MPahenhenlu.com

1Water-GasShiftModelingofCoalGasificationinanEntrained-FlowGasifierXijiaLu*andTingWang504-280-2398,504-280-7183xlv@uno.
edu,twang@uno.
eduEnergyConversion&ConservationCenterUniversityofNewOrleansNewOrleans,LA70148,USAABSTRACTMostofthereactionratesforthewater-gasshift(WGS)reactionwereobtainedfromexperimentsundersimplifiedlaboratoryconditionswithspecificcatalysts.
Afewofthereactionrateswithoutusingcatalystswereobtainedundersupercritical(water)conditions,withthepressuremuchhigherthanthoseinatypicalgasifier.
Ineithercase,itisnotclearhowthepublishedreactionratescanbetrustfullyusedtopredicttheactualWGSreactionrateinagasifierwithoutthepresenceofacatalystandunderdifferenttemperatureandpressureconditionsthanthoseinthelaboratory.
ThisstudyfocusesfirstonreviewingthepublishedWGSreactionrateswithandwithoutthepresenceofcatalysts,followedbycalibratingtheWGSreactionratetomatchtheexperimentaldatatakenfromtheJapaneseCRIEPIresearchgasifier.
The3-DNavier-Stokesequationsandninespeciestransportequationsaresolvedwithsevenglobalgasificationreactions(threeheterogeneousandfourhomogeneous,)andatwo-stepthermalcrackingmodelforvolatiles.
TheChemicalPercolationDevolatilization(CPD)modelisusedforthedevolatilizationprocess.
ThreedifferentcaseswiththreedifferentfiniteratesfortheWGSreaction(Jones'srateundercatalyticconditionsandWade'sandSato'sratesundernon-catalyticconditions)areconducted.
TheresultshowsthatthethreeoriginallypublishedratesarealltoofastandoverpredicttheexperimentalWGSreactionrate.
AddingabackwardWGSreactionratedoesn'tslowdownthereactionrate,resultinginthesamegascompositionandtemperatureatthegasifierexitasthatcalculatedwithoutaddingthebackwardWGSreaction.
Thepre-exponentialrateconstantvalue(A)ofeachreactionrateisthereforeadjustedtomatchtheexperimentaldata.
TheresultsshowthatallthreeWGSreactionratescanmatchtheexperimentaldatareasonablywell.
Theexittemperaturecanbematchedwithin2%(20K).
ThemolefractionsofCOandH2Ocanbematchedfairlywellwithin4percentagepoints(or10%);however,thesimulatedH2molefractionsarealways7-9percentagepoints(orabout40%)higherthantheexperimentaldata.
Thesuggestedcalibratedreactionratesaredocumented.
1.
0INTRODUCTIONGasificationisanincompleteoxidationprocessofconvertingvariouscarbon-basedfeedstocksintocleansyntheticgas(syngas),whichisprimarilyamixtureofhydrogen(H2)andcarbonmonoxide(CO),withminoramountsofmethane(CH4)andinertnitrogengas(N2).
Feedstockispartiallycombustedwithoxygenandsteamathightemperatureandpressurewithlessthan30%oftherequiredoxygenforcompletecombustion(i.
e.
,thestoichiometricamount)beingprovided.
Thesyngasproducedcanbeusedasafuel(usuallyforboilersorgasturbinestogenerateelectricity);itcanalsobemadeintoasubstitutenaturalgas(SNG),orhydrogengasand/orotherchemicalproducts.
Gasificationtechnologyisapplicabletoanytypeofcarbon-basedfeedstock,suchascoal,heavyrefineryresidues,petroleumcoke,biomass,andmunicipalwastes.
TohelpunderstandthegasificationprocessinProceedingsofthe28thInternationalPittsburghCoalConference,Pittsburg,PASeptember12-15,2011Paper45-12gasifiersandsubsequentlyusethelearnedknowledgetoguidedesignsofmorecompact,morecosteffective,andhigherperformancegasifiers,computationalfluiddynamics(CFD)hasbeenwidelyemployedasapowerfultooltoachievethesegoals.
Chenetal.
(2000)developedathree-dimensionalsimulationmodelforentrained-flowcoalgasifiers,whichappliedanextendedcoal-gasmixturefractionmodelwiththeMultiSolidsProgressVariables(MSPV)method.
