OF wELDED BUILT-UP COLUMNS - Digital Scholarship

1 WeldedBuilt-upColumnsEXPERIMENTALINVESTI GATIONOFwELDEDBUILT-UPCOLUMNSbyLambertTa llThisworkhasbeencarriedoutaspartofaninv estigationsponsoredjoint'lybytheColumnRe -searchCouncil,thePennsylvani~Department ofHighways, , (10"x10")boxshapecrosssection,andtheirbe havioriscom-paredwithmediumsize(9"xlO")H shapesandboxshapesandHshapesofsmallersiz esoTheinvestigationincludedtensilecoupon tests,residualstressmeasurements,stubcol umntests, SABSTRACTINTRODUCTION101 ScopeofStudy102 FactorsInfluencing"Column,StrengthPRELIM INARYINVESTIGATION201 CouponTests202 ResidualStressMeasurements203 StubColumnTests, 'Test 'sion40SU:MMARYANDC' ,..~ , ~ ~ ~ ,themajorfactorsareasfollows~1)themagnit udeanddistributionofresidualstress, )thebasic(static)yieldstresslevel,3)thes trainhardeningmodulus(forshortcolumns),a nd4) ,2, NAR,YIN ,preliminarytestsweremadewhichincludedte nsilecouponteststoobtainthestaticyieldst resslevel, ,astress-straindiagram, ~asaroutinecheckon', ,speci-fidd (4) ~ , '''sectioning,,(5) (andconsequentlyresidual-4stresses) , ,ameasureofresidualstress",Figure2showsa typicallayoutforthesectip~ ~chsectioncutfrom,thebeam'isatasufficien tdistancefromthe(1)endstooffsetanyedgeef fectoReference5alsoshowsthatanedgeeffect doesexist,theresidualstressesbeingu~ 'efollowingfactorswereconcerned.

2 1) ,2)theeffectofdiffere~tedgepreparationso ftheplatebeforewelding,and3)theeffectoft hick~ess,thatia,thevariationbetween, ~gthele~gthofthememberoResi-dualstresses weremeasuredatthosesections-marked~,.~an dQinTochecktheeffectofdiffeientplateedge preparationon,theresidualstressdistribut ion,twomethodsofprepari~gtheedgeofthejoi nts, ~erbymachiningorbyflamecutting0 Edgepreparationbymachiningismorelaboriou sand-5henceamoreexpensiveoperationnotnor mallyusedinfabrication~Inthiscase, ,allthetestpieces(exceptforsevenpieces) ,wouldintroduceachangeintheresidualstres sdistribution; ~g(refertoTable2) ~ "Figure4showsthesectionsusedinthestudyoS ection!andFweretakenfrompieceNoo11andsec tions .and!!frompieceNOe120 ThesectionswerecutintoLshapesaccordingto thedetailshowninthefigure0 Additionalgageholeswerelaidoutontheinsid efacesoftheL.

3 C~ttingoperatiouo2~3 StubColumnTestsPriortotnetestingofanycol umn,astubcolumntest'wasmadeon,asectionfr omthesamepiecefromwhichtheactualcolumnwa scuteThelengthofthestubcolumnissuchthatc olumninstability-6cannotoccur(6)butwassu fficientlylongtoretaintheoriginalresi-du alstressdistributionofthesection6 Thestubcolumntestgivesastress-strain,cur veshowingtheeffectofresidualstressoThepr opor-tionallimit,thestaticyieldstresslev elandtheelasticandtheelastic-plasti,cmod uliaretheimportantdatafurnishedbythecurv e0 Datafromthestubcolumntestisnecessaryfort heprediction,ofcolumnstrengthoThestubcol umnspecimensweretestedinan800,000lb, ,ofstressoFigure5showstheinstrumentation ofthestubcolumn6 Four1/1000inchdialgages(Noso1to4)areatta chedatthefourcornerstomeasurethestrainov erthewholelengthoTwo1/10,000inchdialgage saremountedonoppositesidestomeasurethest rainovera10"gagelengthatthemid-heightoTh isdataisusedtodeterminethestres's-strain relati,onship0 Thefourcornergagesareusedforalignment0 Thealignmentofthespecimenwasmadeatloadsn otexceedingone-thirdoftheexpectedyieldst resslevel, ", ,curveofthetestoAbovethe249~13-7proporti onallimit,aloadrelaxation,diagram(loadve rsustime)wasplottedasshownin, , ,aqualitativepictureoftheyieldpatterncou ldbeseenfrom, (hydratedlime) 0 LUM NT E ,weldedbuilt-upcolumnsincludedthreetests ofweldedHsections(9"x1/2"we'band9"x3/4"f langes)withslendernessratiosof59, (7).

