Avramovic Et Al., 2012 Ciceq Empirical Modeling of Ultrasound-Assisted Methanolysis

Embed Size (px)

Citation preview

  • 8/13/2019 Avramovic Et Al., 2012 Ciceq Empirical Modeling of Ultrasound-Assisted Methanolysis

    1/13

    Availableonlineat

    AssociationoftheChemicalEngineersofSerbia ChE

    www.ache.org.rs/CICEQ

    ChemicalIndustry&ChemicalEngineeringQuarterly18(1)115127(2012) CI&CEQ

    115

    JELENA M. AVRAMOVIOLIVERA S. STAMENKOVI

    ZORAN B. TODOROVIMIODRAG L. LAZI

    VLADA B. VELJKOVIUniversity of Ni Faculty ofTechnology Leskovac Serbia

    SCIENTIFIC PAPERUDC 662.756.3

    DOI 10.2298/CICEQ110705053A

    EMPIRICAL MODELING OF ULTRASOUND-ASSISTED BASE-CATALYZED SUNFLOWEROIL METHANOLYSIS KINETICSUltrasound-assistedsunfloweroilmethanolysiscatalyzedbyKOHwasstudied

    to define a simple empirical kinetic model useful for reactor design without

    complexcomputation.Itwasassumedthattheneutralizationoffreefattyacids

    and the saponification reactions were negligible. The methanolysis process

    ratewasobservedtobecontrolledbythemasstransferlimitationintheinitial

    heterogeneous regime and by the chemicalreaction inthe laterpseudo-ho-

    mogeneousregime.Amodel featuringirreversiblesecond-orderkineticswasestablishedandusedfor simulationof thetriacylglycerol conversionand the

    fattyacidmethylestersformationinthelatterregime.Goodagreementbetween

    the proposedmodeland the experimental data in the chemically controlled

    regimewasfound.

    Keywords: biodiesel; kinetics; methanolysis; modeling; sunflower oil;ultrasound.

    Fattyacidalkylesters,knownasbiodiesel,arearenewable,biodegradableandnon-toxicfuel,whichisagoodalternativetofossilfuels.Theuseofbiodiesel

    leads to many environmental benefits such as lessair,waterandsoilpollutionandaminimalimpactonhumanhealth[1].Themostcommonlyusedprocessfor biodiesel production is methanolysis, a reactionbetweentriacylglycerols(TAG)fromavarietyofplantoilsandanimalfatsandmethanol,inthepresenceofa catalyst. Base catalysts such as potassium andsodiumhydroxidesandmethoxidesinhomogeneousmedia are usually utilized in industrial biodieselproduction.

    The operational variables that affect fatty acidmethyl ester (FAME) formation are: the type and

    amount ofcatalyst,molarratioofmethanol tovege-tableoil,reactiontemperature,agitationintensityandpurityofthereactants.MethanolandTAGareimmis-cible[2]andthereactionbetweenthereactantsoc-curs on the interfacial surface [3]. Intensivemixingshouldbeappliedinordertoenhancethecontactsur-face area between the two immiscible reactants, Correspondening author: V.B. Veljkovi, University ofNi, Fa-culty ofTechnology, BulevarOslobodjenja124,16000Lesko-vac,Serbia.E-mail:[email protected]:5July,2011

    Paperrevised:25October,2011Paperaccepted:4November,2011

    causingtheincreaseofthemethanolysisreactionrate[3,4].Basedon the knowneffectsof ultrasonic cavi-tationsuchasmixing,heatinganddisruptionofthein-

    terface,someresearchershavespedupthereactionratebycarryingoutmethanolysisin thepresenceofultrasound[5-12].Ultrasound-assistedtransesterifica-tionisanefficient,time-savingandeconomicallyviableprocessasitrequireslowerquantitiesofalcoholandcatalyst [5] and lower energy than the process em-ployingmechanicalagitation[8,12,13].BymaximizingthecontactareabetweentheimmiscibleTAGandal-coholatalowerexternalenergyinputthanmechani-cally-stirred reactors, ultrasonic reactors might im-provetheperformanceofthetransesterificationpro-cessbyobtaininghigherbiodieselyieldswithinshorter

    reactiontimes[14].During the base-catalyzed methanolysis reac-

    tion,aninitialheterogeneousstagecontrolledbythemasstransferrateandachemically-controlledpseu-do-homogeneousstagearewell-recognized.Theformerstageisusuallyobservedatlowerreactiontempera-turesandinthecaseofnon-adequatemixing.Underthese reaction conditions, the methanolysis kineticmodel should include the mass transfer limitations[15].Thekineticsofthebase-catalyzedmethanolysisreactionhavebeenwidelystudiedbyseveralresear-chers.Inthesestudies,eithercomplex(rigorous)ki-

    neticmodelsbasedonthemechanismofthemetha-

  • 8/13/2019 Avramovic Et Al., 2012 Ciceq Empirical Modeling of Ultrasound-Assisted Methanolysis

    2/13

    J.M.AVRAMOVIetal.:EMPIRICALMODELINGOFULTRASOUND-ASSISTED CI&CEQ18(1)115127(2012)

    116

    nolysisreactionconsistingofthreereversibleconse-cutive-competitive reactions [4,16-19] or simplified(empirical)kineticmodelsinvolvingirreversiblereac-tion[10,11,15,20]havebeenused.Eachofthesemo-dels has its specific advantages and shortcomings.The rigorousmodels have the advantage related totheirpredictionaccuracyoverwiderangesofthere-actionconditions,buttheyareusuallymathematicallycomplicated and require complex computer calcula-tions.Inmodelingthechemicallycontrolledstage,for-wardandreversesecond-orderreactionshavemostoftenbeenused [5,17-19].Exceptionally, Freedmanet al. [16] used a combination of the second-orderconsecutiveandfourth-ordershuntreactionstodes-cribethekineticsofmethanolysis.DarnokoandChe-ryan[21]suggestedasecond-orderkineticfortheini-tialstagesofthereaction,buttheyconsideredthefor-ward reactions only. The empirical models assumethe irreversibleoverall reaction, have the rate cons-tantdeterminedbyfittingtoexperimentaldataandal-lowsimplecomputation.Therefore,thesemodelscannotexplaintherealreactionmechanism.Theirmajoradvantage is their simplicity, although it isoftennotcleariftheycanbeextrapolatedoutsidetherangeofthereactionconditionstheyarebasedon.Stamenko-vi et al. [15] reported such a simple kinetic modelthatconsistedofanirreversiblesecond-orderreactionfollowedbyareversiblesecond-orderreactionclosetothecompletionofthemethanolysisreaction.Beingnot only simple, accurateand able todescribe vari-ationsof FAMEandTAG duringmethanolysis reac-tionbutalsorequiringshortcomputationtime,empi-ricalmodelscanbesuccessfullyusedforsimplifyingreactordesignandevenhelpquickdecisionmakinginindustrialconditions.

