Failures in Concrete Structures - · PDF fileRobin Whittle Failures in ConCrete struCtures Case Studies in Reinforced and Prestressed Concrete A SPON BOOK ISBN: 978-0-415-56701-5 9

Robin Whittle

Failures in ConCrete struCturesCase Studies in Reinforced and Prestressed Concrete

A S P O N B O O K

ISBN: 978-0-415-56701-5

9 780415 567015

9 0 0 0 0

Y106141

Cover image: © Dr Jonathan G M Wood

STRUCTURAL ENGINEERING

“The book provides a fascinating insight into what can go wrong. It illus-trates the importance of approaching problems from first principles and identifying the key load paths, loads, etc. This type of material rarely sees the light of day for reasons of litigation and commercial confidentiality.” —Robert Vollum, Imperial College, London

“This book is essential reading for any experienced engineer and of-fers insight into how designers and builders accidentally create struc-tures that fail. This book both explains how things fall down and what was done or could have been done to overcome these errors. Learning from errors is an essential part of any engineer’s knowledge and expertise.” —David M. Scott, Laing O’Rourke, Connecticut

Failures in Concrete Structures: Case Studies in Reinforced and Prestressed Concrete presents a selection of the author’s firsthand experience with incidents related to reinforced and prestressed concrete structures. He includes mistakes discovered at the design stage, ones that led to failures, and some that involved partial structure collapse—all of which required remedial action to ensure safety. The book focuses on specific incidents that occurred at various points in the construction process, including mistakes related to structural misunderstanding, extrapolation of codes of practice, poor detailing, poor construction, and deliberate malpractice. The author provides this information for readers to gain an understanding of errors that can occur in order to avoid making them. The book also examines issues during the procurement process, as well as the importance of research and development to gain a deeper knowledge of materials and help prevent future failures.


Y106141_Cover_mech.indd 1 8/24/12 2:11 PM

Case Studies in Reinforced and Prestressed Concrete

FAILURES IN CONCRETE STRUCTURES

Case Studies in Reinforced and Prestressed Concrete

Failures in ConCrete struCtures

Robin Whittle

A S P O N B O O K

CRC PressTaylor & Francis Group6000 Broken Sound Parkway NW, Suite 300Boca Raton, FL 33487-2742

© 2013 by Robin WhittleCRC Press is an imprint of Taylor & Francis Group, an Informa business

No claim to original U.S. Government worksVersion Date: 20120815

International Standard Book Number-13: 978-0-203-86127-1 (eBook - PDF)

This book contains information obtained from authentic and highly regarded sources. Reasonable efforts have been made to publish reliable data and information, but the author and publisher cannot assume responsibility for the validity of all materials or the consequences of their use. The authors and publishers have attempted to trace the copyright holders of all material reproduced in this publication and apologize to copyright holders if permission to publish in this form has not been obtained. If any copyright material has not been acknowledged please write and let us know so we may rectify in any future reprint.

Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced, transmitted, or utilized in any form by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying, microfilming, and recording, or in any information stor-age or retrieval system, without written permission from the publishers.

For permission to photocopy or use material electronically from this work, please access www.copy-right.com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization that pro-vides licenses and registration for a variety of users. For organizations that have been granted a pho-tocopy license by the CCC, a separate system of payment has been arranged.

Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe.

Visit the Taylor & Francis Web site athttp://www.taylorandfrancis.com

and the CRC Press Web site athttp://www.crcpress.com

v

Contents

Foreword ixAcknowledgements xiIntroduction xiii

1 FailuresduetoDesignErrors 1

1.1 EdgeBeamandColumnConnection 21.2 ConcreteTruss 51.3 CircularRampstoCarPark 81.4 TransferBeamwithEccentricLoading 101.5 EarlyThermalEffects 101.6 SecondaryEffectsofPrestressing 141.7 TemperatureEffectsonLong-SpanHybridStructure 161.8 LoadingforFlatSlabAnalysis 181.9 PrecastConcreteCarPark 191.10 ArchFloor 211.11 PrecastConcreteStairflights 211.12 ShearStudsonSteelColumntoSupportConcreteSlab 241.13 PiledRaftforTowerBlock 261.14 FloatingPontoonforResidentialBuilding 271.15 PrecastColumnJointDetail 27

2 ProblemsandFailuresduetoErrorsinStructuralModelling 31

2.1 ReinforcedConcreteTransferTruss 312.2 ModellingRigidLinks 312.3 AssessingModelLimitsandLimitations 322.4 EmpiricalMethods 342.5 InitialSizingofSlabs 352.6 AnalysisofFlatSlabswithFiniteElementPrograms 352.7 ScaleEffects 37

vi Contents

3 FailuresduetoInappropriateExtrapolationofCodeofPracticeClauses 39

3.1 CoolingTowers 393.2 DesignBendingMoments 413.3 PileswithHighStrengthReinforcement 423.4 ShearCapacityofDeepSections 43

4 FailuresduetoMisuseofCodeofPracticeClauses 47

4.1 FlatSlabandTwo-WaySlabBehaviour 474.2 RibbedSlabSupportedonBroadBeam 484.3 CarParkColumns 50

5 ProblemsandFailuresduetoInadequateAssessmentofCriticalForcePaths 53

5.1 HeavilyLoadedNibs 535.2 ShearWallwithHolesandCornerSupports 535.3 DesignofBootNibs 56

6 ProblemsandFailuresduetoPoorDetailing 57

6.1 ConcreteOffshorePlatform 576.2 AssemblyHallRoof 606.3 UniversityBuildingRoof 626.4 MinimumReinforcementandCracking 646.5 PrecastConcretePanelBuilding 656.6 Footbridge 67

7 ProblemsandFailuresduetoInadequateUnderstandingofMaterials’Properties 69

7.1 ChangesoverTime 697.2 RebendingofReinforcement 717.3 TackWeldingofReinforcement 737.4 HighAluminaCement 737.5 CalciumChloride 747.6 Alkali–SilicaReaction 757.7 LightweightAggregateConcrete 77

8 ProblemsandFailuresduetoPoorConstruction 79

8.1 FlatSlabConstructionforHotel 79

Contents vii

8.2 SteelPilesSupportingBlockofFlats 818.3 ShearCracksinPrecastTUnits 818.4 CantileverBalconiestoBlockofFlats 828.5 PrecastConcreteTank 838.6 CarPark 858.7 CrackingofOffshorePlatformduringConstruction 878.8 SpallingofLoadBearingMullions 908.9 Two-WaySpanningSlab 928.10 ChimneyFlueforCoal-FiredPowerStation 93

9 ProblemsandFailuresduetoPoorManagement 97

9.1 Column–SlabJoint 979.2 PlacingofPrecastUnits 1009.3 WeakAggregateConcreteinChimney 101

10 ProblemsandFailuresduetoPoorConstructionPlanning 103

10.1 PowerStationonRiverThames 10310.2 TowerBlock 106

11 ProblemsandFailuresduetoDeliberateMalpractice 111

11.1 FloorwithExcessiveDeflection 11111.2 PilesforLargeStructure 11311.3 InSituColumnsSupportingPrecastBuilding 113

12 ProblemsArisingfromtheProcurementProcess 117

12.1 EffectsofDifferentFormsofContracts 11712.2 Workmanship 11812.3 CheckingConstruction 119

13 ContributionsofResearchandDevelopmenttowardAvoidanceofFailures 121

13.1 LinksbetweenPracticeandResearch 12113.2 FlatSlabBehaviour 12113.3 SpanandEffectiveDepthRatiosforSlabs 12213.4 BeamandColumnJoints 12213.5 TensionStiffeningofConcrete 123

References 127

ix

Foreword

Errarehumanumest.…Westructuralengineersarehumanandsohavemadeanumberoferrorsovertheyearsresultinginnarrowescapes,badlyperformingstructures,andevenfatalcollapses.ButasSenecacontinues…sedperseverarediabolicum,wemustnotrepeatourerrors.

Toavoidthismeansthatwemustlearnfromourpastmistakes;wemustknowwhatwentwrongandwhy.Someofthelessonsfromourpasterrorsget embodied in clauses in codes of practice, butmanydo not, and thecollectivememoryoftheprofessiontendstofadeasthegenerationofengi-neerswholearntfromthemishapsandcatastrophesretires.

Pastbooksonthesubjectofstructuralfailurestendedtodealwiththegeneralcausesoffailuresandmethodsofinvestigation,illustratedwiththemorespectacularexamples.However,detailsofsomefailuresthathavenotmadetheheadlines,butneverthelessholdimportant lessons,arehardtofindormaynotevenbeinthepublicdomain.

In thepast,RobinWhittle and Iworked together atArupR&Donavarietyofproblemsofconcretestructures.Someofthesearosefromfail-ures,andotherswereencounteredwhenforestallingundesirableoutcomesof the enthusiasm—untempered by experience—of some of our youngercolleagues.

RobinwasalsoinclosecontactwithresearchersatthenowsadlydefunctCement&ConcreteAssociation,thePolytechnicofCentralLondon,andtheuniversitiesofLeeds,Durham,andBirmingham,andsowasprivytomuchof thebackground for the initialdraftandsubsequent revisionsofCP110.

Nowadays,apreoccupationwiththeever-multiplyingminutiaeofcodes,whetherEuroCommunityorNational,canblinddesignerstotheimpera-tives of first principles. New patterns of procurement and site manage-mentalsowiden thecommunicationgapbetweendesignandexecution,and exert pressures to adopt shortcuts that sometimes have unforeseenconsequences.

x Foreword

Withhisbackground,Robiniswellplacedtopresentaselectionofcasestudiesthathavelessonsforallofus.Thisisabookthatshouldbereadbythosestructuralengineerswhowishtobroadentheirknowledgebylearn-ingfromsomeoftheexperiencesofthelast50yearsandalsoforthosewhowouldliketorefreshtheirmemories.

PoulBeckmann,F.I.Struct.E.,M.I.C.E.,M.I.D.A.,Hon.F.R.I.B.A.

xi

Acknowledgements

TheauthorwouldparticularlyliketothankPoulBeckmannforhiscon-tributiontothisbook.Hehasprovidedthedescriptionsofseveraloftheincidentsandhiscommentsontherestofthetexthavebeengreatlyvalued.

Thecontributionsfromthefollowingpeoplearealsogratefullyacknowl-edged.

IanFelthamTerryBellAlfPerryGeoffPeattieTonyStevensTonyJonesChrisKenyonJohnHirstBrynBird

SteveMcKechnieHowardTaylorRobertVollumRichardScottAndrewBeebyRobKinchGillBrazierAndrewFraser

xiii

Introduction

Thisbookisapersonalselectionofincidentsthathaveoccurredrelatedtoreinforcedandprestressedconcretestructures.Notallhaveledtofailuresandsomeofthemistakeswerediscoveredatthedesignstage.Eachincidentrequiredsome formof remedialaction toensure safetyof the structure.Someoftheincidentswerecausedbymistakesindesignorconstructionorboth.Someinvolvedcollapseofpartofthestructure,butinsuchcasesthecausewasfrommorethanoneunrelatedmistakeorproblem.Afewoftheerrorsandincidentswerecausedbydeliberateintent.

Chapters1to11describespecificincidentssuchasstructuralmisunder-standing,extrapolationofcodesofpractice,detailing,poorconstruction,andotherfactors.Whenaparticularincidentinvolvedmorethanoneofthesecauses,itisdescribedinthemostrelevantsection.Chapters12and13discussissuesrelatedtoprocurementandresearchanddevelopment.

Carehasbeen takennot toname theparticularprojects inwhich theincidents occurred, and the intention in providing the information is toensurethatsuchmistakescanbeunderstoodandavoidedinthefuture.

Someoftheproblemswerediscoveredinassociationwithrequestsforsupportaboutadifferenttopic.Intryingtodiscoverthedetailsoftheprob-lemitbecameclearthatothermoreseriousissueswereatstake.Thisbegsthequestion,howmanyunresolvedproblemsandmistakesareouttherethat have not seen the light of day? It is fortunate that most reinforcedandprestressedconcretestructuresareindeterminateandallowalternativeloadpathstoformandpreventfailuresthatwerenotforeseeninthedesign.

Thereisaworryingtrendinboththedesignandconstructionofbuildingstructures.Materialstrengthsofbothconcreteandsteelhavecontinuedtoincreaseforthepastcentury.Atthesametime,theoverallsafetyfactorusedindesignhasbeenreducing.Thistrendislargelycausedbythepressuretoreducecosts.Inevitablythetimewillcomewhentheeffectofthistrendwillmeanthattheerrorsmadeindesignandconstructionwillnotbeabsorbedbytheglobalsafetyfactorortheabilityofastructuretofindalternativepathsfortheloads.Thisincreaseinriskisnothelpedbythechangesinthe

xiv Introduction

formofcontractsandcontractualprocedure.Healthandsafetyclausesdonotprovide,intheauthor’sopinion,thenecessarysafetynets.

Inordertoimprovethequalityofallaspectsofdesignandconstruc-tion,itisessentialtoinvolvepeoplewhoknowwhattheyaredoing.Moretrainingisrequiredattheoperationallevelsofbothdesignandconstruc-tion. More research and scientific development are required to allowbetterunderstandingofthematerialsandtheiruses.Butthemoresophis-ticatedthematerialsandtoolsbecome,themoreintelligenceisrequiredfortheircontrol.

1

Chapter 1

Failures due to Design Errors

Manyfailures,wheninvestigated,havebeenfoundtoarisefromacombi-nationofcauses.Thetraditionaldesignsequencestartswiththesizingofmembers.Thesearedeterminedfromtheloading(permanentandvariableactions)withreferencetobendingmomentsand,forbeams,shearforces.Thereinforcementisthencalculatedtocaterfortheseforces.Muchofthereinforcementisdetailedonlyaftercompletionofthecontractdocuments.Later,ifproblemsarefoundinfittingtherequiredreinforcementintoanelementor joint, it isdifficulttochangethesizeofsectiononwhichthearchitectandservicesengineersagreed.Manysuchproblemscouldhavebeenavoidedbyproducingsketchesearlyontoshowhowthejointdetailscouldworkbeforesizeswerefinalised.

Computer software is commonly used to provide design information.Thiscannotalwaysbetailoredtosuittheproblemexactly.Alltoooftenadesignerfailstoensure,withmanualchecks,thatthesoftwareprovidesareasonablesolutionfortheparticulardesign.Theeffectsofcreep,shrink-age,andtemperatureareoftennotconsideredbythesoftware,andmanualcalculationsarerequiredtocheckwhethereffectsaresignificant.

Robustnesshasbecomeanimportantconsiderationindesignandthisiscloselylinkedtothedetailingofjoints.Theelementsofaninsitustructurearemechanicallyconnectedtogetherbynormaldetailingandthisshouldprovidesufficientrobustness.However,forhybridstructurescontainingamixtureofprecastandinsituconcrete,muchmorethoughtisrequiredtoobtainreliablejointdetails.

Acode-of-practicementalitycan inhibitaholisticapproach todesign.Incodes,thewholeisbrokendownintopartsthatareanalysedseparately.Theresultisasafestructurebutnotoneinwhichthestrainsandstresseshavemuchsemblancetothecalculatedones.Buildingsaredesignedfortheloadsexpectedtobeapplied,butrarelyforthestrainscausedbyshrinkage,creep,andtemperatureeffectsontheconcrete.

Design-buildcontractshavemeantthatmoreconsiderationofthecon-structionmethodsisgivenatthetimeofdesign.Thishasoftenledtomoreefficient construction (faster and/or cheaper). Unfortunately it has also

2 Failures in Concrete Structures

meantthatthedesignprocessisdominatedbythedemandsofspeedandcost.Sometimesthishasledtoadesignernottoconsideralltheimportanteffectson thefinal structure.Thedesignof inadequatemovement jointsinbuildings(e.g.,carparks)isanexamplethathassometimesbeencom-poundedbypoorworkmanship.

1.1 EDGE BEAM AND COLUMN CONNECTION

Acollapseoccurred involving the connectionofaheavy concrete gutter(1mwide)tothesupportingcolumns.Thestructureconsistedofaseriesofreinforcedconcreteedgecolumnssupportingsteeltrussesthatspannedthewidthof thebuilding.Edgebeamsspannedbetween thecolumns tosupport the concrete gutter. Figure 1.1 shows the collapseof the gutter.Althoughtheedgebeamhadbeendesignedanddetailedtoresistthefullload fromthegutter (bendingmoment, shear,and torsion), thishadnotbeencarriedthroughtothejoint.Figure 1.2showswheretheedgebeamjuststartedtotearawayfromthecolumnandFigure 1.3showsthetopofacolumnwheretheedgebeamhasseparatedandfallenoff.

Analysis—Figure 1.4showsafailuremodelbasedonthetensionstrengthoftheconcrete.Evenifthelinksthathadbeendetailedtopassthroughthejointhadbeenconstructedthisway,theywouldhavebeeninadequatetosupporttheloading.

Figure 1.1 Collapse of concrete gutter.

Failures due to Design Errors 3

Areasonablehandcalculationtochecktheresistanceistoassumethatthe tensile strengthof concrete is abouta tenthof the cube strength. Inthiscase,theconcretestrengthwas20MPasothetensilestrengthmaybetakenas2MPa(nosafetyfactorsapplied).Figure 1.4showshowthetensilestrengthoftheconcreteprovidesthetorsionalresistance.

Figure 1.2 Edge beam starting to sepa-rate from column.

Figure 1.3 Column where edge beam has broken off.

Cracks form

Gutter slab

Column

Compression

T1Appliedtorque

T2

Tension

Tension

Figure 1.4 Failure model relying on tension strength of concrete.


Spacingofcolumns 7.0mSupportingbeamheight,hb 0.6mSupportingbeamwidth,bb 0.3mWidthofcolumnhead 0.6mWidthofbeamsupport 0.1mTorquefromselfweightofgutter 25×7×(0.2×1×(0.9–0.3/2)+0.2 ×(0.8–0.3)×(0.8–0.3)/2+0.3/2) =56.88kNmDensityofsandandbricks 18kN/m3

Volumeofsand 1m3

Loadfromsand 18kNVolumeofbricks 0.6m3

Loadfrombricks 10.8kNTorquefromsandandbricks (18+10.8)×(0.8–0.3/2)=18.72kNm

Totalappliedtorque(unfactored) =56.88+18.72=75.6kNm

T1=0.5×2×0.6/2×0.6×1000 =180kNTorqueresistanceofT1=180×0.6×2/3 =72kNmT2=0.5×2×0.3/(2×3)×0.6×1000 =30kNTorqueresistanceofT2=30×0.1×2/3 =2kNm

Totalresistancefromtensileconcrete =72+2=74kNm

Applied load exceeds resistance

Detailing — The reinforcement for the edge beam–column joint wasbuiltasshowninFigure 1.5.Thetoptwolinksweredetailedtoenclose

T10s@100

T20s

2T8s2T20s

2T16s

Steel truss.

T8 links@130 T8 links @ 300 (stopped off at beam cage)

T8 links @ 300

Figure 1.5 Reinforcement detailing for edge beam and column joint.


allthecolumnmainbars.Evenifthelinksthathadbeendetailedtopassthrough the joint had been constructed this way, they would have beeninadequatetosupporttheloading.Ifthedetailhadbeendrawntoalargescale it would have become apparent that the column links would havebeendifficulttofitthroughthebeamcage,andthisshouldhavetriggeredconcernabouttheconnection.

Construction—The fabricator reduced thedimensionof the top twolinks(seeFigure 1.5)inordertoeasethedifficultiesofconstruction.

Thecontractorhadbeenusingtheguttertopilebricksandsand.Thisloadexceededthedesignloadandfailureoccurred.

Comment — Although the edge beam had been designed and detailedto resist the full load from thegutter (moment, shear, and torsion), thishadnotbeencarriedthroughtothejoint.Thejointreliedonthetensionstrengthoftheconcretewhichwasnotsufficient.Thiswasaseriousdesignerror,butitisunlikelythatthefailurewouldhaveoccurredifthegutterhadbeensubjecttojusttheloadfromrain.Ifthedetailingandconstruc-tionhadbeencarriedoutthoroughlyandcorrectlyitisunlikelythatthecollapsewouldhaveoccurred.

The collapse resulted from combined errors in design, detailing, and construction.

1.2 CONCRETE TRUSS

Aprojectteamrequestedasecondopinionaboutthedesignofthebottomboom of a 19 m span reinforced concrete truss. The project team con-sidered that the tensionwould causeunsightly cracking.Adrawing (seeFigure 1.6showingthedetailofpartofthetruss)andtheanalysisofthetrusswereprovided.Muchattentionhadbeengiventothisdesignandspe-cialcarehadbeentakentoensurethatthestressesintheconcretewerelow.

The requestwas to check thedesignof thebottomboom.This checkwascarriedoutanditbecameapparentthattheshearstrengthoftheendverticalpostconnecting the topandbottombooms (seeFigure 1.7)wasinadequate.Thesectiondimensionswereb=230mmandd=460mm.

fy =460MPaAppliedshearforce,VE =488kNDesigntoBS81101(Cl3.4.5.2),fcu =25MPa

Maximumconcretestrutshearcapacity,Vmax=230×460×0.8×√25/1000 =423kN(notOK)

DesigntoBSEN1992(EC2)2(Cl6.2.3):fcd =13.33MPaαcw =1


Figure 1.6 Detail of part of truss.

Compression

Tension

Shear failure

Figure 1.7 Shear in vertical post at end of truss.


z =0.9dν1 =ν=0.552cotθ=1

Maximumshearcapacity,VRd,max=1×230×0.9×460×0.552×13.33/(2×1000)=350kN(notOK)

Itwasalsoapparentthatthedetailingrequiredsomechanges.Figure 1.6showsthemainT32bars,drawnasiftheycouldbebentwithsharpcornerswheninfacttheminimuminnerradiusofbendrequired(3.5×diameterofbar)is112mm.Whendrawntoscale,itbecameclearthattheconcretewouldfailintwoplacesasshowninFigure 1.8.

Failureofconcreteat support (1 inFigure 1.8)—Thesupportwidthwasonly150mmandthedrawingdidnotshowanyreinforcementintheconnectiontothetruss.Ifreinforcementwasintendedinthejoint,thenitwouldbedifficulttofitthisbetweenthereinforcementofthetruss.Ifnot,theeffectofthecoverandtheradiusofbendoftheT32barsatthecornerofthetrusswouldleadtospallingoftheconcreteandthesupportwouldfail.

Failureofconcreteatinnercorner(2inFigure 1.8)—Sincethebottomboomwasatensionmember,thetopreinforcementinitwouldbeinten-sion.Thewaythatitwasdetailedmeantthatacrackintheconcretewoulddevelopontheinsideofthebendwhereitjoinedtheverticalpostattheendofthetruss.

Section through bottom boom

230

600

1. Spalling of concrete

Compression

Tension

2. Cracking of concrete

Figure 1.8 Concrete failure points.


Comment—Theseproblemswerenoticedwhilstcheckingsomethingquitedifferent.Forthedesignstrengthofconcrete,theshearreinforcementwasinsufficientandtheconcretestrutstrengthinadequate.Ifthecubestrengthhadbeenincreasedfrom25to35MPa,thesectionwouldhavebeenstrongenough.

Theshapeofthetrusswasinherentlyunsuitable;thecentrelinesoftherafterandthetiedidnotintersectoverthesupport.Itmayhaveledtodif-ficultdetailingiftheyhaddoneso,buttherewouldhavebeenagoodsolid‘lump’ofconcretetotaketheshear.

Ifthedetailhadbeendrawntoalargescalethedetailingproblemswouldhavebecomeobviousand simple changes to thedesign couldhavebeenmadetoavoidoverstressingoftheconcrete.

1.3 CIRCULAR RAMPS TO CAR PARK

The construction of this car park (see Figure 1.9) was nearly complete.Theparapet and slabof the circular rampswere supportedon two col-umnsdiametricallyoppositeeachother.Thetorsionandcantilevereffectsoftheloadscausedwidecracksinthesupportingcolumns.Althoughthesecolumns were damaged with shear cracks, it was considered that if thecantileverandtorsionaleffectsfromtherampcouldbereduced,thenthecolumnscouldberepairedwithoutriskthatthecrackswouldreturn.Thiswasachievedbyinsertingsteelcircularcolumnshalfwayaroundateverylevelthroughouttheheightofthestructure(seeFigure 1.10).Inordertoensurethatthesenewcolumnswouldtakethecorrectload,flatjackswereintroducedateachfloorlevel(seeFigure 1.11).

