtransducers because a quick visual assessment of the progression of aload test is essential. However, use of a combined dial gauge/displacement

150 Load testing and monitoringFigure 6.9 Measurements on suspended weights (photograph by courtesy of G.B.G.Structural Services).transducer (Figure 6.10) will enable a visual on-site capability together witha complete data logging of the load test displacements for later retrieval, processingand presentation. Gauges will normally be located at midspan and1/4 points (Figure 6.11) to check symmetry of behaviour. If the memberis less than 150 mm in width, one gauge located on the axis at each pointshould be adequate, but pairs of gauges as in Figure 6.11 should be used forwider members. The selection of gauge size will be based on the expectedtravel, and although gauges can be reset during loading this is not recommended.The gauges must be set so that they can be easily read with aminimum of risk to personnel and so that the chance of disturbance duringthe test is small. Telescopes may often be convenient for this purpose. Readingsshould be taken at all incremental stages throughout the test cycles

Load testing and monitoring 151Figure 6.10 Dial gauge/displacement transducer with portable battery-poweredlogger.described in Section 6.1.1.2 and temperatures noted at each stage. Measurementaccuracy of ±01 mm over a range of 6–50 mm is generally possiblewith dial gauges, and where sustained loads are involved it is prudent tohave a back-up measurement system in case the gauges are accidentallydisturbed. This will generally be less sensitive, and levelling relative to somesuitable datum is probably the simplest method in most cases.

152 Load testing and monitoringFigure 6.11 Gauge location.If the expected deflections are larger than can be accommodated bymechanical gauges, standard surveying techniques can be used in whichscales are attached to the test element and monitored by a level. Thishas safety advantages, but the instrument must not be supported off thestructure under test, which may cause difficulties. The accuracy expectedwill vary widely, but under normal conditions ±15 mm may be possible(167). Shickert (15) has also described the use of laser holography forremote measurement of deflections.Crack-width measurements may occasionally be required, and thesewill normally be made with a hand-held illuminated optical microscope(Figure 6.12). This is powered by a battery and is held against the concretesurface over the crack. The surface is illuminated by a small internal lightbulb and the magnified crack widths may be measured directly by comparisonwith an internal graduated scale which is visible through the eyepiece.A simple unmagnified comparator scale can also be used (Figure 6.13) toassist in the estimation of crack widths.Cracks present or developing in the course of a test should be traced onthe surface with a pencil or coloured pen, with the tip marked at each loadstage. Crack-width readings should also be taken at each load stage at fixedlocations on appropriate cracks. Visual identification of crack developmentmay often be assisted by a surface coating of whitewash.Interpretation of the results will often be a straightforward comparison ofobserved deflections or crack widths with limiting values which have beenagreed previously between the parties concerned. Recommendations havebeen outlined in Section 6.1.1. The effects of load sharing will normallyhave been accounted for in the determination of the test load, so that theprincipal factor for which allowance must be made is temperature. Thismay cause significant changes of stress distributions within a member orstructure between winter and summer, and differentials of temperature

Load testing and monitoring 153Figure 6.12 Crack microscope.across a member such as a roof beam may cause considerable deflectionchanges. Sometimes it may be possible to compensate for these by theestablishment of a ‘footprint’ of movement for the structure for a range oftemperatures (158).As indicated previously, examination of the load/deflection plot can yieldvaluable information about the behaviour of the test member. A typical plotfor a beam is given in Figure 6.14, in which the effects of creep during theperiod of sustained load and recovery can be seen. Since full recovery is notrequired instantaneously, some non-linearity of behaviour under sustained

Figure 6.13 Crack width measurement scale.

Load testing and monitoring 155Figure 6.14 Typical load test result plot.load is to be expected, but any marked non-linearity during the period ofload application, other than that attributed to bedding-in and breakdownof non-structural finishes, must be taken as a sign that failure may beapproaching. Comparison with Figure 6.2 may be useful, but if the memberis over-reinforced, the warning of failure may be small, and experience isrequired to attempt to assess the reserve of strength.6.1.4 Reliability, limitations and applicationsThe reliability of in-situ load tests depends largely upon satisfactory preparatorywork to ensure freedom from unintended restraints, accurate loadapplication to the test member, provision of an accurate datum for deflectionmeasurements and careful allowance for temperature effects. If theserequirements are met, short-term tests should provide a reliable indicationof the behaviour of the member or structure under the test load. This will begreatly enhanced by the examination of load/deflection plots as well as specificdeflection values. Finishes or end restraints, which are not allowed forin design, will frequently cause lower deflections than calculations wouldindicate, but providing these features are permanent this is not important.

156 Load testing and monitoringIt is essential to recognize that a test of this type only proves behaviourunder a specific load at one particular time. The behaviour under higherloads can only be speculative, and no indication of the margin of safetywith respect to failure can be obtained. The selection of load level must inmost cases therefore be a compromise between continued serviceability ofthe test member and demonstration of adequate load-bearing capacity. Ifthere is any possibility of future deterioration of the strength of materials,this must also be recognized in determining the test load together with anydecisions based on the test results.For short-term static tests instrumentation will generally be simple, butfor long-term or dynamic tests, measurement devices must be carefullyselected, particularly where strains are to be monitored. Gauges may beliable to physical damage or deterioration as well as temperature, humidityand electrical instability, and must also be selected and located with regardto the anticipated strain levels and orientation, as well as the gauge lengthsand accuracies available.Cost and inconvenience are serious disadvantages of in-situ load testing,but these are usually outweighed by the benefits of a positive demonstrationof ability to sustain required loads. Applications will therefore tend to beconcentrated on critical or politically ‘sensitive’ structures, as well as thosewhere lack of information precludes strength calculations. In-situ load testsare most likely to be needed in the following circumstances:(i) Deterioration of structures, due to material degradation or physicaldamage.(ii) Structures which are substandard due to quality of design or construction.(iii) Non-standard design methods which may cause the designer, buildingauthorities or other parties to require proof of the concept used.(iv) Change in occupancy or structural modification which may increaseloadings. Analysis is frequently impossible where drawings for existingstructures are not available, but in other cases an inadequate marginmay exist above original design values.(v) Proof of performance following major repairs, which may be politicallynecessary in the case of public structures such as schools, halls andother gathering places.In many cases chemical, non-destructive or partially destructive methodsmay have been used as a preliminary to load testing to confirm the need forsuch tests, and to determine the relative materials properties of comparablemembers. Although estimates of the ultimate structural strength may bepossible in the cases of deterioration of materials or suspect constructionquality, a carefully planned and executed in-situ load test will providevaluable information as to satisfactory behaviour under working loads.

Load testing and monitoring 157Long-term monitoring of behaviour under service conditions willnormally only be necessary if there is considerable doubt about futureperformance, or time-dependent changes are expected. This may occur ifdeterioration is expected, loadings are uncertain, or load testing is impracticableand there is a lack of confidence in strength estimates based on othermethods.6.2 MonitoringThis may range from relatively short-term measurements of behaviour duringconstruction to long-term observations of behaviour during service.6.2.1 Monitoring during constructionThere has been a growing awareness of the need to monitor factors suchas heat generation or strength development during construction of criticalstructures or parts of structures. Techniques for this purpose are outlinedin other parts of this book. In particular, the use of maturity, temperaturematchedcuring and pullout testing to monitor early strength developmentis attracting considerable interest in relation to fast-track construction (113)as identified in Chapter 1. Interest has also grown in the implications ofthe use of newly developing high-performance materials upon stresses andstructural performance and some reports of monitoring are available (168).Sensors may usefully be embedded in the concrete during construction topermit both short-term and long-term monitoring of a range of properties.As well as strains and temperatures these include cover-concrete propertychanges and conditions, reinforcement corrosion development and performanceof prestressing tendons. Examples are considered more fully later inthis chapter.6.2.2 Long-term monitoringIn cases of uncertainty about future performance of an element or structureit may be necessary to undertake regular long-term monitoring of behaviour.The scope of this ranges across the following:(i) modifications to existing structures(ii) structures affected by external works(iii) behaviour during demolition(iv) long-term movements(v) degradation and changes of conditions of materials(vi) feedback to design(vii) fatigue assessment(viii) novel construction system performance.

158 Load testing and monitoringThe Concrete Bridge Development Group in the UK has published acomprehensive guide to testing and monitoring for durability (19) whichoutlines a range of automated monitoring procedures and equipment. Thisencompasses temperature, strain, crack width and corrosion-related parameters,whilst testing specification and contractural issues are also included.Matthews (169) has considered in some detail the issues to be consideredwhen planning the deployment of instrumentation for in-service monitoring.These include both the development and implementation of a monitoringscheme. The purpose will significantly affect the amount of instrumentationnecessary. For example, to gain understanding of structural behaviouris obviously more demanding than simply checking if that behaviour lieswithin acceptable limits. It is necessary to establish clear working proceduresand methodology for both installation and operation of instrumentation.Issues requiring careful planning include:(i) Redundancy of instrumentation, to help detect erroneous readings (forexample due to instrument malfunction).(ii) Use of analytical modelling, to correlate and aid interpretation of measurements.(iii) Scanning schedules, including scope for alteration (possibly automatically)on the basis of results obtained, and their implications for datastorage and processing facilities required.(iv) Datum readings, must be readily established and easily identified.6.2.2.1 Changes in materials properties or conditionSuch monitoring may provide valuable information to assist predictionsof useful service life, and attention is primarily focussed on the near-tosurfaceregions of concrete (cover zone). McCarter (170) has reviewed awide range of embedded sensor systems for electrochemical measurements,which are related to the risk of reinforcement corrosion including pH levelsand chloride content. These are installed during construction and thereport is based on recent work by a RILEM committee. Hansson (171)has also described in-place corrosion monitoring probes for chloride ingresswith sensors at varying depths, which have been used on site for over 10years. These are fixed to the top reinforcement mesh. Half-cell potentials,macro-current flow and linear polarization resistance (see Chapter 7) canbe measured at each level, as well as electrochemical noise. Additionally,McCarter et al. (172) have described the development of a similar deviceconsisting of an array of ten electrode pairs at different depths below thesurface. This can detect water, ionic and moisture movements as well astemperature, all of which are relevant to the development of reinforcementcorrosion. Available techniques are also reviewed by the Concrete Bridge

Load testing and monitoring 159Development Group (19) whilst Malan et al. (173) recently describe a sensorfor long-term moisture monitoring. Two field case studies of remote fieldmonitoring of corrosion using Linear Polarization Resistance measurementare given by Broomfield (174). Embedded fibre optic sensors to monitorpH changes are also under development (175).Acoustic emission (see Chapter 8) can also be used to detect noise generatedby active reinforcement corrosion (176) and wire breaks in posttensionedconstruction with simple field measuring equipment attached tothe concrete surface.6.2.2.2 Structural movement and crackingMoss and Matthews (177) have comprehensively reviewed these applicationsas well as practical issues concerning instrumentation and techniquesto be used. A wide range of techniques are described ranging from simplevisual and surveying methods, often supplemented by automatic datacollection systems, to remote laser sensing systems to detect movements.Automation of data collection and storage is seen as a key aspect of manylong-term monitoring situations and wireless technology is being introduced.It is essential that measurements are taken at regular times eachyear to allow for seasonal effects, which may be considerable. Temperature,and preferably also humidity, should always be measured in conjunctionwith regular testing. Deflections and crack widths will normally be monitored,although strain measurements may also be valuable, particularlywhere the structure is subject to repetitive loading. In cases where thereare large-scale movements of one part of a structure relative to another,permanent reference marks may be established and measured by rule, orif the area is normally not visible, a scale and pointer may be firmly fixedto the adjacent elements to indicate relative movement. This is a simpleapproach: more refined measurement systems will be required where movementsare small.Deflections may be monitored by levelling, using conventional surveyingtechniques, and this will detect major movements. Levels should alwaysbe taken at permanently marked points on test members, preferably on astud firmly cemented to the surface. Midspan readings should be relatedto readings taken as close as possible to the supports for each test memberto determine deflection changes. Variations of this approach includetaut string lines or piano wires stretched between the datum points at thesupports, with a scale attached to the midspan point. A refinement involvinga laser beam provides a means of detecting very small movementsand may be appropriate for critical structures or members. This is suitablefor continuous measurement, and may also be adapted to trigger analarm system by replacing the scale with a suitable hole or slot throughwhich the beam passes. A light-sensitive target would then react if the

160 Load testing and monitoringFigure 6.15 Laser beam for long-term monitoring.beam is cut off by excessive movement of the hole or slot (Figure 6.15).A similar approach using an infrared transmitter and receiver has also beendescribed whilst another automatic system uses a taut stainless steel wirewith proximity switches fixed to the beams to detect relative movement.Temperature-compensating springs maintain the tension, and a wire-breakindicator is included in the control system which can provide an audible orlight warning when a switch is triggered. A maximum wire length of 50 m,with a control box handling up to 30 rooms, each containing 50 switches,has been claimed. A portable laser Doppler vibrometer technique has alsobeen described (178) for static and dynamic remote measurement of deflectionsof bridges loaded by trucks from a range of up to 30 m. Accuraciescomparable to Linear Voltage Displacement Transducers (see Section 6.3.1)are claimed, however this approach is unsuitable for long-term monitoring.Remote sensing photogrammetric techniques are also available (179),including the use of digital methods (180) for crack development. Electricalvariable-resistance potentiometers or displacement transducers may providea useful method of accurate deflection measurement, particularly if automaticrecording is an advantage. These consist of a spring-loaded centrallylocated plunger which is connected to a slide contact and moves up anddown the core of a long wound resistor, usually 50–100 mm in length, anda voltage is applied to the system and recorded by simple digital voltmeter.It is claimed (167) that accuracies of 0.025 mm can be achieved in this way.The problem with long-term deflection measurements involving this type ofequipment lies in providing suitable independent rigid supports and protectionfor the equipment, together with the long-term stability of calibrationof the devices.Crack widths may be measured using the optical equipment described isSection 6.1.3, but this is inevitably subject to the operators’ judgements.A simple method of detecting the continued widening of established cracks

Load testing and monitoring 161is the location of a brittle tell-tale firmly cemented to the surface on eitherside of the crack, which will fracture if widening occurs. Thin glass slips,such as microscope slides, are commonly used but calibrated plastic telltalesare also available. The ‘Scratch-a-Track’ device is one such system. Itcomprises two parts, affixed to the structure with epoxy putty, on eitherside of a crack. One part has a needle, which scratches a mark in the surfaceof the second part as the crack moves. This means that the extent anddirection of movement both can be monitored (Figure 6.16). A mechanicalmethod, using a Demec gauge as described is Section 6.3, will provide directnumerical data on crack-width changes and is simple to use if measuringstuds are fixed permanently on either side of appropriate cracks. Equipmentwith a nominal 100 mm gauge length is recommended, and the same devicemust always be used for a particular test location.If a precise measurement across cracks, or an automatic recording system,is required, the electrical displacement devices described above maybe useful but their long-term performance should be carefully considered.Crack propagation may also be recorded automatically by electrical gaugesconsisting of a number of resistor strands connected in parallel and bondedto the concrete surface in a similar manner to strain gauges. As a crackpropagates individual strands within the high-endurance alloy foil grid willfracture and increase the overall electrical resistance across the gauge. Thiscan be measured with a low voltage DC power supply and ohmmeter,and can be recorded automatically on a strip chart recorder. Additionallow voltage instrumentation can be employed to trigger an alarm if this isFigure 6.16 ‘Scratch-a-Track’ crack monitoring device (photograph courtesy of HammondConcrete Testing Services Ltd/Constructive Group).

162 Load testing and monitoringnecessary. Development of internal cracking due to load or fatigue can bemonitored by acoustic emission (see Chapter 8).Strains may be measured by a variety of methods as described inSection 6.3, but in most cases a simple mechanical approach using a Demecgauge will be the most successful for long-term monitoring. Provided thatthe studs are not damaged, readings may be repeated indefinitely over manyyears and, although the method suffers the disadvantage of a lack of remotereadout, the problems of calibration and long-term ‘drift’ of electrical methodsare avoided. If a remote reading system is preferred, vibrating wiregauges are generally regarded as reliable for long-term testing. The use ofoptical fibres to monitor strains is an important new development. Thishas become possible as a result of new technology, and reported applicationsinclude monitoring crack widths in post-tensioned bridges as well asincorporation into prestressing tendons to monitor their in-service loads(177,181,182). Two types of optical fibre are used:(i) Stranded optical fibre sensors. Measurements can be made of lightthat is lost or attenuated whilst passing through regions containingmicrobends as illustrated in Figure 6.17. The intensity of light emergingas the sensor is strained or relaxed is compared with the light supplied,to enable changes in overall length to be made to an accuracyof ±002 mm for gauge lengths up to 30 m. This accuracy is independentof gauge length, but the precise position of localized straining isnot given. Optical time domain reflectometry apparatus can be usedto transmit pulses of light of about 1 ns duration into the sensor andmeasure the transit times of echoes from reflections at points of attenuation.This backscatter permits positions of attenuation to be locatedwithin ±075 m.(ii) Multi-reflection sensors. These consist of single fibres containing upto 30 partial mirrors at intervals along their length. About 97% ofthe light passes through each partial mirror and the time for a pulseto be reflected back to the source is measured. The position of individualreflectors can be measured to an accuracy of ±015 mm usingpicosecond optical time domain reflectivity equipment.Stranded optical fibres can easily be fitted to an existing structure tomonitor long-term behaviour, and they are typically connected only at nodepoints to enable strain distribution to be assessed. Multi-reflection sensorsare particularly appropriate for incorporation into prestressing strands asa potentially ‘intelligent’ element in a ‘smart’ structure. Further practicaldetails are given by Dill and Curtis (181), including information aboutcontrol systems.This is an area in which there have been a great many developmentsin technology in recent years and Lau (182) reviews these in some detail,

Load testing and monitoring 163Light source directionMicrobendingBackscatter or lightLight lossFigure 6.17 Stranded optical fibre (based on ref. 181).including incorporation into ‘smart’ composites both for structural healthmonitoring and repair of concrete structures.Vibrating wire gauges can be used for long-term monitoring in conjunctionwith a calibration load which is wheeled on a trolley across the floor orroof under test. This method is particularly suitable for monitoring slabs,and may be cheaper than making repeated readings which require scaffoldingor ceiling removal for access.Interpretation of long-term monitoring will normally consist of examinationof load/deflection or load/crack width plots or the development ofcrack maps. Care must be taken to ensure that the effects of seasonal temperatureand humidity changes are minimized or accounted for, and thatloading conditions are similar whenever readings are taken for comparison.The importance of environmental records cannot be over-emphasized andfactors such as central heating or air conditioning must not be overlooked.The frequency of testing will normally be determined according to the levelof risk in conjunction with the observed trends of test readings with time.6.3 Strain measurement techniquesNewly developed optical fibre techniques are described above, whilst establishedtechniques for measurement of strain fall into four basic categories:(i) Mechanical(ii) Electrical resistance(iii) Acoustic (vibrating wire)(iv) Inductive displacement transducers.The field of strain measurement is highly specialized, and often involvescomplex electrical and electronic ancillary equipment. Only the principalfeatures of the various methods are outlined below, together with theirlimitations and most appropriate applications. Other methods which are

164 Load testing and monitoringof more limited value and recent laboratory developments are also brieflydescribed. The selection of the most appropriate method will be basedon a combination of practical and economic factors, and whilst BS 1881:Part 206 (183) offers limited guidance, selection will largely be based onexperience.Multidirectional ‘rosettes’ of gauges may be required in complex situationswhilst strain relief techniques are available to permit estimation ofin-situ concrete stress. One recent example involves cutting a slot into theconcrete surface and inserting and pressurizing a flat jack to restore thepre-existing strain field (184). Change of strain in the relieved area is measuredand stress can be calculated from the data on slot geometry, strainand pressure. Accuracy is improved by performing several cycles of pressurizing/depressurizingto plot a reproducible strain/pressure curve. Theoryand field case studies are both described.6.3.1 Methods availableThe principal features of the available methods for strain measurement aresummarized and compared in Table 6.1; the equipment and operation aredescribed below.(i) Mechanical methods. The most commonly used equipment consists ofa spring lever system coupled to a sensitive dial gauge to magnify surfacemovements of the concrete. Alternatively a system of mirrors and beamof light reflected onto a fixed scale, or an electrical transducer may beused. A popular type, which is demountable and known as a Demec gauge(demountable mechanical), is shown in Figure 6.18.Predrilled metal studs are fixed to the concrete surface by epoxy orrapid-setting acrylic adhesive at a preset spacing along the line of strainmeasurement with the aid of a standard calibration bar. Pins protrudingfrom the hand-held dial gauge holder, which comprises a sprung leversystem, are located into the stud holes, enabling the distance between theseto be measured to an accuracy of about 0.0025 mm. It is important thatthe Demec gauge is held normal to the surface of the concrete. ‘Rocking’the gauge to obtain a minimum reading is recommended. Temperaturecorrections must be made with the aid of an invar steel calibration bar, andreadings may be repeated indefinitely provided the studs are not damagedor corroded. Cleanliness of the gauge and studs is important, and carefulpreparation of the surface is essential prior to fixing of the reference studs,which should be stainless steel for long-term testing. Suitable studs areusually provided by the manufacturer or supplier. This type of gauge canalso be used to monitor expansion of cores with sets of Demec gauge pointsset at 120 angles around the core. This is useful in determining expansiondue to alkali–silica reaction, delayed ettringite formation and suchlike.

Load testing and monitoring 165Table 6.1 Strain gauge summaryType Gauge length Sensitivity Special limitations Advantages(mm) (microstrain)Mechanical 12–2000 50–2 Contact pressurecritical. Notrecommended fordynamic testsElectricalresistance(metal andalloy)0.2–150 20–1 Low fatigue life.Surface preparation,temperature andhumidity criticalSemiconductor 19–100 1 Constant calibrationrequiredAcoustic(vibratingwire)InductivedisplacementtransducersPhotoelasticPiezoelectric12–200 10–1 Assembly andworkmanship critical.Avoid magnetic,physical andcorrosive influences.Not recommendedfor dynamic tests0.6–25 2–100 Expensive electricalancillary equipmentneededDoes not measurestrain accuratelyOnly suitable forlaboratory useCheap androbust. Largegauge lengthGood for shortgauge lengthVery accurateGood forlong-term testsGood fordynamic testsLarge strainrange. Visualindicationof strainconfigurationGood for smallrapid changes instrainThe equipment is available in a range of gauge lengths, although 100,200 and 250 mm are most commonly used. In general the greatest lengthshould be used unless some localized feature is under examination, as inthe case of laboratory model tests. The method is cheap, simple and theequipment relatively robust in comparison with other techniques, withthe long gauge lengths particularly well suited to post-cracking testing ofconcrete and in-situ use. The principal disadvantage is the lack of a remotereadout, and this can lead to very tedious operation where a large numberof measurements are required. A further disadvantage is that differentreadings may be obtained by different operators.A development of the method to give a remote readout is shown inFigure 6.19. This has a gauge length of 100 or 200 mm and uses Demecstuds with a standardized tension spring system to hold the equipment inposition on the concrete surface. The flexible aluminium alloy strip is bent

166 Load testing and monitoringFigure 6.18 Photo of Demec gauge.Figure 6.19 Portal strain transducer.by relative movement of the pins, and this bending is measured by electricalresistance strain gauges. The equipment is particularly suited to laboratoryuse where a large gauge length is required. A similar approach has also beendescribed by Din and Lovegrove (185), in which bending of a formed metalstrip bolted to the concrete surface is monitored by electrical resistance

Load testing and monitoring 167gauges. This version is intended for long-term measurements where cyclicstrains are involved.(ii) Electrical methods. The most common electrical resistance gauge is ofthe metal or alloy type in the form of a flat grid of wires, or etch-cutcopper–nickel foil mounted between thin plastic sheets (Figure 6.20). Thisis stuck to the test surface, and strain is measured by means of changes inelectrical resistance resulting from stretching and compression of the gauge.The resistance changes may be measured by a simple Wheatstone bridgewhich may be connected to multi-channel reading and recording devices.The relationship between strain and resistance will be approximately linear(for these gauges), and defined by the ‘gauge factor’. Characteristics willvary according to the gauge construction, but foil gauges will generally bemore sensitive and have a higher heat dissipation which reduces the effectsof self-heating.Figure 6.20 Electrical resistance strain gauge.

168 Load testing and monitoringThe mounting and protection of gauges is critical, and the surface mustbe totally clean of dirt, grease and moisture as well as laitance and loosematerial. The adhesive must be carefully applied and air bubbles avoided,with particular care taken over curing in cold weather. These difficultieswill tend to limit the use of such gauges under site conditions to indoorlocations, where cotton wool placed over the gauge may provide a usefulprotective covering. Gauges may be cast into concrete if they are held inplace and protected during concreting. They may also be fixed to the surfaceof reinforcing steel before or after casting but may cause local distortionof strains. Scott (186) has however successfully developed a technique inwhich gauges are installed in a milled duct running centrally along thelength of a reinforcing bar leaving the surface unimpaired and this has beenused both in the laboratory and on site.The relationship between strain and resistance will change with temperatureand gauges may be self-compensating or incorporate a thermocouple.Alternatively a dummy gauge may be used to compensate for changes in theambient temperature. Gauges must be sited away from draughts, althoughtemperature will not affect readings over a small timescale of a few minutes.Humidity will also affect gauges, which must be water-proofed if they aresubject to changes of humidity, or for long-term use. Background electricalnoise and interference will also usually be present, and constant calibrationis needed to prevent drift. If the gauge is subject to hysteresis effects theseshould be minimized by voltage cycling before use.It is clear that the use of these gauges requires considerable care, skill andexperience if reliable results are to be obtained. Their fatigue life is low,and this, together with long-term instability of gauge and adhesive, limitstheir suitability for long-term tests. The strain capacity will also generallybe small unless special ‘post-yield’ gauges having a high strain limit areused. In the event that a surface-mounted strain gauge is located at theposition of a subsequent tension crack, it is unlikely that a meaningfulstrain measurement will be obtained and the gauge may even be torn inhalf. Conversely, a gauge located immediately adjacent to a subsequenttension crack will not give representative results relating to the overall straindeformation of the concrete surface. Care therefore should be taken inusing strain gauges on concrete surfaces that are expected to crack and theuse of Demec or other gauges with a large gauge length is often preferable.Electrical gauges with semiconductor elements are very sensitive, consistingof a grid fixed between two sheets of plastic, and are used in the sameway as metal or alloy gauges. The same precautions of mounting, temperaturecontrol and calibration apply, although they do not suffer fromhysteresis effects. The gauges are very brittle however, requiring care inhandling, and are unsuitable for casting into the concrete. The change inresistance is not directly proportional to strain and a precise calibration istherefore necessary, but the accuracy of measurement achieved is high.

Load testing and monitoring 169(iii) Acoustic (vibrating wire) methods. These are based on the principlethat the resonant frequency of a taut wire will vary with changes in tension.A tensioned wire is sealed into a protective tube and fixed to the concrete.An electromagnet close to the centre is used to pluck the wire, and it is thenused as a pick-up to detect the frequency of vibration. This will normallybe compared with the frequency of a dummy gauge by a recording devicesuch as a cathode ray oscilloscope or comparative oscillator, to account fortemperature effects. This type of gauge may be cast into the concrete (oftenencapsulated in precast mortar ‘dog-bones’), and if adequately protectedis considered suitable for long-term tests, although it is not appropriatefor dynamic tests with a strain rate of greater than l microstrain/s (183)because of its slow response time. Particular care is necessary to avoidmagnetic influences, and a stabilized electrical supply is recommended forthe recording devices.(iv) Inductive displacement transducers. Two series-connected coils formthe active arms of an electrical bridge network fed by a high frequency ACsupply. An armature moves between these coils, varying the inductance ofeach and unbalancing the bridge network. The phase and magnitude ofthe signal resulting from this lack of balance will be proportional to thedisplacement of the armature from its central position. The body of thegauge will be fixed to the concrete or a reference frame, so that the armaturebears upon a metal plate fixed to the concrete. In this way lateral or diagonalas well as longitudinal strains may be measured; an adjusting mechanismallows zeroing of the equipment. These gauges are particularly sensitiveover small lengths, but much expensive and more complicated electricalequipment is needed to operate them and to interpret their output. This,together with the extensive precautions necessary, means that considerableexperience is required.(v) Photoelastic methods. These involve a mirror-backed sheet of photoelasticresin stuck to the concrete face. Polarized light is directed at thissurface, and fringe patterns will show the strain configuration at the concretesurface under subsequent loading. The strain range of up to 1.5% islarger than any gauge can accommodate, but it is very difficult to obtain aprecise value of strain from this method. The method may prove useful inexamining strain distributions or concentrations at localized critical pointsof a member.(vi) Piezo-electric gauges. The electrical energy generated by small movementsof a transducer crystal coupled to the concrete surface is measuredand related to strain. This is particularly suitable if small, rapid strainchanges are to be recorded, since the changes generated are very shortlived,and these gauges are most likely to find applications in the laboratoryrather than on site.

170 Load testing and monitoring(vii) Digital photogrammetric techniques (180) for strain and crack developmentmonitoring have been mentioned above and are currently for usein the laboratory.(viii) Laser-interferometry for the same purpose has been reported (187).This involves electronic speckle pattern interferometry to examine the interferenceof reflected beams of coherent laser light, with the inspection areailluminated by two laser beams from different positions to measure smallsurface displacements. By subtracting patterns at different load levels, strainand crack development can be assessed.6.3.2 Selection of methodsThe non-homogeneous nature of concrete generally excludes small gaugelengths unless very small aggregates are involved. For most practical in-situtesting, mechanical gauges will be most suitable, unless there is a particularneed for remote reading, in which case vibrating wire gauges may beuseful and more accurate over gauge lengths of about 100–150 mm. Bothof these types are suitable for long-term use, given adequate protection.Optical fibres may be appropriate where measurements are required overconsiderable lengths.Electrical resistance gauges may be useful if reinforcement strains are tobe monitored, or for examining pre-cracking behaviour in the laboratory,and are usually associated with smaller gauge lengths, offering greater accuracythan mechanical methods. These gauges are, however, not re-usable;mechanical gauges have the advantage of not being damaged by crack formationacross the gauge length. Semi-conductor gauges are very accuratealthough delicate, and are likely to be restricted to specialized laboratoryusage. For dynamic tests, electrical resistance or transducer gauges will bemost suitable, although the operation of the latter will generally be morecomplicated and expensive. Photoelastic and piezo-electric methods havetheir own specialist applications outlined above. A great deal of informationconcerning the wide range of commercially available gauges, includingtheir maximum strain capacity, is to be found in literature supplied by variousequipment manufacturers, and it is recommended that this should beconsulted carefully before selection of a particular gauge.6.4 Ultimate load testingApart from its use as a quality control check on standard precast elements,ultimate load testing is an uncommon but important approach whenin-situ overload tests are impossible or inadequate. The effort and disruptioninvolved in the removal and replacement of members of a completedstructure are considerable, but the results of a carefully monitored test,

Load testing and monitoring 171preferably carried out in a laboratory, provide conclusive evidence relatingto the member examined. Comparison of the tested member with thoseremaining in the structure must be a matter of judgement, often assisted bynon-destructive testing.6.4.1 Testing procedures and measurement techniquesUltimate load tests should preferably be carried out in a laboratory wherecarefully controlled hydraulic load application and recording systems areavailable. For small members it may be possible to use standard testingmachines and for larger members a suitable frame or rig can be assembled.If the size of the test member prevents transportation to a laboratory itmay be possible to assemble a test frame on site, adjacent to the structurefrom which the member has been removed. In this case loads may be appliedby manually operated jacks, with other techniques similar to those usedin the laboratory. The control of load application and measurement willgenerally be less precise, however, and site tests should be avoided wheneverpossible.6.4.1.1 Load arrangementsThe most commonly used load arrangement for beams will consist of thirdpointloading. This has the advantage of a substantial region of nearlyuniform moment coupled with very small shears, enabling the bendingcapacity of the central portion to be assessed. If the shear capacity of themember is to be assessed, the load will normally be concentrated at asuitable shorter distance from a support.Third-point loading can be conveniently provided by the arrangementshown in Figure 6.21. The load is transmitted through a load cell or provingring and spherical seating on to a spreader beam. This beam bears on rollersseated on steel plates bedded on the test member with mortar, high-strengthplaster or some similar material. The test member is supported on rollerbearings acting on similar spreader plates. Corless and Morice (188) haveexamined in detail the requirements of a mechanical test arrangement toavoid unwanted restraint and to ensure stability.Details of the test frame will vary according to the facilities available andthe size of loads involved, but it must be capable of carrying the expected testloads without significant distortion. Ease of access to the middle third forcrack observations, deflection readings and possibly strain measurements isan important consideration, as is safety when failure occurs. Slings or stopsmay need to be provided to support the member after collapse.In exceptional circumstances where two members are available it maybe possible to test them ‘back-to-back’. The ends can be clamped togetherwith a spacer block by means of bolted steel frames, and the centres of the

172 Load testing and monitoringFigure 6.21 Laboratory beam load test arrangement.beams jacked apart by some suitable system incorporating a load cell. Thismethod may be particularly suitable for tests conducted on site.6.4.1.2 MeasurementsCrack widths, crack development and deflections will usually be monitoredusing the techniques described for in-situ testing. Dial gauges will beadequate for deflection readings unless automatic recording is required, inwhich case electrical displacement transducers may be useful. Strain measurementsmay not be required for straightforward strength tests, but arevaluable for tests on prototype members or where detailed informationabout stress distributions is needed. The techniques available for strain measurementhave been described in Section 6.3, and the choice of method willinvolve many considerations, but the advantages of a large gauge length inthe ‘constant moment’ zone are considerable.6.4.1.3 ProceduresBefore testing the member should be checked dimensionally, and a detailedvisual inspection made with all information carefully recorded. If nondestructivetests are to be used for comparison with other similar members,these should preferably be taken before final setting up in the test frame.After setting and reading all gauges, the load should be increased incrementallyup to the calculated working load, with loads, deflections and

Load testing and monitoring 173strains, if appropriate, recorded at each stage. Cracking should be checkedvisually, and a load/deflection plot prepared as the test proceeds. It willnot normally be necessary to sustain the working load for any specificperiod, unless the test is being conducted as an overload test as describedin Section 6.1. The load should be removed incrementally, with readingsagain being taken at each stage, and recovery checked. Loads will thennormally be increased again in similar increments up to failure, with deflectiongauges replaced by a suitably mounted scale as failure approaches.This is necessary to avoid damage to gauges, and although accuracy isreduced, the deflections at this stage will usually be large and easily measuredfrom a distance. Similarly, cracking and manual strain observationsmust be suspended as failure approaches unless special safety precautionsare taken. If it is essential that precise deflection readings are taken up tocollapse, electrical remote reading gauges mounted above the test membermay be necessary. If appropriate, acoustic emission could be used to warnof impending failure.Modern load testing machines usually give the option of testing membersunder either ‘load control’ or ‘displacement control’. The use of loadcontrol will result in a sudden catastrophic ultimate failure, as seen inFigure 6.22. The use of displacement control can enable the behaviour atthe ultimate limit to be examined more carefully, so long as the overallstiffness of the load testing machine is significantly greater than that of thetest member.Figure 6.22 Load/deflection curve for typical over-reinforced prestressed beam.

174 Load testing and monitoringCrack development should be marked on the surface of the test memberand the widths recorded as required. The mode and location of failureshould also be carefully recorded – photographs, taken to show the failurezone and crack patterns, may prove valuable later. If information is requiredconcerning the actual concrete strength within the test member, cores maybe cut after the completion of load testing, taking care to avoid damagedzones.6.4.2 Reliability, interpretation and applicationsThe procedures described above relate to basic load tests to determine thestrength of a structural member. These are relatively straightforward, butthe accuracy obtainable in a laboratory will be considerably higher than ispossible on site. In specific situations, more complex tests may be required,involving dynamic loading, monitoring of steel strains, measurement ofmicrocrack development or detailed strain distributions. Techniques suitablefor such purposes are described elsewhere in this book, but are of aspecialized nature. Testing of models or prototypes will also employ manyof the techniques described but is regarded as being outside the scope ofthis book.In most cases the results of load tests will be compared directly withstrengths required by calculations, and serviceability limits required by specifications.The linear part of the load/deflection curve can be compared withthe working loads carried by the member, and deflections and crack widthsalso compared with appropriate limits. The collapse load can be used tocalculate the moment of resistance for comparison with required ultimatevalues. A typical load/deflection plot for an under-reinforced beam such asis shown in Figure 6.2 can also be used to determine the flexural stiffness.BS 8110 (165) requires that where tests are made on new precast units foracceptance purposes, the ultimate strength should be at least 5% greaterthan the design ultimate load and that the deflection at this load is less than1/40 of the span.Calculation of the collapse moment of resistance on the basis of testson materials will not always provide a value which agrees closely witha measured value, as illustrated by Figure 6.23, which shows measuredcollapse loads for a series of pretensioned beams compared with predictedvalues calculated from small core strength estimates from the same beams.For over-reinforced beams, the dependence of collapse moment on concretestrength is likely to be greater (Figure 6.22). Particular care must be takenwhen testing over-reinforced beams unless displacement control is usedas discussed above because of the sudden nature of failure, which oftenprovides little warning from increasing deflections.A test to destruction is obviously the most reliable method of assessing thestrength of a concrete member, but the practical value of tests on members

Load testing and monitoring 17528Measured collapse moment (kN. m)262422201816Theoretical prediction020 30 40 50 60 70Estimated actual cube strength from cores (N/mm 2 )Figure 6.23 Measured vs. calculated collapse moments for prestressed concretebeams.from existing structures depends upon the representativeness of the membertested. Visual and non-destructive methods may be used for comparisons,but uncertainty will always remain. The most likely application, apart fromquality control checking of precast concrete, will be where a large numberof similar units are the subject of doubt. The sacrifice of some of these maybe justified in relation to the potential cost of remedial works.

Chapter 7Durability testsDeterioration of structural concrete may be caused either by chemical andphysical environmental effects upon the concrete itself, or by damage resultingfrom the corrosion of embedded steel. It is very likely that reinforcementcorrosion in one form or another will form part of the problem experiencedby an Engineer requiring a survey of a structure. The tests described in thischapter are concerned primarily with the assessment of material characteristicswhich are likely to influence the resistance to such deterioration,and to assist identification of the cause and extent if it should occur. Thesetests have been summarized in Table 1.3, although those involving chemicalor petrographic analysis (including carbonation depth, and sulfate andchloride content) are considered in detail in Chapter 9. Other relevant testsrelating to structural integrity and performance are described in Chapter 8,and test selection is discussed in Section 1.4.3.The principal causes of degradation of the concrete are sulfate attack,alkali–aggregate reaction, freeze–thaw damage, abrasion and fire. The presenceof moisture and its ability to enter and move through the concrete arecritical features since both sulfates and chlorides require moisture for mobilityand alkali–aggregate reactions cannot occur in dry concrete. Carbonationrates depend on gas permeability and are also influenced by moisturelevels. Tests which assess water and gas absorption or permeability, andmoisture content, are thus of great importance with respect to durability.Planning and interpretation of a typical corrosion-related investigation areoutlined in Appendix A7.7.1 Corrosion of reinforcement and prestressingsteelReinforcement corrosion is an electro-chemical process requiring the presenceof moisture and oxygen and can only occur when the passivatinginfluence of the alkaline pore fluids in the matrix surrounding the steel hasbeen destroyed, most commonly by carbonation or chlorides (189). Stray

Durability tests 177DC electric currents earthing through the reinforcement in a structure havealso been known to cause severe corrosion (190).A considerable amount of energy is given to steel when it is convertedfrom iron ore in the blast-furnace. Nature prefers materials to have thelowest possible energy state, so iron and steel will revert to iron oxideor rust (the main component of iron ore) given even slightly favourableconditions (i.e. the presence of moisture and oxygen). This is the lowestenergy state for iron. However, corrosion cannot occur if one or both ofmoisture and oxygen is absent. Engineers may recall the simple chemistryexperiment from school, where a nail is placed in water, another in a drytest tube and a third in boiled water with an oil layer over the water toprevent any oxygen re-dissolving. Only the nail in the first tube corrodes,since both air and water are required for corrosion to occur.Corrosion of embedded steel is probably the major cause of deteriorationof concrete structures at the present time. This may lead to structural weakeningdue to loss of steel cross section, surface staining and cracking orspalling. In some instances internal delamination may occur. The corrosionprocess has been described in detail by many authors but is summarized insimple form by Figure 7.1.This may involve either localized ‘micro-corrosion’ cells in which pittingmay severely reduce a bar cross section with little external evidence, orgeneralized ‘macro-corrosion’ cells which are likely to be more disruptiveand easier to detect due to expansion of the rusting steel (24). The formerFigure 7.1 Basic mechanisms of reinforcement corrosion.