4 1)We~dedH-shapedmembersmayhavehighcompre s-siveresidualstresseso2)Columnsbuilt-up byweldingwillcontaintensileresidualstres sesclosetotheyieldpoi~to3)Thestrengthoft heweldedH-columnislessthanthatoftherivet edoras-rolledcolumn0 Alatertest-withthesamecrosssectionb-utwi thaslendernessratioof1205indicatedthatth eultimatestrengthoft1:) ,,excessofthatgivenbytheyieldstresslevel ,failurebeingbylocalbuckling~8)Thecolumn specimenswerefabricatedfromstructuralste elofASTM designation,A7,universalmillplatesandacc ordingtothesche--dulegivenintheAppendix0 The!9-~ldingdetailsaregiveninFig1089 Thejointsweremachineweldedemployinganaut omaticfeedunionmeltofthesubmergedarctype 0 FortheIOux10"boxshape,afirstpassattheroo t_was'depositedmanually0 Inallcases;Jsmalltackweldswerefirst-8-de positedtofixtheshape0 Figures9,10,and11showtheprocessofflamecu ttingtheedges,fixingtheshapeinthejigwith tackwelds,andweldi~ ,exceptfour,weretestedinan800, ,CiO,ellandC12weretestedina5,000,000poun dhydraulic-typeuniversaltestingmachinelJ Thecolumnsweretestedwithpinned-endsuppor tsinthe"weakaxis" ~xed-e11-dsupportsintheperpendicularaxis 0 The.

5 "weakaxis"ofaweldedboxshap'eistheaxisper pendiculartothenarrowerplate(seeFigo12a) .Theendfixturesusedwerestandardcolumnfix turesdevelopedinthelaboratoryoThemaincyl indricalbearingwasdesignedsothattheradiu softhesurfacehaditscenteratthemid-pointo fthecolumnends~9). ,Itcanbeseenfromthisfigurethatatanystage ofthetest)theloadpas-sesthroughthesamepo intIIBeforetesting,thefollowingpreparati onsweremadeonthecolumn~249013-101)Theext ernaldimensionsofthecolumnweremeasuredan dcheckedforanyvariation,not-within, )Thecolumnwaswhitewashedwith~~ ,white-washalsogivesanindication, )T~ ,mechanicalandSR-4gagestomeasurestrain,l evelbarstomeasureendrotationanddialgages tomeasurecross' ,about12incheslong, ~ ,ashortstripscalewasattachedtothefixedcr ossheadofthetestingmachi~etochecklateral move-mentofthetesting' ~ ,wasalsomeasuredatthemid-heightofthecolu mnwithafixed1/1000inchdialgageattachedwi thtaltt'J -thinwire'~.

6 Toasmallscrewtapped-inatthecenterlineoft hecolumn, , ,columnse6, ,astraingageonthreelO-inchgageholeslaido utononefaceofthecolumnandperpendicularto theaxisofbending, ~ , ~ ~ ~ , ~cethespecimenshadsomeslightinitialout-o f-straightness,it-wasalsonecessarytochec kthelateraldeflectionduringthealignmentp rocedureoBybalancingtheeccentricitybetwe entheendsandthemid-height, , ,theSR-4strain, ,apointbypointplotoftheload-deflection, ' (frommeasurementsofthelO-inchgageholes) ,alsoindicatedtheoccurrenceandprogressio nofyielding.,Asinthestubcolumn, ,thedeflec-tionreadingswereobservedcaref ullyinloadingthespecimentomakesurethatth epeak(ultimate)loadofthecolumn,wouldbecl earlydefinedin, ~edinTa,ble4."A,tota1of12columnsweretest edwithsle~ dernessratiosvaryi~ (>In,additiontoactualcolumntests,studies weremadealsoonmaterialproperties"residua l,stressdistribution, , ~ + 'takenfromtheweldrevealedthatthestaticyi eldstressofthedepositedweldmetal, ,thedepositedweldmetalrespectively,Figur e17isastress-straincurverecordedfromates tonastandard8-i~ 'istributionof16sectionsweremeasured.)

