Esterification of a Fatty Acid by Reactive Distillation

反应精馏

3612Ind.Eng.Chem.Res.2003,42,3612-3619

SEPARATIONS

EsterificationofaFattyAcidbyReactiveDistillation

SvenSteinigewegandJu1rgenGmehling*

CarlvonOssietzkyUniversityofOldenburg,IndustrialChemistry,P.O.Box2503,D-26111Oldenburg,Germany

Areactivedistillationprocessfortheproductionofdecanoicacidmethylestersbyesterificationofthefattyaciddecanoicacidwithmethanolispresented.Thereactionhasbeencatalyzedheterogeneouslybyastrongacidicion-exchangeresin(Amberlyst15).ApragmatickineticmodelbasedonaLangmuir-Hinshelwood-Hougen-Watsonapproachhasbeenderivedandthekineticconstantsofthisandapseudohomogeneousmodelhavebeenfitted.Twodifferentcatalyticpackings,Katapak-SandKatapak-SP,havebeenusedforreactivedistillationexperiments.TheseparationefficiencyofKatapak-SPwasdeterminedexperimentallyandreactivedistillationexperimentsinpilot-plantcolumnshavebeenperformed.Operationconditionshavebeenvaried(refluxratioandreactantratio)experimentally.Anequilibriumstagemodeliscapableofdescribingtheexperimentsquantitativelywhentheadsorptionbasedakineticmodelisapplied.Simulationhasbeenusedsubsequentlytodeterminetheinfluenceofimportantoperatinganddesignfactors(reactantratio,refluxratio,pressure,distillate-to-feedratio,sizeofthereactivesection,androleofaprereactor)andtocomparethepackingssystematically.Finally,aprocessisproposedthatispromisingforscale-upandoptimizationwithregardtoeconomicissues.

Introduction

Thecombinationofreactionanddistillationwithinoneunitoperationiscalledreactivedistillation.Espe-ciallyforequilibriumlimitedandconsecutivereactions,reactivedistillationisapromisingprocessalternative.Thedirectremovaloftheproductsorintermediatesresultsinhigherconversionsandselectivitiesincom-parisonwiththeclassical,sequentialapproach.Themostimportantequilibriumlimitedreactionsthataresuitableforreactivedistillationareesterifications,esterhydrolysisreactions,transesterifications,andetherifi-cations.Inrecentyearsresearchfocusedontheinves-tigationofthemethylacetatesynthesisandhydrolysisbyreactivedistillation.Themethylacetatesystemservesasamodelsystemforreactivedistillationprocesses.1,2Besidesmethylacetatesynthesis,alsotheesterificationofotheralcohols(e.g.,n-butanol3)withaceticacidhasbeeninvestigated.Incontrasttothesesystems,informationabouttheesterificationoflong-chaincarboxylicacidssuchasfattyacidsbyreactivedistillationcanhardlybefoundintheliterature.

Esterificationoffattyacidsisacommonpracticeinthechemicalindustry.Fattyacidestersareimportantintermediates,surfactants,lubricants,ortensides.Nev-ertheless,informationaboutheterogeneouslycatalyzedesterificationoffattyacidsbyreactivedistillationcannotbefoundintheopenliterature.Onlyinformationabouthomogeneouslycatalyzedreactivedistillationisavail-able.Investigationsabouttheesterificationofmyristic

*Towhomcorrespondenceshouldbeaddressed.Tel.:+49-441-7983831.Fax:+49-441-7983330.E-mail:gmehling@tech.chem.uni-oldenburg.de.http://www.uni-oldenburg.de/tchemie/.

acidwith2-propanolinatraycolumnhavebeenperformedbyBocketal.4Theseauthorspresentaprocessincludingarecoverycolumnfor2-propanol.Sensitivityanalysesanddynamicexperimentsarepresented.Jerominetal.5describetheesterificationofdifferentfattyacidswithmethanolinatraycolumnandcomparethecontinuousreactivedistillationprocesswithabatchprocess.Schleperetal.6describetheesterificationofafattyacidwith2-propanolinatraycolumn.Theypresentdataconcerningnotonlythereactionkineticsbutalsotheirpilot-plantexperiments.Tothebestofourknowledge,noinformationabouttheheterogeneouslycatalyzedreactivedistillationfortheproductionoffattyacidestersisavailableintheopenliterature.Theesterificationofdecanoicacidwithmethanolformingdecanoicacidmethylesterandwaterwaschosenasamodelsystem.

decanoicacid(DecH)+methanol(MeOH)h

methyldecanoate(MeDec)+water(H2O)(1)Heterogeneouslycatalyzedreactivedistillationoffersadvantagesoverthehomogeneouslycatalyzedprocessalternative(e.g.,sulfuricacid).Sizeandlocationofthereactivesectioncanbechosenregardlessofthermo-dynamicconstraintsandatthesametimecorrosionproblemswillbeminimized.Furthermore,thesepara-tionofthehomogeneouscatalystfromtheproduct(ester)thatisobtainedasbottomproductcanbeavoided.Theimmobilizationoftheheterogeneouscata-lystinsidethecolumncan,forexample,beachievedbystructuredpackingssuchasKatapak-SorKatapak-SP(bothSulzerChemtech).Katapak-Sismadeofcor-rugatedwiremeshsheets.Katapak-SPcombineswiregauzelayers(catalyticlayers)withlayersofMellapak

