2005An investigation of the mechanism of liquid injection into fluidized beds

InjectionintoFluidizedBeds

StefanBruhnsandJoachimWerther

ChemicalEngineeringI,TechnicalUniversityHamburg–Harburg,D-21071Hamburg,Germany

DOI10.1002/aic.10336

PublishedonlineinWileyInterScience(www.interscience.wiley.com).

Themechanismofliquidinjectionintofluidizedbedreactorswasinvestigatedbyinjectingwaterandethanol,respectively,intoapilot-scalebubblingfluidizedbed(crosssection:1ϫ0.5m)thatwaskeptattemperaturesbetween120and180°C.QuartzsandandFCCcatalystwereusedasbedmaterials.Theinjectedliquidwasfoundtoformagglomerateswiththebedparticlesatthenozzleexitandbecometransportedintothebedinteriorbymixingofthelarge-scalesolids.Theretheliquidsevaporatefromthesurfaceoftheparticles.©2005AmericanInstituteofChemicalEngineersAIChEJ,51:766–775,2005

Introduction

Fluidizedbedreactorsarewidelyappliedinchemicalengi-neering.Theirapplicationsrangefromphysicaloperations,likeagglomerationandcoatingofparticles,tochemicalreactorsfortheheterogeneousgas-phasecatalysis.1Withrespecttothelatterapplication,someindustrial-scalereactorconceptsusetheliquidinjectionofreactants,suchasthefluidcatalyticcracking(FCC)process,2theanilinesynthesisbyBASF,3andtheproductionofpolyethyleneinthesuper-condensedmode.4Theliquidsareinjectedintothereactor,evaporateintheinteriorofthefluidizedbed,andthegeneratedgasesundergotherespectivecatalyticreactions.

Theliquidinjectionofreactantsprovidesmanyadvantagescomparedtotheinjectionofreactantsinthegaseousform.Becausetheliquidreactantsevaporateintheinteriorofthefluidizedbed,thistechnologyallowscostsavingsfortheevap-orationinexternalheatexchangers.Italsominimizestheriskofhot-spotformationnearthereactantfeedpointbyprovidinganefficientcoolingbythelatentheatofevaporation,especiallyinthecaseofstronglyexothermicreactions.Adetailedknowl-edgeofthemechanismofliquidinjectionintofluidizedbedreactorsisrequired,notonlyfortheengineeringdesignbutalsoforthesafeandeconomicoperationontheindustrialscale.Intheearlystagesofmodelingtheliquidinjectioninto

S.BruhnsiscurrentlyatBASFAG,GCT/T-L0,D-67056Ludwigshafen,Ger-many.

CorrespondenceconcerningthisarticleshouldbeaddressedtoJ.Wertheratwerther@tu-harburg.de.

©2005AmericanInstituteofChemicalEngineers

fluidizedbedreactors,itwasoftenassumedthattheinjectedliquidevaporatesinstantaneouslyatthenozzleexit.5Thechar-acteristicsofthesubsequentpenetrationwereassumedtobethesameasthoseofgasjets.Anotherapproachfoundintheliteratureistheassumptionoffeeddroplets,whichareinstan-taneouslygeneratedbyatomizationatthenozzleexit.Thesedropletsweremodeledtomoveandevaporateindividuallywithinthegasphaseofthedensegas–solidflowofthefluidizedbed.6–8Inthecaseoffluidizedbedgranulation,theinstanta-neousgenerationofdropletsisalsoassumedbutthesedropletsarethenassumedtohitthefluidizedbedsolidsandtoformaliquidlayerontheparticles’surface,whichevaporatesfromthere.9,10Severalexperimentalinvestigationshavedealtwiththein-jectionofliquidsintofluidizedbeds.Withrespecttofluidizedbedgranulation,SmithandNienow11detected,withtheaidofX-rayimaging,asprayingzonethatextended5cmfromthenozzleexitintothebed.Theyalsomeasureddifferencesbe-tweenthemeanbedtemperatureandthelocaltemperatureinthiszone.Basedonthis,MarongaandWnukowski12measurednotonlylocaltemperaturechangesbutalsochangesintheairhumiditythatwerecausedbytheliquidinjection.Theyiden-tifiedthreezones:(1)asprayingzonenearthenozzle,wheredropletsweregeneratedandparticleswerewettedbythesedroplets;(2)adryingzone,whereliquidsevaporatedfromthesurfaceofwettedparticles;and(3)aheat-transferzone,whereparticleswereheatedup.Similarresultswereobtainedbyotherresearchgroups.13Differentfromfluidized-bedgranulation,wherethemeantemperatureisnearorbelowtheboilingpointoftheinjected

Vol.51,No.3

AIChEJournal

766March2005

Figure1.Atomizationofliquidsingaseousenviron-ments.

AdaptedfromLefebvre.18liquid,fluidized-bedreactorsareusuallyoperatedathighertemperatures.IntheFCCprocess,theliquidfeedstockwithaboilingpointof200–300°Cisinjectedintotheriser,whichisoperatedat600–700°C.Therefore,Guetal.14usedrapidlyevaporatingnitrogenasatestliquid,andconfirmedtheexis-tenceofanevaporationzonewithintheriserbutdisprovedtheassumptionofinstantaneousevaporationatthenozzleexit.ThisfindingwaslaterverifiedbysophisticatedX-raymeasure-mentsbySkouby15andNewtonetal.16Anotherexampleisthefluidizedbedcokingprocess,whereKnapperetal.17investi-gatedthecoatingofatomizedbitumenfeedontosolidcokeparticles,usingasophisticatedtracertechnique,andfoundthatwettedagglomerateswereformedintheimmediatevicinityofthefeednozzle.

Evenwithouttheinteractionoftheinjectedliquidandthegas–solidflowinthefluidizedbeds,theatomizationofliquidsinagaseousenvironmentisalreadyacomplexprocess(cf.Figure1).Thesimplestideaisthattheliquidflowsoutofanorificeandformsaliquidjet.Withinafewcentimetersfromthenozzleexit,theliquidjetbecomesfragmentedbyinternalturbulenceandbytheshearwiththesurroundinggas.Liga-mentsaregeneratedinfragmentation,whichmayfurtherbreakdownintodrops.18Oncedropletsareformed,theyprovideahugesurfaceareaforheatandmasstransfer,whichresultsinfastevaporation.However,highmassfluxesofliquidsneedtobeprocessedintechnicalapplicationsusingso-calleddensesprays.Fordensesprays,thecharacteristicsofevaporationdiffersubstantiallyfromtheevaporationofidealizedsingledroplets,whichisattributednotonlytothehinderedformationofdropletsbyinteractionswithoneanotherbutalsototrans-portlimitationsinheatandvaportransferbetweenthedensespraycoreandthegaseousenvironment.19ThereforeZhuetal.20investigatedboth—theevaporationofaliquidnitrogenjetinco-currentgasflowonlyandthetran-sitiontowardgas–solidflowwithasolidvolumeconcentrationupto1.5%.TheextentoftheevaporationzonewasobservedbyCCDcamerarecordingsandtemperaturemeasurements.Wheninjectingliquidnitrogenataflowrateof0.5g/sintoagasflowat2.1m/s,theymeasuredanevaporationzoneuptoadistanceof12cmfromtheinjector.Theevaporationzonewasmarkedbytemperaturesneartheboilingpoint.Theau-thorsmeasuredadecreaseinthepenetrationlengthfrom12cm,withnosolidspresent,to7cm,withinaverydilutegas–solidflowof1.5%solidsvolumeconcentration.However,theseconditionsdiffersignificantlyfromthesituationattheinjectionlevelofFCCriser-typereactorswithsolidvolumeconcentrationsof20–30%.