7 Oneeachforthe6"x6"boxshapeandthe6"x7"Hsh apeand14forthe10"x10" , ~ ,compressiveresidualstressesand-15abrupt lydecreasingandchangingtotensileresidual stressesattheedges0 Thecompressiveresidualstressatthecentero ftheplatewasabout30ksiandthetensileresid ualstressattheedgewasabout38ksioTheHshap ehadaresidualstresspatternsimilartothato fthestandardrolledshapes(l) ~ , ,esinweldedplatesIIRefer-ence2indicatedt hat,forweldedshapesbuil,t-upfromsimilarp lates',theeffectofrestraintis~ ,t~edistrib~tionofresidualstres ,beobtainedfromtheresi-dualstressdistrib ution,oftheseparatecomponeI1:tplatesoThe fol1owi~ "x10"boxshape~ :.,!!andCofcolumn"C5showedthatthereisnos ignifica~ , 'ses(amaximumof36ksi)ataboutthecenterlin eofthewelded(narrower)' (wider)platesshowedamoreevendistribution ofcompressiveresidualstress(about28ksi)o verthemiddlethree-quartersoftheplate,wit hhightensilestresses(about35kai) ,edgeoftheplatehadlittleeffectontheresid ualstressdis, ,theboxshapeshowedonlyaslig~ ) , ,21,and220 Figure20ist~eaverageofsixtestsmade' "X10" "x 611boxshapeandthe6"x7" "x10"boxshapewasabout710kips, ,itwasnotedthattheproportionallimitwas21 0kips(11ksi)wh~ ~ ,the6"x6"boxshapegaveaload-strain,curvev erysimilartothatofthe10"x1011,boxshape,e xceptforthefactthatst~ain,hardeni~g, , 'toflocalbucklingata~ 71'HshapewasaboutI' " ~Ilc~~destheslendernessratio,thecolumnst rength(P/Py)'a~dtheinitialout-of-straigh t~ ,elbof00001forC4toamaximumof0.

8 024forC6 DFigure23showsthevariationoftheinitialou t-af-straightnessalongthelengthofthecolu mnIITheloadvers,u'smid-heightdeflectionc urvesareshownin"Figs"24and250 Thetestcurvesshowthatthedeflectionsweren egligibleforthelowloads,asexpected,becau setheeccentricitywasbalanc'edbetweenth'e endsandthemid-heightinthealignmentproced ureIIAnotablefeatureoftheloaddeflectionc urvesistheunload-ingportion;forthelonger COLUMNS " ,aftertheulti-matestrengthisattained, ,wide'rra~geofdeflectionthan, ,curvesofcolumnsC6andCIO,bothofwhichhads lendernessratiosof80~ 'olumnsCIahadaslightlyhigherultimateload ( ) ,thetwocurvesweremoreorlesscoincident024 9013 Itmaybe ::' ~ ~ ,thecurvesofthethreefiberswouldcoincideu ptothepointofbifurcationQThiswasshownbyS hanleyinhis(11)classicalpaperoncolumnsfa ilingintheinelasticrange0 Thedataabovecanalsobeplottedinamannerass howninFig~.

9 (1)theinitialstateofuniformstress,(2)the occurrenceofstrainregressionpriortoattai ningultimateload,and(3)theinwardmovement ofthepointofzerostrainregression~ ~13-20atthemid-heightcrosssectionoTheeff ectofresidualstressistakenintoaccountbya ssuminganidealizedresidualstressdistribu -tionbasedontheactualmeasuredresidualstr essdistribution~(2)InFig~28theresultsoft hecolumntestsareplottedoThetangentmodulu sloadcurveandtheultimateloadcurveforthew elded(2)boxsectionarealsoshown0 One'hebasisoftheseresults,itcanbeseentha ttheultimateloadpredictionistoooptimi ticoTheexperimentalresultsshowthattheact ualultimateloadofcolumnswithmedium,slend ernessratiosappearstobeabout20%lessthant hepredictedvalues0 Thisdiscrepancyiscausedmainlybytheinitia lout-of-straightnessofthecolumnandthepre senceofnon-symmetricalresidualstressdist rtbutioninthesection~305 Discussion(a)ItwasshowninRef~2thatthethe oreticalanalysisdidnotgivesatisfactoryre sultsduetothepresenceofinitialout-of-str aightnessintheco1unmspecimensandtoalesse rdegree,duetotheassumptionofanidealizedr esidualstressdistributionq(b)

10 InFigQ293thecolumncurvesfortheeccentrica llyloadedboxcolumnwereplottedforeccentri cityratios,e/bequalto0001~nd04050 Acomparisonofthesecolumncurveswiththecol umncurvefortheaxiallyloadedcolumnshowsth everyappreciableeffectofinitialout-of-st raightnesso249013 Inallthecolumnstested,initialout-af-stra ightnesswaspresentoIf,theinitialout-af-s traightnessofeachtestcolumnis-21takenint oaccount,agoodcorrelationcanbeshowntoexi stbetweentheultimatestrengthofthetestcol umnandtheexpectedultimatestrengthofacolu mnwiththeequivalenteccentricityoAdiscuss ionofthisaspectofcolumnstrengthwillbepre sentedinafuturere-portonthetheoreticalst udyofweldedbuilt-upcolurnnso(c)Asfarasth ewelded10"xlOtiboxsectionswereconcerned, thedeviationfromthespecifieddimensionswa ssatisfactory,witha+maximumof_O~05inchoA scanbeseeninFigo23,thestraightnessofthec olumnwasnotperfect~However3thedeviations indimensionandst~aight~esswerewellwithin theallowabletolerancesgivenbythespecific ations(1956 AWSS pecification,Sec"507).