Table6.KineticEquationandParametersfromReference2a
KineticEquation
reaction:oleicacid+methanolTmethyloleate+waterr)kcxAxM{1-(1/ke)[(xExW)/(xAxM)]}Ccat
activationpre-exponentialenergyheatofreactionparameter
factor(k0)(Ea;kcal/mol)( Hr;kcal/mol)
kc12.93(2.8714.00(0.99
ke
4.17(0.04
2.68(0.05a
Subscripts:A,oleicacid;M,methanol;E,methyloleate;andW,water.Units:kc(moldm3)/(gcat2min);Ccat(gcat/dm3);r(mol/(gcatmin)).
hasbeenmadebyusingasimplifiedreactormodelinwhichthekineticspreviouslydetermined2hasbeenintroduced.InTables3and4arereportedtheaverageReynoldsnumberscalculatedforeachexperimentalrun.Asitcanbeobservedthesevaluesareverylow,andintheseconditionstheflowthroughtheporousmediumisveryprobablyinthecreepingflowregime.12Inthisconditiontheplugflowmodelisnotrigorous;onthecontrary,anaxialdispersionmodelcouldbemoresuitable.Nevertheless,theroughapproximationoftheplugflowreactorhasbeenadoptedforafirsttestingofthepreviouslydeterminedkineticsinthedescriptionofthepresentexperimentaldatacollectedfortubularreactors.Ontheotherhand,Seguinetal.12haveshownamaximumrelativedeviationof10%betweenthepurelyplugflowandtheaxialdispersionmodelsincreepingflowconditions.Theuseofasimplerapproachis,therefore,justifiedasapreliminaryone,consideringboththescatteringofexperimentaldataandtheintroducedapproxima-tions.Moreover,thefluiddynamicbehaviorofacomplexsystemliketheoneusedinthisworkcontainingamixtureofspheresandspringsinthepackedbedisneverdescribedintheliterature.
Referringtoanisothermalplug-flowreactor,thedifferentialmaterialbalanceequationforeachcomponentinthesystem(excludingtriglyceridethatisaninertmedium)canbewrittenas
dFi
dW)Viri)A,M,E,W
(5)
cat
whereFiisthemolarfeedflowrateoftheithcomponent(lettersmeaningA,oleicacid;M,methanol;E,estermethyloleate;W,water),Wcatisthetotalweightofcatalystloadedinthereactor,Viisthestoichiometriccoefficientreferencedtotheithcomponent,andristhereactionrate(seeTable6forparameters,units,andotherdetails).Thenumericalintegrationofthedifferentialequationssystem5,inwhichtheexpressionforrtakenfromourpreviousworkhasbeenintroduced,2resultsincalculatedperformancesthataremuchhigherthanthoseexperimentallyobservedforthethreetubularreactors.Thismeans,inotherwords,thatthepackedbedreactorconfigurationfurnishesalowerconversionwithrespecttothatpredictedonthebasisofthekineticsexperimentallyevaluatedinabatchwellstirredslurryreactor(batchWSSR).ThisparticularbehaviorisreportedinFigure2whereacomparisonisshownbetweentheexperimentalconversionofoleicacidobtainedintherunswiththereactorTR2(squaressymbols)andtheconversionpredictedbythemodelrepresentedbyrelation5(opencirclesymbols).Theconversion,inbothcases,isreportedasafunctionoftheoverallvolumetricfeedflowrate,anditisevidentfromthediagraminFigure2thatthekinetic-basedmodelpredictsahighconversionthatispracticallyinsensitivetotheincreaseinfeedflowratetowhichcorrespondsadecrease
Ind.Eng.Chem.Res.,Vol.46,No.15,20075117
parisonbetweentheexperimentalconversionintubularreactorTR2andthepredictionofthetubularreactormodelbasedonapreviouslydevelopedkineticsfrombatchexperiments.2
inresidencetime.Theexperimentaldatareportedarerelativelyscattered,thevaluesreportedintheplotbeingtheonesobtainedinsteady-stateconditionsafteralongtimeoperation(somehours).Scatteringisduetothefluctuationoftheoperativeconditionssuchasflowrates,temperature,andanalyses.Theflowrateatwhichthemodelstartstopredictadecreaseoftheconversionofoleicacidwasestimatedatabout5000cm3/min.Inthekineticmodelthereactorperformancesarehighenoughthatthesystemreachestheequilibriumcompositionnearlyinalltherunssimulated.Excludinginthissystemtheinfluenceoftheinternaldiffusion,wehavefocusedourattentiontotheinterventionofexternalfluid-to-particlemass-transferlimita-tions.Thefollowingsectionsofthepaperarethendevotedtothefurtherdeepeningofthisaspect.
ExternalDiffusion.Withthepurelykineticmodel(eqs5)unabletocorrectlydescribetheexperimentallyobservedbehaviorofthetubularreactors,anevaluationoftheexternalliquid-soliddiffusionalresistancehasbeenmadebyadoptingthegeneralizedcriterionofMears.11Thiscriterion,referencedtoanisothermalcatalyticpelletandtoann-orderreactionrateequation,isexpressedbythefollowingrelation:
r*RpC<0.15
(6)
0ksn
whereksisthefluid-to-particlemass-transfercoefficient.This
lastparameterhasbeenestimatedinanapproximatedwaybymeansoftheCoeuretcorrelation12,13expressedbythefollowingrelation:
Sh)5.4Re1/3Sc1/4
0.04<Re<30(7)
Thisexpressionhasbeenchosen,amongthewidecorrelationsavailableintheliteratureasreportedinref12,becauseitismoresuitableintermsoflinearvelocity(seethevaluesofReynoldsnumbersreportedinthelastcolumnsofTables3and4)andparticlesizerangesofapplicability.12Thiscorrelationisveryprobablynotfullysuitableforthepackedbedsusedinourexperimentsbecauseitwasoriginallydevelopedforpackedbedsofspheres,while,asmentioned,ourpackinghasamorecomplicatedstructurebeingcomposedofbothparticlesofvarioussizeandofmetallicsprings.However,wehavedevelopedourcalculationalsowithothersimilarcorrelations