Figure 1.9 Car park with circular ramps.


Figure 1.10 Additional steel columns.

Figure 1.11 Flat jacks fitted under each new column.


Alltheflatjackswereconnecteduptoasinglepumpandpressuregauge.Epoxyresinwaspumpedintoeachoftheflatjacksatthecorrectpressureandallowedtoharden.Thisensuredthatthecorrectloadhadtransferredtothenewsteelcolumnsateachlevel.

Comment—Therehadbeenadesignerror from lackofunderstandingof the structural behaviour effects of the shear forces on the columns.Althoughtheexistingcolumnshadcrackedasaresultofashearfailure,inthiscaseitwasnotconsideredtobeultimatelimitstateandtheexist-ingcolumnswerestillcapableofsupportingareducedloadwithoutriskoffurtherdamage.Assoonastheerrorwasrecognisedbythedesigner,asimplesolutionwasprovidedandexecutedquicklywithoutcausingadelaytotheoverallprogramme.

1.4 TRANSFER BEAM WITH ECCENTRIC LOADING

Adesignerwasconsideringalargetransferbeamforpickingupaneccentricload.Thetorsiondesignwasleadingtoaverycomplicatedreinforcementdetail. It was discovered by a chance comment that the support for thetransferbeamwasonmasonryandthatnostepshadbeentakentotransferthetorsiontothesupport.Fortunately,thiserrorwasdiscoveredatanearlystageandthewholestructuraldesignwasrevised.

Comment—Thisisasimpleexampleinwhichadesignerhadnotcheckedhowtheflowofforceswastakenthroughthewholestructure.Itisunfor-tunatethatthisisnotanuncommonfailing.

1.5 EARLY THERMAL EFFECTS

The design of a five-storey car park incorporated long post-tensionedbeams.Theseformedpartofanunbracedconcreteframesupportedby600msquarecolumns;floor-to-floorheightwas2.7m.Thebeamsincludedsix15.6mspansthatwerecastinonepour.Theywerestressedat3daystothefullrequiredprestress.Thebeamswere575mmdeepand2000mmwide,withrebatesinthetopcornerstotake150mmdeephollowcoreunits.Thehollowcoreunitsspannedbetweenthebeamswhichwereat7.2mcentres.Figure 1.12showsthearrangement.

Theconstructionperiodforthecarparkwasextremelyshort(8months)and hence all the prestressing work was carried out under a very tightschedule.Figure 1.13showsatypicalbeamunderconstruction.

Unfortunately,5to6daysafterthetransferofprestressforthefirstsetofbeams,cracksappearedinmanypartsoftheframe.Thecrackingwas


In-situ concrete tie beam (providingmoment frame in transverse direction)

Post-tensioned in-situ concretespine beam (1800 wide × 575 deep)

75 mm concrete �oor screed

150 mm thick precast pre-tensioned concrete hollow-core �oor slabs (supportedon formwork and cast inwith the spine beam)

In-situ concrete columnsDucts for post-tensioning cables

7200

Tableformsystem

Openingsfor tiebars15600

Figure 1.12 Layout of structural members.

Figure 1.13 Typical beam under construction.


widespreadinboththecolumnsandbeamsandexceeded0.7mminplaces.Figure 1.14showsthedifferenttypesofcrackingthatoccurred.

Ittookalongtimetounderstandwhatcausedthecracking.Whilsttheprob-lemwasdebated,inordertoproceedwithconstructionwithminimumdelay,theprestressingwasalteredtoatwo-stageprocess—only50%appliedattrans-ferandtheremainingprestressappliedafter2weeks.Aftermuchdiscussion,itwasconcludedthatwhenearlythermaleffectswereincludedwiththeothershorteningeffects,thetotalshorteningwassufficienttocausethecracking.Atthattime(early1990s)itwasnotcommontoincludeearlythermaleffectsinthedesignsofconcreteframes(forreinforcedorprestressedconcrete).Thecal-culatedvalueofearlythermalmovementattheouterendsofasix-spanbeamwasover8mmwhich,whenaddedtotheelasticshorteningfromprestressof7mm,providedsufficientmovementtocausethecrackingthatoccurred.

Atthattime,thecodeofpracticestated‘unlessthelessersectiondimen-sionisgreaterthan600mmandthecementcontentisgreaterthan400kg/m3thereisnoneedtoconsidertheearlythermaleffect.’AlthoughCIRIA3Report 91 covering early thermal crack control in concrete, provided ameansofcalculatingtheeffectforsituationsofvariousrestraints,itdidnotindicatewhatvalueoftherestraintfactorshouldbetakenforsuchabeaminastructural frame.Furthermore it introducedamodification(‘fudge’)factor,Kandsuggestedthatthisbe0.5.

Therestrainttoshorteningofthebeamsbythecolumnswasnotgreatforthisproject.Theearlythermalmovement,includingframeaction,was8mmthatcomparedwiththefreemovementof10mm.However,itwouldhavebeenincorrecttobasethemovementonthefulltemperaturefallfrom

End elevation

(a) Column �exure (b) Column shear (c) Beam �exure

(d) Beam/column tearingSo�t

(e) Tension in transverse beamBeam 1

Beam 2So�t

Tension�is crack also

appeared intop of beam

Figure 1.14 Different types of cracking.


peaktemperaturesincethebeamstriedtoexpandwhileheatingup.Atypi-calcurveforoneofthebeams,givingthetemperatureriseandfallovertime,isshowninFigure 1.15.

From a layman’s view, as the concrete started to set, the temperaturerose.Duringthisperiod,theconcreteremainedsomewhatplastic.Thestiffcolumns prevented any expansion of the beams (they just bulged a bit).Whenthetemperaturestartedtofall,theconcretehadhardenedandthebeamshortened,causingthecracks.

Comment—Thisisanexamplewheretheeffectsofearlythermalmove-mentshouldhavebeencheckedbecauseofthelongcontinuousmulti-spans.Thebeamshappenedtobeprestressedbutthesameeffectsapplytorein-forcedconcretestructures.Typically,fora300mminternalfloor,anallow-anceof100×10–6 forearly thermalcontractionstrainshouldbemade.Thisprojectprovidedthefirstclearrecordofearlythermaleffectsinacon-creteframestructure.Theresultsemphasizedtheimportanceofincludingtheseeffectsindesign—notcommonpracticebefore1995.

Thisparticularcontractwascomplicatedbythefactthatitwasadesign–buildtype.Thedesignofthebeamswascarriedoutbyaspecialistsubcon-tractor,buttheresponsibilityfortheframedesignwaswiththestructuralengineeringconsultant.Thissplitresponsibilitycontributedtotheconfusionanddelayinresolvingtheproblems.Insuchcontractsit is essential to name a single designer or engineer who retains overall responsibility for the stabil-ity of the structure, the compatibility of the design, and details of the parts and components, even where some or all of the design including detailing of those parts and components are not carried out by this engineer.

ReferenceshouldbemadetoDesignofhybridconcretebuildings.4

25

20

15

Tem

p. °C

Abo

ve A

mbi

ent

10

5

1 2 3 4Time in Days

5 6 7

Figure 1.15 Typical early temperature rise and fall in concrete beam.


1.6 SECONDARY EFFECTS OF PRESTRESSING

Aconcreteframewithatwo-spanprestressedbeamhadbeendesignedforserviceabilitylimitstate(SLS).Whencheckingtheframefortheultimatelimitstate(ULS), thedesignerdiscoveredthatthemoments inducedintotheedgecolumnsexceededtheircapacity,850kNm.Figure 1.16showsthelayoutofthestructuralmodeloftheframe.ThemomentsfromtheULSanalysisare shown inFigure 1.17.Toreduce the transfermomentat theedge,thedesignerconsideredcreatinghingesbetweenthecolumnandslabasshowninFigure 1.18.Thisproposalwasnotfavouredbecause:

3 m

3 m

15.6 m 15.6 m

Figure 1.16 Layout of structural model of concrete frame.

1350 kNm 1730 kNm

Figure 1.17 Moments from ULS analysis.

Provide recess tocreate hinge e�ect

Figure 1.18 Proposal for providing hinges.


• Reinforcementmustbeplacedcentrally toallowahingeeffect tobecreated.

• Compression in the concrete section would automatically createmomenttransfer.

Thequestionraisedwaswhethersecondaryeffectshadbeen included intheULSanalysis.Toanswerthisquestion, it is importanttounderstandthemethodofanalysis.ForprestressedstructuresintheUK,itiscommontocarryouttheprimaryanalysisatSLS.Oftentheequivalentloadmethodisused.Forthistheprestressingeffectsaremodelledasanequivalentload(e.g.,auniformlydistributedloadmodelstheprestressingeffectofapara-bolicdrapeofthetendons).Thistypeofanalysistakesaccountofsecondaryeffects.Forthosenotfamiliarwiththistypeofwork,secondary(parasitic)effectscanbesummarisedbythediagramsshowninFigure 1.19.

AcheckisthencarriedoutatULS.Forthislimitstate,theprestressingactionisoftenconsideredasaffectingonlytheresistanceandnotaspartoftheloading.Thesecondaryforcesandmomentsmustbecalculatedsepa-ratelyandaddedtothoseoftheprimaryanalysis.Forthisparticularframe,thesecondarybendingmomentsareshowninFigure 1.20.Theseeffectshadnotbeen included in theULSanalysis check for thisproject.Whenaddedtotheexistingresults,themomentsbecameacceptableasshowninFigure 1.21.Onceitwasagreedthatthiswasthecorrectanalysis,itwasrealisedthatthecolumnwasnotoverstressedandthemomentsinthespanofthebeamshadincreased.Therewasplentyofroominthesectiontoaddmorereinforcementtotakeaccountofthis.

Unstressed element in structure

Unstressed isolated element

Stressed isolated element

Secondary forces for element

Figure 1.19 Example of prestress secondary effects.


Comment—ThisproblemwaspartlyduetothedesignmethodusedintheUK.Forprestressedbeams,theprimarydesign/analysisisoftencarriedoutattheSLS.Theprestressisconsideredaload(oractioneffect)withintheequivalentloadmethod.

ItisnotpossibletousethisapproachfortheULScheckastheprestress-ingtendonsresisttheappliedactionsinthecriticalpartsofthebeamandthus do not provide an equivalent load. The secondary effects must beconsidered separately at such parts. The debate about the interaction ofsecondaryeffectsandmomentredistributionisongoing.However,inthissituationtherewasnoreasontobelievethatmomentredistributionwouldorcouldtakeplacebeforedamageoccurredtothecolumn.

1.7 TEMPERATURE EFFECTS ON LONG-SPAN HYBRID STRUCTURE

Thegroundlevelofatwo-storeyundergroundcarparkslabwasnotcov-ered.Thestructureconsistedof16mspanninghollowcoreunitsbearingonprecastconcretebeamnibs.Movementjointshadbeenshownonthedrawingsbutthesedidnotfunctioncorrectlyforavarietyofreasons.Theuppersurfaceoftheslabwasexposedtotheweatherandinparticulartolargevariationsintemperature.Thelattercausedmovementandrotationoftheunitsandtheirsupports.Thisresultedinseverecrackingofthesup-portingnibsand in someplacescrackingat theendsof thehollowcoreunits. Even after repair, the cracks reappeared each subsequent year for

1350 kNm1730 kNm

737 kNm

Figure 1.21 Moments from ULS analysis adjusted for secondary effects.

Structural Concrete Failures

613 kNm 270 kNm

Figure 1.20 Secondary effects for frame.


more than5years. Figure 1.22describesdifferentmechanisms that canoccurevenwhereastructuraltoppinghasbeenused.

Ingeneral,ifthebearingmaterialcreateslargefrictionforces,thesecanleadto largetensionstresses inboththesupportandtheprecastslaborbeam(seeFigure 1.22a).Neoprenebearingsorsimilarshouldbeusedtoavoidthis.

Ifthespacebetweentheprecastslaborbeamandthefaceofthesupport-ingmemberisnotadequatefortherequiredmovementoritfillswithhardmaterialovertime,crackingwilloccur(seeFigure 1.22b).Iftheeffectsofmovementand/orrotationcausethelineofactiontomovetooclosetotheedgeofthesupport,localspallingcanoccur(seeFigure 1.22c).

Afurtherproblemariseswherehollowcoreunitsareused.Anycrackingthatoccursclosetotheirendsislikelytocauseanchoragebondorshearfailure of the unit (see Figure 1.23). Anchorage bond failure may occurbecausethecrackingclosetothesupportdoesnotallowthefullanchorageresistancetodevelopandtheprestressingstrandsstarttoslip.Thiscausesthecracktogrowuntiltheunitfails(seeFigure 1.23).Hollowcoreunitsareinherentlyvulnerabletotheeffectsofcrackingclosetothesupportastheirshearresistancereliesonthetensionstrengthoftheconcrete.Unlikeasolidsection,acrosssectionavailableforshearresistanceoftheseelementsismuchreducedduetothepresenceofthecores.

(a) Effect of high friction (b) Effect of hard material in joint (c) Effect of rotation/Movement

Friction cancause cracking

Movement

Hard material canprevent movement

Rotation Rotation

Rotation cancause spalling

Figure 1.22 Examples of potential failures at movement joints.

(a) Anchorage bond failure (b) Shear tension failure

Anchorageslip

Shear tension crack

Large crackclose tosupport

Figure 1.23 Types of end failure of hollow core units.


Comment—Theremedialworktocorrectthissortofproblemislikelytobeveryexpensive.Makinggoodthecracks isnotasatisfactorysolutionasthecracksreappearonanannualcycle.Eventuallythedangeroffallingconcretetousersofthecarparkresultsintemporaryworksthatmaybesoextensive thata rebuildof thewhole structurecanbecomeasensiblesolution.

Whereprecastconcreteelementsformamajorpartofastructure,care-fulcalculationoftolerancesisessential.Thiswasparticularlysoforthisprojectasthedesignofthemovement jointsdidnottaketolerances intoaccountsufficiently.Thisaspectofdesignisnotsocriticalforinsitucon-cretestructuresasmanyofthetolerancesgetabsorbedsatisfactorilybytheconstructionprocess.

Thisprojectwasanotherexampleshowingit is essential to designate a single designer or engineer who retains overall responsibility for the sta-bility of the structure and the compatibility of the design and details of the parts and components, even where some or all of the design including detailing of those parts and components may not be carried out by this engineer.Referenceshouldbemade toDesignofhybridconcretebuild-ings4 and the Concrete Society’s report titled Movement, restraint, andcrackinginconcretestructures.5

1.8 LOADING FOR FLAT SLAB ANALYSIS

Aflatslabspansareasbetweencolumnsandfailurecanoccurbytheformationofhingesalongthelinesofmaximumhoggingandsaggingmoments.ThiscanbemosteasilypresentedusingthefoldedplatetheoryasshowninFigure 1.24.Acomplementarysetofyieldlinescanformintheorthogonaldirection.

Column supports

Hoggingyield lines

Saggingyield lines

Figure 1.24 Simple yield line mechanism for flat slab.


Onemisconceptionofsomeengineers is toconsiderareducedloadingwhenanalysingaflatslab.Eachmomentappliedineachorthogonaldirec-tionmustsustainthetotalloadingtomaintainoverallequilibrium.Thereisnosharingoftheloadbypartialresistanceineachdirection.

Comment—Oneengineer,whohadworkedintheU.S.,foundthatthiserroneousbeliefwentbacktotheearlydaysofflatslabconstruction,whenthe promoters measured the strains in the reinforcement of newly con-structed slabs. They ignored the contribution of the concrete in tensionandpromulgatedtheidea.Theengineerhad,infact,designedsomeware-houseslabsbythatmethod.Manyyearslater,anewowneraskedwhetherthe imposedfloor loading couldbe increased, andoneofhis colleaguescheckeditandfoundthatitwasoverloadedunderthedeadload.

Althoughtheauthorcannotquoteotherinstanceswherethisoccurredin designs, there have been many comments by engineers to this effect.It is clear thatmanyengineersbelieve that it is reasonable toconsiderareducedloading.ReferenceshouldbemadetotheConcreteSociety’stech-nicalreporttitledGuidetothedesignofreinforcedconcreteflatslabs.6

1.9 PRECAST CONCRETE CAR PARK

Thelayoutofthecarparkdeckincludedprecastspinebeamssupportingdoubleteeunits.Figure 1.25showsasectionthroughthedeckclosetotheedgeofthestructure.Astructuralscreedwasplacedoverthewholedeck.Themethodof construction for the cantileverunitswasunconventionalandincludedon-sitewelding.Figure 1.26showsanenlargeddetailthroughthesection.

Thestructuralscreedlaidoverthewholedeckwasdesignedwithastepovertheweldedplateabovethespinebeam.Thefinalscreed,incorporatingawaterproofmembrane,waslaidtofalls.Thedrainagepointsweresitu-atedattheendsofthespinebeams.Figure 1.27showsthefinalarrange-mentthroughasection.Thestepinthescreedcreatedacrackinducerforthetensionstressesatthetopofthecantileverandwaterdrainingoffthe

Structural screed

Double teecantilever

Spinebeam Double tee

Figure 1.25 Section edge of car park deck.


carparkwas likelytopermeate it. Inwinter, thiswaterwas likelytobecombinedwithde-icingsalts.Theriskofcorrosionofanimportantpartofthestructurewashigh.Bythetimethisproblemwasrealized,itwascon-sideredtoolatetochangethedesignandformofconstruction.Insteadthedepthofthewaterproofingmembranewasincreased.

Comment—Thedesignincludedfourmajorerrors:

1.Site welding for the main cantilever reinforcement is not recom-mended.Whencheckedonsite,someoftheweldswerefoundtobeofverypoorstandardandhadtobecondemned.

Site weld

Structural screed

Plan of Plate

720 × 100 × 10 thk plt.

Figure 1.26 Detail through section.

Salty water

Crack inducer

Salty water

Figure 1.27 Section through completed deck.


2.Thedesign includedfilling the spacesbetween the spinebeamanddoubleteeunitswithmortar.Thiswasverydifficulttoachieveandresultedinpoorcompactionthatallowedwaterpassage.Thedesignshouldhaveprovidedabettersolution.

3.Thesharpcornerinthestructuralscreedformedacrackinducer. 4.There was a high probability that salt water would penetrate the

structureandcausecorrosionofthereinforcement.

Ifthemembranewasasphaltorsimilar(madetocarrywheel loads)andfullybondedtothescreed,itwouldwithtimebecomesobrittlethatthecrackinthescreedwouldpropagateintothemembrane.

1.10 ARCH FLOOR

Theconstructionofahotelwasnearingcompletion.Thefirst levelfloorspanning10mhadbeendesignedasaninsituconcreteflatarch.Thedesignkeptthefloorthicknessatmidspantoaminimum(100mm)increasingtowardtheendsto275mmthickatthesupport.Theformworkforpartsofthefloorhadnotbeensupportedsufficientlyanddeflectedtocausethefloorthicknesstobe75mmthickerthandesigned.Theslabwassupportedonmasonrywalls.Somemonthsafterconstruction,crackswerenoticedinthefloorandwereconsideredtohavebeencausedbyshrinkage.Theclient,forotherreasons,complainedabouttheconstructionandanindependentengineerwasaskedtogiveasecondopinion.Duringthisinspection,itwasdiscoveredthatnodesigntiestotheflatarcheshadbeenprovided,althoughnosignificantmovementhadoccurredatthesupports.

AnumberofremedieswereconsideredandtheonethatappearedtobemostsuitablewastoinsertMacalloybarsandanchorthemtotheoutsidesofthesupportingmasonrywalls.Specialfireprotectionfromboxesmadeoffire-resistantmaterialwasrequired.

Comment — It was by pure chance that the structural integrity of thebuildingwascalledintoquestionanditisquitepossiblethatitwouldnothavefailedorcollapsedformanyyears,ifever!

1.11 PRECAST CONCRETE STAIRFLIGHTS

The1968collapseofastaircaseduringconstruction,killingtwopeople,causedarethinkindesign,detailing,andconstruction.Theprecaststair-flightsweredesignedwithhalfjointstositoninsituconcretelandings.Oneoftheupperfloorstairflightshadbeenplacedinpositionbyacranebutitneededtobeshiftedintoitscorrectposition.Temporary‘acro’propswere


usedfromtheflightbelowtosupportandliftitasitwasleveredintoitsfinalposition.Thestairflightbelowwassupportedonaninsituconcretenibthathadbeencastonlyafewdaysearlier.Thisnibdidnothaveitsfulldesign strength and was loaded with almost twice its design load whenfailureoccurred.Thethicknessofthelandingslabwasonly200mmwhichmeantthatthedepthofthenibs,100mm,didnotallowsufficientroomforreinforcement.

Sincethattime,precastmanufacturershavedevelopedavarietyofdiffer-enttypesofjointsusingprecaststairflights.Whenusingsuchaproprietarysystem,itisessentialthatadesignerconsider:

• Themethodofadequately tying the stairflight toadjacentpartsofthestructure

• Thesequenceofconstruction• Thetemporaryworksinvolved• Thechainofresponsibilityinachievingthefinalstructure(Oftenthe

temporaryloadsduetopropsandotherfactorscanprovidethecriti-caldesigncondition.)

Manycurrentsystemsdonotincludeadequatetiesbetweenthestairflightandtheadjacentstructure.Thefollowingexamplesshowhowtyingrein-forcementcanbeprovided.

Half-joints—Figure 1.28showsatypicallayoutofreinforcementwherehalfjointsareused.Thetyingreinforcementthatprojectsfromtheprecastunitsprovidescontinuitywiththereinforcementinthestructuralscreed.

Doweljoints—Toprovidesufficientroomforadowelhole,thedimen-sionsofthenibmustbeincreasedtothoseshowninFigure 1.29.Figure 1.30showsapreferredarrangementofreinforcementforadowelconnection.

375

375

15

15

Tie reinforcement instructural screed

Tie reinforcement instructural screed

Figure 1.28 Typical layout of reinforcement for half joints.


Proprietarysystemusingsteelangles—Figure 1.31showshowapropri-etarysystemcanbeadaptedtoallowcontinuityofties.

Comment—Thisfailurehasbeenincludedwithinthischapteralthoughitisclearthatthedetailingandconstructionfaultswereimportantcontribu-tors.Greatcareisrequiredwhenconstructingstaircaseswithprecaststair-flights.Theimportanceofthiscannotbeoveremphasised.Ahybridform

120

120

35

70

40

Figure 1.29 Half-joint with dowel.

15 375

375 15

Figure 1.30 Reinforcement arrangement for dowel connection.

Structural screedStructural screed

Structuralreinforcement

Structuralreinforcement

PrecastStairflight

PrecastStairflight

Reinforcement welded toangle & lapping with

structural reinforcement

Reinforcement welded to angle& lapping with structural reinforcement

Insitu structure Insitu structure

Figure 1.31 Stairflight system using steel angles.


of construction is commonly used and the dangers get easily forgotten.Although safe construction largelydependsonworkmanship,adesignerhas an important obligation to provide a robust and buildable solution.ReferenceshouldbemadetoDesignofhybridconcretebuildings.4

1.12 SHEAR STUDS ON STEEL COLUMN TO SUPPORT CONCRETE SLAB

Thedesigner’sproposalwastouseshearstudsweldedtoasteelcolumntotransfertheloadfromtheconcreteflatslabtothecolumn(seeFigure 1.32).However,withthisconfiguration,apunchingshearfailurecouldoccurwithonlythebottomrowofstudsprovidinganyshearresistance.Alltheothershearstudswouldremainintheconeofconcreteattachedtothecolumn(seeFigure 1.33).Toavoidsuchafailure,twooptionsprovidesensiblesolutions:

Option1:Providelinkstotransfertheloadbacktotopofslab—Links,inadditiontothoserequiredtoresistpunching,wouldberequiredtotrans-ferthefull loadtothetopoftheslab.Struts intheconcretewouldthentransfertheloadontotheshearstuds(seeFigure 1.34).Thisisunlikelytoprovidethemostpracticaloption.

Punching shearfailure

Figure 1.33 Punching failure of slab.

Steel column

Reinforced concrete slab

Figure 1.32 Junction between steel column and concrete slab.