    Thebase-catalyzedmethanolysiskineticsinthepresenceofultrasoundhashardlybeenstudied.Onlyempiricalmodels have been used and contradictoryresultswerereported.Fora6:1methanol/soybeanoilmolarratioanda25to60C temperature range inthe presenceof low frequency ultrasound (20 kHz),Coluccietal.[20]establishedapseudosecond-orderkineticmodelwithrespecttoTAGforthebase-catal-yzed methanolysis of soybean oil. Mahamuni andAdewuyi[22]concludedfromtheirstudythattheme-thanolysis reaction for conversion of soybean oil tobiodiesel in the presence of ultrasound at the opti-mum conditions (611 kHz and 139 W ultrasound,0.5%catalyst,and6:1molarratioofmethanol/oil)at35Cissecond-order(first-orderwithrespecttobothTAGandmethanol).Ontheotherhand,Georgogiannietal.[10,11]reportedthefirst-andsecond-orderre-actionwithrespecttoTAGforthebase-catalyzedme-thanolysisofsunflowerandcottonseedoilsusingboth

    low frequencyultrasonication andmechanical agita-tion,althoughabetterfitwasobtainedforthefirst-or-derreactionkinetics.Deshmaneetal.[23]verifiedakinetic model based on the first-order reaction fol-lowedbythesecond-orderwithrespecttofattyacidsforultrasound-assistedacid-catalyzedesterificationofpalmfattyaciddistillate.

    We have been investigated the kinetics of theultrasound-assistedsunfloweroilmethanolysiscatal-yzed bypotassiumhydroxide in a continuing study.Recently,wehavereportedtheresultsoftheoptimi-zationofthisprocessobtainedbyusingafullfactorialexperimentdesignwithreplication[24].Theeffectsofmethanol/oilmolarratio,catalystloadingandthere-actiontemperatureonFAMEyieldwereevaluatedbythe analysis of variance and the surface responsemethodology. At the 95% confidence level all threemain operational variablesand the two-way interac-tionofthereactiontemperatureandmethanol/oilmo-larratiowereeffectiveontheFAMEyield.However,themostimportantfactorwasthecatalystloading.Itwas found that the FAME yield generally increasedwiththeincreaseofthemethanol/oilmolarratio,thecatalyst loading and the reaction temperature. Thebest reaction conditions (temperature of40 C,me-thanol/oilmolar ratioof 7.5:1andcatalystloadingof0.7%)ensuredthehighestFAMEyieldofabout95%in60min.However,theoptimumreactionconditionscouldbedifferentforothersystemsdependingonthetypeofsonicator(ultrasonicbathorprocessingprobe),ultrasoundfrequencyandintensity,typeofreactants(oil and alcohol), type of catalyst and the geometryanddimensionsofthereactor.

    In the present work, the kinetics of the ultra-sound-assisted sunfloweroilmethanolysis catalyzedby potassium hydroxide was studied in the samerangesofthreemainoperationalvariablesappliedinthe previous study [24], namely methanol/oil molarratio(4.5:1,6:1and7.5:1),catalystloading(0.3,0.5and0.7%basedontheoilweight)andthereactiontemperature(20,30and40 C).Someotherresear-chershavecarriedouttheirstudiesonultrasound-as-sistedmethanolysisinthesimilarrangesofthereac-tion conditions. The FAME formation in the ultra-sound-assistedmethanolysisof soybeanoilwasnotsubstantially increased with increasing the metha-nol/oilmolar ratio from 6:1 to 12:1 and the catalystloadingfrom0.6to1.0%[22].Forthesamereactiondrivenbylow-frequencyultrasound,Santos etal.[25]

    optimizedthemethanol/oilmolarratioandthecatalystloadingintherangesof3:1to12:1and0.2to0.6%,respectively.By increasing the catalyst loading, theFAMEyieldandreactionratealsoincrease[25]buttheexcesscatalystnegativelyaffectstheFAMEyield

  • 8/13/2019 Avramovic Et Al., 2012 Ciceq Empirical Modeling of Ultrasound-Assisted Methanolysis

    3/13

    J.M.AVRAMOVIetal.:EMPIRICALMODELINGOFULTRASOUND-ASSISTED CI&CEQ18(1)115127(2012)

    117

    [8,11,15,17].Thereactioninthepresenceofsodiumhydroxide at the concentration higher than 1.0%[5,10,12,26] was followed bystrong soap formation,decreasing the FAMEyield.Since the reaction tem-peratureshouldberelativelylowtoreducetheenergyconsumptionandtheuseofultrasoundatboilingtem-peratureisfutile,thetemperaturerangebelow40Cwaschosen.

    Themaingoalofthepresentstudywastode-velop a simpleempirical kinetic model of the base--catalyzed methanolysis reaction in the presenceofultrasound and tosimulate the TAG conversion andtheFAMEformationinthechemicallycontrolledstageoftheprocess.Also,itwasintendedtocomparetheobtainedmodelwiththatforthebase-catalyzedme-thanolysis in the absence of ultrasound to checkwhether the same phenomena control the reactionrateindependentlyofthetypeofmixingapplied(ul-trasoundversusmechanicalagitation).

    Theoretical backgroundThe mechanism of vegetable oil methanolysis

    reactioncanbepresentedbyScheme1,whereC isthebasecatalyst[27].