Afterareviewofthepublishedexperimentaltechniques,itseemsthatasinglemeasurementtechniquealonecannotelu-AIChEJournal

March2005

cidatethecomplexprocesseswithinthethree-phaseflowstruc-ture.Inparticular,laboratory-scaleequipmentlimitstheextentoftheflowstructureandtheportabilitytoindustrial-scaleapplications.Therefore,inthepresentworkanexperimentalinvestigationonthemechanismsofliquidinjectionintofluid-izedbedreactorshasbeenperformedonthepilotscale.Theindividualprocesseshavebeenmonitorednotonlybymeasur-ingthelocaltemperaturesandthelocalconcentrationsofvaporizedfeedbutalsobymeasuringlocalsolidsconcentra-tionsanddetectingliquidswiththeaidofcapacitanceprobes.Thescopeofthepresentworkwasfocusedonansweringthefollowingquestions:

(1)Howdoestheliquidfeedpenetratethefluidizedbed—likeasprayorlikealiquidcolumn?

(2)Howdoestheliquidfeeddistributeinsidethefluidizedbed—byindividuallymovingliquiddropletsorbywettedparticles?

(3)Howstronglyistheflowstructureofthefluidizedbedaffectedbytheliquidinjectionandwhatistheinfluenceoftheoperatingconditions?

ExperimentalSetup

AnoverallsketchoftheexperimentalunitisshowninFigure2.Thekeyelementoftheexperimentalunitisthefluidizedbed.Thefluidizingairwassuppliedbyarootsblower,withamaximumvolumetricflowof2500m3/h.Anelectricalairheaterwasusedtoheatthefluidizingairatapowerof40kW.Toprovidemoreheatfortheevaporationofinjectedliquids,abundleofelectricalheatingrods,havingapowerof70kW,wasimmersedinthefluidizedbed.Athermocouplelocatedwithinthefluidizedbedwasusedfortemperaturecontrolofthebed.Thebedtemperaturewaslimitedto200°Cbecauseofthethermalstabilityofthesealingmaterial.Theapparatuswasthermallyinsulatedtominimizeheatlossesandtopreventthecondensationofliquidsatthewall.Thefluidizingairwascleanedinaninternalcycloneandwassubsequentlycooledinawater-cooledcondenser,wheretheinjectedliquidwascon-densedfordisposal.Theresultingsaturatedgasflowwasmixedwithambientairtoavoidcondensationinthefabricfilter.Finally,thegaswaspassedthroughaninductionfanandventedtotheatmosphere.

AmoredetailedsketchofthefluidizedbedapparatusisshowninFigure3.Thetotalheightoftheapparatuswasapproximately5m.Thefluidized-bedsectionhadacross-

Figure2.Experimentalsetup.

Vol.51,No.3

767

Figure3.Fluidized-bedapparatus.

sectionalareaof1ϫ0.5mandthefreeboardhadacross-sectionalareaof1ϫ1m.Theexpansionofthecrosssectionwasnecessarytoreducetheentrainmentoffluidizedparticlesandtoprovidesufficientspacefortheinternalcyclonewithoutacceleratingthegas–solidflowinthissection.Thecyclone,havingadiameterof0.85m,wasfixedatthetopoftheapparatus.Thediplegofthecyclonewasimmersedintothefluidizedbedforthedirectreturnofseparatedparticles.Thegasdistributorwasequippedwith40bubblecaps.Thetubebankof28U-shapedheatingrodswasmountedatlevelsbetween0.2and0.5mabovethegasdistributor.Theheatingrodswere9mmindiameter.Thespacingbetweentheheatingrodswaschosentobeverywide,atapproximately5cm,tominimizethedisturbanceoftheoverallgas–solidflowinthefluidizedbed.Theinjectornozzlewasmountedinthecenterofthefrontplateataheightof1.1mabovethegasdistributor.Duringtheexperimentsthefluidized-bedapparatuswasfilledwithsolidstoafixedbedheightof1.8m.Twotypesofbedsolidswereusedinthepresentexperimentalinvestigation:spentFCCcatalyst(Sautermeanparticlediameterdpϭ70␮m;minimumfluidizationvelocityumfϭ0.2cm/sinairatambientconditions)andquartzsand(dpϭ120␮m;umfϭ3cm/s).ThespentFCCcatalystwasprovidedbyDeutscheShellAG(Ham-burg,Germany).Thefluidized-bedunitwasoperatedinallexperimentsataconstantsuperficialgasvelocityof0.3m/s,whichiswellabovetheminimumfluidizationvelocityofbothbedmaterials.

Industrialapplicationsofliquidinjectionintofluidizedre-actorsprimarilyprocessorganicliquids,whicharedifficulttohandle.Theseliquidsareeithertoxicorinflammableandinmostcasesevenboth.Twotestliquids—waterandethanol—wereusedbecauseofeasyhandling.Themajorityofexperi-

mentswereperformedwithwater.However,amongtheprop-ertiesofwaterisitsextremelyhighheatofevaporationof2500kJ/kg,whichdifferssignificantlyfromheatsofevaporationoforganicliquidsintechnicalapplications.Theheatofevapora-tionofcrudeoilintheFCCprocess,forinstance,isapproxi-mately460kJ/kg.Thus,additionalexperimentswereper-formedwithethanol,whichhasalowerheatofevaporation,at730kJ/kg.

