ContentslistsavailableatScienceDirect
Industrial Crops and Products
j ourna l h o m e pa 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
Optimization of mechanical oil extraction from Jatropha curcas L.
kernel using response surface method
Erna Subroto
a, Robert Manurung
b, Hero Jan Heeres
a, Antonius Augustinus Broekhuis
a,∗aDepartmentofChemicalEngineering,RijksuniversiteitGroningen,Nijenborgh4,9747AGGroningen,TheNetherlands
bSchoolofLifeSciencesandTechnology,InstitutTeknologiBandung,Jl.Ganesha10,Bandung,Indonesia
a r t i c l e i n f o
Articlehistory:
Received28January2014
Receivedinrevisedform13August2014 Accepted14August2014
Availableonline14November2014
Keywords:
JatrophacurcasL.
Oilextraction Hydraulicpress
Face-centeredcompositedesign Responsesurfacemethodology Processoptimization
a b s t r a c t
ExtractionofoilfromJatrophacurcasL.kernelwasinvestigatedusingalab-scalehydraulicpress.Aface centeredcompositedesignofexperimentswasemployedtostudyandoptimizetheeffectofapplied pressure,pressingtemperatureandmoisturecontentonoilrecovery.Aquadraticpolynomialmodelwas generatedtopredictoilrecoveryandwasfoundtocover98%oftherangeforthefactorsstudied,namely 10–20MPaappliedpressure,60–90◦Cpressingtemperatureand3–5%(w.b.)moisturecontent.Among theprocessparametersstudied,pressingtemperaturehadthemostsignificanteffectontherecovery followedbyappliedpressureandquadraticofmoisturecontent.Modelvalidationexperimentsshow goodcorrespondencebetweenactualandpredictedvalues.Theoptimalextractionconditionforoilyield withintheexperimentalrangeofthevariablesresearchedwasat19MPaappliedpressure,90◦Cpressing temperature,and3.8%(w.b.)moisturecontent.Atthiscondition,theyieldofoilwaspredictedtobe87.8%.
©2014ElsevierB.V.Allrightsreserved.
1. Introduction
Oneofthemostpromisingrenewableandindependentenergy sourcesinruralareasisJatrophaoil(KumarandSharma, 2008;
Makkar and Becker, 2009).It is non-edible oil, thus it will not impairfoodsecurityissues(Pinzietal.,2009).Asitgrowswellon drymarginalnon-agriculturalland,itwillnotcompetewithland neededforfoodproductionorwithnatureconservation(Achten etal.,2007;MakkarandBecker,2009;Pinzietal.,2009).Jatropha isconsideredamoresustainablefeedstockforenergyproduction thananyotherfood-relatedcropsuchaspalm,rapeseed,soybean orsunflower(Achtenetal.,2007;Pinzietal.,2009).
Theextractionoftheoilfromtheseedisdoneindifferentways.
Methodsusedare:solventextraction,mechanicalextraction,enzy- maticextractionandaqueousextraction.Forapplicationinrural areas,mechanicalextractionisconsideredtobethebestoption.
Inthisextractionhydraulicpressesareusedtoremoveoilfrom theseeds.Thismethodisgenerallypreferredbecauseofitslower initialandoperationalcost,andbecauseitcanbeeasilyoperated bysemi-skilledpersonnel.Itproducesrelativelygoodqualityoilas comparedtothesolventextractionprocessanditallowsforthe
∗Correspondingauthor.Tel.:+31503634918;fax:+31503634479.
E-mailaddress:[email protected](A.A.Broekhuis).
useofthecakeresidue(Olajideetal.,2007).However,adisadvan- tageofmechanicalextractionistheloweroilrecoverycompared tosolventextraction.Ithasbeenreportedthatsolventextraction withn-hexanecouldachieveabout70–99%oilrecovery,againsta reportedmaximumof60–80%formechanicalextraction(Achten etal.,2007).
Appliedpressure,pressingtemperature,andpressingtimeare importantprocessparameters,whiletheadjustmentofseedmois- turecontentisshowntobethemostimportantfactoramongst pretreatmentssuchasremovalofhullsorshells,sizereduction orheattreatment.Willemsetal.(2008)reportedhigheroilyield forrapeseed,sesame,linseed,jatrophaseedandjatrophakernel pressedathigherpressuresand/ortemperature.Healsoreported the22%differenceinoilrecoverywhenpressedlinseedatvarious moisturecontentsvariedfrom0to10%.Ourpreviousstudyindeed showsthatappliedpressure,pressingtemperature,andmoisture contentareimportantparametersthatinfluenceoilrecovery.The rateofpressureisfoundtobeoptimumat0.125MPa/s(Subroto etal.,2014).Thisstudyindicatedthattheoptimumoilrecovery iswithintherangeof10–20MPa,60–90◦Cand3–5%(w.b.).This impliesthatmaximizingtheoilrecoveryislimitedtotheoptimiza- tionoftheseprocessparameters.Thisresearchisaimedtostudy andmodeltheeffectofthesevariablesandtheirinteractiononthe percentageofoilextraction.Themodelwillbeusedtooptimizethe extraction,andtheaccuracyofthemodelwillbetested.
http://dx.doi.org/10.1016/j.indcrop.2014.08.050 0926-6690/©2014ElsevierB.V.Allrightsreserved.