10.1021/ie020925iCCC:$25.00©2003AmericanChemicalSociety

PublishedonWeb06/20/2003

反应精馏

Table1.AntoineConstantsAccordingtologPsi)Ai-[Bi/(Ci+T)]WherePs

iIsinkPaandTinK

component

AiBiCimethanol7.205871582.27-33.424water

7.196211730.63-39.724decanoicacidmethylester6.140321590.66-115.835decanoicacid

6.11877

1593.22

-156.557

(separationlayers).ForthemeasurementspresentedheretheKatapak-SP11typewasapplied.7Thispackingconsistsofalternatecatalyticandseparationlayers.Asaheterogeneouscatalyst,thestronglyacidicion-exchangeresinAmberlyst15(Rohm&Haas)wasusedinthisinvestigation.Amberlyst15hasalsobeenappliedsuccessfullyforthemethylacetatesynthesisandhy-drolysis1andforthen-butylacetatesynthesis.3

Thispaperdescribesthedevelopmentofaprocessfortheproductionofdecanoicacidmethylesterbyhetero-geneouslycatalyzedreactivedistillation.Theprocedureappliedhereoffersacomprehensivemethodforthedevelopmentofreactivedistillationprocesses,whichisalsovalidforequilibriumreactionswithaconsiderablylowerreactionrate.Processdevelopmentstartswithathermodynamicanalysisoftheconsideredsystem.Inasecondstepthereactionkineticsofthereactionhastobeinvestigatedunderconditionsexpectedinthereactivedistillationcolumn.Incomparisonwiththemethylacetateandn-butylacetatesynthesis,biningtheresultsofthethermodynamicandthekineticanalysisleadstothedevelopmentofasimulationenvironmentbasedonanequilibriumstagemodel.Subsequently,thesimulationresultshavetobeverifiedbyconductingexperimentsinapilot-plantcolumn.However,becauseofthelargenumberofparameters,itisnotpossibletoinvestigateallprocessalternativesexperimentally.Therefore,simulationstudiesareusedtoidentifytheroleofimportantoperatinganddesignfactors.Ifanappropriateprocesshasbeenidentified,furtheropti-mizationoftheprocesswithregardtoeconomicconsid-erationsandascale-upcanbecarriedout.ThermodynamicAspects

Inthequaternarysystemnoazeotropesoccur.Whiledecanoicacidmethylesteristhehighboilingproduct,whichcanbeexpectedinthebottomstream,leavingthereactivedistillationcolumn,wateristhelowboilingproductthatleavesthecolumnwiththedistillatestream.Atatmosphericpressurehightemperatures(about500K)wouldbeattainedinthereboilerwhennomethanolispresentinthebottomstream.Thesetemperaturesusuallyresultinsignificantlylowerprod-uctqualityofthedecanoicacidmethylester.5Therefore,acertainamountofmethanol(about4mol%)shouldbetoleratedinthereboiler.

TheAntoineequationisusedforthecalculationofthesaturationvaporpressuresofthepuresubstances.TheAntoineconstantsusedinthiswork(Table1)weretakenfromtheDortmundDataBank(DDB),whichwaskindlyplacedatourdisposalbyDDBSTGmbHGer-many.8Theactivitycoefficientsγofphaseandchemicaliareusednotonlyforthecalculationequilibriabutalsoforthekineticexpression.Exceptforthesystemmetha-nol-waternoexperimentalVLEdataareavailableintheDortmundDataBank.8Therefore,modifiedUNI-

Ind.Eng.Chem.Res.,Vol.42,No.15,20033613

FAC(Dortmund)wasusedwiththeparameterspub-lishedbyGmehlingetal.9ExperimentalSection

Chemicals.DecanoicacidsuppliedbyCognisDeut-schlandwithaminimumpurityof99%wasusedwithoutfurtherpurification.Methanolusedfortheexperimentsconcerningthereactionkineticswereofanalyticalgrade(99.8%Scharlau).DecanoicacidmethylesterwassynthesizedinourlaboratoryandverifiedbyNMRspectroscopy.GCanalysisshowedanesterpurityof>99.8%.Waterwasbidistilled.Theorganicchemicalsweredriedoveramolecularsievepriortouse.Forthereactivedistillationexperimentsmethanolwasofreac-tiongrade(99.5%)andwasusedwithoutfurtherpurification.Thepurityofallorganicchemicalswasverifiedbygaschromatography.

Analytics.Allsamplesofthekineticandreactivedistillationexperimentswereanalyzedbygaschroma-tography(HP6890withTCD;Heasacarriergasat2.5cm3min-1hold1for3.05min,rateof10cm3min-2to6.3cm3min-;HP-Innowax30m×0.032mm;split5:1;temperatureprogram:333Kholdfor3min,heatrateof80Kmin-1to403Kholdfor0.2min,heatrateof80Kmin-1to513K).ReactionKinetics

Experiments.Theexperimentswereconductedinathermostatedbatchglassreactorwithavolumeof500cm3.Thetemperatureoftheheatingjacketwaskeptconstantwithin(0.1K.ThestirrerwasplatetypeandmadeofTeflon.Thespeedwasvariedbetween100and800rpm.Toimprovemixing,abafflewasinstalled.Furthermore,arefluxcondenserwasappliedtoavoidalossofvolatilecomponents.PriortouseAmberlyst15waswashedwithwaterseveraltimesuntilthesupernatantliquidwascolorless.Subsequently,Am-berlyst15wasdriedundervacuum(P<1mbar)at353Kuntilthemassremainedconstant(usuallyafter2days).Theion-exchangecapacityofAmberlyst15hasbeendeterminedearlier.11Beforethekineticexperimentwasstarted,bothreactantswerebroughttoreactiontemperatureinseparatevessels.Sincethecatalystwasnotheatedtothereactiontemperature,addingthecatalystwouldleadtoatemperaturedecrease.Toensureisothermalconditions,thecatalystwasaddedabout2minbeforetheexperiment.Temperaturemea-surementindicatedthatthistimeissufficienttoheatthecatalyst.Whenthedesiredtemperaturewasreached,thereactorwasfilledwithbothreactantsandthetimemeasurementwasstarted.Samplesofabout1cm3weretakenusingasyringeandweighed(accuracyofthebalance(0.001g).Afterward,about1cm3ofacetonewasaddedtothesamplestoensureahomogeneousmixture.Duringeverymeasurementseries15-20samplesweretaken.Thesampleswerecooledrapidlyto280Ktoavoidfurtherreactionandthenanalyzedbygaschromatography.Measurementswereperformedinatemperaturerangebetween309and338K.Besidesthetemperaturetheamountofcatalyst,initialreactantratio,andinitialamountofwaterwasvaried.Innoneoftherunssideproductshavebeendetected.