Differenttypesofinjectionnozzlesweretested.Character-isticsoftheinvestigatednozzlesarelistedinTable1.Two-fluidnozzleswereappliedwithafan-spraypatternandfull-conespraypattern,respectively.Besidesthetwo-fluidnozzles,single-fluidfan-spraynozzlesofdifferentsprayingangleswerealsotested.Theslotwidthsofthesingle-fluidnozzleswere0.25and0.8mm,respectively.Single-andtwo-fluidnozzlesdifferedinthesizeofthedropletgeneratedforthegivenoperatingconditions.Incaseofthesingle-fluidnozzlewith60°fan-spraypattern,ameandropletsizeof300–400␮mwasmeasuredbythesupplierfortheatomizationof40–100L/hwaterinagaseousenvironmentatambientconditions.Undersimilarconditions,theair-assistedtwo-fluidfannozzle(no.4inTable1)byLechleryieldedameandropletsizeof40–90␮m.Twodifferenttypesoftwo-fluidnozzleswereusedinthepresentexperimentalinvestigation:internallyandexternallymixednozzles.Internallymixedtwo-fluidnozzlesareoperatedsuchthattheliquidandtheatomizinggasarepremixedinsidethenozzle.Inthecaseofexternallymixednozzles,theliquidandtheatomizinggascomeintocontactdirectlyatthenozzleexit.ThedesignofthenozzlesusedisshowninFigure4.Anumberofdifferentmeasurementtechniqueswereappliedinthepresentexperimentalinvestigation.Fast-responsether-mocouplesof0.25mmdiameterwereusedforthemeasure-mentoflocaltemperaturesneartheinjectionnozzle.Thesethermocouplesregisteraninstantaneousriseintemperaturefrom20to100°C,witharesponsetimeof63ms.Gassuctionprobeswereappliedforthemeasurementoflocalgasconcen-trations:thatis,theconcentrationofvaporizedtestliquids(waterandethanol)andtheconcentrationoftracergas(carbondioxide),whichwasaddedtotheatomizinggasoftwo-fluidnozzles.

TheoverallarrangementofthemeasurementsystemforgasanalysisisschematicallyshowninFigure5.Gaswassuckedofflocallyinthefluidizedbedapparatusbyagas-suctionprobe.Thesuctionprobehadanouterdiameterof10mmandfiltertipofsinteredbronzepreventedthepenetrationofparti-clesintothegasanalyzer.Thetemperaturelevelwithinthetubes,thepump,andthegasanalyzerwascontrolledat150°Ctoavoidcondensation.Theconcentrationofvaporizedwaterorethanolwasdeterminedfromthepartialpressureofoxygeninthegassample(OxygenAnalyzer3001M,Fisher-Rosemount,Orrville,OH).Forthegasanalysisofcarbondioxide,acon-denserwasarrangedinfrontofthecarbondioxidegasanalyzer

Table1.DetailsoftheNozzlesUsed

Nozzle1234

BasicDesign

Single-fluidfanspray60°Single-fluidfanspray120°

Two-fluidexternallymixedfullconespray20°Two-fluidinternallymixedfanspray60°

Supplier

BETEFogNozzle,Inc.,Greenfield,MABETEFogNozzle,Inc.,Greenfield,MALechlerGmbH,Metzingen,GermanyLechlerGmbH,Metzingen,Germany

ModelNo.1/8NF036061/8NF031206156.326.16.11156.442.16.11

768March2005Vol.51,No.3AIChEJournal

Figure4.Designofthenozzlesused.

Single-fluidfannozzleofasprayingangleof60°(1)and120°(2);externallymixedtwo-fluidfull-conenozzlewithsprayingangleof20°(3);internallymixedtwo-fluidfannozzlewithsprayingangleof60°(4)(thedimensionsoftheannulargaparethesameasfortheexternallymixednozzle).

(BINOS,Leybold-HeraeusGmbH,Hu¨rth,Germany).There,thevaporcontentoftheinjectedtestliquidwasremovedbythecondensertoanalyzethegasconcentrationonadrybasis.Thecapacitancemeasurementsystemforthepresentexper-imentalinvestigationwaspreviouslyusedbyWiesendorfandWerther.21Basically,itconsistsofaneedlecapacitanceprobe(10mmindiameter),acapacitancemeasuringinstrument,andadata-acquisitionsystem.Thetwo-channelcapacitanceprobescanbeusedforthemeasurementoflocalsolidsvolumecon-centrationsaswellasthemeasurementofsolidsvelocitiesbythecross-correlationtechnique.Becausethedielectricconstantofliquidwaterisextremelyhigh,thecapacitanceprobescanalsobeusedforthelocaldetectionofliquidwaterwithinthedensegas–solidflowoffluidizedbeds.

ResultsandDiscussion

Inafirstseriesofexperiments,thefluidized-bedapparatuswasoperatedwithspentFCCcatalystasfluidizedbedsolids.

Figure5.Measurementsystemforgasanalysis.AIChEJournal

March2005

Thesuperficialgasvelocitywas0.3m/sandthebedtemper-aturewaskeptconstantat153°C.Waterwasinjectedataflowrateof50L/hwithasingle-fluidnozzleoffan-spraypatternandasprayingangleof60°.Thenozzlewaslocatedatthecenterofthefrontwall(xϭ0cm,yϭ0cm)ataheightofzϭ110cmabovethegasdistributor.Thetemperaturewasmea-suredwithafast-responsethermocouple.Thelocalmeantem-peraturesweredeterminedbyaveragingoverasufficientpe-riodof5–10min.Itshouldbenotedthatthemeasuredmeantemperatureisinfluencedbythevolume-specificheatsofin-jectedliquid,dry/wettedsolidparticles,fluidizinggasandvapor,andthelatentheatofevaporatingliquid.Thetempera-turemeasurementshereinaretakenasoneindicatoramongothersofthemechanismsgoverningtheinjectionofaliquidintoadensebubblingfluidizedbed.Amorein-depthinterpre-tationofthetemperaturemeasurementsispossibleonthebasisofamodelthatincludesliquidspreadingandevaporationaswellastheinteractionbetweenthesprayjetandthefluidizedbed.Suchamodelhasalsobeendevelopedwithintheframe-workofthedissertationofoneofthepresentauthors23andhasbeenpublishedelsewhere.24TheresultsareshowninFigure6,whereadashedlinemarksthenozzle’sidealsprayingangleof60°.Thelocalmeantemperatureinthefluidizedbedrapidlyrisesfromtheinlettemperatureof20to60°Cwithinthefirstfewmillimetersfromthenozzleexit.Withincreasingdistance,themeantemperaturesteeplyrisestotheboilingpointat100°Candreachesthefluidizedbedtemperatureof153°Catadistanceof10cmfromthenozzleexit.

Twointerestingphenomenacanbeobservedfromthemea-sureddistributionofmeantemperaturesinthefluidizedbed.

Vol.51,No.3

769

Figure6.Meantemperaturesnearthenozzleatz؍110

cm.

Single-fluidfannozzleno.1,sprayangle60°;50L/hwater.Liquidmomentumflux0.206N;FCCbedat153°C.

Figure8.Meantemperaturesalongthedistancexfrom

thenozzleatz؍110cmandy؍0cm.