Table1
PropertiesofJatrophasamplesbeforemoistureconditioning.
Properties JatrophaKalimantan JatrophaSubang Weight(%d.b.)
Seed 100 100
Kernel 63.0 63.4
Shell 37.0 36.6
Oilcontent(%d.b.)
Seed 36.8±0.05 35.1±0.06
Kernel 58.3±0.02 55.3±0.01
Shell 0.1±0.01 0.1±0.01
Moisturecontent(%w.b.)
Seed 8.6±0.18 8.5±0.10
Kernel 7.04±0.13 6.97±0.17
Shell 11.3±0.24 11.1±0.21
2. Materialsandmethods 2.1. Material
Jatropha seeds used in the optimization experiment were obtainedfromPalangkaraya,Central Kalimantan,Indonesia.The maturefruitswereharvestedmanuallyinMarch2011.Theseeds weredriedundersunandstoredinjutebagsinawarehousefacil- ityattemperaturesbetween20and30◦Candrelativehumidityof 80–90%foronemonth.InadditiontoJatrophafromKalimantan,Jat- rophafromSubangwasusedforoilqualityanalysis.Jatrophaseed fromSubangwasharvestedmanuallyduringJanuary2011,dried undersunandstoredinjutebagsinawarehousefacilityattem- peraturesbetween20and30◦Candrelativehumidityof70–80%
for3months.AftertransporttotheNetherlandsinApril2011,both seedswerestoredatroomtemperature(withinarangeof18–22◦C) andrelativehumidityof40–50%.Theseedswerede-shelledman- uallyand boththekernelsandshellswereanalyzedforweight fraction,initialmoistureandtotaloilcontent(seeTable1).Theker- nelswereexposedtomoistureconditioningpretreatmentbefore beingpressed(describedbelow).Thepretreatedkernelswereused directlyinthepressingexperimentstoreducetheinfluenceofstor- agetimeonoilquality.Theoilanalyseswereconducteddirectly afterpressinginMay2011forbothsourcesofJatrophaseeds.
Potassiumhydroxide(pellets,85%,Vetec),oxalicacidanhydrous (≥99%,Sigma–Aldrich),ethanol(95%,Sigma–Aldrich),diethylether (≥99%, Sigma–Aldrich), hexane (≥99, Sigma–Aldrich), Hydranal solvent(Fluka)andHydranaltitrant5(Fluka)wereboughtfrom Sigma–Aldrich(Amsterdam,TheNetherlands).
Foroilrecoverymeasurement,thekernelswereconditionedby ovendrying.Thedryingtemperatureswere35,40and50◦Cfor desiredmoisturecontentof5,4and3%w.b.,respectively.Afterdry- ing,thekernelwaswrappedtightlyinalowdensitypolyethylene bagof25mthicknessandthenputinsideadesiccatorcontain- ingsilicagelforaminimumof1daybeforebeingpressed.Foroil qualityanalysis,thekernelswerestoredinsidethedesiccatorcon- tainingsilicageluntilthedesiredmoisturecontentwasreached, andthenwrappedinthepolyethylenebagforequilibration.
Theinitialmoisturecontentofthesampleswasdeterminedby ovendryingof10gofsampleat105◦Cfor24h.Duplicatemea- surementswereperformedfor eachsample andaveragevalues weretaken.Moisturecontentafterconditioningwasdetermined bycalculatingtheweightdifferenceofthesampleafterandbefore conditioning.
2.2. Hydraulicpressing
Aschematicrepresentationofthehydraulicpressisshownin Fig.1.Thepressingchamberwasmadefromstainlesssteelwith adiameterof20mmandaheightof70mm.Itisequippedwitha
Table2
Actualandcodedlevelsoftheindependentvariablesintheexperimentaldesign.
Independentvariable Symbol Level
Actual Coded Actual Coded
Appliedpressure (MPa)
P x1 10 −1
15 0
20 1
Heating temperature(◦C)
T x2 60 −1
75 0
90 1
Moisturecontent (%w.b.)
M x3 3 −1
4 0
5 1
perforatedplate(diameterof1mm)coveredwithfinewiremesh (100mesh).Thiswasplacedatthebottomofthepressingchamber actingasfilterduringextraction.Anelectrical-resistanceheating ringattachedaroundthepressingchamberisusedtopreheatthe pressing chamberduringoperationwithina temperaturerange of60–90◦C.Pressuresupto20MPawereappliedbyahydraulic plunger. Thepressis completedwithathermocouple(±2.5◦C), pressuremeasurement(±1MPa),andalevelindicator(±0.01mm), whichmeasuresthedistancetheplungertraveled.
Approximately7gofkernelswasplacedinthepressingcham- ber.Afterwards,theplungerisputontopofthekernels.Thesample ispreheatedfor5minwithoutapplyingmechanicalpressure.Sub- sequently, themechanical pressure wasincreased linearlyat a pressingrateof0.125MPa/suntilthedesiredpressureisreached.
Totalpressingtimewas10min.Forvalidationexperiment,three replicatemeasurementswereperformedforeachsampleandaver- agevaluesweretaken.