Todeterminetheinfluenceoftheexternalmasstransferonthereactionrate,thestirrerspeedwasvariedbetween100and600rpminadditionalruns.Noinfluenceofthestirrerspeedonthereactionratewas

反应精馏

3614Ind.Eng.Chem.Res.,Vol.42,No.15,

2003

detectedabove200rpm.Therefore,allfurtherexperi-mentswereconductedatastirrerspeedof400rpm.Po¨pkenetal.11provedtheabsenceofinternalmasstransferforthemethylacetatesynthesis.AlsoXuandChuang12statedthatinternaldiffusionisinsignificantfortheesterificationofaceticacidwithmethanolcatalyzedbyAmberlyst15.DuetothefactthatAm-berlyst15iscomposedofverysmallgel-typemicro-sphererswithlargemacropores,13internaldiffusioncanalsobeexcludedfortheesterificationofdecanoicacidwithmethanol.Itcanbeexpectedthatsorptioneffectsaremoreimportantthaninternalmasstransfer.

KineticModeling.Bothapseudohomogeneousmodelandanadsorption-basedmodelhavebeeninvestigatedrecently.11Thepseudohomogeneouskineticmodelislesscomplexandhasasmallernumberofparameters.Esterificationsareknowntobereversiblereactionsofsecondorder.Therefore,thepseudohomogeneousmodelcanbewrittenas11

r)

11dni

m)kcatνidt

1aDecHaMeOH-k-1aMeDecaH2O

(2)

wherebyactivitiesinsteadofconcentrationsormolefractionsareused.Thisleadstoamoreconsistentandaccuratedescription.11Themajordrawbackofthepseudohomogeneousmodelistheneglectofsorptioneffects,forexample,thedifferentaffinitiesofthecomponentstowardtheion-exchangeresin,whichresultindifferencesbetweentheconcentrationsofthecom-ponentsinthepolymericresinandbulkliquid.Espe-ciallyforthesysteminvestigated,ignoringsorptioneffectsisacriticalsimplification.Duetoitssmallmolecularsizeandhighpolarity,waterispreferablysorbedbythestronglypolarion-exchangeresin.Withrisingwatercontentinthebulkliquid,thereactionratedecelerates.TheseassumptionsareingoodagreementwiththeresultsofShimizuandHirai14obtainedfortheesterificationofoctanoicacidwithmethanol.Themostsimplemethodtoaccountforthesorptionofwateristouseasimplifiedadsorption-basedmodelbasedonaLangmuir-Hinshelwood-Hougen-Watson(LHHW)ap-proach.

TheLHHWequationforareversibleesterificationusinganion-exchangeresinasacatalystcanbewrittenas15

r)

1dni

(

νdt

)micat×k//

1KDecHaDecHKMeOHaMeOH-k-1KMeDecaMeDecKH2OaH2O

1+(KDecHaDecH+KMeOHaMeOH+K2

MeDecaMeDec+KH2OaH2O))

(3)

WaterispreferablysorbedandthereforesorptioneffectscanbesummarizedwithasingularsorptionconstantKforSorb.RehfingerandHoffmann16derivedanequationthesynthesisofmethyltert-butyletherusingthissimplification.Theirapproachcanalsobeappliedforanesterificationreaction.Thisleadstothefollowingequation:

r)

1dni

ν)mk1aDecHaMeOH-k-1aMeDecidtcat(

(KSorbaHK2O)2

SorbaH2O)

(4)

Thephysicalbasisofsuchapragmaticapproachshould

Figure1.Resultsoftwokineticexperimentswithdifferentinitial

amountsofwater.(T)333K,x0

H)0.019:(O)exp.,(;)fitted(eq4);x0)0.179:(b)exp.,(;)2fittedOH(eq4)).2O

notbeoveremphasized.Arigorousmodelingofthe

solventuptakebyanion-exchangeresinwouldinvolvemodelingofthepolymer-phaseactivities,forexample,withtheFlory-Hugginsmodel.Fordenselycross-linkedresinwithhighlypolargroupsonalmosteverymonomer(Amberlyst15hasadegreeoffunctionalizationofabout85%)theseeffectsarepoorlyunderstood.Therefore,itshouldbeconsideredasapragmaticmodelthatiscapableofdescribingthemostimportanteffects,forexample,thetemperaturedependencyandthedepen-dencyofthereactionrateasafunctionofwatercontent,withonlyonemoreadjustableparametercomparedtothepseudohomogeneousmodel.

ThetemperaturedependencyoftherateconstantisexpressedbyArrhenius’law:

ki)k0-EA,i

iexp

(

RT

(5)

Itwasfoundthatthetemperaturedependencyofthe

sorptionconstantKSorbcanbeneglectedforthetem-peraturerangecoveredincontrasttothetemperaturedependencyoftherateconstants.

