ContentslistsavailableatScienceDirect
Industrial Crops and Products
j ou rn a l h o m epa g e :w w w . e l s e v i e r . c o m / l o c a t e / i n d c r o p
Experimental and modelling studies on the solvent assisted hydraulic pressing of dehulled rubber seeds
Muhammad Yusuf Abduh
a,b, C.B. Rasrendra
c, Erna Subroto
a, Robert Manurung
b, Hero J. Heeres
a,∗aDepartmentofChemicalEngineering,ENTEG,UniversityofGroningen,TheNetherlands
bSchoolofLifeSciencesandTechnology,InstitutTeknologiBandung,Ganesha10Bandung,40132,Indonesia
cChemicalEngineeringDepartment,FacultyofIndustrialTechnology,InstitutTeknologiBandung,Ganesha10Bandung,40132,Indonesia
a r t i c l e i n f o
Articlehistory:
Received30December2015 Receivedinrevisedform21June2016 Accepted19July2016
Availableonline6August2016
Keywords:
Rubberseeds Hydraulicpressing Solventassisted Regressionmodel Shiratomodel
a b s t r a c t
Asystematicstudyontheexpressionofrubberseedoilfromdehulledrubberseedsinahydraulicpress wasperformedinthepresenceandabsenceofethanol.Theeffectofseedmoisturecontent(0–6wt%, w.b.),temperature(35–105◦C),pressure(15–25MPa)andethanoltoseedratio(0–21%v/w)ontheoil recoverywasinvestigated.Anoptimumoilrecoveryof76wt%,d.b.wasobtained(1.6wt%moisturecon- tent,14%v/wethanol,20MPa,75◦C,10minpressingtime).Theexperimentaldatasetwasmodeledusing twoapproaches,viz(i)theShiratomodeland(ii)anempiricalmodelusingmulti-variablenon-linear regression.Goodagreementbetweenmodelsandexperimentaldatawasobtained.Relevantproperties oftherubberseedoilobtainedatoptimumpressingconditions(freefattyacidcontent,viscosity,density, waterandP-content,coldflowpropertiesandflashpoint)weredetermined.Thepressedrubberseedoil hasarelativelylowacidvalue(2.3mgKOH/g)andissuitableforsubsequentbiodieselsynthesis.
©2016ElsevierB.V.Allrightsreserved.
1. Introduction
Therubbertree(Heveabrasiliensis)isaperennialplantationcrop whichhasbeencultivatedmainlyasasourceofnaturalrubber.
However,thetreealsoproducesarubberseed,ofwhichthevalori- sationhasreceivedlimitedattentiontillnow.Theyieldofrubber seedsisreportedtobeintherangeof100–1200kg/ha/yr(Stosic andKaykay,1981;AbdullahandSalimon,2009).Fromabiorefin- eryperspective,theidentificationofhighaddedvalueoutletsfor therubberseedsishighlyrelevantasitincreasestheoverallvalue ofthevaluechainfromrubberplantationtoprocessedlatex(Abduh etal.,2013).
Theseedsconsistofakernelsurroundedbyahardshell.The kernelcontains40–50%ofoil(Ramadhasetal.,2005;Njokuetal., 1996)embeddedinaproteinrichmatrix.Theoil,alsoknownas rubberseedoil(RSO),maybeavaluablesourceforbiofuelproduc- tion(Ikwuagwuetal.,2000;Morshedetal.,2011).Inaddition,it mayfindapplicationsaslubricants,ingredientinsoapsandalkyd resins(AigbodionandPillai,2000).Theproteinrichmatrixmaybe usedascattlefeed,asafeedforbiogasproduction,forbinderless
∗Correspondingauthorat:DepartmentofChemicalEngineering,ENTEG,Univer- sityofGroningen,Nijenborgh4,9747AG,Groningen,TheNetherlands.
E-mailaddress:[email protected](H.J.Heeres).
boardproduction(Hidayatetal.,2014)andasafeedforthermo- chemicalprocesseslikepyrolysis(Vazetal.,2005;Kootstraetal., 2011).
Anumberofstudieshavebeenreportedontheexpressionof RSOfromtherubberseedsandthesearesummarizedinTable1.
Moststudiesinvolvesolventextractionusingahydrocarbonsol- vent(hexane,petroleumether)orachlorinatedsolvent.Theoil yieldscoverawiderangeandarebetween5and49%.
Threestudieshavebeenperformedusingmechanicalpressing, involvingeitherahydraulicorscrewpress.Theyieldsinthiscase aretypically lowerthanforsolventextractionandbetween5.4 and28.5%.Improvedyieldsarepossiblebyusingsolventassisted hydraulicpressing.Forinstance,Morshedetal.(2011)showedthat theuseofhexaneinmechanicalpressingincreasedtheyieldfrom 5.4%to49%yield(Table1).Additionofasolventduringoilpress- inghasalsobeenappliedsuccessfullytoincreaseoilyieldsforthe extractionofcotton,sunflowerandsoybeanseeds(Abrahametal., 1993;Dufaureetal.,1999).
Table1showthatthefreefattyacid(FFA)contentoftheproduct oilsvariesfrom2to38wt%forthereportedstudies.Forbiodiesel synthesis,anFFAvalueof3wt%(Meheretal.,2006)isacceptable.
ThehighFFAvaluesarenotnecessaryanintrinsicfeatureofthe RSObutwilldependontheprocessingconditionsandtechnology,
http://dx.doi.org/10.1016/j.indcrop.2016.07.025 0926-6690/©2016ElsevierB.V.Allrightsreserved.
Table1
Overviewofliteraturestudiesonoilisolationfromrubberseeds.