    The overall methanolysis reaction can be pre-sentedbythefollowingstoichiometricequation:

    A + 3BCatalyst

    3R + S

    whereAisTAG,Bismethanol,RisFAMEandSis

    glycerol.Forthepurposeofmodelingtheultrasound-as-

    sisted sunflower oil methanolysis, the following as-sumptionsareintroduced:

    a)Themethanolysisprocessoccursviatheini-tial heterogeneous regime, followed by the pseudo--homogeneousregime,wherethemasstransferandthechemicalreactioncontroltheoverallprocesskine-tics,respectively.Theseregimesarewellrecognizedinthestudiesofthemethanolysisreactionkineticsintheabsenceofultrasound,whichischaracterizedbythesigmoidalvariationofFAMEyieldwithtime[15].ThesameshapeofthecurvesdescribingtheFAMEyield with the progress the ultrasound-assisted me-thanolysisofsoybeanoilundercertainsonicationcon-ditions,whichindicatestheexistenceofmasstransferlimitations inthe initialreactionperiod,hasbeenre-centlyreportedbyMahamuniandAdewuyi[22].

    b) Themethanolysis of TAG is an irreversiblesecond-orderreactionwithrespecttoTAGintheche-micallycontrolledregime.Duetotheexcessofme-thanolandthelowproductconcentrationonecanex-pect the reverse reactionsto benegligible.The irre-versiblesecond-orderreactionkineticmodelwithres-pecttoTAGhasalreadybeenemployedforbase-ca-talyzedmethanolysisofpalm[21]andsunflower[15]oilsintheabsenceofultrasonicirradiation,aswellasfortheultrasound-assistedbase-catalyzedmethanol-

    ysis of soybean [20,22], sunflower and cottonseed

    Scheme1.Themechanismofvegetableoilmethanolysisreaction.

  • 8/13/2019 Avramovic Et Al., 2012 Ciceq Empirical Modeling of Ultrasound-Assisted Methanolysis

    4/13

    J.M.AVRAMOVIetal.:EMPIRICALMODELINGOFULTRASOUND-ASSISTED CI&CEQ18(1)115127(2012)

    118

    [10,11]oils.c)Thedispersionandthecompositionofeach

    phaseofthereactionmixtureareuniform.d)Thecontentoffreefattyacidintheoilisne-

    gligible,sotheneutralizationreactioncanbeignored.Fromtheoilandinitialcatalystamounts,aswellasthe acidvalue of the oil, it was estimated that onlylessthan3%ofthecatalystamountcouldparticipateintheneutralizationoffreefattyacids.

    e) The saponification reaction is negligible be-causelowcatalystamountsareusedtoenhanceme-thanolysis[5].

    Masstransfercontrolledregime

    In the modeling of the methanolysis kinetics,onlyStamenkovietal.[15]haveincludedthemasstransferlimitationintheinitialheterogeneousregime

    ofthereactioncarriedoutinabatchstirredreactor.According to assumption (a), the rate of TAG con-version(rA)mustbeequaltotherateofTAGmasstransfer from the oilphase towards themethanol/oilinterfacialarea.SincetheTAGconcentrationontheinterfacialareaisequaltozero(cA,s0),itfollows[15]:

    = = A

    A c A

    d

    d

    cr k ac

    t (1)

    wherekcistheTAGmasstransfercoefficient, aisthespecific interfacial area, cA is TAG concentration in

    theoilphaseandtisthetime.TheTAGconcentrationisrelatedtotheconver-

    siondegreeofTAG,xA:

    A A0 A(1 )c c x= (2)

    So,Eq.(1)istransformedintothefollowing:

    Ac A

    d(1 )

    d

    xk a x

    t= (3)

    where cA and cA0 are the actual and initial TAGconcentrations.

    TheinstantaneousoverallvolumetricTAGmasstransfer coefficient, kca, for the mass transfer con-trolledregimecanbeestimatedfromEq.(3)asfollows:

    Ac

    1 d

    (1 ) dA

    xk a

    x t=

    (4)

    Chemicalreactioncontrolledregime

    Accordingtotheassumption(b),therateoftheTAGconversioninthechemicallycontrolledregimeis:

    2AA 2,app A

    d

    d

    cr k c

    t = = (5)

    Theapparentreactionrateconstant,k2,app,isex-pected to depend on the reaction temperature, theinitial methanol concentration and the catalyst con-centrationas:

    2,app B0 C( , )kf c c =k (6)

    wherekisthereactionrateconstant,cB0istheinitialmethanolconcentrationandcCisthecatalystconcen-tration.

    Equation(6)canbeeasilytransformedinto:

    ( )2A

    2,app A0 A

    d1

    d

    xk c x

    t= (7)

    Forboundaryvalues: t = t1, xA =xA1and t = t,xA=xA(thebeginningandanymomentofthechemi-callycontrolled regime, respectively), the integration

    ofEq.(7)gives:

    2,app A0 1A

    1

    1k c t C

    x= +

    (8)

    wheretheintegrationconstantisdefinedas:

    1 2,app A0 1A1

    1

    1C k c t

    x

    = +

    (9)

    Theapparentreactionrateconstant, k2,app,andthe integration constant,C1, can be estimated fromthe slope and interceptof the lineardependence of

    ( )A1 1 x

    versustime,respectively.Withknownva-lues of the apparent reaction rate and integrationconstant,theTAGconversiondegreecanbecalcul-atedforanymomentofthereactionduringtheche-micallycontrolledregimeusingEq.(9).

    EXPERIMENTALMaterials

    Refined,ediblesunfloweroil(Vital,Vrbas,Ser-bia)wasused.Theacid,saponificationandiodineva-luesoftheoilwere0.09mgKOH/g,190mgKOH/g

    and129gI2/100grespectively,determinedbyAOCSofficialmethods[28].Thecontentofwaterwas0.07%asdeterminedbythevolumetricKarlFishertitration.Absolute methanol of 99.5% purity was purchasedfromZorkaPharma(abac,Srbija).Potassiumhydro-xidepelletsofmin95%puritywerepurchasedfromMos-Lab (Belgrade,Srbija).Hydrochloricacid (36%)wasobtained fromLach-Ner (Neratovce,CzechRe-public).Methanol,2-propanolandn-hexane,allofHPLCgrade,wereobtainedfromLab-Scan(Dublin,Irland).The HPLC standards for methyl esters of palmitic,stearic, oleic and linoleic acids, triolein, diolein and

    monooleinwerepurchasedfromSigmaAldrich.