178 Durability testsare normally associated with the presence of chloride salts, often whereonly a limited oxygen supply is available, owing to the density of the coverconcrete.So called ‘black rust’ often results, with the Fe 3 O 4 or FeO formsof iron oxide predominating, together with iron chloride salts. Stress onthe bar surface seems to be a factor in this and corrosion often initiates atbends in links, for example. Exposing a bar with black rust will often showzones of blackened bar, which turn brown on exposure to air (Figure 7.2).The development of anodic and cathodic regions on the surface of a steelreinforcing bar results in a transfer of ions within the concrete cover and ofelectrons along the bar and hence a flow of corrosion current. The rate atwhich corrosion occurs will be controlled either by the rate of the anodicor cathodic reactions or by the ease with which ions can be transferredbetween them. Thus an impermeable concrete, normally associated with ahigh electrical resistivity, will restrict ionic flow and hence result in lowrates of corrosion. A thick and impermeable cover region will also restrictthe availability of oxygen to the cathode region and further reduce the rateof corrosion. The development of an anodic region requires some smalldifference in the bar or its local environment. This may range from aninclusion in the bar or perhaps an air void next to the bar, in fact air voidsdue to poor compaction are being viewed with increasing suspicion as acause of corrosion initiation (191).The presence of corrosion activity can often be detected by measuringthe electrochemical potentials on the reinforcing bar from the surface of theconcrete with respect to a reference half-cell. The test can be used in conjunctionwith electrical resistivity measurements of the concrete cover zoneto give an indication of the probable rate of corrosion activity. Alternatively,Figure 7.2 ‘Black rust’ on post-tensioned strands revealed during a bridge inspection.

Durability tests 179the rate of corrosion may also be measured directly by one of a range ofdifferent perturbative or non-perturbative electrochemical techniques. Themost popular of these is the linear polarization resistance measurement. Asimple measurement of the thickness of the concrete cover will also givesome guidance to the expected durability of a reinforced concrete structure.7.1.1 Electromagnetic cover measurementElectromagnetic methods are commonly used to determine the locationand cover to steel reinforcement embedded in concrete. Battery-operateddevices commercially available for this purpose are commonly known ascovermeters. A wide range of these is commercially available with differentfeatures and capabilities, and a selection is illustrated below. Their use iscovered by BS 1881: Part 204 (192).7.1.1.1 Theory, equipment and calibrationThe basic principle is that the presence of steel affects the field of an electromagnet.This may take the form of an iron-cored inductor of the typeshown in Figure 7.3. An alternating current is passed through one of thecoils, while the current induced in the other is amplified and measured.The search head may in fact consist of a single or multiple coil system,with the physical principle involving either eddy current or magnetic inductioneffects. Eddy current instruments involve measurement of impedancechanges and will be affected by any electrically conducting metal, whilstmagnetic induction instruments involve induced voltage measurements andare less sensitive to non-magnetic metallic materials. The influence of steelFigure 7.3 Typical simple covermeter circuitry.

180 Durability testson the induced current is non-linear in relation to distance and is alsoaffected by the diameter of the bar. Figure 7.4 shows a model based onmagnetic induction which has a depth range of the order of 300 mm andan audible location signal, with anticipated bar size input by the operator.Estimated cover is indicated on a digital display.Eddy current versions of the equipment shown in Figure 7.4 are alsoavailable, involving more sophisticated electronic circuitry, which can automaticallyestimate and allow for bar size and provide increased accuracybut have a more limited range. Some equipment incorporates allowance forsteel type and provides an audible ‘low cover’ warning facility (Figures 7.5and 7.6), and several recent eddy current instruments operate on a pulseinduction principle. Alternative search heads are available in some casesaccording to the depth range which is of interest, and the extent of barcongestion. Equipment is being constantly upgraded by manufacturers andselection should be based on the nature of information required in particularcircumstances.Developments in covermeter equipment have resulted in several modelsas indicated above that offer to evaluate both the cover to the bar and thebar diameter itself, where this is not known. This may be accomplishedeither by the use of a spacer block (193) or by the use of a specialized searchhead, positioned both orthogonal and parallel to the target bar. The abilityto scan a covermeter over the concrete surface and continuously recordthe output into a data logger for subsequent graphical presentation is alsoavailable with many instruments.Basic calibration of the equipment is important and BS 1881: Part 204(192) suggests several alternative methods. These include the use of a testprism of ordinary Portland cement concrete. A straight clean reinforcingbar of the appropriate type is embedded off centre to project from the prismand to provide an appropriate range of four different covers which canFigure 7.4 Micro covermeter (photograph courtesy of Kolectric Ltd).

Durability tests 181Figure 7.5 Profometer 5 (photograph by courtesy of Proceq).Figure 7.6 Protovale P331 covermeter (photograph by courtesy of Elcometer Ltd).be accurately measured by steel rule for comparison with meter readings.Other methods involve precise measurements to a suitably located bar in air.In all methods it is essential that extraneous effects upon the magnetic fieldare avoided. Under these conditions the equipment should be accurate to±5% or 2 mm, whichever is the greater. While this is true in the laboratory,

182 Durability teststhe influence of additional steel bars in laps or merely adjacent to the steelunder investigation can cause errors of 5–6 mm or even more in some cases.Modern sophisticated covermeters can simultaneously correct for bar sizeand are significantly more accurate in these situations than earlier models.Equipment costs are, however, increased.A useful addition to the covermeter range is the Ferroscan (Figure 7.7).A 600 mm square template is fixed to the structure and the area in questionis then scanned in two directions at right angles to each other, in a seriesof sweeps. A calibrated wheel on the search head records exactly where thesearch head is at all times. Using electromagnetic principles and on-boardcomputer analysis, an image of the underlying reinforcement is then shownon a screen (Figure 7.8). This can be downloaded to a PC and then analysedFigure 7.7 PS200 Ferroscan covermeter (photograph by courtesy of Hilti).Figure 7.8 Covermeter image of reinforcement in a structure (photograph bycourtesy of Hilti).

Durability tests 183in detail to give a readout of bar sizes as well. The data can also be saved toa Microsoft Excel file. In the latest model, the PS200, the data is collectedby the scanning head and then transferred by infra red to the main unit.This avoids the need for a connecting cable and the main unit hung roundthe neck of the operative. Comfort and ease of use are therefore superior,and should not be underestimated when large areas are to be scanned.Site calibration checks should also be performed with the bar and concretetype involved in the investigation. This may involve the drilling of test holesat a range of cover values to verify meter readings, and if necessary to resetthe equipment or develop a separate calibration relationship.Further development is expected to provide a new type of covermeterbased upon the principle of magnetic flux leakage (194). A DC magneticfield normal to the axis of a reinforcing bar is set up via a surface yoke,which partially magnetizes the bar. A sensor moved from one pole of theyoke to the other detects the induced magnetic leakage field, which canbe used to determine both the depth and diameter of the bar. Magneticflux leakage also has the potential to enable detection of a reduction in thesection of the reinforcing bar, such as might be caused by severe pittingcorrosion or fracture of prestressing steel tendons. Efforts are being madeto use neural network artificial intelligence to simplify interpretation of theresults from this and other studies with magnetic induction sensors (195).7.1.1.2 ProcedureMost covermeters consist of a unit containing the power source, amplifierand meter, and a separate search unit containing the electromagnet which iscoupled to the main unit by a cable. In use, the reading will first be zeroedand the hand-held search unit is then moved over the surface of the concreteunder test. The presence of reinforcement within the working range of theequipment will be indicated by a digital value on the display.The search unit is then moved and rotated to obtain a maximum signalstrength (easily seen in Figure 7.9, as the black bar near the top of thescreen moves to the right as the bar is approached, and back to the left asthe bar is passed). This position will correspond to the location of a bar(minimum cover) and indicate its orientation. With some instruments thisis assisted by a variable-pitch audible output. The output will then indicatethe cover on the appropriate scale, whilst the direction of the bar will beparallel to the alignment of the search unit. The use of spacers may alsobe necessary to improve the accuracy of cover measurements which are lessthan 20 mm, with some instruments. More sophisticated equipment can, asindicated above, simultaneously estimate the bar size and correct the coverreading automatically (see Figure 7.9, where the figures in brackets are theestimated bar size and the corrected cover – in this case the same as the set

184 Durability testsFigure 7.9 Data display from covermeter (photograph by courtesy of Elcometer Ltd).values, but often differing if the incorrect bar diameter has been set on themachine).7.1.1.3 Reliability, limitations and applicationsAlthough the equipment can be accurately calibrated for specific reinforcementbars (Section 7.1.1.1), in simpler covermeters the accuracy that canbe achieved will be considerably reduced. The factors most likely to causethis are those which affect the magnetic field within the range of the meter,and include:(i) Presence of more than one reinforcing bar, laps, transverse steel as asecond layer or closely spaced bars (less than three times the cover) maycause misleading results. With some equipment a small, non-directional‘spot probe’ can be used to improve discrimination between closelyspaced bars and to locate lateral bars.(ii) Metal tie wires. Where these are present or suspected, readings shouldbe taken at intervals along the line of the reinforcement and averaged.(iii) Variations in the iron content of the cement, and the use of aggregateswith magnetic properties, may cause reduced covers to be indicated.This has been largely overcome in modern instruments.(iv) A surface coating of iron oxide on the concrete, resulting from the use ofsteel formwork, has been claimed to cause a significant underestimateof the reinforcement cover and should be guarded against.BS 1881: Part 204 (192) suggests that at covers less than 100 mm anaverage site accuracy of about ±15%, with a maximum of ±5 mm, maybe expected and it is important to remember that the calibrated scales aregenerally based on medium-sized plain round mild steel bars in Portland

Durability tests 185cement concrete. If the equipment is to be used in any of the followingcircumstances a specific recalibration should be made:(i) Reinforcement less than 10 mm diameter, high tensile steel or deformedbars. In these cases the indicated cover is likely to be higher than thetrue value. This will also apply if the bars are curved and hence notparallel to the core of the electromagnet.(ii) Reinforcement in excess of 32 mm diameter may require a recalibrationfor some models of covermeter.Estimates of bar diameter may only be possible to within two bar sizes,although the authors have had consistently good results with the Protovalecovermeter. The temperature range over which covermeters can operate isalso generally relatively small, and battery-powered models will not usuallyfunction satisfactorily at temperatures below freezing, which may seriouslylimit their field use in winter. Stability of reading may be a problem withsome types of instrument and frequent zero checking in open space isessential.The most reliable application of this method for in-situ reinforcementlocation and cover measurement will be for lightly reinforced members. Asthe complexity and quantity of reinforcement increases, the value of the testdecreases considerably, and special care should also be taken in areas wherethe aggregates may have magnetic properties. Uomoto et al. have describedbenchmark tests with procedures according to Japanese Standards (196)using both skilled and unskilled technicians with a variety of steel types andconfigurations. Results suggested that in most cases bars could be locatedwithin ±10 mm, size within ±5 mm and depth within ±10 mm. It was alsonoted that difficulties may be encountered near to member edges. Malhotra(63) has described an application to precast concrete quality checking, inwhich the linear scale is calibrated to enable an acceptable range of valuesto be established for routine component monitoring. Snell, Wallace andRutledge (197) have also considered detailed sampling plans for in-situinvestigations and developed a statistical methodology for such situations.Alldred (198) has compared a number of different covermeters on congestedsteel reinforcement and gives some correction factors that may be used toaccommodate errors of measurement.7.1.2 Half-cell or rest-potential measurementThis method has been developed and widely used with success in recentyears where reinforcement corrosion is suspected, and normally involvesmeasuring the potential of embedded reinforcing steel relative to a referencehalf-cell placed on the concrete surface as shown in Figure 7.10. ASTMC876 (199) covers this method.

186 Durability testsFigure 7.10 Reinforcement potential measurement.7.1.2.1 Theory, equipment and proceduresThe reference half-cell is usually a copper/copper sulfate Cu/CuSO 4 orsilver/silver chloride (Ag/AgCl) cell but other combinations have been used(200). Different types of cell will produce different values of surface potential,and corrections of results to an appropriate standardized cell may benecessary during interpretation. The concrete functions as an electrolyteand anodic corroding regions of steel reinforcement in the immediate vicinityof the test point may be related empirically to the potential differencemeasured using a high-impedance voltmeter.It is usually necessary to break away the concrete cover to enable anelectrical contact to be made with the steel reinforcement. This connectionis critical and a self-tapping screw is recommended, but adequate electricalcontinuity is usually present within a mesh or cage of reinforcement toavoid the need for repeated connections (201). Some surface preparation,including wetting, is also necessary to ensure good electrical contact (191).Two-cell methods, avoiding the need for electrical connections to reinforcement,can be used for comparative testing, but are sometimes regarded asless reliable.The basic equipment is very simple and permits a non-destructive surveyof the surface of a concrete member to produce iso-potential contour mapsas illustrated by Figure 7.11.A range of commercially available equipment is available including digitalsingle-reading devices (Figures 7.12 and 7.13), as well as ‘multi-cell’ and‘wheel’ devices (Figures 7.14 and 7.15) with automatic data logging and

Durability tests 187Figure 7.11 Typical half-cell potential contours on a concrete slab (courtesy of MGAssociates Construction Consultancy Ltd).Figure 7.12 Simple half-cell instrumentation.printout facilities designed to permit large areas to be tested quickly andeconomically (202).7.1.2.2 Reliability, limitations and applicationsEarly studies on half-cell potentials (203) were primarily concerned with elevatedbridge decks in the USA, where unwaterproofed concrete was treatedevery winter with large quantities of deicing salts. In conditions where there

188 Durability testsFigure 7.13 Survey equipment comprises a head that reads half-cell potential,resistivity and temperature and interfaces with a software suite thatmanages all of the survey data, covermeter and photographic information(photograph by courtesy Citec GmbH, Germany).Figure 7.14 Multi half-cell instrumentation (photograph by courtesy of Proceq).is a plentiful availability of oxygen and where chloride contamination isingressing from the surface, interpretive guidelines can be given (Table 7.1)to assess the risk of corrosion occurrence. Care should be taken in applyingthose guidelines to different environmental conditions. Studies on Europeanbridge decks (204) where waterproofing membranes are used or wheredeicing salts are applied less frequently have resulted in a different set of

Durability tests 189Figure 7.15 Wheel half-cell instrument (photograph by courtesy of Proceq).Table 7.1 General guides to interpretation of half-cell test results (based on refs 199,202 and 205)Half-cell potential (mV) relative to differentreference electrodesPercentage chance of corrosion activityCu/CuSO 4 Ag/AgCl InterpretationE>−200 E>−120 Greater than 90% probability thatno corrosion is occurring−200

190 Durability testsWhere chlorides are present in the concrete, due to the use of calciumchloride accelerator in the original mix, or the ingress of deicing ormarine salt, corrosion is typically associated with half-cell potential readingsCu/CuSO 4 in the range +100 V to −400 mV or more (206), withoften a very narrow range of potentials separating rapidly corroding and,apparently, uncorroding areas. Half-cell potential gradients are frequentlyshallow, due to the close proximity of adjacent corrosion ‘micro-cells’ onthe surface of the steel reinforcement.In the light of these conflicting interpretive criteria, it is now more commonnot to use absolute potential values as a means of assessing the probabilityof corrosion occurrence. The plotting of iso-potential contour mapssuch as in Figure 7.11 is preferred. Local corrosion risk is identified by‘islands’ of more negative, anodic regions and by steep potential gradients,seen by closely spaced iso-potential lines. Potential differences of more than100 mV between adjacent readings indicate significant corrosion risk (207).In carrying out a potential survey, an initial grid of 0.5–1 m is commonlyused to sample the surface potentials. In regions of particular interest orwhere micro-cell corrosion activity is suspected, a grid as fine as 0.1 mmay be used. It is essential to recognize that the half-cell method cannotindicate the actual corrosion rate or even whether corrosion has alreadycommenced. The test only indicates zones requiring further investigation,and an assessment of the likelihood of corrosion occurring may be improvedby resistivity measurements in these regions.This method is widely used when assessing maintenance and repairrequirements. It is particularly valuable in comparatively locating regionsin which corrosion may cause future difficulties, and those in which it hasalready occurred but with no visible evidence at the surface. Grantham(208) used a −150 mV half-cell potential criterion (Ag/AgCl) to map areasof a car park to target for remedial work, in addition to those areas whichwere visibly damaged. The repairs were then provided with additional protectionwith a car park decking membrane system. Half-cell potential canalso often be of use to confirm that passivity has been restored followingremediation to a corrosion-damaged reinforced concrete structure. Cautionmust be exercised when checking structures treated with electrochemicalrepair to ensure that the potential of the reinforcement has stabilized beforemeasurements are made. It may take some months for this to occur followingelectrochemical chloride removal, for example. Andrade and Martinez(209) are working on a system to confirm passivity of reinforcement undercathodic protection, without switching the current off.It must be understood, however, that half-cell potential is a weatherdependentphenomenon. It measures whether a structure is likely to becorroding not whether it is already corroded! In summer, concrete maydry out and corrosion cells shut down, giving much lower potential readings,despite obvious visible evidence of previous corrosion and spalling of

Durability tests 191concrete. Engineers can and have thus assumed that the test is misleading,because they do not understand that it measures the likelihood of corrosionbeing active at the time of measurement.7.1.3 Resistivity measurementsThe ability of corrosion currents to flow through the concrete can beassessed in terms of the electrolytic resistivity of the material. Proceduresfor in-situ measurement are available for use in conjunction with half-cellpotential measurements, but at the present time the method is less widelyused, in the UK, at least.7.1.3.1 Theory, equipment and proceduresElectrical resistivity tests have been used for soil testing for many yearsusing a Wenner four-probe technique, and this has recently been developedfor application to in-situ concrete. Four electrodes are placed in a straightline on, or just below, the concrete surface at equal spacings as shown inFigure 7.16. A low frequency alternating electrical current is passed betweenFigure 7.16 Four-probe resistivity test.

192 Durability teststhe two outer electrodes whilst the voltage drop between the inner electrodesis measured. The apparent resistivity is calculated as = 2sVIwhere s is the electrode spacing, V is the voltage drop and I is the current. Aspacing of 50 mm is commonly adopted and resistivity is usually expressedin cm or in kcm.Considerable efforts have been made to develop portable equipmentwhich permits satisfactory electrical contact without the need to drill holesinto the surface, and dampened sponge electrode tips can be used to makea good surface contact with the concrete (Figure 7.17). An alternativeapproach (210) has been to use a square wave AC current to accommodatethe effects of a poor surface contact (Figure 7.18). Both approaches havebeen shown (211) to give very similar results in most practical situations.This permits a rapid non-destructive assessment of concrete surface zones.Hand-held ‘two-probe’ equipment is also available (Figure 7.19), althoughthis requires holes to be drilled into the surface of the concrete and filledFigure 7.17 Four-probe resistivity equipment with sponge contacts (photograph bycourtesy of CMT Instruments Ltd).

Figure 7.18 Four-probe resistivity equipment with variable spacing electrodes(photograph by courtesy of CNS Farnell Ltd).Figure 7.19 Two-point resistivity instrumentation (photograph by courtesy of CMTInstruments Ltd).

194 Durability testswith conductive coupling gel to ensure reliable results. The authors’ experiencewith this approach is limited.7.1.3.2 Reliability, limitations and applicationsClassification of the likelihood of significant corrosion actually occurringcan be obtained on the basis of the values in Table 7.2, which can only beused when half-cell potential measurements show that corrosion is probable.The resistivity of concrete is known to be influenced by many factors includingmoisture and salt content and temperature as well as mix proportionsand water/cement ratio. McCarter et al. (212) have shown from laboratorystudies that resistivity decreases as the water/cement ratio increases andsince the resistivity of aggregate may be regarded as infinite relative to thatof the paste, the value for concrete is dependent upon the paste characteristicsand proportion. It is also claimed that the resistivity may be used asa measure of the degree of hydration of the cement in the concrete.For in-situ resistivity measurements there are a number of practical considerationsthat must be accommodated before interpreting the results (213).(i) The presence of steel reinforcement close to the measurement locationwill cause an underestimate in the assessment of the concrete resistivity.(ii) The presence of surface layers due to carbonation or surface wettingcan cause a significant underestimate or overestimate in the underlyingconcrete resistivity.(iii) Taking resistivity measurements on a very small member section orclose to a section edge may result in an overestimate of the actualresistivity.(iv) Resistivity measurements will fluctuate with changes in ambient temperatureand with rainfall. In external conditions in the UK, whereconcrete is normally moist, temperature is found to be the more dominantparameter.Some efforts have been made (214,207) to relate half-cell potential andresistivity measurements to rates of corrosion through computer modelling,but this approach has not yet seen significant practical usage.Table 7.2 Interpretation of resistivity measurementsResistivity (k.cm)Likelihood of significant corrosion when steelactivated (for non-saturated concrete)20 Low

Durability tests 195Although the principal application is for assessment of maintenance andrepair requirements in conjunction with half-cell potential measurements,the technique may be applied on a larger scale with greater spacings toestimate highway pavement thicknesses. Moore (215) has described USFederal Highway Administration work based on the differing resistivitycharacteristics of pavement concrete and subgrade. Electrode spacings arevaried and a change of slope of the resistivity/spacing plot will occur as aproportion of the current flows through the base material.7.1.4 Direct measurement of corrosion rateA number of perturbative and non-perturbative electrochemical techniqueshave been developed to determine the rate of corrosion directly. The principalmethods are:(i) Linear polarization resistance measurement(ii) Galvanostatic pulse measurement(iii) AC impedance analysis(iv) AC harmonic analysis(v) Electrochemical noise(vi) Zero resistance ammetry.These methods have seen increasing use in the field in the UK, especially linearpolarization, which has also seen some limited usage in the rest of Europe(216,217) and in the USA (218), and the galvanostatic pulse method, whichhas been applied to some European bridges (219,220) and on a car park inthe UK (192). A useful summary of methods is given by Bjegovic (205).7.1.4.1 Linear Polarization Resistance MeasurementLinear polarization resistance measurement is carried out by applying asmall electrochemical perturbation to the steel reinforcement via an auxiliaryelectrode placed on the surface of the concrete (Figure 7.20). Theperturbance is often a small DC potential change, E, to the half-cell restpotential of the steel, in the range ±20 mV. From a measurement of theresulting current, I, after a suitable equilibration time, typically 30 secondsto 5 minutes, the polarization resistance, R p , is obtained, whereR p = EIR p is inversely related to the corrosion current, I corr , and cathodic regionson the surface of the steel bar. HenceI corr = B R p

196 Durability testsFigure 7.20 Linear polarization resistance measurement.where for steel in concrete B normally lies between 25 mV (active) and50 mV (passive). The corrosion current density, i corr , is found fromi corr = I corrAwhere A is the surface area of the steel bar that is perturbed by the test.Although the linear polarization resistance method offers a direct evaluationof the rate of corrosion, a number of difficulties with this techniquemust be considered:(i) The time for equilibration to occur may result in excessive measurementtimes or inaccurate corrosion rate evaluations if prematuremeasurements are taken.(ii) When the concrete resistance, R s , between the auxiliary electrode onthe surface of the concrete and the steel reinforcing bar is high, thiscan result in significant errors in measuring R p unless R s is either electronicallycompensated or explicitly measured and deducted from R p .(iii) The use of a correct value for B requires a foreknowledge of thecorrosion state of the steel. Adopting an inappropriate value couldresult in an error of up to a factor of two.(iv) It is tacitly assumed that corrosion occurs uniformly over the measurementarea, A. Where localized pitting occurs then this assumptioncould result in a significant underestimate of the localized rate ofcorrosion.

Durability tests 197(v) Evaluating the area of measurement A is not simple. Some studies(221) recommend using a large (250 mm diameter) auxiliary electrodeand assume that the surface area of measurement is the ‘shadow area’of steel reinforcement lying directly beneath the auxiliary electrode. Analternative approach (222) is to accept that the perturbation currentwill try to spread laterally to steel lying outside the shadow areaand to confine this lateral spread by use of a guard ring positionedaround the auxiliary electrode (Figure 7.21). Both techniques showconsiderable promise in the field, and the guard ring technique cameout well in independent trials in the USA (223).(a)(b)Figure 7.21 Linear polarization resistance measurement, Gecor 8 and schematicshowing guard ring mode of operation (courtesy of Geocisa).

198 Durability tests(vi) No allowance is made for the effect of macro-cell corrosion, whichis inherent in many reinforced concrete structures under corrosionattack.(vii) Corrosion rates measured on one occasion may not be typical of themean annual rate of corrosion. Fluctuations in ambient temperature,moisture levels in the concrete, oxygen availability, etc. may all causethe instantaneous corrosion rate to vary significantly. Only by takinga series of measurements under different environmental conditionscan an overall evaluation of the annual rate of corrosion be made.(viii) It is essential that sufficient time is allowed for the system to reachequilibrium after the perturbation has been applied. Typically it mighttake several minutes and reducing this time could lead to significanterrors in the measurement of the corrosion rate (224).Typical values of i corr and the resulting rate of corrosion penetration areshown in Table 7.3 (225). The effect of corrosion rates on the change indiameter of individual bars is seen in Figure 7.22.7.1.4.2 Galvanostatic pulse methodThe galvanostatic pulse method uses a surface electrode arrangement similarto the linear polarization method (Figure 7.20). A small current perturbation,I, is applied to the steel reinforcing bar and the resulting shift in thehalf-cell potential, E, is measured. The steel bar which is actively corrodingexhibits a much smaller potential shift than a passive bar (Figure 7.23).Typically a current of 5–400A is applied for 10 seconds. The transientbehaviour of E can be used to evaluate the concrete resistance, R, and toevaluate the rate of corrosion, I corr (225). Equipment specifically designedfor field application of the galvanostatic pulse transient response methodis commercially available (Figure 7.24) and the technique holds considerablepromise. Measurements can be taken more quickly than normal linearpolarization measurements and the results are less ambiguous than can beTable 7.3 Typical corrosion rates for steel in concreteRate of corrosionCorrosion currentdensity, i corr (A/cm 2 )Corrosion rate ∗ (loss of sectionof reinforcing bar) (m/year)High >1 >12Medium 0.5–1.0 6–12Low 0.2–0.5 2–6Passive

Area %10090807060504030201000Decrease of Bar Sectional Area for l corr = 1uA/cm 25 10 15 20 25Propagation period (Years)General Corrosion 20 mm dia.General Corrosion 6 mm dia.Pitting Corrosion 20 mm dia.Pitting Corrosion 6 mm dia.Figure 7.22 Graph showing influence of corrosion rate on section loss ofreinforcement in structures (courtesy of Citec GmbH/M.G. AssociatesConstruction Consultancy Ltd).Current density, iΔiΔiΔE activeΔE passivePotential, EFigure 7.23 Effect of a small disturbing galvanostatic pulse on the potential shift,depending on the corrosion state of the bar.

200 Durability testsFigure 7.24 Galvanostatic pulse instrumentation (photograph by courtesy ofGermann Instruments).obtained from half-cell potential mapping. However many of the cautionsgiven in Section 7.1.4.1 with respect to the linear polarization resistancemethod also apply to the galvanostatic pulse method. Grantham and Schneckhave successfully employed the technique to examine the effectivenessof corrosion rate inhibitors applied to a car park deck (226).7.1.4.3 Other electrochemical techniquesBoth the AC impedance and the AC harmonic analysis techniques are laboratoryperturbative electrochemical methods that can evaluate the instantaneousrate of corrosion. By the use of more sophisticated electrochemicalmodelling, both techniques offer better accuracy than linear polarizationor galvanostatic pulse methods but are significantly more complex. ACimpedance is a lengthy test that requires specialist knowledge to carry outand to interpret the results. AC harmonic analysis is carried out over anarrow frequency range and is much quicker but still requires specialistexpertise to obtain meaningful results. Neither of these methods has seenthe field developments that have been applied to the linear polarization andgalvanostatic pulse transient response techniques. Their complexity makesit less likely that they will ever be used in the field for routine corrosionmeasurements.

Durability tests 201Some studies have related small fluctuations (0.1–10 mV) in the halfcellpotential of the reinforcing steel to the spontaneous formation andrepassivation of corrosion anodes. These fluctuations are called ‘potentialnoise’ or electrochemical noise and can be related to the rate of corrosion.Although there may be some reluctance to use this technique in the fielddue to a perceived risk of interference from external signals, it has beenused successfully in Canada in two trials where active corrosion has beenrelated to an increased level of electrochemical noise (225). Measurementscan be treated to a statistical processing, spectral analysis and chaos theoryanalysis to reduce the amount of data and to help determine the form ofcorrosion.A method of embedding a series of mild steel samples at increasing depthhas been used (227) to monitor the ingress of carbonation or chlorides andto evaluate the influence upon the rate of corrosion. Measurement of thecurrent passing between adjacent steel electrodes using a zero-resistanceammeter can give an early indication of corrosion activity. Effective use ofthe technique requires either planning before construction of the structure orsubsequent embedding into a void cut into the hardened concrete. The lattertechnique may place the sensors in an environment unrepresentative of therest of the structure and care should be used in implementing this approach.Other monitoring devices related to corrosion risk and development areconsidered in Section 6.2.2.1.7.2 Moisture measurementMeasurement of the internal moisture content in concrete is surprisingly difficult,but nevertheless of considerable importance when assessing durabilityperformance. Alkali–aggregate reactions are moisture dependent whilst theeffect of moisture conditions upon the interpretation of resistivity, absorptionand permeability test results is critical. The techniques described belowpermit quantitative assessments of varying accuracy, and other methodsdescribed in Chapter 8 such as infrared thermography and sub-surface radarmay detect moisture variations on a comparative basis.7.2.1 Simple methodsOne simple approach that is often adopted is to measure the moisture content,by oven drying and weighing, of a small timber insert or mortar prism(228) which has been sealed into a surface-drilled hole. The insert must beleft in the sealed hole for sufficient time to reach moisture equilibrium withthe surrounding concrete. A humidity meter based on this principle is commerciallyavailable (229) with measurement of the electrical conductivity ofa timber probe after removal from the surface-drilled hole to correlate torelative humidity.

202 Durability tests7.2.2 Neutron moisture gaugesThese work on the principle that hydrogen rapidly decreases the energy of‘fast’ or high-energy neutrons. Hydrogen is present in water in the concrete,and if the retarded or scattered neutrons resulting from interaction of ‘fast’neutrons with a matrix containing hydrogen are counted, the hydrogen, andhence moisture, content can be assessed. The most commonly used sourcesproduce neutrons indirectly by the interaction of -particles generated bythe decay of an X-ray emitting isotope, such as radium, with beryllium.Although a wide range of instruments are commercially available thatcould be used to determine the moisture content of concrete, these are normallycalibrated for a sample of semi-infinite volume and uniform moisturecontent. Since they operate on the basis of measurement of backscatter ofthe neutrons (see Section 8.4.3), results will only relate to a surface zone notexceeding 65–90 mm deep (according to equipment used). Further sourcesof inaccuracy in this approach will be moisture gradients near to the surfaceand the presence of neutron absorbers.Calibration is not simple and accuracy generally improves with increasingmoisture content. Although the equipment is widely used in soils testing,the accuracy possible with the lower moisture levels found in concreteis generally reduced. Nuclear magnetic resonance (NMR) is a techniquewhich has been the subject of recent laboratory studies (230) but is not yetdeveloped for site use.7.2.3 Electrical methodsAs indicated in Section 7.1.3.2, in-situ electrical resistivity measurementsare influenced by moisture content, but as yet this approach cannot be usedto assess moisture based on surface measurements other than in very broadclassification bands and on a comparative basis. Commercially availableequipment involving graphite probes inserted into a surface-drilled hole(229) involves measurement of internal resistivity.The change in dielectric properties of concrete with moisture content has,however, been used by a number of investigators as the basis of tests onhardened cement pastes and concretes. This approach is based on measurementof the dielectric constant and dissipation factor. The propertiesof a capacitor formed by two parallel conductive plates depend upon thecharacter of the separating medium. The dielectric constant is defined asthe ratio of capacitances of the same plates when separated by the mediumunder test, and by a vacuum. When a potential difference is applied tothe plates, opposite charges will accumulate, and if the separating mediumis ideal these will remain constant and no current will flow. In practice,electron drift will occur and a conduction current flows, and the ratio ofthis current to the initial charging current is the dissipation factor.

Durability tests 203It is well established that the dielectric constant for pastes is higher thanfor concretes, and that this constant decreases with age and increasingfrequency and increases with water content. Bell et al. (231) have shownthat the moisture content of laboratory specimens can be determined to±025% for values less than 6% using a 10 MHz frequency, and Jones (232)has confirmed that high frequencies (10–100 MHz) minimize the influencesof dissolved salts and faulty electrode contacts.Simple hand-held electronic equipment for site usage is available whichmeasures the relative humidity of the air within the concrete using similarprinciples. The digital meter is connected to an electronic probe sealed intoa 60 mm deep surface-drilled 25 mm diameter hole (Figure 7.25). The probeis in the form of a small plastic capacitor in a protective guard and isinfluenced by the relative humidity of the air in the hole. A range of 20–90% RH is quoted, although calibration is based on providing a maximumaccuracy at around 75% RH (which is critical for alkali–silica reactions).Accuracy varies with the difference in reading from this value and ±15%of the difference is claimed (e.g. ±375% at 50% RH).This equipment is particularly suited to long-term monitoring situationssince the probes may be left in place and readings taken whenever required.Similar probes may also be monitored by automatic recording equipment.More accurate evaluation of the relative humidity of the air inside a holedrilled into the concrete can be obtained using a chilled mirror hygrometer.Caution should however be exercised as it has been reported that carbonationof the concrete in a drilled hole may release bound moisture, and thusfalsely increase the apparent relative humidity.Figure 7.25 Scribe humidity meter.

204 Durability testsThese devices work on the principle of detecting the dew point temperatureas condensation forms on a small mirror (which may be made of gold),which can be both heated and cooled from the ambient temperature. Theinstrument can measure relative humidity over the entire range 0–100%with an achievable accuracy of ±1%. A probe developed for use in concreteutilizes an expandable gland (Figure 7.26) to enable a seal within a26 mm diameter hole drilled into the concrete element, whilst measurementsmay also be made on drilled dust samples placed in a small container asshown.Studies by Parrott (233) have shown an empirical relationship betweenthe relative humidity of an air void within the concrete and the moisturecontent of the concrete, but this relationship is not unique and varieswith the mix of concrete used. These studies have also demonstrated thehigh sensitivity of the air permeability of the concrete to the moisturecontent.Relative humidity measurements may also be taken from concrete dustobtained by in-situ drilling. A technique is described (234) where drill dustis collected without excessive evaporation of the moisture, and evaluationof the moisture content of the in-situ concrete to an accuracy of ±1% isclaimed.7.2.4 Microwave absorptionMicrowaves are electromagnetic and have a frequency in the range10 9 –10 12 Hz. They are absorbed by water at a higher rate than by concreteFigure 7.26 Chilled mirror relative humidity probe.

Durability tests 205in which the water may be dispersed, hence measurement of attenuationcan provide a method of determining moisture content. The principles ofthis approach have been described by Browne (235).Early efforts used a microwave beam generated by a portable radio transmitterand received by a crystal detector connected to an amplifier with afrequency of 3 × 10 9 Hz modulated at 3 kHz by a square wave. The beamis passed through the specimen with the transmitter and receiver located atfixed distances on either side, and an attenuator attached to the receiveradjusted to give a null reading. This procedure is repeated with the specimenremoved. The difference between the two attenuation readings is dueto the specimen, and may be converted to a moisture content with the aidof an empirically developed calibration chart.Unfortunately, because of the internal scattering and diffraction causedby the heterogeneous nature of concrete, the accuracy of predicted moisturecontent was low and as poor as +30%. Although the method is unlikelyto be of much practical value at this level of accuracy, the techniquesare still under development and improvements are being achieved. Thecommercially available equipment at present also requires two oppositeexposed faces, thus imposing a further restriction on use. An embeddedtransmission line sensor operating at 1 GHz has however recently beendescribed (173) which has potential for long-term monitoring of moisturein the surface zone.More recent studies (236) of the magnitude of the reflection of an electromagneticradar wave in the frequency range 05–10 × 10 9 Hz, using animpulse antenna, have also investigated the moisture content of concrete.Access to only one face is necessary and the moisture content measurementis related to the surface zone of concrete. This approach has becomeaccepted for comparative assessment of moisture conditions and is ableto clearly differentiate between wet and dry zones, possibly through anasphalt layer as in bridge decks. Electromagnetic radiation in the form ofsub-surface radar testing is also used to assess element thicknesses, andlocate reinforcing bars and voids. A fuller description of this technique isgiven in Chapter 8.Figg (237) has described the use of a microwave moisture meter of2450 MHz frequency on sections sawn from commercially produced precastcladding panels. These were constructed from flint and limestoneaggregate concretes with an exposed aggregate finish on one face. Thetests confirmed a linear relationship between microwave attenuation andpercentage of absorbed water, but with a slope which varies with differenttypes of concrete. Prior preparation of a specific calibration chartis therefore necessary to permit an absolute value of moisture contentto be obtained in a practical situation. A further limitation to practicaluse was also indicated by anomalous results obtained during freezingweather.

206 Durability tests7.3 Absorption and permeability testsThese properties are particularly important in concrete used for waterretainingstructures or watertight basements, as well as being critical fordurability. Tests are available for assessing water absorption as well as gasand water permeability. Considerable attention has been paid to this areain recent years and the topic is covered comprehensively in the ConcreteSociety Technical Report 31 (238) and by Basheer (239), which both reviewa wide range of testing techniques and fundamental theory. Only a selectionof these test methods, some of which are no longer in active use, isincluded in this chapter whilst other new methods are under development.RILEM (240) also consider a range of methods and are in the process ofdeveloping a new report on the topic of durability performance assessment,whilst Basheer et al. also provide a recent review of the role of permeationproperties (241).The terms ‘permeability’ and ‘porosity’ are often used as though theirmeanings were synonymous. Porous concrete is often permeable and viceversa.However, the term ‘porosity’ is a volume property that relates thevolume of the pores to the total volume. Unless the pores are interconnected(Figure 7.27) the material will not be permeable. The term ‘permeability’normally relates to the ease with which a fluid will pass through a materialunder a pressure gradient, but is also used to describe capillary, diffusion,adsorption and absorption processes (238).The mechanism of water transport through an interconnected pore systemis schematically shown in Figure 7.28, where a single pore has restrictionson each side. Under dry conditions the principal mechanism is molecularadsorption to the sides of the pores (stage a). Once the pore walls havereached their adsorption limits (stage b), water vapour will diffuse acrossthe pore. Under moisture conditions the restrictions to the pore inlet andoutlet become blocked with water (stage c) and further water movementPorous, impermeable materialPorous, permeable materialHigh porosity, low permeabilityHigh permeability, low porosityFigure 7.27 Porosity and permeability (based on ref. 238).

Durability tests 207Adsorbed layerLiquid meniscusStage (a) Adsorption Stage (b) Vapour diffusion Stage (c) Film transferStage (d) Surface creep plusvapour diffusionStage (e) Partially saturatedliquid flowStage (f) Saturated liquidflowStage (g) Ionic diffusion through partially unsaturated or saturated poresFigure 7.28 Movement of water and ions within concrete pores (based on ref. 238).must involve a transfer across this film. At much higher moisture levels(stages d–f) water is transported by liquid flow through the pore.Ionic diffusion occurs as a result of a concentration gradient rather thana pressure gradient and will occur through the liquid contained in partiallyor fully saturated pores (stage g).Two of the most widely established methods for in-situ usage are theinitial surface absorption test (ISAT), which is detailed in BS 1881: Part208 (242) and assesses water absorption, and the modified Figg air permeabilitymethod. Several other surface-zone gas and water permeabilitymethods (often similar) are commercially available, but reported experienceis relatively limited. The initial surface absorption method was originallydevised for testing precast concrete products, and presents some practicaldifficulties in use on site. Although the Figg method was developed for siteuse, and has grown in popularity, reported field experience is not extensive.In all cases emphasis must be on comparative rather than quantitativeapplication, and results only relate to surface-zone properties. It must berecognized, however, that this is the critical region as far as durabilityperformance is concerned.

208 Durability testsTable 7.4 General relationship between permeability and absorption test results on dryconcrete (238,244)MethodConcrete permeability/absorptionLow Average High‘Intrinsic permeability’ k m 2 10 −17ISAT – 10 min ml/m 2 s 0.50Figg water (s) >200 100–200 300 100–300

Durability tests 209applied head and temperature. Results will be expressed as ml/m 2 s at astated time from the start of test.When water comes into contact with dry concrete it is absorbed bycapillary action, at a rate which is high initially, but decreases as the waterfilled length of capillaries increases. Levitt (246) has shown that this maybe described mathematically by the expressionsp = at −nwhere p = initial surface absorptiont = time from starta = a constantn = a parameter between 0.3 and 0.7 depending on the degree of siltingor flushing mechanisms, but constant for a given specimen.The standard procedure described by BS 1881 involves a constant head of200 mm, with readings of initial surface absorption taken at 10 minutes,30 minutes and 1 hour. Levitt has demonstrated that the factors a and n inthe above expression can be evaluated from the 10 minute and 1 hour rateswith sufficient accuracy to predict rates at other times. Two hour readingswere originally proposed but are seldom used in practice.7.3.1.2 Equipment and procedureThe equipment (Figure 7.29) consists of a cap which may be clamped andsealed onto the concrete surface, together with an inlet connected to areservoir and an outlet connected to a capillary tube with a scale. The watercontact area must be at least 5000 mm 2 whilst the reservoir and horizontalcapillary must be set at 200 ± 5 mm above the surface. It is useful if theFigure 7.29 Initial surface absorption test.