Themodelemployedfourmixturefractionsseparatelytrackthevariablecoaloff-gasesfromthecoaldevolatilization,char-O2,char-CO2,andchar-H2Oreactions.
Bockelieetal.
(2002(a))developedacomprehensiveCFDmodelingtool(GLACIER)tosimulateentrained-flowgasifiers,includingasingle-stage,down-firedsystemandatwo-stagesystemwithmultiplefeedinlets.
TheU.
S.
DepartmentofEnergy/NationalEnergyTechnologyLaboratory(NETL)developeda3DCFDmodeloftwocommercial-sizedcoalgasifiers[GuentherandZitney(2005)].
ThecommercialCFDsoftware,FLUENT,wasusedtomodelthefirstgasifier,whichwasatwo-stage,entrained-flow,slurry-fedcoalgasifier.
TheEulerian-Lagrangianmethodwasusedinconjunctionwiththediscretephasemodeltosimulatetheentrained-flowgasificationprocess.
Thesecondgasifierwasascaled-updesignofatransportgasifier.
TheNETLopensourceMFIX(MultiphaseFlowInterphaseeXchanges)Eulerian-Eulerianmodelwasusedforthisdensemultiphasetransportgasifier.
NETLhasalsodevelopedanAdvancedProcessEngineeringCo-Simulator(APECS)thatcombinesCFDmodelsandplant-widesimulation.
APECSenablesNETLtocoupleitsCFDmodelswiththesteady-stateprocesssimulator,AspenPlus.
From2005to2011,SilaenandWanghavedoneaseriesofstudyofentrained-flowgasificationprocessusingthecommercialCFDsolver,FLUENT.
SilaenandWang(2005)investigatedtheeffectsofseveralparametersongasificationperformance,includingthecoalinputcondition(slurryordrypowder),oxidant(oxygen-blownorair-blown),wallcooling,andvariouscoaldistributionsbetweenthetwostages.
Thesimulationresultsprovidethetemperatureandspeciesdistributionsinsidethegasifier.
In2006,theyinvestigatedtheeffectofflowinjectiondirectionsonthegasificationperformanceusingthesamegenerictwo-stage,entrained-flowgasifier.
In2010,theydidresearchthatinvestigatedtheeffectsofdifferentparametersongasificationperformance,includingfiveturbulencemodels,fourdevolatilizationmodels,andthreesolidcoalsizes.
WithseveralimprovementsintheCFDmodeling,includingupdatingthefiniteratesforheterogeneousreactions,addingChemicalPercolationDevolatilization(CPD)devolatilizationmodel,andaddingtwo-stagevolatilescrackingreactions,SilaenandWang(2011)againconductedaninvestigationontheeffectsofdifferentoperationparametersinthegasificationprocess,includingthecoalinputcondition(dryvs.
slurry,)oxidant(oxygen-blownvs.
air-blown,)anddifferentcoaldistributionsbetweenthetwostages.
IncollaborationwiththeresearchteamofIndustrialTechnologyResearchInstitute(ITRI),WangandSilaeneffectivelyemployedtheCFDgasificationmodeltoinvestigategasificationprocessundertheinfluencesofdifferentpartloads,twodifferentinjectors,andthreedifferentslaggingtapsizes(Wang,etal.
,2006,2007,and2010).
In2011,Wang,etal.
performedthesimulationontheeffectsofpotentialfuelinjectiontechniquesongasificationperformanceinordertohelpdesignthetop-loadedfuelinjectionarrangementforanentrained-flowgasifierusingacoal-waterslurryastheinputfeedstock.
Twospecificarrangementswereinvestigated:(a)coaxial,dual-jetimpingementwiththecoalslurryinthecenterjetandoxygenintheouterjetand(b)four-jetimpingementwithtwosinglecoal-slurryjetsandtwosingleoxygenjets.
3OneoftheimportantreactionsduringthegasificationprocessistheWater-GasShift(WGS)Reaction(CO+H2OCO2+H2.
)TheWGShasbeentraditionallyusedforadjustingtheH2/COratioinasyngastomeetthespecificationsforvariousfinalproducts.
TheforwardWGSreactionfavorsrelativelylowtemperatures(under600oC.