  • 8/13/2019 Avramovic Et Al., 2012 Ciceq Empirical Modeling of Ultrasound-Assisted Methanolysis

    5/13

    J.M.AVRAMOVIetal.:EMPIRICALMODELINGOFULTRASOUND-ASSISTED CI&CEQ18(1)115127(2012)

    119

    EquipmentThereactionwascarriedoutina250cm3three-

    neck round-bottom glass flaskequippedwith a con-denserwhichwasimmersedinanultrasoniccleaning

    bath(Sonic,Ni,Serbia;totalpower:350W;andin-ternaldimensions:30cm15cm20cm)operatingat40 kHz frequency. The ultrasound generator had tobeswitchedoffafter60minandthereforetheultra-sound-assistedreactionwasbeingfollowedfor60min.Thebathwasfilledwithdistilledwaterupto1/3ofitsvolume (about 2.5 dm3). The temperaturewas con-trolledandmaintainedatthedesired level(0.1C)bywatercirculatingfromathermostatedbathbymeansofapump.

    Reaction conditionsThe methanolysis of sunflower oil was carried

    outatmethanol/oilmolarratiosof4.5:1,6:1and7.5:1,KOHloadingsof0.3,0.5and0.7%(basedontheoilweight)andthereactiontemperaturesof20,30and40 C under atmosphericpressure. The experimentsweredoneinduplicate.

    Experimental procedureMethanol and KOH were fed into a three-neck

    glass vessel. The vessel was placed into the ultra-sonicbathandsubmittedtoultrasoundatthedesiredtemperatureuntilallthecatalystwasdissolved.Thesunfloweroil(45.96 g)was thermostated separatelyandaddedtothevessel.Assoonastheoilwasad-ded, the reaction was timed. During the reaction,samples(1cm3)wereremovedfromthereactionmix-ture at certain time intervals, immediately quenchedbyaddingarequiredamountoftheaqueoushydro-chloric acid solution (4.5%) to neutralize KOH andcentrifugedat3500rpm(average700 g)for15minusinga laboratory centrifuge (LC320,Tehtnica, Ze-lezniki, Slovenia). The upper layer was withdrawn,dissolvedina2-propanol/n-hexane(5:4v/v)inanap-propriateratio(1:200)andfilteredthrougha0.45mMilliporefilter.TheresultingfiltratewasusedforHPLCanalysis.

    Composition of the reaction mixtureThecompositionofthereactionmixturesamples

    wasdeterminedbyanHPLCchromatograph(Agilent1100Series)usingthesomewhatmodifiedmethodofHolapek et al. [29] as described elsewhere [3]. AZorbax Eclipse XDB-C18 column (4.6mm150mmwith5mparticlesize)heldatconstanttemperature(40 C)was used for separation. Themobilephasewasmethanol(reservoirA)and2-propanol/n-hexane(5:4v/v;reservoirB).Alineargradientfrom100%A

    to40%A+60%Bin15minwasemployed.Themo-

    bilephaseflowratewas1cm 3/min.Theinjectionvo-lumewas20Landthecomponentsweredetectedat205nm. The calibrationcurveswerepreparedbyusingthestandardmixtureofFAMEandthestandardacylglycerols and used for the quantification of theFAMEand the acylglycerols present in the samplesanalyzed.

    The conversion degree ofTAGwascalculatedfrom the content ofTAG in the FAME/oil fractionofthe reaction mixture, TAG (in %), by the followingequation:

    A 1 /100x TAG = (10)

    TheTAGconversiondegreewasfittedtogiveasigmoidalcurve(Boltzmanfunction),whichdescribesappropriatelythevariationofTAGconversiondegree

    from its lowest (= 0) to the highest (A2) asymptoticvalue:

    1 22

    0

    1

    1 expA

    A Ax A

    t t

    t

    = +

    +

    (11)

    wheret0isthetimeatwhichTAGconversiondegreeishalfwaybetweenthelowestandthehighestvalue,andt1isthesteepnessofthecurve.ThevaluesofA1,A2,t0andt1werecalculatedbythenon-linearregres-sionmethod usingacomputerprogram (SigmaPlot11.0,trial).

    RESULTS AND DISCUSSIONMethanolysis reaction analysis

    Typical variations of the reactionmixture com-positionwiththeprogressoftheultrasound-assistedsunfloweroilmethanolysisareshowninFigure1.Thedecreaseof the TAG concentrationwas followedbytheincreaseofFAMEconcentration.Theconcentra-tions of intermediate products, monoacylglycerols(MAG) and diacylglycerols (DAG), increased at thebeginning of the reaction achieving the maximum,

    thendecreasedandfinallyremainednearlyconstant.The same shapesof the curves representingvaria-tionsofthereactionmixturecompositionwiththere-actiontimehavealreadybeenobservedforthebase-catalyzedsunfloweroilmethanolysisintheabsenceofultrasound[15].

    The shape of the curves representing the va-riations of FAME fraction with time was sigmoidal,showing the change of the reaction rate with time.The initial rate was slow due to mass transfer limi-tationcausedbyaverylowinterfacialareaavailableforthereaction.Asthereactionproceeded,thedrops

    ofmethanolweredisintegratedbyanultrasoundac-

  • 8/13/2019 Avramovic Et Al., 2012 Ciceq Empirical Modeling of Ultrasound-Assisted Methanolysis

    6/13

    J.M.AVRAMOVIetal.:EMPIRICALMODELINGOFULTRASOUND-ASSISTED CI&CEQ18(1)115127(2012)

