2024年4月7日发(作者：主机开了显示器没反应无信号)

Energy&Fuels2002,16,1571-15751571

RapidandAccurateSARAAnalysisofMediumGravity

CrudeOils

y*

NewMexicoPetroleumRecoveryResearchCenter,NewMexicoInstituteofMiningand

Technology,Socorro,NewMexico

ReceivedMay31,2002

alysisiswidely

usedtodividecrudeoilcomponentsaccordingtotheirpolarizabilityandpolarityusingafamily

msarisebecausetheanalyticaltechniquesdonot

fthedata,however,rarelydistinguishbetweenthe

differenttechniques,assumingthatSARAfractionvaluesgeneratedbyanyofthecommonly

inethisassumptionformediumgravity

crudeoilsandthreeSARAanalysismethods:gravity-drivenchromatographicseparation,thin-

layerchromatography(TLC),andhigh-pressureliquidchromatography(HPLC).Resultsfora

suiteofsixcrudeoilsamplesshowthatasignificantvolumeofvolatilematerialthatcontains

ovedHPLCmethodisintroduced

thatgivesanalysescomparabletotheASTM-recommendedchromatographicmethodinlesstime

rnalconsistencytestisrecommendedforevaluating

SARAfractiondata.

Introduction

Analysisofthecompositionofcrudeoilscanbe

endlesslycomplex;theamountofdetailcollectedshould

bedictatedbytheapplicationforwhichthedatais

pleanalysisschemeistodivideanoil

intoitssaturate,aromatic,resin,andasphaltene(SARA)

uratefractionconsistsofnonpolar

materialincludinglinear,branched,andcyclicsatu-

ics,whichcontainoneor

morearomaticrings,ain-

ingtwofractions,resinsandasphaltenes,havepolar

tinctionbetweenthetwoisthat

asphaltenesareinsolubleinanexcessofheptane(or

pentane),whereasresinsaremisciblewithheptane(or

pentane).Thisclassificationsystemisusefulbecause

itidentifiesthefractionsoftheoilthatpertainto

asphaltenestability;itthusshouldbeusefulinidentify-

ingoilswiththepotentialforasphalteneproblems.

SARAanalysisbeganwiththeworkofJewelletal.

1

Threemainapproacheshavebeenusedtoseparate

crudeoilsandotherhydrocarbonmaterialsintoSARA

-geladsorptionchromatographymethod

isthebasisofASTMD2007-93.

2

Thismethodrequires

afairlylargeoilsample,istime-consuminganddifficult

toautomate,andrequireslargequantitiesofsolvents.

irst

grouparehigh-pressureliquidchromatographic(HPLC)

*:1-505-835-6031.E-mail:jill@

.

(1)Jewell,D.M.;Weber,J.H.;Bunger,J.W.;Plancher,H.;Latham,

.1972,44,1391-1395.

(2)ASTMD2007-93:“StandardTestMethodforCharacteristic

GroupsinRubberExtenderandProcessingOilsbytheClay-Gel

AdsorptionChromatographicMethod,”ASTM,1993.

methods,firstintroducedbySuatoniandSwab.

3

Early

HPLCtechniquesusedsilicaoraluminacolumnsto

elopments

inpreparationofthebondedphaseofHPLCcolumnss

especiallyNH

2

-bondedmaterialssmadeitpracticalto

separateheavierfractionsofpetroleumsamples.

4-8

HPLCtechniquesarefaster,morereproducible,and

morereadilyautomatedthantheASTMcolumntech-

cases,however,itisnecessarytoremove

theasphaltenefractionbeforeproceedingwiththe

tenesareeitherirreversibly

adsorbedorprecipitatedduringthesaturateelution

step,andquantitativerecoverycannotbeachieved.

9

Thefastestseparationmethodusesthin-layerchro-

matography(TLC)onquartzrodsthatarecoatedwith

thecolumnandHPLC

techniques,asphaltenesneednotbeseparatedfrom

othercrudeoilcomponentsbeforechromatographic

artechnologyknownastheIatroscan

thatcombinesTLCwithflameionizationdetection

(TLC-FID)wasfirstappliedbySuzuki

10

toautomate

quantitativeSARAseparations,amethodwhichhas

(3)Suatoni,J.C.;Swab,.1975,13,361-

366.

(4)Miller,.1982,54,1742-1746.