210 Durability testscap is manufactured from a transparent material to permit visual checkingthat no air is trapped during a test. The capillary tube should be at least200 mm long with a bore of 0.4–1.0 mm radius which must be determinedprecisely by measuring the discharge under a 200 mm head. Where thepermeability of the concrete is high a length of the order of 1000 mm maybe needed. After correcting for temperature effects on viscosity the radiusmay be calculated from the following simplified formula:r 4 = KLtwhere r = radius (mm)L = length of capillary (mm)t = average time to collect 10 ml of water (s)K = a viscosity factor allowing for the apparatus geometry, ranginglinearly from 0.0167 at 10 C to 0.0100 at 30 C.BS 1881: Part 208 describes a detailed procedure for cleaning the capillaryand performing the above measurement.The bore of the capillary must be known, to permit calibration of thescale which is arranged so that the measured movement of water along thecapillary during a one-minute period of test will be equal to the value ofinitial surface absorption at that time. This is achieved by marking the scalein units of 0.01, spaced 6A/r 2 ×10 −4 mm apart, where A = water contactarea mm 2 . At the start of the test, the cap must be fixed to the surface andsealed to provide a watertight assembly. The choice between greased solidrubber gaskets, foamed rubber gaskets or a knife-edge bedded in a sealingmaterial will depend upon the condition of the surface. Before filling, theseal should be tested by gentle blowing, with soap solution applied to theoutside of the joint to detect leaks. The reservoir and capillary must be set200 ± 20 mm above the surface, this being measured from the mid-heightof concrete under the cap for non-horizontal surfaces, although it must bepossible to lift the end of the capillary to avoid overflow between readings.Timing is started when the reservoir tap is opened and water at 20 ± 2 Callowed to flow into the cap. The capillary should be disconnected fromthe outlet tube until all air has been expelled. The reservoir head must bemaintained, and shortly before a ‘measurement time’ the capillary adjustedso that it fills with water before fixing horizontally at the same level as thereservoir surface.Measurements are made by closing the inlet tap and watching for movementof the capillary. Using a stopwatch, the number of scale units movedin the first five seconds from the start of movement is noted and used todetermine the total measurement time. If less than three divisions measurementsshould be continued for two minutes, if 3–9 divisions, continue forone minute, or if 10–30 divisions, continue for 30 seconds. The measured

Durability tests 211reading should then be factored as necessary to give the number of scaleunits moved in one minute, and hence the value of initial surface absorption.If the movement is more than 30 divisions in five seconds, the resultcan only be quoted as more than 360 ml/m 2 s. This procedure should beperformed 10 minutes, 30 minutes and one hour after the start of the test,and at least three separate samples tested in this way.Major differences of the concrete temperature from the 20 C calibrationvalue will influence measured values due to changes in water viscosity.In-situ results should be corrected to an equivalent 20 C value by multiplyingby a factor ranging from 1.1 at 15 C to 0.8 at 30 C. At 5 C the factoris 1.5 and at 10 C it is 1.3.Laboratory samples for testing should be oven-dry, but for large units orsamples this is not possible. If the test is to be performed in a laboratory,the concrete must have been in the dry laboratory atmosphere for at least48 hours. For in-situ testing a minimum dry period of 48 hours is specified,but these requirements are very unlikely to produce comparable moistureconditions within the concrete and this will cause major problems in theinterpretation of results.An in-situ vacuum drying technique has been developed by Dhir et al.(248), which overcomes the difficulty of the uncertain moisture contentof the in-situ concrete and gives ISAT 10-minute results with a standarddeviation similar to that obtained with oven-dried specimens. Several othersurface-drying procedures involving the use of heaters have been proposedbut, as yet, no standardized procedure has been agreed.A further development of the ISAT test (249) proposes the use of a guardring located around the perimeter of the standard cap containing water atthe same hydrostatic pressure. Water flowing into the concrete surface fromthe guard ring will confine the absorption of water from the central capto a uniaxial flow in the concrete. Using this modification, an improvedcorrelation with adsorption tests carried out on 50 mm oven-dried coreswas found.7.3.1.3 Reliability, interpretation and applicationsIt has been found that tests on oven-dried specimens give reasonably consistentresults, but that in other cases results are less reliable. Particulardifficulties have been encountered with in-situ use in achieving a watertightfixing. Levitt (246) has suggested that specific limits could be laid downas an acceptability criterion for various types of construction, but insufficientevidence is yet available. Values of greater than 050 ml/m 2 s at 10minutes for dry concrete have been tentatively suggested as correspondingto a high absorption (see Table 7.4). The test has been found to be verysensitive to changes in quality and to correlate with observed weatheringbehaviour. The main application is as a quality control test for precast units

212 Durability testsbut application to durability assessment of in-situ concrete is growing. Aminimum edge distance of 30 mm is generally proposed. The pressure usedis low and although this may relate well to normal surface weather exposureconditions it is of little relevance to general behaviour under higherpressures. The method can be applied to exposed aggregate or profiled surfacesprovided that an effective seal can be obtained, but is not suitable forporous or honeycombed concretes.Dhir et al. (250) have considered use of the 10-minute ISAT result toevaluate resistance of the concrete to chloride diffusion and to carbonationand give guidance for lightweight concrete and for concrete with cementreplacements in addition to normal Portland cement.Given that the surface zones of bridges and car parks subjected to deicingsalt ingress may exhibit high surface porosity, the ISAT approach may beuseful in assessing the risk of chloride ingress (251).7.3.2 Figg air and water permeability testsFigg in 1973 originally described the development of a test for air andwater permeability, which involved a hole drilled into the concrete surface,commonly known as the Building Research Establishment Test. A numberof versions of this approach have subsequently been developed in variouscountries (238) but the most widely accepted procedure is that proposed byCather et al. (252) on the basis of extensive experience with the method.This is commonly known as the modified Figg method and is describedbelow. Earlier work was based on a smaller hole depth and diameter (30 and5.5 mm respectively) and comparison of results from different investigationsshould be treated cautiously because of the lack of standardization of holesizes.7.3.2.1 Equipment and procedureA hole of 10 mm diameter is drilled 40 mm deep normal to the concretesurface with a masonry drill. After cleaning, a disc of 3 mm thick polyetherfoam sheet is pressed 20 mm into the hole and a catalyzed liquid siliconerubber is added. This hardens to provide a resilient seal to the small cavityin the concrete, and a gas-and-liquid-tight seal is obtained by a hypodermicneedle through this plug.Air permeability measurements are by means of a hand-operated vacuumpump and digital manometer connected by a three-way tap and plastictubing to the hypodermic needle as illustrated in Figure 7.30. The pressureis reduced to −55 kPa and the pump then isolated, with the manometerand concrete connected together. The time, in seconds, for air to permeatethrough the concrete to increase the cavity pressure to −50 kPa is notedand taken as a measure of the air permeability of the concrete.

Durability tests 213Figure 7.30 Modified Figg air permeability test (based on ref. 252).Water permeability is measured at a head of 100 mm with a very finecanula passing through the hypodermic needle to touch the base of thecavity. A two-way connector is used to connect this to a horizontal capillarytube set 100 mm above the base of the cavity, and a syringe. Water isinjected by syringe to displace all air, and after one minute the syringe isisolated with the water meniscus in the capillary in a suitable position. Thetime, in seconds, for the meniscus to move 50 mm is taken as a measure ofwater permeability of the concrete.An automated system, the ‘Poroscope’, utilizes a preformed silicone plug(Figure 7.31) to avoid delays in waiting for sealant to set and can be usedfor both air and water permeability Figg tests. Results obtained are similarto those using a liquid silicone sealant and manual control. The use of apre-formed plug also enables tests to be carried out on a slab soffit, whichwas not previously possible using a liquid sealant.7.3.2.2 Reliability, limitations and applicationsUsing this method, the relationships between air pressure and time, andmeniscus movement and time, were both found to be nearly linear. Thetest criteria of hole depth, plug thickness, pressure and test time have beeninvestigated. Results have been obtained on laboratory concrete which show

214 Durability testsFigure 7.31 Automatic Figg test equipment – ‘Poroscope-Plus’ (photograph bycourtesy of James Instruments Inc).that both air and water permeability measured in this way correlate wellwith water/cement ratio, strength and ultrasonic pulse velocity. Aggregatecharacteristics have a profound effect on results, limiting the potential usageto comparative testing, but variations of drilling and plugging of the testhole are less significant. As with the initial surface absorption method, themoisture condition of the concrete will considerably influence the results.This seriously restricts in-situ usage, but a general classification for dry concreteis given in Table 7.4. Values should be treated with caution because ofvariations in hole dimensions used by various investigators. It has also beensuggested that use of an even larger hole will improve repeatability (253).The principal use of this method, which involves cheap and simple techniques,is as an alternative to the initial surface absorption method forquality control checking in relation to durability. It is usually the outer50 mm of concrete which is most important when considering durability,since this protects the reinforcement from corrosion. The method tests thisregion, and has the advantage of not being influenced by very localized surfaceeffects such as carbonation of the outer few millimetres of the concrete.

Durability tests 215A water permeability test operating at higher pressure (approx 17 bar)at a depth of about 75 mm has also been described in 1992 (254), butsubsequent reports are few.7.3.3 Combined ISAT and Figg methodsDhir et al. (253) have proposed that the shortcomings of the ISAT and Figgwater permeability methods can be overcome by combining these in theform of the Covercrete absorption test (CAT). It is intended that the testbe performed on 100 mm cores taken from the structure which are ovendried to overcome moisture effects. A 13 mm diameter, 50 mm deep holeis drilled and ISAT equipment with a 200 mm head of water is used butwith a 13 mm internal diameter cap and the inlet tube located inside thesealed hole. A coefficient of variation of about 8% was obtained with thisapproach, which is significantly better than for the Figg water method.7.3.4 Germann Gas permeability testThis Danish method (255) has been developed from work by Hansen andothers for in-situ use. An 18 mm hole is drilled at a shallow angle below theconcrete surface. The pressure rise within this hole is monitored by a sensorinserted in the hole as compressed carbon dioxide is applied to a surfaceregion formed by a test rig clamped to the surface with a sealing gasket.The depth may be up to 25 mm with pressure up to 3 bar. The equipment iscommercially available, although published experience is relatively limited.7.3.5 ‘Autoclam’ permeability systemThe hydrostatic pressure used in the ISAT is low and there are circumstancesrelated to severe exposure conditions or testing of surface coatings wherea higher pressure in-situ test is needed. The Autoclam is a commerciallyavailable system developed at Queen’s University, Belfast (256) from amanual system named CLAM, which is similar in principle to the ISAT,but which uses a hydrostatic pressure of 1.5 bar to measure in-situ waterpermeability (Figure 7.32). The equipment can alternatively be used tomeasure a low-pressure, 0.01 bar water sorptivity. In addition a measureof air permeability, through decay of air pressure of the order of 0.5 barapplied to the concrete surface, can also be measured. All three tests arecarried out by clamping the Autoclam to a 50 mm metal ring which is gluedto the concrete surface. The low-pressure water sorptivity test can be carriedout at the same location as the air permeability test, but it is recommendedthat the high-pressure water permeability test is done at a different testlocation.

216 Durability testsFigure 7.32 AUTOCLAM air and water permeability instrumentation.Once each test is commenced, the control of the test and monitoringof results is fully automatic. Each test takes about 15 minutes and alltests are sensitive to the moisture condition of the in-situ concrete. Onlaboratory cubes, the influence of the internal moisture in the concrete canbe eliminated by oven drying to a constant weight. For in-situ concrete, amethod of surface-drying for 20 minutes using a propane heater followedby cooling for 1 hour was found to give reliable results unaffected by theoriginal moisture condition of the concrete. The Autoclam tests were foundto correlate well with other permeability and sorptivity tests and can beused to rank different concretes into poor, medium and high durabilitycategories. The high-pressure permeability test is thought to test concreteup to 40 mm below the surface and the results have the potential to be usedto evaluate the intrinsic water permeability of the concrete (256).7.3.6 Other non-intrusive water and air methodsAnother commercially available water permeation system, GWT (255), issimilar in principle to the CLAM but can operate at pressures up to 6 barby means of a screw-operated piston and micrometer gauge to maintain therequired pressure. Gomes et al. (257) have recently reported on a detailedand continuing study of the application of this method to a range of concrete

Durability tests 217types, strengths and ages. A standardized pressure of 0.4 bar has been usedand concrete permeability classifications are proposed on this basis.Also in this category is a vacuum method developed in Switzerland byTorrent (258). This has a circular two-chamber cell sealed to the surfaceby rubber rings with the inner measuring chamber surrounded by an outerguard ring cell to achieve unidirectional air flow from the concrete into theinner 50 mm diameter cell. A vacuum is created by a pump and the rateof pressure rise as air is drawn from the concrete is measured to enablea coefficient of air permeability to be calculated, and classifications forconcrete surface zone quality have been proposed. The technique has beenthe subject of several investigations and is covered by a Swiss standard andcommercially available (259). A further similar test has been developed inthe laboratory in the USA by Guth and Zia (260) in which air flows throughthe concrete surface from the outer cell to the inner cell where the vacuum iscreated. This is a low cost device and a test time of the order of 10 minutesis claimed.7.3.7 Flow testsHigh-pressure permeability tests have been used in the laboratory for manyyears and consist of the measurement of the steady flow of water through aconcrete specimen under a specified head. The equipment is usually simpleand an arrangement of the form shown in Figure 7.33 may be suitable. Thespecimen is sealed in a permeameter so that air or water under pressure canFigure 7.33 Typical permeability equipment.

218 Durability testsbe applied to one face and the amount of fluid that permeates and emergesfrom the other face is measured. Leakage must be avoided and a constanthead must be maintained whilst measuring the fluid passing through thesample; precautions must also be taken to avoid evaporation losses fromthe reservoir when water is being used.Particular attention must be paid to the method of sealing the specimen.Whilst this may be reasonably easy for laboratory cast specimens, if theconcrete has been obtained from an in-situ location, by coring, the surfacewill be irregular. It is suggested that a specimen of this type can be castinto an epoxy ring which has an accurate outer surface and can then befitted, with the aid of neoprene seals, into a brass ring mounted in thepermeameter pot.The observed flow will be of the form shown in Figure 7.34, and thetest is usually continued to enable a steady state to be reached. The resultsfrom samples of similar dimensions tested under similar heads may beused comparatively to assess the characteristics of the interior of a body ofconcrete, and this approach is most likely to find applications in the fieldof water-containing structures.7.3.8 BS absorption testBS 1881: Part 122 (243) describes a standard absorption test to be carriedout on 75 mm diameter ±3mm cored specimens. The test is intended asa durability quality control check and the specified test age is 28–32 days.The apparatus required is simple, consisting only of a balance, an airtightvessel, a container of water and an oven, and the test is also straightforwardto perform.The procedure involves drying the measured core specimen, which shouldbe 75 mm long if the member thickness is greater than 150 mm, in an ovenFigure 7.34 Typical flow results.

Durability tests 219at 105 ± 5 C for 72 ± 2 hr. The specimen is cooled for 24 ± 1/2hr in anairtight vessel, weighed and immersed horizontally in the tank of water at20 ± 1 C with 25 ± 5 mm water cover over the top surface. It is immersedfor 30 ± 1/2 minutes, removed, shaken and dried quickly with a cloth toremove free surface water before weighing again.The absorption is calculated as the increase in weight expressed as apercentage of the weight of the dry specimen, and a length correction factoris applied if necessary. Some typical values of this factor which yield theequivalent absorption of a 75 mm long core are given in Table 7.5. Theresults are expressed to the nearest 0.1% and it is recommended that atleast three samples are tested and the result averaged. Results are classifiedin general terms in Table 7.4.The reliability of the method may be influenced by the effects of coring,and the measurements do not relate to the exposed concrete surface. Levitt(246) has also suggested that air trapped at the centre of the specimenaffects readings after more than 10 minutes’ immersion. He suggests thatresults taken at this time would be more reliable and may be related toinitial surface absorption values after 10 minutes, but the method is lesssensitive to concrete quality and the testing accuracy is also lower. Thesurface drying prior to weighing is particularly likely to influence results.If results are to be compared with a specified limit, or with other values,the age at test is particularly important. An appreciably higher value wouldbe expected at ages under 28 days, and lower values at greater ages, dueto a number of factors including the rate of hydration of the cement. Thiswill therefore be a characteristic of the particular mix used and limits thegeneral application of the method other than as a quality control test.7.3.9 ‘Sorptivity’ testThe mechanisms of water penetration into unsaturated concrete can beexpressed in terms of a critical material constant known as the sorptivity. Arange of procedures have been used by different investigators. DevelopmentTable 7.5 Water absorption correction factors(based on ref. 243)Length in mm of75 mm dia. specimenAbsorptioncorrection factor35 0.7350 0.8675 1.00100 1.09125 1.16150 1.20

220 Durability testsof this approach has been reviewed by Dias (261) who has focused particularlyon the critical issue of different drying preconditioning proceduresand concluded that specimen size and coating have relatively little effect.The test is based on measuring the weight change of a cylindrical sampleas water is absorbed through one of the flat faces. The opposite face isexposed to the atmosphere whilst the curved surface is sealed against waterpenetration. The sorptivity is derived from the slope of a plot of sampleweight against the square root of time. The effective porosity can also beobtained by allowing the sample to become saturated and attempts havealso been made to correlate sorptivity with permeability values.Although initial reported results were for laboratory cast specimens, slicesof cores can be used after suitable moisture conditioning.7.3.10 Capillary rise testA porosity test to assess the quality of high alumina cement concrete hasbeen described by the Institution of Structural Engineers (262). A piece ofconcrete at least 50 mm long is sawn flat on one face and placed with thissurface on a block of felt or blotting paper 10 mm thick. This is placed ina dish containing water to a level below the top surface of the felt, and thedish covered by an open-mouthed bell jar to prevent evaporation.Capillary rise is measured by visual observation of colour change, andreadings up to 8 hours and at 24 hours are suggested. The rise of waterin the specimen at these times can be related to percentage conversion ofthe concrete, with values of 15–20 mm at 24 hours quoted for unconvertedconcrete. This represents good quality, well-compacted concrete; 25–35 mmmay be expected for similar concrete if unfavourably converted. A highervalue may indicate serious deterioration. Whilst this method is describedin relation to one specific application, it may have other worthwhile applicationsof a comparative nature both in the laboratory and on site. Asimilar test has certainly been used to assess the presence of waterprooferin cementitious plasters and renders, for example.7.4 Tests for alkali–aggregate reactionReactions between the aggregates and alkaline matrix pore fluids are commonin some parts of the world, and have been the subject of widespreadresearch. A wide variety of such reactions exist which generally cause internalexpansion in the presence of moisture due to the formation of hydroscopicgel at the aggregate/matrix interface and within aggregate particles.In the UK the most common form experienced is alkali–silica reaction.Several tests exist to aid assessment of potential reactivity of materials,which are outside the scope of this book. In-situ tests related to diagnosis,and assessment of existing and future damage are not well developed

Durability tests 221although ultrasonic pulse velocity and pulse-echo techniques may offerpotential. Assessment of existing structures is primarily based on coreswhich may be subjected to expansion tests, alkali-immersion tests, cementcontent and petrographic (sawn surface and thin section) analysis (seeChapter 9). Visual identification of crack characteristics as well as a knowledgeof moisture content will also be important aids to diagnosis, which isconsidered in detail by a British Cement Association report (33).Expansion tests basically involve strain measurements on cores with atleast 4 Demec gauge spans along their length. The reaction may be acceleratedfor diagnosis purposes by a 38 C, 100% RH environment (33),although a number of variations on this are possible to obtain an indicationof likely future performance, especially the use of a test at 20 C.A ‘stiffness damage’ and other tests on cores have also been described inSection 5.1.3.6, and a technique of staining the alkali reaction gel with anultraviolet fluorescent dye to facilitate visual identification is discussed inSection 9.11.1.7.5 Tests for freeze–thaw resistanceThese usually take the form of petrographic analysis of a polished surfaceof a sample removed from the structure to determine entrained air content.ASTM C457 (263) covers this type of testing which is considered further inSection 9.11. Laboratory temperature cycling of samples may also be used.7.6 Abrasion resistance testingAbrasion resistance is likely to be of critical importance for floors of industrialpremises such as factories or warehouses where disputes involving veryhigh repair or replacement costs may often arise. The pore structure ofthe surface zone has been found to be the principal determining factor bySadegzadeh et al. (264) using an accelerated wear apparatus. This consistsof a rotating steel plate carrying three case-hardened steel wheels whichwear a 20 mm wide groove with 205 mm inside diameter in the concretesurface. Abrasion resistance is assessed by measurement of the depth of thisgroove after a 15-minute standardized test period. This leaves some damage,thus efforts have been made to correlate results to other non-damagingtechniques. Field studies using this equipment and rebound hammer testsare reported by Kettle and Sadegzadeh (265). These cover a range of practicalsituations and surface finishing techniques, and it is concluded thatthe relationship between rebound number and abrasion resistance is morecomplex than previously proposed by Chaplin (64).Results from the accelerated-wear equipment correlated well withobserved deterioration, and the classification in Table 7.6 is proposed forslabs in a medium industrial environment. Equipment of this type has been

222 Durability testsTable 7.6 Classification of concrete floor slabs in mediumindustrial environment (based on ref. 265)Quality of slabAbrasion depth (mm)Good 0.4standardized in BS EN 13894-4 (266) whilst classification limits for differenttypes of concrete floor surfaces are given in BS 8204: Part 2 (267).Relatively small differences in equipment characteristics can however leadto significant influence on the results obtained. The simple classificationsin Table 7.6 have recently been extended to deal with a wider range ofcircumstances. Kettle and Sadegzadeh (268) have also shown that the ISATis very sensitive to factors influencing abrasion resistance and may wellprovide an indirect non-destructive approach. Vassou and Kettle (269) havemore recently suggested, on the basis of laboratory studies, that impact testsusing the BRE screed tester (see Section 8.8) may be more reliable indicatorsof abrasion resistance. They also show, with particular reference to fibrereinforcedslabs, that a scratch hardness test is sensitive to both mix designand fibre inclusion, and is suitable for assessment of abrasion resistance.Provisional performance criteria are provided for all these techniques.Another abrasion resistance test with rolling wheels developed in Swedenfor use on specially prepared specimens of materials in the laboratory hasalso been outlined by Alexanderson (270).

Chapter 8Performance and integrity testsThe tests described in this chapter are wide-ranging and measure a varietyof different properties related to performance and integrity which havenot been considered elsewhere in the book. Many of the durability testsdescribed in Chapter 7 and chemical tests in Chapter 9 are also related toperformance, and load testing has been covered in Chapter 6. Integrity testsare generally concerned with the location and development of voids, cracksand other features which are not visible at the surface but may influencestructural or durability performance. The effect of thermal histories onstrength development is also considered.8.1 Infrared thermographyThe use of this technique, in which infrared photographs are taken duringthe cooling of a heated structure, was described by Manning and Holt(271) in 1980. Infrared thermography offers many potential advantagesover physical methods for the detection of delamination of bridge deckswhich are discussed in Section 8.3. The detection of laminations or voidsby infrared thermography is based on the difference in surface temperaturebetween sound and unsound concrete under certain atmospheric conditions.The concrete surface temperature changes throughout the day dueto heating by sunlight and cooling at night, this being particularly markedin summer. During the hottest part of the day the concrete temperaturedecreases with depth below the surface, whereas at night the situation isreversed. Delamination will affect these temperature gradients, and hencesurface temperatures, which may be compared by infrared measurements.Because of the small temperature differentials involved, the results cannotbe directly recorded by infrared film, and the image from the camera mustbe displayed on a cathode ray tube. The temperature of the surface is indicatedby a range of greys, although thermal contours can be automaticallysuperimposed, and colour monitors are widely used. This image is recordedon film to provide a hard copy, or digital cameras enable electronic datastorage.

224 Performance and integrity testsEarly trials have been performed with a camera held by an operatorstanding on the deck (271), but the limited field of view and oblique alignmentrender this impracticable. Greater success was obtained by using anelevated mobile platform to scan the deck from a height of up to 20 m,provided that the surface temperature differentials were greater than 2 C.Airborne testing avoided the necessity for lane closures and reduced thedifficulties of piecing together photographs to form a composite picture(the equipment was mounted on a helicopter and trial flights were made atheights up to 400 m) but the images obtained were often poor.It was found that surface temperature differentials existed in delaminateddecks at most times, but were greatest in summer, peaking in mid-afternoon.Attenuation of the reflected heat by the atmosphere, however, poses amajor problem, and this is increased by wind. Moisture on the surface wasalso found to mask surface temperature differences. While the potentialadvantages of airborne operation are considerable despite the high cost,there remain problems to be resolved before this technique can be consideredfully reliable and economical.Holt and Eales (272) have also described the successful use of thermographyto evaluate defects in highway pavements with an infrared scanner andcoupled real-life video scanner mounted on a 5 m high mast attached to avan. This is driven at up to 15 mph and images are matched by computer.Procedures for thermography in the investigation of bridge deck delaminationare given in ASTM D4788 (273).Other applications of infrared thermography to structural investigationsinvolve comparative assessment of concrete moisture conditions, which willinfluence thermal gradients, as well as location of hidden voids or ducts,services and other construction details (274). Techniques are also availableto detect reinforcing bars which are heated by electrical induction, whilstthere are currently many developments in ‘impulse’ thermography in whichthe surface is heated by a thermal radiator (275). Laboratory trials showthat voids up to 100 mm below the surface can be detected using digitalequipment after heating periods of about 5 minutes. It was noted thatshallow voids are best identified after a short cooling period of a few minutesand deeper voids after a longer cooling time of up to 1 hour. Techniquesfor quantitative assessment are under development using Finite Differencecomputer programs to aid interpretation of results.Recent development of 12-bit equipment has improved the sensitivity totemperature differences to within ±01 C or even better (Figure 8.1) andenabled high-definition imaging and accurate temperature measurement.Clark (276) has indicated that this has enhanced capabilities for detectionof bridge deck delamination in regions with low ambient temperatures anddescribes tests on the underside of decks from ground level. Wavelengthselection is also discussed. Studies of mosaic cladding to multistorey buildingshave been conducted to detect delamination and defective areas (277).

Performance and integrity tests 225Figure 8.1 Infrared thermography apparatus. (Courtesy of FLIR Systems Ltd).The location of ‘hot spots’ during daytime or cold spots at night are usedto identify where remediation is required. The smallest detectable defect isreported to be 200 × 200 mm. Studies of the insulation properties of buildings(274) using infrared imaging can also be used to reduce heat losses athot spots by identifying missing thermal insulation.8.2 RadarOver the past 20 years there has been an increasing usage of sub-surfaceimpulse radar to investigate civil engineering problems and, in particular,concrete structures (278). Electromagnetic waves, typically used in the frequencyrange 500 MHz–1 GHz for tests on concrete, will propagate through

226 Performance and integrity testsFigure 8.2 Investigation of sub-surface anomaly using radar.solids, with the speed and attenuation of the signal influenced by the electricalproperties of the solid materials. The dominant physical properties arethe electrical permittivity which determines the signal velocity, and the electricalconductivity which determines the signal attenuation (279). Reflectionsand refractions of the radar wave will occur at interfaces betweendifferent materials and the signal returning to the surface antenna can beinterpreted to provide an evaluation of the properties and geometry of subsurfacefeatures (Figure 8.2). Concrete Society TR48 (25) on radar testingof concrete was published in 1997 reflecting increased usage of the method.8.2.1 Radar systemsThere are three fundamentally different approaches to using radar to investigateconcrete structures.(i) Frequency modulation, in which the frequency of the transmitted radarsignal is continuously swept between predefined limits. The return

Performance and integrity tests 227signal is mixed with the currently transmitted signal to give a differencefrequency, depending upon the time delay and hence depth of the reflectiveinterface. This system has seen limited use to date on relatively thinwalls (280).(ii) Synthetic pulse radar, in which the frequency of the transmitted radarsignal is varied over a series of discontinuous steps. The amplitude andphase of the return signal is analyzed and a ‘time domain’ synthetic pulseis produced. This approach has been used to some extent in the field(281) and also in laboratory transmission line studies (282) to determinethe electrical properties of concrete at different radar frequencies.(iii) Impulse radar, in which a series of discrete sinusoidal pulses withina specified broad frequency band are transmitted into the concrete,typically with a repetition rate of 50 kHz. The transmitted signal isoften found to comprise three peaks (Figure 8.3), with a well-definednominal centre frequency.Impulse radar systems have gained the greatest acceptance for field usewhen testing concrete and most commercially obtainable systems are ofthis type, for example as shown in Figure 8.4. The power output of thetransmitted radar signal is very low and no special safety precautions areneeded. However, in the UK, a Department of Trade and Industry RadiocommunicationsAgency licence is required to permit use of investigativeradar equipment.8.2.1.1 Radar equipmentImpulse radar equipment comprises a pulse generator connected to a transmittingantenna. This is commonly of a dipole ‘bow-tie’ configuration,which is held in contact with the concrete and produces a diverging beamwith a degree of spatial polarization. A centre frequency antenna of 1 GHzis often used for investigation of relatively small concrete elements (up to500 mm thick), whilst a 500 MHz antenna may be more appropriate fordeeper investigations (283). A reduction in antenna frequency results in lesssignal attenuation and hence a deeper penetration capability, but has thedisadvantages of a poorer resolution of detail and the need for an antennawhich is physically quite large and cumbersome.An alternative to using surface-contact antennae is to use a focused beamhorn antenna with an air gap of about 300 mm between the horn and theconcrete surface. These systems have been used in the USA and Canada(236,284) to survey bridge decks from a vehicle moving at speeds up to50 km/hr, principally to detect corrosion induced delamination of the reinforcedconcrete slab. This technique is now well established and operationaldetails are provided in ASTM D4748 (285).

Figure 8.3 Radar reflections from a concrete slab (based on ref. 279).

Performance and integrity tests 229Figure 8.4 Digital impulse radar system.In the UK, specialized antennae have been developed for specific purposesby British Gas (286) and ERA (287), whilst a complete range of commercialgeneral purpose equipment is principally available from GSSI (288) andSensors and Software (289). A high frequency (4 GHz centre frequency)system is also available in the UK for localized surface zone inspectionsincluding delamination thickness assessments (290).8.2.2 Structural applications and limitationsIn addition to the assessment of concrete bridge decks described inSection 8.2.1, radar has been used to detect a variety of features buriedwithin concrete ranging from reinforcing bars and voids (291) to murdervictims (292). The range of principal reported structural applications issummarized in Table 8.1, although in practice usage is usually restrictedto the first five items on this list. Location of metallic prestressing ducts inpost-tensioned beams has been widely undertaken with particular success.

230 Performance and integrity testsTable 8.1 Structural applications of radar (based on ref. 279)GreatestReliability−−−−−−−−−−−−→LeastDetermine major construction featuresAssess element thicknessLocate reinforcing bars and metal ductsLocate moistureLocate voids/honeycombing/crackingLocate chloridesSize reinforcing barsSize voidsEstimate chloride concentrationsLocate reinforcement corrosionInterpretation of radar results to identify and evaluate the dimensionsof sub-surface features is not always straightforward. The radar ‘picture’obtained often does not resemble the form of the embedded features.Circular reflective sections such as metal pipes or reinforcing bars, for example,will present a complex hyperbolic pattern (Figure 8.5(a)) due to thediverging nature of the beam. The use of signal processing can simplify theimage (Figure 8.5(b)), but there are still three primary hyperbolas, caused bythe three peaks of the input signal, which result from a direct reflection fromthe circular feature as the antenna is scanned across the surface. Below thisare three secondary hyperbolas, resulting from a double reflection betweenthe embedded feature and the surface of the concrete. Evaluating the depthof a feature of interest necessitates a foreknowledge of the speed at whichradar waves will travel through concrete. This is principally determined bythe relative permittivity of the concrete, which in turn is determined predominantlyby the moisture content of the concrete. A value of the relativepermittivity must either be assumed or the concrete calibrated by localizeddrilling or coring.Typical relative permittivity values for concrete range between 5 (ovendryconcrete) and 12 (wet concrete). The propagation of radar throughany medium is governed by complex mathematical expressions and theseare documented more fully in ref. 25. However, for most civil engineeringsituations using non-magnetic concrete constituents the velocity of the radarsignal can be expressed by the simplified equationv =c 1/2rm/secwhere c = speed of light 3 × 10 8 m/sec r = relative permittivity of material.

Performance and integrity tests 231If the properties of materials are known precisely it may be possible tomake depth estimates to within about ±5 mm, but in practice uncertaintiesand concrete variability are likely to increase this range significantly.Work by the authors (293) using laboratory transmission line systems tosystematically examine the effects of a range of constituent materials, moistureconditions, temperatures and salt contents over a range of frequencieshave indicated that moisture content appears to be the dominant factorin determining the relative permittivity and conductivity of all concretes.Both of these properties increase with increasing moisture content, indicatingthe dominant effect of the concrete matrix rather than the aggregateswhere these are of natural origin. Manufactured lightweight aggregates willproduce different results due to increased moisture absorption. The effectof the concrete strength or of cement replacement materials is relativelysmall. The effect of external chloride contamination was investigated byusing a vacuum impregnation technique and was found to cause an increasein conductivity but to have a negligible effect on the relative permittivity.Figure 8.5 Radar scan over steel reinforcing bar: (a) original result; (b) processedresult.

232 Performance and integrity testsFigure 8.5 (continued).Temperature variations within the range 0–40 C were also observed to havelittle effect on the relative permittivity but an increase in temperature causeda small increase in the conductivity for concrete with high moisture content.Some typical results are shown in Figures 8.6 and 8.7, which illustrate thecombined effects of moisture and frequency, and of cement replacements.Because of the difficulties in interpreting radar results, surveys are normallyconducted by testing specialists who rely on practical experience,have knowledge of the fundamental principles involved and have a senseof realism concerning the likely limitations in a given practical situation.Experience has shown (294) that features such as voids can be particularlydifficult to detect if located very deep or beneath a layer of closely spacedsteel reinforcing bars. The use of neural network ‘artificial intelligence’ hasbeen used (295) to facilitate the interpretation of radar results and thesepreliminary studies are encouraging. Systems have been developed whichcan automatically locate and identify a reinforcing bar and can evaluate thedepth and diameter with reasonable accuracy (Figure 8.8).

Performance and integrity tests 23315Water % (by volume)0.10Water % (by volume)RELATIVE PERMITTIVITY1058.8%7.5%6.3%2.0%0%CONDUCTIVITY (S/m)0.058.8%7.5%6.3%2.0% 0%0 500 10000.000 500 1000FREQUENCY (MHz)FREQUENCY (MHz)Figure 8.6 Influence of moisture on relative permittivity and conductivity of concrete.Figure 8.7 Relative permittivity and conductivity for OPC, PFA, GGBS, and high strengthconcrete (HSC) specimens at 500 MHz.Radar reflects most strongly off metallic objects or off an interfacebetween two materials with very different relative permittivities. Thusan air-filled void in dry concrete rair = 1 rair dried concrete = 6 canbe quite difficult to detect, especially if the void is small. A 50 mm

234 Performance and integrity testsNeural cover (mm)300200100Neural diameter (mm)302010100 200 300Actual cover (mm)10 20 30Actual diameter (mm)Figure 8.8 Use of artificial neural networks to determine reinforcement location andsize.void thickness/width has been suggested as a practical lower limit tosize (291). Conversely, the same void if filled with water rwater =81 rwet concrete = 12 may be much easier to detect.Moisture in the concrete will not only influence the speed of the radarsignal and the strength of internal reflections, but in addition a high conductivitywill also cause a greater degree of signal attenuation, resulting ina reduction in the maximum depth of penetration possible. Using a 1 GHzantenna, typical practical penetration limits are around 500 mm for dryconcrete and 300 mm for water-saturated concrete. If the concrete is saturatedwith salt water, the limit of signal penetration is likely to be evensmaller.Experimental studies by the authors (294) on the ability to discriminateindividual reinforcing bars, and the effect of bar size and spacing on maskingof deeper features are illustrated in Figure 8.9(a,b).Antenna orientation is important due to polarization effects, and it hasbeen shown that for the optimum orientation relative to bar directionwith a 1 GHz antenna, individual bars can usually be identified when atspacings of about 200 mm or greater. At closer spacings depth below thesurface becomes important as shown in Figure 8.9(a) but bar size effects arenot clearly defined. Particular difficulties will be encountered with bars at100 mm centres or less at normal covers due to overlap of reflected signalsfrom adjacent bars. Positioning accuracies of the order of ±10 mm may bepossible where clearly defined signals are obtained. The ability to detect atarget below a layer of reinforcement near to the surface was made moredifficult by larger diameter and more closely spaced bars in the layer. Forpractical purposes 100 mm may be regarded as the limiting spacing to avoidmasking when small diameter bars are present, rising to 200 mm for largerbars. Bars running parallel to the scan direction will have a relatively minoreffect on ability to detect transverse bars or other buried features.

Performance and integrity tests 23530036Cover (mm)200100BARSUNRESOLVEDBARSRESOLVEDUpper bar diameter (mm)2412MASKINGNOMASKING00 100 200 300Bar spacing (mm)(a) Detection of individual bars00100 200 300Bar spacing (mm)(b) Masking of deeper target byreinforcement near the surfaceFigure 8.9 Radar measurement of steel reinforcing bars (ref. 294 with permission ofElsevier).Identifying a second mesh of bars below the surface layer poses particulardifficulties of interpretation. This has recently been addressed by Taffe andMaierhofer (296) who describe a system which can generate results in theform of a C-scan at different levels below the surface although this involvesmany measurements over a localized grid and is time-consuming, with theneed for specialized software. A clear advantage of the C-scan form of datapresentation is that the processed radar image bears a visual resemblanceto the embedded target as illustrated in Figure 8.10. This is not the casewith more conventional presentation of radar results.Interpretation can be further assisted by the use of 2- and 3-dimensionalnumerical modelling software using the finite difference approach (297),some of which has been made freely available on the Internet byGiannopoulos (http://www.gprmax.org) (298). This enables anticipatedresponses to be predicted for particular test configurations. Simple raytracingmodels are also available to predict the likely success of a radarsurvey in some situations, particularly where embedded targets are isolatedor widely spaced.Other recent research to assist interpretation includes detailed studiesof antenna characteristics, including the influence of coupling effects oneffective signal centre frequency, which is found to decrease significantlyin concrete, and beam spread, which is found to decrease in concrete containingmore moisture (299). Bungey (300) has recently reviewed many ofthese developments and has provided an extensive list of key references,including work on identification of air voids in plastic post-tensioning ducts.

236 Performance and integrity testsYTXT Rtime sliceFigure 8.10 Time slice (C-scan) presentation of radar measurements (Courtesy ofC. Kohl, BAM).Theoretical and experimental studies are also underway at several centresto attempt to assess concrete dielectric properties in-situ and assess moistureconditions (301,302).8.3 Dynamic response testingTesting ranges from simple localized surface tapping, to large-scale investigationof the response of structures to applied dynamic loads. Simplemethods are useful for assessing localized integrity, such as delamination,whilst more complex methods are commonly used for pile integrity testing,determination of member thicknesses, and examination of stiffnessesof members affected by cracking or other deterioration.8.3.1 Simple ‘non-instrumented’ approachesExperience has shown that the human ear used in conjunction with surfacetapping is the most efficient and economical method of determiningmajor delamination. Delamination may be caused in bridge decks (303) bycorrosion expansion of the top layer of reinforcement causing a fractureplane at that level, and is especially likely where cover is low. In order tocarry out repairs, the deterioration must be detected and its extent determined.Methods currently in routine use are based on the characteristic dullsound when the deck surface is struck over a delamination. Hammers, steelbars and chains are sometimes used to strike the deck and the sound producedis assessed subjectively by the operator. The operation is performed

Performance and integrity tests 237over a grid to ‘map out’ the deck. These manual impact techniques havemany disadvantages, and ‘chain-dragging’ provides a quicker and moreefficient approach. Early methods used a single heavy chain with 50 mmlinks, weighing 2.2 kg/m, with a 2 m length passed over the surface with awhip-like action. More recent usage involves a set of 4 or 5 segments of a6 mm diameter link chain about 460 mm long attached to a 600 mm bar atintervals along its length and drawn across the surface. Sound differencesmay be detected by the operator and a delamination profile marked. Proceduresfor using sounding techniques for locating localized delaminationin bridge decks are given in ASTM D4580 (304).These approaches, however, rely upon a subjective judgement by theoperator to differentiate between sound and unsound regions and resultscannot readily be quantified. An automated chain drag system involving atrolley with microphone and data acquisition system is however recentlyavailable (305). An automated tapping system is also available (306) butless developed.8.3.2 Pulse-echo techniquesVarious attempts have been made to monitor electronically the response toa blow applied to the concrete surface, and have led to pulse-echo techniqueswhich are now established in a variety of forms. Carino has recentlythoroughly reviewed the range of available stress wave propagation methods(307) which are briefly outlined below. The simplest version measuresthe amplitude of the reflected shock waves caused by a surface hammerblow as illustrated in Figure 8.11. If the path of the waves is short a highreading will be obtained, and the measured value will decrease as the pathFigure 8.11 Delamination detection.