Option2:Provideshearkey(insteadofstubs)atbottomofslab—Usingashearkeywithfullstrengthweldsatthebottomoftheslabwouldpro-videadequate resistance.Theshear linkswouldensureagainstapunch-ing failure and the resulting forceflowwould engage the shear key (seeFigure 1.35).Thesizeoftheshearkeywoulddependonthethicknessofslabandtheshearforce.Theupperlimittothissolutionwouldbethecom-pressionstrengthoftheconcreteatthecolumnface:

BS81101:Maximumcompressivestress=0.8√fcubutnotmorethan5MPaBSEN19922:Maximumcompressivestress=0.8×0.6(1–fck/250)MPa

Comment—Thisisanexamplewherethinkingstrutsandtiescanbehelpful.

Extra links to liftload to top of slab

Figure 1.34 Use of links to transfer load to shear studs.

Forceflow

Shear keys welded to column

Figure 1.35 Use of strut-and-tie model to transfer load to shear key.


1.13 PILED RAFT FOR TOWER BLOCK

Thedesignof the raftassumed that thewallsof the two-levelbasementcarparkwouldactwiththeraftoverthepilestotransmittheshearandbendingforcestotheouterpiles.Thewallsofthebasementhadalmostfullheightopenings,placedoneabovetheother,andcontainedonlynominalreinforcement.Thecombinedstrengthofthesewallsplusthe1.5mthickslabwasinadequatetotransmittheloads(seeFigure 1.36).

The mistake was discovered whilst the tower block was being con-structed.Theremedialworkrequiredanewrafttobeconstructedbeneaththeexistingone(seeFigure 1.37).Thenewdesignreliedonthecompositeactionofthenewandoldrafts.Alltheexistingsurfacesofconcretewerescabbledand further reinforcementwas laidbetween the existingpiles.Placing of concrete for the lower part of the new raft was carried outconventionally.

Toachievegoodbondwiththebottomoftheexistingraft,theupperpartofthenewraftwaspackedwithsinglesizedaggregateandthengroutedwitharetardedandfluidcementpaste.Thegroutwas introducedunderpressure to thebackof thepour throughacomplicated systemofmetalpipespinnedtotheundersideoftheexistingraft.Themethodproducedawallofgroutthatextendedfromtoptobottomofthepourandflowedforward toward the peripheral shutters with the top surface behind thebottom.Pipesweresoplacedtolettheairoutinfrontofthegroutsurface,thenindicatewhereandwhenthegroutarrived,andthenallowgroutingtocontinuefromimmediatelybehindtheadvancingwallofgrout.Groutingwascontinuousuntiltheworkwascomplete.

Columns Core walls

Piled raft

Possible line ofshear failure

Basement car parks

Figure 1.36 Arrangement of piled raft and basement car parks.


Comment—Thiserrorwasdiscoveredasafollow-uptoacheckofsomeofthereinforcementdrawingsofthetowerblockthatwerefoundtocontainmistakes.

1.14 FLOATING PONTOON FOR RESIDENTIAL BUILDING

Thedesignandconstructionofafloatingpontoon tocarrya residentialblockwerecomplete.Thepontoonwasafloatinthemarina,readyforworkto start on thebuilding.As the load increased, it becameapparent thatthepontoonwasnotsufficientlybuoyant.Thedesigncheckthatfollowedshowedthatthedensityofconcreteassumedwastoolow.Thiserrorwascompoundedbyerrorsinconstruction(oversizingofthewallsandbaseofthepontoon),allofwhichresultedinaseriousreductioninbuoyancy.Thesituationwasresolvedbyattachinglargeblocksofpolystyrenetothecon-cretestructureofthepontoon.

Comment—During thedesignandconstructionof thepontoon, insuf-ficientthoughthadbeengiventotheessentialrequirementofensuringthatthestructurewasadequatelybuoyant.

1.15 PRECAST COLUMN JOINT DETAIL

Aprecastslopingcolumn,400mmindiameter,wasdesignedtotaketheloadfromseveralfloorsabove(seeFigure 1.38).Ashortsteelstubwascast

Existing raft

Piles scabbled totake new concrete

New supplementaryraft to take shear

3.5 m

Figure 1.37 Schematic arrangement of new raft.


intothebottomoftheprecastelementtosimplifytheconnectiondetailtothebeambelow.Duringconstruction,itwasrealisedthatthedesignofthiscon-nectionwasunsafeduetoinsufficientbondcapacityandanti-burstingsteel.

Theshortstubconsistedofa254×254UCsectionthatextendedintothecolumnabout300mm.Thestubhadfour19×100shearstudsweldedtoeachfaceoftheweb(seeFigure 1.38b).ThelinkssurroundingthestubcolumnwereT12hoopsat250mmpitch.Nootheranti-burstingreinforce-menthadbeenprovided.

Thecolumnswerealreadyconstructedandsupportedfourfloors.Checkcalculationsshowedthatthefactorofsafetyagainstcollapsewasabout1.1andthiswouldreducetolessthan1whenthenextfloorwasadded4dayslater.Workwasstoppedinthatpartoftheprojectandtemporarypropsinstalledintheareatopreventpossiblecollapse.

Theremedialactionwastoencasetheexistingcolumnwithasteelcir-cularhollowsection,457mmdiameter,ofthesamelength.Thiswascutintotwohalvesalongitslengthandthenweldedinplacewithfullstrengthweldsalongthefull length.Thespacebetweenthesteelandprecastcol-umnswasthengroutedup(seeFigure 1.39).

Nocalculationshadbeencarriedouttodeterminetherequiredembed-ment lengthof theUCstubcolumn.Bondandendbearing shouldhavebeenconsidered.

(a) Elevation of column

Precast column

(b) Bottom detail of column

T12 hoops @ 250

300

(c) Section through column

203 × 203 UC

19 × 100shear studs

Figure 1.38 Sloping precast column.


Bond—If the forcewas tobe transferredusingbondstress,adesignshearstrengthof0.6MPa independentofconcretestrengthshouldhavebeenused.ThisisthevaluegiveninBSEN1994-1-17forthedesignshearstrengthdue tobondandfrictionforacompletelyconcrete-encasedsec-tion.Thedesignvalueof studsand shear connectorsmayalsobe takenfromthatstandard.TheconcretearoundshearconnectorsfixedtothewebofHsectionsispartiallyconstrainedbytheflanges,andthiscanresultinhigherdesignresistances.

Endbearing—Someofthecolumnforceistakeninendbearing.Thebearingstressshouldbelimitedtothreetimesthedesignconcretecompres-sivestrength,e.g.,3×0.85fck/1.5=1.7fckfordesignstoBSEN1992-1-1.2

Links—Linksmustbeprovidedovertheembeddedlengthtopreventsplitting.The force tobe resistedby the links (one leg) in aunit lengthshouldbeequaltotheshearforcetransferredinthatlengthdividedby2π.Thelinksabovethestudmustresistaforceequaltotheendbearingforcedividedby2π.

Comment—Itwasafortunatechancethatthisfaultwasfound.Theengi-neerwhodiscovereditwasintendingtocopythedetailforasimilarnearbystructure.

Fullstrength

welds

CHS

Grout

Figure 1.39 Remedial work.

31

Chapter 2

Problems and Failures due to Errors in Structural Modelling

2.1 REINFORCED CONCRETE TRANSFER TRUSS

Amulti-storeybuildingconstructedaccordingtoadesign–buildcontractincluded adeep transfer reinforced concrete truss.The client required asecondopiniononthedesignofthisfromanindependentconsultingengi-neer.Thereportofthisengineercondemnedthedesignasunsafe.Thecon-tractorthenemployedanotherconsultingengineertocarryoutafurthercheckand,ifitconfirmedthefindingsofthereport,tofindasolutionthatprovidedtheleastdisruptiontotheexistingworks.

Thesecondcheckrevealed that the initialdesignof theconcrete trussinvolvedtheuseofacomplicatedfiniteelementprogram.Theendsupporttothetrusswasastiffwallandthishadbeenmodelledwithplateelementsthathadnolateralstiffness(seeFigure 2.1).Thesecondconsultingengi-neerthencarriedoutseparateframeanalyseswithsimplerprogramsthatincludedmorerealisticpropertiesfortheedgewall.Theresultsshowedthatthedesignofthetrusseswassafe—just.Thereinforcementdetailsofthenodesrequiredsomeminorchanges.

Comment—Thisexampleshowsthattheoriginaldesignerhadnotunder-stoodtheeffectoftheendwallstiffnessordidnotunderstandthelimita-tionsofthemodellingsoftware.Bothcasesreflectedsomeincompetence.Thisisanexampleinwhichthetwocheckingengineersactedforpartieswithoppositeviewsaboutthesafetyofthestructure.Itshowshowvulner-ableengineerscanbeintryingtoprovewhattheyareaskedtodoratherthanexaminingtheinformationobjectively.

2.2 MODELLING RIGID LINKS

Itispossibleinsomestructuralframeprogramstosetuprigidlinksinthedata.Theseallowtheuser tospecify thatpartof thestructure thatwillbehaverigidlybetweenstatednodes.Arigidlinkmayincludeanynumber


ofelementsfromasinglepairtoalargegrouprepresentingastiffpartofthestructure;forexample,allnodesonaparticularplanemaybelinkedtorepresentthein-planestiffnessofafloorslabwithoutexplicitlymodel-lingit.Itisoftenpossibletolinkonlysomeofthedegreesoffreedom;forexample,withthefloorslab,theout-of-planetranslationandthein-planerotationsmustnotbelinked.

Forthedesignofoneparticularbuilding,rigidlinkswereusedtomodelfloorplates.Atalevelwherethewallarrangementwithinindividualcoreschanged significantly, the rigid links had to translate very large forcesthrough the floor slabs (transferring forces from one core structure toanother). The program did not provide any output of the forces withinthe rigid linksand thedesignerassumed that loadscouldbe transferredthroughthefloorssafely.Duringadesignreview,itwaspointedoutthatthese forces couldbevery large. In fact, itwas thendiscovered that thedesignforcesexceededthedesignresistanceofthematerialsandaredesignhadtobecarriedoutduringconstructionofthebuilding.

Comment—Thisexamplehighlightstheneedtoknowwhatlevelofinfor-mationisrequiredandwhatlevelofdetailwillbeprovidedbyastructuralmodel.Inthiscase,thedesignerhadnotgiventhoughttotheproblemandthestructuralmodellingsystemdidnotsupplyadequateinformation.

2.3 ASSESSING MODEL LIMITS AND LIMITATIONS

Most analyticalmodels for reinforced concrete, including those forulti-matelimitstate(ULS),arebasedontheelasticpropertiesoftheconcretesection,uncrackedandwithoutreinforcement,i.e.,ahomogeneousmate-rial.Thisallowsa simpleanalysisofa frameor structurewithuniformstiffnessforeachelement,andformostsituationsprovidesanacceptable

RC transfer trussStiff edge wall - analysis assumed aplate element without lateral stiffness

Figure 2.1 Diagrammatic view of concrete truss.

Problems and Failures due to Errors in Structural Modelling 33

solutionfordesign.Thereasonthisisnormallyacceptablecanbeexplainedbyconsideringasimpleindeterminatestructure—aproppedcantileverslabwithUDL.Figure 2.2showstheflexuralresultsofanalysisforseveraldif-ferentsituations.

Thefirstanalysisassumessimpleconcretesectionproperties—ahomoge-neoussectionthroughout.ThisgivesamaximumsupportmomentofWL/8andamaximumspanmomentofWL/14.Theengineer thendesigns thereinforcementforthesemomentsand,ofcourse,thereinforcementrequiredatthesupportisnearlytwicethatrequiredinthespan.Theeffectofinclud-ing this reinforcement (givinggross sectionproperties) is to increase thestiffnessoftheslabnearthesupport.

Ifafurtheranalysisisthencarriedoutwiththerevisedstiffnessproper-ties (notnormallydone), thebendingmoment isaltered to that showingGrossSection(1).Thiswouldcausetheengineertoincreasethereinforce-mentatthesupportandreduceitinthespan,causingafurtherincreaseofstiffnessatthesupportandreductioninthespan.ThenextanalysiswouldresultinmomentsshownasGrossSection(2).Thisiterativeprocesswouldleadtheengineertodesigntheslabasacantilever.However,inreality,theslabwillcrackundertheultimateloads.Thisreducesthestiffnessatthecrackedsections.Acrackedanalysisofthesectionwiththereinforcementrequired by the analysis using simple concrete section properties wouldresultinmomentsshownasCrackedSectionwithTensionStiffening.Thisresultsinasimilarcurvetothatshownfortheanalysisusingsimplecon-cretesectionproperties

If,asaresultofthefirstanalysiswithconcretesectionproperties, theneutralaxisdepthatacriticalsectionisgreaterthanthatforabalancedsection(i.e.,thereinforcementdoesnotyieldattheultimateload),andthe

Cracked sectionwith tension

stiffening

WL/8

WL/14

Gross section (2)

Gross section (1)

Concrete section

W

L

Figure 2.2 Design analysis of propped cantilever.


appliedmomentislessthanthecrackingmoment,thereisapossibilityofaprematurebrittlefailure.Boththeseconditionsshouldbecheckedtoensurethatthecriticalsectionsarecapableofmomentredistribution.Otherwisethedesignmaybeunsafe.

Many reinforced concrete continuous beam and slab and frame pro-gramsmodeltheelementswithhomogeneouspropertiesandtheengineerinputsthesewithjusttheconcretesectionwithoutreinforcement.Itisthusimportant that the software, after it designs the reinforcement, containschecks for neutral axis depth and cracking to ensure against prematurebrittlefailure.

Thisdemonstratesthatdesignisheavilyreliantonamaterial’scapacitytoredistributetheforcesachievedbytwoindependentmeans:(1)crackingoftheconcrete,and(2)yieldingofthereinforcement.Theeffectofthesecanoftenprovideaverydifferentultimateresistanceofastructurethanthatpredictedfromtheelasticresults.

Although elastic results are normally conservative, in some modellingsituationsthisisnotso.Atypicalexampleisintheuseofaplaneframeprogramforanalysingflatslabs.Themodellingsimplifiestheflatslabasthough it is a continuous slab spanning on continuous supports (walls).Thebendingmoments from this analysispeakover the supportsbutdonotvaryacrossthewidthoftheslab.Infact,thebendingmomentspeakovertheactualcolumnsupportssothetotalbendingmomentcalculatedforthefullwidthofslabshouldbedistributedacrossthelineofsupport,unevenlypeakingoverthesupport.Thecodeprovidesrulestocompensateforthiseffect(usingcolumnandmiddlestrips).Anincorrectdistributionofmomentswillalsocauseincorrectdistributionofshearforces.

2.4 EMPIRICAL METHODS

Thedevelopmentofgoodpracticeshasreliedonempiricalandmorerigor-ousanalyticalmethods.Manyempiricalmethodshavebeenhoned fromexperience to provide very efficient (cost effective) solutions. Sometimesthishascausedthesemethodstobelessconservativethanamorerigorousapproach.

Onereasonforthisisthatthematerialpropertiesoftheconcreteandreinforcementhavechangedsincetheempiricalmethodswereintroducedand the original assumptions for them are no longer valid. This hascausedsomeconcernanddiscussionamongstthecodewriters.Generallyitisfeltthattheempiricalsolutionsshouldhaveanoverallbuilt-insafetyfactorthat isgreaterthanthatformorerigorousmethods.Overrecentyearsthishasbeenimplementedforultimatelimitstatesbyeditingthecodeclauses.


2.5 INITIAL SIZING OF SLABS

During the past 50 years, the reduction of partial safety factors andincreases instrengthsofmaterialshaveresulted inthethicknessofslabsoften being controlled by serviceability limit state (SLS; deflection andcracking).However,determiningaccuratedeflectioninformationofrein-forcedconcreteslabsandbeamsatthedesignstageisalmostimpossible.Itrequiresknowledgeofthefollowingfactorsand/orproperties:

1.Materials used (type of aggregate, type of cement, and amount ofwater)

2.Weather conditions at timeof construction (hot or cold, humidordry)

3.Mechanical properties of hardened concrete (compression and ten-sionstrength,modulusofelasticity)

4.Mechanical properties of reinforcement and characteristics of itsbondwiththeconcrete

5.Actualloadingontheelement,notonlyatthepointintimerequired,butalsotheloadhistoryuptothatmoment

6.Effectofcontinuitywithadjacentstructuralelement 7.Exactconstructionprocessandsequence

Whereaccurateinformationexists,forexample,whencheckinganexistingstructure,itispossibletomakeanexcellentcalculationassessmentofthedeflection(certainlywithin5mm).

Incontrast,atthetimeofdesign,verylittleaccurateinformationexistsabout concrete properties or loading. Nevertheless, for the purposes ofdesign, codesofpracticemake reasonableassumptions forall theabovefactors and properties and provide simplified methods to aid the designengineer.Themostimportantoftheseisthespan/effectivedepthmethod.Howeveritisimportanttorealisethatthismethodcannotbeconsideredmorethanaroughguideandthosewhorelyonittoparethedesignthick-nessofaslabover7minspantoanaccuracyoflessthan25mm,withoutthebenefitofexperienceorspecificinformation,arelivingin‘cloudcuckooland’!Itisthuswisetoerrontheconservativesidewhenusingthespan–effectivedepthmethodtoassessthedesigndepthofaslab.

2.6 ANALYSIS OF FLAT SLABS WITH FINITE ELEMENT PROGRAMS

When using finite element programs to analyse and design flat slabs, itis important to realise that to reinforce for the peak moments over the


supports is normally considered overly conservative. Common practiceassumessomelateralredistributionofthepeakmoments.Figure 2.3showsthemomentprofileresultsacrossthefullwidthofatypicalslabatthesup-portfromthefiniteanalysis.Oftenthemomentsaroundthesupportnodeareinconsistent.Itisreasonablewithinhalfthecolumnstrip(thecolumnstripisdefinedashalfthetotalwidthofslab)toassumemeanmomentsacrossthiswidth.Thismaycausecrackingunderworkingconditionsandoneortwoofthereinforcingbarsoverthecolumnmayevenreachtheiryieldstress.Thisisnormallyconsideredacceptablebutadesignershouldbecarefultoensurethattheanalysisissatisfactoryfortheparticularsituation.

Thisisoneexampleinwhichadesignerreliesontheductilitypropertiesofaslab.Thisductilityresultsfrombothcrackingoftheslabandyieldingofthereinforcement.Thecrackingoftheslabcausesitsstiffnesstoreduceandhencetheredistributionofmoments.Theyieldingofthereinforcementcausesaplastichingewithsimilareffects.

Section

Plan

Middle strip Middle strip

Column strip, CC/2

C/4

M/2

Mean of node momentsacross half column

strip width

Inconsistent node momentacross face of column

Width of column

Moment of resistanceMoment diagram

Calculate mean ofthese nodes results

Column centrelineCentreline of panel Centreline of panel

Column strip

Figure 2.3 Bending moment results from typical finite element program.


ThechoiceofthestiffnessforaflatslabandsupportingcolumnsforULSanalysisdependsonengineeringjudgement.Takinghalftheuncrackedcon-cretesectionpropertiesfortheslabandthefulluncrackedconcreteprop-ertiesforthecolumnsisconsideredreasonableformostsituations.ThisisdiscussedinmoredetailintheConcreteSociety’sGuidetothedesignofreinforcedconcreteflatslabs.6

2.7 SCALE EFFECTS

Designengineerssometimesfindthattheyareinvolvedinverylargesizestructural projects (e.g., multi-storey blocks, heavy transfer slabs, deepraftsonlargepiles,etc.).Thereisatemptationtocontinuetousetherulesofthumbandempiricalmethodsthatarecommonforsmallandmediumsizedstructures.Oftentheseortheassumedsimplereinforcementdetailinglayoutarenotapplicableandmayleadtounsafedesigns.

Itisimportanttothinkmorefundamentallyhowtheforcesaretransmit-tedand ensure that reinforcement isprovidedand fully anchoredwheretensionforcesneedtoberesisted.Atypicalexamplewouldbeapilecapthathas3mdiameterpilesandrequiresathicknessover5m.Thestrutforcesmustbeconsideredcarefullyandthenodesreinforcedtoensuretheconcretecancontaintheforces.Thiscanleadtodesignreinforcementinadditiontothenormalbottombarsthatenclosethecaptoresisttheburst-ingforces.

Strut-and-tie methods are very useful in helping explain the flow offorcesbuttheyareoftendifficulttoapplyquantitatively.BSEN1992-1-12providessomehelpfulrulesforsuchapplications.Whendealingwithdeepsections,itisimportanttorememberthattheleverarmforcalculatingthetension in thebottomreinforcementmustnotexceed0.6 times thespan(maximumheightofthearch).

39

Chapter 3

Failures due to Inappropriate Extrapolation of Code of Practice Clauses

Duringthe1960sand1970s,therewasastronginclinationamongstsomeengineerstoextrapolateclausesinthecodesofpracticebeyondreasonablelimits. This may have been a result of the growing pressure to increasespans without reducing the depth of floors and without adding furthercoststodesigns.Rule-of-thumbmethodshadbeendevelopedandrefinedforfloorspansupto7mbutincreasingdemandforlongerspansmadeitalltooeasyforengineerstoextrapolatethecurrentmethodswithoutensuringthattheywerestillvalid.

Codesofpracticegiverulesthatcanatbestprovideapproximationsandwillworkinmost,butnotallcases;theydonotnecessarilygivefactualinformation.

3.1 COOLING TOWERS

InNovember1965,threeoutofeightcoolingtowerscollapsedingalesofover85mph(seeFigure 3.1).Eachtowerwas375feethighandtheyhadbeenconstructedclosertogetherthanusualandhadgreatershelldiametersandshellsurfaceareasthananyprevioustowers.Thehighwindswerecon-sideredtohavetriggeredthecollapse,butaninquiryfound theexactcausetobeanamalgamationofseveralotherfactorsinthetowerdesign:

• BritishStandardwindspeedshadnotbeenusedinthedesign;asaresult,designwindpressuresatthetopsofthetowerswere19%lowerthantheyshouldhavebeen.

• Basic wind speed was interpreted and used as the average over a1-minuteperiod,whereas, inreality thestructuresweresusceptibletomuchshortergusts.

• Thewindloadingwasbasedonexperimentsusingasingleisolatedtower.Thegroupingofthetowerscreatedturbulenceontheleewardtowers—theonesthatdidactuallycollapse.

• Safetymarginsdidnotcoveruncertaintiesinthewindloadings.


Based on the findings, wind loading in the initial design was seriouslyunderestimated.

ThecodeofpracticeusedatthattimewasCP1148.Itusedoverallsafetyfactorsanddidnotprovidefactoredloadswithspecialcombinations.Thecollapse occurred a month before the publication of a paper by Rowe,Cranston,andBestonNewconceptsinthedesignofstructuralconcrete.9Thisnewapproachtodesignwouldhaveensuredthatthedesignloadingcombinationwouldhaveresultedinasafestructure.

Figure 3.2showsasimplifiedviewoftheoveralldesignforces,whereWisthewindload;Gistheselfweightofthecoolingtower;hwistheheightofthecentreofthewindforce;andbisthewidthofthebaseofthecool-ingtower.Thetwodesignapproachesgiveverydifferentrequirementstoensurethatnooveralltensionforcesarecreatedinsuchstructures.

PermissiblestressCode(CP 114)approach:Appliedwindloadmoment=W×hw

Restoringgravitymoment =G×b/2G×b/2≥W×hw

Figure 3.1 View of collapsed cooling towers. (Courtesy of Gillian Whittle.)

Failures due to Inappropriate Extrapolation of Code of Practice Clauses 41

Newlimitstateapproachwithfactoredloads:Appliedwindloadmoment=1.4(W×hw)Restoringgravitymoment =1.0(G×b/2)

1.0(G×b/2)≥1.4(W×hw)

Comment—This incidentshouldreallybeexplained inachapterof itsownasthefailurewasaresultofaphilosophybuiltintothethencurrentcodeofpractice,CP114.Althoughthereweresomedefectsinwallthick-ness,themaincauseofthecollapsewasbecausethedesignvaluechosenforthewindloadwastoosmall.ThiscollapseensuredtheearlyadoptionofalimitstatecodeofpracticeintheUK,resultinginacompletelynewapproachtodesign.Thefirstdraftoftheunifiedcodeappearedin1968andin1972itwaspublishedasCP110.10Thiswasthefirstcomprehensivelimitstatecodeofpracticeeverpublished.