2.3. Statisticalanalysis
Levelsfortheindependentvariables,i.e.appliedpressure,X1, pressingtemperature,X2,andmoisturecontent,X3,werebasedon resultsobtainedinapreviousstudy(Subrotoetal.,2014).Athree- factor-three-levelfacecenteredcentralcompositedesign(CCRD) wasappliedwherethevaluesoftheindependentvariablesXwere codedasthevariables,xintherangeof−1and+1level(shown inTable2).Themathematicaltransformationofanyactuallevelof appliedpressure,temperature,andmoisturecontentintothecoded levelcanbeobtained,respectively,fromthefollowingequations:
x=(X−XM) XD
(1)
XM= (Xmax+Xmin)
2 (2)
XD=(Xmax−XM) (3)
x1=(X1−15)
5 , x2=(X2−75)
15 , x3=X3−4 (4)
whereXM,XD,Xmax,andXministhemeanvalue,intervalofvariation, maximumandminimumvalueofX,respectively.While,x1,x2and x3arethecodedvaluesandX1,X2andX3aretheactualvaluesfor appliedpressure,temperatureandmoisturecontent.
Theexperimentalplanwasdesignedandtheresultsobtained wereanalyzedusingDesignExpertversion8.0.0software(State- EaseInc.,StatisticsMadeEasy,Minneapolis,MN,USA)tobuildand evaluatemodelsandtoplotthethree-dimensionalresponsesur- facecurves.Theexperimentaldatawereanalyzedfortheresponse.
Twentyexperimentswereperformedwhichconsistedofeight factorialpoints,sixextrapoints(starpoints)andsixreplicatesfor thecenterpoint.Thesixreplicatesforthecenterpointwereusedto estimatetheexperimentalerror.Ananalysisofvariance(ANOVA)
Fig.1.Schematicrepresentationofthehydraulicpress.
andR2(coefficientofdetermination)statisticwereusedtocheck theadequacyofthedevelopedmodel.UsinganF-test,itwaspossi- bletotestthevariationofthedataaroundthefittedmodel(lackof fit).Thesignificancelevelwasstatedat95%,withp-value0.05.Con- firmatoryexperimentswerecarriedouttovalidatetheequations usingthecombinationsofindependentvariables.Thesewerenot partoftheoriginalexperimentaldesign,butwithintheexperimen- talregion.TheoptimalconditionsfortheJatrophaoilrecoverywere obtainedusingthesoftware’snumericaloptimizationfunction.
2.4. Totaloilcontentandoilrecovery
Totaloilcontentofthekernelswasdeterminedbytheweight ofsubstancesthatareextractedbyn-hexaneaccordingtostan- dardSoxhletextractionmethod.Thekernelsweredriedovernight inanovenat105◦C.Thedriedkernelsweregrindedusingacof- feegrinder(Princess242195,PrincessHouseholdAppliancesB.V., TheNetherlands)andapproximately10gweretransferredintoa celluloseextractionthimble(Whatman,I.D.×H18×55mm,GE).
Thiswasextractedwithhexaneatitsboilingpointfor24h.The hexane–oilmixturewasevaporatedinarotaryvacuumevaporator whilethecakewasdriedintheovenat80◦Covernight.Duplicate measurementswereperformedandaveragevaluesweretaken.The totaloilcontentwasdefinedasweightofoilextractedoverthedry weightofthesampletaken.Theoilrecoverywasdefinedasthe ratiooftheamountofoilexpressedduringmechanicalpressingto thetotaloilcontentofthesampleatdryweightbasis.
2.5. Oilqualityanalysis
Hydraulicpressedoilswhichstillcontainedsomefineparticles werecentrifugedat2000×gfor10min.Theclearoilswererecov- eredandusedforoilanalysis.Chemicalandphysicalanalysesofthe sampleswerecarriedoutaccordingtothestandardtestmethods:
DINEN14111,DINEN14104,DINENISO12937,DINEN14112, DINEN14107andDINEN14538foriodinevalue,acidvalue,water content,oxidativestability,phosphoruscontent,andcalciumand magnesium(Ca+Mg)content,respectively.Densityanalysiswas
carriedoutbymeasuringthemassofthesampleswithrespect totheirvolumeusinga 10mLpycnometer.Thedensityat15◦C wasobtainedbyextrapolationofdatacollectedfrom30to100◦C, in10◦Cincrements.Dynamicviscosityanalysiswascarriedout byusingacone-and-plateviscometerAR1000-N(40mm2◦alu- minumcone)at40◦Cwithashearrateof40/sfor10min.Kinematic viscositywascalculatedfromthecorrespondingdynamicviscos- ityanddensityatcorrespondingtemperature.Theflashpointof thesampleswasmeasuredaccordingtothemethodsdescribedin ASTMD6450usingaMINIFLASHFLP/H/L.Cloudpoint(CP)andpour point(PP)weremeasuredusingaTanakaScientificLimitedType MPC-102LaccordingtomethodsdescribedinASTMD6749and ASTMD2500,respectively.
AccordingtotheGermanfuelstandard DIN51605: 2010-10 forpureplantoil,theacidvalue,phosphoruscontent,andwater contentshouldnotexceed2mgKOH/goil, 3ppm,and750ppm respectively,andoxidative stabilityshouldbeatleast6h.Most chemicalpropertyanalysesontheplantoilswereconductedin ourlaboratorywiththeexceptionofphosphoruscontent,which wasconductedbyASGAnalytik-ServiceGmBH,Germany.Dupli- catemeasurementswereperformedoneachsampleandaverage valuesweretaken.