(5)Radke,M.;Willsch,H.;Welte,.1984,56,2538-

2546.

(6)Grizzle,P.L.;Sablotny,.1986,58,2389-2396.

(7)Fe´lix,G.;Thoumazeau,E.;Colin,J.M.;Vion,.

Chromatogr.1987,10,2115-2132.

(8)Chaffin,J.M.;Lin,M.S.;Liu,M.;Davison,R.R.;Glover,C.J.;

Bullin,togr.,l.1996,19,1669-1682.

(9)McLean,J.D.;Kilpatrick,Fuels1997,11,570-

585.

(10)Suzuki,Y.21stAnnualMeetingoftheJapanSocietyfor

AnalyticalChemistry,1972;47(inJapanese).

10.1021/ef0201228CCC:$22.00©2002AmericanChemicalSociety

PublishedonWeb10/29/2002

1572Energy&Fuels,Vol.16,No.6,2002

ilSampleProperties

oil°API

25.2

22.6

31.3

28.8

37.2

31.1

densityat20°C

(g/cm

3

)

0.8956

0.9161

0.8673

0.8795

0.8409

0.8685

MW

(g/mol)

236

268

235

240

213

270

RIat20°C

1.5128

1.5137

1.4851

1.4976

1.4769

1.4906

P

RI

1.4513

1.4231

1.4444

1.4465

1.4223

∼1.44

FanandBuckley

n-C

7

asph

(%)

8.7

2.8

1.9

5.8

1.3

4.1

A-95

C-LH-99

C-R-00

S-Ven-39

SQ-95

Tensleep-99

sincebeenusedextensively.

11,12

Barman

13

compared

SARAanalysesofheavyhydrocarbondistillatesbythe

-FIDusesvery

actionsinacrudeoil

sampleareoftenwell-resolvedusingestablisheddevel-

tativeresultsareobtained

frompeakareas,assumingthateachSARAfractionhas

anidenticalFIDresponsefactor.

ComparisonsofSARAfractionmeasurementsby

differenttechniques,usuallyfromdifferentlaboratories,

extent,these

differencesmightberealscausedbytheuseofdifferent

work,identicalsampleswereexamined

byseveraldifferenttechniquestoilluminatethe

ovedHPLC

techniqueandatestforinternalconsistencyinSARA

dataarepresented.

ExperimentalMaterialsandMethods

ium-gravity,deadcrudeoils,varying

inAPIgravityfrom22.6to37.2°,wereusedinthisstudy.

SelectedphysicalandchemicalpropertiesareshowninTable

1(reproducibletoatleast(5intheleastsignificantdigit

shownforeachmeasurement).Densitiesweremeasuredusing

aMettler/PaarDMA40withacirculatingwaterbathfor

vitieswerecalculatedfrom

measureddensities,correctedto60°emolecular

weight(MW)wasdeterminedbyfreezingpointdepression

(PrecisionSystemsCryoscope5009).Kinematicviscosities

weremeasuredinCannon-Fenskeviscometersandconverted

untsofasphaltenepre-

cipitatedbyn-heptane(1goil:40mLheptane)rangedfrom

1.3to8.7%.Refractiveindex(RI)wasmeasuredwithanIndex

InstrumentsGPR11-37automaticrefractometer.P

RI

isthe

RIofamixtureofoilandtheleastamountofheptaneinwhich

asphalteneaggregatescanbeobservedmicroscopically(ata

magnificationofabout320×).ThedifferencebetweenRIof

theoilsampleandP

RI

isameasureofasphaltenestability.

14

thecrudeoilswas

testedusingthefullASTM-recommendedprocedure(ASTM

D2007-93

2

)

ASTMprocedureisachromatographicseparationofthenon-

asphalticoilcomponentsthroughtwocolumns:anAttapulgite

clay-packedcolumnadsorbstheresinsandasecondcolumn

packedwithactivatedsilicagelseparatesaromaticsfromthe

saturatefraction.A50:50mixtureoftolueneandacetoneis

aromaticscanberecoveredbySoxhletextractionofthesilica

lecomponentslostduringtheprocess

arecalculatedbyweightdifference.

-5Iatroscan(IatronLabsInc.,Tokyo),

equippedwithaflameionizationdetector(FID),interfaced

(11)Karlsen,D.A.;Larter,m.1991,17,603-617.

(12)Vela,J.;Cebolla,V.L.;Membrado,L.;Andres,J.M.J.