Isolationtechnique Conditions OilYielda FFAcontent(wt%)c Reference
Solvent 68–69◦C,n-hexane,6h 41.6% 7.6 AbdullahandSalimon(2009)
Solvent 40–60◦C,petroleumether 45.6% 2 Ikwuagwuetal.(2000)
Solvent 27◦C,carbontetrachloride,overnight 38.9%b – Haqueetal.(2009)
Solvent 68–69◦C,n-hexane 49% 4 Zhuetal.(2011)
Solvent 68–69◦C,n-hexane,4h 45% – Ebeweleetal.(2010)
Mechanical(hydraulic) 70◦C,8MPa,10wt%m.c 28.5% 38 Ebeweleetal.(2010)
Mechanical 27◦C 5.4% – Morshedetal.(2011)
Mechanical+solvent 27◦C,hexane/seedwt.ratio:0.8% 49% 2 Morshedetal.(2011)
aKernel(dehulledseed)unlessstatedotherwise.
b Seedestimatedfromacidvalue(mgKOH/g).
c Estimatedfromacidvalue(mgKOH/g).
andalsobythestorageconditionsoftheseeds(Zhuetal.,2011;
Ebeweleetal.,2010).
Thispaperpresentsasystematicstudyoftheinfluenceofpres- sure,temperature,moisturecontentandtheuseofasolventonthe pressingbehaviourofdehulledrubberseedsinalaboratoryscale hydraulicpress.Theseprocessvariableshaveshowntobeofhigh importanceforbothoilyieldsandproductquality(Venteretal., 2007;Willemsetal.,2008;Subrotoetal.,2015).Alargenumber ofexperimentswereperformedand modeledusingappropriate models.Ethanolwasselectedasthesolventofchoiceasitmaybe obtainedfromrenewableresources.
2. Theory:theShiratomodel
Severaldifferentmathematical modelsfor theexpression of oilseedshavebeendeveloped,whichmaybecategorisedas(i)mod- elsbasedonthenatureofcellstructures(Mremaand McNulty, 1985), (ii)empiricalmodels(Fasinaand Ajibola,1990),and (iii) Terzaghi-typemodels(Shiratoetal.,1986).Thefirstmodeltype providesfundamentalinsightsintheexpressionprocess.However theinformationofthecellstructureandcelldimensionsisnoteasy toobtained,whichlimitstheapplicability.
Empiricalmodelsenablethepredictionofoilyieldsbutareoften limitedtospecificseedsandprocessingequipment.Terzaghimod- elsallowforagooddescriptionoftheexpressionprocess,however, themodelassumesthatthecakethicknessremainsconstantdur- ingpressingwhichis oftennota goodassumption.TheShirato modelisamodifiedTerzaghi-typemodel.Ithasbeenpreviously appliedsuccessfullytomodelthehydraulicpressingofdrycocoa nibsandseveraloilseeds(Venteretal.,2007;Willemsetal.,2008).
Itisadynamicmodelwhichusesthecakethicknessofasample asafunctionoftimeandprocessingparametersasinput.Thecake thicknessisexpressedastheconsolidationratio(Uc),definedas thedifferencebetweencakethicknessesatthestartoftheprocess andthecakethicknessattimetdividebythemaximumdifference incakethickness(beforeandaftertheprocess).Thisconsolidation ratiocanbedescribedasafunctionoftime,pressureandmaterial properties(Eq.(1)).
Uc(t)= L (0)−L (t)
L (0)−L (tend)=(1−B)
1−exp
−2Cet 420+B
1
−exp
−
EG
.t where
Ce= P
1s˛ıPıe (1)
with
Ucconsolidationratio(−) Locakethicknessattinitial(m) L(t)cakethicknessattinitial(m)
Lendcakethicknessattend(m)
Brelativecontributionofsecondaryconsolidation Ceconsolidationcoefficient(m2/s)
ωovolumeofsolidsperunitarea(m3/m2) ttime(s)
E/Gcreepconstant(s−1) Ppressure(Pa)
1liquidviscosity(Pa.s) ssolidsdensity(kg/m3)
␣filtrationresistance(m/kg) evoidratio(−)
TheShiratomodelconsistsofthesumoftwoterms,primary andsecondaryconsolidation(creep).Therelativecontributionof thesecondtermisgivenbythe(fit)parameterB.Foranindividual pressingexperiment,theconsolidationratioversustimeisdeter- minedataconstantvalueofthepressureP.Parameterfittingallows calculationofthevalueof␣,BandE/Gforthisparticularexper- iment.Whenperformingexperimentsatdifferentpressures,the valuesfortheindividualcakeresistances␣maybecorrelatedusing thefollowingrelation:
␣=␣0.
1+ P Pa
(2) where
˛omaterialconstantforfiltrationresistance(m/kg)
materialconstantforfiltrationresistance(−)
Parameter fitting for theexperiments at differentpressures allowscalculationof␣0,thepressureindependentfiltrationresis- tance and thevalue of, which is a measure for thepressure dependenceofthecakeresistance.Alargevalueforisindicative forahardmaterial.
Besidesthecakethickness,theoilyieldintheformofanoil recoveryisalsodeterminedforeachpressingexperiment.Theoil recoverycanberelatedtomaterialpropertiesusingEq.(3).
Oil recovery (wt%)=
(1−F0)o(1−ε)sFo −1
×100% (3)
With:
finalaverageporosity(m3nonsolid/m3total) Fooriginaloilcontentoftheseeds(wt%,d.b.) ooildensity(kg/m3)
ssoliddensity(kg/m3)
ThevaluesforF0,o ands weredeterminedexperimentally inseparateexperiments(seeexperimentalsectionfordetails).In combinationwiththeexperimentallydeterminedoilrecoveryfor anindividual experiment,the cakeporosity for each individ- ualexperimentmaybedetermined.values forexperimentsat differentpressuresmaybecorrelatedusingEq.(4):
(1−)=(1−0).
1+ P Pa
n(4)
Here,
εomaterialconstantforporosity(m3nonsolid/m3total) nmaterialconstantforporosity(−)
Pathresholdpressure(Pa)
Thethresholdpressurewassetat5MPa.
3. Materialandmethods
3.1. Material
Seedsfromtherubbertree(Heveabrasiliensis)wereobtained from Bengkulu,Indonesia. Ethanol (absolute, proanalysis) was obtainedfromMerck(Darmstadt,Germany).
3.2. Moisturecontentconditioning
Thetotalmoisturecontentofthesampleswasdeterminedusing MethodB-14of theGermanstandard methods(DGF,2002). It involvesheatingthedehulledrubberseedsintheovenat103◦C untilconstantweight.
