EXPERIMENTALANDNUMERICALANALYSISOFLADLETEEMINGPROCESSGastnM.MazzaferroA,1,MarceloPivaB,2,SergioP.FerroA,3,PabloBissioB,4,MarinaIglesiasB,AdrianaCalvoB,5,MarcelaB.GoldschmitA,6.ACenterforIndustrialResearch,FUDETEC.Dr.Simini250,2804Campana,ARGENTINA.BPorousMediaGroup,EngineeringSchool,UniversidaddeBuenosAires,PaseoColn850,1063,BuenosAires,ARGENTINA.1E-mail:2E-mail:mpivafi.uba.ar3E-mail:4E-mail:.ar5E-mail:acalvofi.uba.ar6E-mail:ABSTRACTInthecontinuouscasting,themoltensteelispouredfromtheladletothetundishthroughanozzlelocatedatthebottomoftheladle.Thisprocess,however,mustbestoppedbeforetheladleiscompletelyemptiedtoavoidslagcarryovertothetundish.Theamountofsteelthatremainsunteemedintheladleisusuallysignificantsosteelplantsarehighlyinterestedinstudyingdifferentwaystoimprovetheprocess.Inthepresentwork,experimentalstudiesonwatermodelsandnumericalsimulationsareusedtoanalyzetheconditionsneededforvortexformationandtheinfluenceofgeometricalandflowparametersintheamountofwastedsteel.Bothexperimentalandnumericalresultsleadtotheconclusionthatnovortexformationisexpectedtotakeplaceduringladledrainageunderindustrialconditions.1.INTRODUCTIONDuringthecontinuouscastingprocess,moltensteelispouredfromladletotundishthroughanozzlelocatedinaneccentricpositionofladlefloor.Aladleisavesselofgenerallycylindricalshapewithadiameterof2-3metersandaheightofapproximately3meters.Itcontainsbetween100and200tonsofliquidsteelwhicharedrainedthrougha5/10cmdiameternozzle.Moltensteelinladleiscoveredbyaslaglayer(whichthicknessvariesbetween5and20cm)whichpreventsitsoxidationbyaircontact.Asthedrainingprocessprogresses,theinterfacethatseparatesthesteelfromtheslag,eventuallydeflectstowardsthedrainagenozzleandadoptstheformofa“funnel”.Funnelformationleadstoslagcarryoverfromladletotundish.Thedrainageisstoppedwhenfirsttracesofslagaredetectedinthenozzle,leavingasignificantamountofsteel(upto3tons)unteemedintheladle.Funnelformationeffecthasbeenanalyzedbyseveralstudies1-7,generallybasedonexperimentsperformedonphysicalmodelswherewaterisusedinsteadofsteel.Intheseexperimentswaterdrainagefromcylindricalorsquareshapedvesselswithflathorizontalfloorsisstudiedfordifferentnozzlediameters(0.54cmto8cm)andeccentricities(upto0.7).Thesubjacentfluiddynamicsofthephenomenonhasalsobeenanalyzedbyseveralauthors8-14.Accordingtobibliography1-8twodifferentmechanismscanleadtothedeflectionofthesteelsurface:vortexsinkordrainsink.Vortexsinkischaracterizedbyhightangentialvelocitiesintheneighborhoodofthenozzle(seeleftschemeonFigures1and2)andcanbedevelopedevenwithahighcolumnofsteelintheladle.Vortexformationcantakeplaceinvesselswherethefluidhasimportantinitialtangentialvelocitiesespeciallyifthedischargenozzleiscentered.Theamountofliquidinladlewhenvortexsinkreachesthenozzledependsoninitialrotationalvelocityandonnozzleeccentricity.Ontheotherhand,thedrainsinkischaracterizedbyradialflow(rightschemeonFigures1and2)anddevelopsinthelaststageoftheteemingprocess,whenlessliquidsteelisleftintheladle.Drainsinkisalwayspresentattheendoftheprocessanddoesnotdependonthepreviousformationofavortexsink.Theheightoftheliquidcolumnleftunteemedinladlewhendrainsinkirruptsinthenozzle,isapproximatelyequaltothediameterofthenozzle.Drainsinkirruptioninthenozzle,unlikevortexsinkirruption,leadstosignificantproportionofslagcarryover1,4,6.Takingintoaccounttheseconsiderationsfoundinliterature,itseemsthatduetonozzleeccentricityandtheabsenceofpreferredrotationsense,vortexsinkisnotexpectedtotakeplaceundergeneralplantconditions.Theamountofsteelusuallydowngradedinplantsduetoslagcarryoverisconsistentwiththishypothesis.InthisworkwefirstpresentwatermodelexperimentsSection2carriedouttoconfirmthelastassumptionbyestimatingtheprobabilityofvortexformationinactualcastingpractice.InSection3wepresentnumericalsimulationsofindustrialladledrainageprocess,withfocusondrainsinkformationandpossibleinfluenceofladlefloorgeometry.AnanalysisofexperimentalandnumericalresultsispresentedinSection4andconclusionsarepresentedinSection5.2.WATERMODELEXPERIMENTSWatermodelexperimentswerecarriedoutbythePorousMediaGroup,attheEngineeringSchoolofBuenosAiresUniversity.TheexperimentalsetupusedtoanalyzetheflowofwaterduringthedrainingprocessispresentedinFigure3.AcylindricalcontainerofdiameterD=20cmispartiallyfilledwithwateruptoaconstantheightH0,throughtwotangentialpipeslocatedatoppositesidesofthecontainerbottomandat1cmofthewall.Thiswayoffillingprovidesthefluidthetangentialvelocitynecessarytoinducevortexformation.Waterisdrainedthroughanozzleofdiameterdn=0.5cmandlengthhn=10cmlocatedatthebottomofthevesselandatadistancerfromitscenter.Oncethefillingprocesshasconcludedthevelocityfielddecayswithtimeuntilthefluidremainsatrest.So,wecanselecttheinitialvelocityfieldbythewaitingtime,ti,betweentheendofthefillingandthebeginningofthedrainage.AssumingthattheflowfieldisessentiallytangentialandaxisymmetricwecharacterizetheinitialflowconditionthroughV,definedasthemaximuminitialtangentialvelocitymeasuredatthefreesurface.Oncethenozzleisopenthemassofwaterleavingtheladle,m,iscollectedinasecondvesselwhereitsweightismonitoredasafunctionoftimewithaprecisionbalance.Inthiswaytheflowrateatthenozzleoutput,Q=dm/dt,canbeobtained.ThecriticalheightHc,definedasthelevelofwaterintheladlewhenairirruptioninthenozzletakesplace,wasanalyzedasfunctionofthenozzleeccentricity=2r/D,andinitialtangentialvelocity,V.InFigure4aanexampleshowingthetimeevolutionoftheflowrateQthroughthenozzleispresented.ThefigurecorrespondstothedrainageofaH0=11cmwatercolumnthroughacenterednozzle(=0)andlargeinitialtangentialvelocity,V2.5cm/s.Itcanbeseenthattheinitialflowrate,Qi32g/s,smoothlydecreaseswithtimeuntilt=28s.Inthisperiodthefreesurfaceofthefluidremainsalmostflatexceptinthevicinityofthenozzleaxiswheretheformationofasmallfreesurfacedepression(ordimple)wasobservedafterfewseconds.ThetheoreticalflowratecalculatedfromBernoullilawwasfoundtofitquitewelltheexperimentaldataindicatingtheinviscidnatureofthedrainageprocessinthisstage.Theabruptdropintheflowrateattc28sindicatesthebeginningofthevortexsink.ThecriticalheightisHc=8.6cmwhilethedrainedpercentageisabout22%.Simultaneouslythefreesurfacedimpleevolvesdownwardtoformalongvortexfunnelwhichgoesthroughtheoutputproducingtheirruptionofairinthenozzle.Afterthiscriticaleventtheflowratemeanvaluedecreasessmoothlywithfluctuationsduetothepresenceofairinthenozzle.Inthefinalpartofthedrainage,t190s,itisobservedthatthefunneldisappears:theremainingcirculationisnotenoughtosustainthevortexsinkandthefreesurfaceofthefluidbecomesflat.ThisphenomenonoccursatHdn.Thenozzleisagainfilledwithwaterandtheflowrateincreasessuddenlyduringafewsecondsproducingasignificantdrainage.Then,thedrainsinkoccursatt=198s.InFigure4banewexampleisconsideredwithidenticalconditionsasthepreviouscasebutwithamuchlowerinitialtangentialvelocity(muchlongerwaitingtime).Inthiscasetheinitialcirculationisnotlargeenoughtoproduceavortex,theladledrainswiththefreesurfacealmostflatuntilt=109swhenthedrainsinkoccurs.Thedrainedpercentageisabout85%.ToanalyzetheinfluenceofnozzleeccentricityweperformedexperimentsinsimilarconditionsasinFigure4a(H0=11cmandlargeinitialtangentialvelocity)butwithtwodifferentvaluesofthenozzleeccentricity:=0.5and=0.9.InFigure4ctheresultsfor=0.5areshown,itcanbeseenthatavortexdevelopsatt=96s.ThecriticalheightisHc=1.6cmandthedrainedpercentageis86%.For=0.9theflowrateevolutionissimilarasthatshowninFigure4cthatis,novortexisobservedinspiteoftheconsiderableamountofinitialcirculation.Again,thedrainsinkoccursattheendofthedrainagewhenHdn.AcompletesetofresultsarepresentedinFigure5wherethecriticalheightHcisplottedasafunctionoftheinitialtangentialvelocityVfor=0(circles),=0.5(downtriangles)and=0.9(squares).Inagreementwithpreviousstudies1-8wefoundthattherollplayedbynozzleeccentricityistodelaythevortexformationandeventoinhibititforthelargestvalueof.Thus,forthelargesteccentricity(=0.9)novortexwasobservedinanyoftheconsideredvaluesofV.Amoredetailedanalysisoftheseexperimentscanbefoundelsewhere15.InordertorelatetheresultsjustshownwithatypicalfullscalemoltensteeldrainageonecanusethedimensionalanalysisproposedbyGuthrieetal8.TheyconcludethatthedimensionlesscriticalheightHc/H0dependsonlyontwohydrodynamicsparameters:aReynoldsnumberdefinedasRe=VoutH0/(