.1995,33,417-424.

(13)Barman,.1996,34,219-225.

(14)Buckley,J.S.;Hirasaki,G.J.;Liu,Y.;VonDrasek,S.;Wang,

J.X.;Gill,l.1998,16,251-285.

tionofparaffinsfromsingle-anddouble-

ringaromatics.

withamodel203PeakSimpledatasystem(SRIInstrument),

wasusedtoscansilica-coatedquartzrods(Chromarod-SIII,

IatronLaboratories).Therodswere15.2cmlongand1.0mm

indiameter,withauniformcoatingof5.0µmsilicaparticles

(porediameter60Å).TheFIDdetectorwasoperatedwitha

puregradeofhydrogenataflowrateof160mL/min;airata

flowrateof2.0L/minwassuppliedbyapump;scanspeed

was60s/scan.

CrudeoilsamplesweredissolvedinHPLCgradedichloro-

methane(DCM)ataconcentrationof20mg/mL.A1µL

repeatingsyringe(AlltechAssociate,Inc.)wasusedtospot

10to20µpment

stepsincludedexposuretoHPLCgradehexanefor30min,

HPLCgradetoluenefor10min,anda95:5mixtureofDCM

sweredriedinairfor3min

eatmentproducedfourwell-

resolvedpeaksrepresentingsaturates,aromatics,resins,and

ativelossesduringthedevelopmentsteps

wereevaluatedbytestsofsimilarlytreatedpreparativethin-

layerchromatographyplatesforwhichchangesinweightcould

bemeasuredandbycomparisonofpeaksizesforundeveloped

rods.

Cchromatographicseparationsystem

consistedofaModel110Apump(Beckman),anR401dif-

ferentialrefractometer(Waters),aU6Kuniversalinjector

(Waters),aModel7040high-pressureswitchingvalve(Rheo-

dyne),andaWaters486UVdetector(Millipore).Analogue

signalsfromtheRIdetectorwereinterfacedtothemodel203

PeakSimpledataacquisitionsystem(SRIInstrument).Two

3.9×300mmµBondapakNH

2

columnswith10µmpacking

(Waters)wereusedinseriesforchromatographicseparation

etectorwasoperatedatawavelengthof

254nmtomonitorelutionofeachfraction.

Selectivityofthecolumnswastestedusingmixturesof

esofn-decane(C

10

),n-tridecane

(C

13

),n-pentadecane(C

15

),andn-octadecane(C

18

)inhexane

elutedasasinglepeak,asdidhexanesolutionsoftolueneand

1,3-diisopropylbenzene(DIPB).Hexanesolutionsofasingle-

ringaromatic(DIPB)withatwo-ringcompound(1-methyl-

naphthaleneor1-MN)elutedastwowell-resolvedpeaks.

MixturesofthehexanesolutionsofDIPB,1-MN,andC

15

elutedasthreedistinctpeaks(Figure1).

Crudeoil(1-mL)wasweighedanddissolvedin40mLof

hexaneinanopen-topscrewcapvialwithaTeflon/silicon

48h,asampleofmaltenesdissolvedinhexane

waswithdrawnthrougha0.2µ

withdrawnmaltenesolutionwassealedina5-mLcrimp-top

SARAAnalysisofMediumGravityCrudeOils

parationofC-R-00maltenes.

ctorCalibration

classhydrocarbonsignal(area/mg)

saturatesn-heptadecane4647

1-ringaromaticsDIPB7088

2-ringaromatics1-MN12517

3-ringaromaticsphenanthrene16022

glassvialwithanaluminumsealandPTFE/siliconseptum.

Theprecipitatedasphalteneswererecoveredbyfiltration

througha0.22µmfilter,dried,l0.5mL

aliquotsofthemaltene/hexanemixturewereinjectedontothe

HPLCcolumns,usingagastightHPLCsyringe(Hamilton).

Saturatesandaromaticswereelutedwithhexaneataflow

rateof1.5mL/untsofsaturatesandaromatics

werecalculatedfrompeakareasusingcalibrationfactors

measuredforknowncompounds(Table2).Standarddeviations

forpeakareasofknowncompoundswerelessthan1%ofthe

alchromatogramisshowninFigure

2.