Single-fluidfannozzleno.1andtwo-fluidfannozzleno.4,sprayingangle60°;50L/hwater;FCCbedat153°C.Theliquidmomentumfluxesofthetwonozzlesare0.206and0.125N,respectively;theblackarrowdenoteshereandinthefollowingfigurestheinlettemperatureoftheinjectedliquid.

First,theregionwheretemperaturedeviationsarecausedbytheliquidinjectionisonly10cmindepth.Second,thespatialpatternofthischangeinlocaltemperaturecorrespondsremark-ablywelltothenozzle’sidealsprayingangleof60°.Itseemsthatthesmallextentoftheinjectionzoneisattributabletothewell-knowntemperaturehomogeneityoffluidizedbeds,whichiscausedbytheintensegas–solidmixing.However,thecon-servationofthespraypatternisratherremarkableundertheconditionsofvigorousgas–solidsmotioninthefluidizedbed.Anotherexperimentwasperformedunderthesameoperat-ingconditionsbutwithafan-typesingle-fluidnozzleofadifferentsprayingangle,thatis,120°.Themeasureddistribu-tionoflocaltime-averagedtemperaturesinthefluidizedbedisshowninFigure7.Awidersprayingangleofthenozzleobviouslyleadstoawiderdistributionoftheliquidfeed.Asaconsequence,thepenetrationdepthbecamefurtherreducedtoadistanceof8cmfromthenozzleexit,whereatemperature

Figure7.Meantemperaturesnearthenozzleatz؍110

cm.

Single-fluidfannozzleno.2,sprayangle120°;50L/hwater.Liquidmomentumflux0.206N;FCCbedat153°C.

differencewasdetectablebetweenthemeanbedtemperatureof153°Candthemeasuredtemperature.Asindicatedbythedashedlineinthechart,thesprayingangleof120°isalsoconservedinthemeasuredtemperaturedistribution.Thiscon-servationofthespraypatternunderfluidized-bedconditionswasalsoobservedforothertypesofnozzles,suchassingle-andtwo-fluidnozzleswithconeandfan-spraypattern,respec-tively.

ThespraypatternshowninFigures6and7isinacharac-teristicwaydifferentfromwhatisknownforspraysinapuregasenvironment.Fortheatomizationinagaseousenviron-ment,theradialdistributionoftheliquidfluxisstronglyinfluencedbytheentrainmentofgasfromthesurrounding.Theentrainmentofgasresultsinatransportoffinedropletsintothecoreofthesprayand,consequently,alowfluxatthe“edge.”Asshowninthepresentexperiments,thesituationinafluid-izedbedisdifferentwheretheprocessisdominatedbythepresenceofadensegas–solidflow.Becausetheformationoffinedropletsishindered,whichinapuregasenvironmentwouldmigrateintothecore,thefluxismorepronouncedattheedgesunderfluidizedbedconditions.

Despitethefactthatthespraypatternsofdifferentnozzlescanbeidentifiedfromthemeasuredtemperaturedistribution,thereshouldbesomedifferenceinpenetrationdepthfortheinjectionwithsingle-fluidnozzlesandwithtwo-fluidnozzles,giventhedifferencesinthegenerateddropletsizedistribution.Acomparisonwasthusmadebetweentemperatureprofilesmeasuredfortheliquidinjectionwithsingle-fluidandtwo-fluidnozzles(cf.Figure8).Thetestliquidwaterwasinjectedataflowrateof50L/husingbothtypesofnozzleswithafan-spraypatternandanopeningangleof60°.Theinternallymixedtwo-fluidnozzlewassuppliedwithatomizingairof8.3and4.2m3/h,respectively.Theseairvolumeflowsareequivalenttomomentumfluxesof1.93and0.96N,respectively,wherethemomentumfluxwascalculatedastheproductofairexitvelocityandairmassflow.

Forthepresentlyusedflowrateof50L/hwaterthesingle-fluidnozzlegeneratesdropletswithameandiameterof400

Vol.51,No.3

AIChEJournal

770March2005

␮minagaseousenvironmentatambientconditions,accordingtothesupplier’sinformation.Underthesameoperatingcon-ditions,thetwo-fluidnozzlegeneratesinapuregasenviron-ment90-␮mdropletsforthesupplyof4.2m3/hofatomizingairand50-␮mdropletsforanatomizingairfluxof8.3m3/h.Themomentumfluxesoftheairinthelattercaseare1.93and0.96N,respectively.Inthecaseofthesingle-fluidnozzle,theliquidmomentumfluxesarebasedontheequivalentdiameterofthenarrowestcross-section.Theresultingliquidmomentumfluxis0.206Nforthesingle-fluidnozzleand0.125Nforthetwo-fluidnozzle.WeseefromFigure8thatneithertheaddi-tionalhelpoftheatomizingairinthecaseofthetwo-fluidnozzle,comparedtothesingle-fluidnozzle,northeincreaseofatomizingairflowleadstoasubstantialeffectwithrespecttothepenetrationofthejet.Onthecontrary,themeasuredtem-peratureprofilesarealmostidenticalforthetwotypesofnozzles.Weobservethatonlythehigherflowrateoftheatomizingaircausesaslightlydeeperpenetration.

Afterall,theperformanceofdifferentnozzlesfortheliquidinjectionintofluidizedbedsisalmostindependentofthechar-acteristicdropletdistributionmeasuredinagaseousenviron-ment.Itratherseemsthatnoatomizationoccursunderfluidizedbedconditionsandthemomentumoftheliquidfeedistheonlyfactorthatpromotesthepenetrationintothefluidizedbed.However,amoredetailedanalysisontheperformanceofinternallyandexternallymixedtwo-fluidnozzlesrevealedfur-therinterestinginsightsintothemechanismofliquidinjectionintofluidizedbeds.

Thedifferencebetweenexternallymixedandinternallymixedtwo-fluidnozzlesisintheprocessofatomization.Forexternallymixednozzles,theliquidfeedcomesintocontactwiththeatomizinggasatthenozzleexitonly.Forinternallymixednozzles,theliquidfeedandtheatomizinggasarealreadymixedintheinteriorofthenozzle.Therefore,inter-nallymixedtwo-fluidnozzlesareexpectedtopreatomizetheliquidfeedwithinthenozzle.18Fortheseexperimentsthefluidizedbedwasoperatedatthesameconditionsasdiscussedabove.Forbothtypesoftwo-fluidnozzles,theflowrateofatomizinggaswaskeptconstantat8.3m3/h.Becausethegeometryoftheannulusforthegasflowinsidethenozzleisthesameforbothnozzlesusedheretheexitvelocitiesarethesame.Moreover,althoughtheairmomentumfluxeswith1.93N,theflowrateoftheinjectedliquidwasincreasedfrom0to25,50,and75L/hintherespectiveexperiments.Thetemperatureprofilesweremea-suredandtheresultsareshowninFigures9and10.Itwasobservedfortheinternallymixednozzlethatthehighertheflowrateoftheinjectedliquid,thehigherthepenetrationdepthinthemeasuredtemperatureprofile.Themaximumextentofthetemperatureprofilewas8cmfortheinjectionof25L/hofthetestliquidwater.Anincreaseinthefeedrateoftheinjectedliquidto75L/hleadstoamaximumextentofthetemperatureprofileof14cm.