    120

    tion and additionally stabilized by MAG, DAG andsoapsthatactedasemulsifiers[30],whichresultedintheincreaseoftheinterfacialareaavailableformasstransfer.Wedidnot observe thatultrasoundwasasmuchefficientinemulsifyingthereactantsatthebe-ginningofthereactionaspreviouslyreportedbySta-varacheetal.[7]accordingtowhomthemajorpartoftheultrasound-assistedbase-catalyzedalcoholisysofvegetableoilstakesplaceinthefirst3to10minofthereaction.Theobserveddisagreementbetweenthatandthepresentstudyisprobablyduetodifferentre-actorgeometriesanddimensions(a250 cm3round--bottomflaskinthepresentstudyanda100cm 3Er-lenmeyer flask in the study of Stavarache and co-workers [7],whichwere filledwith approximately 64and69cm3ofthereactionmixture,respectivelyat6:1methanol/oilmolarratio)andsonicationintensity(150and840Woftotalpower,respectively)appliedinthetwo studies. The specific power density was likelylowerin thepresentstudy,causingaloweremulsifi-cation rate of the oil into methanol, a smaller inter-facial area between the immiscible reactants and alower methanolysis reaction rate. This conclusionseemedtobesupportedbytheresultsofMahamuniand Adewuyi [22] who first reported the sigmoidalshapeofcurves representingthe variationofFAMEyieldduringthebase-catalyzedmethanolysisofsoy-beanoilinthepresenceofhigh-frequencyultrasoundatcertain reactionconditions.Theyshowed thatpo-wer and frequency of ultrasound played significantrolesinthemethanolysisreaction,affectingthedropbreakage process and the critical size of the cavi-tationbubble,respectively.Theinitialinductionperiodshortened and the formation rate increasedwith in-creasing the powerofultrasound,whileanoptimum

    frequency formaximumFAMEformationfora set ofotherreactionconditionswasobserved.

    ThesigmoidalshapeofthecurvesrepresentingthevariationoftheFAMEyieldmeansthatthemetha-nolysisreactionoccurredviatheinitialheterogeneousregimewhereTAGmasstransferfromthebulkoftheoilphasetotheinterfacialarealimitstheoverallpro-cessrate(assumption(a)).Thus,itwouldbeusefultoanalyze the variation of the overall volumetric TAGmasstransfercoefficientwithreactiontimeatdifferenttemperatures. For this purpose,Figure 2 shows thevariationsoftheoverallvolumetricTAGmasstransfercoefficientwiththereactiontimeat20,30and40 Cin the initial mass transfer controlled regime of themethanolysisprocessperformedatmethanol/oilmolarratioof7.5:1andcatalystloadingof0.7%.Theoverallvolumetric TAG mass transfer coefficient was cal-culatedusingEq.(4)andsigmoidal-fittedvaluesofxA(Eq.(11)).

    Athigherreactiontemperature,theoverallvolu-metric TAG mass transfer coefficient was higher atany moment of the reaction and increased rapidlywith themethanolysisprogress. This was explainedbyafasterMAGandDAGformationathighertempe-ratures,whichwasfavorabletothestabilizationoftheemulsionformed.Theincreaseofthereactiontempe-raturedecreasedthereactionmixtureviscosity,whichfavoredtheagitationactiononthedropbreakagepro-cess[15].ThespecificinterfacialareawasincreasedduetothedropbreakagecausingtheincreaseoftheoverallvolumetricTAGmasstransfercoefficient.Onthe other hand, the TAG mass transfer coefficientdecreasedwith thedropsize reductiondueto lowerinternalcirculationinsidesmalldrops.Thedecreaseof TAGmass transfer coefficient with the drop size

    Figure1.Variationsofthereactionmixturecompositionwiththeprogressofthesunfloweroilmethanolysis(FAME-;MAG-

    ;

    DAG-andTAG-;reactionconditions:30C,methanol/oilmolarratio:6:1andKOHamount:0.7%).

  • 8/13/2019 Avramovic Et Al., 2012 Ciceq Empirical Modeling of Ultrasound-Assisted Methanolysis

    7/13

    J.M.AVRAMOVIetal.:EMPIRICALMODELINGOFULTRASOUND-ASSISTED CI&CEQ18(1)115127(2012)

    121

    reduction was predicted by a simulation model ap-plied for alkaline transesterification reaction of soy-beanoilwithalcoholscatalyzedbypotassiumhydro-xidewithinanultrasonic reactor [31].Therefore, theincreaseofthespecificinterfacialareahadaprimaryeffectontheincreaseoftheTAGmasstransferrateandtheFAMEformationrate,bothwiththeincreaseinthereactiontemperatureandwiththemethanolysisprogress.

    Reaction kinetic modelAccordingtotheassumption(a),themethanol-

    ysisprocessoccurs viatheinitialheterogeneousre-gime followed by the pseudo-homogenous regimewhere the mass transfer and the chemical reactioncontrol the overall process kinetics. To model thekineticsofTAGmasstransfer,experimentaldataonthe alcoholic drop size in the reaction mixture wasneeded [15]. Since the alcoholic drop size was notmeasuredduringthemethanolysisprocess,thepre-sentkineticmodelincludedonlythelatterregime.Ty-

    picalvariationsof ( )A1 1 x withtimeatthreetempe-ratures(20,30and40C)andcatalystloadings(0.3,0.5and0.7%)atthe7.5:1molarratioofmethanoltosunfloweroilare shown inFigure 3. Itwas obviousthatEq.(7)correspondingtotheirreversiblepseudosecond-order reaction (assumption(b))agreed quitewell with theexperimentaldata during the chemicalreaction controlled regime. This was in agreementwith the results of Colucci et al. [20] and Georgo-giannietal.[10,11],whoestablishedtheirreversiblepseudo second-order kinetic model with respect toTAGforthebase-catalyzedmethanolysisofsoybean,

    sunflowerandcottonseedoilsinthepresenceoful-

    trasound.Theirreversiblepseudosecond-orderkine-ticshasalsobeenreportedforthepseudo-homoge-neousstageofvegetableoilmethanolysisintheab-sence of ultrasound such as for sunflower [3] andpalm [21] oil. The proposed kinetic model satisfac-torily fitted the changes of the FAME concentrationwith the reaction time in the chemical reaction con-trolled regime, as it couldbeconcluded from the li-nearcorrelationcoefficientandtherelativedeviations

    of the calculated and the experimentalTAG conver-siondegrees.ThevaluesofR2werehigherthan0.91,and the relativedeviationsofthecalculatedand theexperimentalTAGconversiondegrees (in the rangewheretheyfitted)werebetween13.04%inallexpe-riments,showingthatthekineticmodelisvalid.So-mewhat higher deviations were observed under thereaction conditions close to the equilibrium (for ins-tance,curveforthecatalystloadingof0.7%inFigure3), where a small error in the FAME determinationmightresultinagreaterrorinthecalculatedvalueof