DataAnalysis.Fiveadjustableparameters(k00

E,E1,kgiven-1,inA,1TableA,-1,K2.ToSorb)havetobefitted.Theresultsareevaluatetheinfluenceofwateronthereactionrateinadditionalkineticexperiments,waterwasaddedtobothreactantsbeforethemeasurementwasstarted.Thisallowsonetodeterminetheinfluenceofwateronthereactionratequantitatively.ForfittingthesorptionconstantKbeenperformedunderSorbninekineticexperimentshavethesameconditions(i.e.,initialamountofacidandalcohol,temperature,andamountofcatalyst)fordifferentamountsofwaterinitiallypresent.Figure1showstheweightfractionofdecanoicacidasafunctionoftimefortwoofthesekineticexperiments.Furthermore,thecalculatedcon-centrationsusingtheconstantsgiveninTable2areshown.Itcanbeseenthattheadsorption-basedmodelisabletodescribethereactionratefordifferentwaterconcentrations.FromFigure2itcanbeseenthatthetemperaturedependencyofthereactionratescanbedescribedusingArrhenius’law.

Toevaluatetheinfluenceofthekineticmodelonthesimulationresultsofthereactivedistillationprocess,additionalparametersforthepseudohomogeneouski-neticmodelhavebeenfitted.Thekineticexperimentsforthedeterminationofthesekineticparametershavebeenconductedwithoutanywaterinitiallypresentandwithvacuum-drycatalyst.Thefouradjustableparam-etersofthepseudohomogeneousmodel(k0,k0

ETable3.Pleasenotethat1the-1,ErateA,1,constantsA,-1,)aregiveningivenrelatetotheamountofdrycatalyst.

反应精馏

Figure2.Arrheniusdiagramoftherateconstantsfortheesterificationreactionk1(b)andthehydrolysisreactionk-1(O)oftheheterogeneouslycatalyzedreaction.Thelinesrepresenttheresultsoftheoverallfit(eq4).

Table2.KineticParametersfortheModifiedLHHWModel(eq4)(Ksorb)2.766)

reactionik0g

-1i(mols-1)EA,i(kJmol-1)

esterification13.1819×10672.23

hydrolysis

-1

3.5505×105

71.90

Table3.KineticParametersforthePseudohomogeneousKineticModel(eq2)

reactionik0g

-1i(mols-1)EA,i(kJmol-1)

esterification19.1164×10568.71

hydrolysis

-1

1.4998×104

64.66

ReactiveDistillationExperiments

Setup.Theexperimentsinapilot-plantscaleareperformedinaglasscolumnwithaninnerdiameterof50mmsuppliedbyQVF(QVFEngineering).Inthereboilertheliquidwasheatedelectricallybyrod-shapedquartzheaters(VogelsbergerQuarzglastechnik).Thereboilerdutywascontrolledwithatransformeranddeterminedwithadigitalmultimeter(VoltcraftM-3860M)within(1%.ThereactivesectionofthecolumnconsistedeitherofKatapak-SorKatapak-SPelementsfilledwithAmberlyst15,whereasthenonre-activesectionconsistedofSulzer-BXpackings.Eachcolumnsectionhadaheightof1.2mandaneffectivepackingheightof1m.TheheatlossofthecolumnhasbeendeterminedearlierandthecorrelationproposedbySteinigewegandGmehling3wasusedforthesimula-tion.Thecolumnwasequippedwithatotalcondenserandarefluxsplitter.Theliquidlevelinthereboilerwaskeptconstanttocontrolthebottomflowrate.Membranepumps(Gamma4-RS,ProMinent)wereusedasfeedpumps.Sincethemeltingpointofdecanoicacidis304K,theacidwasstoredinathermostatedglassvesselwithaholdupof6dm3workingat323K.Furthermore,avesselequippedwithheatingbandslocatedonthebalancewasapplied.Allfeedstreamsweremeasuredbydeterminingthemassflowusingbalanceswithanaccuracyof(1%.ThecolumnwascontrolledbyaprocesscontrolsystembasedonOptoboards(Opto-ware),whichareconnectedtoaPC(WinNT-worksta-tion).ThetemperaturesweremeasuredusingPt-100thermometerswithanaccuracyof(0.1K.Thermom-eterswereinstalledatthelowerendofeachcolumnsection(exceptatthefeedpositions),inthereboiler,inthevaporstreamenteringthecondenser,andintheliquidrefluxfromthecondenser.Thetopandthebottompressureaswellasthepressuredropwererecordedbypressuretransducers(Bosch,accuracy(0.1%).Atthelowerendofeachcolumnsection,inthereboilerandfromthedistillatestream,liquidsamplesweredrawnbymeansofasyringeandimmediatelycooledto277K

Ind.Eng.Chem.Res.,Vol.42,No.15,2003

3615

Figure3.Setupofthereactivedistillationexperiments(28theoreticalstagesforKatapak-Sand20theoreticalstagesforKatapak-SP).

Figure4.Profileofliquid-phasecompositionforS-2(Katapak-S).Experimentaldata:MeOH,4;DecH,0;H2O,×;MeDec,O.Simulationresults:MeOH,blackcontinuousline;DecH,---;Hgreycontinuousline.

2O,- -;MeDec,toavoidfurtherreactionandanalyzedbygaschroma-tography.Belowthereactivesectiontheliquidloadofthecolumnwasmeasuredbyrecordingthetimeneededtocollectaspecifiedamount(40cm3)ofliquidinagraduatedvesselinsidethecolumn.