Differentresultswereobtainedfortheexperimentswiththeaidofanexternallymixedtwo-fluidnozzle.There,aflowrateof25L/hoftheinjectedliquidcausedamaximumextentofthetemperatureprofileof13cm.Thisisahigherpenetrationdepthcomparedtothatoftheinternallymixedtwo-fluidnozzle.However,anincreaseintheflowratefrom25to75L/hleadstoanunexpectedreductioninthemaximumpenetrationdepthof9cm.Forexternallymixednozzles,itseemsthatthe

AIChEJournal

March2005

Figure9.Meantemperaturesalongthedistancexfrom

thenozzleatz؍110cmandy؍0cm.

Internallymixedtwo-fluidfannozzleno.4,sprayangle60°.Theliquidmomentumfluxesare0.031,0.125,and0.282N;8.3m3/hatomizingair;FCCbedat153°C.

momentumoftheatomizinggasdissipatesnotonlyintheprocessofliquidinjectionbutalsointransportingfluidizedbedparticles.Becausethefractionoffluidizedbedsolidsremainsconstant,moremomentumdissipatesintheprocessofliquidinjection.Ontheotherhand,thepremixingofliquidfeedandatomizinggasseemstoprovidemomentumforamoreeffec-tiveinjectionandtransportoffluidizedbedsolids.

Itshouldbenotedthatthetemperatureprofilesofinternallyandexternallymixedtwo-fluidnozzlesdifferfortheinjectionofaironly,althoughthesamevolumetricfluxesofairhavebeenused.Theexplanationissimplythedifferenceinthespraypattern.Theinternallymixednozzlehasafansprayof60°,whereastheexternallymixednozzlehasafull-conesprayof20°sprayingangleonly,whichresultsinadeeperpenetrationofthejet.

Suchdifferencesbetweenexternallymixedandinternallymixedtwo-fluidnozzlesarenotknownfortheatomizationinagaseousenvironment.18Fromtheseresultsofthepresentexperimentalinvestigation,itcanthusbededucedthatthefluidizedbedsolidshaveasignificantimpactontheprocessof

Figure10.Meantemperaturesalongthedistancexfrom

thenozzleatz؍110cmandy؍0cm.

Externallymixedtwo-fluidfullconenozzleno.3,sprayangle20°.Theliquidmomentumfluxesare0.01,0.039,and0.088N;8.3m3/hatomizingair;FCCbedat153°C.

Vol.51,No.3771

Figure11.Localtime-averagedtracergasconcentra-tionalongthedistancexfromthenozzleaty؍0cmandz؍115cm.

Internallymixedtwo-fluidfannozzleno.4,sprayangle60°.Theliquidmomentumfluxesare0.031,0.125,and0.282N;8.3m3/hatomizinggas;FCCbedat153°C.

atomization.Thefluidizedbedsolidsseemtohinderthepro-cessofatomization,ifnottoinhibitthegenerationofdroplets.Furtherevidencewasgivenbyanexperimentalinvestigationonthefateoftheatomizinggas.Theatomizinggassuppliesthemomentumnotonlyforthedispersionoftheliquidfeedbutalsoforthepenetrationintothefluidizedbed.Copanetal.22foundthattheinjectionwithatwo-fluidnozzleperformsinamannersimilartothepenetrationofgasjetsintofluidizedbeds.Thiswouldimplythatalmostnomomentumisneededforthedispersionoftheliquidfeed,whichneedstobeinvestigatedforhigherliquidloading.Theaboveexperimentsontheliquidinjectionwithinternallymixedandexternallymixedtwo-fluidnozzleswererepeatedinsuchawaythatCO2tracerwasmixedintheatomizinggas.Theconcentrationoftracergasnearthenozzleexitwasmeasuredwithsuctionprobes.Itwasfoundthatan“equilibriumconcentration”isadjustedwithsufficientdistancefromthenozzleexit.ThisequilibriumconcentrationiscausedbytheequilibriumofadsorptionanddesorptionofCO2onthesurfaceofFCCcatalystparticles.Theequilibriumconcentrationcorrespondstothevolumeconcentration,whichisadjustedwhenthetracergasisperfectlymixedwiththefluidizingair.Therefore,themeasuredlocalgasconcentrationswerenormalizedbytheequilibriumconcentration.Thetracergasconcentrationwasmeasured5cmabovethesprayaxis.Itshouldbenotedthatthetracergasconcentrationwasmeasuredonadrybasis.Thevaporizedwaterwasremovedfromthegassamplebeforethegasanalysis,andisthusindependentoftheliquidfeedrate.TheresultsofthesemeasurementsareshowninFigures11and12.

Fortheinternallymixedtwo-fluidnozzle(Figure10),thetracergasconcentrationincreasesfrom150to200%oftheequilibriumconcentrationatxϭ1cm,reachesitsmaximumof300%atxϭ4–5cm,anddecreaseswithfurtherdistancefromthenozzleexituntiltheequilibriumconcentrationisreached.Almostnodifferencecanbefoundinthetracerprofilefortheinjectionofthegasjetonly(noliquid)andtheinjectionof75L/hofthetestliquid.Giventhemeasuredchangesinthecorrespondingtemperatureprofile,theatomizinggasofinter-nallymixedtwo-fluidnozzlesseemstooccupyaconstantregiononly.

772

March2005

Differentresultswereobtainedfortheinjectionwiththeexternallymixedtwo-fluidnozzle(Figure12).Theprofilefortheinjectionofgasonlyissimilartotheprofiledescribedabove.Then,themaximumofthetracergasconcentrationshiftstowardthenozzleexitwithincreasingflowrateoftheinjectedtestliquidwater.Sothepenetrationdepthoftheatomizinggasintothefluidizedbecomeshinderedwithin-creasingliquidloading.Thisisinaccordancewiththefindingthatthepenetrationdepthoftheliquidfeedisalsodecreasingwithincreasingliquidloading,asseeninthecorrespondingtemperatureprofiles(cf.Figure10).

Thediscussionofthemeasuredprofilesofthetemperatureandofthetracergasconcentrationgivesfirstindicationsofthemechanismoftheliquidinjectionintofluidizedbeds.However,amajorfactorinthisprocessisthedistributionofthevaporizedliquidfeed.