3. Resultsanddiscussions
3.1. Fittingthemodelandanalysisofvariance(ANOVA)
In order todetermine theoptimum process parameters, i.e.
appliedpressure,pressingtemperatureandmoisturecontentfor maximumoilrecovery,theexperimentsweredesigned accord- ingtoafacecenteredcentralcompositedesigninthreevariables followingtheResponseSurfaceMethodology(RSM). Theexper- imental set-up and corresponding experimental responses are shown inTable3.Theoilrecovery fromthesamplesexhibited anincreasewithincreasingtemperatureandpressure.Thelow- estvalueof68.9%wasobtainedat10MPapressure,60◦Cand3%
(w.b.)moisture;thehighestvalueof86.4%wasobtainedat20MPa, 90◦Cand3%(w.b.)moisture.
Table3
Experimentallayoutoffacecenteredcentralcompositedesignanditscorrespondingobservedvaluesofoilrecovery.
Run Variableproperties
Pressure(MPa) Temperature(◦C) Moisturecontent(%w.b.) Oilrecovery(%d.b.)
1 10 60 3 69.6
2 20 60 3 78.6
3 10 90 3 79.1
4 20 90 3 86.4
5 10 60 5 73.7
6 20 60 5 78.6
7 10 90 5 79.0
8 20 90 5 82.3
9 10 75 4 78.2
10 20 75 4 84.6
11 15 60 4 79.8
12 15 90 4 86.1
13 15 75 3 80.5
14 15 75 5 79.3
15 15 75 4 83.3
16 15 75 4 84.2
17 15 75 4 83.2
18 15 75 4 83.7
19 15 75 4 84.7
20 15 75 4 82.6
Fitting of the data to various models (linear, two factorial, quadratic and cubic) and their subsequentanalysis of variance showsthatthehydraulicpressingofJatrophakernelismostprop- erlydescribedwithaquadraticpolynomialmodel.TheAdjusted R2ofthequadraticmodel(0.9752)washigherthanthatoflinear (0.5431)andtwofactorial(0.5186)models.Thecubicmodelwas foundtobealiased.Thesecond-orderpolynomialmodelsusedto expresstheoilrecovery(Y)asafunctionofindependentvariables (Eqs.(5)and(6))areshownbelow(intermsofcodedandactual levels):
Oilrecovery(coded)=83.41+3.09x1+3.26x2+0.13x3
−0.41x1x2−1.01x1x3−1.04x2x3
−1.71x21−0.16x22−3.21x23 (5)
OilRecovery(actual)=51.43+3.89X1+0.68X2+33.77X3
−0.005X1X2−0.20X1X3−0.07X2X3
−0.06X12−0.001X22−3.21X32 (6)
Table4summarizestheANOVA(F-test)andp-valuethatare usedtoestimatethecoefficientsofthemodel,tocheckthesignif- icanceofeachparameter,andtoindicatetheinteractionstrength ofeachparameter.ItwasobservedfromtheANOVAanalysisthat theconfidencelevelwasgreaterthan95%whilethep-valueofthe modelwaslessthan0.0001.Themodelwiththep-valuebelow0.05 wasstatisticallysignificant,whichimpliedthatthemodelwassuit- ableforthisexperiment.Meanwhile,the“lackoffit”ofthismodel wasinsignificantwiththep-valuebeing0.76.Themaineffects,i.e.
X1 andX2,andinterationeffects,i.e.X12,X32,X1X3,X2X3,aresig- nificantbasedonthecalculatedp-values.Theeffectofmoisture contentonoilrecoveryexhibitedap-valueof0.54.Thisexceedsa p-valuelevelof0.05andindicatesthattheeffectisnotsignificant.
Thecoefficientofdetermination(R2)andadjustedcoefficient ofdetermination(R2adj)were0.987and0.975,respectivelywhich indicatedthattheestimatedmodelfitstheexperimentaldatasat- isfactorily.Leeetal.(2010)suggestedthatforagoodfitofamodel, R2shouldbeatleast0.80.TheR2fortheseresponsevariableswas higherthan0.80,indicatingthattheregressionmodelsexplained themechanismwell.
Fig.2ashowstheexperimentalversuspredictedoilrecovery obtainedfromEq.(6).Alineardistributionisobservedwhichis indicativeofawell-fittingmodel.ThevaluespredictedfromEq.(6) wereclosetotheobservedvaluesofoilrecoveryfromJatropha kernel. Thenormalprobability plotisalsopresented inFig.2b.
Theplotindicatesthat theresiduals(difference betweenactual andpredictedvalues)follow anormaldistributionand forman approximatelystraightline.
3.2. Modelvalidation
The adequacy of the model equations to predict optimum responsevalueswastestedusingtheconditionsshowninTable5.