ExperimentalResults.Experimentshavebeenperformedusingthecolumnsetup,whichisshowninFigure3.ThereactivesectionshowninFigure3consistedofKatapak-SorKatapak-SP.Atypicalcom-positionprofileforthesetupusingthepackingsKata-pak-SisgiveninFigure4.Figure5showsatypicalcompositionprofileforthesetupusingKatapak-SP.ExperimentaldetailsforKatapak-SsetuparegiveninTable4andforKatapak-SPsetupinTable5.Figure6showstheinfluenceoftherefluxratioontheconversion.TheinfluenceofthemolarfeedratioontheconversionispresentedinFigure7.Theprimaryaimoftheexperimentswastheverificationofthesimulationresultsandtoinvestigateexperimentallywhetherahigherseparationefficiency(Katapak-S)orahigheramountofcatalystperstage(Katapak-SP)ismorebeneficialfortheprocessbutnotprimarilytoobtainhighconversions.Therefore,experimentswereper-formedwithlowerreboilerdutiesthanneededforhighconversionstoensurethecompliancewiththesite’ssafetyrequirements.

反应精馏

3616Ind.Eng.Chem.Res.,Vol.42,No.15,

2003

Figure5.Profileofliquid-phasecompositionforSP-1(Katapak-SP).Experimentaldata:MeOH,4;DecH,0;H2O,×;MeDec,O.Simulationresults:MeOH,blackcontinuousline;DecH,---;H- -;MeDec,greycontinuousline.

2O,Table4.ExperimentalDatafortheSetupUsingKatapak-S

runnumber

S-1

S-2S-3S-4S-5P1(mbar)10121000999995995 P(mbar)0.680.820.560.490.47F˙DecH(mol/h)1717171118F˙MeOH(mol/h)2929293427B˙(mol/h)2727262124D˙(mol/h)

1919202421refluxratio(mol/mol)20.510.50.5xD(DecH)0.0000.0000.0000.0000.000xD(MeOH)0.9450.8030.8270.8220.765xD(MeDec)0.0000.0000.0000.0000.000xD(H2O)0.0550.1970.1230.1780.235xB(DecH)0.4670.4070.4400.2380.465xB(MeOH)0.2800.3150.3040.4090.237xB(MeDec)0.1720.2200.2000.3110.272xB(H2O)0.0810.0580.0560.0410.026X(%)26.5836.6532.6456.5436.76Q˙(W)940793810633716wL(mh-1)2.932.622.822.162.82 Feed,DecH(°C)56.1355.0056.4462.0656.69 Feed,MeOH(°C)64.5064.2564.2564.1364.25 1(°C)65.2067.3767.0468.2368.19 14(°C)71.8472.3873.0673.0275.13 28(°C)

85.42

87.35

87.00

78.02

93.03

SeparationEfficiencyofthePacking

TheseparationefficiencyofKatapak-Shasbeendeterminedearlier1andaNTSMvalue1(numberoftheoreticalstagespermeter)of4m-wasfoundformediumwatercontents.Athigherwatercontentstheseparationefficiencyofastructuredpackingisknowntodecreaseduetopoorerwettingcharacteristics.ForSulzer-BXpackingsaNTSMvalueof5m-1wasdetermined.1MeasurementsoftheseparationefficiencyofKatapak-SPpackingshavebeenperformedusingthetestsystemwater-aceticacid.ForthesemeasurementsacolumnconsistingofonesectionequippedwithKata-pak-SPworkingattotalrefluxandatmosphericpres-surehasbeenapplied.Theliquidloadwasmeasureddirectlybelowthecondenser.Thereboilerdutyandthusthevaporandliquidloadwasvariedinarangeexpectedforthereactivedistillationexperiments.Afterthecolumnwasallowedtoreachequilibrium(usuallyafter2h),liquidsamplesweredrawnaboveandbelowthe

Table5.ExperimentalDatafortheSetupUsingKatapak-SP

runnumber

S-1

S-2S-3S-4S-5P1(mbar)10131019102010181017 P(mbar)1.632.052.441.661.61F˙DecH(mol/h)1515151118F˙MeOH(mol/h)2930293427B˙(mol/h)2323242224D˙(mol/h)

2122202321refluxratio(mol/mol)0.511.50.50.5xD(DecH)0.0000.0000.0000.0000.000xD(MeOH)0.7610.8150.8850.7610.721xD(MeDec)0.0000.0000.0000.0000.000xD(H2O)0.2390.1850.1150.2390.279xB(DecH)0.4080.4250.4940.2450.459xB(MeOH)0.2710.3040.2890.4610.249xB(MeDec)0.3140.2540.1600.2450.286xB(H2O)0.0080.0170.0570.0490.006X(%)42.9937.1923.1850.1338.57Q˙(W)

754860900693790wL(mh-1)2.932.722.933.013.19 Feed,DecH(°C)58.7559.0058.0661.8856.25 Feed,MeOH(°C)64.’5064.8864.8164.8164.81 1(°C)68.1267.1366.0067.8768.77 10(°C)74.1372.7574.1973.5074.75 20(°C)

90.38

88.49

86.00

77.86

98.25

Figure6.Conversionasafunctionofrefluxratio.Experimentaldata:Katapak-S,b;Katapak-SP,O.Simulationresults:blackcontinuousline,Katapak-SP,adsorptionbasedkinetics(eq4);greycontinuousline,Katapak-S,adsorption-basedkinetics(eq4);---,Katapak-SP,pseudohomogeneouskinetics(eq2);---,Katapak-S,pseudohomogeneouskinetics(eq2).