For experimentsat different oilseedmoisture contents,the dehulledseedswereheatedat60◦Cfora certaintime untilthe desiredmoisturecontentwasreached.Themoisturecontentofthe samplewasdeterminedusingMethodB-14.Thesampleswere storedinsealedplasticbags.Themoisturecontentbeforeanactual hydraulicpressingexperimentwasexperimentallydeterminedto ensurethatthemoisturecontentwasretainedandnotaffectedby storage.
3.3. Oilcontentmeasurement
Theoilcontentoftherubberseedsusedinthisinvestigationwas determinedusingsoxhletextraction,basedonmethodB-15ofthe Germanstandardmethods(DGF,2002).Theseedsweremanually dehulledanddriedovernightat103◦Cbeforeanalysis.Thedried kernelsweregrindedusingacoffeegrinder.Approximately5gof samplewasweighedwithanaccuracyof0.0001gandtransferred toasoxhletthimble,coveredwithcottonwoolandextractedwith n-hexaneforatleast6h.Thesolventwasevaporatedinarotary evaporator(atmosphericpressure, 69◦C) andthesampleswere subsequentlydriedinanovenat103◦Cuntilconstantweight.The oilcontentisreportedasgramoilpergramsampleonadrybasis.
3.4. Hydraulicpressingexperiments
Aschematicrepresentationofthehydraulicpressusedinthis investigationisshowninFig.1.Thesamplewasplacedonaper- foratedplate(holesof1mmdiameter)inthepressingchamber andcoveredwithastainlesssteelgrid(100mesh).Thetempera- tureinthepressingchamberisadjustableandwasbetween30and 120◦C.Thepressingchamberwasmadefromstainlesssteelwitha diameterof20mmandaheight70mm.Pressuresupto25MPaare possiblebyusingahydraulicplunger.Thepressisequippedwitha thermocouple(±1◦C)and,apressureindicator(±1MPa),aswellas device(VoltcraftVC820)connectedtoacomputerforonlinemon- itoringofthecakethicknessasafunctionoftimeduringpressing.
Approximately7gofsamplewasplacedinthepressingchamber andpreheatedfor5minbeforepressing.Thedefaultpresssetting andthevariationofeachparameterisshowninTable2.
Incaseasolventwasused,thesamplewasaddedtothepressing chamberandtheappropriateamountofsolventwasintroduced beforethepressurewasappliedtothesample.
Fig.1.Schematicrepresentationofthelaboratoryhydraulicpressusedinthisstudy (Subrotoetal.,2015).
Table2
Basecaseandrangeofvariablesforhydraulicpressingofdehulledrubberseeds.
Parameter Basecase Range
Pressure,P(MPa) 20 15–25
Temperature,T(◦C) 35 35–105
Moisturecontent,MC(wt%,w.b.) 2 0–6
Sampleweight(g) 7 –
Solventtoseedratio,SR(%v/w) 0 7–21
Pressingtime(min) 10 –
3.5. Designofexperiments,statisticalanalysisandoptimization
Non-linear multi-variableregression was usedto model the experimentaldataandforthispurposetheDesignExpertVersion 7.0.0softwarepackagewasused.Thedataweremodeledusingthe followingequation:
y=b0+
4i=1
bixi+
4i=1
biix2i +
3i=j
4j=i+1
bijxij+e (5)
Here yis a dependentvariable(oil recovery), xi and xj are the independentvariables(pressure,temperature,moisturecontent, solventamount),bo,bi,biiandbijareregressioncoefficientsofthe modelwhereaseisthemodelerror.
Theregressionequationswereobtainedbybackwardelimina- tionofnon-significantcoefficients.Acoefficientwasconsidered statisticallyrelevant when the Pvalue wasless than 0.05. The optimumconditionsforthesolventassistedhydraulicpressingof dehulledrubberseedsweredeterminedusingthenumericalopti- mizationfunctionprovidedinthesoftwarepackage.
3.6. Dataanalyses
Oilyieldandoilrecoveriesweredeterminedforallexperiments.
TheoilyieldisdefinedastheamountofpressedRSOobtainedfrom acertainamountofdehulledrubberseeds(Eq.(6)).Theoilrecov- eryisdefinedastheactualamountofoilobtainedfromasample dividedbythemaximumamountofoilthatcanbeobtainedfrom acertainamountofdehulledrubberseeds,thelatterdetermined
usingsolventextraction(Eq.(7)).
Yield (wt%,d.b.)= amount ofpressedoil (g)
intake ofrubberseed sample (g)×100%
(6)
Oilrecovery (wt%,d.b.)= amount ofpressedoil (g)
oil contentbysolvent extraction (%d.b.) xintakeof rubberseedsample (g)×100% (7) Allmeasurementswereperformedatleastinduplicateandthe
averagevaluesarereported.
3.7. Analyticalmethods
Thedensityoftheoilswasmeasuredwithapicnometerat10◦C intervalsbetween30and100◦C.Forthispurpose,10mlofasam- plewasplacedinthemeasuringcellandequilibratedtowithin 0.1◦Cofthedesiredtemperature.Reportedvaluesaretheaverage ofduplicatemeasurements.
Theviscosityofthesamplewasdeterminedusingarheometer AR1000-NfromTAinstrument.Acone-and-plateviscometerwas usedwithaconediameterof40mmanda2◦angle.Themeasure- mentwasperformedatdifferenttemperatureswithashearrateof 15s−1.
The fatty acid composition of the oil was analyzed by gas chromatography–massspectrometry (GC–MS) usinga Hewlett- Packard (HP) 5890 series II Plusdevice in combination with a withHPchemstationG1701BAusingtheB0100/NISTlibrarysoft- ware.DetaileddescriptionoftheGCmethodandotheranalytical methods forwater content, acidvalue, flashpoint, cloud point andpourpointareasdescribedelsewhere(Abduhetal.,2013).
Thephosphorus contentanalysis of theRSO wasperformed at ASGAnalytik-ServiceGmbH,Neusass,Germanyaccordingtothe methoddescribedinEN14107.