3.2 DESIGN BENDING MOMENTS

Itiscommonforengineerstousesimplifiedcoefficientstoobtainbendingmoments for continuous beams and slabs. Although it is generally con-servativetodesigncontinuousbeamswithpinsupports(i.e.,nomomenttransfer to the supports), itmaybeunsafe toassume thatnomoment istransferredtothecolumnsorendsupportsofthebeams.Asanexample,designers using BS 8110,1 Clause 3.4.2 and Table 3.5, often incorrectlyassumepinnedsupports.

The coefficients given in Table 3.5 of the standard are unsuitable forsituationswheremomentscanbetransferredtothesupports.Thisispar-ticularlysoforendsupports.Fortunately,thedetailingrulesrequiresomereinforcementtoresistnominalhoggingmomentsattheendsofbeams.Itisnotuncommonforengineerstoignoreanymomenttransfertosupportsevenwhentheadjacentspansdifferbymorethanthelimitationsgiveninthe codes (e.g., variations in span length should not exceed 15% of thelongest).Evenmasonrysupportsprovidesomemomentresistance.

W

b

hw

G

Figure 3.2 Equilibrium requirement.


Thedesignofaparticularconcreteframeincludedatwo-spanbeamforwhichonespanwastwicethelengthoftheother.Theanalysisforverti-calloadsassumedthatmomentswerenottakenintotheinternalcolumn.Thiswasquestionedatadesignreview.Acheckshowedthatthecolumnmoment capacity was not sufficient to take the moments caused by thebeamrotationatthatsupport.Itwasfortunatethatthedesignloadscouldbereducedsufficientlysothataredesignofthecolumnwasnotrequired.

Comment — Although it is not possible to cite particular failures fromthiscase, it isanotherexamplewhere theductilityof reinforcementandtheredundantcapacityofmoststructuresallowalternativeloadpathstopreventfailure.

3.3 PILES WITH HIGH STRENGTH REINFORCEMENT

Anengineerwashopingtoreducethenumberofpilesrequiredforabuild-ingbyprovidingreinforcementwithmuchhigherstrengththanrequiredfornormalhighyieldbars.Figure 3.3showsasectionthrougha225mmdiameterpilewithaproposedDywidagthreadedbar36mmindiameter(DywidagSystemsInternational,Munich).Theyieldstressfyofsuchabaris1200MPa.

Beforegoingintoproductionitwasdecidedtochecktheloadcarryingcapacitywithatestpile.Theresultwasthatthepilefailedatasimilarloadtothatofapilewithnormalreinforcementwithfykof460MPa.

After some thought, the reason for the unexpected low pile strengthbecameapparent.Theactualconcretecompressionstress–straindiagramissimilartothatgiveninPart2ofBS81101andFigure 3.2ofBSEN1992(EC2)2anddiffersinshapefromthatusedinULSdesigninthat,insteadofhavingaplateauofmaximumstress,itpeaksatabout0.002strain.Thisstrainiscloseinvaluetothatoftheyieldstrainofnormalreinforcement(500/200000=0.0025).Thismeansthatthemaximumresistanceoftheconcreteoccursatthesamestrainwhenthereinforcementreachesitsyieldstress(maximumresistance).

Dywidag threaded bar36 mm diameter

fy = 1200 MPa

225 mm dia. pile

Figure 3.3 Proposed pile with Dywidag bar as reinforcement.


TheyieldstressofaDywidagbarisabout1200MPawhichcorrespondstoayieldstrainof1200/200000=0.006.Thiscompareswiththeultimatestrainoftheconcreteof0.0035,sotheconcretewillfailinacompressiontestlongbeforetheDywidagbarreachesitsyieldstrainorstress.Figure 3.4showsthisgraphically.

Comment—Thisisanexamplewhereextrapolatingfromthecodeclauses(BS 8110, Table 3.1) can lead to unexpected results. It is important tounderstandthelimitsassumedinthecode,iftheyarelikelytobeexceededinaparticulardesign.

3.4 SHEAR CAPACITY OF DEEP SECTIONS

Thedesignofaprestressedbridgesectionfollowedacodeofpracticethatsomeconsideredtooverestimatetheshearcapacitiesofprestressedbeams.Soonafterconstruction,cracksappeared.Therewasalongdebateaboutthecauseofcrackingand,untilremedialworkwascarriedout,temporaryadditionalexternalshearreinforcementwasputinplace(seeFigure 3.5).

0.00350.002

Stress

Strain

Stress

Strain

Yield stress of normal high yield reinforcement

Yield stress of dywidag bars

Steel stress/strain curve

Concrete stress/strain curve

0.67fcu

0.8fcu

Actual curveULS design curve

Figure 3.4 Stress–strain curves for concrete and steel.


ThecodeinquestionhadsimilarclausestothoseofBS81101thatpro-vide an expression that increases the shear resistance according to howmuchmoment is required tonullify the effects ofprestress.The follow-ingexamplecomparingcalculationsinBS81101andBSEN1992(EC2)2showshowsuchananomalycanoccur.

Designinformation:ShearforceV=9000kNM0/M =0.4Prestress =0.8MPafpe/fpu =0.5d =2700mmb =800mmfck =30MPafcu =37MPafy =500MPa

Mainreinforcement:12 bars (32 mm diameter) to simulate both pre-stressingtendonsandreinforcement

Links:16mmdiameter(sixpersection)at300mmspacing

BS 81101:(Cl.3.4.5.4)vc=0.79{100As/(bd)(fcu/25)}1/3(400/d)1/4/1.25=0.55MPa(where(400/d)1/4≥1)

Vc=bdvc /1000 =1189kN(Cl.4.3.8.5)Vcr=(1–0.55fpe/fpu)bdvc+VM0/M=4462kN

Figure 3.5 Temporary repair work on bridge.


VcrandVcaretheconcreteshearresistanceswithoutshearreinforcementwithandwithoutprestress.Thelargedifference,theshearresistancefromprestress,3255kN,isquestionedbysome.

Theshearresistanceoftheshearreinforcementwas

Vs=(Asv/s)0.87fyd =4723kN

Thetotalshearresistanceisthus

Vt=Vcr+Vs =9185kNOK

EC22: The calculation of the concrete shear resistance without shearreinforcement,VRd,c, is similar to thatofVcr inBS81101, except that itincludestheeffectsofprestressbyfactoringtheaxialprestress.

(Cl.6.2.2)VRd,c=[0.12(1+√(200/d))(fck100As/bd)1/3+0.15σcp]bd/1000=1042kN

Thereductionfactorforthissectioncomparedwithasection200mmdeepis0.64.

Thecalculation for shear resistancewith shear reinforcementuses thevariabletrussmethod.Choosingcotθ=2forthisexampletheshearresis-tanceofthelinksis

(Cl.6.2.3)VRd,s=(Asv/s)zfywdcotθ/1000 =8497kNNotOK

andtheshearresistanceoftheconcretestrutis

VRd,max=αcwbzν1fcd/{1000(cotθ+tanθ)} =8540kNNotOK

ThetotalshearresistancepermittedbyBS81101isthesumoftheresis-tancesoftheconcreteandshearreinforcement.Noreductionisrequiredfor thedeepsection.This is incontrast tocalculation toEC22whicha)reducestheshearresistanceoftheconcretewithoutshearreinforcementbyalargefactorandb)onlypermitstheresistanceoftheshearreinforcementorconcretestrut.

Comment—Althoughthisdoesnotrepresentaveryunsafesituation, itdoesemphasizetheimportanceoftheincreasedreductionfactorfordeepsectionsinEC22.

Thisexamplealsodemonstratestheverydifferentapproachesofthetwocodesofpracticetoshearresistanceandespeciallyshearresistanceofpre-stressedsections.

47

Chapter 4

Failures due to Misuse of Code of Practice Clauses

4.1 FLAT SLAB AND TWO-WAY SLAB BEHAVIOUR

Aflatslabisdefinedasaplatesupportedonindividualcolumns.Atwo-wayslabisaslabsupportedbybeamsateachedge.TheUKcodesofpracticedifferentiatebetweenflatslabsandtwo-wayslabs.

Acarparkwasdesigned(basedonadesign–buildcontract)asacofferedslabsupportedbycolumns.Thecofferswereomittedaroundthecolumnssothatthesolidsectioninthisregionprovidedpunchingshearresistance.Thedesignerclearlyunderstoodthatthesystemwouldbehaveasaflatslabbutdecidedtousethesimplifiedappliedmomentcoefficientsforatwo-wayslab,ignoringanybeameffects.

Themaximummomentsforatwo-wayslab,takenfromTable 3.14ofBS8110,1are:

Hogging 0.031nl2

Sagging 0.024nl2

Themaximummomentsforaflatslab,takenfromTable 3.12ofBS8110are:

Hoggingandsaggingforinteriorspans 0.063Flor0.063nl2

Hoggingandsaggingforouterspan 0.086Flor0.086nl2

Thereinforcementdetailedwasthuslessthanhalfthatrequiredforaflatslab! For some reason, the contractor fitted only half the reinforcementdetailedforthesupports.Theslabfinishedupwithonlyaquarteroftherequiredreinforcementinsomeareas.

After10yearsofservice,crackswereappearingintheslabandadecisionhadtobemadetodeterminewhatremedialworkwasrequired.Althoughthe car park did not appear to be about to collapse, the remedial workrequiredtoensureitssafetywasconsideredtooextensiveandthestructurewasdemolished.


Comment—Thiswasaclearexampleofadesignerrorcompoundedbyfurthererrorsonsite.Onereasonthatthestructuredidnotcollapsewasthattheactualloadingwasmuchlessthanthedesignvalueof1.5kN/m2.Theactualloadonacarparkmaybeaslittleashalfthisvalue.Anotherreasonfortheapparentstrengthislikelytobemembraneactioneffectsnotincludedinthedesign.

Thisisanexampleofthebenefitsofanindeterminatestructure.Thereexistedalternativeloadpathstothatconsideredindesign,whichpreventedcollapseofthestructure.

4.2 RIBBED SLAB SUPPORTED ON BROAD BEAM

Several buildings of a university included ribbed floor construction (seeFigure 4.1).Theribbedslabwasdesignedasasimplysupportedelementspanningbetweenedgesofabroadbeam.Thecurtailmentofthelongitudi-nalreinforcementintheribswasinaccordancewiththecodeclausestatingthat,‘eachtensionbarshouldbeanchoredbyoneofthefollowing:…b)aneffectiveanchoragelengthequivalentto12timesthebarsizeplusd/2fromthefaceofthesupport….’Inthiscase,thecurtailmentwasmeasuredfromtheedgeofthebroadbeam.Topreinforcementintheslabwasprovidedintheformofastructuralfabric.

Thedepthofthebroadbeamwasthesameasthetroughslabandthedesignwronglyassumed that it couldbe consideredaone-way slabandthereforedidnotrequirelinks(seeFigure 4.2).

Inorderfortheedgeofthebeamtobeconsideredinthedesignasthesupportface,theverticalforcefromtheloadontheribbedslabhadtobe

Trough slab

Broad beam with samedepth as trough slab

Figure 4.1 Plan view of trough slab.

Failures due to Misuse of Code of Practice Clauses 49

transferredtothetopofthebeam.Verticalreinforcementshouldhavebeenprovidedforthis.Linksinthebroadbeam(inadditiontoanylinksneces-saryforshearresistance)couldhaveprovidedsufficientresistance.

Failureoftheribbedslaboccurredatthesupportwiththebroadbeam(seeFigure 4.3).Completecollapsewasavoidedbyimmediatetemporarypropping.Thesequenceofthefailurewaslikelytobe:

• Yieldingofthefabricreinforcementinthetopflangeoftheslab• Largeflexuralcracksopeningnearthesupport• Redistributionofmomenttransferredtensionstressestothebottom

reinforcement• Combinationofanchorageandshearfailure

Assumed width of beamSingle wayribbed slab

Figure 4.2 Section through broad beam.

Flexure cracks

Single way ribbed slab

Bond failure resultingin shear failure

Assumed beam

Figure 4.3 Section through ribbed slab at failure.


Comment—Failurecouldhavebeenavoidedif(1)thespanoftheribbedslabhadbeentakenasthecentre-to-centredistanceofthebroadbeams,or(2)linkshadbeenprovidedinthebroadbeam.Abetterdesignwouldhaveincludedbothfeatures.

4.3 CAR PARK COLUMNS

Figure 4.4showsthebeamsonthesidesoftheinternalcolumns.Theyaresteppedtoallowrampingbetweenlevels(asisoftenthecase).Theinternalcolumnsweremuchstifferthantheedgecolumnsandinconsequencethesupportmomentsandshearforceswithinthecolumnwerehighcomparedwiththeedgecolumns.Thecolumnshearforcescausedlargeshearcracks(upto2mm)withinmanyofthecolumns.

At first sight, this might be considered as an ultimate limit state.Howevertherewasstillaloadpathfortheverticalforcesthroughthecolumn and when the columns cracked, the moment in the adjacentbeamwasredistributedtothespan.Fortunately,thespanresistancewasadequate.

Althoughafailuremechanismwasnotpresent,aproblemoccurredfromrepairingthecracks.Assoonasthecarparkfilledwithvehiclesaftertherepair,thecracksreopenedandeventuallythedeteriorationofthecolumnconcretecouldhaveledtoacollapse.

Thedesignhadbeencarriedouttotheexistingcodeofpracticeofthetimewhichdidnothavespecificclausesforshearincolumns.Bythetimeremedialworkwascarriedout,newcodeclausesinplacerequiredshearchecksofcolumns.Thedesignofthisparticularjointwasinadequatetothenewclauses.Theinsurancecompanyrequiredremedialworktobecarriedoutsothatthebuildingwouldcomply.Thiswasconsideredtobesocostlythatthedecisionwasmadetodemolishtheexistingstructureandreplaceitwithanewdesign.

Edge columns Internal columnsShear cracks

in column

Figure 4.4 Structural elevations of beams and columns of car park.

Failures due to Misuse of Code of Practice Clauses 51

Comment—Althoughtheexistingdesigncodedidnotcoverthisparticu-larsituation,thedesignershouldhavebeenawareoftheproblemandtakenaction in thedesign topreventoverstressingwithin the joint.This is anexampleofcommonpracticeinthe1960sand1970swhenmanyengineersextrapolatedtheexistingclausesofthecodesofpracticeforsituationsout-sidethescopesofthecodes.

Typicallycontinuousbeamsweredesignedassumingmomentswerenottransferredtocolumns.Thisensuredsafedesignsforthebeamsbutmeantthatcolumnscouldbecomeoverstressed,asoccurredinthissituation.

53

Chapter 5

Problems and Failures due to Inadequate Assessment of Critical Force Paths

5.1 HEAVILY LOADED NIBS

Itispossiblethataheavilyloadednibmayrequiremorethanoneloadpathtotransferaloadsafely.Strut-and-tiemodelscandemonstratealternativemethodsforreinforcing.However,itshouldberealisedthattheleastdirectpathswillcausethemostdistortionandcracking,andshouldnotbeusedfortheserviceabilitystate.

Figure 5.1showsthemostdirectstrut-and-tie(primary)model.Theforcepathsareclosesttothatofanelasticmodelandwillcreatetheleastinter-naldistortiontoachieveequilibrium.Figure 5.2showsasecondarystrut-and-tiemodel.Thismaybeaccompaniedbydistortionandcrackingoftheconcretebeforeitcanachieveequilibrium.

Iftheforcesonthenibaretoogreatfortheprimarymodel,itisreason-able to superimpose the secondarymodel toprovide sufficient resistancefor the total ultimate loads.However, it is important to ensure that theprimary model is sufficient to resist the serviceability loads and providecrackcontrol.

Comment—Thisapproachwasusedforamajorviaduct.Thecombinationofthetwomodels(seeFigure 5.3)enabledallthereinforcementtofit—just!

5.2 SHEAR WALL WITH HOLES AND CORNER SUPPORTS

Amulti-storeyshearwallrequiredsomanyopenings(windows,doors,etc.)that the loadpathbecamevery complicated.Thedesigner assumed thattheloadwouldflowtothecornersandthentrackverticallydowntheedgeofthewall(seeFigure 5.4a).Infact,sincethewallwasbuiltinsituasahomogeneousstructure,straincompatibilitycausedtheloadtoflowbackintothefullwidthofthewall.Theresultwasthatseveralstoreysofload


weresupportedbyadeepbeamthattransferredtheloadtoitsendsupports(seeFigure 5.4b).

The limitingheightof thenaturalarchofadeepbeam(0.6×span)wasnot considered (seeFigure 5.4b)and this resulted in theomissionin thedesignofmuchof the reinforcementneeded for thebottomtie.Constructionhadreachedseveralfloorsupbythetimethemistakewasrecognisedandthisledtoaredesignofthewallduringconstructionandheavyremedialwork.Eachpartofthewallrequiredcarefulre-appraisal.This led to the requirementofmuchmore reinforcementat eachfloorlevel.Thebottomcornerreinforcementdetailsrequiredspecialattentiontoensurethatthejunctionbetweenthetieandcompressionstrutswasadequatelydesigned.

Figure 5.5 shows in a simplediagrammatic formhow the forcepathsautomatically flow out and back again. The assumed force path downtheedgeswouldnotrequiretiesattopandbottom,butwithoutthesetheactual forcepathwouldcause largecracks toopenupfromthe topandbottomsurfaces.Evenaftercracking,theanglestrutswouldstillexistandsowould theconsequentialhorizontal component.Without sufficient tieforcetoresist,thesupportjointwouldmoveoutwardandeventuallyfailurewouldfollow.

Figure 5.3 Example of use of the two strut-and-tie models. (Courtesy of Gill Brazier.)

Figure 5.1 Primary strut-and-tie model.

Cracking

Figure 5.2 Secondary strut-and-tie model.

Problems and Failures due to Inadequate Assessment of Critical Force Paths 55

Comment—Theconsequenceofmissingthissimpleprincipleofdeepbeambehaviourbeforeconstructionreachedsuchanadvancedstatemeantthatitrequiredtheredesignofthestructureandreprogrammingofconstructionwhichwereextremelycostly.

(a) Incorrect simple modelling (b) Correct simple modelling

Figure 5.4 Multi-storey shear wall.

Without tie rein-forcement large

cracks form

Assumedforce path

Actualforce path

Tie

Tie

Figure 5.5 Modelling deep beams.


5.3 DESIGN OF BOOT NIBS

Wherenibsareattachedtothebottomofabeamitisimportanttounder-standtheloadpathoftheforces.Figure 5.6showsatypicalsectionofsuchanib.

Theconventionalassumptionforashortcantileverofdcandzc(showninredinFigure5.6)isunsafeforsuchanib.Thedesigncompressionzoneforsuchamodelwouldbeclosetothebottomfaceofthebeamandlikelytofalloutsidethebeamreinforcement(boththelinksandmainreinforce-ment).Strut-and-tiemodellingishelpfultoexplainwhythisisso.Thestrut(showninredinFigure5.6)wouldjustcausethecovertothereinforcementtospalloff.Thestrutmustbesupportedmechanicallybythereinforcementofthesupportingbeam(showninblackinFigure5.6).Theeffectiveleverarmbecomesmuchsmallerandthetensionforceinthenibtopreinforce-mentmuchlargerthanassumedbytheshortcantileverapproach.

Itshouldalsobenotedthattheforceinthesupportinglinksofthebeam,Ft2d,islikelytobemuchgreaterthantheappliedloadonthenib,FEd,tosatisfyequilibrium.ForthesituationshowninFigure 5.6,itisconservativetoassumethecompressionactsatthecentroidofatriangularcompressionstressblock.Hencetheforceinthelink,inadditiontoanyshear,maybecalculatedasfollows:

Fc=FEd×ac/zb

Ft2d=Fc+FEd=FEd(1+ac/zb)

Comment—There are probablymanynibs of this type that have beendesignedincorrectlyandsurvivebecauseofbuilt-insafetyfactorsandthefactthattheloadassumedinthedesignhasnotoccurred.

Theerrordescribedinthiscasestudywasfoundinthedesignofanibforaveryprestigiousproject.Itwasveryfortunatethatitwasdiscoveredbeforeconstructionstarted.

db

FcFEd

Ft1d

Ft2dHEd

zn

ac

zcdc

zb

Figure 5.6 Nib attached to bottom of a beam.

57

Chapter 6

Problems and Failures due to Poor Detailing

Poordetailingisoftenconnectedwithalackofsufficientdesignthoughtandaccompaniedbypoorworkmanshipinconstruction.Thecombinationcanleadtostructuralfailure.Manyofthecasestudiesinthischapterhaveresultedfromextrapolationsfrompreviousjobs.Smallmodificationsweremadetosaveconstructiontimeandcost.Thiswasnotaccompaniedbysuf-ficientcheckstoensureasafestructure.

6.1 CONCRETE OFFSHORE PLATFORM

TheplatformincludedalargecellularconcretestructurebelowthethreetowersasshowninFigures 6.1and6.2.Duringconstruction,theplatformunderwentsubmergingfordeckmatingafterwhichtheplanwastoraiseitagainandtowittoitsfinalpositionintheoilfield.Itwasduringthesub-mergingpriortodeckmatingthatoneofthetri-cellsfailed.Thiscausedfloodingofthestructureandfurtheruncontrolledsinkingthat ledtoanimplosionofthestructureandcompletecollapse.

Figure 6.3showsadetailofthetri-cellwallthatwasdesignedtoresistthewaterpressurewhenthecellularbasewassubmerged.Figure 6.3bshowstheoriginalformofthecellswithcylindricallyshapedwalls.ThenaturalarchactionprovidedbythatgeometrywasnotpresentinthemodifiedformshowninFigure 6.3a.

Theanalysisofthecellstructurewascarriedoutusingafiniteelementsoftwarepackage.Theaccuracyoftheanalysiswasreducedasthearrange-mentofthequadrilateralelementsmeantthatthoseintheregionofthetri-cellcornersweredistortedfromtheidealsquareshape.Thisledtoerrorsintheresults.Itwasrealisedafterthefailurethattheshearstressresultsfromtheanalysiswereinerrorontheunsafeside.

ThecriticalshearsectionwasreinforcedwithT-headedbars.Thesoft-ware package gave the required areas of reinforcement, and the designrequiredthatthelengthoftheT-headedbarswouldextendacrossthefull


See detail

Figure 6.2 Plan section of cell structure.

Figure 6.1 Concrete offshore platform during construction. (Courtesy of Gillian Whittle.)

Problems and Failures due to Poor Detailing 59

widthofthesection.Astheyweredifficulttofixthroughtheouterlayerofreinforcement,itwasdecidedtoreducetheirlength(seeFigure 6.4).

Acrack formedatacornerof thecellandspreadto theendof theTbar.Thewaterpressurebecameactiveinthecrack,makingthesituationworse.Ashearcrackdevelopeduptothecompressionzoneofthesectionandthis failed inabrittlemanner.Theresultingmassive leak ledtothestructuresinkingfurtheruntiltheincreasedwaterpressurecausedprogres-sivefailureofthewholecellularstructure.Itfinishedupontheseabedasamassofrubble.

(a) As built (b) Original form

5800

800

550

Waterpressure

Figure 6.3 Detail of tri-cell.

Compressionfailure

Initial cracking

Water pressure

‘T’ headed baras required

‘T’ headed baras fixed

Figure 6.4 Section through tri-cell where failure occurred.


Comment—Thiscatastrophicfailurewastheresultofanumberoferrors:

• Theanalysisprogramwassetupwithafiniteelementmeshthatwastoocoarsetoprovideaccurateshearresults.

• TheT-headedbarsweretooshortandallowedtheshearresistancetobecomeunsafe.Thiswasprobablytheprimarycauseofthefailure.

• There was minimal checking of the design and detailing, possiblybecausethiswasatypeofstructurethatwaswellestablished.Inthesameperiod,thesamedesignerwasinvolvedinthreequitedifferentandmorecomplicatedplatforms.