238 Performance and integrity testslength increases. The readings can only be used comparatively to detectchanges in path length due either to variation in member thickness or tomajor internal crack planes.Standardization of the impact has been one of the major obstacles toquantitative assessment. The associated equipment for processing the signalobtained from the receiving transducer is complex and has been developedfor application to the location of voids, ducts, honeycombing and other‘flaws’ in structural elements. Success has been achieved in the laboratoryand reports of site usage are increasing. A ‘thickness gauge’ is available inthe UK.Differentiation should be made between two basic types of test.(1) Impulse response testing in which the concrete surface is impacted bya relatively heavy hammer (typically 1 kg) with a load cell, which maybe built-in, to record details of the blow. Response is usually measuredby a geophone located adjacent to the impact position. Stress wavesare generated through the element, with the maximum compressivestress at the impact point related to the elastic properties of the hammertip. In reviewing the development of this technique Davis (308)reports values ranging from 5 MPa for hard rubber to 50 MPa for aluminiumtips. Whilst originally developed for the time-domain testing ofpiles (see Section 8.3.2.1), the advent of digital technology has greatlyimproved data storage and analysis capabilities, including frequencyanalysis, leading to application to a much wider range of problems. Inthe USA, plate-like structures such as slabs, bridge decks and walls havebeen successfully inspected with results presented in the form of contourmaps. Features detected include voiding beneath slabs, delamination,cracking and similar through study of dynamic stiffness, mobility anddamping. A 1 kg hammer will typically produce a zone of influenceof the order of 600 mm with frequencies in the range 0–1 kHz, leadingto typical test grid spacings of 300–900 mm. Davis (308) presentsseveral field case studies of this approach using commercially availableequipment (309).(2) Impact-Echo testing is more recently developed, and involves impactcontact stresses approx 10 times less than for Impulse-Response andwith frequencies in the range of about 3–40 kHz. Penetration is thusmuch less and interpretation is more localized and based on frequencydomain analysis. Impact is normally by ball-bearing or similardevice, with frequency affected by the size and contact time (310) ofthe impactor. Sansalone and Streett (311) have described the developmentof this technique, which can be applied to a range of situations(see Section 8.3.2.2), in some detail, including backgroundtheory.

Performance and integrity tests 2398.3.2.1 Time domain measurementThe pulse-echo technique is most widely developed for the testing of concretepiles, and is based on the analysis of reflected pulse traces to detectdefects or varying soil support conditions. This is Impulse-Response testingas outlined above. A single blow is applied by hammer to the pile top andthe subsequent movements, including reflected shock waves, are detected byan accelerometer. This is connected to a signal processor which integratesthe signal, and a trace of vertical displacement against time can be displayed.The basic equipment and features of a typical trace are shown in Figure 8.12.The trace may be used for reliable assessment of pile length and to detectany defects or lack of uniformity of the pile, which will show as distortionsto the trace. These may be interpreted by a specialist to determine theirtype and position. Equipment can record and store several hammer blowsto check signal consistency, and the traces may be recorded by camera ordigitally to provide hard copies as required. The method is simple and cheapto use, but cannot determine the cross-sectional area of a pile or its bearingcapacity.This method is most suitable for permanently cased piles or for singlelengths of precast piles. The maximum length which can be tested dependsupon the ability to produce well-defined peaks due to reflections from thetoe. Considerable damping will be provided by the surrounding soil andan approximate limiting length/diameter of 30 is often assumed for piles inmedium clay. Defects will cause intermediate but lower peaks although itis difficult to differentiate between bulbs and necking.A wide range of systems is commercially available for this application,some involving up to a 6 kg hammer falling through a height of 1 m withFigure 8.12 Pulse-echo method; time domain measurement.

240 Performance and integrity testsvarious electronic refinements of the signal and trace presentation formats.Practical application of the method has been described (312) with a rate of50–100 piles a day claimed, and the approach has gained widespread acceptancefor quality control and integrity testing of piles. Whilst use continuesin Europe, it has recently declined in the USA in favour of cross-hole soniclogging methods involving tomography (313) and 3D visualization (314).Time domain measurements have also been used to detect voids in concreteslabs (73) using ultrasonic pulse-echo techniques. The presence of anair void is seen as a faster return of the reflection signal (Figure 8.13).8.3.2.2 Frequency domain measurementWhilst the direct measurement of pulse echo in the time domain is appropriatefor relatively large structural elements, such as pile foundations,Figure 8.13 Imaging of defect in concrete using ultrasonic pulse-echo technique: (a) Crosssection of test specimen; (b) B-scan (based on ref. 73).

Performance and integrity tests 241measurements of thin concrete elements such as slabs or walls result inshort-duration, multiple echoes that are time-consuming to measure andcan be difficult to interpret. A quicker and simpler approach is to carryout a frequency analysis of the reflection signal. The time domain signalis transformed into the frequency domain using the fast Fourier transformtechnique. Features such as voids or delamination can be detected by a shiftin the amplitude of the higher frequency components of the return signal(Figure 8.14).The presence of a layer of reinforcing bars will result in a high-frequencysignal component, resulting from a reflection from the upper surface of thesteel bars (315). By reducing the duration of the input impulse the effect ofthese bars can be enhanced. By increasing the duration of the impulse thebars can be made ‘invisible’, thus facilitating the detection of voids beneaththe reinforcement. The use of a long-duration impulse does however limitthe resolution of voids detectable to sizes greater than the size of the bardiameter.Impact-echo methods can also be used to detect a loss of contact betweenthe upper surface of a reinforcing bar and the surrounding concrete, but itis difficult to detect a loss of contact on the underside of a bar (315).ForceImpactDisplacementContactTimeTimeRecieverTime0.5mRelativeAmplitudeSolid slabRelativeAmplitudeVoid in slab010 20 30Frequency (kHz)010 20 30Frequency (kHz)Figure 8.14 Frequency domain analysis of pulse-echo signal (based on refs 318and 320).

242 Performance and integrity testsCommercial impact-echo systems are available from more than one manufacturerfor field use, and one of these is shown in Figure 8.15 (316). Ithas been used in the USA to monitor the thickness and integrity of concretepavements and walls and also for detecting cracking in beams (317). Usingan impact duration of 20–80 ps, sections up to 1 m thick have been studied.Studies on the use of impact-echo techniques to detect air voids resultingfrom inadequate grouting of steel prestressing ducts are promising (318)including site usage (319), and studies (320) have also investigated the useof neural network ‘artificial intelligence’ to facilitate the interpretation ofcomplex frequency domain results. Research has also been reported into arange of other possible applications including evaluation of early age concretestrength (321), damage assessment of bridge decks (322) and otherelements (323), and bond of overlays (324). Potential improvements in analysisbased on signal processing and numerical modelling have also recentlybeen reported by Abraham (325). It should be noted that most successfulapplication of the approach is on plate-like elements with interpretationbecoming increasingly more difficult as member geometry becomes morecomplex. It is however a technique which is likely to see further refinementand increased field use in the future, and has been included in the HighwaysAgency guidance notes (26) for detecting voids in prestressing ducts.Figure 8.15 DOCter impact-echo apparatus (photograph by courtesy of GermannInstruments).

Performance and integrity tests 2438.3.3 Analysis of surface waves(1) Spectral analysis. The use of fast Fourier transformation of impulseresponses has also been used for spectral analysis of surface waves(SASW) that are transmitted laterally through a medium (326). A pairof surface transducers are positioned adjacent to the impulse location(Figure 8.16) and are used to measure variations in the surface wavevelocity at different frequencies. The results are then further analysedto determine both the thickness and elastic modulus of the materialand of underlying layers. This technique has the potential to detectinterlayered good and poor quality concrete, but the complex signalprocessing required has resulted in infrequent usage in the field. Effortshave been made to automate the analysis of results (327).This is a relatively recent technique which is the subject of muchongoing research worldwide, including theoretical principles of surfacewave propagation arising from a range of possible sources. The optimumcharacteristics of the source will be influenced by the stiffnessesof the layers and the depth to be investigated, as well as the spacingbetween the two receivers. Potential applications, again to plate-likestructures, considered in laboratory studies include assessment of deterioratedconcrete (328) and in-situ compressive strength (329).(2) Time of flight crack depth assessment. Some recent developments,involving theoretical and laboratory studies, focus on the use of surfacewaves to determine the depth of surface breaking cracks. Proceduresinvolve impact source and receiver equidistant on opposite sides of thecrack (330) which is similar to the procedure using ultrasonics shownin Figure 3.20, or receivers equidistant on either side of the crack withan impact on each side also equidistant from the receivers (331). Thislatter approach allows wave transmission measurements across a rangeHammerSpectralanalyserReceiver 1 Receiver 2d >xxFigure 8.16 Spectral analysis of surface waves.

244 Performance and integrity testsof frequencies to be self-compensated. Both of these techniques arelikely however to be affected by the practical difficulties encounteredby ultrasonic assessment of crack depth discussed in Section 3.4.2.2.8.3.4 Testing large-scale structuresDynamic response testing of entire structures may similarly involve hammerimpacts or application of vibrating loads. In either case the responseis recorded by carefully located accelerometers and complex signal processingequipment is required. Maguire and Severn (332) have described theapplication of hammer blow techniques to a range of structures includingchimneys and bridge beams. Vibration methods have been reported by anumber of authors including Williams (333) and Maguire (334) and arecommercially available.Modal analysis and dynamic stiffness approaches have been under developmentin the UK and elsewhere but are not yet widely used on site. Onerecent report (335) uses model reinforced concrete beam and slab bridgedecks to assess the effects of incremental overloading, but identifies limitationsin procedures for comparing vibration modes for different damagestates.The aim of testing will usually be to monitor stiffness changes due tocracking, deterioration or repair. The ‘dynamic signature’ of a structure maybe obtained and compared at various time intervals to monitor changes. Thisis particularly valuable when monitoring the effects of a suspected overloador deterioration with time, and delamination of screeds or toppings fromslabs may be detected. Comparisons may in some cases also be possiblewith theoretically calculated frequencies of vibration and member stiffness.A range of vibration and ‘shock’ methods is also regularly used for testingpiles. The basic principles have been described by Stain (336) and permitestimates with varying degrees of confidence of dynamic pile head stiffness,cross-sectional area and limiting stiffness values of end-bearing piles.As with pulse-echo techniques, the principal advantage is speed of test incomparison with traditional static methods.8.4 Radiography and radiometryRadioactive methods have developed steadily over recent years, andalthough generally expensive with important safety issues and more appropriateto laboratory conditions, their field applications are increasing innumber. There are three basic methods currently in use for testing concrete:X-ray radiography, -ray radiography, and -ray radiometry.The radiographic methods consist essentially of a ‘photograph’ takenthrough a specimen to reveal a picture of the interior, whereas radiometry

Performance and integrity tests 245involves the use of a concentrated source and a detector to pick up andmeasure the received emissions at a localized point on the member.X-rays and -rays are both at the high energy end of the electro-magneticspectrum and will penetrate matter, but undergo attenuation according toits nature and thickness. The energies of these rays are expressed in electronvolts, and the sources available for concrete testing will typically be in therange 30–125 keV for X-rays and 0.3–1.3 MeV for -rays. The attenuationof radiation passing through matter is exponential and may be expressedin simplified terms asI = I 0 e −mwhere I = emergent intensity of radiationI 0 = incident intensity = mass absorption coefficientm = mass per unit area × thickness of material traversed.In the X-ray energy range, attenuation is dependent on both the atomicnumber and density of the material, whilst for the -ray range, density isthe principal factor. The cost and immobility of X-ray equipment, whichrequires high voltages, has been a major limitation to the development offield usage although the method is of considerable value in the laboratory.Sources of -rays are more easily portable, and so this has become theprincipal radioactive method for on-site use.BS 1881: Part 205 (337) provides guidance on radiographic work, includingsuitable sources of radiation, safety precautions and testing procedures.Forrester (338) has described the techniques in detail.8.4.1 X-ray radiographyLaboratory applications of this approach have been principally aimed at thestudy of the internal structure of concrete, and are summarized by Malhotra(63). Studies have included the arrangement of aggregate particles, includingtheir spacing and paste film thicknesses, three-dimensional observation ofair voids, segregation and the presence of cracks. Although some fieldapplications were reported in the 1950s, little attention seems to have beenpaid to the use of X-rays in the field since that time, apart from the use oflorry-mounted high energy 8 MeV Linac equipment capable of penetratingup to 1600 mm of concrete (339).Concern over corrosion of post-tensioned prestressing strands as a resultof inadequate grouting has led to a recent focus of interest in systems capableof investigating the interior of steel prestressing ducts of bridge beams.The Scorpion II system, developed in France, comprises a miniature 4 MeVlinear accelerator which produces a beam of high-powered X-rays (340).

246 Performance and integrity testsA lorry-mounted system is used to investigate the integrity of post-tensionedbridge beams by tracking the accelerator on one side of a beam web andlocating a detector on the other side. Results are obtained as real-time videodisplay. However, the risk of backscatter radiation reflected back from theconcrete surface has necessitated a 250 m exclusion zone below bridges tosafeguard the health of pedestrians and boatmen.8.4.2 Gamma radiographyThe use of -rays to provide a ‘photograph’ of the interior of a concretemember has gained considerable acceptance. It is especially valuable fordetermining the position and condition of reinforcement or prestressingsteel, voids in the concrete or grouting of post-tensioned construction, orvariable compaction.The source will normally be a radioactive isotope enclosed in a portablecontainer which permits a beam of radiation to be emitted. The choice ofisotope will depend upon the thickness of concrete involved: Iridium 192 for25–250 mm thickness and Cobalt 60 for 125–500 mm thickness are mostcommonly used. The beam is directed at the area of the member under investigationand a photograph produced on a standard X-ray film held againstthe back face. In cases where particularly precise details are required, suchas specific identification of reinforcing bars or grouting voids, the image canbe intensified by sandwiching the film between very thin lead screens. Afterdevelopment of the film, reinforcement will appear as light areas due tothe higher absorption of rays by the high-density material, whilst voids willappear as dark areas. If it is required to determine the size and position ofreinforcement or defects, photogrammetric techniques can be used in conjunctionwith stereoscopic radiographs. A multi-incidence angle techniquewith digital detection using Cobalt 60 has recently been trialled in the fieldfor detection of multi-layered reinforcement (341). This removes the needfor film and produces a 3-dimensional image using computed laminographytomo-synthesis algorithms. Although this technique has become establishedfor examination of steel and voids, it is expensive and requires stringentsafety precautions. It is also limited by member thickness; although 600 mmis sometimes quoted as an upper limit, for thicknesses greater than 450 mmthe exposure times become unacceptably long.Studies on the use of in-situ Cobalt 60 sources (1.17 and 1.33 MeVgamma rays) for bridge beams between 300 and 800 mm thick have reportedthat exposure times of several hours may be required. An alternative is to usea 6 MeV Betatron as a switchable X-ray source, providing a higher-powerand more effective radiographic facility (342). This has enabled exposuretimes to be reduced to 15–30 minutes for a section of 600 mm thickness.A portable system is currently commercially available with reported exposuretimes of up to 20 minutes for 750 mm thickness using 7.5 MeV and a

Performance and integrity tests 247patented image capture system (343) requiring only limited access restrictions.This is potentially suitable for testing post-tensioned concrete for ductgrouting and wire breaks.8.4.3 Gamma radiometryAs in radiography, -rays are generated by a suitable radioisotope anddirected at the concrete. In this case, however, the intensity of radiationemerging is detected by a Geiger or scintillation counter and measured byelectronic equipment. This approach will primarily be used for the measurementof in-situ density of the concrete, although it may possibly alsobe applied to thickness determination.As the high-energy radiation passes through concrete some is absorbed,some passes through completely, and a considerable amount is scatteredby collisions with electrons in the concrete. This scattering forms the basisof ‘backscatter’ methods which may be used to examine the properties ofmaterial near the surface as an alternative to ‘direct’ measurements of theenergy passing through the member completely.The first use of this approach seems to have been in the early 1950s primarilyapplied to highway applications. More recently the backscatter techniquehas developed and applications widened. Mitchell (344) has revieweddevelopments and techniques in detail (together with radiography).8.4.3.1 Direct methodsA variety of test arrangements have been adopted, but the basic equipmentconsists of a suitably housed radioactive source, similar to that used forradiography, together with a detector. The detector will usually consist ofa counter housed in a thick lead sheath to exclude signals other than thosecoming directly from the source. The radiation beam may pass directlythrough the concrete member as shown in Figure 8.17 or alternativelyFigure 8.17 Direct radiometry.

248 Performance and integrity testssource and/or detector may be lowered into predrilled holes in the body ofthe concrete if density variations with depth are required.Calibrations may be made by cutting cores in the path of radiation aftertest and using these to provide samples for physical density measurements.As with radiography, the thickness of concrete tested is limited to about600 mm, which poses a serious restriction to the use of the method. Asan alternative to measuring density, this approach may also be adapted toassess member thickness or the location of reinforcing bars.Computerized tomography analysis methods are being studied (345) todetermine if accurate mapping of internal features of concrete elementscan be derived from the differential absorption of gamma rays. This is aspecialist technique that has yet to be used on full-sized structural elements,but initial laboratory studies are promising.8.4.3.2 Backscatter methodsIt is generally considered that this method tests the density of the outer100 mm of concrete. The -ray source and detector are fixed close togetherin a suitably screened frame which is placed on the concrete surface. A typicaldevice of this type has source and detector angled at approximately45 to the surface and at a spacing of approximately 250 mm. In this casethe rays propagate through the concrete at an angle to the surface and theintensity of radiation returning to the surface at this fixed distance from thesource is measured as illustrated in Figure 8.18.If density measurements are required at a greater depth, a device consistingof a screened source and detector assembly which may be lowered into asingle borehole may be used. Calibrated gauges have been developed basedon the backscatter technique to permit quantitative assessment of the bulkdensity of concrete, but the results obtained from other forms of equipmentcan be used for comparative measurements. While this method appears tobe simple, difficulties may arise from the non-uniform radiation absorptionFigure 8.18 Backscatter radiometry.

Performance and integrity tests 249characteristics of concrete, and inaccuracies may result where there is adensity gradient. A number of instruments using the backscatter methodare commercially available, with provision also for near-surface direct measurementsup to 300 mm deep, with a drilled hole; some instruments alsopermit neutron moisture measurement (see Section 7.2.2).8.4.3.3 Limitations and applicationsA major problem with -radiometry is the expense of the detecting equipment,which requires skilled personnel for its operation, although this hasbeen reduced with the development of commercially available gauges. Scintillationdetectors have been favoured in recent years because of the highermaximum count rate which permits the use of a less intense source. Thedirect method offers the greatest versatility, despite path restrictions andcan be used to detect reinforcement or member thickness in addition todensity measurements. Backscatter methods are limited to surface densitymeasurements in most cases. Although valuable in specialized in-situ situationswhen large numbers of repetitive measurements are required, or as aquality control method for precast construction, radiometry is unlikely toreplace conventional gravimetric methods of density determination whenthe number of specimens is small.8.5 Holographic and acoustic emission techniquesAttempts have been made in recent years to apply these established techniquesto the examination of concrete structures. Both provide means ofmonitoring the initiation of cracking under increasing load. Successful applicationof holography has so far been reported only for laboratory use, butmay be further developed for in-situ load test monitoring.8.5.1 Holographic techniquesHolographic techniques provide a method of measuring minute surface displacementsby examination of the fringe patterns generated when the surfaceis illuminated by a light beam and photographed under successive loadingconditions, and the results superimposed. Application of these methods toconcrete models in the laboratory indicates that holographic interferometry,which is capable of measuring out-of-plane displacements of the order ofone wave-length of the light used, and speckle holography, which can detectin-plane movements of less than one wavelength, are the most useful. Inboth cases the elimination of vibrations and rigid body motions is essential.Whilst the latter can be achieved by rigidly fixing the optics to the memberunder test, the elimination of vibrations poses a serious problem even under

250 Performance and integrity testslaboratory conditions. Although these methods may permit the examinationof crack development in the laboratory, the equipment is complex andthe practical problems associated with site usage appear to be difficult toovercome. Recent developments in laser technology may however help inthis respect.8.5.2 Acoustic emission8.5.2.1 TheoryAs a material is loaded, localized points may be strained beyond theirelastic limit, and crushing or microcracking may occur. The kinetic energyreleased will propagate small amplitude elastic stress waves throughout theelement. These are known as acoustic emissions, although they are generallynot in the audible range, and may be detected as small displacements bytransducers positioned on the surface of the material.An important feature of many materials is the Kaiser effect, which isthe irreversible characteristic of acoustic emission resulting from appliedstress. This means that if a material has been stressed to some level, noemission will be detected on subsequent loading until the previously appliedstress level has been exceeded. This feature has allowed the method to beapplied most usefully to materials testing, but unfortunately Nielsen andGriffin (346) have demonstrated that the phenomenon does not alwaysapply to unreinforced concrete. Concrete may recover many aspects of itspre-cracking internal structure within a matter of hours due to continuedhydration, and energy will again be released during reloading over a similarstress range.More recent tests on reinforced concrete beams (347) have shown thatthe Kaiser effect is observed when unloading periods of up to 2 hours havebeen investigated. However, it is probable that over longer time intervalsthe autogenic ‘healing’ of microcracks in concrete will negate the effect. The‘felicity ratio’ – defined as the load at which emissions start/previous maximumload – may nevertheless be a useful indicator of increasing damagewith the value decreasing as the material approaches failure.8.5.2.2 EquipmentThe signal detected by the piezo-electric transducer is amplified, filtered,processed and recorded in some convenient form (Figure 8.19). An array oftransducers is located on the structure or element to encompass the areasof interest.Specialist equipment for this purpose is available worldwide from severalmanufacturers as integrated systems and lightweight portable models maybe used in the field. The results are most conveniently considered as a plot

Performance and integrity tests 251Figure 8.19 Acoustic emission equipment.Figure 8.20 Typical acoustic emission plot.of emission count rate against applied load (Figure 8.20) although moredetailed interpretation is possible if amplitude and frequency values are alsorecorded. This has been greatly facilitated by the development of digitalequipment.8.5.2.3 Applications and limitationsIt has been reported (63) that as the load level on a concrete specimenincreases, the emission rate and signal level both increase slowly and consistentlyuntil failure approaches, and there is then a rapid increase upto failure. Whilst this allows crack initiation and propagation to be monitoredduring a period of increasing stress, the method cannot be usedfor either individual or comparative measurement under static load conditions.Colombo (348) has however recently reported field trials usingnormal bridge traffic to permit comparisons of the performance of individualbeams for short-term condition survey tests. Differentiation between‘primary’ activity due to new crack development and ‘secondary’ activity

252 Performance and integrity testsdue to friction within pre-existing cracks can be made on the basis of signalenergy. A key factor in the success of acoustic emission testing (includinglong-term acoustic monitoring (349) for features such as prestressing wirebreaks) is the correct setting of threshold levels to filter out backgroundnoise, as well as sensor calibration. In the UK the Highways Agency arecurrently developing guidance notes for applications to bridges as part ofref. 26.Interest in the technique applied to concrete has received a significantincrease in the past 20 years, most notably in Japan and more recently inthe UK. Ohtsu (350) provided a comprehensive review in 1996 and workhas continued (predominantly in the laboratory) to develop interpretationprocedures. General principles of acoustic emission are provided by BSEN 13554 (351).Mindess (352) has shown that mature concrete provides more acousticemission on cracking than young concrete, but confirms that emissions donot show a significant increase until about 80–90% of ultimate stress. Thepossible absence of the Kaiser effect for concrete effectively rules out themethod for establishing a history of past stress levels. Several authors includingBalazs (353) have, however, described laboratory tests which indicatethat it may be possible to detect the degree of bond damage caused by priorloading (both static and dynamic) if the emissions generated by a reinforcedspecimen under increasing load are filtered to isolate those caused by bondbreakdown, since debonding of reinforcement is an irreversible process.Titus et al. (354), Idriss (355) and Lyons et al. (356) have all suggested thatit may be possible to detect the progress of microcracking due to corrosionactivity. Long-term creep tests at constant loading by Rossi et al. (357)have shown a clear link between creep deflection, essentially caused bydrying shrinkage microcracking, and acoustic emission levels (Figure 8.21).However, in tests on plain concrete beams, together with fibre-reinforcedSTRAINEventsStrainStrainEventsCONCRETE 1CONCRETE 2NUMBER OF ACOUSTICEVENTS0Time (hours)60Figure 8.21 Acoustic emission tests on creep specimens (based on ref. 357).

Performance and integrity tests 253and conventional steel-reinforced beams, Jenkins and Steputat (358) perhapssurprisingly concluded that acoustic emission gave no early warningof incipient failure.The application to concrete of acoustic emission methods has not yet beenfully developed, and as equipment costs are high they must be regarded asessentially laboratory methods at present although field use is beginning todevelop. However, there is clearly future potential for use of the methodin conjunction with in-situ load testing as a means of monitoring crackingorigin and development and bond breakdown, and to provide a warning ofimpending failure. Monitoring corrosion development and wire fracture isalso potentially valuable.8.6 Photoelastic methodsPhotoelastic and similar techniques for tests on concrete members will berelated to load testing or monitoring (Chapter 6). Photoelastic coatings maybe bonded to the concrete surface and will develop the same strains and actlike a continuous full-field strain gauge. Commercially produced sheets of1–3 mm thickness are available for flat surfaces, and ‘contoured’ sheetingmay be produced to fit curved surfaces. A portable reflection polariscopecan be used to produce an overall strain pattern, and it is suggested thatmeasurement of fringes by an experienced operator can yield concrete stresssensitivities of ±015 N/mm 2 in the laboratory or ±035 N/mm 2 under fieldconditions.Moire fringe techniques, in which two sets of equidistant parallel linesare placed close together and their relative movement measured, are alsoavailable. A grid is cemented to the structure and compared with a masterby reflected light. The technique may be used in similar situations to a photoelasticcoating but is restricted to flat surfaces and has a lower sensitivitythan photoelastic methods.The principal use of these methods for the testing of concrete in structureswill be to determine patterns of behaviour at stress concentrations, or incomplex localized areas under an increasing load or with time. They arebest suited to a laboratory environment, although field use is possible.8.7 Maturity and temperature-matched curingThese two techniques may be useful when attempting to monitor performancein terms of in-situ concrete strength development for timing of safeformwork or prop removal, application of loading (including pre-stress)or some similar purpose. As indicated in Section 1.4.3, there is particulargrowth of interest related to fast-track construction. Companion test cubesor cylinders stored in air alongside the pour will not experience the sametemperature regime as the concrete in the pour, and will usually indicate

254 Performance and integrity testslower early age strength values. Even if covered by damp Hessian or plasticsheeting, test specimens in steel moulds will remain close to air temperaturewhile in-situ concrete temperatures may commonly rise initially by up to 20or 30 C within the first 12 hours after casting and remain above ambient forseveral days. The effect of this is particularly critical in cold weather, whenthe true in-situ strength may be seriously underestimated by companionspecimens. This has recently been illustrated by Hulshizer (359). Maturityand temperature-matched curing approaches both involve measurement ofwithin-pour temperatures to overcome this problem, but location of testpoints is critical because of the internal temperature variations which willexist.8.7.1 Maturity measurementsStrength development is a function of time and temperature, and for a particularconcrete mix and curing conditions this may be related to a maturityfunction to yield a maturity index which quantifies these combined effects,and can be used to estimate in-place strength. Several different maturityfunctions have been proposed. The simplest commonly used expressionassumes that the initial rate of strength development is a linear function oftemperature. Maturity is thus defined as the product of time with temperatureabove a predetermined datum and is commonly calculated asMt = T a − T 0 twhere Mt is maturity in degree-hours or degree-dayst is time interval in hours or daysT a is average concrete temperature during time intervalT 0 is datum temperature.This is known as the Nurse-Saul function, and the datum is the temperatureat which concrete is assumed not to gain strength with respect to time andis commonly taken as −10 C.The approach is detailed by ASTM C1074 (360) and multi-channel equipmentis commercially available to monitor in-situ temperatures by means ofthermocouples or thermistors embedded in the concrete. Simple chemicalbaseddevices are also available (316), as shown in Figure 8.22, which areinserted into the surface of the newly placed concrete but these offer a lowerlevel of precision.Unfortunately correlation between maturity and strength, as illustratedin Figure 8.23, is specific to a particular concrete mix and curing regimeand maturities cannot be used alone because of the risk of variations inthe mix composition and the possibility of insufficient water available insitufor complete cement hydration. Maturity measurements must thus be

Figure 8.22 ‘Coma’ mini maturity meter.Figure 8.23 Typical strength–maturity results.

256 Performance and integrity testsbacked up by some other form of strength testing, but do provide a usefulpreliminary indicator of strength development.Carino (361) discusses in great detail the development and use of concretematurity expressions to estimate early age strengths and treatment in thischapter is thus limited. He concludes that the relationship between Mtand strength for a given concrete mix is sensitive to the early age curingtemperature and that a higher temperature will result in a higher early agestrength but a lower long-term strength (Figure 8.24). However, there isa unique relative strength vs. maturity relationship which can be used topredict accurately the early age strength as a proportion of the final longtermstrength or of the 28-day strength. Estimation of longer-term strengthsfrom early age in-situ maturity measurements for a particular concrete isnevertheless not easy.The other widely used maturity function which is sometimes regarded asmore reliable is more complex and assumes that the early rate of strengthgain varies exponentially with temperature. This is sometimes known as theArrhenius function, and requires a knowledge of the activation energy (Q)of the cement used, to yield maturity as an equivalent age t e at a specifiedtemperature T s . The equivalent age in this case is given byt e = ∑ e −Q 1Ta − 1 Ts tA Q-value of 5000 K is recommended by ASTM C1074 (360) for USType 1 cement, and guidance is given on procedures for determining thisfor other cements by Carino (361) who gives values of activation energiesfor a range of cement types.High temperatureCompressive strengthLow temperatureCrossoverMaturity IndexFigure 8.24 The effect of early age curing temperature on the strength–maturityrelationship (based on ref. 361).

Performance and integrity tests 257In Europe a specified reference temperature of 20 C is normally used,whilst in the USA a value of 23 C is usually assumed.The equivalent age based on maturity calculated by the Nurse-Saulmethod is given by the simplified expressiont e =MtT s − T oMore complex expressions for the equivalent age, giving a greater accuracy,are discussed fully in ref. 361.In summary, maturity measurements can be used to account for the effectsof temperature and time on strength development but cannot be used inisolation to detect batching errors. Particular care is needed when using‘automatic’ commercially available ‘maturity meters’ to ensure that the correctvariables for the concrete used are programmed. A potential applicationis to use the maturity approach in conjunction with supplementary in-situstrength tests to determine whether the as-placed concrete will comply withthe contractual strength requirements.8.7.2 Temperature-matched curingThe temperature at a pre-selected point within the concrete element may bemonitored and used to control the temperature of a water bath in whichtest specimens (cubes or cylinders) are placed. Their temperature regime,and hence maturity, will thus be identical to that at the selected point inthe pour and, as with maturity testing, careful location of sensors withrespect to the aims of the testing is most important. These specimens maythen be tested for strength in the normal way when required, and may berelated to the in-situ properties with due allowance for likely difference incompaction.The fundamental requirements are detailed by BS 1881: Part 130 (362).Features include a tank with sufficient capacity to take at least four testspecimens in their moulds and with a water-circulating device and heatercapable of producing a temperature rise of at least 10 C/hr. The temperaturesensor in the concrete is coupled to control equipment which maintainsthe water temperature within 1 C of the sensor temperature. A continuousrecord of both temperatures must be produced to permit confirmation ofthe correct functioning of the equipment and calculation of maturity. Thisis regarded as a very reliable approach, but principal difficulties include theneed for sizeable water tanks located close to the pour and the potentialfor accidental damage or vandalism to connecting cables and the loss ofpower supply in a vulnerable environment. Commercially available equipmentexists which fulfils these requirements. Cannon (363) has discussedthis technique, which is growing in popularity, in greater detail.

258 Performance and integrity tests8.8 Screed soundness testerFloor screeds are prone to failure if laid ‘semi-dry’, i.e. with inadequatewater to allow proper compaction. The result is a thin compacted surfaceskin and poorly compacted debris beneath. Such floor screeds often fail bydeformation under a concentrated load, such as a chair leg, and Pye andWarlow (364) have reported a test method to measure soundness of densefloor screeds. The equipment, commonly known as the BRE screed tester,includes a 4 kg weight which is dropped 1 m down a vertical rod on to a‘foot’ to impact the surface over a 500 mm 2 area. The surface indentationcaused by four blows on the same area is measured, using a simple portabledial gauge device. If the indentation is more than 5 mm the screed is unsatisfactory.The equipment is covered by BS 8204: Part 1 (365) which alsogives classification limits for results according to floor type. These limits willapply to screeds which are at least 14 days old and where the test area is flatand free of all loose dirt and grit. The test needs to be modified if the screedis laid over a weak or soft insulating material, as a core may be punched out,but is particularly useful in identifying zones of poor compaction beneathan apparently good surface. It has also been shown to be useful for assessingabrasion resistance of concrete slabs (269) – see Section 7.6.8.9 Tests for fire damageVisual observation of spalling and colour change (10) possibly aided bysurface tapping is the principal method of assessment of fire damage.Ultrasonic pulse velocity measurements may perhaps help (366) – see alsoChapter 3 – but use on site for this application poses many problems.Surface zone residual concrete strength can be assessed by appropriatepartially destructive tests (see Chapter 6), whilst thermoluminescence, asdescribed in Section 9.12, may help to determine fire history. Furtherguidance regarding the assessment of fire damage is given by ACI SpecialPublication No. 92 (367).Damage to concrete attributable to fire has been summarized in the ConcreteSociety Technical Report 33 (10). Three principal types of alterationare usually responsible.(i) Cracking and microcracking in the surface zone: This is usually subparallelto the external surface and leads to flaking and breaking awayof surface layers. Cracks also commonly develop along aggregate surfaces– presumably reflecting the differences in coefficient of linearexpansion between cement paste and aggregate. Larger cracks canoccur, particularly where reinforcement is affected by the increase intemperature.(ii) Alteration of the phases in aggregate and paste: The main changesoccurring in aggregate and paste relate to oxidation and dehydration.

Performance and integrity tests 259Loss of moisture can be rapid and probably influences crack development.The paste generally changes colour and various colour zones candevelop. A change from buff or cream to pink tends to occur at about300 C and from pink to whitish grey at about 600 C. Certain typesof aggregate also show these colour changes which can sometimes beseen within individual aggregate particles. The change from a normal tolight paste colour to pink is most marked. It occurs in some limestonesand some siliceous rocks – particularly certain flints and chert. It canalso be found in the feldspars of some granites and in various otherrock types. It is likely that the temperature at which the colour changesoccur varies somewhat from concrete to concrete and if accurate temperatureprofiles are required, some calibrating experiments need to becarried out.(iii) Dehydration of the cement hydrates: This can take place within theconcrete at temperatures a little above 100 C. It is often possible todetect a broad zone of slightly porous light buff paste which representsthe dehydrated zone between 100 and 300 C. It can be important, inreinforced or pre-stressed concrete, to establish the maximum depth ofthe 100 C isotherm.Changes in fire-damaged concrete

260 Performance and integrity testsResearch by Short (368) has recently indicated the potential for opticalmicroscopy and colour image analysis on polished or petrographic thinsections (see Section 9.11.2) to yield more detailed information than ispossible by simple visual inspection. Suitable samples can be cut from corestaken in the damaged areas. Crack density measurements using similartechniques may also prove useful.Cores are also sometimes used for compressive strength testing, but usefulinformation about surface zones most affected by fire and property changeswith depth will be limited. Recent work (369) has however suggested that15 mm thick slices of cores at varying depths can be used to measurewater absorption and tensile strength (from diametric compression loading).Both of these parameters have been shown to reflect gradients between thedeteriorated surface and sound concrete. In laboratory tests these have alsobeen related to temperature history. Other recent studies have examined thepotential for using thermography to assess fire-damaged concrete (370).

Chapter 9Chemical testing and alliedtechniquesChemical testing of hardened concrete is mainly limited to the identificationof the causes of deterioration, such as sulfate or chloride attack, or to specificationcompliance, involving cement content, aggregate/cement ratio oralkali content determination. Water/cement ratio, and hence strength, isdifficult to assess to any worthwhile degree of accuracy, and direct chemicalmethods are of limited value in this respect. Water/cement ratio can, however,often be determined to a good accuracy using petrographic methods.Some chemical tests are expensive, and will often only be used in casesof uncertainty or in resolving disputes, rather than as a means of qualitycontrol of concrete.Specialist laboratory facilities are required for most forms of chemicaltesting. Basic procedures for the principal tests are outlined below, andemphasis has been placed on the interpretation and reliability of results.Techniques and procedures are generally complex, and extreme care mustbe taken both during sampling and testing if accuracies are to be achievedwhich are of practical value. One of the major problems of basic chemicaltesting is the lack of a suitable solvent that will dissolve hardened cementwithout affecting the aggregates, and, if possible, samples of the aggregatesand cement should also be available for testing. Other instrumental techniques,such as differential thermal analysis, electron microscopy and X-raymethods, require expensive and complex equipment together with a highdegree of skill and experience, but are growing in usage. The range of techniquesavailable to the cement chemist is wide, and many are of such a highlyspecialized nature that they are outside the scope of this chapter. Attentionhas therefore been concentrated on those methods that are most commonlyused for in-situ investigations, whilst the more important of the other techniquesare indicated together with their most commonly used applications.ASTM standards are available for commonly used chemical tests, but BS1881: Part 124 (371) provides more comprehensive guidance and proceduraldetails for many tests. These include cement content, aggregate contentand grading, aggregate type, cement type, original water content and bulkdensity, as well as chloride, sulfate and alkali contents. These procedures

262 Chemical testing and allied techniquesapply to calcareous cements, and to natural or inorganic artificial aggregates.Additional background information and details are given in a comprehensiveConcrete Society Technical Report which offers detailed guidance (372).It is particularly important that an engineer requiring chemical analysisof concrete should be aware of the limitations of the methods available,and in particular the effect that some materials’ properties may have on theaccuracy of analysis. The most likely causes of lack of accuracy are:(i) Inadequate sampling or testing(ii) Aggregates contributing to the analysis(iii) Cements with unusual and unknown composition(iv) Changes to the concrete from chemical attack or similar cause(v) Presence of other materials.It is also essential that an experienced concrete analyst should beemployed. He must be given a clear brief of the information required fromtesting, and all relevant data concerning the constituents and history of theconcrete must be made available.9.1 Sampling and reporting9.1.1 SamplingThe tests undertaken on a sample will relate only to the concrete at thatparticular location in a structure. It is therefore essential that sufficientsamples be taken to represent the body of concrete under examination.Results from particular samples cannot be assumed to provide informationon the concrete at points other than those from which the samples weretaken. The natural variability of concrete properties as well as the mobilityof some chemicals (e.g. chloride salts) must thus be considered. At leasttwo, and preferably four, separate samples are recommended for volumesup to 10 m 3 of concrete, and at least ten samples should be used for verylarge volumes under examination, followed by more extensive investigationat particular locations if necessary. It is important, in cases of dispute, thatagreement is reached between all parties involved about the location andmethod of sampling, together with the extent of the material which is to beconsidered to be represented by the sample.The basic requirements for a sample for chemical analysis are that:(i) The minimum linear dimension should be at least five times the maximumaggregate size.(ii) The sample should be in a single piece.(iii) The sample should be free from reinforcement and foreign matter unlessthey are the subject of the test.

Chemical testing and allied techniques 263For drilled dust samples, which are often used for assessment of chloridecontent and gradient:(i) The sample should be taken with a 20 mm masonry drill bit.(ii) A suitable means of collecting the sample without loss of fines is used.(iii) The minimum mass of sample should be 25 g. This may necessitatetwin holes for gradient samples, to ensure each increment meets the25 g requirement.(iv) The first 5 mm of drilled material is rejected, as it may be unrepresentativeof the bulk of the concrete.The sample should be clearly labelled with the date, precise location andmethod of sampling with any other relevant details and sealed in a heavydutypolythene bag, which is also labelled. The size of the sample will varyaccording to the tests to be performed; some tests may require only a smalldrilled dust sample whilst others will need a sizeable lump of concrete.Most chemical tests can be carried out on a sample obtained from a coreafter compressive testing, although if an undamaged specimen is requiredfor original water content determination an untested core will be required.In many cases it will be necessary to determine the cement content of theconcrete in addition to the particular component under examination (e.g.water content, sulfate content, etc.). The original sample must thereforebe large enough to allow for all the tests to be performed. The quantitiesof sample subjected to preparation, such as grinding, should also takeaccount of this. It is important that the sampling method does not influencethe sample in relation to the property to be measured, for example bywashing-out.9.1.2 ReportingIt is desirable that the reports contain the following information in additionto that required by BS 1881: Part 124 (371):(i) Methods used(ii) Variation to standard methods, with justification(iii) Assumptions made, such as aggregate type or properties, with justification(iv) Raw analytical data(v) Conclusions with a statement of their reliability.BS 1881: Part 124 further requires that full details of identification marksshould be given with a full qualitative description of the sample. Particularreference must be made to factors that are likely to reduce the accuracyof results. Details of other results obtained coincidentally to the principaltests, or any tests done at the analyst’s discretion, should also be given.