)Whenthetemperatureishigherthan1200oC,thereversereactionstartstodominate.
ThistrendcanbeseenfromtheequilibriumconstantshowninTable1.
Table1Water-gasshift(WGS)reactionequilibriumbalanceconstantTheWGSreactionratehasbeendiscoveredtoplayanimportantroleinaffectingtheaccuratepredictionofthesyngascompositionduringsimulationsofthegasificationprocess(SilaenandWang,2009).
TheearliestdatarecordingtheWGSreactiondatesbackto1888(Rhodesetal.
,1995),anditsprominencecamewiththeHaberammoniasynthesisprocessandcatalystdevelopmentbyBoschandWildein1912(Twigg,1989).
MostofthereactionratesfortheWGSreactionwereobtainedfromexperimentswithspecificcatalystsunderlaboratoryconditionsofrelativelynarrowrangesofpressureandtemperature.
Afewofthereactionrateswithoutusingcatalystswereobtainedundervarioussupercritical(water)conditionsbecausealargeexcessofwatersolventcouldpossiblydrivethereactiontoproducehydrogenwithoutacatalyst.
However,thepressureunderasupercriticalconditionismuchhigherthanthatinanoperatinggasifier.
Ineithercase,itisnotclearhowthepublishedreactionratescanbetrustfullyusedtopredicttheactualWGSreactionrateinagasifierwithoutthepresenceofcatalystsandunderdifferenttemperatureandpressureconditionsthanthoseusedinthelaboratory.
DuetotheunavailabilityofappropriateWGSreactionratesforbroadoperatingconditionsinactualgasifiers,theobjectiveofthisstudyistoobtainanappropriaterepresentativeWGSglobalreactionrateundernon-catalyticconditionsbycalibratingtheWGSrateagainsttheexperimentaldataformanactualgasifier.
1.
1ReviewofWGSReactionRates1.
1.
1WGSCatalyticReactionsChenetal.
(2008)investigatedthecharacteristicsofcarbonmonoxideconversionandhydrogengenerationfromtheWGSreactionexperimentallyusingahigh-temperaturecatalystandalow-temperaturecatalyst.
Theimportantparameters,includingthecatalysttype,residencetimeofthereactantsinthecatalystbed,reactiontemperature,andCO/steamratio,wereaddressedastheinfluentialfactorsthataffectedtheperformanceoftheWGSreaction.
Theexperimentalresultsshowedthatwhentheresidencetimewasaslongas0.
09s,theWGSdevelopedwellnomatterwhichcatalystwasused.
ItalsorevealedthattheWGSreactionswiththehigh-temperaturecatalystandthelow-temperaturecatalystweregovernedbychemicalkineticsandthermodynamicequilibrium.
ItisdifficulttonarrowdowntheexpressionfortheWGSreactionwithacatalystbecausetherateofthereactionisdependentonvariousparameters,includingthecompositionofthecatalyst,theactivesurfaceareaandstructureofthecatalyst,thesizeofthecatalyst,ageofthecatalyst,theoperatingT(oC)600800100012001400logKp1.
3960.
5530.
076-0.
222-0.
4244temperatureandpressure,andthecompositionofthegases.
Smith,etal.
(2010)madeacomprehensivereviewofresearchontheWGSreactionrateandthedevelopmentsinmodelingapproachesfordesigningWGSreactors.
Theyconsolidatedalistingofthevariousimportantkineticexpressionspublishedforboththehightemperatureandthelowtemperaturewater-gasshiftreactionsalongwiththedetailsofthemake-upofthecatalystsandtheoperatingconditionsatwhichthekineticswereobtained.
SelectedstudiesfromSmithetal.
'sreviewareshowninTable2-4.
Table2WGSReactionRatewithNobleMetalCatalysts(Smithetal.
,2010)Table3WGSReactionRatewithHigh-TemperatureCatalysts(310°C-450°C)(Smithetal.
,2010)CatalystOpertatingConditionsArrheniusParametersReferenceAE(kJ/mol)Ru300°C-1000°C0.
008to0.
05contacttimeCoatedonaluminasupport5wt%loading1.
6*107(1/s)80Wheeleretal.
(2004)Ru/Ceria5.
0*107(1/s)80Ni8.