• In previous designs, the geometry of tri-cells had been formed byintersectingcylinders.Althoughsomecrackingoccurred,thegeom-etryensuredthatsufficientarchingactiontookplacewithoutinduc-inglargeshearstresses.Thegeometryoftri-cellswasalteredonthisproject to make the formwork simpler to construct. Unfortunatelythenewformdidnotallowarchingactiontotakeplaceandthesharpcornersactedascrackinducers(thesewerealsopresentintheoriginaldesign).

• The rebuild retained the cylindrical geometry in the tri-cells andthe reinforcement was detailed to ensure mechanical linkage. TheT-headedbarswereextendedtotheouterreinforcement.

6.2 ASSEMBLY HALL ROOF

Thisdisastercouldalsobecalledthemiracleofthedecade.On13June,1973,lateintheevening,theroofofanassemblyhallcrashedtotheground.Inthewordsofthecaretaker,heheardaloudrumble,wenttoinvestigatebytorchlight,andfoundthewholeroofweighingmanytonshadcollapsed(see Figures 6.5 to 6.7). Twenty-four hours before this event, some fivehundredparentshadattendedameeting in thehallandthechairswerestillinplace.

The principal cause of the collapse was inadequate bearing for beamseatingsanddeteriorationofconcreteatbeamends.Thehallwasoneofthefirstbuildingsfoundtohavesufferedfromtheeffectsofhighaluminacement(HAC;seeSection7.4).

Comment—Thiswasanexampleofinadequatedesignandpoordetail-ingoftheendbearingnibsbuiltintothesupportingbeamfortheprecastbeams. The reduction in strength caused by HAC left no margin fortemperature effects.The combinationwas likely to have triggered thecollapse.


Bearings for precast beams

Figure 6.5 Assembly hall showing edge beam that supported the precast beams. (From Scott, G.A., Building Disasters and Failures, Construction Press Ltd., Lancaster, 1976. With permission.)

Figure 6.6 Part of roof that collapsed on chairs below. (From Scott, G.A., Building Disasters and Failures, Construction Press Ltd., Lancaster, 1976. With permission.)


6.3 UNIVERSITY BUILDING ROOF

OnaJunemorning,followingaperiodofcoolnightsandveryhotsunnydays, aportionof a roof, including threeof theprestressedbeams, col-lapsedontothefloorbelow.Afourthbeamhadpulledoutofitsseatingbutremainedinplacejammedagainsttheedgebeam.

The roofwasconstructedofprestressedprecast concretebeamsmadewithHACspanning12.6m, supportingprecastconcrete slabsactingaspermanentformworkforaninsituconcretetopping(seeFigure 6.8).Theouterendsofthesebeamsrestedin50mmdeeppocketscastintotheverti-calinsidefacesoftheedgebeams.Inadditiontothese,theedgebeamshadrecessesjustbelowtheirseatings.Theseencroachmentsonthealreadynar-rowsectionledtotheadoptionofareinforcementarrangementthathadnomainsteelundertheseatingpocketsfortheprestressedbeams.

Thecauseof thiscollapsewasadisastrouscombination.ThecolumnsandedgebeamswereprecastasTunits(withwebsfacingoutward).Thearchitecthad,inordertoobtainadarkcolour,specifiedHACconcretetobemadewithaggregatethatwashighlyalkaline.Thespecificationhadbeenmistypedormisreadtocall foronepartHACtofourpartssandtotwoparts aggregate (thesemixproportionswere confirmedby analysis aftertheevent).CoresdrilledfromoneoftheTunitsafterthecollapserevealedstrengthsaslowas7MPaandsomecouldnotbeextractedinonepiece.

Figure 6.7 Collapsed roof lying on floor below. (From Scott, G.A., Building Disasters and Failures, Construction Press Ltd., Lancaster, 1976. With permission.)


Thedepthof seating for thepretensioned roofbeams satisfied thecoderequirementforbearingpressure,butleftnomarginforerectiontolerancesortemperatureeffects.Therewasinadequatereinforcementdetailedunderthebeamseatings,andastheTunitswerecastwithexternalfacesdown,eventhatreinforcementwasdisplaceddownward,awayfromtheseatingledge.

Bearing stresseson the shallowseatingandvertical stressesunder thepocketswerewithinthecoderecommendations,andonlyaninvertedhatbarwasprovidedunderthebeamseating;thiswasdetailedinawaythatmade it difficult to construct in the correct position as the edgeprecastbeamswerecastwiththeoutsidesfacedowntoproduceasmoothfascia.

Linksprojectedfromtheprestressedbeamsintotheinsitutoppingandtheedgebeamshad9.5mmmildsteelbarsprojectingintotheinsitutop-ping,thusprovidingsometyingtogether,butnotattheleveloftheseating.

Theactualmechanismthat led to the failurewas found tobe thermalhoggingof theprestressedbeams.The tie steel projecting from the edgebeamintotheinsitutoppingactedasahinge,andtheendrotationaccom-panyingthethermalflexurebetweenthetwoextremescorrespondedtoahorizontalmovementattheleveloftheseatingofapproximately2mm.ThiscausedcrackingofthebearingnibasshowninFigure 6.9.Thetensionstressintheconcretesupportreduceditsshearresistancesufficientlytocausethefailure.Theprestressedbeamhadbeenmadewithhighaluminacementandalthoughitaggravatedthesituationitwasnotthecauseofthefailure.

Onanotherbuilding,anidenticaldetailwasfound,buttheedgebeamswereconstructedwithordinaryPortlandcementconcrete.Inthatcase,acracksimilartotheoneshowninFigure 6.9wasfound,butpresumablythereinforcementhadnotbeenbadlydisplaced,as fullcollapsehadnotoccurred.Thisdid,however,demonstratethatHACwasnottheprimarycauseofthecollapseofthisroof,butacceleratedthecrackinguntilcollapsehadbecomeinevitable.

Dry seat(no mortar bed)

Links projectinginto in situ topping

Prestressedbeam

150

50Bar under seat

Precast soffitslabs

Tie steel projecting into in situ topping

Figure 6.8 Prestressed beam and support.


Comment—Insummation,thecollapsewascausedby:

• Unfortunate selection of aggregate and ignorance of the material’sproperties

• Poordetailingoftheedgejoint• Insufficientbearingintheedgebeam• Insufficienthorizontalrestraintatthelevelofthebearingtoresistthe

relativemovementduetotemperatureeffects• Thereinforcementdetailedunderthebearingwasdisplaced

6.4 MINIMUM REINFORCEMENT AND CRACKING

Thedesignofpartsofabridgedeckslabdidnotrequiremorethanmini-mumreinforcement.Somemonthsafterconstructionlargecracksbegantoappear.Thesizesofsomecracksincreasedto2mmandthereasonwasnotimmediatelyclear.

Onerequisiteofacodeofpracticeforcrackcontrolistosatisfythemini-mum reinforcement percentage.Unless this is provided, the spacing andsizingofbarshavelittleeffectoncrackwidth.Thereasonforthisisbasedonthetensionstrengthoftheconcrete.Ifforanyreason(shrinkageetc.)theconcretecracks,itisessentialthatthereinforcementdoesnotyieldatthecrack.Ifitdoesyield,thecrackwillbecomelargeandthiswillpreventtheoccurrenceofsmallcracksatsmallspacings.Inordertoavoidsuchasituation,thetensionyieldstrengthofthereinforcementshouldbeatleastequaltothetensionstrengthoftheconcrete.Codesofpracticestipulateaminimumamountof reinforcementbasedon the tension strengthof theconcrete.

Thedesignforthisprojectspecifiedaconcretestrengthoffck=30MPaandtheamountofminimumreinforcementpercentagewascalculatedandprovided for this value. Unfortunately, the contractor provided concrete

Point ofrotation

�ermal gradient causesbeam to hog and rotate

Crack

Outwardmovement

Figure 6.9 Detail at support.


withastrengthof50MPa.Becauseofitshighertensilestrength(4.1MPacompared with 2.9MPa), the reinforcement yielded when the concretecrackedandthecracksbecamelarge—upto2mm!

Comment — This is one of the few, but not insignificant, cases whereincreasingamaterial’sstrengthmakesthesituationworse.

6.5 PRECAST CONCRETE PANEL BUILDING

Intheearlyhoursof16May,1968,agasexplosioninabathroomonanupperfloorshookabuilding,resultingintheinstantaneouscollapseofpartofonewing(seeFigure 6.10).Fourpeoplewerekilled.Figure 6.11showsacloserviewoftheupperfloorsaftertheexplosion.

Figure 6.10 Progressive collapse showing point of explosion. (Ronan Point building col-lapse, May 16, 1968, London. Courtesy of Building Research Establishment.)


Thereasonsforthecollapsewere:

1.The possibility of unusual, and hence non-codified, loads was notconsidered.

2.Thestructurewasinadequatelytiedtogether.

Traditionalpre-WorldWarIItwo-storeyhousingwouldnothavehadanyengineeringinput;brickwallthicknessesandtimberfloorjoistsizeswereprescribedbytheLondonCityCouncilBuildingby-laws,andsimilarregu-lationsoutsideLondon.Therehadbeengasexplosionsbeforethisincidentinsimilar typesofdwellings,but thedamageandcasualtieshadusuallybeenlimitedtoonehousehold.Theriskwasacceptedasa‘factoflife.’

Therewerethereforenoprecedentsforprogressivecollapse,whensystembuildingwasintroduced.Forfour-storeywalk-upblocks,thematerialsandthicknessesofload-bearingwallswouldsimilarlybeprescribed.Anyfire-breakingfloorswouldbestraightforwardreinforcedconcreteslabs,designedforoccupationalloading.Theywerecastinsituontothewallsbelowandtherefore‘stucktothewalls.’Evenifthedesignspanwasparalleltothewallbelow,theslabwouldimposesomeloadonit.That,togetherwiththeloadfromthewallabove,wouldprovidesomuchpreloadthatalowerlevelwallwouldbeunlikelytobeblownoutduetoitsprescribedthickness.

Figure 6.11 Closer view of upper floors showing point of explosion. (Ronan Point building collapse, May 16, 1968, London. Courtesy of Building Research Establishment.)


There was therefore a degree of tying together, albeit reliant on fric-tion,intraditionalconstruction.Inlarge-panelconstruction,mostofthisinherenttyingtogetherwaslostandnotreplaced(withsteelties)untiltherequirementforresistanceagainstprogressivecollapseenteredtheregula-tionsandcodes.

Drypackingofthepaneljoints,whetherwithmortarorfineconcrete,iscarriedoutasamenialtaskafterthe‘spectacular’eventofplacingandplumbing the panel. It is likely to attract less careful workmanship andsupervision.

Comment—ThiscollapsewasasignificanteventfortheindustryintheUKandmarkedthepartialdemiseoftheprecastindustry.Largeprecastpanelandframeconstructionbecamemuchlesspopularinthefollowingtwodecades.InformationgatheredfromtheincidentledtomajorchangestotheUK’sBuildingRegulations(1970)11andcodesofpractice(startingwiththeCP11612precastconcretecodein1970)withregardtoprogres-sivecollapseandrobustness.Morerecently,theEurocodeshaveincludedaccidentalloadandrobustnessclauses.

6.6 FOOTBRIDGE

Afootbridgewasbeingconstructedoveramotorway.ThebridgehadtwospanswithaninsituconcreteT-sectiondeck(seeFigure 6.12).Thespanlengthswere20and40m.Afterremovingtheformworkandprops,twolargetransversecracks(2and3mm)appearedinthesidesandsoffitofthelongerspan.Thecracksoccurredatthepositionsofthelapsofthemainbottomreinforcement.

Thereinforcementhadbeendetailedsothatallthemainbarsinthebot-tom,16T40s,werelappedatpositionsofhighstress.Itisgenerallyconsid-eredpoordetailingpracticetolapbarsinpositionsofhighstressandlapping

Reinforcement not shown

T12s @ 150 inpairs

16T40s lappedat one position

(no staggers)

Figure 6.12 Section through footbridge.


all thebars ina congested situation compounds the error.Thebarswerecrankedtoallowthemtobeascloseaspossible(seeFigures 6.13and6.14).Links,T12s@150inpairsenclosedallthemainbars.Noextralinkswereprovidedatthelapsorcranks.

Thelapjointsshouldhaveincludedverticallinkssurroundingeachpairof lapped bars. The presence of the vertical cranked bars increased theforcetoberesistedatthatendofthelapandadditionallinksshouldhavebeenprovidedatthisposition.Theforceduetothecrankedbarsactedasacrackinducer.

CoresweretakenandX-rayexaminationsshowedthatthe lapsofthemainreinforcementhadbeendetailedasshownonthedrawingsandthatconstruction had followed the drawings. It was decided that the bridgeshouldbedemolished.

Comment—Thisfailurewasaresultofpoordetailing.LappingofT40barscreateslargetransverseforcesthatmustberesistedwithlinks.Ifthecoderules(BS81101)hadbeenapplied,muchmoretransversereinforce-ment should have been specified. Good detailing practice would haveavoidedlappingthebarsatthepositionofmaximumstress.

13490

46

4T40s

T12s @ 150 in pairs40 cover

Figure 6.13 Layout of bars at laps.

3 mm crack

Figure 6.14 Side view of bottom reinforcement showing position of crack.

69

Chapter 7

Problems and Failures due to Inadequate Understanding of Materials’ Properties

7.1 CHANGES OVER TIME

Overtheyears, thechanges inmaterialsandreductions insafetyfactorsmake itmore important tounderstand thebehaviourof reinforced con-creteandprovidemorecare.Rulesofthumbandempiricalmethodsmayhave been developed for different conditions and may not be applicablefortoday’smaterials’propertiesanddesigncriterianeedtobecheckedtodeterminewhethertheyarestillapplicable.Anotherresultofthedevelop-mentofandchangestomaterialpropertiesisthattheultimatelimitstateisoftennolongercritical,andadesignnowoftendependsontheservice-abilitylimitstates,apartfrompunchingshear.

Concrete — There has been a continuous increase in the strength ofconcreteoverthelasthundredyears;muchoftheincreasehasdevelopedsince1980(seeFigure 7.1).Aroundthattime,thevalueofblendedcementsand theuseofadmixtureswas realised.Modernconcreteshavebecomecomplexwithalmostinfinitevariationsavailabledependingontherequire-ments.Theunderstandingofhowtochangethepropertiesofconcreteandreinforcementisdevelopingrapidly.Itincludes:

• Theuseofadmixturesandblendedcements.Admixturesareessen-tialformodernconcrete.Self-compactingconcreteisoneimportantexample.Blendedcementsallow thecontrolof the rateof strengthgainandtheamountofheatcreated.

• Theuseofstainlesssteelwillincreaseforsituationswheredurabilityisparamount.

• Theuseof higher strength concretewill becomemorepopular forfloorslabs,particularlyflatslabs.Thiswillresultinthinnerandlon-gerspanslabs.

• Serviceability limit states have already become critical to flat slabdesignanditwillbecomemorecommontocheckvibrationoffloors.

• Theuseoffibreswillincrease;theuseofsteelfibreshasalreadybeenprovenforgroundfloorslabs.


Allthesedevelopmentsaddcomplexityandcosttoconcreteconstruction.Aboveacylinderstrengthof50MPa,thestress–strainpropertieschangewithincreasesinstrength.Figure 7.2showsthischangediagrammatically.Theconcrete itselfbecomesmorebrittleas the strength increases,but itshouldbenotedthatinflexuralmembers(beamsandslabs),theductilityandbrittlenessaredependentmostlyonthepropertiesofthereinforcement.

Theincreaseinconcretestrengthandreductioninoverallfactorofsafety(seeFigure 7.3)havemeantthat,formanystructuralelements,thedesignfortheserviceabilitylimitstateisbecomingmorecriticalthanthatfortheultimatelimitstate.

Up to C50/60

Stre

ss

Strain

C90/105

Strain at maximumstress increases

Ultimate strainreduces

Figure 7.2 Change in stress block for high strength concrete.

0

10

20

30

40

50

60

70

80

1915 1933 1938 1948 1957 1965 1969 1972 1980 1983 1985 2005

Concrete Cube Strength (MPa)

Figure 7.1 Increase of concrete strength during 20th century.

Problems and Failures due to Inadequate Understanding of Materials’ Properties 71

Reinforcement—Asimilarpatternofchangehasoccurredforreinforce-mentbothinstrengthandpartialsafetyfactors(seeFigures 7.4and7.5).

7.2 REBENDING OF REINFORCEMENT

In1964,theconstructionofa35mhighdustbunkerforacoal-firedpowerstationincludedanexternalconcretecantileverstaircasetobebuiltontothe faceof theoutsidewallof thebunker.The constructionof thewall

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

1915 1933 1938 1948 1957 1965 1969 1972 1980 1983 1985 2005

Overall Factor of Safety for Concrete

Figure 7.3 Reduction in concrete partial safety factor during 20th century.

0

100

200

300

400

500

600

1915 1933 1938 1948 1957 1965 1969 1972 1980 1983 1985 2005

Yield Stress of High Yield Reinforcement (MPa)

Figure 7.4 Increase in steel yield strength during 20th century.


meantthatthereinforcementrequiredforthestairwaywouldbecastflushwith thewalland thenbentoutafter removalof the formwork.At thattime,proprietaryreinforcementsystemsforsuchasituationdidnotexistandthebarswerebentbeforefixingwithintheshutter.Theradiusofbendwouldhavebeentoastandardofthreetimesthebardiameter.

Aftertheformworkhadbeenremoved,thesurfaceoftheconcretewasscabbled toexpose thesebarsandthescaffold tubeswere threadedoverthem.Thescaffoldtubeswerethenusedtoleverthebarsoutofthewallintotheirfinalpositions.About30%ofallthebarsbentoutsnappedoffduringtheoperation.Thereasonwasacombinationoffactors:

• Thebarsshouldhavebeenbentoutwithaspecialtoolthatensuredthattheradiusofbendwasatleastthreetimesthebardiameter.

• Theparticularbatchofreinforcementwasfoundtobemorebrittle(lessductile)thanspecified.

• Theworkwascarriedoutatatemperaturejustabovefreezing.

Theremedialactiontakenwastodrillholesintotheconcreteandgroutinreplacementbars.

Comment—Thebendingoutofreinforcementcast intowalls isacom-monprocedureand,alltoooftenisdonewithscaffoldtubesthatareread-ilyaccessibleonsite. It is regrettable thataproperrebendingtool isnotoften used which is a reflection of poor understanding of the material’sphysical and chemical properties. In the manufacture of reinforcement,specialproceduresare inplace tocheck the rebendingofbars toensurethat the reinforcement is sufficientlyductile. It is unfortunate that some

0.00

0.50

1.00

1.50

2.00

2.50

3.00

1915 1933 1938 1948 1957 1965 1969 1972 1980 1983 1985 2005

Reinforcement Overall Factor of Safety

Figure 7.5 Reduction in reinforcement overall safety factor during 20th century.


manufacturershavecontinuallytriedtoeliminatesuchtestsfromtherein-forcementstandard.

7.3 TACK WELDING OF REINFORCEMENT

The design of a building with large columns required 32 mm diameterstarterbarsprojectingfromthepilecaps.Thetemporaryworksforcon-struction included tack welding some small diameter bars to the starterbars.Whenthetimecametofixthecolumnreinforcementtothestarterbars,thecontractorattemptedtobendthestarterbarstoensurethattheywouldfitintothecolumnshutterwithsufficientcovertotheconcreteface.Alargesledgehammerwasusedtoeffectthis.Duringthisoperation,twoofthe32mmdiameterbarssnappedoff.

Thereasonwasthattackweldingthesmallbarsontothelargerdiameterstarterbarschangedthemolecularstructureofthelatter.Unlikestructuralwelding, tackweldingheats just the local spot,and theheat sinkof themainbarcoolsitveryrapidly.Theresultwasthatthestarterbarsbecamebrittle and requiredonlya sharpblow to fail. In thepast, tackweldingonsitewasforbidden.Todayitissometimespermittedifcarriedoutbyaskilledspecialist.Unfortunatelyoncepermitted,itisalltooeasyforanon-specialisttodothiswork,believingthatitwilldonoharm.

Comment—Toomanypeopleareunawarethattackweldingcanhavesig-nificantstructuraleffects.Thisisanothercasewhereamaterial’schemicalandphysicalbehaviourwasnotproperlyunderstood.

7.4 HIGH ALUMINA CEMENT

Highaluminacementconcretehasachievedacertainnotorietyfollowingthecollapseofseveralbuildings inthe1970s.Bytheendof1974,upto50,000buildingshadbeenreportedassuspectandamajoreffortwasmadetochecktheirsafety.Fortunately,manyoftheaffectedbeamsstoodindryconditions and the chemical deterioration had not reached an advancedstage.Theworstaffectedelementswerepositionedindampenvironments.

Description—Highaluminacementismanufacturedfromlimestoneorchalkandbauxite(theorefromwhichaluminiumisobtained).Thetwomaterialsarecrushedandfiredtogetherusingpulverisedcoalasafuel.Thematerials fusetogether,andaftercoolingarecrushedandgroundintoadarkgreypowder.

Thepredominantcompoundsarecalciumaluminates;calciumsilicatesaccountfornomorethanafewpercent.Thecalciumaluminatesreactwithwaterandtheprimaryproductiscalciumaluminatedecahydrate(CAH10).


Oneofitsmaincharacteristicsisthattheconcretemadewithitachievesitsfullstrengthafter24hourscomparedwith28daysforaconcretewithPortlandcement.However,itscrystalstructureisunstableandchangestotricalciumaluminatehexahydrate (C3AH6) spontaneously (albeit slowly).Thisprocessoccursatroomtemperatureandisacceleratedbyanincreaseintemperature.Thecrystalstructuretransformsitselftoamorecompactform,withtheresultthatthecementmatrixoftheconcretebecomesporousandweaker.Theextenttowhichthisconversion,asitisknown,occursislargelyafunctionofthe:

• Originalwater/cementratiooftheconcrete• Temperatureriseintheconcreteduringhardening• Temperatureandmoisturetowhichthehardenedconcreteissubse-

quentlyexposed

Degreeofconversion—ItwasfoundthatatthetimeofthecollapsesmostHACconcreteusedinbuildingswas90%ormoreconverted.Aconcretefrom a wet mix exposed subsequently to the sun was found to have itsstrengthreducedfrom40MPaat24hourstoanaverageofabout10MPaafter less than 10 years. In contrast, concrete from prestressed precastbeamswithalowwater-to-cementratioandhencea24hourstrengthof65MPafromthesamebuildingbut inadryenvironmentwas foundtohaveretainedastrengthofabout35MPa.

7.5 CALCIUM CHLORIDE

Manyreinforcedconcretestructureshavesufferedfromtoomuchchlorideintheconcretemix.Thiscausesthebreakdownofthehighlevelofalkalin-ity.Whenmoistureandoxygenarepresent,carbonationoccurs.Thisallowsthereinforcementtorustandleadstospallingoftheconcretesurface.

Before1980,calciumchloridewasusedextensivelyforinsituconcreteworks,frequentlywithoutadequatesupervision.Itwasusedprincipallyforfrostprotectionandtofacilitatetherapidstrippingofshutters.However,alltoooften,toomuchwasadded.Inthe1980s,thecodesofpracticeandconcretespecificationsweretightenedtoensurethattherustingandspall-ingshouldnothappenagain.Thefollowingthreeexamplesdescribewheretoomuchchlorideinconcretecausedstructuralfailures.

Example1—Aprimary school (built in1952)was shut in1973duetoextensivecorrosionof the reinforcementof factory-madeprecastcon-cretebeams.Thiswasduetothepresenceof toomuchcalciumchlorideaddedduringthemanufactureofthebeamstohastenthehardeningofthecement.The condensationunder thebeamsaccelerated the corrosionbycombiningwiththecalciumchloridetoproducehydrochloricacid.


Example2—In1974,theconcreteroofofaschoolcollapsed.Therea-sonwasfoundtobetoomuchcalciumchlorideintheconcrete,causingthereinforcementtodeteriorateandeventuallyfail.

Example3—Anindependentinvestigationofthecollapseofa100mlongpedestrianbridgefoundthecausetobehighlevelsofcalciumchlorideinthegroutusedintheductsfortheprestressedtendons.Thisledtocorro-sionandfailureoftheprestressingtendons.