264 Chemical testing and allied techniques9.2 Cement content and aggregate/cement ratio9.2.1 TheoryThe most common methods for determining the cement content of a hardenedmortar or concrete are based on the fact that lime compounds andsilicates in Portland cement are generally much more readily decomposed by,and soluble in, dilute hydrochloric acid, than the corresponding compoundsin aggregate. The quality of soluble silica or calcium oxide is determinedby simple analytical procedures, and if the composition of the cement isknown, the cement content of the original volume of the sample can becalculated. Allowance must be made for any material that may be dissolvedfrom the aggregate, and representative samples of aggregate should be analysedby identical procedures to permit corrections to be made. It will not bepossible to treat all concretes in an identical manner because of differencesin aggregate properties, but the aggregate/cement ratio will be found as wellas cement content, whichever method is used.9.2.2 ProceduresASTM C85 (373) uses a crushed sample, dehydrated at 550 C for threehours and treated with 1:3 hydrochloric acid:distilled water. Dissolved silicais determined by standard chemical methods, and the filtrate from this isthen tested for calcium oxide content. Calculations depend upon the natureof the aggregate (a sample of which must be available) and figures areprovided to assist interpretation. The techniques recommended by BS 1881:Part 124 (371) are more complex, but follow the same basic procedureswith variations in the significance placed upon analytical methods accordingto the type of aggregate. Aggregates are classified into three broad groupsfor this purpose:(i) Type I - Natural aggregates essentially insoluble in dilute HCl(ii) Type S - Natural aggregates mainly soluble in dilute HCl(iii) Type O - Other aggregates.9.2.2.1 Sample preparationThe sample is initially broken into lumps not larger than about 50 mm,taking care as far as possible to prevent aggregate fracture. These are driedin an oven at 105 C for 15–24 hours, allowed to cool to room temperature,and divided into sub-samples.A portion of the dried sample is crushed, ground and subdivided to providea powder, which passes a 150 m fine mesh sieve. The recommendedprocedure is detailed, but care is essential if a representative sample is to be

Chemical testing and allied techniques 265obtained. The operations should also be performed as quickly as possibleto minimize exposure to atmospheric carbon dioxide.The portion is first crushed to pass a 5.0 mm sieve and a sub-sample of500–1000 g obtained, which is then crushed to pass a 2.63 mm sieve andquartered to give a sample which is ground to pass a 600 m sieve. Thisis also quartered and further ground to pass the 150 m sieve. Where amechanical means of sample preparation is employed, it is possible to omitsome of these steps and go directly from a 5.0 mm sample to a samplepassing 150 m.9.2.2.2 Determination of calcium oxide (CaO) content aloneA portion of the prepared analytical sample weighing 5 ± 0005 g is treatedwith boiling dilute hydrochloric acid. Triethanolamine, sodium hydroxideand calcein indicator are added to the filtered solution, which is then titratedagainst a standard EDTA solution. The CaO content may be calculatedto the nearest 0.1% in this way. Atomic absorption spectrophotometricmethods are also often used.Reliance upon measurements of CaO content alone may be consideredacceptable if the calcium oxide content of the aggregate is less than0.5%, but additional determination of soluble silica content is recommended.It should be noted that the calcium in the concrete is not inthe form of calcium oxide, CaO, but it is reported as being CaO byconvention.9.2.2.3 Determination of soluble silica, calcium oxideand insoluble residueSoluble silica is extracted from a 5±0005 g portion of the prepared sampleby treatment with hydrochloric acid, followed by sodium carbonate solutionand the insoluble residue collected by filtering. The filtrate is reduced byevaporation and treated with hydrochloric acid and polyethylene oxidebefore again being filtered and diluted to provide a stock solution.The filter paper containing the precipitate produced at this last stage isignited in a weighed platinum crucible at 1200 ± 50 C until constant massis achieved, before cooling and weighing. The soluble silica content can becalculated to the nearest 0.1% from the ratio of the mass of the ignitedresidue to that of the analytical sample. It is also possible to measure thesoluble silica content directly on the stock solution extract, using a nitrousoxide/acetylene flame and atomic absorption spectrophotometric methods.This can be considerably faster.The calcium oxide content is determined from the stock solution usingprocedures similar to those in Section 9.2.2.2. The insoluble residue isdetermined from the material retained during the initial filtration process

266 Chemical testing and allied techniquesby repeated treatment with hot ammonium chloride solution, hydrochloricacid and hot water followed by ignition in a weighed crucible to925 ± 25 C.9.2.2.4 Calculation of cement and aggregate contentThe cement content should be calculated separately from both the measuredcalcium oxide and soluble silica contents, unless the calcium oxide contentof the aggregate is less than 0.5% (see Section 9.2.2.2) or greater than 35%in which case results based on calcium oxide are not recommended. In thelatter case, if the soluble silica content of the aggregate is greater than 10%,analysis should be undertaken to determine some other constituent, suchas an iron or aluminium compound, known to be present in substantiallydifferent amounts in the cement and aggregate. This should preferably bepresent in the larger quantity in the cement. It is assumed that the combinedwater of hydration is 023× the percentage cement content, and that 100%oven-dried concrete consists of C% cement +R% aggregate +023C%combined water of hydration. Thus if,a = calcium oxide or soluble silica content of cement (%)b = calcium oxide or soluble silica content of aggregate (%)c = measured calcium oxide or soluble silica content of the analyticalsample (%)d = oven-dried density of concrete in kg/m 3 ,then percentage cement contentC =c − b × 100%to nearest 01%a − 123band percentage aggregate contentR = a − 123c × 100%to nearest 01%a − 123bThus the aggregate/cement ratio = R/C to the nearest 0.1.The cement content in kg/m 3 is given byC × d100 kg/m3 to the nearest 1 kg/m 3If coarse and fine aggregates have differing calcium oxide and solublesilica contents the overall weighted means should be used as the valuesfor b determined on the basis of assessed grading, mix design or visualinspection.

Chemical testing and allied techniques 267Use of these expressions requires an analysis of both the cement and theaggregate to be available. If an analysis for the cement is not available,ordinary and rapid hardening cements complying with the relevant BritishStandard may be assumed with little loss in accuracy to have a calciumoxide content of 64.5%. An assumed soluble silica content of 20.7% isless reliable. Appendix A of BS 1881: Part 124 (371) provides detailedtypical analyses of these and other common cement types. Absence of anaggregate analysis poses greater problems, although in some cases the silicacontent may justifiably be assumed to be very low (many laboratories areknown to use a value of 0.2–0.5%, silica correction as routine), and fortype I aggregates the calcium oxide content may be taken as zero. Otherwisemicroscopic examination may be necessary (see Section 9.11).If the two estimated cement contents are within 25 kg/m 3 or 1% bymass, the average value is adopted. If a greater difference exists, and noreason can be found for the discrepancy, both results may be quotedwith an indication of the preferred value based on the factors outlinedabove.Where pulverized fuel ash or natural pozzolans have been used as reactivecomponents it is difficult to determine the content of the addition, butan approximate estimate may be made of the ordinary Portland cementcontent provided the added material has a calcium oxide content belowabout 2%.9.2.2.5 XRF Methods for the determination of cement replacementsMathematical methods involving the use of X-ray fluorescence techniquesto determine slag and PFA contents have however been reported (374),and are in regular use in some UK laboratories. Other techniques involvingchemical analysis of individual grains are believed to be in use. Providedthat such additions have not been used, a value of aggregate contentmay be calculated from the insoluble residue values of the analytical sampleand of the aggregate assuming no insoluble residue in the cement.Where the aggregate is type I, the percentage insoluble residue valueobtained for the analytical sample may be taken as the percentage cementcontent.Example of method for determination of GGBS (ground granulatedblastfurnace slag) and cement content of a hardened concreteThe hardened concrete is examined using thin sections with a petrologicalmicroscope in conjunction with control concretes to assess the originalwater/binder ratio of the concrete.A large-area polished plate is prepared from the sample and examinedby the method of point counting in accordance with ASTM C457 (263) todetermine the volume proportions of aggregate, binder and void.

268 Chemical testing and allied techniquesA high-quality polished surface is prepared from the sample and examinedwith an electron microscope fitted with an energy dispersive X-raymicroanalysis system calibrated using mineral standards and hydrated standardcements to determine the mean composition of the GGBS, cement andbinder (excluding aggregate and aggregate dust).The chemical data obtained with the electron microprobe is analysedusing ratios rather than absolute quantities. The ratio Al 2 O 3 /Al 2 O 3 +CaOshows a suitable large range between Portland cement and GGBS as inFigure 9.1.In this example the level of cement replacement by GGBS would be givenfrom the equation of the curve using the mean ratio Al 2 O 3 /Al 2 O 3 + CaOdetermined from the binder area analyses excluding aggregate made withthe electron microprobe.The petrographically determined water/binder ratio is used in conjunctionwith the calculated ratio GGBS/GGBS + PC and measured volumeproportions of aggregate paste and void to calculate the composition ofthe sample in terms of weight fractions using assumed densities for theaggregate, cement and GGBS.9.2.3 Reliability and interpretation of resultsThe procedures used above will yield the Portland cement content of a hardenedconcrete, whilst the content of blended cements can only be determinedif the actual analysis of the blend is known. The Concrete Society TechnicalGGBS/(GGBS + PC)1.00.9 y=–6.1211x 2 +7.5235x – 0.55730.80.70.60.50.40.30.20.10.00.00 0.100.20 0.30Al 2 O 3 /(Al 2 O 3 + CaO)Figure 9.1 Example of calibration graph for the determination of GGBS byEDAX/Electron Microscopy (Courtesy of Geomaterials ResearchServices).

Chemical testing and allied techniques 269Report (372) provides details of extensive precision trials involving severaltest laboratories. Errors may be divided into those caused by sampling, forwhich ±50 kg/m 3 for one sample reducing to ±25 kg/m 3 for four samplesmay be assumed, and those due to testing. Testing accuracy is dependentupon the aggregate type and test procedures, but may be of the order of±15 kg/m 3 for limestone ranging to ±40 kg/m 3 for flint aggregates. Combiningthese effects typically gives an overall accuracy of between ±30 and±50 kg/m 3 based on the mean of four independent samples. A figure of±45 kg/m 3 is often regarded as typical for a gravel concrete with a cementcontent of the order of 350 kg/m 3 . This value will be considerably worsefor atypical cements, or when blast furnace slag or pulverized fuel ashare present. Whenever assumptions about the contribution of acid-solubleaggregate components must be made because of lack of availability of aseparate sample, the accuracy will decrease. Similarly, assumptions of CaOand SiO 2 proportions in the cement will affect accuracy, but since the rangeof CaO in a typical UK Portland cement is less than that of SiO 2 , assumptionsbased on CaO should be preferred whenever possible. If the aggregateeffect is high, the silica approach may be used, but although this is lessaggregate-sensitive it is not so easy or reliable.The importance of careful sampling and sample preparation cannot beoveremphasized. If the aggregates prevent analysis of cement content andaggregate/cement ratio by chemical means, it may be possible to obtainestimates based on micrometric methods outlined in Section 9.11, or byinstrumental techniques such as X-ray fluorescence spectrometry.Whatever method is used to determine cement content, it is essentialto recognize that no statistical statement about a particular batch of concretecan be made from a test on a single sample. Where several separatesamples are used, an estimate of an average value may be possible, and aminimum of four individual samples should be used to obtain a reliableestimate. Selection of the location for these should preferably take accountof variations caused by position in mixer discharge, bearing in mind thatfor a truck mixer the first portion will generally be richer than averagewhilst the later portions may be leaner. Differences of up to 100 kg/m 3have been reported (304) between the top and bottom of walls andcolumns.9.3 Original water content9.3.1 TheoryThe quantity of water present in the original concrete mix can be assessedby determining the volume of the capillary pores which would be filled withwater at the time of setting, and measuring the combined water present

270 Chemical testing and allied techniquesas cement hydrates. The total original water content will be given by thesum of the pore water and combined water. This approach forms thebasis of the BS 1881: Part 124 method (371). It requires a single sampleof concrete that has not been damaged either physically or chemically.Usually the water/cement ratio is of greatest interest, so that a cement contentdetermination will also be necessary. Alternative methods of assessingwater/cement ratios include thin section microscopy and reflected light fluorescencemicroscopy (375). These latter methods claim a higher degree ofaccuracy.9.3.2 ProcedureAn undamaged sample is normally obtained as a saw-cut slice approximately20 mm thick and with a single face area of not less than 10 000 mm 2(e.g. 100 mm square). The sample should be taken from sufficient depthwithin the concrete to avoid laitance and other surface effects, and mayconveniently be obtained as a ‘vertical’ slice on a core. Care must be takento minimize material loss from the cut faces, and carbonation is preventedby storage in an airtight container.The sample is oven dried at 105 C for at least 16 hours, cooled in adesiccator, weighed and immersed in a liquid of known density (commonlytrichlorethane), in a vacuum desiccator. The pressure is reduced, causingthe air from the capillaries to be evolved. The vacuum is then releasedand the sample kept immersed in the liquid for a further 5 minutes, thenweighed in a sealed polythene bag to prevent loss by evaporation. The massof liquid filling the pores can thus be calculated, and the % capillary waterderived:% capillary water =mass of liquid absorbedLiquid density × wt. of dry sample × 100After the capillary porosity measurements are complete the slice is heatedat 105 C until constant mass is obtained, and crushed to pass a 150 msieve. Approximately 1 g of this sample is ignited at 1000 C in a streamof nitrogen or dried air, and the evolved water is weighed after it hasbeen absorbed by ‘dried’ magnesium perchlorate. A further portion of thisground sample is used for the cement content measurement, using the mostappropriate method of those given in Section 9.2, according to aggregateproperties.Unfortunately the aggregate will frequently also have a porosity andcombined water content that must be allowed for. These can be assessedby applying identical procedures to those above to a sample of coarseaggregate.

Chemical testing and allied techniques 271It must be assumed that the coarse aggregate values are typical for allaggregates in the concrete slice, and hencecorrected capillary water = Q − qR100 %and corrected combined water = X − YR100 %where Q = determined capillary porosity of slice, %q = determined capillary porosity of aggregate, %R = aggregate content, %X = combined water of concrete, %Y = combined water of aggregate, %C = cement content, %.If it is assumed that no water of hydration has been replaced by carbondioxide, the original free water/cement ratio is given by the sum of theabove expressions divided by the cement content C%. The original totalwater/cement ratio can similarly be derived using the uncorrected percentagecapillary porosity Q with the corrected percentage of combined water.If aggregate control samples are not available it must be assumed thatthe combined water of hydration is typically 0.23 times the percentagecement content. This value is often adopted as a matter of routine becauseof the inaccuracies that are likely in combined water measurements. Thecapillary porosity of the aggregate may also be taken to be equal to thewater absorption value of the aggregate, if this is known, thus permittingan estimate of original free water/cement ratio. If the aggregate absorptionvalue is not known, then only the original total water/cement ratio can bequoted and is given by Q/C + 023.9.3.3 Reliability and interpretation of resultsSince this determination requires a sound specimen of concrete, strengthtestedcores or cubes cannot be used. It is suggested that a precision betterthan ±01 for the original water/cement ratio is unlikely to be achievedeven under ideal conditions (372). A major source of difficulty lies inthe corrections for aggregate porosity and combined water, which maybe overestimated by the procedures used. This will lead to an underestimateof the true original water content although it is also possible thatcontinued hydration in older concretes may lead to low apparent originalwater/cement ratios. The original cement content must also be measured

272 Chemical testing and allied techniqueswith reasonable accuracy, and the problems associated with this are discussedin Section 9.2.The method is not suitable for semi-dry or poorly compacted concretealthough in the case of air entrainment, Neville (376) suggests that since thevoids are discontinuous they will remain air-filled under vacuum and willabsorb no solvent although this view is regarded by others as doubtful. If thisis the case, entrained air will not affect the results, which are influenced onlyby capillary voids. Carbonated concrete must also be avoided. Where theaggregates are very porous or contain an appreciable amount of combinedwater the corrections required will be so large that the results may be oflittle value. This shortcoming is likely to limit the value of the method forartificial aggregates.9.4 Cement type and cement replacements9.4.1 TheoryThe basic type of cement used in a concrete may be established by separationand chemical analysis of the matrix for comparison with established analysesof particular cement types. Alternatively, microscopic examination canbe used to detect unhydrated cement particles, which can be compared withknown specimens. The complexity of the analysis varies according to cementtype, admixture, replacements and aggregates, and BS 1881: Part 124 (371)strongly recommends that chemical analysis should be supplemented bymicroscopic examination. It is not possible to distinguish between ordinaryand rapid hardening Portland cements, although cement replacements canusually be detected. Complex techniques such as differential thermal analysismay also be used to establish cement type on a comparative basis.A simple site test has been developed for high alumina cement identification(see Section 9.4.2.4).9.4.2 Procedures9.4.2.1 Analysis of matrixA very fine sample is obtained by sieving material from a broken piece ofconcrete through a 90 m mesh. It is essential that this contains no morethan minute traces of aggregate which may contribute to the analysis. Thevery fine sample can then be analysed for basic cement components suchas SiO 2 , CaO, Al 2 O 3 ,Fe 2 O 3 , MgO and SO 3 and the resulting compositioncompared with analyses of known cements. It is also possible to analyseareas of the cement paste on polished specimens using electron probemicroanalysis methods (see Section 9.2.2.5).

Chemical testing and allied techniques 2739.4.2.2 Microscopic methodA small solid piece of concrete, about 20 mm cube, should be selected,preferably not containing any large aggregate pieces nor having a finishedsurface. This is dried for 12 hours at 105 C and embedded in epoxy mortarbefore saw cutting and grinding with carborundum powder. The surfaceis finally polished with diamond powder and examined by reflected lightmicroscopy. Coarse unhydrated cement particles will be visible, and thesemay be etched to show the phases characteristic of the cement type. BS 1881:Part 124 (371) offers guidance concerning the most appropriate etchingmethods, and comparative reference specimens of known cement type maybe valuable for identification. French (377) has also described approachesusing thin microscopic sections.9.4.2.3 Cement replacementsThe most commonly used cement replacements are blast furnace slag(GGBS) and pulverized fuel ash (PFA). Blast furnace slags contain considerablyhigher levels of manganese and sulphides than are to be found innormal cements, and this can be used for their identification by chemicalanalysis. However, the sulphide is readily oxidized by the air and thismeans that precautions must be taken to protect samples from the air. Italso means that older samples may have oxidized and the sulphide may notbe detectable. If the slag is from a single source of known composition, itmay be possible to obtain quantitative data, provided there is no slag in theaggregate. A characteristic green or greenish black coloration of the interiorof the concrete may aid identification, and microscopic methods may alsobe used. PFA is most easily detected by microscopic examination of theacid-insoluble residue of a sample of separated matrix. Particular characteristicspherical particle shapes may be recognized. Microscopic examinationof thin sections as described by French (377) may also be used.9.4.2.4 Identification of high alumina cement (simple method)A simple rapid chemical test for the identification of high alumina cementhas been developed by Roberts and Jaffrey (378). Whilst this type of cementcan be identified by microscopic or complex methods such as X-ray spectrometer,the need for a simple test that could be used on site arose followingproblems with this material in the UK. This method is based on the assumptionthat an appreciable quantity of aluminium will be present in HACconcrete dissolved in dilute sodium hydroxide, whereas little will be foundin a similar solution of Portland or other types of cement.A powdered sample is obtained by drilling by masonry drill, or by crushingsmall pieces of concrete with a pestle and mortar and grinding after

274 Chemical testing and allied techniquesremoval of aggregate particles. Approximately 1 g of this powdered sampleis placed in a 25 ml test tube with about 10 ml of 0.1 M cold sodiumhydroxide solution, sealed and then shaken by hand for 2–3 minutes. Thesolution is then filtered through a medium-grade paper and the filtrateacidified with five drops of dilute HCl. Ten drops of ‘Oxine’ reagent and1 ml ammonium acetate are then added, and if aluminium is present inappreciable quantities, as in HAC, turbidity and the formation of a yellowprecipitate will occur. If the solution remains clear or only slightly cloudy,HAC is not likely to be present. The presence of gypsum, which may resultfrom contamination of the sample by plaster, can interfere with the testand produce misleading results. Care must therefore be taken to avoid suchcontamination during sampling.9.4.3 Reliability and interpretation of resultsBoth chemical and microscopic methods of cement type determination relyupon a carefully prepared sample and a great deal of specialist skill andexperience. Whilst under favourable conditions it should be possible toidentify cement type, neither method is accurate enough to tell whether thatcement complied with a particular specification. Cement replacements canusually be identified, but the level can only be determined chemically forblast furnace slag, and then only if conditions concerning the source andaggregate properties are favourable (though PFA content can be determinedby XRF techniques as described earlier). The simple HAC determinationmethod has been developed for site use and has proved to be a reliableindicator of the presence of this type of cement provided that sampling iscarefully executed.9.5 Aggregate type and grading9.5.1 Aggregate typeThis is best determined in a cut slice or core which will allow the cut sectionsof aggregate to be examined visually, physically or chemically to determineits type. ASTM 856 (379) provides guidance on petrographic examination.The description and group classification of aggregates is dealt with in BS812: Part 104 (380) but petrographical examination by a qualified geologistis required. If a more detailed analysis relating to mineralogy, textureand microstructure is needed, microscopic methods described by French(377) may be useful. These can yield information on hardness, porosity,permeability, specific gravity and thermal properties of the aggregate aswell as the presence of potentially deleterious substances. This is also the

Chemical testing and allied techniques 275only reliable method of distinguishing between freeze–thaw damage, alkali–aggregate reaction, delayed ettringite formation and sulfate attack as thecause of deterioration of hardened concrete.In many instances a detailed aggregate identification will be unnecessary,and a broad chemical classification based on reaction with acid will beadequate. This will apply particularly to the selection of the analyticalmethod to apply to the concrete for cement content determination.9.5.2 Aggregate gradingThis can only be reliably achieved for aggregates which are essentially insolublein dilute hydrochloric acid. An initial sample of at least 4 kg will benecessary, and a sub-sample of approximately one quarter of this will betested. This is initially broken down into coarse and fine fractions usinga 5 mm sieve without fracturing the aggregate, assisted as necessary byheating in a furnace to about 300 C. Placing the concrete in a microwaveoven has also been used, but it is essential that the concrete contains noembedded metal and is securely wrapped in a fire blanket, as there is arisk of the concrete fracturing explosively. Coarse aggregate particles arecleaned by chipping off the paste and both fractions treated with hydrochloricacid to dissolve the paste prior to sieve analysis. Care must be takento remove chipped portions of coarse aggregate from the fines, and thegradings obtained are not of sufficient accuracy to assess compliance withdetailed aggregate specifications.9.6 Sulfate determinationThe sulfate content is obtained by chemical analysis of a weighed, groundsample of concrete which is expected to contain about 1 g of cement.The use of a concentrated sample of fine materials, obtained as describedin Section 9.2.2.2, may be worthwhile. This is treated with hydrochloricacid, ammonia solution and barium chloride solution to obtain a precipitatewhich is ignited at 800–900 C and weighed. If the sulfate content ofthe original sample exceeds about 4% of the cement content for Portlandcement concretes, chemical attack may be indicated. A cement content determinationmust therefore be performed in conjunction with sulfate tests,and an accuracy better than ±03% SO 3 /cement is unlikely. Specializedinstrumental techniques (see Section 9.13) are growing in popularity andmay improve on this accuracy.Cases of suspected sulfate attack may be identified by measuring thesulfate distribution between sulphoaluminate and gypsum. Since sulphoaluminateis stable only in alkaline conditions and is converted to gypsumby acid, attack by groundwater may be detected and distinguished from theproducts of sulphur-oxidizing bacteria.

276 Chemical testing and allied techniquesSulfate attack may also be identified by petrographic methods (seeSection 9.11) involving the microscopic study of thin sections to reveal thepresence of crystalline calcium sulphoaluminate. This is commonly knownas ettringite and may be found along cracks when reactions have occurredbetween sulfate-bearing solutions and the cement paste. However, cautionmust be exercised as sulfate minerals can line cracks and voids in concretesthat have been subjected to leaching by water.More recently, a new form of sulfate attack has been identified. In 1998,the foundations to a number of motorway bridges in the UK were foundto be suffering from serious erosion and crumbling of the outer part of theconcrete in the foundations. The problem was diagnosed as being due toan unusual form of sulfate attack, known as thaumasite attack. For theproblem to occur, a number of factors have to be present.(i) A source of sulfate(ii) Water (usually plenty of moisture)(iii) A source of carbonate (as aggregate, or filler, or possibly even as fill orcarbonated groundwater)(iv) Low temperatures

Chemical testing and allied techniques 277It should be noted that sulfate-resisting cement has not proved to be anymore resistant than normal Portland cement in resisting this type of attack.When high alumina cement is used, complex methods such as differentialthermal analysis may be required (see Section 9.13).9.7 Chloride determinationThe need to assess the chloride content of hardened concrete is most likelyto arise in relation to the corrosion risk to embedded reinforcement or ties.Small quantities of chloride (up to approx 0.01% chloride ion by mass ofconcrete) will normally be present in concrete, but substantially more canresult if calcium chloride (as an admixture) or sea-dredged aggregates areused, and may present a potential hazard. Chlorides may also be absorbedfrom the surface, as in the presence of seawater or deicing salts, or enterthrough cracks in the concrete. In some parts of the world, such as theMiddle East, contaminated sand may also present a major problem.9.7.1 ProceduresIn addition to the ‘Volhard’ laboratory method described in BS 1881: Part124 (371), X-ray fluorescence spectrometry and other sophisticated techniquesmay be used to determine chloride content. These include remotelaser induced breakdown spectroscopy (LIBS) which is under developmentfor site use (382). Automatic titrators such as the Orion 960 (Figure 9.3)Figure 9.3 Automated Chloride Analyser using Potentiometric Titration. Thisequipment also interfaces with a carousel to allow automated analysis ofmultiple samples (photo courtesy of Thermo Electron).

278 Chemical testing and allied techniquesand others can allow 200 or more samples a day to be tested with goodaccuracy. Careful attention to quality control is critical, however, whenautomated methods are used. Simplified methods are also available whichmay be suitable for site use, although the accuracy of these will be lowerthan is possible from a detailed analysis. The Building Research Establishmenthas outlined procedures in information sheets (383,384) for twocommon methods, and other available methods have been compared byFigg (385).Whichever analysis technique is to be used, a powdered sample of concreteis required with a total mass of at least 25 g. This can be obtained bythe methods described previously (Section 9.2.2), usually from sliced coresor else by drilling with a masonry bit in a hand-held slow-speed rotarypercussion drill (Figure 9.4). Care must be taken to avoid steel, whichmay be located by covermeter. Views differ concerning sampling techniquesusing drilling (385) although it is generally accepted that a drill diameterat least equal to the size of the coarse aggregate is necessary to obtain arepresentative sample, and that a similar depth for sample increment is alsorequired. A diameter of 20 mm is thus often used but some engineers preferto combine drillings from several adjacent smaller diameter holes. If surfacelayer concentrations are required, shallower depth samples or incrementsshould be used, but if results are intended to relate to the interior of a bodyof concrete, the drillings from a 5 mm surface zone should be discarded.Externally exposed surfaces may have been leached by rainwater and giveFigure 9.4 Field collection of drilled dust sample.

Chemical testing and allied techniques 279unrepresentative results, so if shallow drilling is used it should preferablybe done from the inside of a building but avoiding contamination fromgypsum plaster or other materials. Surface zones exposed to seawater penetrationmay similarly give unrepresentative results in relation to the bodyof concrete.Drillings can be collected in a clean container pressed against the concretesurface below the drill hole, with care taken to avoid the loss of fine material.It is most important that all the drill dust is carefully collected and is notallowed to blow away under windy weather conditions. Where chlorides arepresent in the sample, the fine dust will often contain a significantly higherlevel of chloride than the coarse dust. A variety of funnels and collectiontube devices have been developed (385) to assist sample collection. Vacuumdrilling techniques are also available, which prevent loss of fine particlesfrom the sample.9.7.1.1 Laboratory analysis (Volhard method)This method, which is detailed in BS 1881: Part 124 (371), is relativelysimple and reliable but does require specialized laboratory facilities andexperience. Approximately 5 g of a ground or powdered sample is treatedwith hot dilute nitric acid and then filtered and cooled. Silver nitrate andnonyl alcohol are added with ferric alum indicator and back titrated withammonium thiocyanate. It is usual to express the results as percentagechloride ion by mass of concrete or cement.9.7.1.2 Instrumental methodsX-ray fluorescence spectrometry (Section 9.13) may be used to determineboth chloride and cement content of compressed samples. The wavelengthand intensity of fluorescent radiation generated by bombardment with highenergyX-rays is compared with the characteristics of standard samples ofknown composition. The method requires specialized sample preparationand test equipment, but analysis is quick, and chloride and cement contentare obtained from the same sample, which is available for a repeat analysis ifthat should be necessary. More rapid methods such as ion chromatographyand use of ion-selective electrodes are growing in popularity and a rapidsite version of the latter approach is available (14).9.7.1.3 ‘Hach’ simplified methodThis makes use of a commercially available kit (386) for a drop counttitration with silver nitrate solution in the presence of potassium chromateindicator. The kit includes a bottle of silver nitrate solution with a dropdispenser, capsules of indicator, a plastic measuring tube and a glass bottle

280 Chemical testing and allied techniquesfor the titration. A powdered 5 g sample is weighed and dissolved in 50 mlof 1 M nitric acid and 5 g sodium bicarbonate is dissolved in this solutionwhich is decanted and filtered. A 5.75 ml sample of filtrate is measuredand placed in the titration bottle with the contents of an indicator capsule,and the silver nitrate solution is added drop by drop until there is a colourchange from bright yellow to a faint reddish brown (Figure 9.5) (see ref.383 for a detailed procedure). For a test on one 5.75 ml measure of filtrate,the percentage chloride ion by mass of concrete is given directly by thenumber of drops × 0.03, whilst if less than four drops are required twoadditional measures of filtrate will be added and the total number of dropsrequired by the three measures recorded. In this case the required result isgiven by total number of drops × 0.01. This method is straightforward andquick, and although a working location is required, should be suitable forsite use by staff without specialist experience.9.7.1.4 ‘Quantab’ simplified methodThis method uses a commercially available ‘Quantab’ test strip (387) tomeasure the chloride concentration of a solution. The solution containing5 g powdered concrete is obtained as described in Section 9.7.1.3. Thestrip is plastic, approximately 75 mm long and 15 mm wide, with a verticalFigure 9.5 Titration using the ‘Hach’ method.

Chemical testing and allied techniques 281Figure 9.6 ‘Quantab’ chloride test.capillary column impregnated with silver dichromate (see Figure 9.6). Atthe top of the column is a horizontal air vent containing a yellow moisturesensitiveindicator which changes to blue when the capillary is full. Thelower end of the test strip is placed in the chloride solution until the capillaryis full, and the reddish brown silver dichromate in the capillary tubereacts with the chloride to form white silver chloride. The tip of this colourchange is related to a vertical scale, and the reading converted to mg chlorideion/litre by reference to calibration tables. The procedure is discussedin detail in ref. 384 and the use of ‘low range’ test strips (Type 1175) fornormal purposes is recommended. Caution should be taken with the ‘highrange’ strips that may overestimate chloride concentrations in some situations.Although facilities for sample preparations are required, the methodshould be suitable for site use by staff without specialist experience, andwill be of sufficient accuracy to indicate the presence and level of significantchloride contents for most practical purposes. The total time required fora test is approximately 30 minutes, and in the authors’ experience this hasbeen found to be the most reliable of the simplified methods, if used byexperienced personnel.

282 Chemical testing and allied techniques9.7.2 Reliability and interpretation of resultsChloride distribution may vary considerably within a member due to migrationand other effects and an adequate number of samples must be obtained.The Building Research Establishment (383) recommends single samplesfrom at least 10% of a group of building components under investigation,for the identification of the presence of significant quantities of chloridewithin members of the group. If an individual member is under examination,a number of samples should be taken according to size. Results byGrantham and van-Es (388) and by Reknes of the Norwegian BuildingResearch Institute (389) showed that inter-laboratory accuracy and reproducibilityin chloride testing of hardened concrete were poor.In most instances, the main requirement will be to establish the presence oflevels of chloride higher than would be normally expected and the simplifiedmethods will be adequate. The site use of ‘Quantab’ test strips is quick andcan determine concentrations within the range 0.03–1.2% chloride ion bymass of concrete, or establish if even less is present. A similar accuracymay be possible with the ‘Hach’ method. Results expressed in the formof percentage chloride ion by mass of concrete will be adequate for thesimplified methods, and should be used if big differences are expectedbetween the mixes specified and obtained. In doubtful or borderline cases,or where the precise level is required, proper laboratory determinations willhowever be necessary.In these cases the cement content of the concrete must also be determinedas in Section 9.2, so that the result can be expressed as percentagechloride ion by mass of cement. Table 9.1 shows a comparison of resultsusing a wide range of different testing methods for a series of laboratoryspecimens when a known percentage of sodium chloride was added tothe original concrete mix. It is very important that the basis of presentationof results is clearly indicated to avoid confusion at the interpretationstage.If the cement content is unknown, a typical value of 14% may beassumed, and hence the percentage chloride content by mass of concreteis multiplied by a factor of seven to estimate the relationship to cement(384). An adequate mass of sample >25 g is critical for dust samples toensure that the cement content of the sample is representative of that ofthe concrete. Even if the cement content ranges between 11 and 20%,the accuracy should be within ±02% chloride by mass of cement, usingthis assumption for chloride contents which are less than 0.10% by massof concrete. A precise chloride determination will normally be necessaryas the referee method in disputes. Whichever method of analysis is used,when obtaining samples by drilling, large aggregate particles should beavoided, since these may lead to an underestimate of the true percentage ofchloride in the concrete. It is difficult to improve on an overall accuracy of

Table 9.1 Comparison of chloride analysis results (based on ref. 389)Mix no. 1 2 3 4Chloride Content %by weight ofconcrete0.0225 0.0474 0.0731 0.1681Analysis Technique Number ofanalyses foreachchlorideconcentrationMeanValue% byweightofconcreteStandardDeviationMeanValue% byweightofconcreteStandardDeviationMeanValue% byweightofconcreteStandardDeviationMean Value% by weightof concreteStandardDeviationRCT 43 0.0165 0.0039 0.0361 0.0064 0.0546 0.0116 0.1416 0.0251Quantab 19 0.0243 0.011 0.0501 0.0078 0.0745 0.0137 0.1622 0.0142Volhard 11 0.0162 0.0096 0.0394 0.0148 0.0669 0.0065 0.1572 0.0213PotentiometrictitrationPotentiometricmeasurementwith ion selectiveelectrodeSpectrophotometricmeasurement9 0.0221 0.0009 0.0485 0.0017 0.0742 0.0024 0.1809 0.01063 0.0204 0.0099 0.0505 0.0156 0.0841 0.0273 0.1730 0.04751 0.028 — 0.054 — 0.0785 — 0.1860 —

284 Chemical testing and allied techniques±005% chloride ion/cement at a 95% confidence level, whatever methodis used, due largely to sampling errors.BS 8500 (390) suggests that the maximum acceptable percentage of chlorideion by mass of cement is 0.1% for prestressed concrete, or upper 95%confidence limit of 0.4% for reinforced concrete. Deterioration caused bychlorides is discussed in detail by Kay et al. (391) who also suggest that avalue of 0.4% chloride ion by mass of cement may be sufficient to promotecorrosion of reinforcement, although it is now generally accepted that aslittle as 0.25% chloride ion by mass of cement ingressing into hardenedconcrete may be required to depassivate the steel reinforcement. Whilstthese values are generalized, they may assist the interpretation of results onchemical analysis. When judging chloride gradients, a peak in chloride levelis often found a little way below the concrete surface, usually correspondingwith the depth of carbonation of the concrete. This is quite normal andresults from the release of chemically bound chloride by the carbonationprocess. This is why a combination of carbonation and even moderate chloridecontents can be so damaging to reinforced concrete – their combinedeffect is synergistic.9.8 Alkali reactivity testsWith the increased concern in the UK about the risks of alkali–silica reaction(ASR) causing damage to concrete (see Section 7.4), tests for alkali contentand alkali-induced expansion have been developed.9.8.1 Alkali contentAn analysis method for hardened concrete has been incorporated into BS1881: Part 124 (371), which involves the use of a calibrated flame photometerto assess the sodium oxide and potassium oxide contents of atreated powdered sample. A number of uncertainties exist with this method,including the possible contribution of sodium and potassium compoundscontained in aggregates but which are not readily available for reaction inconcretes. It is common practice to remove as much as possible of the coarseaggregate and to analyse a fines-rich fraction to minimize this problem.Migration or leaching of alkalis within test specimens during sampling orstorage is a further cause of uncertainty. Precision and accuracy data forthis approach are not currently available.9.8.2 Alkali immersion testThis is intended to give a rapid indication of the presence in a concrete sampleof some potentially alkali–silica reactive particles, but does not provethe past or future occurrence of damaging ASR in the concrete. The test

Chemical testing and allied techniques 285procedure is described in detail by Palmer et al. (33). A diamond-cut surfaceis lapped flat, and the sample fully immersed in a 1 M concentrated alkalisolution taking appropriate safety precautions. Precautions are also necessaryto prevent evaporation during the test period, which is usually up to28 days at 20 C, although this may be accelerated by higher temperatures.Small gel growths on the prepared surface may be monitored visually, andany reactive particles identified. Comparative photography before and aftertesting is essential to confirm and record results.9.9 AdmixturesAdmixtures are frequently organic and are not easy to identify withoutthe aid of sophisticated equipment such as infrared absorption spectrophotometersor high-pressure liquid chromatographs. Assessment of the dosageused (which is frequently the prime object of the test) is dependent upona precise knowledge of the admixture. Owing to the very small amountsof admixture often used, analysis is fraught with difficulty and prone tointerference by impurities in the concrete and aggregates. A recent paperby Roberts (392) suggests that sulfate content and phase distribution inconcrete can have dramatic effects on the performance of admixtures inconcrete. A book by Rixom (393) discusses the use and applications ofadmixtures.9.10 CarbonationCarbonation of concrete by attack from atmospheric carbon dioxide willresult in a reduction in alkalinity of the concrete, and increase the risk ofreinforcement corrosion. This will normally be restricted to a surface layerof only a few millimetres thickness in good quality concrete but can bemuch deeper in poor quality concrete, with results as high as 30 mm beingnot uncommon. The extent of carbonation can be easily assessed by treatingwith phenolphthalein indicator the freshly exposed surfaces of a piece ofconcrete which has been broken from a member to give surfaces roughlyperpendicular to the external face. Alternatively, incrementally drilled powderedsamples may be sprayed or allowed to fall on indicator-impregnatedfilter paper. Drilled cores may also be split and sprayed with indicator toshow the carbonation front. A purple-red coloration will be obtained wherethe highly alkaline concrete has been unaffected by carbonation, but nocoloration will appear in carbonated zones.The colour change of phenolphthalein corresponds to a pH of about 9.2whilst reinforcement corrosion may possibly commence at a pore solutionpH of less than about 11. Thus the carbonation front indicated must berecorded as approximate in relation to steel corrosion. The width of carbonationfront between these values is of the order of 2–3 mm (394), but

286 Chemical testing and allied techniquesan overall accuracy of ±5 mm is sometimes quoted for a single test usingthis approach. Accuracy can be improved by taking the mean of severalreadings, for example five readings can give an accuracy of ±2 mm at the95% confidence level. Multicoloured ‘Rainbow’ indicators are also available(316) which may provide slightly more detailed information.A number of practical difficulties may sometimes arise as discussed byTheophilus (394). These include:(i) Freshly broken concrete giving an initially clearly defined colour changeboundary may all become coloured within about 1 minute, making theboundary indistinguishable.(ii) Concrete may immediately register entirely uncarbonated, despitemicroscopic studies showing a significant carbonated depth. (This isless common.)(iii) Freshly broken concrete initially registers entirely carbonated, but onsubsequent standing the pink coloration appears. This can occur withboth white concretes and also those made with so-called ‘reconstitutedstone’.These effects are all due to patchy, diffuse carbonation and the effect ofhydration of cement grains between the carbonated areas. The speed ofdevelopment of the colour in this test is therefore critical in judging thedepth of carbonation. The colour change should be almost instantaneous.A slow change usually indicates partial or patchy carbonation.Doubtful results may often be clarified by spraying the surface withdeionized water immediately prior to spraying with phenolphthalein, andexcessive quantities of indicator should be avoided. Only the finest mistspray is required.Direct chemical tests may be used to determine the carbonation front withgreater precision and may be justified in critical situations. These include themeasurement of evolved carbon dioxide from slices of cores, about 5 mmthick (not for calcareous aggregates), using a range of specialized techniquessuch as thermogravimetry.Microscopy is probably the most precise method of measuring carbonation,including localized effects of surface cracks (395). Thin sections areviewed under cross-polarized light to reveal calcium carbonate crystals, butcalcareous fines may hinder identification of carbonation.The progression of carbonation with time is often estimated by theexpressionD = K √ twhere D is the depth in mm at age t years and K is a constant for theconcrete and conditions prevailing (395).