0*107(1/s)85Ni/Ceria1.
7*108(1/s)85Pd4.
0*106(1/s)100Pd/Ceria4.
0*107(1/s)100Pt1.
0*106(1/s)80Pt/Ceria2.
5*107(1/s)80Rh/SiO2350°C3.
23*106(molecules/s/site)22.
8±2.
5(kcal/mole)Grenobleetal.
(1981)Pt/Al2O3270°C1.
9*106(molecules/s/site)19.
6±2.
5(kcal/mole)Pt/SiO2340°C1.
9*106(molecules/s/site)19.
1±0.
8(kcal/mole)Pt/C340°C3.
84*106(molecules/s/site)25.
5±1.
4(kcal/mole)CuO.
1CeO.
8O2-y(Cuceria)200°C-350°C1.
8*103(1/s)61Henriketal.
(2006)CuO.
1CeO.
8O2-y(Cuceria)300°C-350°C4.
0*103(1/s)78CatalystOperatingConditionsArrheniusParametersReferenceAE(kJ/mol)Powerplantdata1/4"*3/8",2.
20g/cm39.
4*107(1/s)21.
4(kCal/gmol)Rase(1977)Fe3O4/Cr2O33-5bar,573°C-633°ClnA=26.
195Keiskietal.
(1996)Fe3O4/Cr2O38wt%Cr2O31atm,350°C-440°ClnA=11.
5112Rhodesetal.
(2003)Fe3O4/Cr2O31bar,380°C-450°ClnA=10.
1±0.
2118±1Fe3O4/Cr2O36bar,380°C-450°ClnA=12.
0±0.
2124±1Fe3O4/Cr2O327bar,350°C-450°ClnA=7.
4±0.
1111±180-95%Fe3O4,5-10%Cr2O3,1-5%CuO1atm,450°C100.
659(mol/gcat-s)88Sanetal.
(2009)5Table4WGSReactionRatewithLow-TemperatureCatalysts(200°C-250°C)(Smithetal.
,2010)1.
1.
2WGSnon-catalyticreactionsTheabovestudiesreviewedbySmith,etal.
(2010)areallinvolvedwithcatalysts,whereasthosestudieswithoutusingcatalystswereallconductedundersupercritical(water)conditions.
Watanabeetal.
(2001)didtheresearchonthepartialoxidationofn-hexadecaneat673Kinsupercriticalwaterandfoundthatwhenacarbontooxygenratioof3:1wasused,theCOconcentrationintheproductscouldreachalmost40%.
So,itispossibletousepartialoxidationofhydrocarbonstoproduceCOviatheWGSreaction.
Inotherwords,itispossibletodevelopanon-catalyticwayofreforminghydrocarbonswithoutusinghightemperatures(1073-1273K).
HirthandFranck(1993)reportedtheequilibriumconstantKaofWGSreactionat773-873Kand40MPa.
ItisalsomentionedthatKaislargelydifferentfromthatintheidealgasstateandtheequilibriumshiftedtothereactants'sidewithincreasingpressure.
Holgateetal.
(1992)proposedaglobalrateexpressionoftheWGSreactionbyconductingexperimentsinnon-catalytic,supercriticalconditionsat712-866Kat24.
6MPa,RWGS=102.
6±1.
2exp[(-67±11)/RT](CO)0.
81±0.
19.
Satoetal.
(2004)studiedthekineticsoftheWGSreactionundernon-catalytic,supercriticalconditions(653-713Kand10-30MPa)withaninitialCO/H2Oratioof0.
03inaflow-typereactor.
Byanalyzingtherateconstantsobtainedbytheirstudyandthosereportedpreviously,aglobalreactionmodelfortheWGSreactionundersupercriticalconditionswasproposedask=105.
58±1.
38exp(-1.
16±0.
19*105/RT)/sat10-59.
6MPaand653-866K.
Wadeetal.
(2008)conductedexperimentsontheWGSreactionnon-catalyticallyinthetemperaturerangeof770to1050Kwithanoperatingpressureof24MPa.
TheyobtainedtherateconstantsofA=2.
512x105andE=1.
325x105J/kmol.
1.
1.
3WGSreactionratesusedinCFDmodelingUsually,therearetwoapproachestomodelingtheWGSreactionrateinCFD.