4. Resultsanddiscussion
4.1. Rubberseedcharacteristics
The experiments werecarried out using fresh rubberseeds obtainedfromBengkulu,Indonesia.Theseedshadanaveragewidth andlengthof2.3–2.5cm and2.8–3.0cmrespectively.Theseeds consistofashell(39wt%d.b.)andakernel (61wt%d.b.).Initial moisturecontentoftheseedsandkernelsasreceivedwere10and
Fig.2.Effectofmoisturecontentoftherubberseedsonoilrecoveryandacidvalue oftheproductoil(20MPa,35◦C,solventfree).
8wt%,w.b.,respectively.Thedehulledseedshadanaverageoilcon- tentof49±0.3wt%,asdeterminedusingastandardizedSoxhlet extractionwithn-hexane.Thisvalueisatthehighendoftheoil contentrange(39–49wt%)reportedintheliterature(Table1).
4.2. Nonsolventassistedhydraulicpressing(NSHP)
Exploratorypressingexperimentsintheabsenceofasolvent wereperformedinalaboratoryhydraulicpressingmachine(Fig.1) togaininsightontheinfluenceofpressingconditionsontheoil recoveryofthedehulledrubberseeds.Pressingconditionsforthe non-solventhydraulicpressingexperiments,particularlythemois- turecontent,temperatureandpressurewerevariedsystematically (Table2).Theresultsofeffectofmoisturecontentandtemperature areshowninFigs.2and3,respectively.Thepressurewithinthe experimentalrange(15–25MPa)didnothaveasignificanteffect ontheoilrecovery(63–66wt%,d.b.,figurenotshown).
Theeffectofmoisturecontentontheoilrecoveryandacidvalue ofthepressedoilaregiveninFig.2.Theoilrecoveryshowsan optimumregardingthemoisturecontentandthehighestrecovery atthesenon-optimisedconditionswasabout55wt%atamoisture contentofabout2–3wt%,w.b.Apossibleexplanationisthatatopti- mumMCvalues,thepressureisdistributedequallyinalldirections, causingmoreoilcellstobedeformedleadingtohigheroilrelease.
Athighermoisturecontents,theliquidphaseabsorbsmostofthe pressureload,resultinginaloweroilrecovery(Sivalaetal.,1991).
Anoverviewof optimum moisturecontents for themechanical pressingofvariousoilseedsisgiveninTable3.Thevaluesrange between2.1and 10wt% forwholeand dehulledseeds. Thus, it canbeconcludedthattheoptimummoisturecontentdependson thenatureoftheoilseedsandthepressingconditions.Ahigher optimummoisturecontentwaspreviouslyreportedfordehulled rubberseeds(10wt%)pressedat70◦Cand8MPa(Ebeweleetal., 2010).Thereportedoilyieldwas28.5wt%,d.b.,slightlyhigherthan theyield(27.4wt%,d.b.)obtainedinthisstudy.
Themoisturecontentalsohasasignificantimpactontheacid value of the product oil, see Fig. 2 for details. The acid value increasesfrom1.0to3.0mg/kgwhengoingfromnegligiblerubber seedmoisturecontentto6wt%,w.b.moisture.Athighermoisture contents,thetriglyceridesarelikelymorepronetohydrolysesto formfreefattyacids(FFA)withaconcomitantincreaseintheacid value.
Thus,itmaybeconcludedthatthemoisturecontentisanimpor- tantprocessvariablethataffectsboththeoilrecoveriesandthe productqualityintermsofacidity.Clearly,theuseofrubberseeds withlowmoisturecontentisfavored.Theeconomicallyoptimum moisturecontentwilldependonthebalancebetweenprocessand
Table3
Optimummoisturecontentformechanicalpressingofdifferentoilseeds.
Oilseed Optimum
moisturecontent (wt%)
References
Sesame(whole) 2.1 Willemsetal.(2008)
Linseed(whole) 4.7 Willemsetal.(2008)
Flax(whole) 2.1 FaborodeandFavier(1996)
Cotton(whole) 5.4 MpagalileandClarke(2005)
Sunflower(whole) 6 DedioandDorrell(1977)
Jatropha(dehulled) 4 FasinaandAjibola(1990)
Rubber(dehulled) 10 Ebeweleetal.(2010)
Fig.3. Effectoftemperatureonoilrecoveryandacidvalue(20MPa,2wt%,w.b.
moisturecontent,solventfree).
productrequirements.Seedswithreducedmoisturecontentwill leadtoproductswithalowFFAcontentandhigheroilrecoveries, thoughthisgoesattheexpenseofhighercostforseeddrying.
TheeffectoftemperatureontheoilrecoveryisgiveninFig.3.
The oil recovery increases at higher temperatures(25–110◦C), thoughseemstoleveloffbetween100and110◦C.Thisincrease islikelyduetotemperatureinducedchangesinthephysicalprop- ertiesofthedehulledseeds.Atelevatedtemperatures,thedehulled seedtissuesaresoftenedandtheviscosityoftheoilwillalsobelow- ered.Asaconsequence,thepermeabilityincreaseswhichenhances theflowofoilthroughthematrix(KhanandHanna,1983;Singh etal.,1984).
Thus,onthebasisoftheNSHPexperiments,theoptimumcon- ditionstoobtainareproducibleoilrecoveryof69±0.4wt%d.b.and ayieldof34±0.2wt%,d.b.wereatapressureof20MPa,seedswith amoisturecontentofabout2wt%andatemperatureof85◦C.The yieldishighercomparedtootherstudiesreportedonthemechani- calpressingofrubberseeds(5.4wt%at27◦C(Morshedetal.,2011) and28.5wt%at70◦C(Ebeweleetal.,2010),seeTable1fordetails).
Thisismostprobablyduetotheuseofacombinationofahigher temperatureandlowermoisturecontentinourstudy.
4.3. Solventassistedhydraulicpressingexperiments(SAHP) Solventassistedhydraulicpressingexperiments(SAHP)were carriedoutusingethanol.Ethanolwaschosenasasolventasitis availablefromrenewableresourcesandposeslesshandlingrisks thann-hexane(Ferreira-Diasetal.,2003).Theethanolamountto (dehulled)seedwasvariedbetween7and21%v/w.Thepressure andmoisturecontentweresetconstantattheoptimumconditions obtainedintheNSHPexperimentsexceptthetemperature,which wassetat75◦C(slightlybelowtheboilingpointofethanol).