Resinswerestronglyadsorbedanddidnotelutewith

hexane.A30%(v/v)dichloromethane/hexanebackflushwas

usedtoelutetheresinsaccumulatedfromatleastthree

replicateinjectionstoensurethatsufficientresinscouldbe

collectedforaccurategravimetricdeterminationoftheamount

ofresinremainingaftersolventevaporation.

TheprincipalimprovementsintheHPLCmethoddescribed

hereoverpreviouslyproposedHPLCtechniquesincludethe

following:(i)analysisofthesaturateandaromaticfractions

withoutsolventevaporation,thusavoidingtheuncertainties

duetoevaporationofsomeofthemorevolatilematerialalong

withthesolvent;and(ii)improvedgravimetricquantification

oftheamountofresinselutedbybackflushingthecolumns

afterrepeatedinjectionsofmaltenes.

ResultsandDiscussion

TheSARAfractionsmeasuredbyallthreetechniques

ultsoftheASTM

methodprovideabaselineagainstwhichotherSARA

3showsthe

resultsoft-testsappliedtoTLC-FIDandHPLCresults

pairedwiththoseobtainedbytheASTMmethodforthe

sameoiltodeterminestatisticalsignificanceofthe

differencesinthedata.

isonsbetweenASTM

andTLC-FIDresultsshowconsistent,statistically

significantdifferencesforallfractionsexcepttheas-

phaltenes(Table3).Theamountofvolatilemateriallost

beforethedetectorresponseisrecordedisnotroutinely

ountisashighas60%

inthisstudy,allofwhichcomesfromthesaturateand

tionswereapportionedonthe

basisofpeakareassummedto100%,ignoringthe

volatilematerial,theamountsofresinsandasphaltenes

Energy&Fuels,Vol.16,No.6,20021573

set-testsforSignificanceofDifferences

betweenSARAMethods

a

p-values

b

fractionASTMvsTLC-FIDASTMvsHPLC

saturates<0.00010.0236

saturates+volatiles0.00490.8192

aromatics0.00640.1634

resins0.04430.0914

asphaltenes0.2214

a

Statisticallysignificantdifferencesarehighlightedinbold

italictype.

b

pistheprobabiltythatthesamplescomefrom

differentpopulations;p<0.05indicatesthatdifferencesbetween

theresultsofdifferentmeasurementmethodsarestatistically

significant.

heamount

ofvolatilematerialhasbeenmeasuredinthisstudy,it

mightbeincludedinthesaturateandaromaticfrac-

tions,butadditionalinformationisneededonhowto

ion

ofthevolatileandsaturateamountsfromtheTLC-

FIDmethodwouldsignificantlyoverestimatethesatu-

ratefractionandunderestimatetheamountofaromat-

separatingallofthevolatilematerialupto

somefairlyhighboilingpoint(KarlsenandLarter

11

measuredlossesofapproximately50%ofn-C

14

,which

boilsat253.5°C;whereasthelossofn-C

16

,whichboils

at286.8°C,wasnegligible)andanalyzingthatfraction

separately,thencombiningthetwosetsofdatawould

onebeabletoapplytheTLC-FIDtechniquetomedium

gravityoilsampleswithanyconfidence,buttheadvan-

tageofrapidanalysissthemainreasonforusingthis

ioningthehighmolec-

ularweightpolarmaterialsbetweenresinsandasphalt-

enes,whichisalwaysbased,ratherarbitrarily,on

solvency,inevitablywillvaryfromtheASTMandHPLC

methodsaswellsincethesolventsusedarenotidenti-

asisofthesecomparisons,theTLC-FID

techniquecannotberecommendedforroutineSARA

analysisofmediumgravityoils.

encesbetweentheimproved

HPLCSARAfractionsandthosemeasuredbythe

ASTM-recommendedmethodweremainlyinthevolatile

erenolossestoevapora-

tionintheHPLCtechnique,unliketheASTMmethod

inwhichthereweresomelosses(duringevaporationof

thesolventafterremovaloftheasphaltenesandinthe

variousfractionrecoveryandsolventevaporationpro-

cesses).However,ifallofthevolatilemateriallostin

theASTMmeasurementswasassumedtobelongtothe

saturatefraction,theamountsofsaturates,aromatics,

andresinsdeterminedbytheASTMandHPLCmethods

werestatisticallyindistinguishable(seeTable3).The

measurementtechniquefortheamountofasphaltenes

isthesameforbothmethods,henceasphalteneswere

ordifferencesbetween

thetwotechniquesaretheamountsoftime,sample,

htechniques,thereisa

two-dayperiodrequiredforseparationoftheasphalt-

that,however,theHPLCtechniquere-

quiresonlyabout1hpersampleincludingthree

replicateseparationsofsaturatesandaromaticsand

backflushingofthecolumntoeluteresins,compared

mof

oilisadequatetogivethesameaccuracybytheHPLC

1574Energy&Fuels,Vol.16,No.6,2002FanandBuckley

isonofSARAfractionsdeterminedbytheASTMmethod,bytheimprovedHPLCmethod,andbyTLC-FID.