Aseriesofexperimentswasperformedonthemeasurementofthevaporconcentrationinthefluidizedbed.Oneofthekeyexperimentswasontheinjectionofthetestliquidwaterwithafannozzlehavingasprayingangleof120°,wherethetemperaturedistributionisdiscussedinthecontextofFigure7.Therethenozzle’sspraypatternwasrecognizableinthemea-suredtemperaturedistribution.ThecorrespondingdistributionofmeasuredvaporconcentrationsisshowninFigure13.Amazingly,tosomeextentthespraypattern(fannozzle120°)isalsorecognizableinthedistributionofthevaporconcentra-tions,giventhatitisinthedistributionofmeantemperaturesinthefluidizedbed.Themeasuredvaporconcentrationreachesalmost100%forthefirstfewcentimetersnearthenozzleexit.AllvaporconcentrationsϾ50%aremarkedbythecolorblackinthechart.Thevaporconcentrationdecreaseswithfurtherdistancefromthenozzleexitandasymptoticallyreachesanequilibriumconcentrationof14.6vol%.TheadjustmentofanequilibriumconcentrationissimilartothefindingfortheinjectionoftheadsorptivetracergasCO2,whichiscausedbytheequilibriumofdesorption(drying)andadsorptionofwateronthesurfaceoftheporouscatalystparticles.Theequilibriumconcentrationof14.6%correspondstothevolumeconcentra-tion,whichisadjustedwhenthevaporizedliquidfeedisperfectlymixedwiththefluidizingair.

Figure12.Localtime-averagedtracergasconcentra-tionalongthehorizontaldistancexfromthenozzleaty؍0cmandz؍115cm.

Externallymixedtwo-fluidfullconenozzleno.3,sprayangle20°.Theliquidmomentumfluxesare0.01,0.039,and0.088N;8.3m3/hatomizinggas;FCCbedat153°C.

Vol.51,No.3AIChEJournal

Figure15.Time-averagedvaporconcentrationsalong

thex-axisaty؍0cmandz؍115cm.

Figure13.Time-averagedvaporconcentrationnearthe

nozzleatz؍110cm.

Single-fluidfannozzleno.2,sprayangle120°;50L/hwater.Theliquidmomentumfluxis0.206N;FCCbedat153°C.

Two-fluidfull-conenozzleno.3,sprayangle20°;50L/hwater.Theliquidmomentumfluxis0.039N;8.3m3/hatomizingair,bedat153°C.

Foramoredetailedanalysis,thevaporconcentrationwasmeasurednotonlyalongthesprayaxisbutalso5cmaboveand5cmbelow.TheresultsareshowninFigure14,wherethelocalvaporconcentrationsarenormalizedbytheequilibriumconcentration.Theconcentrationsonthesprayaxisexhibithighvaluesnearthenozzleexit.Thesehighvaporconcentra-tionsseemtobecausedbyliquidsthatstickandevaporateatthetipofthesuctionprobes.Itisstrikingthatthevaporconcentrations5cmaboveandbelowthesprayaxisarealmostsimilar.Ifaninstantaneousevaporationwouldoccuratthenozzleexit,onewouldhavefoundasignificantlyhighervaporconcentrationdownstream(thatis,atthemeasurementposition5cmabovethesprayaxis).Becausethevaporizingfeedishomogeneouslydistributedbothup-anddownstream,atrans-portofliquiddropletswiththeup-flowinggasphaseseemstobeunlikely.Suchahomogenousdistributionmaybecausedbytheerraticmotionofthefluidizedbedsolids,whicharewettedatthenozzleexitandtransportedintotheinteriorofthe

Figure14.Time-averagedvaporconcentrationalong

thedistancexfromthenozzleaty؍0cm.

Single-fluidfannozzleno.2,sprayangle120°;50L/hwater.Theliquidmomentumfluxis0.206N;FCCbedat153°C.

fluidizedbed,wheretheliquidthenevaporatesfromtheparti-cles’surfaces.

Thisfindingofwettedparticlesisinkeepingwiththealreadydiscussedideathatthefluidized-bedsolidshindertheatomi-zationofliquidfeed.Inthecaseofporouscatalystparticles,theliquidfeedmaybetakenupintotheporesandthereleaseofvaporissloweddownbythemoderatedryingratesofporousmaterial.Theadsorptionofvaporatthesurfaceofdryparticlesisalsoanimportantfactor.

However,thedominanceofadsorptioneffectsisunlikelyinhigh-temperatureapplicationssuchasBASF’sanilineprocess3orthephthalicanhydridesynthesis.25Tomodelthissituation,an-othersetofexperimentswasperformedwithquartzsandasthefluidizedbedmaterial.Quartzsandisnonporous;thusneitherisliquidtakenupintotheporesnorisvaporizedfeedadsorbed.Acomparisonwasmadebetweentheprofileofvaporconcentration,whichwasmeasuredinafluidizedbedofquartzsand,andofFCCcatalyst.ThelocalvaporconcentrationwasnormalizedbytheequilibriumconcentrationandtheresultsareshowninFigure15.ThemeasuredvaporconcentrationinabedofFCCcatalystdecreasessteeplywithincreasingdistancefromthenozzleexitandasymptoticallyreachestheequilibriumconcentrationatxϭ25cm.Theprofileinabedofquartzsanddoesnothavesuchacharacteristic.Itwasobservedforquartzsandthatthemeasuredvaporconcentrationsexhibitnumericalvaluesabovetheequilib-riumconcentrationoveradistanceofupto70cmfromthenozzleexit.Thevaporconcentrationdecreaseswithfurtherdistancefromthenozzleexitandconcentrationsweremeasuredbelowtheequilibriumconcentration.

Itisunlikelythatenhancedevaporationoccursforthenonpo-rousfluidized-bedmaterialofquartzsand.Onepossibleexplana-tionmaybethatagglomeratesareformednearthenozzleexit,thuscapturingtheinjectedliquid.Theseagglomeratesaremorestableandsticktothetipofthesuctionprobesoverawiderdistancefromthenozzleexit.AgglomeratesofporousFCCcat-alystareobviouslylessstablebecausethecapturedliquidistakenupandtheagglomeratesbreakupmoreeasily.However,theseeffectsaredifficulttoconcludefromtheresultsofvaporconcen-trationandtemperaturemeasurementsonly.

Therefore,thecapacitancemeasurementtechniquewasusedforthedetectionofliquidsthatmoveeitherfreelyorinag-Vol.51,No.3

773

AIChEJournalMarch2005

Figure16.Sequenceofcapacitancemeasurementsig-nalsatx؍7cm,y؍0cm,andz؍110cm.