Fig.4showstheeffectofsolventtoseedratioontheoilrecovery.
Additionofethanolasasolventhasapositiveeffectontheoilrecov- ery.Whenusing14%v/wofethanolontheseeds,theoilrecovery increasedfrom66intheabsenceofasolventto74wt%,d.binthe presenceofethanol.Sofar,wedonothaveasoundexplanationfor thepositiveeffectofethanolonoilrecoveries.Itiswellpossiblethat thepermeabilityoftheoilisenhancedbyethanolassistedrupture ofcellstructuresinthematrix.Theeffectofthesolventtoseedratio isnotverypronouncedwithintheexperimentalrange(7–21%v/w) andthisis confirmedbysubsequentexperimentsandmodeling activities(videinfra).Apreviousstudyonsolvent-assistedextru-
Fig.4.Effectofsolventtoseedratioonoilrecovery(20MPa,75◦C,2wt%,w.b.
moisturecontent).
sionofsunflowerseedsshowedanincreaseof6wt%inoilrecovery when2-ethylhexanolwasusedasasolvent(Dufaureetal.,1999), closetothevaluefoundinthisstudy(8wt%improvement).Based onthesefindings,a14%v/wratioofethanolondehulledseedswas selectedasthebasecaseforsubsequentmodelingstudies.
4.4. DatamodelingusingtheShiratomodelforNSHPandSAHP 4.4.1. Estimationofmaterialproperties
TheShirato modelwasappliedtomodeltheexperimentally determinedcakeheightsintheformofaconsolidationratioversus thetime.Incombinationwiththeexperimentallydeterminedoil recoveries,italsoallowsdeterminationofamongothersrelevant materialpropertiesofdehulledrubberseeds.Atotalof20exper- imentswereperformedinarangeofpressingconditionsandthe resultsareshowninFig.5andTables4–5.
TheShiratomodelgives agooddescriptionoftheSAHPand NSHPofrubberseeds,seeFig.5fordetails.ThevaluesforB,amea- sureforsecondaryconsolidation(creep)isbetween0.04and0.17, indicatingthatprimaryconsolidationisbyfarmoreimportant.The creepconstantE/Gvariesbetween0.056and0.092.Theexperimen- taldataobtainedisothermallyatdifferentpressures(exp.9–17in Table4)allowscalculationofthevaluesfor0,n,␣0,(section 2.3formoredetails).ThesearegiveninTable5forbothSAHPand NSHPandwillbediscussedinthefollowing.
ThevaluesforBforSAHPandNSHPareequalandindicatethat thecontributionofsecondaryconsolidationtothetotalprocessfor bothSAHPandNSHPiscomparable.Thesameholdsforthecreep constant(E/G),indicatingthatsecondaryconsolidationisnotinflu- encedbythepresenceofethanol.However,thematerialconstants likeporosityandfiltrationresistancedifferconsiderablyforSAHP andNSHP,seeTable5fordetails.Dehulledrubberseedsformsa verydensecakeatallpressuresinvestigatedleadingtorelatively highfiltrationresistances(highvalueof␣o).Thecalculated␣ofor theNSHPmodelisinthesameorderofmagnitudeasthosereported forcellularbiologicalsolids(Schwartzberg,1997).Thevalueof␣o
forSAHPofrubberseedkernelsisoneorderofmagnitudelower thanintheabsenceofasolvent.Possibly,theadditionofethanol increasesthepermeabilityofoilandruptureofcellstructuresinthe matrix(Gandhietal.,2003).Theadditionofsolventalsoreduced thepressuredependencyofthefiltrationresistanceasshownby lowervalueofforSAHPincomparisonwithNSHP.
Table4
Overviewofexperimentsfordehulledrubberseedsatdifferentpressingconditions.a
No. T(◦C) MC(wt%,w.b.) SR(%v/w) P(MPa) Bb(−) E/Gb(−) ˛b(m/kg)
1 35 0 – 20 0.04 0.0056 3.8×1011
2 35 2 – 20 0.07 0.0068 3.6×1011
3 35 4 – 20 0.09 0.0056 4.1×1011
4 35 6 – 20 0.10 0.0051 3.9×1011
5 65 2 – 20 0.10 0.0055 3.5×1011
6 85 2 – 20 0.14 0.0057 3.6×1011
7 105 2 – 20 0.17 0,0092 3.8×1011
8 75 2 – 15 0.08 0.0056 2.5×1011
9 75 2 – 20 0.07 0.0051 3.4×1011
10 75 2 – 25 0.16 0.0075 4.0×1011
11 65 2 14 15 0.09 0.0066 2.2×1010
12 65 2 14 20 0.11 0.0066 2.6×1010
13 65 2 14 25 0.12 0.0061 3.4×1010
14 75 2 14 15 0.12 0.006 3.6×1010
15 75 2 14 20 0.09 0.0052 6.6×1010
16 75 2 14 25 0.12 0.0061 9.7×1010
17 55 2 14 20 0.12 0.0069 1.8×1010
18 75 2 14 20 0.1 0.006 6.4×1010
19 65 2 7 20 0.12 0.0069 2.3×1010
20 65 2 21 20 0.09 0.0074 2.5×1010
aT:temperature,MC:moisturecontent,SR:solventtoseedratio,P:pressure.
b ObtainedbyparameterfittingusingEq.(1).
Table5
MaterialpropertiesestimatedfromtheShiratomodel.
Parameter Dehulledrubberseeda Dehulledrubberseedb Dehulledrubberseedc DehulledJatrophaseedd Wholelinseede
0(−) 0.67±0.004 0.66±0.002 0.67±0.003 0.32 0.56
n(−) 0.02±0.007 0.02±0.001 0.04±0.002 0.09 0.19
R2 0.93 0.96 0.96 0.99 0.97
␣0(m/kg) 9.5×1010 9.9×1010 6.7×1011 6.8×1010 3.7×109
B 1.08±0.1 1.16±0.12 1.44±0.02 0.48 1.55
R2 0.97 0.97 0.97 0.63 0.99
B(−) 0.11±0.02 0.11±0.02 0.11±0.06 0.64±0.08 0.12±0.06
E/G(s−1) 0.006±0.0003 0.006±0.0004 0.006±0.001 0.005–0.006 0.006–0.008
a65◦C,15–25MPa,2wt%,w.b.,14%v/wofsolvent.
b 75◦C,15–25MPa,2wt%,w.b.,14%v/wofsolvent.
c 75◦C,15–25MPa,2wt%w.b.,solventfree.
d 40◦C,20–70MPa,dryseeds,solventfree(Willemsetal.,2008).
e40◦C,10–70MPa,dryseeds,solventfree(Willemsetal.,2008).