Figure7.Conversionasafunctionofreactantratio.Experimen-taldata:Katapak-S,b;Katapak-SP,O.Simulationresults:blackcontinuousline,Katapak-SP,adsorption-basedkinetics(eq4);greycontinuousline,Katapak-S,adsorption-basedkinetics(eq4);---,Katapak-SP,pseudohomogeneouskinetics(eq2);---,Katapak-S,pseudohomogeneouskinetics(eq2).

packing.TheNTSMvaluewasdeterminedusingmea-surementsfortheknownVLEofthesystemwater-aceticacid.1,8TheresultsareshowninFigure8.ThisfigurealsoincludestheNTSMvaluesobtainedforKatapak-S.1Itcanbeexpectedthatlowtomediumwatercontentsarepresentinthereactivesectionthroughoutthereactivedistillationexperimentsper-

反应精馏

Figure8.Separationefficiencyofthepackingsused.LiquidloadforKatapak-SP:1.5m3m-2h-1.ResultsforKatapak-Sweretakenfromliterature.1(Katapak-SP,O;Katapak-S,b).

formedinthisinvestigation.Therefore,theseparationefficiencyofKatapak-SPhasbeendeterminedatme-diumwatercontents.AdecreaseoftheseparationefficiencycanalsobeexpectedforKatapak-SPathigherwatercontents.Furthermore,itshouldbenotedthatreactivedistillationexperimentsareperformedatlowliquidandvaporloadstoachieveareasonableresidencetimeofthereactantsinthereactivesectionofthecolumn.F-Factors2ofabout0.8-1Pa0.5andliquidloadsofabout3m3m-h-1canbeexpected.TheseparationefficiencyisconstantundertheseconditionsandnodependencyfromtheF-Factorwasfound.Therefore,theNTSMvalueforKatapak-SPissetto2m-1forfurthercalculations.ThelowerseparationefficiencyofKatapak-SPisaresultofthelowerspecificsurfaceareaofthepacking.7Simulation

Allsimulationswerecarriedoutwiththesteady-statemodelRADFRAC(Aspen-Plus17Version11.1).ThemodelisbasedonarigorousequilibriumstagemodelforsolvingtheMESHequations.Besidespurecompo-nentpropertiesandmod.UNIFAC(Do)interactionparameters,dataaboutthecolumnheatlossandthereactionkineticsareincorporatedintotheprocesssimulator.Stagesarenumberedfromthetoptothebottom,withstage1ascondenserandstageNasreboiler.ThisresultsinatotalnumberoftheoreticalstagesofN)28forthecolumnequippedwithKata-pak-SandN)20forthecolumnequippedwithKatapak-SP,whileKatapak-Scontains150gofdryAmberlyst-15/mandKatapak-SPcontains210gofdrycatalyst/m.

Results.ItcanbeseenfromFigures4-7thatexperimentaldataareinagoodagreementwithsimu-lationresults.Thesefiguresshowthatonlyanadsorp-tion-basedkineticmodelensuresgoodagreementbe-tweenexperimentalandsimulatedresults.Deviationsbetweenexperimentaldataandsimulationresults(adsorption-basedkineticmodel)arewithinexperimen-talerrorforbothpackingtypes.Asexpected,thecalculatedconversionforthepseudohomogeneousmodelismuchhigherthanthedataobtainedexperimentally.Itcanbeconcludedfromtheexperimentsthatthesimulationmodelisabletodescribebothpackingtypes(Katapak-SandKatapak-SP)quantitativelywhenap-plyinganadsorption-basedkineticmodel.Processsimu-lationcanbeusedforareliableevaluationofimportantdesignfactorsandtoproposeaprocesswhereascale-upispromising.

Ind.Eng.Chem.Res.,Vol.42,No.15,2003

3617

Figure9.Conversionasafunctionofthesizeofthereactivesection(P)3bar,r)0.01,D:F)0.62,fivenonreactivestagesaboveandbelowthereactivesection,F˙MeOH)2 F˙DecH)34mol h-1).(Katapak-S,;;Katapak-SP,---).

ProcessDevelopment

Forthepurposeofgeneralvalidityallfurthercalcula-tionswereconductedwithzerocolumnheatloss.Inafirststep,theroleoftheoperatingconditions(refluxratio,reactantratio,pressure,andD:Fratio)areexamined.Itwasinvestigatedwhetherseparationef-ficiency(Katapak-S)oramountofcatalystperpacking(Katapak-SP)ismorebeneficial.Furthermore,theroleofaprereactorandtheinfluenceofthenumberofreactiveandnonreactivestagesonthecolumnconver-sionwasevaluated.

ReactantRatio.AscanbeseenfromFigure7,anexcessofmethanolinthefeedstreamresultsinhigheracidconversions.Therefore,forallfurthercalculationsatotalfeedratioofdecanoicacid:methanolof1:2isapplied.

RefluxRatio.FromFigure6itcanbeseenthatlowrefluxratiosresultinhigherconversions.Sincewaterisanintermediateboiler,ahigherrefluxratioleadstoahigheramountofwaterinthereactivesection.Therefore,asmallrefluxratioof0.01isusedforfurthercalculations.

Pressure.Elevatingthecolumnpressureleadstohighertemperaturesandhencehigherreactionratesforkineticallycontrolledreactions.Otherwise,usingAmberlyst15itisnotpossibletooperateattempera-turesabove393Ksincethisisthemaximumoperatingtemperaturerecommendedbythemanufacturer.Simu-lationstudiesshowthattemperaturesbelow393Kcanberealizedwhenpressuresupto3bararerealized.D:FRatio.Thedistillate-to-feedratiohasamajorinfluenceonthecountercurrentflowofthereactantsinthereactivesectionofthecolumn.Anoptimalcountercurrentflowisachievedwhentheexcessmetha-nolandthewateriscompletelywithdrawninthedistillatestream.Foratotalmolarfeedrateofdecanoicacid:methanolof1:2thisresultsinanoptimalD:Fratioof0.66.Undertheseconditions,nomethanolispresentinthereboiler,resultingintemperaturesabove500Kinthereboiler.Toreducethetemperature,aD:Fratioof0.62isappliedduringthecalculations,ensuringasufficientamountofmethanolinthereboiler.