7.6 ALKALI–SILICA REACTION

Thealkali–silicareaction(ASR)isaheterogeneouschemicalreactionthattakesplace inaggregateparticlesbetween thealkalinepore solutionofcement paste and silica in the aggregate particles. Hydroxyl ions pene-trate the surface regionsof theaggregate andbreak the silicon–oxygenbonds. Positive sodium,potassium, and calcium ions in thepore liquidfollowthehydroxyl ions so thatelectro-neutrality ismaintained.Waterisimbibedintothereactionsitesandeventuallyanalkali–calcium–silicagelisformed.

Thereactionproductsoccupymorespacethantheoriginalsilicasothesurfacereactionsitesareputunderpressure.Thesurfacepressureisbal-ancedbytensilestressesinthecentresoftheaggregateparticlesandintheambientcementpaste.Atacertainpoint,thetensilestressesmayexceedthetensilestrengthandbrittlecrackspropagate.Thecracksradiatefromtheinterioroftheaggregateoutintothesurroundingpaste.

Thecracksareempty(notgel-filled)whenformed.Smallorlargeamountsofgelmaysubsequentlyexudeintothecracks.Smallparticlesmayundergocompletereactionwithoutcracking.Formationofthealkali–silicageldoesnot cause expansion of the aggregate. Observation of gel in concrete isthereforeno indicationthat theaggregateorconcretewillcrack.ASRisdiagnosedprimarilybyfourmainfeatures

• Presenceofalkali–silicareactiveaggregates• Crackpattern(oftenappearingasthree-pointedstarcracks)• Presenceofalkali–silicagelincracksand/orvoids• Ca(OH)2depletedpaste

In mainly unidirectional reinforced members, the cracks become linearandparalleltothereinforcement.Thedegreeofcrackingdependsontheamountofconfiningreinforcement,i.e.,links,etc.OnemajorconcernwasthatASRcausedcrackingthatledbitsofconcretetofalloffstructuralele-mentsandhitpeoplebelow.Thisledtodemolitionofthestructuresinsomecases.ExamplesofASReffectsaregiveninFigure 7.6.


Figure 7.6 Examples of alkali–silica reaction. (Top: From the US Department of Transportation Highway Administration; middle: From Dr. Ideker, http://web.engr. oregonstate.edu/~idekerj/; bottom: From the US Department of Transportation Highway Administration.)


7.7 LIGHTWEIGHT AGGREGATE CONCRETE

Duringthe1960s,amedium-sizecivilengineeringcontractorwantedtojointhehousingdrive,thenatitspeak.Atthetime,anAustrianconstructionfirmusedcrushedbrickrubbleasaggregateinun-reinforcedconcretewallsforsix-andseven-storeyblocksofflats.Inspiredbythis,itwasdecidedtotrytodevelopasimilarformofload-bearingwallwithadequatethermalinsulation,madeoflean-mixplainconcretewithlightexpandedclayaggre-gate(LECA).A12-storeyblockwasconstructedasapilotproject.

Thestrengthofthewallconcretewasreducedinstages—about2000psi(14MPa)at28daysforthefourbottomstoreys,1600psi(15MPa)forthenextfour,and1200psi(8MPa)forthetopstoreys.Thefloorslabswereoftraditionalreinforcedconcrete,buttheroofslabwasreinforcedLECAconcretewithastrengthof3000psi(21MPa).

Therewasnosignificantadversefeedbackfromthetenantsnorthebuild-ingauthority.Theblockremainedstandingandinuseforover40years.Encouragedbytheapparentsuccess,thecontractorstartedpromotingthe‘system.’Aboutthesametime,lightweightaggregateconcretewasincludedinthecodeofpracticeandaminimumstrengthof3000psi(21MPa)wasstipulated.Thisrequiredarichermixthanthatusedforthewallsoftheearlierblock.The resultingeffectsof thison the thermal insulationandshrinkagepropertiesoftheLECAconcreteappeartohavebeenoverlookedbythedesignteam.

AfewblockswerebuiltforlocalauthoritiesoutsidetheLondonCountyCouncil area. These were higher than the first block utilising the higherconcretestrengthrequiredbythecodeinthewalls.Manyoftheflatswereallocatedtotenantsinpoorfinancialcircumstances,whocouldnotaffordthe charges for the underfloor heating and used paraffin heaters instead.This,combinedwiththereducedthermalinsulationoftheexternalwalls,ledtoseverecondensation.

Structurally more important, however, were the diagonal cracks thatdevelopedonthetopfloorofoneoftheblockswithinashorttimeafterhand-over.Fromtheirgeometry,theyappearedtobeduetolowershrinkageandgreaterthermalexpansionoftheroofslabrelativetothewallconcrete.

Definitelyalarmingwastheoccurrenceofhorizontalcracksinoneofthe200mmthickinternalcrosswallsconnectedtothe300mmthickexternalwallatrightangles.Oneofthesecracksonthe13thfloorofa16-storeyblock opened suddenly with a noise like a gun shot. The wall was 200mmthickand,accordingtothedesignassumptions,carriedthefloorslabsthatspannedabout3.5moneitherside.Thismeantthatthebuildingcon-tainedthreestoreysofunreinforcedconcretecrosswall,withtheloadfromapproximately3.5mwidthoffloorsplustheroofhangingorcantileveringofftheexternalwall!


Structurally,theonlyexplanationforthesecracksseemedtobethatthe internal wall was drying out, and therefore shrinking and short-ening, while the external wall with very little load to carry (at leastinitially)andexposedtotheBritishweatherwasnotshorteningatthesamerate.

Discussion—TheporousLECApelletsweresoakedjustbeforethemix-ingoftheconcrete,topreventthemfromabsorbingwaterfromthefreshmixandthusmakingittoostiff.Theythereforeconstitutedareservoirofwater,overandabovethatrequiredforthehydrationofthecement.ThisextrawatermeantthattheLECAconcreteneededmoretimetodryoutandthe300mmexternalwallswouldhaveaslowerrateofdryingoutthanthe250mminternalwallseveniftheyhadthesameenvironmentonbothfaces.

Comment—Thesecrackswereduetothechangedpropertiesofthewallconcrete.Aproper studyof thepropertiesof thematerialsalongwithareviewofthedesignwouldhaveshownthatthetwo-stageextrapolationfrommediumrise tohigh riseand from leanmix,brick rubble,unrein-forcedconcretetodense,albeitlightweightaggregate,concretecouldnotbesustained.

79

Chapter 8

Problems and Failures due to Poor Construction

8.1 FLAT SLAB CONSTRUCTION FOR HOTEL

Forashorttimeintheearly1970s,thegovernmentprovidedloansfortheconstructionofhotels.Inorderforaprojecttobeeligible,theconstructionperiodhadtobeverytight.Theworkmanshipofsomeofthehotelsbuiltthenwasshoddy.Foronesuchhotel,theshoddyconstructionwasnotdis-covereduntil20yearslaterwhenamajorrefurbishmentwastakingplace.

Figure 8.1 shows the structural layoutofa typicalflat slabfloor.Thedepthoftheslabwas250mm.Thespansalongthebuildingwere7.2mandacrossthebuildingwere6.1and7.4m.Thetopsurfaceoftheslabwasveryunevenanddidnotappeartohavebeenlevelled(byhandorpowerfloat).Insomeplaces,bootmarkshadbeenleft.Cracks(generallynotlargerthan0.3mmwidth)occurredontheuppersurfaceradiatingfromthecornersofthecolumnswithoneortwosmallcracksrunningtangentially.Largecracks(upto1mmwidth)appearedatsomeoftheconstructionjoints.Thedeflectionofoneoftheslabbaysofanupperfloorwaslarge—over75mm.

Severalorganisationsbecameinvolvedinassessingthesituation.Itwasunfortunatethatoneofthemconcludedthatthestructurewasunsafeandthatoneofthefloorswasindangerofimminentcollapse.Onebayofanupperfloorinquestionwasthensetasideforinstrumentation(bothcompli-catedandcostly).Afternearlyayearofmonitoringmovement,theresultsshowednoperceptibleincreaseofdeflection.However,duringthatperiod,additionalsteelbracketshadbeendesignedtopreventtheslabsfailingbypunchingshear,andtheyhadalreadybeenfittedtothecolumn–slabjunc-tionsofseveralfloors.Afurtherindependentcheckwasinstigatedandthefollowingwereitsfindings:

Excessivedeflection—Inordertounderstandhowtheexcessivedeflec-tionhadcomeabout,itwasnecessarytochecknotonlythedesign,buttheinsituconcretestrength,reinforcementdetails,andcovertothereinforce-mentasbuilt.Thischeckfoundthattheonlyfactornotasspecifiedwasthetopcovertothereinforcementnearthecolumnsupports.Thiswasfoundtobeonaverage30mmmorethanspecified.Thisreducedthemomentof


resistanceatthesupport.However,afterreasonablemomentredistributionwasincludedinthecalculations,therewassufficientoverallmomentcapac-ityintheslabwithoutrequiringanyreductiontothedesignsafetyfactors.

This conclusion raised the question ofwhy such large deflections hadoccurred.Oncloseexamination,itwasnoticedthattheedgedeflectionoftheslabalonggridlineBwasalsoverylarge.Inoneofthebays,theskirtingboardbetweentwoedgecolumnshadbeenmadeintwoequallengthssplitinthemiddle(seeFigure 8.2).Deflectionsof15to20mmoccurredbeloweachhalfoftheskirtingboard.Thisrepresentedanedgedeflectionupto50mm.Sincetheskirtingboardwasattachedtothewall,itwaslikelythatitwasfittedthiswayandthatmuchofthedeflectionhadtakenplacebeforeconstructionofthewall.Thiswasconfirmedbyfindingthatthebottomcoursesoftheexternalwallhadbeenlaidonthesaggingshapeoftheslabandthefollowingcoursesadjustedsothattheywerelevelatthewindowsillsabovethefloor.

The conclusion was that much of the deflection occurred during con-struction,someofwhichwasduetothesaggingoftheformworkandsomeduetoearlyremovaloftheformwork.

Edge column

Skirting board

Floor surface15–20 mm

Edge column

Figure 8.2 Skirting board deflection.

1 2 3 4 5 6 7

7200 (all bays)

Down

Down

Down

7400

6100

DC

B

Column size900 × 300 mm

Figure 8.1 Structural layout of hotel floor.

Problems and Failures due to Poor Construction 81

Punchingshear—Punchingshearfailureisdifficulttopredict.Failurecanoccurwithlittlewarning.Thepresenceofcracksintheslab,tangentialtoacirclearoundthecolumninthetopsurfaceoftheslab,indicatedthatthepossibilityofapunchingshearfailureshouldbeconsidered.Thecovertothereinforcementwaslargerthanspecifiedinthisareawhichmeantthatthecracksizewasmagnified.However,thefactthatthesurfacecrackswerelargedidnotnecessarilymeanimpendingshearfailure.

Areliablemethodofpredictingpunchingshear failure isbyusing thecodedesignexpression(BS81101orBSEN1992-1-12)forshearwiththecharacteristicvaluesfortheconcretestrengthinsteadofthefactoreddesignvalues and the as-built information concerning the reinforcement (i.e.,size,spacingandcovertothebars).Anassessmentofsafetycanbemadebycomparingthe ‘worstcredible’ loadswiththeresistanceascalculatedabove.Thecalculationsforthiscaseshowedthattheworstcredibleloadscouldbecarriedwithasufficientsafetyfactor.

Comment—Itwasunfortunatethatthereasonfortheexcessivedeflectionwas not found earlier and that calculations were not made to check thepunchingshearcapacity.Muchcostlyremedialworkcouldhavebeensaved.

8.2 STEEL PILES SUPPORTING BLOCK OF FLATS

Thepilecapsupportingthestructureforamulti-storeyblockofflatsincor-poratedsteelHpiles.Theseshouldhavebeencutoffclosetothebottomofthepilecapbutweretakenuptowithin100mmofthetop.Thepileswerecoatedwithbitumenthathadnotbeeneffectivelyremovedfromtheprotrud-inglengthsthathadtotransfertheloadfromthepilecapsthroughbond.Theresultwasthatasthebuildingworkcontinued,thepilecapsstartedtosettlewiththepilesactingaspistons.Theconcreteatthetopofthepilecapsfailedinpunching.Theremedialwork includeddiggingunder thepilecapsandweldinglargecollarsaroundthepilestopreventfurthermovement.

8.3 SHEAR CRACKS IN PRECAST T UNITS

RemedialworktothebearingofaprecastcolumnrequiredlocaljackingofprecastTfloorunitsforseveralfloorsabovetoreleasetheirload.Thejacksandpropsweretakenrightdownthestructuretotheground.Atonefloor,thepropsaboveandbelowdidnotlineup.TheeccentricloadcrackedtheTunitatthatfloor.Figure 8.3showstheopeningupof2mmshearcracks.

The subcontractor proposed to repair the damaged Tunit bydrillingangledholesdownthecentreofthewebandgroutinginstraightdeformedreinforcingbars.Hewas sufficientlyconfident toagree to then test load


thatunitwiththefullultimatedesignloadtocheckthatitwasserviceable.Figure 8.4showshowthebarswerearranged.Therepairworkrequiredcarefuldrillingtoensurethattheprestressingtendons(showndashed)werenotdamaged.Theunitpassedtheextremetest loadandpermissionwasgivenforittoremainaspartofthestructure.

Comment—Itwassurprisingthatstraightbarswerecapableofgeneratingenoughbondandprovidingsufficientshearresistance.

8.4 CANTILEVER BALCONIES TO BLOCK OF FLATS

Fiveyearsafterconstructionoftheflats,cracksappearedonthetopsur-facesofthecantileverbalconies.Beforecarryingoutrepairs,investigationrevealedthatthereinforcementrequirednearthetopsurfacewasplacedhalf way down the section. Although the amount of cracking varied ateachfloorlevel,itwasdecidedtorebuildthebalconiesateveryfloorlevel.

Shear crack

Figure 8.3 Cracked T unit.

(a) Elevation showing the extra shear reinforcement (b) Section through repair

6 No. 20 mm reinforcingbars grouted into web

Figure 8.4 Repair of T unit.


Theengineerwascuriousbecausethefifthfloorlevelshowednosignsofcrackingatall.Whenthebalconyatthatlevelwasdemolished,itwasdis-coveredthatthereinforcementhadbeenleftoutaltogether!

Theconcretehadsurvivedsowellbecauseitstensionstrengthwassuf-ficient to resist all the loads that the slab sustained. The reason for thecrackingoftheotherbalconieswasthatthereinforcementrestrainedtheshrinkageoftheconcreteandthuscreatedtensionstressesintheconcrete.Normallythereinforcementshouldbedetailedwithbarsatsmallspacing.Ifthebalconieshadbeenconstructedwiththespecifiedconcretecover,anycrackingwouldhavehadasmallwidthandbeenatasmallpitch.

Comment—Althoughtheupperbalconyshowednosignsoffailureitwasmuch more dangerous than those for the floors below. If subjected to alargeimpactload,itwouldhavecollapsedsuddenly.

8.5 PRECAST CONCRETE TANK

Aliquidstoragetankwasconstructedwithprecastwallpanels.Thediam-eterandheightofthetankwere12.3and7m,respectively(seeFigure 8.5).The vertical panelswere held inplace byunbondedprestressed tendonsthreadedthroughhorizontalPVCductsembeddedintheconcreteandfullyencirclingthetankatsetlevelsthroughouttheheight.Thetankcollapsedwithoutwarningwithin2yearsofconstruction.Failurewascausedbyanumberofseparatemistakes.

Watertightness—Inorderforthetanktobewatertight,eachprecastverticalpanelhadtobuttuptotheadjacentpanelevenlythroughoutthe7mheight.Arubberstripwasinsertedwithinthejointbetweeneachsetofadjacentpanelsand incorporatedholes throughwhich theprestress-ingtendonswerethreaded.Theprestressingwasintendedtocreateuni-formcompressionthroughouttheheightofthetank.However,toachieve

12.2mAnchor unit

Figure 8.5 Precast elements of tank.


watertightconstruction,theedgesofthewallunitshadtobebuiltwithverysmalltolerances.Thedesignoftheseprecastpanels(seeFigure 8.6)allowedforseveraldifferentdiameter tanks.Thismeant that theangleof the edge formwork for the panel was adjustable and had to be setwith extreme care for each diameter of tank. The water tests showedleaks. Several attempts were made to seal them before water tightnesswasachieved.

Prestressing ducts — During the water tests and possibly afterward,waterpenetratedthePVCprestressingducts.Thedesignassumedthatevenifwaterhadpenetratedtheducts,therewouldbesufficientprotectionoftheprestressingtendonsfromtheenclosingsheathandgrease.

Sheathingandgrease—Thesheathandgreaseprovidedacontinuouscoveringofthetendonsuptotheanchoragezone.Inordertoattachtheprestressingjacksandinsertthewedges,thetendonswerestrippedofthesheathingforsomedistance(seeFigure 8.7).

The amount of sheathing required to be cut back depended on theextentoftendonstretchduringprestressing.Inevitablyaftertheoperationsof stressing and locking off, there remained a length of bare tendon. Itwouldhavebeenoverlyoptimistictoassumethatthegreasecovertothestrippedstrandswouldbeintactafterthreadingthemthroughtheanchor-age.Althoughpossible, itwouldhavebeenadifficulttasktoreplacethegreasearoundthebaretendonafterstressingandunlikelytobecompletedsuccessfully.Anywaterthatpenetratedtheprestressingductswouldhavereachedthispartofthetendon.

23 mm PVC ductInterface withadjacent unit

Figure 8.6 Section through wall panel.

7-wire greased tendonPVC duct castinto concrete

Anchorage castinto concrete

Screw in capfilled with grease

Sheath over tendoncut back from end

Figure 8.7 Detail of anchorage zone.


Grease—Thegreaseusedinthisparticulartypeofunbondedtendon(12.5mmdiameterTyesaseven-wirestrandmanufactured inSpain)wasfoundtoemulsifywhenincontactwithwater.Thisallowedanywaterthatpenetrated theanchorzone tonotonlycome intocontactwith thebarepartsofthetendonsbutalsotopenetratethesheathing.

Stress-corrosioncracking—Thealloysteeloftheprestressingtendonsused in this structure had a microstructure susceptible to stress–corro-sioncracking,andthestressinthetendonswasgreaterthan50%oftheyieldstrength.Moistureincontactwiththetendonsprovidedacorrosiveenvironment.Onexaminationafterthecollapse,itwasfoundthatstress–corrosioncrackinghadtakenplaceinmanypartsoftheunbondedtendons.

Comment—IfsimilarstrandmanufacturedintheUKhadbeenused,thegreasewouldnothaveemulsified.Nevertheless,itwouldstillhavebeenadifficulttasktoensurethatthebaredendsofstrandsneartheanchorageswereadequatelyprotected.Althoughitwaslesslikelythatstress–corrosioncrackingwouldtakeplace,therewasstillasignificantrisk.Thereisnorea-sontosupposethatothertanksofthistypedidnotleakwhenwatertestedasitwasanexactingtasktoensurethateachverticalfaceofeverypanelmatchedupexactlywithitspartner.Henceitislikelythattheprestressingductsformanyofsuchtankswouldbefloodedatsomestageintheirlives.Intheopinionoftheauthor,thevulnerabilityofsuchconstructioncastsdoubtontheviabilityandsafetyofsuchsystems.

8.6 CAR PARK

InMarch1997,a120tonnesectionoftheroofofacarparkcollapsedontothefloorbelow(seeFigure 8.8).Thisoccurredat3a.m.when,fortunately,nopeoplewereinthestructure.Itwasimmediatelyclearfromthedebristhatapunchingshearfailurehadtakenplace.

Figure 8.8 Collapse of roof of car park. (From Jonathan Wood, http://www.hse.gov.uk/research/misc/pipersrowpt1.pdf.)


Thecarparkwasconstructedusingtheliftslabmethod.Thisinvolvedcasting the slabsoneon topof anotheron theground.Precast columnswerepositionedandthentheslabswerejackedupthecolumnstothecor-rectlevel.Finalconnectionbetweentheslabandthecolumnwasmadeviaasteelcollarintheslabandasteelinsertinthecolumnintowhichwedgeswerefixed.ThesteelcollarsupportedtheslabonanglesthateitherformedasquareoranHinplan.Figure 8.9showsthetypicalHconfigurationusedfor internalcolumns.Thecolumnconnection isverydifferent fromthatfoundinatypicalflatslab.

The230mmthick slabwasconstructedwithconcreteofhighlyvari-ablequality.Areasoflowqualityconcretedeteriorated,probablythroughfreeze–thawaction.Insomeplaces,thisdeteriorationoccurredtoadepthof100mmandhadbeenrepaired.Therepairwaspoorlybondedtotheparentmaterial.This leftaslabthatwaseffectivelysplit into two layerswith the only connection being the longitudinal steel passing throughtherepairintotheoriginalconcrete.Furtherdeteriorationoftheoriginal

Top reinforcementonly shown

Shear headcollar

Punching shearperimeter (1.5d)

Wedge

19

16

130

130

241241H

H22

Figure 8.9 Plan layout at column.


concrete,andinparticularitsbondstrengthtothetopsteel,reducedwhatcompositeactionexisteduntilfailureoccurred.

Comment—ThisformofconstructionhadbeenusedinmanyplacesintheUKduringthe1970sand1980sandhasbeenacommonformofconstruc-tionintheU.S.Ithasprovidedreasonablyrobuststructures.Theverynatureoftheconstructionmethodfocusesattentiononthecolumn–slabjoint.Insomesituations,thestructurehasreliedonthemomentresistanceofthesejoints,i.e.,unbracedframes.Inothersituations,separateinsitucorestruc-tureswerebuilttohandletheswayforcestowhichthecompletestructuremayhavebeensubjected.Insomeofthelaterexamples,Ubarswereweldedtothesteelcollarsandembeddedintothesurroundingconcrete.

8.7 CRACKING OF OFFSHORE PLATFORM DURING CONSTRUCTION

Thesubstructuretothisplatformconsistedofthreeconcretetowerssetonacellularconcretestructurethatwouldsitontheseabedinthefinalloca-tion(seeFigure 8.10).Itwasconstructedonshoreinadrydock.

Duringtheoperationoffloatingtheplatformouttotheoilfieldandsink-ingitintothecorrectposition,itwasnecessarytofillsomeofthecellswithwater(seeFigure 8.11).Thisensuredthattheplatformfloatedattherightlevelasitwastowedtotheoilfield.Thecellsandshaftswerethenfilledinacontrolledmannertoallowtheplatformtosettleontheseabedintothecorrectposition.Thecellswerecompletelyfilledwithwaterduringopera-tionandnonewasusedforoilstorage.

Whilstthemovingoperationwastakingplace,thepartiallyfilledcellscreateddifferentialpressuresamongthecells.Itwasimportanttoensurethat the cells remained watertight during this operation. Arrangements

Figure 8.10 Elevation of offshore platform.


weremadeduringconstructiontotestthewatertightnessofthecells.Theintentionwastofilleachoneofthecellstoitstoponly.Eachcellwascon-structedwithaplastictubethatledtothetopoftheshaftstoallowairtoescape from thecellwhen itwasfilledduring thefinal installation.Themainwater supplyused tofill cellswas capableof providing apressuremuchmorethanthatrequired,andthewaterlevelwasallowedtoriseuptheplastictube(seeFigure 8.12).

Water surface

Water ballast

Figure 8.11 Water ballast pumped into some of the cells during sinking operation.

Plastic tube to let air outof cell to sea surface

Excessive pressure causedwater to rise up plastic tube

To test water tightness of cellswater let into cell from the mains

Figure 8.12 Water tightness test during construction.


Onlyaverytinyamountofwaterwasneededtofilltheventpipe,butthisvastlyincreasedthehydrostaticpressure.Thegreatlyincreasedwaterpres-sureinthecellsexceededthedesignpressureandcausedoneofthewallstocrack(Figures 8.13and8.14).TherepairworkincludedcastingreinforcedconcretebuttresseseithersideofthecrackedwallandprestressingthemtothewallwithMacalloybars(seeFigure 8.15).

Comment—Thisfailurecouldhavebeenavoidedifithadbeenrealisedthatthetestingarrangementcouldpermittoohighapressuretobeappliedtothestructure.Thefailurelayintheinadequatemonitoringandcontrolof thewaterpressure in the cell,not in thedesignof the structure.Theoverloadwasaboutthreetimesthedesignpressure.

To test water tightness of cellswater pumped into cell.