FromTable5,itcanbeobservedthatdehulledrubberseedshave alowervalueofn(0.02–0.04)ascomparedtodehulledjatropha (0.09)andwholelinseed(0.19).Thisindicatesthatthepressure dependencyof theporosityis relativelylimited(Willemsetal., 2008).Alowervalueofnimpliesthatdehulledrubberseedisless compressibleascomparedtodehulledjatrophaatthestudiedcon- ditions.Additionofasolventslightlydecreasedtheporosityandits pressuredependence.Therelativelyhighvaluesofincompari- sontodehulledjatrophaseed(0.48)indicatethatdehulledrubber seedscanbeconsideredtobehighlycompressiblematerialatthe conditionsstudied(Willemsetal.,2008).
4.4.2. Effectofoperatingconditionsonconsolidationratiofor NSHP
FrompreliminaryNSHPscreeningexperiments(videsupra),the temperatureandmoisturecontentwereshowntohavethelargest effectsonoilrecoveriescomparedtopressureandsolventtoseed ratio(Figs.2–4).Assuch,thesevariableswerestudiedinmoredetail byperformingadditionalexperiments(Table4)forNSHPandthe resultsweremodeledusingtheShiratomodel.Theexperimental rangeswerebetween0and6wt%forthemoisturecontentand 35–105◦Cforthetemperature(Table2).
Theeffectofmoisturecontentontheconsolidationratioversus timeisgiveninFig.6.Thefinalconsolidationratioisessentially similarfor allMC’sbut thefinal value isachieved ata shorter timeforlowermoisturecontents.Thecontributionofsecondary consolidationincreaseswithanincreaseinmoisturecontentas
illustratedinFig.7.Thecreepconstantandspecificfiltrationresis- tanceareapproximatelyindependentofthemoisturecontent(see Fig.1,Supplementarydata).Thesetrendsareinagreementwiththe resultsreportedforthehydraulicpressingofsesameseed(Willems etal.,2008).
Theeffectofthetemperatureontheconsolidationratioversus timeisgiveninFig.7.Highertemperaturesintherange35–85◦C resultsinamorerapiddecreaseinthefiltercakethickness.These findingsmaybeexplainedbyconsideringthattheelasticityofthe solidmatrixincreasesathighertemperaturesandbecomeshighly compressible(Venteretal.,2007;Willemsetal.,2008;Subroto etal.,2015).However,afurtherincreasefrom85to105◦Cdoes notleadtoanincreaseintherateofexpression.Thus,onthebasis oftheexperimentaldataandsupportedbytheShiratomodel,we canconcludethattherateofexpressionforNSHPincreaseswith(i) pressingtemperaturetillamaximumat85◦Cand(ii)whenusing rubberseedswithalowmoisturecontent.
4.5. Empiricalmodelingofoilrecoveriesusingdesignof experimentsforSAHP
Togainfurtherdetailedinsightsintheeffectsofprocessvari- ables on oilrecovery for SAHP, a new set of experiments was performedandthedataweremodeledusingmulti-variablenon- linear regression. The pressing conditions and particularly the moisturecontent,temperature,pressureandsolventtoseedratio werevariedsystematicallyusingafour-factorfacecenteredCen-
Fig.5.ConsolidationratioversustimeandparityplotfortypicalSAHPandNSHP experimentsusingdehulledrubberseeds(SAHP:20MPa,75◦C,2wt%,w.b.moisture content,14%v/w,NSHP:20MPa,75◦C,2wt%,w.b.moisturecontent,solventfree).
Table6
LevelandrangeofvariablesfortheCCDforSAHP.
Factors Levels
−1 0 1
Pressure,P(MPa) 15 20 25
Temperature,T(◦C) 55 65 75
Moisturecontent,MC(wt%,w.b.) 1 2 3
Solventtoseedratio,SR(%v/w) 7 14 21
tralCompositeDesign(CCD,Table6)andatotalof30experiments wasperformed.
Theexperimentaloilrecoverieswerebetween53.3–73.8wt%, d.b(Table7),indicatingthatthepressingvariableshavea large impactontheoilrecovery.Theeffectofpressingconditionsonthe oilrecoverywasmodeledandthemodelcoefficientsaregivenin Table8.TheANOVAdataareprovidedinTable9andrevealthat themodeldescribestheexperimentaldataverywell(lowp-value, highR-squaredvalues).Thisisalsoillustratedbyaparityplotshow- ingtheexperimentalandmodeledoilrecovery(Fig.8).Thesolvent toseedratio(SA)wasnotstatisticallyrelevant(p>0.05)andwas
Fig.6.Consolidationratioversustimeatdifferentmoisturecontents(35◦C,20MPa, solventfree).
Fig.7.Consolidationratioversustimeatdifferenttemperatures(20MPa,2wt%, w.b.,solventfree).
excludedfromthemodel.Avisualisationoftheeffectofprocess variablesontheoilrecoveryisgiveninFig.9.Highertemperatures haveapositiveeffectontheoilrecovery.Anoptimuminoilrecov- eryforbothmoisturecontentandpressurewasobserved,theexact vauebeingafunctionoftheotherprocessvariables.
TheDesignExpertsoftwareallowscalculationoftheoptimum conditionstoattainthehighestoilrecoveryfortheSAHPinthe experimentalwindow.Anumberofoptima(5)withoilrecoveries ofabout75%werecalculated,allatapressureof20MPA,atemper- atureof75◦C,andamoisturecontentbetween1.3and1.9wt%.An experimentwasperformedatoneoftheseoptima(moisturecon- tent1.6wt%,solventtoseedartioof14%v/w)toverifythemodel predictions.Good agreementbetweenexperimental(75.7%)and modeledoilrecovery(75.4%)wasobserved.