Chemical testing and allied techniques 287The use of this expression to predict the time for the carbonation frontto reach a specified depth will magnify the effects of the accuracy range towhich the measured carbonation value has been obtained. Watkins and PittJones (396) have suggested that this above expression is a simplificationand that a more reliable relationship isD = Kt xwhere x lies between 0.5 and 1.0. Good agreement has been shown forextensive site data from Hong Kong with K and x varying according tostrength level. Carbonation rates are usually higher in dry or shelteredexternal concrete than in those exposed to rain; thus differing faces of astructure or element are likely to show different carbonation depths. Thesefactors have been discussed more fully by Somerville (3), whilst assessmentof resulting corrosion risk is summarized by Sims (395).9.11 Microscopic methodsThese fall into two basic categories, namely methods involving examinationof a prepared concrete surface by reflected light and those requiring a‘thin section’ to be obtained. These, and other petrographic methods, areconsidered in detail by the Concrete Society (372) and by French (377).9.11.1 Surface examination by reflected lightMajor internal crack patterns caused by alkali–silica reactions can be examinedby viewing a polished cut concrete surface which has been sprayed orimpregnated with a fluorescent dye under ultraviolet light. A portable fieldinspection kit is also available for use on a treated in-situ surface (397),but care is needed to allow for weak fluorescence, which may be caused bycarbonated concrete (398). Specialized laboratory techniques can also beused to determine cement, aggregate and air content of samples taken fromin-situ concrete.9.11.1.1 TheoryA varnished sawn face of a dried concrete specimen can be examined bystereomicroscope to give the volumetric proportions of a hardened concrete.Polivka (399) has described a ‘point count’ method based on the principlethat the frequency with which each constituent occurs at equally spacedpoints along a random line on the surface will reflect the relative volumesof the constituents in the solid. This is because the relative volumes ofthe constituents of a heterogeneous solid are directly proportional to the

288 Chemical testing and allied techniquesrelative areas on a plane section, and also to the intercepts of these areasalong a random line on the section. An alternative approach is the lineartraverse technique, in which the intercepts of the constituents are measuredalong a series of closely spaced regular transverse lines. In either case theaggregate and voids can be identified, and the remainder is assumed to behydrated cement. The total volume of this can be computed and convertedto a volume of unhydrated cement if the specific gravity of the dry cementand non-evaporable water content of the hydrated cement are known.9.11.1.2 ProcedureThe cut surface must be carefully prepared so that the constituents arereadily distinguishable. This is best achieved by grinding, polishing andimpregnation with a suitable dye before varnishing. The prepared samplewill normally be examined by a stereomicroscope with a travelling specimenstage, which may be manually or motor driven. Counting will bemanually controlled but may conveniently be linked to a microcomputerwhich is coupled to the moving stage to record its location automatically.A magnification of 50× is generally used.ASTM C457 (263) concerns the measurement of entrained air using eitherthe linear traverse or modified point count techniques. In the linear traversemethod, records are kept of the total number of sections of air voids, thetotal distance across air voids and the total distance across the remainder.The modified point count method involves the recording of the frequencyof each component coinciding with a regular system of points, coupledwith the frequency of intersection of each component by regularly spacedlines.Both methods permit calculation of the number, size and spacing ofthe air voids to be determined. The size is usually expressed in terms of thespecific surface, whilst the commonly used ‘spacing factor’ represents themaximum distance from any point in the paste to the periphery of an airvoid. Markestad (400) has described the application of this approach tomeasurement of air voids in hardened concrete from offshore platforms, andalso gives details of the use of an automatic image-analysing microscope forthis work. This has the advantage of being much quicker than the tediousstandard microscopic techniques. The optical image of a point on the groundsurface is projected onto the photosensitive surface of a TV camera tubeand converted to an electrical signal. The position and strength of eachsignal can be classified and stored for processing, and a data printout givesthe area or volume percentage, and specific surface, of air voids togetherwith details of scanning of a particular area. Special surface preparation isnecessary to fill the pores with white gypsum whilst the remainder of thesurface is blackened.

Chemical testing and allied techniques 2899.11.1.3 Reliability and applicationsIt is claimed that this method can measure the cement content to ±10%, andthe total aggregate content and coarse/fine aggregate ratio can be assessedsimilarly. However, the total water content cannot be assessed, since waterand air voids cannot be distinguished, except for entrained air voids whichwill be identifiable by their spherical nature and uniformity of size. In situationswhere cement content cannot be determined chemically this approachmay be valuable. Details of precision experiments are given by the ConcreteSociety (372).This method has also become accepted for measurement of air entrainment.The choice between point count or linear traverse techniques willdepend on circumstances, but Markestad (400) has indicated problemsof component identification leading to uncertainties with the point countapproach. Whichever method is used, the procedures are tedious andrequire specialized equipment and skill in both sample preparation andmeasurement. Modern electronic aids have eased the burden of data collation,and automatic image-analysing microscopes will probably be morewidely used.9.11.2 Thin-section methodsApplications of microscopic examination of thin sections have been outlinedin earlier sections of this chapter, and include identification of mix components(372), carbonation and causes of deterioration. Considerable growthin usage of this approach has been experienced in connection with alkali–silica reaction where the method is invaluable in confirming that reactionhas occurred, examining the size and extent of cracks, and identifying reactiveaggregate particles as illustrated in Figure 9.7. When diagnosing ASR asa contributory or sole cause of damage, it is essential that cracks radiatingfrom a reactive particle, with gel present, are observed (the so-called ‘sitesof expansive reaction’). The mere presence of gel is not conclusive evidenceof expansive ASR.Sample preparation involves cutting a slice of concrete from a core bydiamond saw (preceded if necessary by vacuum resin impregnation), dryingand impregnation by low viscosity epoxy resin. This will then be cut andground using standard petrographic procedures to a 30 m thickness usingoil lubrication to avoid the dissolution of water-soluble materials. Detailedprocedures have been described by Poulsen (401), and are summarized byPalmer (33).Samples will typically be examined with a petrographic microscope underordinary and polarized light. Micrographs may be produced for recordpurposes and to illustrate interpretation, which is highly specialized.

290 Chemical testing and allied techniquesFigure 9.7 Photomicrograph of alkali–silica reaction (photograph by courtesy of ProfessorE. Poulsen). Note: An opaline chert particle is seriously affected by ASR, withcracks radiating into the surrounding paste. The air void in the lower right-handcorner is also partially filled with gel.9.12 Thermoluminescence testingIt has been proposed by Placido (402) that the thermoluminescence ofsand extracted from concrete can form the basis of a test for fire-damagedconcrete, which is a measure of the actual thermal exposure experienced bythe concrete.9.12.1 TheoryThermoluminescence is the visible light emission which occurs on heatingof certain minerals, including quartz and feldspars, and it is known thatthe curve of light output vs. temperature for a given sample depends uponits thermal and radiation history. This forms the basis of established techniquesof mineral identification, radiation dosimetry and pottery dating.In naturally occurring quartz sand, this light emission occurs within thetemperature range of 300–500 C, but if samples are reheated there is noemission of light up to the temperature of the preceding heating. The subsequentpattern of light emission has also been shown to depend on theperiod of exposure to the particular temperature (Figure 9.8).

Chemical testing and allied techniques 291Figure 9.8 Typical thermoluminescence trace (based on ref. 402).9.12.2 Equipment and procedureThe equipment required is somewhat complex, including a servo-controlledglow oven to heat the sample, of only a few mg, in an oxygen-free nitrogenatmosphere. Radiation is detected by a photomultiplier and fed intoa photon-counter which drives the Y-axis of an X–Y recorder whilstthe X-axis is driven by a thermocouple monitoring the heating platetemperature. The small sample required should be drilled by slow drillingwith a small battery-powered masonry drill in order to reduce drill bittemperatures, and the sample should be washed in concentrated acid solutionto remove minerals which may give spurious outputs. The remainingquartz sand may then be tested, although if there is none present it may bepossible to utilize quartz from coarse aggregate, or to modify procedures touse other minerals which may be present in the aggregates.9.12.3 Reliability, limitations and applicationsThe practical value of this test for concrete assessment hinges upon the factthat the emission temperature range of 300–500 C is critical in the deteriorationexperienced by concrete subjected to fire, and it is also believedthat loss of strength is influenced by thermal exposure as well as maximumtemperature. Whilst little data relating thermoluminescence results to

292 Chemical testing and allied techniquesconcrete strength are available, Chew (403) has shown that correlationsbetween these properties do exist. Sampling is very quick and cheap andthe test may be particularly valuable in checking where high temperaturesare suspected but there is no visible damage. The depth of heat penetrationinto a member can also be conveniently monitored by taking samples atprescribed depths. The method is still at the development stage but doesappear to offer considerable potential for the detailed investigation of theextent and severity of fire damage.9.13 Specialized instrumental methodsA number of highly specialized laboratory techniques are available to thecement chemist. These require expensive equipment and considerable experience,but may be used to identify various characteristics of the hardenedconcrete. Some of those which are more commonly used are outlinedbelow.9.13.1 X-ray fluorescence spectroscopyA sample of concrete is bombarded by high energy X-rays and the fluorescentemission spectrum so caused is collimated into a parallel beam, directedon to the analysing crystal within a spectrometer and reflected into a detector.The wavelengths and densities of the fluorescent emission are measuredand the constituent elements, together with their proportions, can be calculatedfrom these data. The sample of concrete must be in the form of a pelletof suitable density, formed by compressing a dried, finely ground sampletogether with a binder under very high pressure. The preparation of such asample (40 mm in diameter) by compaction of 10 g concrete for 10 secondsby a load of 20 tonnes is described in ref. 383, for use in the determinationof cement content and chloride content. Actual analysis time is very short,and samples may be reused as required. The method is comparative – theemission results are compared with samples of known proportions in termsof the component under investigation. An alternative method is to fuse thesample in another material such as lithium metaborate. This latter method,while more time-consuming, can produce more consistent results.9.13.2 Differential thermal methodsDifferential thermal analysis (DTA) is the best known of these methods,because of its important role in the assessment of HAC concrete. This methodinvolves heating a small sample of powdered concrete in a furnace togetherwith a similar sample of inert material. The rate of temperature rise of theinert sample is controlled to be as nearly uniform as possible, and is measuredby thermocouple. The test sample is similarly monitored to provide a

Chemical testing and allied techniques 293trace of the temperature difference between the two specimens, and this tracewill have a series of peaks at particular temperatures which are characteristicfor the minerals in the sample. These correspond to the loss of waterof crystallization of the various mineral forms or thermal decomposition ofmineral components, and in general form will identify the presence of particularminerals. The method may also be made quantitative by calibratingthe apparatus against suitable pure minerals to relate peak height to mass.Differential thermal analysis, together with the similar methods of differentialscanning calorimetry (DSC) and derivative thermogravimetry (DTG),has been described by Midgley (404) and used for following the hydrationprocesses of high alumina concretes. Midgley and Midgley have alsodescribed their application to measuring the degree of conversion of HACconcrete (405). Both DTA and DSC have become established techniquesfor this purpose. As the relative proportions of the hydration productschange with time due to conversion, so do the heights of the correspondingendotherms, enabling the composition of the matrix and degree of conversionto be determined. Although primarily used to establish degree ofconversion, the method will also indicate the presence of sulfate attack orother cement type. Particular care is required in sampling to avoid heatingof the sample and resulting moisture loss from the contained materials,which will reduce the reliability of results. Further problems can occur ifclay minerals or mica are present in the aggregate. A powdered sample willnormally be obtained by drilling the in-situ concrete using a rotary percussiondrill, although a larger sample, such as a core, may also be groundand used. In experienced hands the method is reliable, and an accuracyof ±5% is possible for HAC percentage conversion measurements. It isnow considered that in the UK most HAC concrete in the field which waspre-early 1970s will have fully converted and the conversion test is nowlittle used. Assessment of HAC concrete is based on visual inspection, petrographicexamination and in-situ strength estimation, using, for example,the BRE Internal Fracture Test. More emphasis is now placed on the estimationof carbonation in HAC, because of the possible effects on corrosionof prestressing tendons (406).Whilst HAC applications have made this method known to many engineers,other applications are also possible, although they are likely to beprimarily of a research nature.9.13.3 Thermogravimetry, X-ray diffraction, infraredand atomic absorption spectrometry and scanningelectron microscopyThese are all highly specialized techniques which may be used for analysisof the constituents of hardened concrete. Although involving high capitalexpenditure, they are likely to permit more rapid throughput than chemical

294 Chemical testing and allied techniquesapproaches and to be able to deal with more complex problems. Manytesting organizations have developed their own expertise and procedureswith specific methods, and use is growing steadily, often in conjunctionwith classical techniques. Descriptions of the techniques and applicationsto the testing of cement and concrete have been given by Ramachandran(407). If, however, precise details of the testing procedures are required,reference should be made to texts such as that by Willard et al. (408). Lasermethods under development can remotely assess chloride content (382) andother features of the hardened concrete at the surface.

Appendix ATypical cases of test planningand interpretation of resultsThe engineer has complete and absolute authority as to whether concrete iscondemned or accepted. The problem of testing, and interpretation of theresults, will however be approached in a variety of ways – specifications,which will be used as the basis for decisions, vary widely and in some casesmay legally empower the engineer to condemn concrete if the cubes orcylinders fail irrespective of the condition or quality of the in-situ concrete.Many factors can however vitiate cube results including variations due tofailure to observe the required standardized procedures for sampling, manufactureand curing of the cubes. Further errors may also be introduced bythe testing operative or inaccuracies in the testing machine, although theseshould be checked by regular comparative reference testing. Whilst testingof the in-situ concrete eliminates most of these sources of error, specificationsrarely mention in-situ strength and Codes of Practice do not definethe in-situ strength required. Current British and European Standards, however,imply that an in-place strength of f cu /15 is expected for in-situ work bythe adoption of a partial factor of safety of 1.5 when calculating the designconcrete strength to use in calculations. If a design is based on some otherCode of Practice this value may vary and may reflect the use of cylinderrather than cube values (as in the USA for example), but the basic principlethat in-situ strength is recognized as being lower than standard cube specimenstrength remains. It is to be hoped that engineers facing the problem offailed cubes will consider these aspects, as well as the in-situ requirements ofthe concrete and possible errors in the cube results, before taking decisions.The following examples are included to assist the planning of tests andinterpretation of results for this and other commonly occurring situations.A128-day cubes fail (cube results suspect)(i) Problem28-day cubes from concrete used for in-situ beam construction are of lowstrength.Visualexaminationofthecubessuggeststhattheywerepoorlymade.

296 Appendix A(ii) Aims of testingThe principal aim will be confirmation of specification compliance of thein-situ concrete, followed by determination of structural adequacy if noncomplianceis indicated.(iii) ProposalsSchmidt hammer and/or ultrasonic pulse velocity testing of the beams issuggested, either by way of comparison with similar members of knownacceptable standard cube strength, or to yield an absolute in-situ strengthprediction based on a specifically prepared calibration for the particularmix. Visual comparison may also prove valuable. Tests should be locatedat mid-depth or spread evenly to provide a representative average value ofin-situ strength for comparison with the specified value after application ofsafety factors. If doubt still exists then cores can be used, located to giverepresentative values from the suspect concrete.(iv) InterpretationIf similar members are available for comparison with those that are suspect,the combined raw non-destructive test results should be plotted inhistogram form and the overall coefficient of variation calculated to indicatethe uniformity of concrete between members. The mean values for eachgroup should also be calculated for comparison, bearing in mind possibleminor variations due to age differences. If non-uniformity is indicated, relativestrengths may be estimated from the mean non-destructive results foreach group.If similar satisfactory members are not available, or comparative nondestructivetesting suggests borderline values, strength calibration charts forthe particular mix may be used. If these are not available, or cannot beobtained, then cores cut to examine representative concrete of the suspectmembers should be used to estimate the equivalent ‘standard’ cube strength(or potential cube strength).(v) Numerical exampleSpecified 28-day characteristic cube strength = 30 N/mm 2 . Schmidt hammerand UPV comparisons with similar beams cast one week earlier show onlyone peak in histogram form, and have the following values.Schmidt hammer: mean rebound no = 32coefficient of variation = 6%

Appendix A 297UPV: mean velocity = 415 km/scoefficient of variation = 4%Mean standard 28-day cube strengths for comparison beams = 34 N/mm 2 ,henceEstimated mean standard 28-day cube strength for suspect beams= 32 N/mm 2 based on Schmidt hammer results= 35 N/mm 2 based on ultrasonic pulse velocities(using calibration curves of standard form).Whilst NDT results suggest only one supply and reasonable constructionquality, the estimated mean equivalent standard 28-day cube strength forthe suspect concrete is low in relation to the expected value of 30 + 164swhere s = 5N/mm 2 (corresponding to ‘normal’ standards – Table 1.8)giving 38 N/mm 2 in-situ strength.The effects of age difference on these values can be assumed to lead toa small underestimate of true strength, but this cannot be relied upon. Itappears therefore that whilst the mean value of strength for the suspectbatches is above the minimum specified, the proportion of concrete likelyto be below this value may be greater than normally permitted.If further evidence is required cores should be used. These should betaken near mid-depth of a typical beam and at least four should be usedto provide an estimate of potential strength as described in Appendix C.This estimate, which will have at best an accuracy of ±15%, can then becompared with the absolute minimum specified (085 × 30 = 255N/mm 2for BS 8500).Core resultsestimated potential strength = 27293235Mean = 31 ± 45N/mm 2Since the mean is above the characteristic specified value, and all resultsexceed the minimum acceptable value, it cannot be proved conclusivelythat the concrete does not meet the specification. Such results could wellform part of an acceptable spread of values with a characteristic strengthof 30 N/mm 2 and the concrete should not be rejected.

298 Appendix AA228-day cubes fail (cube results genuine)(i) ProblemA suspended floor slab has been cast from several batches and the 28-daycube strengths (6 cubes) are below the characteristic value required. Thereis no reason to suspect the validity of the cube results.(ii) Aims of testingThe initial aim will be to establish whether the low cubes are representativeor relate to isolated substandard batches. The subsequent aim will then beto assess structural adequacy.(iii) ProposalsVisual inspection may indicate uniformity or otherwise of the slab in thefirst instance. This can be followed by Schmidt hammer tests on the soffitto confirm uniformity. Direct ultrasonic tests may prove difficult, and withindirect readings on the soffit being unreliable cores should be taken fromtypical zones. In cases of doubt, in-situ load testing may be necessary.(iv) InterpretationSchmidt hammer readings should be taken on a regular grid, and plottedon a ‘contour’ plan to indicate the degree of uniformity. A histogram plotmay also be worthwhile to provide confirmation of this. The results fromcores can be used to estimate an average actual in-situ strength which canthen be related to the required design strength with an allowance for likelystandard deviation of in-situ results.(v) Numerical exampleSpecified characteristic 28-day cube strength = f cu = 40 N/mm 2Mean ‘standard’ 28-day cube strength = 38 ± 2N/mm 2Mix design mean strength = 46 N/mm 2Visual inspection and non-destructive test results suggest uniform constructionacross the slab.Average estimated in-situ cube strength from six cores = 30 ± 15N/mm 2Estimated characteristic in-situ strength range ={ 315 − 164 s ′ N/mm 2285 − 164 s ′ N/mm 2

Adopt estimated in-situ standard deviation s ′ = 45N/mm 2Appendix A 299based on scatter of Schmidt hammer results and past site cube records whichshow ‘good’ control (see Table 1.8).Hence{ 315 − 75 = 24 N/mm2estimated characteristic in-situ strength =285 − 75 = 21 N/mm 2If the design is based on a partial factor of safety on concrete strength of1.5, minimum acceptable design strength = 40/15 = 26 N/mm 2 . Since theestimated range of characteristic in-situ strength lies below the minimumdesign strength, the concrete must be considered unacceptable. If the slabis to be permanently dry, the estimated in-situ values may be increased by10%, but this factor will not be sufficient to accept the concrete unless theslab is not critically stressed. It is recommended that an in-situ load testbe undertaken to establish directly the serviceability behaviour of the slab.Likely durability performance must also be considered based on exposureconditions.A3Cubes non-existent for new structure(i) ProblemA large number of columns have been cast, and the cubes lost.(ii) Aims of testingThe principal aim will generally be confirmation of acceptability of thein-situ concrete from the point of view of strength and durability.(iii) ProposalsA comprehensive comparative survey, using surface hardness and/or ultrasonicpulse velocity. Visual inspection may also indicate lack of uniformity.Plot raw results to indicate patterns and then follow-up with a limitednumber of cores at points of apparently lowest and highest strength, unlessreliable calibrations for the mix are available or can be obtained. If similarmembers are available of concrete which is known to be acceptable it maybe adequate to rely on non-destructive comparisons with these, using coresonly in cases of extreme doubt. Reserve crushed cores for chemical analysisto determine the cement content if strengths indicate durability doubts.(iv) InterpretationIt is important that the non-destructive results to be used comparativelyare taken at comparable points in relation to the members tested. This is

300 Appendix Abecause of the likely within-member strength variations, and measurementsshould preferably be taken at points of expected lower strength (i.e. nearthe top of the columns). Sufficient readings should be taken to encompassthe various batches of concrete that may have been used.Results should be plotted in histogram form to detect the weakest areas;coefficients of variation may also provide valuable confirmation of constructionuniformity. The cores should provide values for minimum strength tobe compared with a calculated minimum acceptable value from the design,as well as a rough calibration for the non-destructive tests. If attemptsare to be made to relate results to specifications, the likely within-membervariations and in-situ/standard specimen strength must not be overlooked.Under normal circumstances a minimum in-situ strength of characteristic/1.5would be acceptable for design to British or European Standards,whilst an even lower value may be adequate for low stress areas, subject toadequate durability as indicated by cement content.(v) Numerical exampleSpecified characteristic cube strength f cu = 30 N/mm 2For design to British Standards assumeminimum acceptable in-situ strength = f cu= 20 N/mm215For four cores taken to correspond with lowest pulse velocities and reboundnumbers:estimated in-situ cube strengths = 205N/mm 2250N/mm 2225N/mm 2210N/mm 2Mean = 220N/mm 2Thus estimated minimum in-situ strength = 22 ± 15N/mm 2 .Hence the mean and all results are above minimum acceptable value, and theconcrete will be considered adequate. It will follow that the remainder of theconcrete is also acceptable since these results relate to the worst locations.Note (1): If either individual results or the mean estimated in-situ strengthare below the minimum acceptable as calculated above, detailed considerationshould be given to design stress levels and service moisture conditions.Note (2): If the concrete strength is critical to the design, or if calibrationshave been used with surface hardness or UPV results to estimate strength

Appendix A 301without cores, it may be appropriate to include a factor of safety to accountfor this, e.g. maximum acceptable design stress = 22/12 = 18 N/mm 2 basedon the factor of safety of 1.2 recommended by BS 6089 (13).A4Cubes damaged for new structure(i) ProblemA series of elements have been cast but the cubes damaged. The scope forin-situ testing is very limited due to access difficulties.(ii) Aims of testingThe principal aim is to estimate the in-situ characteristic strength.(iii) ProposalsUse statistical procedures to establish the likely in-situ characteristic strengthbased on tests at six accessible locations.(iv) InterpretationFor the six results obtained, the mean estimated in-situ strength is36 N/mm 2 and the coefficient of variation of these results is 12%.Using the ‘normal’ distribution relationship illustrated in Figure 1.10,it can be seen by interpolation that for six values, the estimated in-situcharacteristic strength with 95% confidence limits isf cu ⋍ 07f c= 07 × 36 = 25 N/mm 2 A5Cubes non-existent for existing structure(i) Aims of testingThe aim of testing will be to provide a concrete strength estimate for use indesign calculations relating to a proposed modification of the structure.(ii) ProposalsSurvey by ultrasonic pulse velocity, Capo, pull-off or Windsor probe correlatedwith a limited number of cores according to practical limitations.Tests to be spread as representatively as possible over members underexamination.

302 Appendix A(iii) InterpretationUse mean estimated in-situ cube strength to obtain a value of design strengthwhich takes account of the standard of construction quality as well asuncertainties about the adequacy of the in-situ test data.(iv) Numerical exampleEstimated mean in-situ cube strength = 25 N/mm 2Assumed mean ‘standard’ cube strength = 25 × 15 = 375N/mm 2for ‘normal’ construction quality (unless evidence from scatter of testresults suggests otherwise), standard deviation of control cubes estimatedat 5 N/mm 2 (Table 1.8).Hence estimated characteristic ‘standard’ cube strength = 375 − 164 × 5= 29 N/mm 2 Allowance should be made for errors in test data by a factor of safety (1.2suggested by BS 6089 (13)).Hence maximum design stress using a partial factor of safety of 1.5 onconcrete strength for design equals:2912 × 15 = 16 N/mm2 Alternatively, if the test results are taken to correspond to locations oflowest anticipated strength:Maximum design stress = estimated minimum in-situ cube strength, givenby mean of test results/1.2.A6Surface cracking(i) ProblemThe wing wall to a highway bridge abutment shows random surface cracksand spalling several years after construction.(ii) Aims of testingThe principal aim will be identification of the cause of deterioration followedby an assessment of present and future serviceability. Apportionmentof blame may follow.

Appendix A 303(iii) ProposalsVisual inspection of crack patterns, and their development with time,may permit preliminary classification of cause as (a) structural actions,(b) shrinkage or (c) material deterioration. This may be followed bystrength assessment as in A2 if structural actions are suspected, or chemical/petrographictesting if material deterioration is likely. Cores may convenientlybe used to provide suitable samples and should be taken from theareas most seriously affected. Chemical testing to detect chlorides or sulfateswill be selected according to the crack pattern whilst microscopic examinationcan check for frost action, alkali–aggregate reaction and entrained aircontent.Serviceability will be determined on the basis of the extent of deteriorationand the ability to prevent worsening of the situation.(iv) InterpretationReference to Table 1.2 will assist preliminary identification. Shrinkagecracks are likely to occur at an early age and follow a recognizable pattern,as do cracks due to structural actions. Material deterioration is thereforeindicated in this case, and may be due to chemical attack from internal orexternal sources or due to frost action. Chloride attack is unlikely sincethe cracks do not follow the pattern of reinforcement, thus initially test forsulfate and cement content. If the results of these tests indicate acceptablelevels, petrographic examination will be necessary to attempt to identifyaggregate–alkali attack or frost action. If frost action is indicated, micrometricexamination will yield an estimate of the entrained air content forcomparison with the specified value. Expansion and alkali-immersion testson cores may be required if alkali–aggregate reaction is found.If future deterioration can be prevented by protection of the concretefrom the source of attack, this should be implemented after such cuttingout and making good as may be necessary. If the source of deterioration isinternal and not of a localized nature it may prove necessary to replace themember once it reaches a condition of being unfit for use.A7Reinforcement corrosion(i) ProblemA major modern in-situ concrete structure is showing numerous rust-stainedcracks, and in some cases pieces of concrete have spalled exposing seriouslycorroded reinforcement. Repairs are obviously necessary to preventcontinued deterioration and to restore the appearance of the structure.

304 Appendix A(ii) Aims of testingThe principal aim will be to establish the reason for the corrosion as wellas its (present and likely future) extent (which may not be visible) to permitthe design of repair proposals. Litigation may follow. Structural adequacymust also be checked.(iii) ProposalsA comprehensive covermeter survey coupled with phenolphthalein carbonationdepth measurements should be undertaken. If access costs are high,this may be limited initially to about 25% of the areas involved althougha full survey will be necessary prior to repair. A limited number of chloridetests at the level of the steel using simple methods (Hach or Quantab)should indicate acceptable or excessive chloride levels. A limited number ofcores should also be taken for compressive strength testing. If carbonation isfound to be excessive these should also be tested for absorption and cementcontent. Further cores or more detailed chemical analysis may be required(chloride, cement content and possibly water/cement ratio) depending uponresults for litigation purposes.A comparative half-cell potential survey may be useful in determining thecorrosion risk in apparently undamaged regions unless this can be establishedfrom the covermeter and carbonation survey. ISAT measurementsmay also indicate the extent of highly absorptive concrete if that is foundto be present.(iv) InterpretationThe most likely causes of the problem are the existence of inadequatecover and/or excessive carbonation. These will be immediately apparentfrom the covermeter and phenolphthalein survey. The reasons for excessivecarbonation may be deducible from the tests on cores, and may include mixdeficiencies indicated by low (or variable) strength or low cement content,and inadequate curing possibly indicated by high water absorption. Corestrength results coupled with careful inspection of corroded steel shouldenable structural adequacy to be checked, and the need for additional steelor strengthening to be determined. If chloride content is found to be high,more extensive testing to provide surface zone gradients may help to identifylikely sources. A half-cell potential survey may be useful in this situation tohelp plan the repair strategy.The information provided by these tests may be adequate to permit thedesign of repairs, but deterioration is often the result of a combinationof several factors and it is important that all reasonable possibilities areconsidered before conclusions are reached.

Appendix BExamples of pulse velocitycorrections for reinforcementSee Section 3.3.2.5 for full details of notation and procedures.Example B1The beam contains 16 mm diameter main bars with 10 mm links as shownin Figure B1. The measured pulse velocity across the top of the beamaway from links is 4.2 km/s. Determine the anticipated measured apparentvelocity when readings are made directly in line with a link.As main bar will have no practical influence on measured values V c =42km/s. Hence from Figure 3.15 for V c = 42km/s and 10 mm bar, = 088Figure B1

306 Appendix Bthus allowing for 25 mm cover at each end of linkthusL sL = 150200 = 075k = 1 − 0751 − 088= 091∴ anticipated V m = 42 = 462 km/s091Example B2An uncracked beam shown in Figure B2 has well bonded 20 mm bars with50 mm end cover. Measurements at 150 mm offset yield V m = 465 km/s.Estimate the true pulse velocity in the concrete (V c ).a>2b and a/L = 0075For trial values of V c evaluate kV m as in Table B1, and plot as shown inFigure B3.On the basis of these values, try V c = 410 then = 079k= 088 henceV m = 410 = 466 km/sOK088Figure B2Table B1Trial V c (Fig. 3.15) k (Fig. 3.16) kV m V c − kV m 4.5 km/s 0.87 0.94 4.37 km/s +0133.5 km/s 0.68 0.79 3.67 km/s −017

Appendix B 307Figure B3thusestimated V c = 410 km/s(note RILEM ‘maximum effect’ factors suggest V c = 375 km/s).Example B3Measurements across a 300 mm wide beam containing three No. 32 mmbars give a value of V m = 44km/s. Estimate V c .ThenL sL = 3 × 32300 = 032For trial values of V c , evaluate kV m as in Table B2 and plot Figure B4.Table B2()Trial V c Fig318 k = 1 − L s1 − LkV mV c − kV m 4.5 km/s 0.92 0.974 4.28 km/s +0224.0 km/s 0.85 0.952 4.19 km/s −019

308 Appendix BFigure B4Thus try V c = 425 km/s, then = 089 and k = 0965, henceV m = 425 = 440 km/s OK0965thusestimated V c = 425 km/s(note: RILEM ‘maximum effect’ factors suggest V c ≃ 39km/s)

Appendix CExample of evaluation of coreresultsA 100 mm diameter core drilled horizontally from a wall of concrete with20 mm maximum aggregate size contains one 20 mm reinforcing bar normalto the core axis and located at 35 mm from one end.Measured water-soaked concrete density = 2320 kg/m 3 after correction forincluded reinforcement (Section 5.1.3.4).Measured crushing force = 160 kN (following BS EN 12504 (135) procedure)Failure mode – normal.Measured core length after capping = 120 mm.(i) Concrete Society (36) in-situ cube strength160 × 103Measured core strength = = 205N/mm 2 × 10024Core length/diameter ratio = 120/100 = 1225Estimated in-situ cube strength = (15 + 1 ) × 205 for a horizontal core12= 22 N/mm 2( 20Reinforcement correction factor = 1 + 15100 × 35 )120= 109Corrected in-situ cube strength = 22 × 109 N/mm 2 ± 12% for an individualresult.= 240 ± 3N/mm 2

310 Appendix C(ii) Concrete Society (36) potential cube strengthMeasured core strength = 205N/mm 2Core length/diameter ratio = 12325Estimated potential cube strength = (15 + 1 ) × 205 for a horizontal12core= 285N/mm 2Reinforcement correction factor = 109Potential density of concrete = 2350 kg/m 3 (mean value from cubes)( )2350 − 2320Excess voidage =× 100%2350 − 500= 16%Strength multiplying factor = 114 (Figure 5.4)Corrected potential cube strength = 285 × 109 × 114= 355N/mm 2(Note: Accuracy cannot be realistically quoted for a single result but themean estimated potential cube strength from a group of at least fourcores may be quoted to between ±15% and ±30% subject to a proceduredescribed by Technical Report No. 11 (36) to eliminate abnormallylow results. In this instance because of the use of reinforcement and excessvoidage correction factors, the estimated accuracy for a group of four isunlikely to be as high as ±15%.)(iii) Procedure to eliminate abnormal results (36)This requires that for n cores (where n is at least 4) the lowest value isseparated and the value t calculated fromt =mean of remainder − lowest√mean of remainder × 6× 1 + 1100n − 1The value t is then compared with Table C1, and if greater than the valuein column A, corresponding to the total number of cores n, the lowest coreresult is discarded if there is any evidence of abnormality in relation to theothers.This may be in terms of location, reinforcement, compaction, cracks ordrilling damage. If the value of t is greater than that in column B, the lowestresult is discarded irrespective of other considerations. The mean value

Appendix C 311Table C1No. of cores (n)tAB4 29 435 24 326 21 287 20 268 19 25obtained from the remaining n − 1 cores is then taken as the estimatedpotential strength.If an abnormally high result is obtained, although this is less likely, thesame procedure can be adopted but substituting the highest value for thelowest. Applying this procedure to the results used in Section A1, i.e. fourcores with potential strengths 27 29 32 35 N/mm 2 ,32 − 27t = √32 × 61 + 1 100 3= 226This is less than the value 2.9 from column A, Table C1, and the quotedmean and accuracy may be considered valid.

References1 Petersen, C.G. and Poulsen, E. Pull-out testing by Lok-test and Capo-test. DanskBetoninstitut A/S, 1992.2 BS EN 206-1: Concrete: Specification, performance, production and conformity.British Standards Institution, London.3 Somerville, G. The design life of concrete structures. Structural Engineer, 64A,No. 2, Feb. 1986, 60–71.4 Clifton, J.R. Predicting the service life of concrete. ACI Materials Journal, 90,No. 6, Nov./Dec. 1993, 611–617.5 Inspection and Maintenance of Reinforced and Prestressed Concrete Structures.Federation Internationale de Pre-contrainte (FIP), Thames Telford Ltd, London,1986.6 Appraisal of existing structures. Institution of Structural Engineers, London, 2ndEdn, 1996.7 RILEM. Draft recommendation for damage classification of concrete structures.Materials and Structures, 27, 1994, 362–369.8 ACI Committee 364. Guide for evaluation of concrete structures prior to rehabilitation.ACI Materials Journal, 90, No. 5, Sept./Oct. 1993, 479–498.9 Currie, R.J. and Crammond, N.J. Assessment of existing high alumina cementconstruction in the UK. Proc. ICE, Structs & Bldgs, 104, Feb. 1994, 83–92.10 Assessment and repair of fire-damaged concrete structures. Tech. Rept. 33,Concrete Society, London, 1990.11 Fleischer, C.C. and Chapman-Andrews, J.F. Survey and analysis of bomb damagedreinforced concrete structures. Proc. Struct. Faults and Repairs 93, Eng.Technics Press, Edinburgh, 2, 1993, 111–118.12 BS 1881: Part 201 Guide to the use of non-destructive methods of test forhardened concrete. British Standards Institution, London.13 BS 6089 Guide to assessment of concrete strength in existing structures. BritishStandards Institution, London.14 Bungey, J.H. Testing concrete in structures – a guide to equipment for testingconcrete in structures. TN 143, CIRIA, 1992, p. 87.15 Shickert, G. NDT in civil engineering in Germany. Insight, 36, No. 7, July 1994,489–495.16 Wiggenhauser, H. and Borchardt, K. The NDT-CE compendium on the Internet.Proc. NDT-CE 2003, DGZfP, Berlin, 2003, CD-ROM.

References 31317 Carino, N.J. Nondestructive testing of concrete: History and challenges. Spec.Publ. SP 144-30, American Concrete Institute, Detroit, 1994, pp. 623–678.18 Bungey, J.H. Recent developments in Civil Engineering NDT in the UK.INSIGHT – Journal of British Institute of NDT, 45, No. 12, Dec. 2003,796–799.19 Concrete Bridge Development Group. Guide to testing and monitoring the durabilityof concrete structures. Tech Guide 2, Concrete Society, Crowthorne, 2002.20 Breysse, D. and Abraham, O. Nondestructive assessment of damaged reinforcedconcrete structures: From needs to answers. A national state of the art. Proc.NDT-CE 2003, DGZfP, Berlin, 2003, CD-ROM.21 Uomoto, T. Utilisation of NDI to inspect internal defects in reinforced concretestructures. Proc. NDT-CE 2003, DGZfP, Berlin, 2003, CD-ROM.22 ACI Committee 228. Nondestructive test methods for evaluation of concrete instructures. ACI 228, 2R-98. American Concrete Institute, Detroit, 1998.23 ACI Committee 228. In-place methods to estimate concrete strength. ACI 228,1R-03. American Concrete Institute, Detroit, 2003.24 Electrochemical tests for reinforcement corrosion assessment. Tech. Rep. 60,Concrete Society/Institute of Corrosion, Crowthorne, 2004.25 Guidance on radar testing of concrete. Tech. Rept. 48, Concrete Society,Crowthorne, 1997.26 Highways Agency. Advice notes on the non-destructive testing of highway structures.BA 86/04, Vol. 3, Section 1, Part 7, 2004.27 DeBreto, J., Branco, F.A. and Ibanez, M. A knowledge based concrete bridgeinspection system. Concrete International, 16, No. 2, Feb. 1994, 59–63.28 Smith, J.R. Assessment of concrete buildings. Concrete International, 13, No. 12,Dec. 1991, 48–49.29 Pollock, D.J., Kay, E.A. and Fookes, P.G. Crack mapping for investigation ofMiddle East concrete. Concrete, 15, No. 5, May 1981, 12–18.30 Non-structural cracks in concrete. Tech. Rept. 22, Concrete Society,Crowthorne, 1992.31 Higgins, D.D. Diagnosing the causes of defects or deterioration in concretestructures. Current Practice Sheet 69, Concrete, 15, No. 10, Oct. 1981, 33–34.32 Fewtrell, A. Special access techniques for inspection works. Proc. Struct. Faultsand Repairs 93, Eng. Technics Press, Edinburgh, 2, 1993, 19–24.33 Palmer, D. The diagnosis of alkali–silica reaction. Report of Working Party,British Cement Association, Slough, 1988.34 Bungey, J.H. and Madandoust, R. Evaluation of non-destructive strength testingof lightweight concrete. Proc. ICE, Structs & Bldgs, 104, Aug. 1994, 275–283.35 Frigerio, T., Mariscotti, M.A.J., Ruffolo, M. and Thieberger, P. Developmentand application of computed tomography in the inspection of reinforced concrete.INSIGHT – Jnl. Brit. Inst. NDT, 46, No. 12, Dec. 2004, 742–745.36 Concrete core testing for strength. Tech. Rept. 11, Concrete Society, London,1987.37 Bungey, J.H. Concrete strength variations and in-place testing. Proc. 2nd AustralianConf. on Engineering Materials, University of New South Wales, Sydney,1981, pp. 85–96.38 Bungey, J.H. and Madandoust, R. Strength variations in lightweight concretebeams. Cement and Concrete Composites, 16, 1994, 49–55.