Two-fluidfull-conenozzleno.3,sprayangle20°;50L/hwater.Theliquidmomentumfluxis0.039N;8.3m3/hatomizingair;FCCbedat153°C.

glomerates.Becausethedielectricconstantofliquidwaterisextremelyhighcomparedtothedielectricconstantsofgasandsolids,respectively,liquidwaterwillcauseclearsignalsofthecapacitancemeasurementsystem.Additionally,themeasure-mentsystemwithneedlecapacitanceprobesassistsnotonlyinthelocaldetectionofthetestliquidwaterbutalsointhedeterminationofthelocalsolidsvolumeconcentration.

AtypicaltimeseriesisshowninFigure16thatwasmea-suredatadistanceofxϭ7cmfromthenozzleexit.Thesignallevelwiththecharacteristicvalueof7.74Vforairisdepictedinthechartasadashedline.Thesignallevelwiththecharac-teristicvalueof4.27VforthefixedbedofFCCcatalystparticlesisdepictedinthechartasadottedline.Thesignallevelsofairandofafixedbedofsolids,respectively,wereobtainedfromexternalcalibrationofthecapacitanceprobe.Theflowstructureofafullydevelopedbubblingfluidizedbedcanbeidentifiedinthegiventimeseries.Bubblesarecharacterizedbyapeakfromthesignalbaselineofgas–solidsuspensionuptothesignallevelofair,asmarkedinthechart.Bubblescanbeidentifiedatatypicalmeanfrequencyofbubblingfluidizedbedsat2–3Hz.25Becausethisbubblingfrequencywasmeasuredinafluidizedbedwithandwithoutliquidinjection,itwasconcludedthatthegas–solidflowstruc-tureisalmostunaffectedbytheliquidinjection.Occasionally,peaksweremonitoredduringthemeasurementwithahighercapacitancethanthatofthefixedbedofsolids.Thesepeaksareattributedtothepresenceofliquids.Itwasfoundthatliquidsweredetectedinthesuspensionphaseonly,thusprovidingfurtherevidencefortheideathatthesesignalswerecausedbywettedparticlesand/oragglomerates.Thefrequencyofag-glomeratesinthesetimeserieswashigherinfluidizedbedsofquartzsandthanthatinFCCcatalyst.

Thedirectportabilityoftheaboveexperimentalresultstotechnicalapplicationsoftheliquidinjectionintofluidizedbedreactorsislimitedbecauseofthetestliquidwater.Waterexhibitsanextremelyhighheatofevaporationof2500kJ/kg,whichdifferssignificantlyfromthatofheatsofevaporationofliquidsintechnicalapplications.Thus,additionalexperimentswereperformedwiththetestliquidethanol.Ethanolisan

774

March2005

organicsubstanceandhasalowerheatofevaporationof730kJ/kg.Withsuchalowheatofevaporationandaboilingpointof78°Catambientpressure,ethanolisalsohighlyvolatileatmeanfluidizedbedtemperaturesof150°C.

Thetestliquidethanolwasinjectedintothefluidizedbedbyuseoftheexternallymixedtwo-fluidnozzleoffull-conespraypattern.ThefluidizedbedmaterialwasFCCcatalystinonesetofexperimentsandquartzsandinanothersetofexperiments.Thefluidizedbedwasoperatedatasuperficialgasvelocityof0.3m/sandthemeantemperaturewaskeptconstantat153°C.TheresultsofthetemperaturemeasurementsareshowninFigure17.Toassesstheperformanceofthetestliquidwater,temperatureprofilesoftherespectiveexperimentswithwaterarealsoshowninthegraph.Itcanbeseenthatthetemperatureprofilesofbothtestliquidsexhibitasimilarcharacteristic.Evidently,noinstantaneousevaporationoccursatthenozzleexitnotevenforthehighlyvolatiletestliquidethanol.Thetemperaturerisessteeplyfromtheinlettemperatureof20°Ctoacharacteristictemperatureat10Kbelowtheboilingpoint(thatis,about90°Cforwaterand70°Cforethanol).Thischaracteristictemperatureisanotherindicationfortheforma-tionofagglomeratesatthenozzleexit.Ifthereweredropletsorindividualparticleswithwettedsurfaces,thetemperaturewouldbeeitherthewetbulbtemperatureforexcessairortheboilingtemperatureforlimitedmasstransferofvapor.Thereductionintemperatureaccountsforagglomerates.

Itmightbearguedthatthedifferenceinthevapordensitiesbetweenwaterandethanol,respectively,mightexplainthedifferenceinthepenetrationdepth.However,becausenoin-stantaneousevaporationoftheliquidswasobserved,butratheramoderatedryingprocessofwettedsolids,thedifferencesinthevapordensitiesshouldnothaveasignificanteffectonthelocaltemperatureandthegas–solidflowinthefluidizedbed.Withfurtherdistancefromthenozzleexit,thetemperatureincreasestothemeantemperatureofthefluidizedbed.Theagglomeratesdecayinthisregionandbecomemixedintotheinteriorofthefluidizedbedbythelarge-scalemixingofthesolids.Ingeneral,thetemperatureprofilereachesdeeperintothefluidizedbedfortheinjectionintoafluidizedbedofquartzsandthanintoafluidizedbedofFCCcatalyst.Theagglomer-

Figure17.Localtime-averagedtemperaturesalongthe

distancexfromthenozzleatz؍110cmandy؍0cm.

Two-fluidfull-conenozzleno.3,sprayangle20°;25L/hethanolorwater.Theliquidmomentumfluxesare0.0078and0.010N;8.3m3/hair;fluidizedbedat153°C.

Vol.51,No.3AIChEJournal

Acknowledgments

TheauthorsgratefullyacknowledgethefinancialsupportofthisresearchprojectbytheDeutscheForschungsgemeinschaft,Bonn.

LiteratureCited

1.WertherJ.Fluidized-bedreactors.Ullmann’sEncyclopediaofIndus-trialChemistry.Weinheim,Germany:Wiley–VCH;2001.

2.WilsonJW.FluidCatalyticCrackingTechnologyandOperations.Tulsa,OK:PennWellPublishing,1997.

3.KahlT,SchroederKW,LawrenceFR,MarshallWJ,HoekeH,JaehkR.Aniline.Ullmann’sEncyclopediaofIndustrialChemistry.Wein-heim,Germany:Wiley–VCH;2001.

4.BurdettID,EisingerRS,CaiP,LeeKH.Gas-phasefluidizationtechnologyforproductionofpolyolefins.In:YangWC,LiJ,KwaukM,eds.FluidizationX.NewYork,NY:UnitedEngineeringFounda-tion;2001:39-52.

5.TheologosKN,MarkatosNC.Advancedmodelingoffluidcatalyticcrackingriser-typereactors.AIChEJ.1993;39:1007-1017.