HeightoftheReactiveSection.Figure9showstheinfluenceofthelengthofthereactivesectionontheconversionforKatapak-SandKatapak-SP.Aboveandbelowthereactivesectionanonreactivesectionof1meachwasusedforcalculations.ItcanbeseenthatKatapak-Sismorefavorablefortheprocesssincehigherconversionscanbeobtainedwithasmallerreactivesection.Butdifferencesbetweenbothpackingsarerathersmall,indicatingthatbothseparationefficiency

反应精馏

3618Ind.Eng.Chem.Res.,Vol.42,No.15,

2003

Figure10.Overallconversionofthecombinationofprereactorandreactivedistillationcolumn.Conversionisgivenasafunctionofthemethanolfractionfedintotheprereactor.Remainingmethanolisfedintothecolumndirectly.(Columnconditions:P)3bar,r)0.01,D:F)0.62,fivenonreactivestagesaboveandbelowthereactivesection,F˙MeOH)2 F˙DecH)34mol h-1).(Katapak-S,;;Katapak-SP,---).

andamountofcatalystperstageareimportant.The

separationefficiencyiscrucialforthesuccessoftheprocesssincewaterhastoberemovedfromthereactivesectionnotonlytoshiftthechemicalequilibriumtohigherconversionsbutalsotoavoiditsdetrimentalsorptioneffectonreactionkinetics.

Prereactor.Theapplicationofaprereactormightbeadvantageousinreactivedistillationprocesses.3Intheprereactorequilibriumconversionisachievedandthereactivedistillationcolumnisusedforshiftingreactiontohigherconversions.Forthecalculations,aprereactorworkingat323Kisused(KtheDDBsoftware8).Thereactivex)1.8calculatedusingsectionwassetto4m;aboveandbelowthereactivesectionanonre-activesectionof1meachwasapplied.Fortheevalu-ationofthisprocess,analternativecasestudywasperformed.Acertainfractionoftheoverallmethanolwasfedintotheprereactorandtheremainingmethanolwasfedintothecolumnbelowthereactivesection.Thestreamleavingtheprereactorenteredthecolumndirectlyabovethereactivesection.Figure10showstheconversionasafunctionofthemethanolfractionenteringthecolumndirectly.Itcanbeseenthathighconversionscanbeexpectedwhennomethanolisfedintotheprereactor.Anoptimalcountercurrentflowofthereactantsinthereactivesectionofthecolumnisobviouslymoreimportantfortheprocessthantheconversionsobtainedintheprereactor.Thisindicatesthattheapplicationofaprereactorisnotanadvanta-geousprocessalternative.

TechnicallyOptimizedProcess.Thepreviousevaluationpermitsproposalofapromisingprocessforfurtherscale-upandoptimizationwithregardstoeconomicissues.Additionalsimulationstudiesindicatedthatthenonreactivesectionsaboveandbelowthereactivesectiondonothaveanybeneficialeffect.Therefore,theprocessconsistsofacolumnconsistingof6-mKatapak-Spackings.Boilingmethanolisfedintothecolumndirectlyabovethereboiler;decanoicacidentersthecolumndirectlybelowthecondenserwithatemperatureof323K.Thecolumnheadpressureissetto3bar,therefluxratiois0.01,andtheD:Fratioissetto0.62.Areactantratioofdecanoicacid:methanolof1:2ischosen.Figure11showstheconversionasafunctionofamodifiedspacevelocity(SV*)asgivenbyeq6.Here,V˙meansthesumofthevolumeflowsofthe

SV*)

V˙˙

V)

VReactiveSection

π (rColumn)2 h(6)

ReactiveSection

Figure11.Conversionasafunctionofspacevelocityfortheoptimizedprocess(P)3bar,r)0.01,D:F)0.62,nononreactivestages,F˙MeOH)2 F˙DecH)(Katapak-S,;;Katapak-SP,---).

feedstreams,VReactiveSectionthetotalvolumeofthe

reactivesection,rhColumntheradiusofthecolumn,andReactiveSectionthelengthofthereactivesection.Forcomparisontheresultsobtainedforacolumnconsistingof6mofKatapak-SParealsogiven.Itcanbeseenthathighconversionscanbeexpectedevenwhenhighspacevelocitiesareapplied.AgainitcanbeseenthatKata-pak-Sshowsabetterperformanceduetoitshigherseparationefficiency.Conclusion