Theprocessing conditionsfor maximumrecoverywereused to experimentallyvalidateandpredictthevalues oftheresponses usingthemodelequation.Closeagreementexistsbetweenvalues calculatedusingthemodelequationandtheexperimentalvalues oftheresponsevariablesatthepointofinterest.TheX2goodness- of-fittestwasusedtoexaminethevalidityofthemodel(Table5) (MooneyandSwift,1999).Thetestshowsthatthereisnotasignif- icantdifferencebetweenthepredictedandactualvaluessincethe X2value(0.02)ismuchsmallerthanthecut-offvalueofX2for95%
confidencelevelfor3degreesoffreedom(7.81).Thisindicatesthat thegeneratedmodelisvalidat95%confidencelevel.
Fromtheestablishedequation,themaximumoilrecoverypre- dicted is 87.8% at process parameters equal to19MPa applied pressure,90◦Cpressingtemperatureand3.8%(w.b.)moisturecon- tent.Theactualoilrecoveryobtainedis87.4±0.5%.
3.3. Effectofindependentprocessingparameters
Theeffectofthefourindependentvariablesontheoilrecovery ofJatrophaisshowninFig.3.Oilrecoveryimprovedwithincreas- ingpressureasshowninFig.3a;inparticularwhentheapplied pressurewasincreasedfrom10to15MPa.However,uponincreas- ingtheappliedpressurefrom15to20MPa,oilrecoveryseemedto leveloff.Forexample,forJatrophakernelcontaining4%(w.b.)mois- tureandpressedat75◦Ctheoilrecoveryincreasedby5pointsfrom 78.6to83.4%byraisingtheappliedpressurefrom10to15MPa.The oilrecoveryonlyincreasedby1pointto84.8%whenthepressure wasraisedto20MPa.Athigherpressures,theinter-kernelvoidis muchsmallerthanatlowpressures,whichrestrictstheflowofoil.
Theseresultsindicatethatexcessivepressuredidnotnecessarily
Table4
ANOVAforresponsesurfacequadraticmodel.
Sourceofvariation Degreeoffreedom Sumofsquares Meansquare F-value p-Valueprobability
Model 9 323.92 35.99 83.85 <0.0001a
Pressure,P 1 95.48 95.48 222.44 <0.0001a
Temperature,T 1 106.28 106.28 247.58 <0.0001a
Moisture,M 1 0.17 0.17 0.39 0.5444b
PT 1 1.36 1.36 3.17 0.1053b
PM 1 8.20 8.20 19.11 0.0014a
TM 1 8.61 8.61 20.06 0.0012a
P2 1 8.03 8.03 18.71 0.0015a
T2 1 0.07 0.07 0.16 0.6957b
M2 1 28.32 28.32 65.98 <0.0001a
Residual 10 4.29 0.43
Lackoffit 5 1.46 0.29 0.52 0.7563b
Pureerror 5 2.83 0.57
Correctiontotal 19 328.22
R2 0.987
Adj.R2 0.975
Valuesof‘p-value’<0.0500indicatemodeltermsaresignificant.
aSignificant.
b Notsignificant.
haveapositiveinfluenceonoilrecovery.Asimilarobservationwas observedbyWillemsetal.(2008)whilepressingJatrophaseedand kernelfrom20to70MPa.
Pressingtemperaturesignificantlyraisedoilrecoveryasshown inFig.3b.Thiswasobservedfor Jatrophakernel with4%(w.b.) moisturecontentwhen pressed under15MPa, theoilrecovery increasedfrom80.0to86.5%byincreasingthetemperaturefrom60 to90◦C.Increasingthetemperaturecoagulatestheprotein,soften- ingthesolidstructureanddecreasestheoilviscosity.Kabuteyetal.
(2012)mentionedpre-heatingtemperaturesofthejatrophaseeds decreasedseedhardness(N/mm),contrary,seeddeformation(mm) increasedin relationto theeffect of pre-heating temperatures.
Burubaietal. (2007)reportedthat seeddeformationof African nutmegincreasedwithincreasingtemperatures.Increaseindefor- mationrelatedtothevolumeofoilreleasesfromcompressedcake.
Tambunanetal.(2012)foundthatincreasingpressingtemperature from50to80◦Cincreaseoilrecoveryfrom70to80.9%inpressing ofJatrophaseed.
TheeffectofmoisturecontentonoilrecoveryisshowninFig.3c.
Ata pressing temperature of 75◦C and an applied pressure of 15MPa,oilrecoveryincreasedfrom80.3to83.4%withanincrease inmoisture contentfrom3 to 4%(w.b.) butthen decreased to 80.1%whenthemoisturecontentwasraisedto5%(w.b.).Moisture contentaffectsthehardnessandcompactnessofthekernel.Sam- plewithlowermoisturecontenthashigherhardness(Karajand Müller,2010)whileathighermoisturecontentsplasticizationpro- motescakecompactionandrestrictstheflowofoil.Asreported byKabuteyet al.(2011)increasing moisturecontent from1to 10%reducesthedeformation.Theresultshowsthatthevalueof 4%(w.b.)wasfoundtobetheoptimummoisturecontentforJat- rophakernel.Achehebetal.(2012)alsofound3.95%astheoptimum moisturecontentforpistachionut.
3.4. Effectsofinteractivefactors
Theeffectofinteractionbetweenfactors,i.e.appliedpressure, temperatureand moisturecontentisshown in Figs.4–6.Fig.4 depictstheeffectofappliedpressureandpressingtemperatureon thepressingofJatrophakernelwithmoisturecontentof4%.The effectoftemperatureismoresignificantatlowerappliedpressure.