6.TheologosKN,LygerosAI,MarkatosNC.FeedstockatomizationeffectsonFCCriserreactorsselectivity.Chem.Eng.Sci.1999;:5617-5625.

7.GaoJ,XuC,LinS,YangG.Simulationsofgas–liquid–solid3-phaseflowandreactioninFCCriserreactors.AIChEJ.2001;47:677-692.8.GuptaA,RaoDS.EffectoffeedatomizationonFCCperformance:Simulationofentireunit.Chem.Eng.Sci.2003;58:4567-4576.9.HeinrichS,Mo¨rlL.Fluidizedbedspraygranulation—Anewmodelforthedescriptionofparticlewettingandoftemperatureandconcen-trationdistribution.Chem.Eng.Process.1999;38:635-663.10.BecherRD,Schlu¨nderEU.Fluidizedbedgranulation:Gasflow,

particlemotionandmoisturedistribution.Chem.Eng.Process.1997;36:261-269.

11.SmithPG,NienowAW.Onatomizingaliquidintoagasfluidizedbed.

Chem.Eng.Sci.1982;37:9-956.

12.MarongaSJ,WnukowskiP.Establishingtemperatureandhumidity

profilesinfluidizedbedparticulatecoating.PowderTechnol.1997;94:181-185.

13.FanLS,LauR,ZhuC,VoungK,WarsitoW,WangX,LiuG.

Evaporativeliquidjetsingas–liquid–solidflowsystem.Chem.Eng.Sci.2001;56:5871-51.

14.GuW,TuzlaK,ChenJC.Measurementofliquidsprayevaporationin

fastfluidizedbeds.AIChESymp.Ser.1996;92:153-1.

15.SkoubyDC.Hydrodynamicstudiesina0.45-mriserwithliquidfeed

injection.AIChESymp.Ser.1999;95:67-70.

16.NewtonD,FiorentinoM,SmithGB.TheapplicationofX-rayimaging

tothedevelopmentsoffluidizedbedprocesses.PowderTechnol.2001;120:70-75.

17.KnapperBA,GrayMR,ChanEW,MikulaR.Measurementofeffi-ciencyofdistributionofliquidfeedinagas–solidfluidizedbedreactor.Int.J.Chem.ReactorEng.2003;1:A35.

18.LefebvreAH.AtomizationandSprays.London:Hemisphere;19.19.SirignanoWA.FluidDynamicsandTransportofDropletsandSprays.

Cambridge,UK:CambridgeUniv.Press;1999.

20.ZhuC,WangX,FanLS.Effectofsolidsconcentrationonevaporative

liquidjetsingas–solidflows.PowderTechnol.2000;111:79-82.

21.WiesendorfV,WertherJ.Capacitanceprobesforsolidsvolumecon-centrationandvelocitymeasurementsinindustrialfluidizedbedreac-tors.PowderTechnol.2001;119:143-157.

22.CopanJ,ClarkeN,BerrutiF.Theinteractionofsingle-andtwo-phase

jetsandfluidizedbeds.In:YangWC,LiJ,KwaukM,eds.FluidizationX.NewYork,NY:UnitedEngineeringFoundation;2001:77-84.23.BruhnsS.OntheMechanismofLiquidInjectionintoFluidizedBed

Reactors.PhDDissertation,TechnicalUniversityHamburg–Harburg,Germany;2002.

24.BruhnsS,WertherJ.Modelingtheinjectionofliquidreactantsinto

fluidizedbedreactors.Proc.of8thChina–JapanSymposium,Fluidi-zation2003—ScienceandTechnology,Tokyo,Japan:SocietyofChemicalEngineersJapan,2003:386-396.

25.KuniiD,LevenspielO.FluidizationEngineering.Stoneham,MA:

Butterworth-Heinemann;1991.

ManuscriptreceivedJan.12,2004,andrevisionreceivedJun.28,2004.

Figure18.Mechanismoftheliquidinjectionintoadense

(bubbling)fluidizedbed.

atesofquartzsandparticlesareobviouslymorestablethanagglomeratesofFCCcatalyst.Thiswasalsoobservedintheprofilesofvaporizedfeed,asdiscussedinthecontextofFigure15.Anotherfactoristhetake-upofliquidsintoporesandtheresultingslowerdryingrateofporousFCCcatalyst.

Conclusions

Basedontheresultsofthepresentexperimentalinvestiga-tion,anewmodelissuggestedforthemechanismoftheinjectionofliquidreactantsintofluidizedbedreactors,whichisillustratedbyFigure18.Theliquidsarelocallyintroducedintothefluidizedbedreactorwiththeaidofaninjectornozzle.Despitethefactthatthereactorisoperatedattemperatureswellabovetheboilingpointoftheinjectedliquid,noinstantaneousevaporationoccursatthenozzleexit.Instead,theliquidjetpenetratesthefluidizedbedandwetsthefluidizedbedsolidsinthevicinityofthenozzleexit.Thepresenceofsolidsatahighsolidsvolumeconcentrationof30–40%seemstosuppresstheatomizationoftheinjectedliquids.Becausetheinjectedliquidsandthesolidscomeintodirectcontactatthenozzleexit,itseemsthat—contrarytowhatisknownfromtheatomizationinagaseousenvironment—noliquiddropletsareformedwithinthedensebubblingfluidizedbed(cf.Figure1).

Particlesbecometransportedintotheliquidjetandimmediatelyformagglomeratesafterall.Theseagglomeratesbecomerapidlytransportedintotheinteriorbythegrosssolidsmixingofthefluidizedbed.Theliquidevaporatesfromtheinterioroftheagglomerates.Theagglomeratesdecaybecauseoftheimpactofthevigorousparticle–particleinteractionsinthefluidizedbed.Inthecaseofporousparticles,theliquidisalsotakenupintotheinnerpores.Therefore,agglomeratesofporousparticlesbreakupmoreeasilybutthedryingrateissloweddownbecauseoftheporediffusion.Ifthevaporadsorbsattheparticles’surface,amorehomogeneouspatternofreleasedvaporcanbeobservedtoarisefromtheequilibriumofdesorption(drying)andadsorption.

Basedonthesefindingsamodelforthescale-upoffluidizedbedreactorswithliquidinjectionhasbeendeveloped,23,24amodelthatassumestheinstantaneouswettingofthefluidizedbedparticlesatthenozzleexitandconsidersthetransportofthewettedparticlesbythegrosssolidsmixingofthefluidizedbed.Thereleaseofthevaporizedfeedshouldbemodeledbyexperimentallydetermineddryingkinetics,whichaccountnotonlyforthepropertiesoftheinjectedliquidbutalsoforthepropertiesofthefluidizedbedparticles.

AIChEJournal

March2005

Vol.51,No.3775