314 References39 Maynard, D.P. and Davis, S.G. The strength of in-situ concrete. Structural Engineer,52, No. 10, Oct. 1974, 369–374.40 Davies, S.G. Further investigations into the strength of concrete in structures.Tech. Rept. 42.514, Cement and Concrete Association, Slough, 1976.41 Bartlett, F.M. and MacGregor, J.G. Variation of in-place concrete strength instructures. ACI Materials Journal, 93, No. 2, March–April 1999, 261–269.42 Miao, B., Aitcin, P-.C., Cook, W.D. and Mitchell, D. Influence of concretestrength on in-situ properties of large columns. ACI Materials Journal, 90, No. 3,May/June 1993, 214–219.43 Bartlett, F.M. and MacGregor, J.G. Statistical analysis of the compressivestrength of concrete in structures. ACI Materials Journal, 96, No. 2, March–April, 1996, 158–168.44 Watkins, R.A.M., Pang, H.W. and McNicholl, D.P. A comparison between cubestrengths and insitu strength development. Proc. ICE Structures and Buildings,116, May 1996, 138–153.45 Murray, A.McC. and Long, A.E. A study of the in-situ variability of concreteusing the pull-off method, Proc. ICE, Part 2, 83, Dec. 1987, 731–745.46 Tomsett, H.N. The development of a nondestructive test analysis. Proc. NondestructiveTesting in Civil Engineering, Brit Inst NDT, 2, 1993, 435–450.47 Tomsett, H.N. Ultrasonic pulse velocity measurements in the assessment ofconcrete quality. Magazine of Concrete Research, 32, No. 110, March 1980,7–16.48 Leshchinsky, A.M., Leshchinsky, M.Yu. and Goncharova, A.S. Within-test variabilityof some non-destructive methods for concrete strength determination.Magazine of Concrete Research, 42, No. 153, Dec. 1990, 245–248.49 Stone, W.C., Carino, N.J. and Reeve, C.P. Statistical methods for in-placestrength predictions by the pull-out test. ACI Journal, 83, No. 5, Sept./Oct.1986, 745–756.50 Carino, N.J. Statistical methods to evaluate in-place test results. Proc. Int. Workshopon testing during Concrete Construction (RILEM), ed. H.W. Reinhardt,Chapman & Hall, 1992, pp. 432–448.51 Leshchinsky, A.M. Non-destructive testing of concrete strength: Statistical control.Materials and Structures, 25, No. 146, March 1992, 70–78.52 Bartlett, F.M. and MacGregor, J.G. Equivalent specified concrete strength fromcore test data. Concrete International, 17, No. 3, March 1995, 52–58.53 Reynolds, W.N. Measuring concrete quality non-destructively. British Journalof NDT, 26, No. 1, Jan. 1984, 11–14.54 Samarin, A. Combined methods. Handbook on Non-destructive Testing of Concrete,ed. V.M. Malhotra, and N.J. Carino, 2nd Edn, CRC Press, Ch. 9, 2004.55 RILEM. Draft recommendation for in-situ concrete strength determination bycombined non-destructive methods. Materials and Structures, 26, 1993, 43–49.56 Miretti, R., Carrasco, M.F., Grether, R.O. and Passerino, C.R. Combinednon-destructive methods applied to normal weight and lightweight concrete.INSIGHT – Jnl. Brit. Inst. NDT, 46, No. 12, Dec. 2004, 748–753.57 Hola, J. and Schabowicz, K. New technique of non-destructive assessment ofconcrete strength using artificial intelligence. NDT & E International, 38, No. 4,June 2005, 251–259.

References 31558 BS EN 12504-2 Testing concrete in structures: Non-destructive testing: Determinationof rebound number. British Standards Institution, London.59 ASTM C805 Rebound number of hardened concrete. American Society forTesting and Materials, Philadelphia.60 Akashi, T. and Amasaki, S. Study of the stress waves in the plunger of a reboundhammer at the time of impact. Spec. Publ. SP 82-2, American Concrete Institute,Detroit, 1984, pp. 17–34.61 Kolek, J. Non-destructive testing of concrete by hardness methods. In NondestructiveTesting of Concrete and Timber. Institution of Civil Engineers,London, 1970, pp. 19–22.62 Willetts, C.H. Investigation of the Schmidt concrete test hammer. Misc. Papers,S-627, US Army Engineer Waterways Experiment Station, Vicksburg, Miss.,June 1958.63 Malhotra, V.M. Testing hardened concrete: Non-destructive methods. Monograph9, American Concrete Institute. Detroit, 1976.64 Chaplin, R.G. Abrasion resistant concrete floors. In Advances in Concrete SlabTechnology, Pergamon, Oxford, 1980, pp. 532–543.65 Nogueira, C.L. and Willam, K.J. Ultrasonic testing of damage in concreteunder uniaxial compression. ACI Materials Journal, 98, No. 3, May–June 2001,265–275.66 Pavlakovic, B., Lowe, M.J.S. and Cawley, P. The inspection of tendons in posttensioned concrete using guided ultrasonic waves. INSIGHT - Journal of theBritish Institute of Non-Destructive Testing, 41, No. 7, July 1999, 446–452.67 Krause, M., Mielentz, F., Milmann, B., Streicher, D. and Muller, W. Ultrasonicimaging of concrete elements: state of the art using 2D synthetic aperture. Proc.NDT-CE 03, DGZfP, Berlin, 2003, CD-ROM.68 Koehler, B., Hentges, G. and Mueller, W. Improvement of ultrasonic testingof concrete by combining signal conditioning methods, scanning laser vibrometerand space averaging techniques. NDT&E International, 31, No. 4, 1998,281–287.69 Krause, M., Wiggenhauser, H., Krieger, J. and Schickert, M. NDE of a posttensioned concrete bridge girder using ultrasonic echo and impact echo. Proceedingsof Struct. Faults and Repair 2003. Eng. Technics Press, Edinburgh,2003, CD-ROM.70 Popovics, J. Ultrasonic testing of concrete structures. Materials Evaluation, 63,No. 1, 2005, 50–55.71 Andrews, D.R. Future prospects for ultrasonic inspection of concrete. Proc. ICEStructs & Bldgs, 99, Feb. 1993, 71–73.72 Bungey, J.H. Ultrasonic testing to identify alkali–silica reaction in concrete. Brit.J. NDT, 33, No. 5, 1991, 227–231.73 Hillger, W. Imaging of defects in concrete by ultrasonic pulse-echo technique.Proc. Struct. Faults and Repairs 93, Eng. Technics Press, Edinburgh, 3, 1993,59–65.74 Kroggel, O. Ultrasonic examination of crack structures in concrete slabs. Proc.Struct. Faults and Repairs 93, Eng. Technics Press, Edinburgh, 3, 1993, 67–70.75 Sack, D.A. and Olson, L.D. Advanced NDT methods for evaluating concretebridges and other structures. Proc. Struct. Faults and Repairs 93, Eng. TechnicsPress, Edinburgh, 1, 1993, 65–76.

316 References76 BS EN 12504-4. Testing Concrete – determination of ultrasonic pulse velocity.British Standards Institution, London.77 ASTM C597 Standard test method for pulse velocity through concrete. AmericanSociety for Testing and Materials. Philadelphia.78 Instruction manual for V-meter. James Electronics Inc., Chicago, Illinois,www.ndtjames.com.79 PUNDIT, CNS-Farnell, Borehamwood, UK, www.cnsfarnell.co.uk.80 Ikpong, A.A. The relationship between the strength and non-destructive parametersof rice husk ash concrete. Cement and Concrete Research, 23, 1993,387–398.81 Yiching Lin, Chao-Peng Lai and Tsong Yen. Prediction of ultrasonic pulse velocity(UPV) in concrete. ACI Materials Journal, 100, No. 1, Jan.–Feb. 2003,21–28.82 Jones, R. and Facaoaru, I. Recommendations for testing concrete by the ultrasonicpulse method. Materials and Structures, 2, No. 10, 1969, 275–284.83 Bungey, J.H. The validity of ultrasonic pulse velocity testing of in-place concretefor strength. NDT International, IPC Press, Dec. 1980, pp. 296–300.84 Swamy, R.N. and Al-Hamed, A.H. The use of pulse velocity measurements toestimate strength of air dried cubes and hence in-situ strength of concrete. Spec.Publ. SP 82-13, American Concrete Institute, Detroit, 1984, pp. 247–276.85 Testing of concrete by the ultrasonic pulse method. NDT 1, RILEM, Paris,Dec. 1972.86 Bungey, J.H. The influence of reinforcement on pulse velocity testing. Spec. Publ.SP 82-12, American Concrete Institute, Detroit, 1984, pp. 229–246.87 Hirata, T. and Uomoto, T. Detection of ultrasonic pulse echo through steel barin concrete crack depth measurement. Proc. NDT-CE 2000, ed. T. Uomoto,Elsevier, 2000, pp. 383–390.88 Hashimoto, S., Hashimoto, C., Watanabe, T. and Mizuguchi, H. Study onevaluation of settlement cracks occurring on the concrete surface above steelbars of reinforced concrete slab by NDT. Proceedings of Structural Faults andRepair 2003. Eng. Technics Press, Edinburgh, 2003, CD-ROM.89 Chhabra, Y., Jamaji, R. and Lim, B. Structural investigation and retrofitting ofheritage shop-houses at hotel Rendezvous, Singapore with the Tyfo® Fibrwrap®composite system. Proceedings of Concrete Solutions 2003, 1st InternationalConference on Concrete Repair. GR Technologie Ltd, Barnet, 2003, CD-ROM.90 Benedetti, A. On the ultrasonic pulse propagation into fire damaged concrete.ACI Structural Journal, 95, May–June 1998, 259–271.91 Bungey, J.H. Ultrasonic pulse testing of high alumina cement concrete on site.Concrete, 8, No. 9, Sept. 1974, 39–41.92 Swamy, R.N. and Laiw, J.C. Evaluation of concrete deterioration due to alkali–silica reactivity through nondestructive tests. Proc. Nondestructive Testing inCivil Engineering, Brit Inst NDT, 1, 1993, 315–328.93 BS 1881: Part 207 Recommendations for the assessment of concrete strength bynear-to-surface tests. British Standards Institution, London.94 Windsor Probe Test System. James Instruments Inc., Chicago, Illinois,www.ndtjames.com.95 ASTM C803 Penetration resistance of hardened concrete. American Society forTesting and Materials, Philadelphia.

References 31796 Nasser, K.W. and Al-Manaseer, A.A. New non-destructive test. Concrete International,9, No. 1, Jan. 1987, 41–44.97 BS 4078: Part 1 Powder actuated fixing systems – Code of practice for safeuse. British Standards Institution, London.98 Bungey, J.H. Testing by penetration resistance. Concrete, 15, No. 1, Jan. 1981,30–32.99 Nasser, K.W. and Al-Manaseer, A.A. Comparison of non-destructive testersof hardened concrete. Materials Journal, American Concrete Institute, Detroit,Sept./Oct. 1987, 374–380.100 Swamy, R.N. and Al-Hamed, A.H. Evaluation of the Windsor probe test toassess in-situ concrete strength. Proc. ICE, Pt1,77, June 1984, 167–194.101 Al-Manaseer, A.A. and Aquino, E.B. Windsor Probe test for non-destructiveevaluation of normal and high-strength concrete. ACI Materials Journal, 96,No. 4, July/Aug. 1999, 440–447.102 Shoya, M., Tgsukinaga, Y., Sugita, S., Matsui, M. and Hara, T. Estimatingthe compressive strength of concrete from pin-penetration test method. Proc.Nondestructive Testing in Civil Engineering, Brit Inst NDT, 1, 1993, 379–390.103 Carino, N.J. Pull-out test. Handbook on Non-destructive Testing of Concrete.Chap. 3, 2nd Edn, CRC Press, Boston, USA, 2004.104 British Cement Association. Early age strength assessment of concrete on site.Best Practice Guide, Crowthorne, 2000.105 Germann Instruments. Lok-test method for determination of the compressivestrength of concrete in place. Emdrupvej 102, DK 2400, Copenhagen,www.germann.org.106 BS EN 12504-3. Testing concrete in structures – determination of pull-outforce. British Standards Institution, London.107 Petersen, C.G. LOK-test and CAPO-test pullout testing – twenty years experience.Proc. NDT-CE 97, 1, ed. J.H. Bungey. Brit. Inst. NDT, 1997, 77–96.108 Ottosen, N.S. Nonlinear finite element analysis of pull-out test. Journal ofStruct. Division ASCE, 107, No. ST4, April 1981, 591–603.109 Yener, M. Overview and progressive finite element analysis of pullout tests,ACI Structural Journal, 91, No. 1, Jan./Feb. 1994, 49–58.110 Stone, W.C. and Carino, N.J. Deformation and failure in large scale pull-outtests. ACI Journal, Nov./Dec. 1983, 501–513.111 Stone, W.C. and Giza, B.J. The effect of geometry and aggregate on the reliabilityof the pull-out test. Concrete International, 7, No. 2, Feb. 1985, 27–36.112 Bickley, J.A. The evaluation and acceptance of concrete quality by in-placetesting. Spec. Publ. SP 82-6, American Concrete Institute, Detroit, 1984,pp. 95–109.113 Soutsos, M.N., Bungey, J.H., Long, A.E. and Henderson, G.D. Insitu strengthassessment of concrete – The European Concrete Frame Building Project.INSIGHT – Jnl. Brit. Inst. NDT, 42, No. 7, July 2000, 432–435.114 ASTM C900 Pull-out strength of hardened concrete. American Society forTesting and Materials, Philadelphia.115 Chabowski, A.J. and Bryden-Smith, D.W. A simple pull-out test to assessthe strength of in-situ concrete. Current Paper CP 25/77, Building ResearchEstablishment, Garston, June 1977.

318 References116 Chabowski, A.J. and Bryden-Smith, D.W. Assessing the strength of in-situPortland cement concrete by internal fracture tests. Magazine of ConcreteResearch, 32, No. 11, Sept. 1980, 164–172.117 Bungey, J.H. Concrete strength determination by pull-out tests on wedgeanchorbolts. Proc. ICE, Pt. 2, 71, June 1981, 379–394.118 Domone, P.L. and Castro, P.F. An expanding sleeve test for in-situ concrete andmortar strength evaluation. Proc. Struct. Faults and Repairs 87, EngineeringTechnics Press, Edinburgh, 1987, pp. 149–156.119 Keiller, A.K. Assessing the strength of in-situ concrete. Concrete International,7, No. 2, Feb. 1985, 15–21.120 Long, A.E. and Murray, A.McC. The pull-off partially destructive test forconcrete. Spec. Publ. SP 82-17, American Concrete Institute, Detroit, 1984,pp. 327–350.121 Jaegermann, C. A simple pull-out test for in-situ determination of early strengthof concrete. Magazine of Concrete Research, 41, No. 149, Dec. 1989, 235–242.122 Cleland, D.J. In-situ methods for assessing the quality of concrete repairs. Proc.ICE, Structs & Bldgs, 99, Feb. 1993, 68–70.123 McLeish, A. Standard tests for repair materials and coatings-Part 1. Pull-OffTesting. CIRIA, TN 139, 1992, p. 41.124 ASTM C1583. Tensile strength of concrete surfaces and bond strength ortensile strength of concrete repair and overlay materials by pull-off method.American Society for Testing and Materials, Philadelphia.125 BS EN 1542. Products and systems for the repair of concrete structures –Test methods – Measurements of bond strength by pull-off. British StandardsInstitution, London.126 Bungey, J.H. and Madandoust, R. Factors influencing pull-off tests on concrete.Magazine of Concrete Research, 44, No. 158, March 1992, 21–30.127 Long, A.E., Montgomery, F.R. and Cleland, D. Assessment of concrete strengthand durability on site. Proc. Struct. Faults and Repair 87, Engineering TechnicsPress, Edinburgh, 1987, pp. 61–73.128 Naderi, M. Friction-transfer test for the assessment of insitu strength andadhesion of cementitious materials. Construction and Building Materials, 19,No. 6, July 2005, 454–459.129 Christiansen, V.T., Thorpe, J.D. and Yener, M. In-situ testing of concrete usingbreak-off method. Proc. Struct. Faults and Repair 87, Engineering TechnicsPress, Edinburgh, 1987, pp. 101–112.130 Johansen, R. In-situ strength evaluation of concrete – the break-off method.Concrete International, 1, No. 9, Sept. 1979, 45–51.131 Dahl-Jorgensen, E. and Johansen, R. General and specialized use of the breakoffconcrete strength testing method. Spec. Publ. SP 82-15, American ConcreteInstitute, Detroit, 1984, pp. 293–308.132 Carlsson, M., Eeg. I.R. and Jahren, P. Field experience in the use of the breakofftester. Spec. Publ. SP 82-14, American Concrete Institute, Detroit, 1984,pp. 277–292.133 Naik, T. The break-off test method. Handbook on Non-Destructive Testing ofConcrete. ed. V.M. Malhotra and N.J. Carino, 2nd Edn, CRC Press, Chap. 4,2004.

References 319134 Stoll, U.W. Compressive strength measurement with the Stoll tork test. ConcreteInternational, 7, No. 12, Dec. 1985, 42–47.135 BS EN 12504-1. Testing concrete in structures – Cored specimens – Taking,examining and testing in compression. British Standards Institution, London.136 ASTM C42 Standard method of obtaining and testing drilled cores and sawnbeams of concrete. American Society for Testing and Materials, Philadelphia.137 ACI 318 Building code requirements for reinforced concrete. American ConcreteInstitute, Detroit.138 Insitu Concrete Strength. Project Report 3, Concrete Society, Crowthorne,2004.139 Neville, A. Core tests – Easy to perform, not easy to interpret. Concrete International,23, No. 11, November 2001, 59–68.140 Monday, J.G.L. and Dhir, R.K. Assessment of in-situ concrete quality by coretesting. Spec. Publ. SP 82-20, American Concrete Institute, Detroit, 1984,393–410.141 Robins, P.J. The point-load test for tensile strength estimation of plain andfibrous concrete. Spec. Publ. SP 82-16, American Concrete Institute, Detroit,1984, 309–325.142 Robins, P.J. and Austin, S.A. Core point-load test for steel fibre reinforcedconcrete. Magazine of Concrete Research, 37, No. 133, Dec. 1985, 218–242.143 Clayton, N. Fluid pressure testing of concrete cylinders. Magazine of ConcreteResearch, 30, No. 102, March 1978, 26–30.144 Chrisp, T.M., Walron, P. and Wood, J.G.M. Development of a non-destructivetest to quantify damage in deteriorated concrete. Magazine of ConcreteResearch, 45, No. 165, Dec. 1993, 247–253.145 Bartlett, F.M. and MacGregor, J.G. Effect of moisture condition on concretecore strengths. ACI Materials Journal, 91, No. 3, May/June 1994, 227–236.146 Bartlett, F.M. and MacGregor, J.G. Effect of core length to diameter ratio onconcrete core strengths. ACI Materials Journal, 91, No. 4, July/Aug. 1994,339–348.147 Lesinskij, M. and Lichtmann, M. Special features of concrete strength testing.Concrete Plant International, 1, No. 1, Feb. 2002, 82–94.148 Jones, A.E.K., Clark, L.A. and Amasaki, S. The suitability of cores in predictingthe behaviour of structural members suffering from ASR. Magazine of ConcreteResearch, 46, No. 167, June 1994, 145–150.149 Yip, W.K. Estimating the potential strength of concrete with prior load history.Magazine of Concrete Research, 45, No. 165, Dec. 1993, 301–308.150 Ahmed, A.E. Does core size affect strength testing? Concrete International, 21,No. 8, Aug. 1999, 35–39.151 Bungey, J.H. Determining concrete strength by using small diameter cores.Magazine of Concrete Research, 31, No. 107, June 1979, 91–98.152 Bowman, S.A.W. Discussion on ref. 151, Magazine of Concrete Research, 32,No. 111, June 1980, 124.153 Swamy, R.N. and Al-Hamed, A.H. Evaluation of small diameter core tests todetermine in-situ strength of concrete. Spec. Publ. SP 82-21, American ConcreteInstitute, Detroit. 1984, pp. 411–440.154 Schupack, M. Durability study of a 35 year old post-tensioned bridge. ConcreteInternational, 16, No. 2, Feb. 1994. 54–58.

320 References155 Azizinamini, A., Shekar, Y., Barnhill, G. and Boothby, T.E. Old concrete slabbridges: Can they carry modern traffic loads? Concrete International, 16, No. 2,Feb. 1994, 64–69.156 Ladner, M. In-situ load testing of concrete bridges in Switzerland. Spec. Publ.SP 88-4, American Concrete Institute, Detroit, 1985, pp. 59–80.157 Ladner, M. Unusual methods for deflection measurements. Spec. Publ. SP 88-8,American Concrete Institute, Detroit, 1985, pp. 165–180.158 Moss, R.M. and Currie, J.R. Static load testing of building structures. InformationPaper, IP9/89, Building Research Establishment, Garston, 1989.159 Jones, D.S. and Oliver, C.W. The practical aspects of load testing. StructuralEngineer, 56A, No. 12, Dec. 1978, 353–356.160 Garas, F.K., Clarke, J.L. and Armer, G.S.T. Structural Assessment. Butterworths,London, 1987.161 Lindsell, P. and Cole, G. Bridge load testing in Surrey. Structural Engineer, 76,No. 5, 3 March 1998, 86–90.162 Menzies, J.R., Sandberg. A.C.E. and Somerville, G. Instrumentation of structures.Structural Engineer, 62A, No. 3, March 1984, 83–86.163 Report of working party on high alumina cement, Institution of StructuralEngineers. Structural Engineer, 54, No. 9, Sept. 1976, 352–361.164 Moss, R.M. Load testing of beam and block concrete floors. Proc. ICE Structs& Bldgs, 99, May 1993, 211–223.165 BS 8110: Part 2 Structural use of concrete – code of practice for special circumstances.British Standards Institution, London.166 Lee, C.R. Load testing of concrete structures, with particular reference toCP11O and experience in the HAC Investigations. Building Research Establishment,Garston, Doc. B507/77, 1977.167 Guedelhoefer, O.C. Instrumentation and techniques of full scale testing of structures.Experimental Methods in Concrete Structures for Practitioners, AmericanConcrete Institute, Detroit, 1979.168 Miao, B., Chaallal, O., Perraton, D. and Aitcin, P.-C. On-site early-age monitoringof high-performance concrete columns. ACI Materials Journal, 90, No. 5,Sept./Oct. 1993, 415–420.169 Matthews, S.L. Deployment of instrumentation for in-service monitoring.Structural Engineer, 78, No. 13, 4 July 2000, 28–32.170 McCarter, W.J. and Vennesland, O. Sensor systems for use in reinforced concretestructures. Construction and Building Materials, 18, 2004, 351–358.171 Hansson, C., Seabrook, P.T. and Marcotte, J.D. In-place corrosion monitoring.Concrete International, 26, No. 7, July 2004, 59–65.172 McCarter, W.J., Chrisp, T.M., Butler, A. and Basheer, P.A.M. Near-surfacesensors for condition monitoring of cover zone concrete. Construction andBuilding Materials, 15, 2001, 115–124.173 Malan, F.S., Ahmet, K., Dunster, A. and Hearing, R. Development of amicrowave frequency sensor for the long-term localized moisture monitoringof concrete. Magazine of Concrete Research, 56, No. 5, 263–271.174 Broomfield, J., Davies, K., Hladky, K. and Noyce, P. Monitoring of reinforcementcorrosion in concrete structures in the field. Proceedings of ConcreteSolutions 2003, 1st International Conference on Concrete Repair. GR TechnologieLtd, Barnet, 2003, CD-ROM.

References 321175 Basheer, P.A.M., Grattan, K.T.V., Sun, T., Long, A.E., McPolin, D. andXie, W. Fibre optic chemical sensor systems for monitoring pH changes inconcrete. Advanced environmental, chemical and biological sensing technologiesII, Proc. SPIE, 5586, Bellingham, WA, 2004, 144–152.176 Ing, M., Austin, S.A. and Lyons, R. Cover zone properties influencing acousticemission due to corrosion. Cement and Concrete Research, 35, 2005, 284–295.177 Moss, R.M. and Matthews, S.L. In-service structural monitoring – a state ofthe art review. The Structural Engineer, 73, No. 2, 1995, 23–31.178 Nassif, H.H., Gindy, M. and Davis, J. Comparison on laser Doppler vibrometerwith contact sensors for monitoring bridge deflections and vibration. NDT &E International, 38, No. 3, April 2005, 213–218.179 Olaszek, P. The photogrammetric method for dynamic bridge investigations,Proc. Struct. Faults and Repairs 93, Eng. Technics Press, Edinburgh, 1, 1993,89–93.180 Hegger, J., Gortz, S. and Hausler, F. Analysis of the crack characteristics undershear load of reinforced concrete structures by using photogrammetry. Proc.NDT-CE 2003, DGZfP, Berlin, 2003, CD-ROM.181 Dill, M.J. and Curtis, I.L. Monitoring concrete structures using optical fibresensors. Concrete, 27, No. 5, Sept./Oct. 1993, 31–35.182 Lau, K.T. Fibre-optic sensors and smart composites for concrete applications.Magazine of Concrete Research, 55, No. 1, 2003, 19–34.183 BS 1881: Part 206 Recommendations for determination of strain in concrete.British Standards Institution, London.184 deWit, M., Horhanessian, A. and Diouron, T. Le. The strain relief method formeasuring the stress in concrete. Proc. Struct. Faults and Repair 2003, Eng.Technics Press, Edinburgh, 2003, CD-ROM.185 Salah el Din, A.S. and Lovegrove, J.M. A gauge for measuring long-term cyclicstrain on concrete surfaces. Magazine of Concrete Research, 33, No. 115,June 1981, 123–127.186 Scott, R.H. and Gill, P.A.T. Short term distributions of strain and bond stressalong tension reinforcement. Structural Engineer, 65B, No. 2, June 1987,39–43/48.187 Hegger, J., Gortz, S. and Niewels, J. The use of laser-interferometry (ESPI) inanalysis of reinforced concrete structures. Proc. NDT-CE 03, DGZfP, Berlin,2003, CD-ROM.188 Corless, R.C. and Morice, P.B. The kinematics of structural testing. StructuralEngineer, 64B, No. 4, Dec. 1986, 100–102.189 Raupach, M. Corrosion of steel in the area of cracks in concrete – Laboratorytests and calculations using a transmission line model. Corrosion ofsteel reinforcement in concrete construction, Proceedings of an InternationalConference, Cambridge, SCI, 1996, 13–23.190 Roberge, P. Handbook of corrosion engineering. McGraw Hill, 2000.191 Glass, G.K. and Buenfeld, N.R. The inhibitive effects of electrochemicaltreatment applied to steel in concrete. Corrosion Science, 42, No. 6, 2000,1923–1927.192 BS 1881: Part 204 Recommendations on the use of electromagnetic covermeters.British Standards Institution, London.

322 References193 Alldred, J.C. An improved method for measuring reinforcing bars of unknowndiameter in concrete using a cover meter. Proc. Non-Destructive Testing inCivil Engineering, Brit. Inst. NDT, 2, 1993, 767–788.194 Scheel, H. Magnetic detection of prestressing steel fractures in prestressed concrete.Proc. Structural Faults and Repair – 99, Eng. Technics Press, Edinburgh,1999, CD-ROM.195 Zaid, M., Gaydecki, P., Quek, S., Miller, G. and Fernandez, B. Extractingdimensional information from steel reinforcing bars embedded in concreteusing neural networks trained on data from an inductive sensor. NDT & EInternational, 37, No. 7, Oct. 2004, 551–558.196 Uomoto, T., Ohtsu, M., Ikenaga, H., Tanano, H., Hamasaki, H., Kishi, K. andYoshimura, A. Identification of reinforcement in concrete by electro-magneticmethods. Proc. NDT-CE 03, DGZfP, Berlin, 2003, CD-ROM.197 Snell, L.M., Wallace, N. and Rutledge, R.B. Locating reinforcement in concrete.Concrete International, 8, No. 4, April 1986, 19–23.198 Alldred, J.C. Quantifying the losses in covermeter accuracy due to congestionof reinforcement. Proc. Struct. Faults and Repairs 93, Eng. Technics Press,Edinburgh, 2, 1993, 125–130.199 ASTM C876 Half-cell potentials of uncoated reinforcing steel in concrete.American Society for Testing and Materials, Philadelphia.200 Figg, J.W. and Marsden, A.F. Development of inspection techniques – a stateof the art survey of electrical potential and resistivity measurements for useabove water level. Offshore Technology Report OTH 84 205, HMSO, London,1983.201 Vassie, F.K. The Half-cell Potential Method of Locating Corroding Reinforcementin Concrete Structures. TRRL Application Guide 9, 1991, 30pp.202 Langford, P. and Broomfield, J. Monitoring the corrosion of reinforcing steel.Construction Repair, 1, No. 2, Palladian Pubs., May 1987, 32–36.203 Stratfull, R.F. Half cell potentials and the corrosion of steel in concrete. HighwayResearch Record, 1973, 433, 12–21.204 Raherinaivo, A. Contrôle de la corrosion des armatures dans les structures enbéton armé. Bulletin De liason des Laboratoires ales Ponts et Chaussés, 158,Nov./Dec. 1988, 29–38 (in French).205 Bjegovic, D., Mikulic, D. and Sekulic, D. Non-destructive corrosion rate monitoringfor reinforced concrete structures, 15th World Conference on Non-Destructive Testing, Italian Society for NDT and Monitoring Diagnostics,Rome, 15–21 October, 2000. www.ndt.net/article/wcndt00/index.htm.206 Baker, A.F. Potential mapping techniques. Proc. Seminar on Corrosion in Concrete.London Press Centre, Global Corrosion Consultants, Telford, May 1986,3.1–3.21.207 Andrade, C., Martinez, I., Ramirez, M. and Garcia, M. Non-destructive techniquesfor on-site measurements of reinforcement corrosion in concrete structures.Proc. NDT-CE 03, DGZfP, Berlin, 2003, CD-ROM.208 Grantham, M. and Ross, R. Repairing concrete car parks, a case study fromthe client’s perspective. Proceedings of Concrete Solutions, 1st InternationalConference on Concrete Repair, St Malo, France. GR Technologie Ltd, Barnet,Hertfordshire, 2003, CD-ROM.

References 323209 Andrade, C. and Martinez, I. Cathodic protection efficiency measurementsmade on concrete structures using the new technique called passivity verification(PVT) without disconnecting the protection current, Concrete Solutions2006, 2nd International Conference on Concrete Repair, St Malo, GR TechnologieLtd, 2006, CD-ROM to be published.210 Ewins, A.J. Resistivity measurements in concrete. Brit. Jnl NDT, 32, No. 3,March 1990, 120–126.211 Millard, S.G., Harrison, J.A. and Gowers, K.R. Practical Measurement of concreteresistivity, INSIGHT-Brit. Jnl NDT, 33, No. 2, Feb. 1991, 59–63.212 McCarter, W.J., Forde, M.C. and Whittington, H.W. Resistivity characteristicsof concrete. Proc. ICE, Pt. 2, 71, March 1981, 107–117.213 Gowers, K.R. and Millard, S.G. Measurement of concrete resistivity for assessmentof corrosion severity of steel, ACI Materials Journal, 96, No. 5, Sept.–Oct.1999, 536–541.214 Naish, C.C., Harker, A. and Carney, R.F.A. Concrete inspection: Interpretationof potential and resistivity measurements. Proc. Corrosion of Reinforcementin Concrete, Elsevier Applied Science, 1990, pp. 314–332.215 Moore, R.W. Earth resistivity tests applied as a rapid non-destructive procedurefor determining thickness of concrete pavements. Highway Research Record.Highway Research Board, Washington DC, No. 218, 1968, 49–55.216 John, G., Hladky, K. and Dawson, J. Recent developments in electrochemicalinspection techniques for corrosion damaged structures. Industrial Corrosion,11, No. 2, Feb./Mar. 1993, 12–13.217 Filiu, S., Gonzalez, J.A. and Andrade, C. Electrochemical methods for on-sitedetermination of corrosion rates of rebar. ASTM, STP 1276, 1995.218 Sehagal, A., Kho, Y.T., Osseo-Asare, K. and Pickering, H.W. Comparison ofcorrosion rate measuring devices for determining corrosion rate of steel-inconcretesystems. Corrosion, 48, No. 10, 1992, 871–880.219 Elsener, B., Müller, S., Suter, M. and Böhni, H. Corrosion monitoring of steel inconcrete: Theory and practice. Proc. Corrosion of Reinforcement in Concrete,Elsevier Applied Science, 1990, 348–357.220 Elsener, B., Wojtas, H. and Böhni, H. Inspection and monitoring of reinforcedconcrete structures – electrochemical techniques to detect corrosion.INSIGHT – Jnl. Brit. Inst. NDT, 36, No. 7, July 1994, 502–506.221 Dawson, J.L., John, D.G., Jafar, M.I., Hladky, K. and Sherwood, L. Electrochemicalmethods for inspection and monitoring of corrosion of reinforcingsteel in concrete. Proc. Corrosion of Reinforcement in Concrete, ElsevierApplied Science, 1990, 358–371.222 Broomfield, J.P., Rodriguez, J., Ortega, L.M. and Garcia, A.M. Field measurementof the corrosion rate of steel in concrete using a microprocessor controlledunit with a monitored guard ring for signal confinement. Proc. Techniques toAssess the Corrosion Activity of Steel-Reinforced Concrete Structures, ASTM,STP 1276, Philadelphia, 1995.223 Strategic Highway Research Programme, SHRP. Condition evaluation of concretebridges relative to reinforcement corrosion, Volume 2: Method for measuringthe corrosion rate of reinforcing steel, SHRP-S/FR-92-104, 1992.224 Tullmin, M.A.A., Hansson, C.M. and Roberge, P.R. Electrochemicaltechniques for measuring reinforcing steel corrosion, InterCorr/96, 1st

324 ReferencesGlobal International Corrosion Conference, 1996. www.corrosionsource.com/events/intercorr/index.html.225 Corrosion of reinforcement in concrete: Electrochemical monitoring, Digest434, Building Research Establishment, Garston, 1998.226 Grantham, M. and Schneck, U. Evaluation of corrosion inhibitors for remediationof St Mary’s multi-storey car park, Colchester, Proceedings of GlobalConstruction, Ultimate Concrete Opportunities, ‘Repair and Renovation ofConcrete Structures’, University of Dundee, Concrete Technology Unit, ThomasTelford, July 2005, 269–278.227 Raupach, M. Corrosion behaviour of the reinforcement under on-site conditions,15th International Corrosion Congress, Frontiers in Corrosion Scienceand Technology, International Corrosion Council, Granada, September 2002,Paper 096, 9 pages (www.icc-net.org/).228 Parrott, L.J. Moisture profiles in drying concrete. Advances in CementResearch, 1, No. 3, July 1988, 164–170.229 HUM-Meter, Germann Instruments, Copenhagen, www.germann.org.230 Marfisi, E., Burgoyne, C.J., Amin, M.G.H. and Hall, L.D. The use of MRI toobserve the structure of concrete, Magazine of Concrete Research, 57, No. 2,March 2005, 101–109.231 Bell, J.R., Leonards, G.A. and Dolch, W.L. Determination of moisture contentof hardened concrete and its dielectric properties. Proc. ASTM 63, 1963,996–1007.232 Jones, R. A review of the non-destructive testing of concrete. Proc. Symp. onNDT of Concrete and Timber, ICE, London, 1969, 1–7.233 Parrott, L.J. Moisture conditioning and transport properties of concrete testspecimens. Materials and Structures, 27, 1994, 460–468.234 Newman, A.J. Improvement of the drilling method for the determination ofmoisture content of building materials. BRE, CP 22/75, 1975.235 Browne, J.D.I. Non-destructive testing of concrete – a survey. Non-destructiveTesting, Feb. 1968, 159–164.236 Maser, K.R. Bridge deck condition survey using radar. Transport ResearchBoard, Paper 91-0530, 1991, 19pp.237 Figg, J.W. Determining the water content of concrete panels. Magazine ofConcrete Research, 24, No. 79, June 1972, 94–96.238 Permeability of concrete – a review of testing and experience. Tech. Rept. 31,Concrete Society, London, 1988.239 Basheer, P.A.M. A brief review of methods for measuring the permeationproperties of concrete in situ. Proc. ICE, Structs & Bldgs, 99, Feb. 1993,74–83.240 Kropp, J. and Hilsdorf, H.K. (eds). Performance criteria for concrete durability,RILEM Report 12, TC 116-PCD, Spon, 1995.241 Basheer, L., Kropp, J. and Cleland, D.J. Assessment of the durability of concretefrom its permeation properties: A review. Construction and Building Materials,15, No. 2/3, 2001, 93–104.242 BS 1881: Part 208 Methods for the determination of initial surface absorption.British Standards Institution, London.243 BS 1881: Part 122 Methods for determination of water absorption. BritishStandards Institution, London.

References 325244 Kreijger, P.C. The skin of concrete: Composition and properties. Materials andStructures, 17, No. 100, July/Aug. 1984, 275–283.245 Basheer, P.A.M. and Nolan, E. Near-surface moisture gradients and insitupermeation tests. Construction and Building Materials, 15, No. 2/3, 2001,105–114.246 Levitt, M. Non-destructive testing of concrete by the initial surface absorptionmethod. Proc. Symp. on NDT of Concrete and Timber, ICE, London, 1969,23–36.247 Wilson, M.A., Taylor, S.C. and Hoff, W.D. The initial surface absorption test(ISAT): An analytical approach. Magazine of Concrete Research, 50, No. 2,June 1998, 179–185.248 Dhir, R.K., Shaaban, I.H., Claisse, P.A. and Byars, E.A. Preconditioning in situconcrete for permeation testing – Part 1: Initial surface absorption. Magazineof Concrete Research, 45, No. 163, June 1993, 113–118.249 Price, W.F. and Bamforth, P.B. Initial surface absorption of concrete: Examinationof modified test apparatus for obtaining uniaxial absorption. Magazineof Concrete Research, 45, No. 162, March 1993, 17–24.250 Dhir, R.K., Hewlett, P.C., Byars, E.A. and Bai, J.P. Estimating the durabilityof concrete in structures. Concrete, 28, No. 6, Nov./Dec. 1994, 25–30.251 McCarter, W.J. and Chrisp, M. Monitoring water and ionic penetrationinto cover-zone conrete. ACI Materials Journal, 97, No. 6, Nov./Dec. 2000,668–674.252 Cather, R., Figg, J.W., Marsden, A.F. and O’Brien, T.P. Improvements to theFigg method for determining the air permeability of concrete. Magazine ofConcrete Research, 36, No. 129, Dec. 1984, 241–245.253 Dhir, R.K., Hewlett, P.C. and Chan, Y.N. Near-surface characteristics of concrete:Assessment development of in-situ test methods. Magazine of ConcreteResearch, 39, No. 141, Dec. 1987, 183–195.254 Meletiou, C.A., Tia, M. and Blooquist, D. Development of a field permeabilitytest apparatus and method for concrete. ACI Materials Journal, 89, No. 1,Jan./Feb. 1992, 83–89.255 Gas and Water Permeation Tests. Germann Instruments, Copenhagen,www.germann.org.256 Basheer, P.A.M., Long, A.E. and Montgomery, F.R. The ‘Autoclam permeabilitysystem’ for measuring in-situ permeation properties of concrete, Proc. Non-Destructive Testing in Civil Engineering, Brit. Inst. NDT, 1, 1993, 235–260.257 Gomes, A.H., Costa, J.O., Albertini, H. and Agular, J.E. Permeability of concrete:A study intended for the insitu valuation, using portable instruments andtraditional techniques. Proc. NDT-CE03, DGZfP, Berlin 2003, CD-ROM.258 Torrent, R. A two chamber vacuum cell for measuring the coefficient of permeabilityto air of the concrete cover on site. Materials and Structures, 25,No. 150, July 1992, 358–365.259 Proceq, S.A., Zurich, Switzerland, www.proceq.ch.260 Guth, D.L. and Zia, P. Evaluation of new air-permeability test device forconcrete. ACI Materials Journal, 98, No. 1, Jan./Feb. 2001, 44–51.261 Dias, W.P.S. Influence of drying on concrete sorptivity. Magazine of ConcreteResearch, 56, No. 9, Nov. 2004, 537–543.