Forexampleat10and20MPa,theoilrecoveryincreasefrom74.8to 82.1%andfrom81.8to87.5%,respectively.Thiseffectwasobserved byMpagalileandClarke(2005)whenpressingcoconut.Theinter- actionbetweenappliedpressureandpressingtemperatureonoil recoverywasexplainedbyBargaleetal.(1999).Theeffectmaybe attributedtotheinteractionoftemperatureandpressurewhichat higherlevelstendtobecomecounteractive:increasingthetemper- aturedecreasestheviscosityoftheoiltherebyincreasingitsfluidity throughthecompressedmediumwhereasanincreaseinpressure makesthecakeharderwhichrestrictstheflowofoil.
Acomparativestudyoftheresultsshowedthattheoilrecovery atanypressurewasaffectedbythemoisturecontentofthesample.
Theeffectofappliedpressureandmoisturecontentonoilrecovery isshowninFig.5.Anincreaseinoilrecoveryisobservedwhenthe appliedpressurewasincreasedfrom10to15MPa.However,the oilrecoverytendstoleveloffathigherpressurelevelsfrom15to 20MPa.Thiseffectwasobservedforalllevelsofmoisturecontent usedinthisstudy.Forexample,whenJatrophakernelwith3%(w.b.) moisturecontentispressedat75◦C,theoilrecoveryincreasedfrom 74.5to80.3%uponincreasingpressurefrom10to15MPaandthen slightlyincreasedto82.7%whenthepressureisraisedto20MPa.
AsimilartrendwasobservedforJatrophakernelwith5%(w.b.) moisturecontent:theoilrecoveryincreasedfrom76.3to80.1%
byincreasingpressurefrom10to15MPaandthenleveledoffto 80.4%towardsapressureof20MPa.Thislevelingoftheoilrecovery
Table5
Validationofmodelequation.
Run Variableproperties Oilrecovery(%)
P T M Experimental(E) Predicted(P)
1 19 90 3.8 87.4±0.5 87.8
2 15 60 3 75.1±1.1 75.9
3 15 60 5 77.7±1.7 77.7
4 10 60 4 75.3±0.6 74.8
5 20 60 4 81.2±1.1 81.8
X2=
(E−P)2P =0.02.
Fig.2. Correlationofactualconversionsandvaluespredictedbythemodel(a)and normalprobabilityofresiduals(b).
wasappreciablynoticeableinsampleswithhighermoisturecon- tent,whichshowsthatundersuchconditionshighpressureisnot effectiveinincreasingtheoilrecovery.Thiseffectwasobservedby MpagalileandClarke(2005)whenpressingcoconut.
Moisturecontentaffectsthehardnessandcompactnessofthe kernel.Atlowermoisturecontents,evaporationcausesthesur- faceofthesampletoharden,thusrequiringahigherpressureto beusedtoovercomethehardenedsampleduringpressing.Thus increasingpressureincreasestheoilrecovery.Athighermoisture contents,thepresenceofwaterwillactsasplasticizerbetweenthe protein-richcakeandoilwhichformspaste-likeplastizedmate- rial.Thispromotescakecompactionandrestrictstheflowofoil.
Thusincreasingpressureathighermoisturecontentgavehardly noticeablechangesorevensomereductioninoilrecovery.
Theeffectoftemperatureandmoisturecontentonoilrecovery isshowninFig.6.ThiswasobservedforJatrophakernelwith3%
(w.b.)moisturecontentwhenpressedat15MPa:theoilrecovery increasedfrom75.9to84.5%byincreasingthetemperaturefrom60
Fig.3. Effectofvariousindividualparameters:appliedpressure(a),temperature (b),andmoisturecontent(c),onthepressingofJatrophakernel.Oneparameteris variedwhiletheothersarekeptconstantattheircenterpoints.
Fig.4.Responsesurface(a)andcontourplots(b)ofoilrecoveryasfunctionof appliedpressureandtemperatureatmoisturecontentof4%(w.b.).
to90◦C.AsimilartrendwasobservedforJatrophakernelcontain- ing5%(w.b.)moisturecontent:oilrecoveryincreasedfrom77.7to 82.1%byincreasingpressingtemperaturefrom60to90◦C.Increas- ingtemperatureatlowmoisturecontentgaveamorenoticeable effectonoilrecoverythanathighermoisturecontent.
Theinteractiveeffectbetweenpressingtemperatureandmois- ture content can be explained as: increasing the temperature reduces the viscosity of the oil thereby increasing its fluidity throughthecompressedmedium.Anincreaseinmoisturecontent makesthecakelesscompressibleandrestrictstheoilflow.Thus increasingpressingtemperatureathighermoisturecontentgave lessnoticeablechangesinoilrecoveryincrement.
Manipulatingmoisturecontentenabledhighoilexpressioneffi- cienciesto be attained even at relatively low pressure or low temperature.Lowmoisturecontentsamplesneedlowertemper- atureorappliedpressuretogetthesameoilrecoverycompared tohighermoisturecontent.Forexample,oilrecoveryof83.4%was attainedatapressureof15MPaandatemperatureof75◦Cforker- nelswith4%moisturecontent,whileanoilrecoveryof82%was obtainedatapressureof20MPaandtemperatureof90◦Cforker- nelswith5%moisturecontent.Similarresultswereobtainedby MpagalileandClarke(2005)andEbeweleetal.(2010)inthestudy ontheeffectofmoisturecontent,appliedpressureandpressing temperatureoncoconutgratingandrubberseedexpression.