Thefirstapproachistousethedetailedkineticswithbothforwardandbackwardelementaryreactions.
Inthisapproach,theratesoftheelementaryreactionsareusuallytoomanytobecalculatedintheCFDmodel,sothereactionratesarecalculatedseparatelyinanothersoftwarepackage,likeChemkin,ateachiterationasthelocaltemperatureandpressurechange.
Theadvantageofthisapproachisthatitprovidesthemostappropriatevehicletomodelthecorrectkineticsforthereactionrates,whilethedisadvantagesarethattheelementarykineticsmaynotbeadequatelyknownandthatitcouldbedifficulttoachieveconvergenceinCFDcomputation.
ThesecondapproachistouseGlobalReactionRatesthatareobtainedbyexperiments.
Sincetheratesareglobal,theratesusuallyreflectthenetratebetweenthedifferenceoftheforwardandbackwardrates.
TheadvantagesarethattheglobalratessimplifytheCatalystOpertatingConditionsArrheniusParametersReferenceAE(kJ/mol)ICI52-1(Copperbasedcatalyst)density=5.
83g/cm31atm,200°CK=5.
37*10-7(mol/m2s)/atm1+mSalmietal.
(1989)Cu-ZnO-Al2O3(EX-2248)SudChemie200-250m,120-250°ClnA=12.
647.
4Choietal.
(2003)42%CuO-ZnO-Al2O3123°C-175°CCO/H2O=1/34.
9*106(S-1)71Henriketal.
(2006)CuO-ZnO-Al2O31atm,200°C-79Koryabkinaetal.
(2003)6complexCFDmodelingandconservethecomputationalpower,whilethedisadvantagesarethat(a)theexperimentaldataareusuallyobtainedinrelativelynarrowtemperatureandpressureranges,(b)theratesareusuallyobtainedwhentheproductsareleanandthepresenceofotherspeciescommonlyinagasifierisnonexistent,and(c)thereisstillalackofsufficientdatacoveringtheentirespanofthetemperaturerangeforthegasificationprocess.
WatanabeandOtaka(2006)performedanumericalsimulationwiththecoalgasificationmodelontheJapanese2tons/day,researchscalecoalgasifiersupportedbytheCentralResearchInstituteofElectricPowerIndustry(CRIEPI).
TherateconstantsoftheWGSreactionthattheyusedisA=2.
75x1010andE=8.
37x107J/kmolfortheforwardreactionrateandA=2.
65x10-2andE=3.
96x103J/kmolforthebackwardreactionrate.
Theinfluenceoftheairratioongasificationperformance,gastemperaturedistribution,andproductgascompositionwerepresentedanddiscussedintheirpaper.
Thenumericallysimulatedresultswerecomparedfavorablywiththeexperimentaldata,andmostfeaturesofthegasificationprocesswereclaimedtohavebeencapturedadequately.
Ajilkumaretal.
(2007)usedthesameWGSfiniterateconstantasemployedbyWatanabeandOtakatosimulatethesamecoalgasificationprocessintheCRIEPIgasifier,buttheircomputationaldomainwasasmallsectionofasimplified,axisymmetriccylinder(i.
e.
thiswasessentiallya2-Dcomputation.
)Itwasnotclearhowtheinletconditionsandfuelinjectionwerescaleddownfrom3Dto2D;however,theyclaimedthatthepredictedresultsshowedgoodagreementwiththeexperimentaldataoftheCRIEPIgasifier.
IfAjikumaretal.
'sclaimiscorrect,itimpliesthatthegasfier'sgeometryandinjectionlocationsarenotcriticalfordesigningagasifier.
Furtherstudiesareneededtoverifythisimplication.
SilaenandWang(2011)usedJonesandLindstedt'srate(1998,abbreviatedasJones'sratelater)andcomparedtheirCFD-simulatedsyngasresultswiththatfromtheactualproductionofacommercial,slurry-fed,entrained-flowcoalgasifierfedfromthebottom.
PerhapsduetothefactthatJones'sratewasobtainedundercatalyticconditions,theyfoundthatJones'sratewastoofastandtheyhadtopurposelyreducethereactionrateconstanttoA=2.
75x102tomatchtheoperatingdata.