solventrequirementsareontheorderof300mLand2

LpersamplefortheHPLCandASTMtests,respec-

tively.

InternalConsistencyofSARAFractionData.

Therearemanyvariationsofthespecificanalyses,all

ofwhicharereferredtobytheacronymofSARA,but

hedetails

oftheseparationtechniquearenotalwaysprovided,a

methodisneededtoevaluateSARAdatafromunspeci-

fiedtestssothatASTMandHPLCdatacanbedistin-

eevaluation

schemeisshownhere,basedoncomparisonsbetween

theSARAfractiondataandAPIgravity,whichisalmost

universallyreportedforstock-tankoilsamples.

Agroupof87crudeoilswasevaluatedusingthe

Afraction

datathusobtainedwereusedtofindregressioncoef-

ficientsthatcorrelatethepercentagesofeachfraction

withtheAPIgravitiesmeasuredforthesamesamples.

Thefollowingempiricalrelationshipwasobtained:

˚API

calculated

)74.5-0.306×S-0.385×

A-1.08×R-0.763×As(r

2

)0.68)

whereSisthewt%ofsaturates,Aisthewt%of

aromatics,Risthewt%ofresins,andAsisthewt%

fmeasuredvscalculatedAPI

gravityisshowninFigure4.

TotesttheconsistencyofanysetofSARAfraction

data,similarplotscanbeprepared,asillustratedfor

esome

scatter,theASTMdataclearlyfollowthepredicted

SARAAnalysisofMediumGravityCrudeOils

mateofAPIgravityobtainedfromSARA

fractiondatacanbecomparedtothemeasuredAPIgravity

asatestoftheinternalconsistencyoftheSARAdata.A

correlationwasdevelopedwith87differentcrudeoilsamples

spanningtherangefromabout15toover40°API,testedwith

Mdatasetcloselyreproduces

thecorrelation,whereasthereisnocorrelationbetweenthe

SARAfractiondatafromTLC-FIDmeasurementsandAPI

gravityoftheoil.

trend(r

2

)0.76),whereasthereisnocorrelation

betweentheTLC-FIDdataandAPIgravitymeasured

atitisnotsufficientto

etrendforoilswith

varyingpropertiesoverarangeofAPIgravitythatis

relationshouldbeappliedonlyover

Energy&Fuels,Vol.16,No.6,20021575

thegravityrangeofthesamplesthatwereusedinits

,fromalowvalueofabout15toahigh

ofabout40°API.

Conclusions

•MeasurementsofSARAfractionsarehighlydepend-

sofdifferentmethodsmay

nisneededincomparing

results,evenfromsimilarmethods.

•TheTLC-FIDtechniquecannotbeusedtotest

mediumgravityoilswithoutconsiderableadditional

analysistoaccountforcomponentsthatboilattemper-

aturesupto250°ortioningofhighmolecular

weight,polar,aromaticmaterialintoasphaltenesand

resinsdoesnotcorrespondtotheasphaltenesandresin

fractionsdefinedbytheASTMmethod.

•AnefficientHPLCmethodologyhasbeendeveloped

andshowntocorrespondcloselytothemoretime-

consumingASTM-recommendedmethodofSARAanaly-

sis.

•SARAfractiondatacanbetestedforinternalcon-

sistencybycomparingAPIgravityvaluescalculated

fromtheSARAfractionswithmeasuredvaluesofAPI

gravityforthesameoilsample.

n

rkwas

supportedbytheNationalEnergyTechnologyLabora-

tory(NETL)oftheUSDOEthroughcontractDE-AC26-

99BC15204andbysupportfromindustrialsponsors

includingBP,Chevron,GazdeFrance,IFP,Norsk

Hydro,ilswereprovidedby

ARCO,Chevron,Shell,andtheUniversityofWyoming.

EF0201228