Fig.5.Responsesurface(a)andcontourplots(b)ofoilrecoveryasfunctionof appliedpressureandmoisturecontentatpressingtemperatureof75◦C.
3.5. Oilquality
OilexpressionexperimentswerecarriedoutwithJatrophaker- nels from different plantations. The best processing conditions showedtobe:19MPaappliedpressure,90◦Cpressingtemperature and3.8%(w.b.)moisturecontent.Dataontheoilsaresummarized in Table6.The qualityof theobtainedoilmettheDIN 51605:
2010-10standardforplantoilbasedfuelexceptonphosphorus contentandgroupIImetals.Acidvalueisameasurementofthe hydrolyticdegradationofoilduringstorageorprocessing,i.e.the hydrolysisofester bondsin lipidsbyenzymeaction orbyheat andmoisture,resultsintheliberationofFFAs.Theoxidativestabil- ityindexisassociatedwiththeoxidativedegradationofoilwhich resultsindevelopmentofrancidity.AhigherOxidativeStability Index(OSI)expressedinhoursInductionPeriod(IP)meansmore resistancetooxidation.Phosphoruscontentindicatesthepresence ofphosphor-derivedcomponentsinoilsuchasphospholipidsand phytates.Theiodinevalueisameasureofthedegreeofunsatura- tionoftheoil(thehighertheiodinevalue,thegreaterthedegree ofunsaturation).JatrophaoilfromKalimantanhasahigherphos- phoruscontentcomparedtoJatrophaoilsfromSubangasisshown inTable6.Soilnutritionorfertilizerisconsideredasacausefor thisresultwithJatrophafromKalimantangrownonmorenutri- tiouslandthanthatfromSubang.AccordingtoLickfettetal.(1999), phosphoruscontentinseedisaffectedbythelevelofphosphorus
Table6
PhysicalandchemicalcharacteristicsofJatrophacurcasL.oil.
Parameters Testmethods Units DIN51605:2010-10 Value
Flashpoint ASTMD6450 ◦C 101(min) 200b–204a
Coldpoint ASTMD6749 ◦C (−1)b–(−2)a
Pourpoint ASTMD2500 ◦C (−3)b–(−4)a
Iodinevalue DINEN14111 g/100g 125(max) 108a–113b
Density(15◦C) Pycnometer kg/m3 910–925 923a–925b
Kinematicviscosity(40◦C) Viscometer mPas 36(max) 35.8a–37.2b
Oxidativestability(110◦C) EN14112 hours 6(min) 9.49a–13.99b
Acidvalue DINEN14104 mgKOH/g 2(min) 0.25b–0.6a
Phosphorus DINEN14107 ppm 3(max) 2.8a–25.9b
Group2metal(Ca+Mg) DIN51627-6 ppm 2(max) 3.8a–31.2b
Moisturecontent DINENISO12937 ppm 750(max) 697a–747b
Note:JatrophasampleusedforanalysisarefromSubang(a)andKalimantan(b).
Fig.6.Responsesurface(a)andcontourplots(b)ofoilrecoveryasfunctionof temperatureandmoisturecontentatappliedpressureof15MPa.
fertilizersupplyorphosphoruscontentinsoil.Thehigheroxida- tivestabilityofJatrophaoilfromKalimantanisattributedtothe presenceofphospholipids,phytatesandothernaturalantioxidants.
Phospholipidisknownforitsantioxidantpropertiesandcanwork togetherwithnaturalantioxidantssuchastocopherols(Choeand Min,2006).
4. Conclusions
A face centered central composite RSMdesign was usedto determinetheoptimumconditionsfortheprocessingparameters
appliedintheextractionofoilfromJatrophacurcasL.kernelby hydraulic pressing. It is found that applied pressure, pressing temperature,andthequadraticsofappliedpressureandmoisture content, as well as interaction between applied pressure and moisturecontentandbetweenpressingtemperatureandmoisture contentaresignificantfactorsaffectingtheoilrecovery.Thesecond orderpolynomialequationdevelopedinthisstudyshowsahigh correlationbetweenobservedandpredicatedoilrecoveryvalues.
Responsesurfaceanalysiswasfoundtobeagoodapproachforvisu- alizingprocess-parameterinteraction.Themodelsdevelopedby RSMshallbeusefulforpredictingtheoptimumprocessingcondi- tiontoachievemaximumoilrecoveryofJ.curcasL.kernelpressing.
Oilextractionexecutedatanappliedpressureof19MPa,apressing temperatureof90◦Candakernelmoisturecontentof3.8%(w.b.) gaveanactualoilrecoveryof87.4±0.5%whichcloselymatchesthe predictedvalueof87.8%.Theoilqualitysatisfiestherequirements setbytheDIN51605:2010-10normforplantoilbasedfuels.
Conflictofinterest
Theauthorshavedeclarednoconflictofinterestonfinancialand orcommercialapplicationofthisstudy.