Shear cracksoccurred

Figure 8.13 Shear crack in wall.

Shear cracksoccurred

Water pressure

Figure 8.14 Enlarged view of shear crack.


8.8 SPALLING OF LOAD BEARING MULLIONS

The facadesof a tallofficeblockweremadeofprecast concrete sculp-tured panels (see Figure 8.16) and incorporated load bearing mullions.Thehorizontalpanelsandmullionjointswereabout600mmabovethefloorlevel.Ithadbeenspecifiedthatthepanelsbelevelledbymeansofsteelshimsandthenmortarbedsbelaidtoaleveljustproudoftheshims.Thepanelswiththemullionsattachedwouldthenbelowereddownontothebedswith the intention that the loadwouldbe sharedbetween themortarandshims.

Sometimeafterconstruction,someofthemullionsbegantospallaroundthebearings.Itwasdiscoveredthatsomeofthemortarbedsweremissing.Inplaceof these, the contractorhadundertaken todrypack the joints,butthishadnotbeencarriedoutthoroughly,probablybecauseitwasanimpracticaltask.Themortarusedwasmuchsofterthanthesteelshimsanddidnotcarryitsshareoftheloadthatwaslargelytransferredthroughtheshims.Theconcentrationoftheloadonsuchasmallareaofconcreteinthemullionscausedseverespalling(seeFigures 8.17and8.18).

Therepairworkincludedinsertingasetofsteelcolumnsateveryfloor(seeFigure 8.19).Toensurethecorrectloadwastakenbyeachcolumn,flatjackswereinsertedateachfloorlevel.Eachjackwasinflatedwithresinthatwasallowedtohardenoncethecorrectpressurewasattained.

Comment—Thisjointdetailmighthaveworkedhadhardwoodplywoodbeenusedfortheshimsinsteadofsteel.TheEvaluewouldhavebeenclosetothatofthesetmortarand,ifpre-soaked,theshimswouldhaveshrunkastheydried,thustransferringtheloadtothemortar.

Prestressed withMacalloy bars

Concrete buttresses

Figure 8.15 Repair work.


Figure 8.16 Office block with sculptured panel facades. (Courtesy of Poul Beckmann.)

Steel shim

Figure 8.17 Spalling of mullions. (Courtesy of Poul Beckmann.)


Althoughthisincidenthasbeenincludedinthechapterbasedonpoorconstruction,thedesigndetailforthejointbetweenmullionswasdifficult,ifnotimpossible,toconstructasintended.Itwascriticallyimportantforthestructuralintegrityofthefacadeandmorethoughtshouldhavebeengiventoproducingamorepracticaldetail.

8.9 TWO-WAY SPANNING SLAB

Atwo-wayspanningslabwasdesignedinaccordancewiththedesignrules.Onesidewassignificantly longer thantheotherandthismeantthat the

Concrete profileafter spalling

Steel shims

Figure 8.18 Plan section through mullion after spalling.

Additional steel columns

Figure 8.19 Layout of additional columns. (Courtesy of Alan Steele.)


reinforcementprovided for the short spanwasconsiderablygreater thanthatforthelongspan.

Whenthedrawingarrivedonsite,theengineerdecidedthatthedrawingcontainedamistakeandthatthereinforcementinthelongspanshouldbethegreater.Heinstructedthereinforcementlayouttobeswappedaroundandtheslabwasbuiltthatway.Later,whenthefullloadwasapplied,theslabcollapsed.

Comment—Thisfailurecouldhaveeasilybeenavoidedifthesiteengineerhadcheckedwiththedesignerbeforemakinghisdecision.

8.10 CHIMNEY FLUE FOR COAL-FIRED POWER STATION

Thewindshieldofthischimneywas270mhighand20mindiameter.Atlowlevel,three‘portals’forthehorizontalboilerfluesweretoconnectwiththethreeverticalflues.Theflueswereabout8mindiameterandlinedwithfirebricks.Theirwall thicknesswasreducedabove theportalsbyataperon theoutside surface.The reinforced concrete shaftsof thewindshieldandtheflueswereconstructedsimultaneously,usingslip forming.Theconcretewasmadewithahighproportionofblast-furnaceslagcementreplacement;thePortlandcementandtheslagwerenotpremixed,butsup-pliedseparatelyandstoredinseparatesilosatthebatchingplantthathadasinglemixer.

TheroofslabandoneortwointernalfloorslabswereseparatedfromthefluewallswithFlexcelorasimilarjointmaterialtoallowunrestrainedtem-peratureexpansion.Thefirebrickliningstothefluesweresupportedoncor-bels,andthecastingandthelayingofthebrickworkfollowedtheslipforming.

Abovetheroofslabatthetop,theflueswerealsocladexternallywithbrickwork.Thebricklayershadjustfinishedcappingthelastfluewhenafracturelowerdowncausedthetophalfofoneofthefluestoslidedown,disintegrate,andfillthebottomofthewindshieldwithdebristhatspilledoutoftheportalforthatflue.Oneworkerwaskilled(seeFigure 8.20).

Itwasstillpossibletoenterthechimneyspacethroughoneoftheotherportals, and the inside of one of the intact flues could be examined. Asubstantialhorizontalcrackwasfoundonpartofthewallsurface;insert-ingaknifebladeintothecrackindicatedthatthefracturesurfaceslopedupward.Subsequently,morecracksofasimilarnaturewerefoundontheinsideofthewindshield.

The fracture surfaces on the concrete fragments in the debris seemedmore square and sharper than expected. This was later explained as afeature of sudden disintegration of high-strength concrete by impact, asopposedtotheslowfailuresobservedinlaboratorytests.Coresweretaken


atvariouslocations.Anumbershowedpatchesofbluecolouronthefreshlydrilledsurfaces,somedistancefromthewallfaces.Hydratedslagisblueuntil ithasbeenoxidizedbyexposuretotheatmosphere.Thisthereforeindicated that the cementand slaghadnotbeenproperly intermixed.Anumberofcores,someofwhichwereabout0.75mlong,weresawnintotestablelengthsandallshowedadequatestrengths.Othercoresweretakenacrossthehorizontalcracksandshowedclearseparation.

Anexperiencedslipformingexpertwascalledintogiveasecondopin-ion.Heexplainedthatthehorizontalcrackswere liftingcracksthatcanoccurwhenliftingisresumedafterconcretinghasbeeninterrupted.Ifsomeofthehalf-setconcretethenstickstotheform,itisliftedwithit,untiltheweightofthefreshconcretedepositedontopforcesittodropback.Astheconcretemaynotalldropneatlyinonepiece,cavitiesmayresult,obviouslyweakeningthewall.Liftingcrackscanbecausedoraggravatedbyuneven-nessofformworksurfaces;theyareusuallyhiddenbytheslurryrub-downthatiscustomarilycarriedoutfromafinishingplatformsuspendedsome2mbelowaworkingplatform.Recordsandanecdotalevidenceindicatedanumberofoccurrences:

• Ashortdistanceabovethefoundation,awholebandhadbeencastwith practically pure slag. This was discovered early, cut out, andre-concreted.

• Where the wall thickness was reduced above the flue portals, thewholeoftheformworkhadtobeliftedclearoftheconcretetoallowadjustmentoftheoutsideformworktothesmallerdiameter.Thishadtobedonetwice—atthebeginningandattheendofthetaper.Therewassomecongestionofreinforcementinthiszone,someofitduetodesign,someofitduetoineptsteelfixing.

See detail

Detail: Close up of debris

Figure 8.20 Debris from collapsed flue. (Courtesy of Alan Steele.)


• Deliveries of materials were sometimes late, causing concreting tohalt.

• Themixerbrokedown,causingconcretingtohalt.• Supervisionandinspectionwereinadequate.

Theconclusionofhowthecollapseoccurredwasasfollows:

• Onoccasionwhenconcretingstopped,asevereliftingcrackformedintheflue,extendingaboutthreequartersofthecircumference.Becauseofitslocationsomewayuptheshaft,theintactquadrantcouldcarrytheweightoftheshaftabove,albeitwithasubstantiallyreducedfac-torofsafety.

• Solarheatingcaused thewind shield tobend.Theodolitemeasure-mentsshowedthatthetopofthewindshieldonsunnydaysdescribedanirregularhorizontalorbitwithadiameterintheorderof0.75m.

• Theflueshadtofollowthismovement,whichatcertaintimesofthedayincreasedthecriticalstressontheintactquadrantofthecrackedflue. Italsocausedthe liftingcracktoopenandcloseonadiurnalcycle.Thiscausedtheconcretetogrindawayjustbeyondtheendsofthecrack,graduallyreducingtheareaofloadcarryingconcreteuntilfailureoccurred.

Comment—Morethoroughsupervisionwouldhavepreventedthisfailurefromoccurring.The slip formingprocess requires carefulmonitoringatalltimes.

97

Chapter 9

Problems and Failures due to Poor Management

Manyofthecasestudiesintheotherchaptersmayhavebeenavoidedwithbetter management. The three cases reported in this chapter emphasisehowimportantanengineeringbackgroundistothemanagementofstruc-turalengineeringprojects.

9.1 COLUMN–SLAB JOINT

Attimeswhensomuchworkmakesaprojectteamover-committed,itiscommon to assignpackagesofwork toother teams. For this particularjob,aprojectteamdesignedtheslabsandhandedoverthedesignofthecolumnstoanotherteam.Thereinforcementdetailingwascarriedoutbyanothergroup.Allthedesignworkshouldhavebeencheckedbytheprojectteambutinthiscasetheseparatepartsofthedetaildesignwerepassedtothedetailinggroupwithoutanoverallcheck.Itwascommontodetailtheslabsandcolumnsonseparatedrawings.Figure 9.1showsthearrangementoftheedgecolumn–slabjoint.

The reinforcement arrangement intended for this joint is shown inFigure 9.2.Construction reached the secondfloorwhena younggradu-ate visited the building to gain some site experience. He reported backhisconcernaboutthereinforcementlayoutalongtheedgeoftheslabthatwasabouttobeconcreted.Figure 9.3showswhathesaw.Therewasnomechanical linkbetween the slab and the column reinforcement! Itwasimmediatelyrealisedthatnotonlyhadthisoccurredatalledgesoftheslabonthesecondfloor,butalsoonthefirstfloorthatwascompletedseveralweeksearlier.

Immediateremedialworkwasputintoaction.Temporarysupportswereputinplaceforbothfloorsandthenholesweredrilledthrougheachjointandlongboltsgroutedinplace(seeFigure 9.4).


Slab

100

Edge column800 × 400

Figure 9.1 Plan layout of edge column–slab joint.

T16s

T12s

T10s

Figure 9.2 Intended layout of reinforcement.

Problems and Failures due to Poor Management 99

Comment—Oneofthereasonsthissituationoccurredwasthatnocheckhadbeenmadeinthedesignofficetoensurethatamechanicallinkcouldbemade. In fact,when thecoverand theactual sizeofbarswere takenintoaccount,itwasclearthatthereinforcementcouldnotfitasintended.Thefabricatorofthereinforcementfittedthebarsasclosetothecorrectpositionaspossiblewithoutthinkingthatamechanicallinkwasessential.

Nevertheless,theunderlyingcauseofthismistakewasinfailuretoman-agethedesign.

T12s

T16s

T10s

No mechanicalconnection

Figure 9.3 Actual layout of reinforcement.

T16s

T12s

T10s

Figure 9.4 Remedial work.


9.2 PLACING OF PRECAST UNITS

Aspinebeamcarryingprecastplankslostitsbearingbecausealabourertrying to jack one of the final planks into position actually levered out

Section

Plan View

spine beam

Wall supportingspine beam

Precast slab jackedinto position

Wall shifted outwards causingspine beam to fall off its bearing

Precast slab jackedinto position

Lacer bars not inplace at time of jacking

Precastslabs

Figure 9.5 Adjusting positions of precast planks.

Problems and Failures due to Poor Management 101

thewallpanelsupportingtheendofthespinebeam(seeFigure 9.5).Thiscaused the spinebeam to lose its bearing and led to the collapseof thefloor.TheconnectionbetweenthespinebeamandthewallpanelhadbeendesignedtohavetwooverlappingUbarslockedtogetherwithlacingbarsthreadedthroughthespacebetweenthem.Theselacerbarshadnotbeeninsertedatthetimeoferectingandlayingofthefloorelements.Iftheyhadbeeninplace,theywouldhavepreventedthewallpanelfrommovingawayfromthespinebeam.

Comment—This isanexample inwhichmanagementshouldhavehadmorecontrolonhowtheerectionandplacingofprecastunitstookplaceand,moreimportantly,ensuredthatthelacerbarsattheendsofthespinebeamwereinplacebeforetheerectionoffloorunitstookplace.

9.3 WEAK AGGREGATE CONCRETE IN CHIMNEY

A tall chimney with a single flue shaft for a coal-to-petrol synthesizingplantwasbeingslip-formed.Half-wayup,the3-daycylindertestsresultssuddenly showed a serious dip in the strengths. The samples had beentakenfromconcreteplacedbetweenmidnightandthefollowingmorning.Subsequenttestresultsweresatisfactory.Aftersomeprobingandquestion-ing,thefollowingscenarioemerged.

Thereweretwobatchingplantsonthesite,eachwithitsownday-workstockpile.Onewassuppliedwithlocallyavailable,butsomewhatinferior,aggregateforuseinlow-gradeconcreteforplantfoundationblocks.Theotherhadtoobtain‘good’aggregateforthechimneyfromsomedistanceaway.Thebatchingplantswereatoppositecornersofthesite.

Onthenight inquestion,themidnightaggregate lorrydidnotappearon time.The foreman inchargeof slip-forming foundhimself facinganunplanned stop to the sliding thatwouldhave causeda calamity.Apartfromtheresultingdelay,itcouldeasilyhavecausedaproblemwhenrestart-ingtheslip-forming,forwhichhecouldhavebeenblamed.Notawareofthedifferentqualitiesoftheaggregates,hissolutionwastogetacoupleofdumpertruckstofetchaggregatefromtheotherbatchingplantthatwasshutdownforthenight.Theslidingcouldthencontinue.

However, this left the other batching plant short of aggregate. Whenthelorrywiththe‘good’aggregatearrivedinthesmallhours,theslidingforemandirectedittounloadatthe low-gradebatchingplanttoavoidadaybreakdisputewiththatforeman.Normalservicewasthenresumed.

Bythetimeallthiscametolight,asubstantialheightofgoodconcretehadbeenplacedontopofthelow-gradeaggregatebelt.Toavoiddemolish-ingsome20mofshaft,itwasproposedtoconstructa“collar”or“sleeve”ofgoodconcreteorgunitearoundtheweakzone.Compositeactioncouldbe provided by drilled-in anchor bolts acting as shear connectors. Thiswouldnecessitateaplannedinterruptionoftheslidingtoenableaworking


platformtobesuspendedfromtheslipform,sotheremedialworkcouldbecarriedout,butthiswasacceptedasanecessarydelay.

Comment—Carefulmanagementisessentialfortheslipformingprocess.Thereshouldhavebeenproceduresinplacethatensuredconsistentsuppliesofthecorrectaggregate.Thestockpileshouldnothavebeenallowedtofalltosuchalevelthattheslipformingproductionlinewaswaitingforalorrytoarrive.

103

Chapter 10

Problems and Failures due to Poor Construction Planning

Planningofanyconstructionworkrequirestheinputofengineeringthoughtandthisincludesthetimeneededtoconsider‘whatifs?’.Handsketchescanbeveryusefultoexplainwhatshouldorcouldhappen.

10.1 POWER STATION ON RIVER THAMES

ApowerstationwasconstructedonthenorthbankoftheThamesintheearly1960s.Originallyitwastobecoalfiredtoproduce1500MW.Thefoundationsofthepowerstationsaton20,000reinforcedconcretepilesandaspecialcastingyardwassetuponsitetoproducethem(seeFigure 10.1).

Thepileswere430mmsquareand18mlong.Severalpilerigsweresetupwithdiesel-drivenhammers(seeFigure 10.2).Apilewashoistedintopositionandthengivenatapbythehammertogetthepointofthepilethroughthetopcrustofthemarshland.Thenunderitsownweightthepiledropped15mthroughthemud.Eachpilewasthendrivenintothegraveluntilaspecifiedsethadbeenreached.

Pilingcommencedfromtheedgeofthesiteclosesttotheriverandcontin-uedinlandforadistanceofover250m.Pileswereplacedat1.5mcentres(onaverage).Theexactpositionsofpilesshownonthedrawingsrelatedtothetypeoffoundation(turbine,boiler,culvert,ormiscellaneous).Thetimeperiodforthispartoftheprojectwasabout18months.Excavationforthefoundations,alsostartingfromtheriverendofthesite,commenced6monthsafterthestartofpiling.Thedepthofexcavationvariedfrom1to3m.Thisexposedthepilesthatwerethencutdowntotheleveloftheconcreteblinding.Concretingofthefoundationsthencommenced,startingfromthesameendofthesiteasthepilingandexcavation.

A year after the start of piling, when concreting of the foundationsprogressedaboutathirdofthewayalongthesite,itwasdiscoveredthatthetopsofthepilesthatwerestillexposedweremoving.Measurements showed that the movement was up to 1.5m!


Figure 10.2 Pile driver.

Figure 10.1 Pile casting yard.

Problems and Failures due to Poor Construction Planning 105

Thefirstmajorconcernwastocheckwhetherthepileswerestillintact.Onepilewasextracted.Itcameoutstraightandappearedundamaged.Onclose examination,fine cracks couldbe seen throughout its length.Thisindicatedthatithadbentevenlyoveritsfulllength.

Ittooksomeweekstofullyunderstandwhathadhappened.Figure 10.3showsaplanof the site andattempts todemonstrate the situation.Thesteady addition of piles into the ground built up ground pressure. Thispressurewasappliedincreasinglyfromonedirection.Atthesametime,theexcavationforthefoundationsofthestructureclosesttotheriverreleasedanygroundpressure thatbuilt up.The combinationof these twomajoractionscausedthemovement.

Theresultingremedialworkincluded

(1)addinganadditional600verticalpilestocompensatethereductioninverticalcapacityoftheexistingpiles

(2)adding200morerakedpilestocompensateforthehorizontalforcecomponentcausedbythebentpiles.

Large amounts of remedial and extra work were required because ofthemovementofallthesepiles.Forexample,theexistingpilesnolongerfollowedtheplanlayoutfortheeightinletandoutletculvertsthatwound

Turbinefoundations

Suction

Boilerfoundations

Chimneyfoundations

Chimneypiles

River �ames

Pressure

Figure 10.3 Diagram showing cause of movement.


theirway through the site tobring coolingwater to the condensers andreturn it to theRiverThames.On-sitedecisions tomakechanges to thedesign had to be made each day. It became very tricky trying to fit theextrapiles,especiallyforthe170mchimneypilecaps.Theoriginaldesignincludedacarefularrangementof rakedpileswhichhadbeendrivenallpointingtowardsthecentreofthecap.Theyhadalltippedover,makingacompletetangle.Astheextrapilesweredrivendowntheykepthittingexistingpiles.Thismeantthattheyhadtoberepositionedandangledtoobtainaclearroute.

Comment—Theprogrammeforthecontractsonthisprojectdidnotfore-seetheproblemscausedastheworkprogressedfromoneendofthesitetotheother. Inprevioussimilarprojects,asignificantdelaybetweenpilingandthestartofexcavationallowedenoughtimeformuchofthesoilpres-suretodissipate.Theneedtoreducetimeandcostsonthisprojectwasnotbalancedbycarefulconsiderationoftheconsequence.

Wherepilinglayoutdrawingsaremadeforlargesites,eachpileisindi-catedbyacrossonthedrawing,sothevolumetriceffectofeachpileandtheentiregroupisnotimmediatelyapparent.Tokeepatightprogramme,onepossiblesolutionmighthavebeentostartthepilingfrombothendsofthesite.

10.2 TOWER BLOCK

An underground car park was to be added to one of several residentialtowerblocks.Excavationcommencedonthesouthsideoftheblockandthe excavated soil was piled on the north side to a height of 10 m (seeFigure 10.4).Aheavyrainfalloccurredsometimeaftertheexcavationfin-ishedanditincreasedthelateralgroundpressureonthepiles.Thebuildingbegan tomove sideways toward theexcavation.This caused thepiles tofailastheirshearresistancewasexceeded(seeFigure 10.5).Thebuildingbecameunstableasitmovedtowardtheexcavationandeventuallyfellover(seeFigure 10.6).

Itwasveryfortunatethattheothertowerblockswerenotcloseenoughtocauseadominoeffect.Figures 10.7and10.8showhowclosethistowerblockwasfromitsneighbours.Figure 10.8showsaclose-upviewofthepilesthatfailedinshear.Thepilesweremadeofcircularhollowunrein-forcedconcrete.Theywereconstructedwithshortlengthsatthetopwithsolidreinforcedsections(seeFigure 10.9).

Comment—Thereappearstohavebeennoengineeringplanningtopre-ventthecombinationofexcavationandpilingupofthesoilonopposite


sidesofthebuilding.Somehandsketchesshowingtheintentionsoftheplanwouldhavehelpedidentifythelikelyproblem.

Itwasprobablyreasonableforthepilestobedesignedforcompressiononlysincetheloadingtowhichthefoundationsweresubjectwasbeyondthescopeofthedesign.

Figure 10.4 Diagrammatic view after excavation.

Figure 10.5 Diagrammatic view of pile shear failure.


Figure 10.6 Diagrammatic view of collapse.

Figure 10.7 View of underside of collapsed building. (Courtesy of Gillian Whittle; redrawn from Reuters, http://www.telegraph.co.uk/news/worldnews/ asia/china/ 5685963/ Nine-held-over-Shanghai-building-collapse.html.)


Figure 10.8 View of piles that failed in shear. (Courtesy of Gillian Whittle; redrawn from Reuters, http://www.telegraph.co.uk/news/worldnews/asia/china/5685963/Nine- held-over-Shanghai-building-collapse.html.)

Figure 10.9 Unreinforced piles with short reinforced length at tops. (Courtesy of Gillian Whittle; redrawn from Reuters, http://www.telegraph.co.uk/news/world-news/asia/china/5685963/Nine-held-over-Shanghai-building-collapse.html.)

111

Chapter 11

Problems and Failures due to Deliberate Malpractice

11.1 FLOOR WITH EXCESSIVE DEFLECTION

Thebuildinginquestionwasatelephoneexchangebuiltinthemid-1970s,10yearsearlier.Figure 11.1showsaplanandsectionofatypicalendbay.Theslabhadbeendesignedassinglewayspanningbetweentwoshallowhaunchedbeams.Thedesignincludedaspanof9mwithaslabonly250mmthickwhichmanyengineerswouldconsidertoothin.

Ten years after the building had been completed, the operators com-plainedthatthedeflectionwasstillincreasingandhadcausedsomeoftheswitchgeartobecomefaulty.Thedesignersaskedforasecondopiniononthedesignoftheslab.Thecalculationsanddrawingswerecheckedandnomajorflawswerefound.Itwasconceivablethatcreepandshrinkageeffectswerestillincreasing.

The designers also asked the local university to run an independentcheck using its finite element modelling package.The subsequent reportconcluded that there was some major overstressing that could lead to ashear failure. The finite element analysis assumed elastic behaviour thatledtohighstressesatthesupports.Thiswasemphasisedbytheshallowsupporting beams that allowed the slab to behave more like a flat slab.However,ifareasonableamountofmomentredistributionwasassumed,althoughsomecrackingoftheconcretecouldbeexpected,thedesignwassafe.Neverthelessitwasdifficulttoconvincethedesignerthatthiswassoandavisittothesitewasarranged.

Thevisittositeincludedtheinspectionoftheslabclosetoacolumn.Thescreedhadbeenremovedtoexposethetopsurfaceofthestructuralslab.On close inspection, the reinforcement appeared tobebadly rusted andbreakingthroughthetopsurface.Asacrudecheckofthehardnessoftheconcretesurface,itwasscratchedwithapenknife.Quiteunexpectedly,thebladeoftheknifepenetratedintotheconcretesurfacerightuptothehilt!Afurthercheckofthesoffitoftheslabgaveasimilarresult.