    ( )A1 1 x , which is inherent to this reciprocal func-

    tion.Tochecktheorderofthebase-catalyzedmetha-

    nolysisof vegetableoils different from sunfloweroilunderultrasoundirradiation,weanalyzedthekineticdataofStavaracheetal.[7]ontheTAGconversion.Figure 4 shows the pseudo second-order reactionwith respectto TAG forcommercial,cornandgrapeseedoilsduringthewholereactionperiodandonlyintheinitialperiodofthereactionforcanolaandpalmoils.Sincedifferentvegetableoilscontainvariousty-pesoffattyacidsindifferentconcentrations,thekine-ticmodelandthevalueofthereactionrateconstant

    aregenerallyaffectedbythe typeofrawmaterial. It

    Figure2.VariationsoftheoverallTAGmasstransfercoefficient,estimatedbyEq.(4),atvarioustemperatures

    (reactiontemperature,C:20-;30-and40-;methanol/oilmolarratio:4.5:1andKOHamount:0.7%).

  • 8/13/2019 Avramovic Et Al., 2012 Ciceq Empirical Modeling of Ultrasound-Assisted Methanolysis

    8/13

    J.M.AVRAMOVIetal.:EMPIRICALMODELINGOFULTRASOUND-ASSISTED CI&CEQ18(1)115127(2012)

    122

    Figure3.ThekineticmodelofTAGmethanolysisatvarioustemperatures:20(a),30(b)and40C(c)(reactionconditions:

    methanol/oilmolarratio7.5:1;catalystloading,%:0.3-0.5-and0.7-).

    appears that base-catalyzed methanolysis of vege-tableoilswithhighpercentageoflinoleicacidfollowsthe second-order reaction kinetics such as for the

    commercial,corn,grapeseed(52.5,66.5and82.2%,

    respectively) [6] sunflower (68.7%) [11], cottonseed(57.4%) [10] and soybean (typically 55.5%) [22,32]oils.However,fortheoilswithhighercontentsofpal-

    miticandoleicacids,suchascanola(60.0%ofoleic

  • 8/13/2019 Avramovic Et Al., 2012 Ciceq Empirical Modeling of Ultrasound-Assisted Methanolysis

    9/13

    J.M.AVRAMOVIetal.:EMPIRICALMODELINGOFULTRASOUND-ASSISTED CI&CEQ18(1)115127(2012)

    123

    acid)andpalm(44.0and39.4%ofpalmiticandoleicacid, respectively) oil [6], it appears that the base-catalyzedmethanolysis is the second-order reactiononly in the initialperiod. Darnoko and Cheryan [21]alsoobservedthatpalmoilmethanolysiswith1%ofKOHas catalystandamolar ratio of 6:1in theab-senceofultrasoundwaspseudosecond-orderduringthefirst30minofthereaction.ThismeansthatthespecificdistributionofthefattyacidsinTAGofvege-tableoilsaffectthemethanolysiskinetics[7].Indeed,thesaturationdegreeandchainlengthoffatty-acidspresent in a vegetableoilseems toaffectbiodieselconversion by influencing both the FAME yield andthe reaction time [33],or inotherwords themetha-nolysisrate.Vegetableoilscomprisingprimarilylonger,unsaturatedfatty-acids(oleic,linoleicorlinolenicacid)needhalfofthereactiontimerequestedbythoseoilscontainingshorter,saturatedfatty-acids(lauric,miris-tic,palmiticorstearicacid)toachievethemaximumFAMEyield.

    Whentheirreversiblepseudosecond-orderap-parentreactionrateconstantwaspreliminarilycorre-latedwiththecatalystconcentration,theexperimentaldataobtainedatamethanol/oilmolarratioandare-action temperature gathered around a straight linepassingthroughtheoriginpoint.Sinceseparatelineswere obtained for the possible combinations ofme-thanol/oil molar ratio and the reaction temperature,we concluded that the irreversible pseudo-secondorder apparent reaction rate constant was affected,asexpected,notonlybythereactiontemperatureandthecatalystconcentrationbutalsobytheinitialme-thanolconcentration.Inthenextstep,wecorrelated

    2,app B0k c with cC atdifferent reaction temperaturesandlinearcurveswereobtained,asitcanbeseeninFigure5.Theslopeofthelinearcurvesrepresentedthe reaction rate constant, k. Now, Eq. (6) can berewrittenasfollows:

    2,app B0 Ckc c=k (12)

    When the ln k was correlatedwith 1/T then astraightlinewasobtained,asitcanbeseeninFigure6.Thus,theArrheniusequationcanbeappliedforde-termining the activationenergy for themethanolysisreactions:

    ( )aexpk A E RT = (13)

    whereA is the pre-exponential factor,Ea is the ac-tivationenergy,andRisthegasconstant.Theacti-vationenergy and the pre-exponential factor for theirreversible pseudo second-order reaction were cal-culatedtobe46.2kJmol-1and2.45107L3mol-3.min-1,

    respectively(R2=0.999).Bothhigherandlowerva-luesoftheactivationenergy(53.1and30.9kJmol -1)has been recently published for the KOH-catalyzedmethanolysisofsoybeanoilinthepresenceoflow-frequency [20] and high-frequency [22] ultrasound,respectively.Generally,theactivationenergyforthemethanolysis reaction in the presence of ultrasonicirradiationisofthesameorderasthatinitsabsence(53.3 kJ mol-1) [15]. This suggests that there is nochangeinthereactionmechanisminthepresenceofultrasonicirradiation,whichcanbeexplainedbypu-rely physical nature of the ultrasound effect on the

    methanolysisreaction[34].

    Figure4.Thekineticirreversiblesecond-ordermodelofTAGmethanolysisofseveralvegetableoils(commercialoil:o,cornoil:,grape

    seedoil:,canolaoil:andpalmoil:)atthemethanol/oilmolarratioof6:1,thereactiontemperatureof36C

    andKOHloadingof0.5%basedonthedataofStavaracheetal.[7].

  • 8/13/2019 Avramovic Et Al., 2012 Ciceq Empirical Modeling of Ultrasound-Assisted Methanolysis

    10/13

    J.M.AVRAMOVIetal.:EMPIRICALMODELINGOFULTRASOUND-ASSISTED CI&CEQ18(1)115127(2012)

    124

    The kinetic model and the experimental datawere also comparedbasedon the variations of themolar concentrations of TAG and FAME. Figure 7

    shows that the kinetic model agrees well with theexperimentaldatainthechemicalreactioncontrolledregime.ThemolarTAGconcentrationwascalculatedby Eq. (2) and themolar FAME concentration wascalculated using the TAG conversion degree cor-rectedfortheMAGandDAGformation.

    CONCLUSIONSUltrasound-assisted base-catalyzed sunflower

    oilmethanolysiswas studied undervarious reactionconditionsinordertomodelthemethanolysisreactionkinetics. A sigmoidal shape of the FAME yield va-riationwiththeprogressofthemethanolysisreaction

    wasobservedundertheappliedreactionconditions,thus indicating that the overall process included theinitialmasstransfercontrolledregimefollowedbythe

    chemical reaction controlled regime. The former re-gimewasnotincludedinthemodelingoftheoverallprocesskineticsbecausethealcoholphasedropsizewas not measured. In the second regime, the ultra-sound-assisted methanolysis was the irreversiblepseudo second-order reactionwith respect to TAG.Being simple, in good accordance with the experi-mentaldata and useable without complex computa-tion, this empiricalkineticmodelis valuableforche-mical engineeringapplicationssuchasdesign, ope-rationandscale-upofreactorandprocessanalysis,simulationandoptimization.Additionally,beingactu-

    allythesameasthatforthebase-catalyzedmetha-

    Figure5. 2,app B0k c versuscCatdifferentreactiontemperatures(20C-o;30C;and40C-).

    Figure6.Arrheniusplotofthereactionrateconstantversustemperature.

  • 8/13/2019 Avramovic Et Al., 2012 Ciceq Empirical Modeling of Ultrasound-Assisted Methanolysis

    11/13

    J.M.AVRAMOVIetal.:EMPIRICALMODELINGOFULTRASOUND-ASSISTED CI&CEQ18(1)115127(2012)

    125

    Figure7.ThecomparisonofTAGandFAMEconcentrationscalculatedbythekineticmodel(solidlines)withtheexperimentaldata

    (TAG-andFAME-)ata)20,b)30andc)40C(reactionconditions:methanol/oilmolarratio4.5:1;catalystloading:0.7%).

    nolysisintheabsenceofultrasound,thismodelindi-catesthatthesamephenomenacontroltheprocessrateindependentlyofthetypeofmixingapplied,in-directly confirming that the ultrasound effect on themethanolysis reaction is of purely physical nature.Finally,whenapplyingtheempiricalkineticmodelforpredictingthemethanolysisreactionrateandthebio-

    diesel yield under certain reaction conditions, onemustbeawareoftheeffectsofotherfactorssuchasmixing and reactor fluid-dynamics on the reactionrate.Thus,furtherstudiesontheoveralmethanolysisprocessinultrasonicreactorsareneededfortheirre-liabledesignandscale-up.

  • 8/13/2019 Avramovic Et Al., 2012 Ciceq Empirical Modeling of Ultrasound-Assisted Methanolysis

    12/13

    J.M.AVRAMOVIetal.:EMPIRICALMODELINGOFULTRASOUND-ASSISTED CI&CEQ18(1)115127(2012)

    126

    AcknowledgmentThiswork has been funded by theMinistry of

    Education and Science of the Republic of Serbia(Project45001).

    Nomenclaturea-Specificinterfacialarea,m-1A-Pre-exponentialfactor,Eq.(13),L3mol-3min-1A1,A2-ParametersinEq.(11),1cA-ConcentrationofTAGintheoilphase,molL

    -1cA0-InitialconcentrationofTAG,molL

    -1cA,s - Concentration of TAG on the interfacial area,molL-1cB0-Initialconcentrationofmethanol,molL

    -1cC-Concentrationofcatalyst,molL

    -1C1-Integrationconstant,1cR-ConcentrationofFAME,molL-1

    Ea-Activationenergyofthereaction,kJmol-1

    kc-TAGmasstransfercoefficient,min-1

    kca - Instantaneous overall volumetric TAG masstransfercoefficient,min-1k-Reactionrateconstant,L3mol-3min-1k2,app - Apparent reaction rate constant for theirreversible pseudo second-order reaction, L3 mol-3min-1rA-RateofTAGdisappearance,mol

    -1L-1min-1R-Gasconstant,Jmol-1K-1t-Time,min

    t0,t1-ParametersinEq(11),minT-Temperature,KTAG-ContentofTAGintheFAME/oilfractionofthereactionmixture,%xA-TAGconversiondegree,1

    AbbreviationsDAG-DiacylglycerolsFAME-FattyacidmethylestersMAG-MonoacylglycerolsTAG-Triacylglycerols

    REFERENCES[1] J.Hill,E.Nelson,D.Tilman,S.Polasky,D.Tiffany,Proc.