4.6. CompositionandrelevantproductpropertiesofRSO
Thefattyacidcompositionofthepressedoilobtainedatopti- mumpressingconditionsforSAHP(20MPa,1.6wt%,w.b.,75◦C,
Table7
Experimentalconditionsandthepercentageofoilrecovery(wt%,d.b.).a
No P(MPa) T(◦C) MC(wt%,w.b.) SR(%v/w) Oilrecovery(wt%,d.b.) Actual Predicted
1 25 55 3 7 54.7 53.7
2 25 65 2 14 69.0 69.1
3 15 65 2 14 67.5 68.2
4 15 75 3 7 58.9 57.1
5 15 55 3 21 55.4 52.7
6 15 75 3 21 56.3 57.1
7 15 55 3 7 50.0 52.7
8 15 75 1 21 70.9 71.4
9 25 55 1 7 64.7 63.9
10 20 65 2 7 68.7 70.5
11 25 55 3 21 53.3 53.7
12 15 55 1 21 64.6 62.9
13 20 65 2 21 72.6 70.5
14 25 75 3 7 57.5 58.1
15 25 75 3 21 57.9 58.1
16 20 65 2 14 71.0 70.5
17 25 75 1 7 72.8 72.3
18 20 65 2 14 73.0 70.5
19 20 65 2 14 69.0 70.5
20 20 65 2 14 69.0 70.5
21 20 65 2 14 70.0 70.5
22 20 65 3 14 56.4 57.2
23 20 65 1 14 69.5 69.5
24 15 75 1 7 70.4 71.4
25 15 55 1 7 62.4 62.9
26 25 55 1 21 62.0 63.9
27 25 75 1 21 73.2 72.3
28 20 65 2 14 72.0 70.5
29 20 75 2 14 73.8 73.7
30 20 55 2 14 66.9 67.3
aP:pressure,T:temperature,MC:moisturecontent,SR:solventtoseedratio.
Table8
CoefficientsfortheempiricalmodelofoilrecoveryforSAHP(wt%,d.b.).
Variable Coefficient
Constant −11.61
P 3.06
T 0.53
MC 29.1
T.MC −0.1
P2 −0.074
MC2 −7.15
P:pressure(MPa),T:temperature(◦C),MC:moisturecontent(wt%,w.b.).
Table9
AnalysisofvariancefortheSAHPofdehulledrubberseeds.
SS DF MS F p-value R2values
Model 1410 6 235 104 <0.0001 R2 0.96
Error 52 23 4.7 R2adjusted 0.96
Total 1462 29 R2predicted 0.94
14%v/w)wasdetermined (GC) and shown to consistmainly of palmiticacid(12.2%),stearicacid(7.3%),oleicacid(28.1%),linoleic acid(38.2%) andlinolenicacid(14.2%). Themeasuredfattyacid compositionisinthesamerangeasreportedbyRamadhasetal.
(2005)viz.;10.2%palmiticacid,8.7%stearicacid,24.6%oleicacid, 39.6%linoleicacidand16.3%linolenicacid.
RelevantproductpropertiesofthepressedRSOafterquantita- tiveethanolremoval(GC)areshowninTable10.Theacidvalue oftheoilis(2.3)relativelylowcomparedtheacidvaluereported for RSO in the literature (2–38mg KOH/g, Table1).A possible explanationforthelowvalueisthattheseedsusedinthisstudy werefreshlyobtainedfromtheplantationanddirectlydriedtoa MCbelow7wt%beforestorage(Ebeweleetal.,2010).Theflash point (290◦C) is within therange as reported in theliterature
Fig.8.ParityplotfortheempiricalmodelforSAHP.
Fig.9. Responsesurfaceshowingtheinteractionbetweentwoparametersonoil recovery(a)temperatureandpressure(2wt%,w.b.,14%v/w)(b)moisturecontent andtemperature(20MPa,14%v/w).
Table10
PropertiesofpressedRSOatoptimumconditions(20MPa,1.6wt%,w.b.,75◦C, 14%v/wethanol/seedratio).
Property RSO
Acidvalue(mgKOH/g) 2.3
Watercontent(mg/kg) 300
Pcontent(mg/kg) 57.7
Flashpoint(◦C) 290
Pourpoint(◦C) −4
Cloudpoint(◦C) 0
Fig.10.ColdflowpropertiesofRSOandotherplantoils.
(198–294◦C)(Ramadhasetal.,2005;Njokuetal.,1996),aswell asthecloudpoint(0◦Cversus−1–0◦Cinliterature).Dataforthe pourpointofRSOarenotavailableintheliterature.Thepourpoint forRSOisclosetothereportedvalueforRSOmethylester(−5to
−8◦C)(Ramadhasetal.,2005;Njokuetal.,1996).Thecloudpoint andpourpointofRSOin comparisontopalmoil(Crabbeetal., 2001),groundnutoil(Gunstoneetal.,2007)andrapeseedoil(Balat, 2007)arepresentedinFig.10.Thepourpoint,whichisafunction ofthedegreeofunsaturationofthefattyacidchainsandtypically decreaseswithhigherunsaturationlevel,isintheexpectedrange forplantoils(Mingetal.,2005).
The P content (58ppm) is higher than the threshold limit (3mg/kg)setbythe pureplant oilqualitystandard DIN51605 (www.eia.gov,2014).Forbiodieselsynthesis,aphosphorouscon- tentabove50ppmmayreducetheyield by3–5%(VanGerpen, 2005).Thus,apurificationandparticularlyadegummingprocedure willberequiredbeforetheRSOcanbeusedforefficientbiodiesel synthesis.
Thetemperaturedependenceofthedensityandviscosityofthe pressedrubberseedoilarerequiredinputfortheShiratomodel(Eq.