Acknowledgements
The authors are grateful to the Koninklijke Nederlandse AkademievanWetenschappen,ScientificProgrammeIndonesia– Netherlands(SPIN-KNAW),TheNetherlandsforthefinancialsup- port. We also would like toacknowledge AnnaApeldoorn and JanvanderVeldefortechnicalandanalyticalsupport.Finally,the authorsthankBoyA.Fachri,YusufAbduhandallcolleaguesatthe UniversityofGroningen,DepartmentofChemicalEngineering,for allsupportandvaluablesuggestions.
References
Acheheb,H.,Aliouane,R.,Ferradji,A.,2012.Optimizationofoilextractionfrom PistaciaatlanticaDesf.seedsusinghydraulicpress.AsianJ.Agric.Res.6,73–82.
Achten,W.M.J.,Mathijs,E.,Verchot,L.,Singh,V.P.,Aerts,R.,Muys,B.,2007.Jatropha biodieselfuelingsustainability?BiofuelsBioprod.Bioref.1,283–291.
Bargale,P.C.,Ford,R.J.,Sosulski,F.W.,Wulfsohn,D.,Irudayaraj,J.,1999.Mechan- icaloilexpressionfromextrudedsoybeansamples.J.Am.OilChem.Soc.76, 223–229.
Burubai,W.,Akor,A.J.,Igoni,A.H.,Puyate,Y.T.,2007.Effectsoftemperatureandmois- turecontentonthestrengthpropertiesofAfricannutmeg(Monodoramyristica).
Int.Agrophys.21,217–223.
Choe,E.,Min,D.B.,2006.Mechanismsandfactorsforedibleoiloxidation.Comp.Rev.
FoodSci.FoodSafety5,169–186.
Ebewele,R.O.,Iyayi,A.F.,Hymore,F.K.,2010.Considerationsoftheextractionprocess andpotentialtechnicalapplicationsofNigerianrubberseedoil.Int.J.Phys.Sci.
5,826–831.
Kabutey,A.,Hérak,D.,Sedláˇcek,A.,2011.Behaviourofdifferentmoisturecontentof JatrophacurcasL.seedsundercompressionloading.Res.Agric.Eng.2,72–77.
Kabutey,A.,Sedláˇcek,A.,Diviˇsová,M.,Svato ˇnová,T.,2012.HeattreatmentofJatropha curcasL.seedsundercompressionloading.Sci.Agric.Bohemica43,116–121.
Karaj,S.,Müller,J.,2010.Determinationofphysical,mechanicalandchemicalprop- ertiesofseedsandkernelsofJatrophacurcasL.Ind.CropsProd.32,129–138.
Kumar,A.,Sharma,S.,2008.Anevaluationofmultipurposeoilseedcropforindus- trialusesJatrophacurcasL.:areview.Ind.CropsProd.28,1–10.
Lee,A.,Chaibakhsh,N.,Rahman,M.B.A.,Basri,M.,Tejo,B.A.,2010.Optimizedenzy- maticsynthesisoflevulinateesterinsolvent-freesystem.Ind.CropsProd.32, 246–251.
Lickfett,T.,Matthäus,B.,Velasco,L.,Möllers,C.,1999.Seedyield,oilandphytate concentrationintheseedsoftwooilseedrapecultivarsasaffectedbydifferent phosphorussupply.Eur.J.Agron.11,293–299.
Makkar,H.P.S.,Becker,K.,2009.Jatrophacurcas,apromisingcropforthegeneration ofbiodieselandvalue-addedcoproducts.Eur.J.LipidSci.Tech.111,773–787.
Mooney,D.,Swift,R.,1999.ACourseinMathematicalModeling,firsted.TheMath- ematicalAssociationofAmerica,UnitedStatesofAmerica.
Mpagalile,J.J.,Clarke,B.,2005.Effectofprocessingparametersoncoconutoilexpres- sionefficiencies.Int.J.FoodSci.Nutr.56,125–132.
Olajide,J.O.,Igbeka,J.C.,Afolabi,T.J.,Emiola,O.A.,2007.Predictionofoilyieldfrom groundnutkernelsinanhydraulicpressusingartificialneuralnetwork(ANN).
J.FoodEng.81,643–646.
Pinzi, S., Garcia, I.L., Lopez-Gimenez, F.J., Luque de Castro, M.D., Dorado, G., Dorado,M.P.,2009. Theidealvegetableoil-based biodieselcomposition:a review of social, economical andtechnical implications. EnergyFuels23, 2325–2341.
Subroto,E.,Manurung,R.,Heeres,H.J.,Broekhuis,A.A.,2014.Mechanicalextrac- tionofoilfromJatrophacurcasL. kernel:effectofprocessingparameters, http://dx.doi.org/10.1016/j.indcrop.2014.06.018.
Tambunan,A.H.,Situmorang,J.P.,Silip,J.J.,Joelianingsih,A.,Araki,T.,2012.Yieldand physicochemicalpropertiesofmechanicallyextractedcrudeJatrophacurcasL oil.BiomassBioenergy43,12–17.
Willems,P.,Kuipers,N.J.M.,deHaan,A.B.,2008.Hydraulicpressingofoilseeds:
experimentaldeterminationandmodelingofyieldandpressingrates.J.Food Eng.89,8–16.