    Natl.Acad.Sci.U.S.A.103(2006)11206-11210[2] H.Zhou,H.Lu,B.Liang,J.Chem.Eng.Data51(2006)

    1130-1135

    [3] O.S.Stamenkovi,M.L.Lazi,Z.B.Todorovi,V.B.Velj-kovi, D.U.Skala,Bioresour. Technol.98 (2007) 2688-2699

    [4] H.Noureddini,D.Zhu,J.Am.OilChem.Soc.74(1997)1457-1463

    [5] C. Stavarache, M. Vinatoru, R. Nishimura, Y. Maeda,Ultrason.Sonochem.12(2005)367-372

    [6]

    J.Lifka,B.Ondruschka,Chem.Eng.Technol. 27(2004)1156-1159

    [7] C.Stavarache,M.Vinatoru,Y. Maeda,Ultrason.Sono-chem.13(2006)401-407

    [8] S.C.Stavarache,M.Vinatoru,Y.Maeda,Ultrason.Sono-chem.14(2007)380-38

    [9] J.Ji,J.Wang,Y.Li,Y.Yu,Z.Xu,Ultrasonics44(2006)411-414

    [10] N.G.Siatis,A.C.Kimbaris,C.S.Pappas,P.A.Tarantilis,M.G.Polissiou,J.Amer.OilChem.Soc.83(2006)53-57

    [11] K.G.Georgogianni,M.G. Kontominas,P.J. Pomonis,D.Avlonitis,V.Gergis,EnergyFuels22 (2008)2110-2115

    [12] K.G.Georgogianni,M.G. Kontominas,P.J. Pomonis,D.Avlonitis, V. Gergis, Fuel Process. Technol. 89 (2008)503-509

    [13] P.Chand, V.R. Chintareddy, J.G.Verkade,D.Grewell,EnergyFuels24(2010)2010-2015

    [14] V.B.Veljkovi,J.Avramovi,O.S.Stamenkovi,Renew.Sust.Energ.Rev.(2011),submittedforpublication

    [15]O.S.Stamenkovi,Z.B.Todorovi,M.L.Lazi,V.B.Velj-kovi, D.U.Skala,Bioresour. Technol.99 (2008) 1131--1140

    [16] B. Freedman, R.O. Buttefield, E.H. Pryde, J. Am. OilChem.Soc.63(1986)1375-1380

    [17] K.Komers,F.Skopal,R.Stloukal,J.Machek,Eur.J.Li-pidSci.Technol.104(2002)728-737

    [18] G.Vicente,M.Martinez,J.Aracil,A.Esteban,Ind.Eng.Chem.Res.44(2005)5447-5454

    [19] F.F.P.Santos,S.Rodrigues,F.A.N.Fernandes,FuelPro-cess.Technol.90(2009)312-316

    [20] G.Vicente,M.Martinez,J.Aracil,EnergyFuels20(2006)17221726

    [21] J.A. Colucci,E.E. Borrero, F.Alape, J.Am. Oil Chem.Soc.82(2005)525-530

    [22] D. Darnoko, M. Cheryan, J. Am. Oil Chem. Soc. 77(2000)1263-1267

    [23] N.N.Mahamuni,Y.G.Adewuyi, EnergyFuels23(2009)2757-2766

    [24] V.G. Deshmane, P.R. Gogate, A.B. Pandit, Ind. Eng.Chem.Res.48(2009)7923-7927

    [25] J.M.Avramovi,O.S.Stamenkovi,Z.B.Todorovi,M.L.Lazi, V.B. Veljkovi, FuelProcess. Technol.91 (2010)15511557

    [26] K.G.Georgogianni,A.K.Katsoulidis,G.Pomonis,M.G.Kontominas,FuelProcess.Technol.90(2009)671-676

    [27] E. Lotero, Y. Liu, D.E. Lopez, K. Suwannakarn, D.A.Bruce,J.G.Jr.Goodwin,Ind.Eng.Chem.Res.44(2005)5353-5363

    [28] AOCS. Official and Tentative Methods. American OilChemistsSociety,Chicago,1980

    [29] M.Holapek,P.Jandera,J.Fischer,B.Proke,J.Chro-matogr.,A858(1999)1331

    [30] M.ernoch, M. Hjek, F. Skopal, Bioresour. Technol.101(2010)20712075

    [31] F.A. Benitez, Ph.D. Thesis, University of Puerto Rico,Mayagez, Puerto Rico, 2004, (available at http://grad-works.umi.com/31/40/3140277.html)

    [32] F.Ma,M.A.Hanna,Bioresour.Technol.70(1999)1-15

  • 8/13/2019 Avramovic Et Al., 2012 Ciceq Empirical Modeling of Ultrasound-Assisted Methanolysis

    13/13

    J.M.AVRAMOVIetal.:EMPIRICALMODELINGOFULTRASOUND-ASSISTED CI&CEQ18(1)115127(2012)

    127

    [33] S.Pinzi,J.M.Mata-Granados,F.J.Lopez-Gimenez,M.D.LuquedeCastro,M.P.Dorado,Bioresour.Technol.102(2011)10591065

    [34] A.Kalva,T.Sivasankar,V.S.Moholkar,Ind.Eng.Chem.Res.48(2009)534-544.

    JELENA M. AVRAMOVIOLIVERA S. STAMENKOVI

    ZORAN B. TODOROVIMIODRAG L. LAZI

    VLADA B. VELJKOVIUniversity of Ni Faculty of Technology

    Leskovac SerbiaNAUNI RAD

    EMPIRIJSKO MODELOVANJE BAZNO-KATALI-ZOVANE METANOLIZE SUNCOKRETOVOGULJA U ULTRAZVUNOM REAKTORUBazno(KOH)katalizovanametanolizasunokretovoguljauultrazvunomreaktorujepro-

    uavanaradidefinisanjajednostavnogempirijskogkinetikogmodelakojiebitiprimen-

    ljivbezsloenogizraunavanjapomouraunara.Pretostavljenojedasereakcijeneu-

    tralizacijeslobodnihmasnihkiselinaireakcijesaponifikacijemoguzanemariti.Uoenoje

    dajebrzinametanolizekontrolisanamaseno-prenosnimogranienjimaupoetnomhete-

    rogenomreimui hemijskomreakcijomu kasnijempseudo-homogenomreimu.Potvr-

    enjemodelkojiukljuujekinetikunepovratnereakcijedrugogreda,kojijezatimkori-

    enzasimulacijukonverzijetriacilglicerolaumetilestremasnihkiselinaukasnijemrei-

    mu.Konstatovanojedobroslaganjeizmeupretpostavljenogmodelaieksperimentalnih

    podatakauhemijskikontrolisanomreimu.

    Kljunerei:biodizel;kinetika;metanoliza;modelovanje;sunokretovoulje;ultra-zvuk.