Amethodforthedevelopmentofreactivedistillationprocessesisappliedfortheesterificationofdecanoicacidwithmethanol.Thethermodynamicaspectshavebeendiscussed.Thereactionkineticshavebeeninvestigated.AkineticapproachbasedupontheLHHWmodelisderived.Ithasbeenshownthatthewateruptakebythepolymericcatalysthastobetakenintoaccountandthatthemodelproposediscapableofdescribingthereactionrateasafunctionofthewatercontent.Fur-thermore,kineticconstantsforthepseudohomogeneousmodelbasedonactivitieshavebeenfitted.TheresultshavebeenincorporatedintotheprocesssimulatorAspen-Plus(AspenRADFRAC).Severalreactivedistil-lationexperimentsatapilot-plantcolumnhavebeenperformedwherebytwodifferentcatalyticpackings(Katapak-SandKatapak-SP)havebeenemployed.Acomparisonoftheexperimentaldatawithsimulationresultsindicatethatanequilibriumstagemodeliscapableofdescribingthecolumnprofilesquantitativelywhenanappropriatekineticmodelisused.Apseudo-homogeneouskineticmodelisnotabletodescribetheexperimentalresults.Withthehelpofreliablesimula-tiontheinfluenceofseveralimportantdesignfactorshasbeeninvestigated.Ithasbeenshownthatbothseparationefficiencyofthepackingandamountofcatalystofferedbythepackingsareimportantforthesuccessoftheprocess.Duetoabetterseparationefficiency,Katapak-Sshowedabetterperformance.Thenumberofreactivestagesnecessarytoachieveconver-sionscloseto100%hasbeendeterminedandtheinfluenceofthetotalmolarfeedonthenumberofthereactivestageshasbeenstudied.Thenumberofnon-reactivestagesaboveandbelowthereactivesectionisofminorimportancefortheperformanceoftheoverallprocess.Furthermore,ithasbeenshownthatapre-reactorisnotadvantageoussincetherequiredcounter-currentflowofthereactantsinthereactivesectioncannotberealized.Finally,atechnicallyoptimizedprocesshasbeenproposed,whichisthebasisforthescale-upoftheprocess.

反应精馏

Acknowledgment

Wethankthe“FondsderChemischenIndustrie”forprovidingascholarshiptoS.St.andCognisDeutschlandforprovidingthedecanoicacidrequiredforthemea-surements.Also,weexpressourthankstoSulzerChemtechforthepackings.Furthermore,wewouldliketothankM.Steinmannforperformingapartofthekineticmeasurements.LiteratureCited

(1)Po¨pken,T.;Steinigeweg,S.;Gmehling,J.SynthesisandHydrolysisofMethylAcetatebyReactiveDistillationUsingStructuredCatalyticPackings:ExperimentsandSimulation.Ind.Eng.Chem.Res.2001,40,1566.

(2)Krafczyk,J.;Gmehling,J.EinsatzvonKatalysatorpackun-genfu¨rdieHerstellungvonMethylacetatdurchreaktiveRekti-fikation.Chem.Ing.Tech.1994,66,1372.

(3)Steinigeweg,S.;Gmehling,J.n-ButylAcetateSynthesisviaReactiveDistillation:ThermodynamicAspects,ReactionKinetics,PilotPlantExperiments,andSimulationStudies.Ind.Eng.Chem.Res.2002,41,5483.

(4)Bock,H.;Wozny,G.;Gutsche,B.DesignandControlofaReactionDistillationColumnIncludingtheRecoverySystem.Chem.Eng.Process.1997,36,101.

(5)Jeromin,L.;Bremus,N.;Peukert,E.KontinuierlicheVer-esterunginReaktionskolonnen.Fette,Seifen,Anstrichmittel1981,83,493.

(6)Schleper,B.;Gutsche,B.;Wnuck,J.;JerominL.EinsatzeineseinfachenSimulationsmodellszurVersuchsplanungfu¨reineGegenstrom-Veresterungskolonne.Chem.Ing.Tech.1990,62,226.(7)Go¨tze,L.;Bailer,O.;Moritz,P.;vonScala,C.ReactiveDistillationwithKatapak.Catal.Today2001,69,201.

Ind.Eng.Chem.Res.,Vol.42,No.15,20033619

(8)DortmundDataBank;DDBSTGmbH(www.ddbst.de),Oldenburg,Germany,2002.

(9)Gmehling,J.;Wittig,R.;Lohmann,J.;Joh,R.AModifiedUNIFAC(Dortmund)Model.4.RevisionandExtension.Ind.Eng.Chem.Res.2002,41,1678.

(10)Gmehling,J.;Kolbe,B.Thermodynamik,2nded.;VCH:Weinheim,1992.(11)Po¨pken,T.;Go¨tze,L.;Gmehling,J.ReactionKineticsandChemicalEquilibriumofHomogeneouslyandHeterogeneouslyCatalyzedAceticAcidEsterificationwithMethanolandMethylAcetateHydrolysis.Ind.Eng.Chem.Res.2000,39,2601.

(12)Xu,Z.P.;Chuang,K.T.EffectofInternalDiffusiononHeterogeneousCatalyticEsterificationofAceticAcid.Ind.Eng.Chem.Res.1997,52,3011.

(13)Pitochelli,A.R.IonExchangeCatalysisandMatrixEffects;RohmandHaasCo.:Philadelphia,PA,1980.

(14)Schimizu,S.;Hirai,C.KineticStudyofLiquid-PhaseEsterificationwithSulfonicAcidCation-ExchangeResinoftheMacroreticularType.I.Heterogeneous-PseudohomogeneousResinCatalysts.Bull.Chem.Soc.Jpn.1986,59,7.

(15)Song,W.;Venimadhavan,G.;Manning,J.M.;Malone,M.F.;Doherty,M.F.MeasurementofResidueCurveMapsandHeterogeneousKineticsinMethylAcetateSynthesis.Ind.Eng.Chem.Res.1998,37,1917.

(16)Rehfinger,A.;Hoffmann,U.KineticsofMethylTertiaryButylEtherLiquid-PhaseSynthesisCatalyzedbyIonExchangeResin.I.IntrinsicRateExpressioninLiquid-PhaseActivities.Chem.Eng.Sci.1990,45,1605.

(17)AspenPlus,version11.PhysicalPropertyMethodsandModels;AspenTechnologiesInc.:Cambridge,MA2001.

ReceivedforreviewNovember18,2002RevisedmanuscriptreceivedMay11,2003

AcceptedMay16,2003

IE020925I

本文来源：https://www.bwwdw.com/article/k4j4.html