326 References262 Guidelines for the appraisal of structural components in high alumina cementconcrete. Institution of Structural Engineers, London, 1975.263 ASTM C457 Air void content in hardened concrete. American Society forTesting and Materials, Philadelphia.264 Sadegzadeh, M., Page, C.L. and Kettle, R.J. Surface microstructure and abrasionresistance of concrete. Cement and Concrete Research, 17, 1987, 581–590.265 Kettle, R.J. and Sadegzadeh, M. Field Investigations of abrasion resistance.Materials and Structures, 20, 1987, 96–102.266 BS EN 13894-4. Methods of test for screed materials – Determination of wearresistance – BCA. British Standards Institution, London.267 BS 8204: Part 2 Screeds, Bases and Insitu Floorings. Abrasion resistance ofconcrete floor surfaces. British Standards Institution, London.268 Sadegzadeh, M. and Kettle, R.J. Indirect and non-destructive methods forassessing abrasion resistance of concrete. Magazine of Concrete Research, 38,No. 137, Dec. 1986, 183–190.269 Vassou, V. and Kettle, R.J. Near surface characteristics of fibre reinforcedconcrete floors: An investigation into non-destructive testing. INSIGHT – Jnl.Brit. Inst. NDT, 47, No. 7, July 2005, 400–407.270 Alexanderson, J. Resistance of floor screeds with floor coverings to the actionof rolling wheels. Concrete, 35, No. 1, Jan. 2001, 44–46.271 Manning, D.G. and Holt, F.B. Detecting delamination in concrete bridge decks.Concrete International, 2, No. 11, Nov. 1980, 34–41.272 Holt, F.B. and Eales, J.W. Non-destructive evaluation of pavements. ConcreteInternational, 9, No. 6, June 1987, 41–45.273 ASTM D4788 Detecting delaminations in bridge decks using infraredthermography.274 Titman, D.J. Applications of thermography in non-destructive testing of structures.NDT & E International, 34, No. 2, March 2001, 149–154.275 Maierhofer, C., Brink, A., Rollig, M. and Wiggenhauser, H. Quantitativeimpulse-thermography as non-destructive testing method in Civil Engineering –Experimental results and numerical simulations. Construction and BuildingMaterials, 19, No. 10, 2005, 731–737.276 Clark, M.R., McCann, D.M. and Forde, M.C. Application of infra-red thermographyto the non-destructive testing of concrete and masonry bridges. NDT& E International, 36, No. 4, June 2003, 265–275.277 Stanley, C. and Balendran, R.V. Non-destructive testing of the external surfacesof concrete buildings and structures in Hong Kong using infrared thermography.Concrete, 28, No. 3, May/June 1994, 35–37.278 Cantor, T.R. Review of penetrating radar as applied to the non-destructivetesting of concrete. Spec. Publ. SP 82-29, American Concrete Institute, Detroit,1984, pp. 581–602.279 Bungey, J.H. and Millard, S.G. Radar inspection of structures. Proc. ICE,Structs & Bldgs, 99, May 1993, 173–186.280 Olver, A.D. and Cuthbert, L.G. FMCW radar for hidden object detection,Proc. Instn Elect. Engnrs, 135, Part F, No. 4, Aug. 1988, 354–361.281 Cariou, J. Perspectives sur les applications des technique radar aux bétons.Bulletin de Liason des Laboratoires des Ponts et Chaussées, Paris, No. 163,Sept./Oct. 1989, 25–29.

References 327282 Bungey, J.H., Shaw, M.R., Millard, S.G. and Thomas, C. Radar testing ofstructural concrete. Proc. 5th Int. Conf. Ground Penetrating Radar, GPR ’94,1, 1994, 305–318.283 Hobbs, C.P., Temple, J.A.G., Hillier, M.J., Silk, M.G. and Tattersall, H.G.Radar inspection of civil engineering structures. Proc. Non-Destructive Testingin Civil Engineering, Brit. Inst. NDT, 1, 1993, 79–96.284 Carter, C.R., Chung, T., Holt, F.B. and Manning, D.G. An automated signalprocessing system for the signature analysis of radarwave forms from bridgedecks. Can. Elect. Eng., 11, No. 3, 1986, 128–137.285 ASTM D4748 Determining the thickness of bound pavement layers using shortpulse radar. American Society for Testing and Materials, Philadelphia.286 British Gas Plc. Ground probing radar. Publicity material, British Gas Plc,London.287 ERA Technology Ltd, Surface penetrating radar. Publicity material, Leatherhead,Surrey, www.era.co.uk.288 GSSI Inc. Subsurface interface radar. Publicity material, N. Salem, New Hampshire,www.geophysical.com.289 Sensors & Software Inc., Mississauga, Ontario, Canada, www.sensoft.on.ca.290 Utsi, V. Minimum delamination width for GPR, Proc. Structural Faults &Repair 03, Eng. Technics Press, Edinburgh, 2003, CD-ROM.291 Bungey, J.H. and Millard, S.G. Radar inspection of concrete structures. 3rd Int.Kerensky Conf., Global Trends in Structural Engineering, Singapore, 1994, 8pp.292 Cheshire, K. Police search for human remains in Gloucestershire using surfacepenetrating radar. INSIGHT – Jnl. Brit. Inst. NDT, 36, No. 5, May 1994.293 Soutsos, M.N., Bungey, J.H., Millard, S.G., Shaw, M.R. and Patterson, A.Dielectric properties of concrete and their influence on radar testing. NDT &E International, 34, No. 6, Sept. 2001, 419–425.294 Bungey, J.H., Millard, S.G. and Shaw, M.R. The influence of reinforcing steelon radar survey of concrete structures. Construction & Building Materials, 8,No. 2, 1994, 119–126.295 Shaw, M.R., Molyneaux, T.C.K., Millard, S.G., Taylor, M.J. and Bungey, J.H.Assessing bar size of steel reinforcement in concrete using ground penetratingradar and neural networks. INSIGHT – Brit. Jnl. of NDT, 45, No. 12, Dec.2003, 813–816.296 Taffe, A. and Maierhofer, C. Guidelines for NDT methods in Civil Engineering:Proc. NDT-CE 03, DGZfP, Berlin, 2003, CD-ROM.297 Millard, S.G., Shaw, M.R., Giannopoulos, A. and Soutsos, M. Modelling ofsubsurface radar for non-destructive testing of structures. Jnl. Materials in CivilEngineering, ASCE, 10, No. 3, Aug. 1998, 188–196.298 Giannopoulos, A. Modelling ground penetrating radar by Gpr Max. Constructionand Building Materials, 19, No. 10, 2005, 755–762.299 Millard, S.G., Shaari, A. and Bungey, J.H. Field pattern characteristics of GPRantennas. NDT & E International, 35, 2002, 473–482.300 Bungey, J.H. Sub-surface radar testing of concrete – a review. Constructionand Building Materials, 18, No. 1, 2004, 1–8.301 Gordon, M.O., Hardy, M.S.A. and Giannopoulos, A. Semi-intrusive determinationof ground penetrating radar wave velocity. NDT & E International, 34,No. 2, March 2001, 163–171.

328 References302 Klysz, G., Balayssac, J.P. and Laurens, S. Spectral analysis of radar surfacewaves for non-destructive evaluation of concrete cover. NDT & E International,37, No. 3, April 2004, 221–227.303 Khan, M.S. Detecting corrosion – induced delaminations. Concrete International,25, No. 7, July 2003, 73–78.304 ASTM D4580 Measuring delaminations in concrete bridge decks by sounding.American Society for Testing and Materials, Philadelphia.305 Henderson, M.E., Costley, R.D. and Dion, G.N. Acoustic inspection of concretebridge decks with the Hollow Deck, Structural Materials Technology IV: AnNDT Conference, ed. S. Adampalli, Atlantic City, NJ, 2000, 184–189.306 Clark, P. Concrete evidence: Sampling performance of non-destructive rotarypercussion sounding for delaminations in steel reinforced suspended concretestructures. Proc. Struct. Mats. Tech 4, ed. S. Alampalli and G. Washer, Buffalo,New York, Sept. 2004, 239–243.307 Carino, N.J. Stress wave propagation methods, Handbook on NondestructiveTesting of Concrete, ed. V.M. Malhotra and N.J. Carino, 2nd Edn, CRC Press,Chap. 14, 2004.308 Davis, A.G. The non-destructive impulse response test in North America:1985–2001. NDT & E International, 36, No. 4, June 2003, 185–193.309 Germann Instruments, S’MASH. Impulse Response System, Copenhagen,www.germann.org.310 Colla, C. and Lausch, R. Influence of source frequency on impact-echo dataquality for testing concrete structures. NDT & E International, 36, No. 4, June2003, 203–213.311 Sansalone, M.J. and Streett, W.B. Impact-Echo non-destructive evaluation ofconcrete and masonry, Bulbrier Press, Jersey Shore, PA, 1997.312 Byles, R. Dutch strike a blow for integrity testing. New Civil Engineer, 11 June1981, 26.313 Hollema, D.A. and Olson, L.D. Crosshole sonic logging and velocity tomographyimaging of drilled shaft formations. Proc. NDTCE-03, DGZfP, Berlin,2003, CD-ROM.314 Volkovoy, Y. and Stain, R. Ultrasonic crosshole testing of deep foundations –3D imaging. Proc. NDTCE-03, DGZfP, Berlin, 2003, CD-ROM.315 Cheng, C. and Sansalone, M. Effect on impact-echo signals caused by steelreinforcing bars and voids around bars. ACI Materials Jnl, 90, No. 5, Sept./Oct. 1993, 421–434.316 Germann Instruments, Copenhagen, Denmark, www.germann.org.317 Casas, J.R. An experimental study on the use of dynamic tests for surveillanceof concrete structures, RILEM. Materials & Structures, 1994, 27, 588–595.318 Carino, N.J. and Sansalone, M. Detection of voids in grouted ducts using theimpact-echo method. ACI Materials Jnl, 89, No. 3, May/June 1992, 296–303.319 Jaeger, B.J., Sansalone, M.J. and Poston, R.W. Using impact-echo to assesstendon ducts. Concrete International, 19, No. 2, Feb. 1997, 42–46.320 Pratt, D. and Sansalone, M. Impact-echo signal interpretation using artificialintelligence. ACI Materials Jnl, 89, No. 2, Mar./Apr. 1992, 178–187.321 Pessiki, S. and Rowe, M.H. Influence of steel reinforcing bars on the evaluationof early-age concrete strength using the impact echo method. ACI StructuralJournal, 94, No. 4, July/Aug. 1997, 378–388.

References 329322 Tawhed, W. and Gassman, S.L. Damage assessment of concrete bridge decksusing impact echo method. ACI Materials Journal, 99, No. 3, May/June 2002,273–281.323 Kesner, K., Sansalone, M.J. and Poston, R.W. Detection and quantificationof distributed damage in concrete using transient stress waves. ACI MaterialsJournal, 101, No. 5, July/Aug. 2004, 318–326.324 Lin, J.-M. and Sansalone, M. Impact-echo studies of interfacial bond quality inconcrete: Part 1 – effects of unbonded fraction of area. ACI Materials Journal,93, No. 3, May/June 1996, 223–232.325 Abraham, O., Leonard, C., Cote, P. and Piwakowski, B. Time frequency analysisof impact-echo signals: Numerical modelling and experimental validation.ACI Materials Journal, 97, No. 6, Nov./Dec. 2000, 645–657.326 Olson, L.D. and Wright, C.C. Non-destructive testing for repair and rehabilitation.Concrete International, 12, No. 3, Mar.1990, 58–64.327 Nazarian, S. and Desai, M.R. Automated surface wave method: Field testing,ASCE Jnl Geotech. Eng., 119, No. 7, July 1993, 1094–1111.328 Rhazi, J., Hassaim, M., Ballivy, G. and Hunaidi, O. Effects of concrete nonhomogeneityon Rayleigh waves dispersion. Magazine of Concrete Research,54, No. 3, June 2002, 193–201.329 Cho, Y.S. and Lin, F.-B. Nondestructive evaluation of in-place cement mortarcompressive strength using spectral analysis of surface waves. Constructionand Building Materials, 19, No. 10, 2005, 724–730.330 Sansalone, M., Lin, J.-M. and Streett, W.B. Determining the depth ofsurface-opening cracks using impact generated stress waves and time-offlighttechniques. ACI Materials Journal, 95, No. 2, March/April 1998,168–177.331 Popovics, J.S., Song, W-.J., Ghandehari, M., Subramaniam, K.V., Achenbach,J.D. and Shah, S.P. Application of surface wave transmission measurementsfor crack depth determination in concrete. ACI Materials Journal, 97, No. 2,March/April 2000, 127–135.332 Maguire, J.R. and Severn, R.T. Assessing the dynamic properties of prototypestructures by hammer testing. Proc. ICE, Pt. 2, 83, Dec. 1987, 769–784.333 Williams, C. Vibration testing of large structures to determine structural characteristics.Proc. Struct. Faults and Repair 87, Engineering Technics Press,Edinburgh, 31–35.334 Maguire, J.R. Engineering measurement and testing of land-based structures.Structural Engineer, 79, No. 7, April 2001, 33–37.335 Pearson, S.R., Owen, J.S., Tam, C.M. and Choo, B.S. Monitoring damage inreinforced concrete bridge decks using vibration data. Proc. Struct. Faults andRepair 03, Eng. Technics Press, Edinburgh, 2003, CD-ROM.336 Stain, R.T. Integrity testing. Civil Engineering, April 1982, 54–59 and May1982, 71–73.337 BS 1881: Part 205 Recommendations for radiography of concrete. British StandardsInstitution, London.338 Forrester, J.A. Gamma radiography of concrete. Proc. Symp. on NDT of Concreteand Timber, ICE London, 1969, pp. 9–13.339 Pullen, D. and Clayton, R. The Radiography of Swathling bridge. Atom,No. 301, Nov. 1981, 283–288.

330 References340 Parker, D. X-rated video. New Civil Engineer, 21 April 1994, 8–9.341 Redmer, B., Likhatchev, A., Weise, F. and Ewert, U. Location of reinforcementin structures by different methods of gamma-radiography. Proc. NDT-CE2003, DGZfP, Berlin, 2003, CD-ROM.342 Kear, P. and Leeming, M. Radiographic inspection of post-tensioned concretebridges. INSIGHT – Jnl. Brit. Inst. NDT, 36, No. 7, July 1994, 507–510.343 Brown, K.D. and St. Leger, J. Use of the Megascan imaging process in inspectionsystems for post-tensioned bridges and other major structures. Proc. NDT-CE2003, DGZfP, Berlin, 2003, CD-ROM.344 Mitchell, T.M. Radioactive/Nuclear methods, Handbook on Non-destructiveTesting of Concrete, ed. V.M. Malhotra and N.J. Carino, 2nd Edn, CRC Press,Chap. 12, 2004.345 Martz, H.E., Schneberk, D.J., Robertson, G.P. and Monteiro, P.J.M. Computerisedtomography analysis of reinforced concrete. ACI Materials Jnl, 90,No. 3, May/June 1993, 259–264.346 Nielsen, J. and Griffin, D.F. Acoustic emission of plain concrete. Journal ofTesting and Evaluation (JTEVA) USA, 5, No. 6, Nov. 1977, 476–483.347 Lim, M.K. and Koo, T.K. Acoustic emission from reinforced concrete beams.Magazine of Concrete Research, 41, No. 149, Dec. 1989, 229–234.348 Colombo, S., Forde, M.C. and Halliday, J. AE monitoring of concrete bridgebeams insitu. Structural Engineer, 81, No. 23/24, Dec. 2003, 41–46.349 Cullington, D.W., MacNeil, D., Paulson, P. and Elliot, J. Continuous acousticmonitoring of grouted post-tensioned concrete bridges. NDT &E International,34, No. 2, March 2001, 95–105.350 Ohtsu, M. The history and development of acoustic emission in concrete engineering.Magazine of Concrete Research, 48, No. 177, Dec. 1996, 321–330.351 BS EN 13554. Non-destructive testing – Acoustic emission – General principles.British Standards Institution, London.352 Mindess, S. Acoustic emission and ultrasonic pulse velocity of concrete. Int.Journal of Cement Composites and Lightweight Concrete, 4, No. 3, ConstructionPress, Aug. 1982, 173–179.353 Balazs, G.L., Grosse, C.U., Koch, R. and Reinhardt, H.W. Damage accumulationon deformed steel bar to concrete interaction detected by acoustic emissiontechnique. Magazine of Concrete Research, 48, No. 177, Dec. 1996, 311–320.354 Titus, R.N.K., Reddy, D.V., Dunn, S.E. and Hartt, W.H. Acoustic emissioncrack monitoring and prediction of remaining life of corroding reinforced concretebeams. Proc. 4th European Conf. on NDT, Vol. 2, Pergamon, Oxford,1988, pp. 1031–1040.355 Idriss, H. and Limam, A. Study and characterisation by acoustic emission andelectrochemical measurements of concrete deterioration caused by reinforcementsteel corrosion. NDT & E International, 36, No. 8, Dec. 2003, 563–569.356 Lyons, R., Ing, M. and Austin, S.A. Infleunce of diurnal and seasonal temperaturevariations on the detection of corrosion in reinforced concrete by acousticemission. Corrosion Science, 47, No. 2, Feb. 2005, 413–433.357 Rossi, P., Godart, N., Robert, J.L., Gervais, J.P. and Bruhat, D. Investigationsof the basic creep of concrete by acoustic emission. Materials and Structures,27, 1994, 510–514.

References 331358 Jenkins, D.R. and Steputat, C.C. Acoustic emission monitoring of damageinitiation and development in structurally reinforced concrete beams. Proc.Struct. Faults & Repairs 93, Eng. Technics Press, Edinburgh, 3, 1993, 51–57.359 Hulshizer, A.J. Benefits of the maturity method for cold-weather concreting.Concrete International, 23, No. 3, March 2001, 68–72.360 ASTM C1074 Estimating concrete strength by the maturity method. AmericanSociety for Testing and Materials, Philadelphia.361 Carino, N.J. The maturity method. CRC Handbook on Non-Destructive Testingof Concrete. CRC Press, ed. V.M. Malhotra and N.J. Carino, 2nd Edn,Chap. 5, 2004.362 BS 1881: Part 130 Method of temperature matched curing of concrete specimens.British Standards Institution, London.363 Cannon, R.P. Temperature matched curing: Its development, application andfuture role in concrete practice and research. Concrete, 20, No. 7, July 1986,27–30; 20, No. 10, Oct. 1986, 28–32, and 21, No. 2, Feb. 1987, 33–34.364 Pye, P.W. and Warlow, W.J. A method of assessing the soundness of somedense floor screeds. Current Paper, CP72/78, Building Research Establishment,Garston, 1978.365 BS 8204: Part 1 Screeds, Bases and Insitu floorings – Concrete bases and cementlevelling screeds to receive floorings – Code of Practice. British StandardsInstitution, London.366 Chung, H.W. and Law, K.S. Assessing fire damage of concrete by the ultrasonicpulse technique. Cement, Concrete and Aggregates, CCAGDP, 7, No. 2,American Society for Testing and Materials, Philadelphia. Winter 1985, 84–88.367 Evaluation and repair of fire damage to concrete. Spec. Publ. SP-92, AmericanConcrete Institute, Detroit, 1986.368 Short, N.R., Purkiss, J.A. and Guise, S.E. Assessment of fire damaged concreteusing colour image analysis. Construction and Building Materials, 15, No. 1,2001, 9–15.369 Soutsos, J.R. dos, Branco, F.A. and Brito, J. de. Assessment of concrete structuressubjected to fire – the FB Test. Magazine of Concrete Research, 54, No. 3,2002, 203–208.370 Zhang, X., Du, H.X., Zhang, B. and Phillips, D.V. Assessment of fire damage ofconcrete by using infrared thermal imaging method. Proc. Int. Conf. Concretefor Extreme Conditions, ed. R.K. Dhir, M.J. McCarthy and M.D. Newlands.Thomas Telford, Sept. 2002, 595–604.371 BS 1881: Part 124 Methods of testing concrete: Analysis of hardened concrete.British Standards Institution, London.372 Concrete Society Analysis of hardened concrete, Technical Report, 32, 1989.373 ASTM C85 Cement content of hardened Portland cement concrete. AmericanSociety for Testing and Materials, Philadelphia.374 Grantham, M.C. Determination of slag and pulverised fuel ash in hardenedconcrete – the method of last resort revised. STP 1253, Amer. Soc. for Testingand Materials, 1995.375 Mayfield, B. The quantitative evaluation of the water/cement ratio using fluorescencemicroscopy. Magazine of Concrete Research, 42, No. 150, March1990, 45–49.376 Neville, A.M. Properties of Concrete. Wiley, 4th Edn, 1996.

332 References377 French, W.J. Concrete petrography: A review. Quarterly Jnl of Eng. Geology,Geological Soc., 24, 1991, 17–48.378 Roberts, H.M. and Jaffrey, S.A.M.T. Rapid chemical test for the detection ofhigh alumina cement concrete. Information Sheet IS 15/74, Building ResearchEstablishment, Garston 1974.379 ASTM 856 The petrographic examination of hardened concrete. AmericanSociety for Testing and Materials, Philadelphia.380 BS 812: Part 104 Testing aggregates – Method for qualitative and quantitativepetrographic examination of aggregates. British Standards Institution, London.381 Concrete in aggressive ground. Special Digest 1, Building Research Establishment,Garston, 2005.382 Wilsch, G., Schaurich, D., Weritz, F. and Wiggenhauser, H. Laser-inducedbreakdown spectroscopy for on-site determination of chloride in concrete. Proc.NDT-CE 03, DGZfP, Berlin, 2003, CD-ROM.383 Determination of cement and chloride contents of hardened concrete. InformationPaper IP21/86, Building Research Establishment, Garston, 1986.384 Simplified method for the detection and determination of chloride in hardenedconcrete. Information Sheet IS 12/77, Building Research Establishment,Garston, 1977.385 Figg, J.W. Methods to determine chloride concentrations in in-situ concrete.TRRL Contractor Report, 32, 1986.386 ‘Hach’ Chloride test kit, www.aquakemtechnology.co.uk.387 ‘Quantab’ Chloride filtrators, www.materialstestingequip.com.388 Grantham, M. and Van-Es, R. Admitting that chlorides are variable. ConstructionRepair, 9, No. 6, Nov./Dec. 1995.389 Reknes, K. Measurement of chloride content in concrete. A Norwegian interlaboratorytest comparison. Corrosion of Reinforcement – Field and LaboratoryStudies for Modelling and Service Life. Proceedings of the Nordic Seminar inLund, Report TVBM – 3064, February 1995.390 BS 8500-1 Concrete: Complementary British Standard to BS EN 206-1.Method of specifying and guidance for the specifier, British Standards Institution,London.391 Kay, E.A., Fookes, P.G. and Pollock, D.J. Deterioration related to chlorideingress. Concrete, 15, No. 11, Nov. 1981, 22–28.392 Roberts, L. Recent advances in the understanding of cement/admixture interactions.Institute of Concrete Technology Annual Symposium, Institute ofConcrete Technology, Crowthorne, April 2005.393 Rixom, R. Chemical admixtures for concrete, Spon, 1999.394 Theophilus, J. Uncertainties in assessing the durability of concrete. Civil Engineering,Aug. 1986, 10–13.395 Sims, I. The assessment of concrete for carbonation. Concrete, 28, No. 6,Nov./Dec. 1994, 33–38.396 Watkins, R.A.M. and Pitt-Jones, A.A. Carbonation: A durability model relatedto site data. Proc. ICE, Structs. and Bldgs, 99, May 1993, 155–166.397 Spectronics Corporation. Detector keeps bridges from vanishing. ConcreteInternational, 16, No. 7, July 1994, 78.398 Natesaiyer, K., Stark, D. and Hover, K.C. Gel fluorescence reveals reactionproduct traces. Concrete International, 13, No. 1, Jan. 1991, 25–28.

References 333399 Polivka, M., Kelly, J.W. and Best, C.H. A physical method for determining thecomposition of hardened concrete. Spec. Tech. Publ. STP 205, 1958, AmericanSociety for Testing and Materials, Philadelphia, 135–152.400 Markestad, A. Changing over from standard microscopic technique to an imageanalysing microscope for measuring quality parameters of concrete. Proc. Int.Conf. on Concrete Structures, Trondheim, Tapir, 1978, 57–67.401 Poulsen, E. Preparation of samples for microscopic investigation. ProgressReport M1, Committee on alkali reactions in concrete, Danish National Instituteof Building Research and Academy of Technical Sciences, Copenhagen,1958.402 Placido, F. Thermoluminescence test for fire damaged concrete. Magazine ofConcrete Research, 32, No. 111, June 1980, 112–116.403 Chew, M.Y.L. Effect of heat exposure duration on the thermoluminescence ofconcrete. ACI Materials Journal, 90, July/Aug. 1994, 319–322.404 Midgley, H.G. The mineralogy of set high alumina cement. Transactions of theBritish Ceramic Society, 66, No. 4, June 1967, 161–187, Current Paper CP19/68, Building Research Establishment, Garston.405 Midgley, H.G. and Midgley, A. The conversion of high alumina cement. Magazineof Concrete Research, 27, No. 91, June 1975, 59–77; Current Paper CP72/75, Building Research Establishment, Garston.406 Neville, A. Should high alumina cement be re-introduced into design codes?Structural Engineer, 81, No. 24, Dec. 2003, 35–40.407 Ramachandran, V.S. Calcium Chloride in Concrete, Applied Science (London),1976, pp. 22–38.408 Willard, H.H., Merritt, L.L. and Dean, I.A. Instrumental Methods of Analysis,5th Edn, Van Nostrand, London, 1974.

Indexabrasion resistance 7, 15, 34–5, 48,50, 176, 221absorption tests 7, 12–13, 15, 33–4,120, 176, 206–20, 304AC harmonic analysis 195–200AC impedance 7, 12, 195–200access 9, 11, 12, 15, 27, 82, 141, 146,163, 171acousticemission 6, 7, 15, 34, 140, 159,162, 173, 250–3gauges see vibrating wire gaugesactual concrete strength 4, 67,88, 174admixtures 3, 118, 285age correction factors 22, 35, 135aggregate/cement ratio 41, 261, 264–7, 269grading 261, 274–5type 3, 4, 14, 20, 41–2, 61, 63,88–90, 106–7, 109, 113, 118,124, 127, 261, 264, 267, 269,274–5air content 7, 34, 126, 221,288–9air permeability 2, 4, 13, 31,212–15, 217alkali/aggregate reaction 4, 10, 52, 78,120, 126, 128, 134, 164, 176,201, 220–1, 275, 284–5,289–90, 303content 261, 284immersion test 284–5, 303antennas 26–30, 235backscatter methods 202,247–9beams 15, 17, 18, 21–2, 34, 64,73, 88, 101, 107, 111, 121,142–6, 148, 153, 171–5, 242,250, 252break-off test 7, 13, 16, 26, 82,116–18calculations 4, 25, 29, 30–1, 47, 71,79, 140, 155, 174, 202, 266,270–1, 286, 305–11calibration 13–14, 16, 17, 21–3,26–7, 32–3, 40, 43–5, 47–9,60–2, 66–7, 74, 76–8, 81,87–95, 97–8, 106–9, 112–13,118, 139–40, 160, 162–5, 168,179–83, 210capillary rise test 220Capo test 98, 101, 109–10, 301capping of cores 125–7, 130carbonation 4, 7, 12–13, 40, 43–4,47, 93, 105, 107, 111, 176, 285–7,289, 304cast-in inserts 94–7cementcontent 3, 6, 7, 23, 28, 34–5, 40–1,261, 264–7, 304replacements 118, 232–3, 267–9,272–3type 7, 22, 76, 98, 106, 117, 120,136, 261, 272–4, 292–3chain-dragging 14, 236–7characteristic strength 29–32, 141

Index 335chemicalanalysis 6, 34, 120–35, 176,261–94, 303–4attack 10–11, 77–8, 261, 275–7,282–7, 289chlorides 4, 7, 10, 12, 35, 158, 176,178, 188–90, 230–1, 261, 277–84,294, 303–4coefficient of variation 18, 24–7,29–30, 47, 74, 91, 98, 107, 113,133, 137, 297, 301collapse load test 17, 33, 174columns 17, 19, 21–2, 106, 121combined tests 32–4compaction 2–3, 17–18, 21, 40, 43,49, 67–8, 76, 118, 129, 134–5,139, 258comparative testing 9, 13–14, 16, 33,49, 68, 77, 81, 185–91, 205,223–4, 230, 248compliance testing 1–3, 16–18, 28,83, 92, 94, 98, 121, 128compressive strength see strength,compressivecontour plots 18–20, 22, 24, 68,73–4, 76, 168, 187–90control testing 1, 2, 29, 44, 49, 67–8,74, 116, 118conversion of HAC 193, 273–4, 293cores 6–7, 9, 13–14, 16–18, 21, 23,26–7, 30, 83, 91–3, 109, 111,120–39, 164, 174–5, 260,296–300, 309–11compression testing 120–1, 125–6,136, 309–11large 16–17, 142–3small 16–17, 121, 135–9tensile testing 126–8corrosion see reinforcement, corrosioncover 2, 4, 7, 10, 12, 28, 33–5, 51, 70,119, 121, 157, 178–85, 236, 304covermeters 34, 121, 135, 179–85crackscauses 8–11, 64, 74–6, 177, 258,302–3depth 75–6, 243–4detection 15, 51, 58, 64, 74–6, 81,92, 130, 160–1, 223, 230, 245,250–3identification 5, 10, 122, 124, 137,152, 177, 258mapping 10, 221creep 8, 11, 23, 153, 252curing 2–4, 7, 13–14, 17–18, 21–2,34, 40, 43, 45, 67, 73, 79, 89, 94,98, 112–13, 117–18, 128, 133–4,157, 168, 253–7damage 11–15, 24, 48, 51, 77–8, 82,91–3, 102, 106, 109, 111, 118,121, 124, 128, 130, 134, 139,141, 146, 149, 156, 165, 170,173, 174, 177, 250–3deflections 8, 15, 23, 141–5, 149,152–3, 155, 159–60, 163, 171,173–4delamination 14, 35, 177, 223, 227,236–8, 244‘Demec’ gauge 161–2, 164–8density measurement 4, 32, 35, 53,124–6, 135, 139, 246–9destructive tests 6, 13, 14, 17, 21, 27,33, 49, 74, 140, 170–5deterioration 3–4, 7–9, 11–13, 31,33, 52, 73, 77–8, 128, 140, 144,156–7, 176, 243, 258–60, 275–7,282–7, 289, 302–4differential scanning calorimetry 293differential thermal analysis 292–3displacement transducers see inductivedisplacement transducersdocumentation 5–6, 9–10, 23, 34–5drilled hole methods 6, 33, 93–4,101–9durability 2–4, 6–8, 12, 15, 33, 50,158, 176–222, 223dynamic response 7, 15, 170, 236–44elastic modulus 4, 47, 52–3, 60–1,78–9, 243electrical conductivity 226, 231electrical methods 7, 53, 176–201,202–4electrical resistance gauges 161,163–8, 170electrical resistivity 4, 7, 12, 33, 178,191–5, 201–2ESCOT 102, 109excess voidage 125–6, 128–9, 139expansion tests 7, 12–13, 126, 164,221, 303factors of safety 4, 31–2, 156, 295,299, 300–2Figg test 213–14

336 Indexfire damage 4, 5, 11, 15, 77–8, 176,258–60, 290–2flow tests 217–18formwork stripping 4, 83, 93–4, 99,111, 253frost damage 10–11, 77, 176, 221galvanostatic pulse 12, 195, 198–200gammaradiography 7, 12, 14, 34, 246radiometry 7, 14, 32, 247–9gas pressure test 127–8, 139, 206, 215gauges 149–52, 156, 161–73‘Hach’ method 279–80, 282, 304half-cell potential tests 7, 12, 33, 35,158, 185–91, 194–5, 200hardness testing 7, 13–15, 32, 36–50,83, 86–9, 93high alumina cements 5, 41, 78, 101,111, 125, 127, 144, 220, 273–4,292–3high strength concrete 19, 20, 66, 83,85, 95, 132histograms 24–5, 74, 296, 298, 300holographic methods 249–50, 300honeycombing 2, 8, 74, 76, 124–6,230–8impact-echo 14, 24, 35, 238–42impulse response 52, 238–9inductive displacement transducers163, 165, 169infraredabsorption spectrometry 293–4beam 160thermography 7, 14, 35, 201,223–5, 260initial surface absorption test 13, 15,33–4, 207–12, 215, 222integrity 2–4, 7, 14, 24, 223–60internal fracture test 7, 13, 16, 26,101–9internal reactions 10–11, 52, 78, 262interpretation 1–35, 40, 46, 52, 60,87, 120–1, 128, 135, 149–52, 158,163, 174–5, 188–90, 194, 196,198–200, 202, 208, 211, 219,230–7, 241, 246, 248, 251–2, 256,259, 295–304in-situ strength see strength, in-situlasers 152, 159–60, 170, 250, 277, 294layer thickness 35, 77–9, 238,240, 243lightweight concrete 14, 18, 20–1, 33,41–2, 61, 63, 83, 89, 91, 93, 98,107–8, 115–16, 129–30, 136–7,139, 231linear polarization resistance 7, 12,158–9, 179, 195–8load/deflection curve 15, 23, 144–5,153, 155, 163, 173–4load testing 140–75in-situ 4, 5, 15, 17, 23, 24, 33, 49,140–57, 299instrumentation 7, 141, 149–55,157–70, 171–4, 252–3methods of loading 141–3, 144–9,171–3ultimate 33, 140, 156, 170–4location of tests see test, locationLok-test 94–101, 105, 109–10long-term monitoring see monitoringmagneticflux leakage 183induction 179–80, 183methods 179–85maturity 13–14, 26, 35, 61, 106, 120,157, 253–7microscopic methods 7, 260, 270,273–4, 276, 286, 287–90, 293microwave absorption 204–5mix proportions 23, 62, 264–72moistureconditions 12–14, 18, 26, 31, 40,43–4, 49, 62, 66, 76–7, 88–9,106, 128–9, 176–7, 189, 190,194, 198, 206, 208, 211,214–16, 221, 224, 230content 12, 13, 66, 77, 201–5,231–3gauges 201–5monitoring 2, 4, 15, 33, 54–5, 79, 82,98, 116, 118, 140, 157–70, 250–3neural networks 24, 33, 183,232–4, 242neutron moisture gauges 202nuclear methods 7, 35, 202, 244–9number of tests 4, 8, 15–18, 28–31,47, 122, 135, 138–9

Index 337optical fibres 162–3, 170over-reinforced beams 155,173–4partially destructive tests 6, 13–14,27, 33, 74, 82–119pavement thickness 35, 227, 230,236, 242penetration resistance 7, 13, 14, 23,36, 82–93, 118permeability 2, 4, 7, 12, 13, 31, 126,206–20absorption tests 7, 12, 13, 15,33–4, 208, 218–20, 260air 212–17‘Autoclam’ 215–16BRE test see Figg testCapillary rise test 220combined test 215‘Figg’ test 208, 212–15flow tests 208, 217–18GWT 216–17ISA test 208–12, 215sorptivity test 215, 219–20perturbative methods 4, 179,195–200petrographic methods 3, 6, 7, 12–13,33, 35, 221, 258, 260–1, 267,273–4, 276, 287–90photoelasticity 253pile integrity 236, 239–40pin penetration test 36, 83, 93planning 1–35, 120, 158, 257,295–304point load test 126–7, 139Poisson’s ratio 53, 60, 79potential noise 201‘potential’ strength see strength,potentialprestressing ducts 80, 178, 183, 229,230, 235, 238, 242, 245–7pull-off tests 7, 14, 16, 26, 34–5, 82,111–16, 118, 301pull-out tests 7, 13, 14, 16, 23, 26,34–5, 82, 93–101, 102–3, 106–7,109, 111, 118pulse-echo tests 7, 32, 237–42pulse velocity see ultrasonic, pulsevelocityPUNDIT 54–6quality control tests 2, 27–30, 32, 80,116, 118–19, 170, 175, 185, 211,214, 240‘Quantab’ method 280–3, 304radar 6–7, 14, 24, 32, 35, 201, 205,225–36radioactive methods 7, 202, 244–9radiography 7, 12, 14, 34, 244–7radiometry 7, 14, 32, 244, 247–9rebound hammer 26, 32, 34, 36–7,39–50, 77, 82, 91, 98, 296–9reinforcementcorrosion 4, 6, 7, 10–13, 15, 24,33, 157–8, 176–9, 185–201,227, 230, 236, 253, 277, 284–5,303–4location 7, 31, 136, 139, 179–85,224, 229, 230, 232, 234, 246,249pulse velocity 51, 68–9, 72, 305–8size 72, 180, 183–5, 230, 234–5see also covermetersreinforcement corrections; cores130–1, 310relative humidity 7, 201, 203–4, 221relative permittivity 226, 230–1,233–4, 236repairs 2, 4, 23, 34–5, 76, 111,116, 156resistivity tests see electrical resistivitysafety 4, 7, 9, 11, 15, 84, 91, 127,141, 144, 146, 149, 152, 156,171, 173, 227, 245, 247safety factors see factors of safetysample preparation 6, 56, 264, 270,272–5, 280sampling 6, 8, 34, 74, 120, 185,262–3, 278scanning electron microscopy 293–4Schmidt hammer see rebound hammerscratch test on aggregate 86–7screed tester 34, 222, 258secondary testing 2, 3, 9selection of tests see test, selectionserviceability 4, 8, 121, 139–40,156, 174shearing rib method 119shrinkage 8, 10, 11, 118, 122,252, 303

338 Indexsignal processing 52, 55, 80, 230,235, 238–44slabs 14, 17, 18, 21–2, 74, 78, 91–2,95, 121, 139, 146–9, 163, 188,222, 227, 238, 240–1, 244sorptivity test 219–20soundness of screeds 34, 258specification compliance 1–3, 16–18,28, 83, 92, 94, 98, 121, 128,295–7spectral analysis 243–4speed of tests 12, 13, 16, 80standard deviation of strength 25–7,29–30, 299, 302‘standard’ specimens 2, 20–2, 26, 28,29, 67–8, 133, 302standards 2, 5, 6, 14, 34–5, 39, 68,76, 82, 99, 116, 120, 121, 135stiffness damage test 128‘Stoll tork’ test 118–19strain measurement 7, 140–1, 159,163–70, 171–2, 253strengthcalibration 16–17, 21–2, 26–7,32–3, 40, 43–9, 61, 66, 74,76–7, 81, 89, 98, 109compressive 5, 27, 32, 34, 35, 47,63, 76, 87–8, 93–9, 103–7, 109,112, 115, 118, 127–8, 135, 139,309–11design 31, 299, 301–2development 3, 14, 17, 22, 33, 35,43, 47, 62, 82, 111, 119, 128,157, 223, 254–7in-situ 3–7, 13, 14, 20–2, 26–31,67–8, 77, 98, 110–11, 116, 119,132–3, 243, 297–302member 47, 52, 64, 73, 76, 78,140, 170–5potential 125, 133–4, 310–11tensile 35, 64, 106, 111–12, 115,117–18, 120, 126, 128, 260variability 3, 14, 15–18, 20–1,23–4, 29–31, 73, 136–8structural safety 4, 31–2, 141–57subsurface radar see radarsulfates 7, 10, 176, 261, 275–7surfacecracking 8–9, 11, 58, 302–3hardness 7, 13–15, 23, 32, 36–50,94, 221, 258, 296–9wave techniques 52, 243tapping 7, 14, 236–7, 258temperature effects 15, 39–40,44, 63–4, 118, 155, 169, 185,194, 210–11, 221, 223, 232,253–7, 259temperature matched curing 7, 13–14,253, 257tensile strength see strength, tensiletestlocation 4, 8, 11, 15, 18, 23–4, 39,47, 49, 74, 92, 135, 161,295–302selection 5, 8, 9, 12–15, 295–304thaumasite 276thermography see infrared,thermographythermoluminescence 7, 15, 258,290–2tomography 14, 52, 80, 240, 248transducers 60, 62, 68, 70, 75–6, 78,80, 149, 150–2, 160, 163–6,169–72, 250ultrasonicpulse attenuation 52, 55, 58,65, 74pulse velocity 7, 13–14, 16, 24, 26,32–4, 51–81, 88, 98, 240, 258,305–8under-reinforced beams 144–5,174uniformity 2, 11, 13, 24, 47–9, 73‘V’-meter 54vacuum test 212, 217variability 17–18, 20–1, 98, 101–4,106–7, 109, 118, 121, 127, 130,136–7, 139vibrating wire gauges 162–3, 165,169–70visual inspection 6, 8–12, 17, 35,120, 122, 135, 142, 172, 258–60,295–304void location 74–5, 223–5, 229–30,232–4, 238, 240–1‘Volhard’ method 277, 279, 283walls 20–2, 92, 121, 238, 242,302water/cement ratio 61–2, 88, 98, 117,194, 261, 270–2

Index 339water content 261, 263, 269–72wear tests 7, 15, 221–2Windsor probe 16, 26, 82–93,301within-member variability 18, 20–1,23–4, 73wood-screw method 111workmanship 2, 3, 8, 17, 24,135, 165X-raydiffraction 293radiography 7, 12, 14, 34,244–6spectroscopy 267, 269, 274,279, 292zero resistance ammetry 195, 201

eBooks – at www.eBookstore.tandf.co.ukA library at your fingertips!eBooks are electronic versions of printed books. You canstore them on your PC/laptop or browse them online.They have advantages for anyone needing rapid accessto a wide variety of published, copyright information.eBooks can help your research by enabling you tobookmark chapters, annotate text and use instant searchesto find specific words or phrases. Several eBook files wouldfit on even a small laptop or PDA.NEW: Save money by eSubscribing: cheap, online accessto any eBook for as long as you need it.Annual subscription packagesWe now offer special low-cost bulk subscriptions topackages of eBooks in certain subject areas. These areavailable to libraries or to individuals.For more information please contactwebmaster.ebooks@tandf.co.ukWe’re continually developing the eBook concept, sokeep up to date by visiting the website.www.eBookstore.tandf.co.uk