Anadditionalinterestingfeatureofthesoffitwasthepresenceofanum-ber of shallow disc shaped (‘flying saucer’) pieces of concrete (150 mm


diameter)thatwereseparatingfromthesurface.Onesuchpiececameawayasitwasbeingexamined.Althoughtheslabhadbeendesignedtospanoneway, thesupportingbeamwassufficientlyflexible for theslabtobehavemorelikeaflatslab.The‘flyingsaucers’appearedinthecompressionareasofthesoffitandwereconsideredtobetheeffectsofspalling.Itwasclearthattheslabinquestionrequiredimmediateadditionalsupportandtherestofthebuildingrequiredcoretesting.

Aftercoreshadbeentakenthroughoutthebuilding, itwasdiscoveredthattheconcretecubestrengththatshouldhavebeen25MPawasonaver-age only 5MPa. The subcontractor had deliberately reduced the cementcontentinthespecifiedmix.Majorremedialworkfollowed.

Comment — It is very surprising that the building structure remainedintactforsolonganddemonstratesthat,ifthereisapossibleforcepathtoholdastructuretogether,thestructurewillfindit.Theclientwasrelievedtolearntherealcauseoftheproblems.Itisremarkablethatthetelephoneexchangeremainedinoperationthroughouttheperiodofremedialwork.

250 thick slab

Excessive deflection(still increasing after 10 years)

9 mA

AA - A

600

300

Figure 11.1 Plan and section of typical end bay.

Problems and Failures due to Deliberate Malpractice 113

11.2 PILES FOR LARGE STRUCTURE

Overthirtyreinforcedconcretepiles(3mdiameter)wererequiredtosup-portacomplexstructure includinga3mthicktransferslabandseveraltowerblocksabove.Whentheconstructionofatowerblockabovereachedseveral storeys, it was discovered that the piles were less than half therequiredlength.

Manyideastoresolvethesituationwereconsidered,butintheendthechosensolutionwastoprovideasetofninesteelHsectionpilesaroundtheexistingpilesanddrivethemtotherequireddepth.TheHpileswerethenconnectedtotheexistingstructurethroughnewpilecaps(oneperoriginalpile).Tocarryout the remedialwork,oneof the intermediatebasementfloorshadtoberemovedtoallowthepilingrigstooperatewithsufficientheadroom.Access intothebasementwasgainedbycuttinga largeholethroughtheexistingperimeterretainingwall.

Comment—Thepileshorteningwasadeliberateactofthesubcontractorthatresultedinalongdelaytothecompletionoftheprojectatenormousextracost.

11.3 IN SITU COLUMNS SUPPORTING PRECAST BUILDING

Thisbuildingwasconstructedwithprecastelementsaboveground.Belowground,thefoundations,columns,andbeamswereconstructedinsitu(seeFigure 11.2).Constructionhad reached an advanced stagewhen cracksappearedintheinsitucolumnsjustbelowtheconnectionswiththepre-castcolumns.

TheconstructionoftheinsitucolumnsshouldhaveproceededasshowninFigure 11.3.Thecentralcircularhollowsection(CHS)dowelwascastintothetopoftheinsitucolumnreadytoreceivetheprecastcolumn.

Inordertocheckthattheconcreteinthebox-outwasinplaceasintended,holesweredrilledthrougheachface.Itwasdiscoveredthatthepolystyrenehadbeenleftinandonlyathinlayerofconcretehadbeenplacedatthetop(seeFigure 11.4).

Theloadintheprecastcolumnfromsevenfloorsabovetransferredtotheouterrimoftheinsitucolumn.Thisloadwasintendedtobetakenbythewholesectionoftheinsitucolumns.However,thethinlayerofconcreteatthetopofthebox-outandthepolystyrenebelowwerequiteincapableoftakinganysignificantload.Theoutershellofinsituconcretethathadtotaketheloadbecameoverstressedandconsequentlycracked,providingthefirstsignofimminentfailure(seeFigure 11.5).

Inordertorepairthetopsoftheinsitucolumns,theloadfromthepre-castbuildinghadtoberemoved.Thiswasachievedbyprovidingpropsand


jacksclosetotheexistingprecastcolumnsateachfloorlevelandcreatinganewloadpathtotheground.Thisreleasedtheloadontheinsitucolumnsbelowandallowedtherequiredremedialwork—reconstructionofthetopsoftheinsitucolumns—totakeplace.

Street level Transfer beams

Precast beams, columns and slabs

Existingretaining

wall

In situ beams andcolumns

See detail of column connection

Figure 11.2 Section through lower part of building.

Column reinforcedas normal

Column cast with largepolystyrene box-out

Polystyrene totallyremoved; CHS 114 diadowel cast in with freshconcrete filling box-out

Figure 11.3 Intended construction of top of in situ column.

Problems and Failures due to Deliberate Malpractice 115

Comment—Althoughtheactthatcausedtheproblemwasnotintendedtocausecollapseofthebuilding,itwas,intheauthor’sopinion,adeliberateacttoleavemostofthepolystyrenebox-out.Thiswasdonetosavetimeandeffortforthecontractor.Itwasextremelyfortunatethatthefaultwasfoundbeforecollapseofthebuildingoccurred.

Only thin layer ofconcrete cast in topof column

CHS dowel pushedinto polystyrene

Existing insitucolumn

Only top layer ofpolystyrene removed

Figure 11.4 Actual construction of top of in situ column.

Load from 7floors above

Precast column

Grouting tube

Load from precast unitsupported on thin outershell of insitu column

Insitu column

Severe cracking ofinsitu column wall

Figure 11.5 Cracking of outer rim of in situ column.

117

Chapter 12

Problems Arising from the Procurement Process

12.1 EFFECTS OF DIFFERENT FORMS OF CONTRACTS

Typicallytheformofprocurementliesbetweenthetwoextremes:

• Traditionalarchitect-ledcontracts(e.g.,oneoff)• Design–buildcontractor-ledcontracts(e.g.,systembuild)

Theformerprovidestheclientandarchitectwiththegreatestfreedomoflayoutandform.Thelattercanleadtothemosteconomicandfastestcom-pletiontime.Bothrelyontheintegrityandcareofthewholeteam.

Therearesignificantdifferencesinsomeaspectsofthedesignanddetail-ing between these two forms of contracts. The reasons are not difficulttounderstandbuttheyhaveamarkedeffectonthewaytheinformationisproduced.Whereaproject isarchitect led, thecontractor thatwillbeawardedtheworkisnotknownatthetimeofthedesign.Thismeansthatthedesignanddetailing shouldbecarriedoutwithout involvingpropri-etarysystemsunlesstheyarerequiredbythedesign.Thisisbecauseeachcontractorislikelytofavouraparticularproprietarysystemthatisnotpre-ferredbyothers.Thedetailingshouldincludeonlyreinforcementrequiredbythedesignandnotbecontrolledbypossiblesiterequirements(suchashealthandsafety).Thiscanbebestexplainedthroughexamples.

Example1—Onemethodthatcontractorsusetoresolvethedangerthatpeoplewilltripoverslabreinforcementbeforeandduringplacingofcon-creteistoplacetopreinforcingbarsatsmallcentres(say,200mmorless).Thisisoftennotadesignrequirement.Infact,veryoftenthereisnodesignneedfortopreinforcementinthemiddleareaofaslab.

Hence, if the drawings show only the reinforcement required for thedesign,thecontractorwillhavetodecidewhethertoproposeextrarein-forcement or provide some other method of ensuring safe passage forpeople(suchasprovidingtemporaryplanksandwalkways).Theclientisinterestedinthecheapestmethod,notnecessarilythemostconvenientone


for thecontractor.All toooften, theextrareinforcementalongwith theextra cost is assumed tobe a requirementofdesignwhen theremaybecheaperwaysofachievingtherequiredsafety.

Example2—Anothercommonexampleconcernstheprotectionofcol-umnreinforcementstarterbars.Healthandsafetyregulationshaverequiredtheendsofthestarterbarstobebentintohooks.Thisiscostlytoachieveandacheaperalternativewouldbetoplaceanopenplasticboxoverthetopofeachgroupofcolumnbars.Theboxescouldbereusedforeachfloor.

Example3—Asignificantdifferencebetweenthetwoformsofcontractsconcernsthepositionsofconstructionjoints.Forbuildingstructures,goodpracticeensuresthatreinforcementislappedatpositionsoflowstress.Thelapandanchoragelengthsusedbydetailersareoftensetbyasimple“ruleof thumb.” Often40×bar diameter is used.This is consideredbymostengineerstobeasafeapproachandnormallynofurtherchecksaremade.

However,somecontractorsrealisedthatthisisapointwheresavingscanbemadeandfordesign–buildcontracts,theyinstructthedetailertouse35×bardiameter.Thiswouldprobablybeacceptable fora traditional typeofcontractwherethespecificationdemandsthatconstructionjointsbeposi-tionedawayfromhighlystressedareas.However,fordesign–buildcontracts,acontractoroftendecidestopositionconstructionjointsatpositionscon-venientforconstruction—wherereinforcementmaybehighlystressed.Therequiredlaplengthinsuchapositioncouldbeashighas60×bardiameter.

Thiscausesadilemmaforadetailer.Shouldheorsheuse60×bardiam-eteranchorageandlaplengthsfordesign–buildcontractstoensuresafetyinallsituations?

12.2 WORKMANSHIP

Thequalityofworkmanshipislikelytoremainthegreatestissuewithintheindustryfortheforeseeablefutureandwillrelyontheintentandenthusi-asmofthemanagerstoimprovestandards.Itisnotjustaquestionoftick-ingboxesonaform.Itisaboutensuringthattheengineeringofeachpartofajobisunderstoodandcarriedoutcorrectly.Itrequiresindividualstotakeontheownershipandresponsibilityfortheexecutionofgoodqualitywork.Workmanshipqualityisstronglyaffectedby:

• Anintelligentworkforceandrequirementforchecking• Achievingindividualsatisfactionandprideinwork• Utilizingsensibleprocedurestoavoidmistakes

Onecriticismlevelledagainstthetraditionalformofcontractisthatmanydesignsare tooconservativeandasa result concreteconstruction is lesseconomicalthanconstructionusingothermaterials.Itiscertainlytruethat

Problems Arising from the Procurement Process 119

manydesignerserrontheconservativesidewhensizingelements.Thishascomeabout frompast experiencewhen itbecamecommon forarchitectclientstomakechangeslateoninthejob,leadingtoincreasedloadingorchangesinstructure(movingcolumnpositions,addinglargeholes,etc.).

Duringtheconstructionphase,acontractormayseewaystoreducehiscostsandoftenthisispossiblebecauseoftheconservatismofthedesign.Forastandardformofstructure(e.g.,officeorresidentialblock),thiscon-servatismcanbeconsideredinefficientandwasteful.

Attheotherextreme,forsomedesign–buildcontracts,theworkmanshipisconsideredtobeoflowquality.Occasionallymethodsadoptedforapar-ticulartypeofconstruction(e.g.,hybridcarparks)areextrapolatedforlon-gerspansandareillthoughtout,sometimesproducinganunsafestructure.

It isnotuncommonforadesign–buildcontracttousehybridconcreteconstruction(insituandprecastconcrete).Thismaywellleadtothedesignoftheindividualelementsbydesignersworkingfordifferentcompanies.In such situations it is essential that there should be a single responsibility of one engineer for the stability of the structure, and the compatibility of the design and details of the parts and components, even where some or all of the design, including the details of those parts and components, are not carried out by this engineer.

The lackof systematicor thirdpartycheckinghas led topoorand insomecasesunsafeconstruction(e.g.,ungroutedprestressingducts).Thirdparty checking shouldbe carriedoutwhere considerednecessaryby theengineer and, in the author’s opinion, should be incorporated more fre-quentlyinallformsofcontract.

12.3 CHECKING CONSTRUCTION

Ononemajorproject, in situpost-tensioningwas required for thefloorslabs.Anindependentresidentengineerwasnot includedinthecontractanditwasassumedthatthecontractorwouldprovidesufficientsupervisionofthework.

Afterconstruction,waterwasdiscovereddrippingfromthesoffitoftheslab. Investigationrevealedthat theprestressingduct justabove the leakhadnotbeengrouted.Afurther investigationrequiredthatall theductsinthebuildinghadtobecheckedandthisrevealedthatmanyotherductshad not been grouted. The possibility of stress corrosion and failure ofprestressingtendonscanbegreatlyincreasedbythepresenceofwateraswasthecasewiththeseungroutedducts.Theremedialworktoresolvethismistakewascostlyandtimeconsuming.

Comment—Inaprestressedfloor,theamountofreinforcementisnegli-giblecomparedtoafloorwithreinforcementonly.Thespacingoftendons


canbeupto2or3m.Ifonetendonweretofail,upto6mwidthofslabwouldremainunsupported.Theriskofcollapseoftheslabandprogressivecollapseof thewhole structurewouldbe significant.There isagrowingconcernamongstmanyengineersthatthequalityofdesignandconstruc-tionhasdeclinedasaresultoftheincreasingpressuretocutthetimeandcostofprojects.

IthasbeenreportedtoCARES13thatasurveyhasshownthatanumberofducts,approaching1%onaverage,havebeeneitherpartiallyorcom-pletelyungrouted,therebypotentiallyaffectingtheintegrityofanumberofbuildingsinthelongterm.Thisisanaveragefigure,anditislikelythattheaveragehasbeenexceededgreatlyinanumberofstructures.

Inmany situations, the checkingof constructionworkhas reduced tounacceptablelevels.Theremustbeabalancebetweenrigorousproceduresandtheamountofcheckingrequired.Thepresenttrendistoincreasetherequirementswithin the specifications.Theauthor isnot convinced thatthis issufficientorthebestwayto improvethequalityofwork.Havingathirdpartycheckforcriticalpartsofastructurewouldappeartobeasensiblesolution.

121

Chapter 13

Contributions of Research and Development toward Avoidance of Failures

13.1 LINKS BETWEEN PRACTICE AND RESEARCH

Theneedforresearchanddevelopmentcontinuestoincreaseasmaterials,designs, and constructionmethods are refined. It is very important thatgood communication links aremaintainedbetween engineeringpracticeandresearchinstitutions.Thisisoftenbestachievedthroughpersonalcon-tacts. It isunfortunatethatsuch linksarebecoming lesscommonintheUK,probablyduetoreductionsinavailablefunds.Thefollowingexamplesdescribehowtheauthorhasbeeninvolvedinsuchlinks.

13.2 FLAT SLAB BEHAVIOUR

Inthelate1970s,flatslabconstructionwasbecomingpopularasitresultedinthinnerslabs.Italsoallowedsimpleandquickmeansofconstruction.However,designerswerelookingformoretoolstohelpthemanalysesuchstructuresanddemandingmoreinformationabouttheirstrengthandper-formance. The author became involved in a research project to test flatslabsatthePolytechnicofCentralLondon.Thisresultedfromquestionsabout thebehaviourand strengthof this formof construction.Dr.PaulRegansetupthetests(seeFigure 13.1)andproducedagroundbreakingresearchreport forCIRIA(Report89:Behaviourof reinforcedconcreteflatslabs14).

Oneimportantareaofdoubthadbeenpunchingshearbehaviour.Regan’stestsprovidedvaluableinformationonthissubject,takingintoaccounttheeffectofmomenttransferbetweenslabandcolumn.ThisinformationhassincebeendevelopedandincludedinboththeUKandEuropeancodesofpractice.


13.3 SPAN AND EFFECTIVE DEPTH RATIOS FOR SLABS

BS 81101 and previous UK codes of practice provided clauses on spanand effective depth. Modification factors were provided for tension andcompressionreinforcement.In1994,duringthedraftingoftheConcreteSociety’sTechnicalReport49:Designguideforhighstrengthconcrete,15itwasrealisedthatsomeconcretepropertiesimprovedwithstrength.Theseincludedanincreasedmodulusofelasticity,increaseofcrackingmoment,andreductionsinshrinkageandcreep.

Professor Beeby, who provided the supporting information for theexistingcodeclauses,wasaskedwhether itwasappropriate tomakeanadjustmenttothespanand/oreffectivevalueforthestrengthofconcrete.Afterfurtherresearch,heprovidedanexpressiontoincludethisvariable.Figure 13.2showsaplotoftheexpressionthatwasadoptedinthetechnicalreportandlater,in2004,wasalsoincludedwithinEurocode2.2

13.4 BEAM AND COLUMN JOINTS

TestswerecarriedoutatDurhamUniversitytodeterminethestressesinboththeconcreteandthereinforcementastheloadwasincreasedtofail-ure.Someof the reinforcingbarshadspecial treatment to inserta largenumberofstraingaugesintoslotsthroughthecentresofthebars.Eachbarconsistedoftwoseparatebarsmilledlongitudinallysothatcross-sections

Figure 13.1 Flat slab tests carried out at Polytechnic of Central London. (Courtesy of Jonathan Wood.)

Contributions of Research and Development toward Avoidance of Failures 123

werehalved.Aslotwascutdownthecentreofeachofbarandthestraingaugesplaced alongonebar at varyingpitches (as little as50mm)andgluedintoposition.Thetwohalveswerethenjoinedtosimulateasinglecompletebar.Allthewiresfromthestraingaugeswereledoutoftheendsofthebarstoacontrolbox.Figure 13.3showsoneofthetestsatthepointwherethejointfailed.

Differentarrangementsofreinforcementweredetailedandtested.Oneimportant result showed that if thebeam topbarswerebentupward inthe column, with gravity loading on the beam, the joint resistance wasdrasticallyreducedcomparedwiththebarsbentdown.ThereasonisthatthehighcompressionforcecreatedwithinthejointwasnotresistedbytheenclosingbarasshowninFigure 13.3a.

13.5 TENSION STIFFENING OF CONCRETE

The importance of time effects of tension stiffening of concrete becameapparentaftertestingwascarriedoutatLeedsandDurhamUniversities.ThisworkwasundertakenasaresultofsomeunansweredquestionsfromaBriteEuramprojectonhighstrengthconcretecarriedoutin1995.Aspartof thatprojectTaywoodEngineeringbuilt a full-scale thinflat slab (seeFigures 13.4and13.5).

10

15

20

25

30

35

40

45

50

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

Practical limit

C30/37 C50/60 C70/85 C90/105C20/25 C40/50 C60/75 C80/95

l/d

100 As,rqd/bd (mean value for span)

BS8110

Figure 13.2 Effect of concrete strength on span and effective depth ratios.


Thebehaviouroftheslabwascarefullymonitoredduringitsearlylifeandtheauthorcomparedthemeasurementswiththeresultsfromafiniteelementmodel.Theresultscorrespondedverycloselyuptoandjustafterthefulltestloadhadbeenapplied.Itcameasasurprisetodiscoverthat2weeksafterunloadingtheslab,itdeflectedmorethanthemodelhadpre-dicted.Dr.RobertVollumofImperialCollegesetupanindependentmodelbuttheresultsdidnotexplainthedifferences.

(b) Failure(a) Severe cracking

Figure 13.3 Loading column joint to failure. (Courtesy of Richard Scott.)

3 m

3 m

9 m 250 thick solid slab

300 × 300 columns

Figure 13.4 Layout of Taywood test flat slab.

Contributions of Research and Development toward Avoidance of Failures 125

Thesolutiontothisproblemcameaboutasaresultofachancemeet-ingwithProf.AndrewBeebytwoyearslater.HesuggestedthatheandDr.RichardScottcarryout testsatLeedsandDurhamUniversities toexaminetheconcretetensionstiffeningeffectsinmoredetail.Theproj-ectwasanunmitigated successand resolved theproblemwith respectto theTaywoodEngineering slab. It revealed thatwhathadbeencon-sidered short term effects of tension stiffening lasted much less timethanassumedandthatthelongtermeffectstookoverafterafewdays.Figure 13.6showshowtensionstiffeningaffectsthebehaviourofacon-creteelement.

Theknock-oneffectofthatprojecthasbeenasignificantchangeintheapplicationoftheBritishandEuropeancodes.Eachofthesecodesstatedthattensionstiffeningishalvedinthelongterm.Untilrecently,longterm

Figure 13.5 Construction of test flat slab. (Courtesy of Richard Scott.)

Calculated assumingconcrete has notensile strength

Calculated assumingno cracking

Actual behaviour

Deflection

Load

Figure 13.6 Effect of concrete tension stiffening on deflection of slabs.


was assumed to be several years after loading instead of the actual fewdays. This means that for design purposes the long term value shouldalmostalwaysbeused.

This research has provided an advance in the understanding of thebehaviourofreinforcedconcretewhenitcracks.ThereportontheprojecthasbeenconvertedtoConcreteSocietyTechnicalReport59:Influenceoftensionstiffeningondeflectionofreinforcedconcretestructures.16

127

References

1. BritishStandardsInstitution.BS8110-1Thestructuraluseofconcrete.Part1:Codeofpracticefordesignandconstruction,London,1997.

2. BritishStandardsInstitution.BSEN1992-1-1:Eurocode2,Designofconcretestructures.Generalrulesandrulesforbuildings,London,2004.

3. ConstructionIndustryResearchandInformationAssociation.Report91:Early-agethermalcrackcontrolinconcrete,1992.ReplacedbypublicationC660in2007.

4. WhittleR.andTaylorH.Designofhybridconcretebuildings,ConcreteCentre.2009.

5. ConcreteSociety.TechnicalReport67:Movement, restraintandcracking inconcretestructures,2008.

6. ConcreteSociety.TechnicalReport64:Guidetothedesignofreinforcedcon-creteflatslabs,2007.

7. BritishStandardsInstitution.BSEN1994-1-1:Eurocode4,Designofcompos-itesteelandconcretestructures.Generalrulesandrulesforbuildings,London,2004.

8. BritishStandardsInstitution.CP114:Thestructuraluseofreinforcedconcreteinbuildings,London,1957.

9. RoweR.E.,CranstonW.B., andBestB.C.Newconcepts in thedesignofstructuralconcrete.StructuralEngineer,43,399–403,1965.

10. BritishStandardsInstitution.CP110:Thestructuraluseofconcrete,London,1972.

11. Department for Communities and Local Government. Building regulations,5thamendment,HMSO,London,1970.

12. BritishStandardsInstitution.CP116:Thestructuraluseofprecastconcrete,London,1965(revised1970).

13. UnitedKingdomCertificationAuthorityforReinforcingSteels(CARES). 14. ConstructionIndustryResearchandInformationAssociation(CIRIA).Report

89:Behaviourofreinforcedconcreteflatslabs,1981. 15. ConcreteSociety.TechnicalReport49:Designguideforhighstrengthconcrete,

1998. 16. ConcreteSociety.TechnicalReport59:Influenceoftensionstiffeningondeflec-

tionofreinforcedconcretestructures,2004.

Robin Whittle


A S P O N B O O K

ISBN: 978-0-415-56701-5

9 780415 567015

9 0 0 0 0

Y106141

Cover image: © Dr Jonathan G M Wood

STRUCTURAL ENGINEERING

“The book provides a fascinating insight into what can go wrong. It illus-trates the importance of approaching problems from first principles and identifying the key load paths, loads, etc. This type of material rarely sees the light of day for reasons of litigation and commercial confidentiality.” —Robert Vollum, Imperial College, London

“This book is essential reading for any experienced engineer and of-fers insight into how designers and builders accidentally create struc-tures that fail. This book both explains how things fall down and what was done or could have been done to overcome these errors. Learning from errors is an essential part of any engineer’s knowledge and expertise.” —David M. Scott, Laing O’Rourke, Connecticut

Failures in Concrete Structures: Case Studies in Reinforced and Prestressed Concrete presents a selection of the author’s firsthand experience with incidents related to reinforced and prestressed concrete structures. He includes mistakes discovered at the design stage, ones that led to failures, and some that involved partial structure collapse—all of which required remedial action to ensure safety. The book focuses on specific incidents that occurred at various points in the construction process, including mistakes related to structural misunderstanding, extrapolation of codes of practice, poor detailing, poor construction, and deliberate malpractice. The author provides this information for readers to gain an understanding of errors that can occur in order to avoid making them. The book also examines issues during the procurement process, as well as the importance of research and development to gain a deeper knowledge of materials and help prevent future failures.


Y106141_Cover_mech.indd 1 8/24/12 2:11 PM

Documents

Failures in Concrete Structures - · PDF fileRobin Whittle Failures in ConCrete struCtures Case Studies in Reinforced and Prestressed Concrete A SPON BOOK ISBN: 978-0-415-56701-5 9