(1)).Bothpropertiesweremeasuredatarangeoftemperatures(see Fig.2Supplementarydata)andfittedusingEqs.(8)and(9):
=1−0T (8)
=0 exp
−1/(RT)
(9) Goodfitswereobtained(R2of0.99)withvaluesforoand1
of6.94×10−4g/(cm3.◦C)and0.948g/cm3,respectivelyandoand
1valuesof4.59×10−6Pa.sand23.1×103J/mol,respectively.
5. Conclusionsandoutlook
Systematic experiments on rubberseed oil expression have beenperformedbothintheabsence(NSHP)andpresenceofethanol (SAHP).Intheabsenceofasolvent(NSHP),thehighestoilrecov- eries(69wt%)wereobtainedat2wt%moisturecontent,20MPa, 85◦C and10minpressingtime.A7%improvementinoilrecov- erywaspossiblebyexpressioninthepresenceofethanol(SAHP) at1.6wt%moisturecontent,14%v/wethanol,20MPa,75◦C and 10minpressingtime.
Theexperimentaldatasetwasmodeledusingtwoapproaches, viz.(i)afundamentaldynamicmodelknownastheShiratomodel fortheconsolidationratioversusthetimeprofiles(NSHPandSAHP) and(ii)anempiricalmodelforoilrecoveriesusingmulti-variable non-linearregression(forSAHP).Bothmodelsgaveagooddescrip-
tionoftheexperimentaldata.ParameterestimationfortheShirato modelindicatesthatthedehulledrubberseedsarerelativelyhard materialsasindicatedbythelowvalueofn(0.02–0.04)ascom- paredtodehulledjatrophaseeds(0.09)andwholelinseed(0.19).
Inaddition,wecanconcludethattherateofexpressionforNSHP increaseswith(i)pressingtemperaturetillamaximumat85◦Cand (ii)whenusingrubberseedswithalowmoisturecontent.Thenon- linearregressionmodelfortheoilrecoveryusingSAHPsuggests thatthemoisturecontentofthedehulledseedsandtemperature havethelargesteffectontheoilrecoveryfollowedbypressureand solventtoseedratio.Atoptimumconditions,a reproducibleoil recoveryof76andanoilyieldof37wt%,d.b.wereobtained.
RelevantpropertiesoftheRSOweredeterminedandindicate thattheRSOobtainedinthisstudycanbeusedasafeedstockfor biodieselproduction,providedthatthePcontentisreducede.g.by degumming.TheuseofSAHPwithethanolmayhaveadvantages whenaimingfortheproductionoffattyacidethylesters(FAEE).
IntegrationofaninitialSAHPoftheseedsfollowedbysubsequent ethanolysisoftheRSOproducedisanattractiveprocessoptionas it(i)leadstohigheroverallbiodieselyieldsduetoimprovedRSO recoveriesinthefirststepwhenusingethanolassistedoilexpres- sionand(ii)eliminatestheuseofethanolseparationfromtheRSO aftertheoilexpressionby e.g.distillation,which isenergy and capitalintensive.
Acknowledgement
Theauthors thankNWO/WOTROfor a researchgrant in the frameworkoftheAgriculturebeyondFoodprogram.
AppendixA. Supplementarydata
Supplementarydataassociatedwiththisarticlecanbefound,in theonlineversion,athttp://dx.doi.org/10.1016/j.indcrop.2016.07.
025.
References
Abduh,M.Y.,VanUlden,W.,Kalpoe,V.,VandeBovenkamp,H.H.,Manurung,R., Heeres,H.J.,2013.BiodieselsynthesisfromJatrophacurcasL.oilandethanolin acontinuouscentrifugalcontactorseparator.Eur.J.LipidSci.Technol115, 123–131.
Abdullah,B.,Salimon,J.,2009.PhysicochemicalcharacteristicsofMalaysianrubber (HeveaBrasiliensis)seedoil.Eur.J.Sci.Res.31,437–445.
Abraham,G.,Sr,R.H.,Kuk,M.,Wan,P.,1993.Wateraccumulationinthealcohol extractionofcottonseed.J.Am.OilChem.Soc.70,207–208.
Aigbodion,A.,Pillai,C.,2000.Preparation,analysisandapplicationsofrubberseed oilanditsderivativesinsurfacecoatings.Prog.Org.Coat.38,187–192.
Balat,M.,2007.Productionofbiodieselfromvegetableoils:asurvey.Energy SourcesPartA29,895–913.
Crabbe,E.,Nolasco-Hipolito,C.,Kobayashi,G.,Sonomoto,K.,Ishizaki,A.,2001.
Biodieselproductionfromcrudepalmoilandevaluationofbutanolextraction andfuelproperties.ProcessBiochem.37,65–71.
Dedio,W.,Dorrell,D.,1977.Factorsaffectingthepressureextractionofoilfrom flaxseed.J.Am.OilChem.Soc.54,313–315.
Dufaure,C.,Mouloungui,Z.,Rigal,L.,1999.Atwin-screwextruderforoil extraction:II.Alcoholextractionofoleicsunflowerseeds.J.Am.OilChem.Soc.
76,1081–1086.
Ebewele,R.,Iyayi,A.,Hymore,F.,2010.Considerationsoftheextractionprocess andpotentialtechnicalapplicationsofNigerianrubberseedoil.Int.J.Phys.Sci.
5,826–831.
Faborode,M.,Favier,J.,1996.Identificationandsignificanceoftheoil-pointin seed-oilexpression.J.Agric.Eng.Res.65,335–345.
Fasina,O.,Ajibola,O.,1990.Developmentofequationsfortheyieldofoilexpressed fromconophornut.J.Agric.Eng.Res.46,45–53.
Ferreira-Dias,S.,Valente,D.G.,Abreu,J.M.,2003.Comparisonbetweenethanoland hexaneforoilextractionfromQuercussuberL.fruits.GrasasAceites54, 378–383.
Gandhi,A.,Joshi,K.,Jha,K.,Parihar,V.,Srivastav,D.,Raghunadh,P.,Kawalkar,J., Jain,S.,Tripathi,R.,2003.Studiesonalternativesolventsfortheextractionof oii-Isoybean.Int.J.FoodSci.Technol.38,369–375.
Gunstone,F.D.,Hardwood,J.L.,Dijkstra,A.J.,2007.TheLipidHandbook.CRCPress, BocaRaton.