Inthisstudy,inadditiontothemodifiedJonesrate,theothertwoWGSreactionrates(Sato'sandWade'srates)obtainedundernon-catalyticconditionsaretobeemployedforcomparisonasshowninTable5.
Table5WGSReactionRatesusedinthisstudySourcesoftheWGSrateCatalystk=ATnexp(-E/RT)withn=0JonesandLindstedt,1998YesA=2.
75x1010,E=8.
38x107J/kmolWadeetal.
,2008NoA=2.
512x105,E=1.
325x105J/kmolSato,etal.
2004NoA=105.
58±1.
38,E=1.
16x105J/kmol1.
1.
4ExperimentaldataInordertogetanappropriateWGSreactionrate,detailedandaccurateexperimentaldatasetsareneededtocalibratetheCFDmodel.
However,mostoftheexperimentaldataavailableinthepublicdomainwasobtainedafterthesyngascoolingorafterthegasclean-upprocess.
Forlimiteddatatakeninthegasifierbeforethesyngascoolingsection,typicallynoinformationonthewatervaporconcentrationisavailable.
Lackofadequate"rawdata"hasmadecalibrationofthegasificationmodelandtheWGSreactionrateuncertainanddifficult.
Forexample,thedataofsyngascompositionpublishedfromtheWabashRiverCoalGasificationRepoweringProject(2000,2002)didn'tshowthe7molefractionofwatervaporattheexitofthegasifier,althoughthewatervaporinformationwasshownforsyngascompositionafterdesulfurization.
TheexperimentaldataprovidedbytheTampaElectricPolkPowerStationIGCCProject(2002)wasthecleanedsyngascomposition,whichwastakenafterthesyngascoolerandgascleanupprocesses.
Hughesetal.
(2010)providedthesyngasdatafromatwo-tonneperday(slurryfeedrate)pilot-scalegasifier,and,similarly,nowatervapormolefractionwasgiven.
WainedandWhitty(2010)performedtestsina1ton/daypressurized,slurry-fed,oxygen-blown,entrained-flowcoalgasifier.
Theexperimentaldataalsolackedinformationforthewatervaporcontentattheexit.
Sofar,totheauthors'knowledge,theonlypublishedexperimentalsyngasdataobtainedbeforesyngascoolingwithwatervaporcontentinformationisfromtheCRIEPIresearchscalecoalgasifierpresentedinthepaperbyWatanabeandOtaka(2006).
Therefore,theCRIEPIdataisusedforcalibratingtheWGSreactionrateinthispaper.
1.
2GlobalGasificationChemicalReactionsThisstudydealswiththeglobalchemicalreactionsofcoalgasification(SmootandSmith,1985)thatcanbegeneralizedinreactions(R1.
1)through(R1.
10)inTable6.
Table6SummaryofreactionrateconstantsusedinthisstudyReactionsReactionTypeReactionheat,ΔH°R(MJ/kmol)k=ATnexp(-E/RT)(n=0)ReferenceAE(J/kmol)HeterogeneousReactionsR1C(s)+O2→COPartialcombustion-110.
50.
0526.
1*107Chenetal.
(2000)R2C(s)+CO2→2COGasification,Boudouardreaction+172.
00.
07321.
125*108R3C(s)+H2O→CO+H2Gasification+131.
40.
07821.
15*108HomogeneousReactionsR4CO+O2→CO2Combustion-283.
12.
2*10121.
67*108WestbrookandDryer(1981)R5CO+H2O(g)CO2+H2WaterGasshift-41.
02.
75*10108.
38*107JonesandLindstedt(1998)R6CO+3H2CH4+H2OMethanation-205.
7kf=4.
4*10111.
68*108kb=5.
12*10-142.
73*104BenyonP.
(2002)R7CH2.
2538O0.
3015→0.
3015CO+0.
3025H2+0.
3168CH4+0.
1908C2H2Two-stepVolatilesCracking+12.
088EddydissipationN/AR8CH4+O2→CO+2H2Volatilesgasifi-cationviaCH4-35.
71R9C2H2+O2→2CO+H2Volatilesgasifi-cationviaC2H2-447.
83R10H2+O2→H2OOxidation-2426.
8x10151.
68x108JonesandLindstedt(1998)1)AllΔH°Rat298Kand1atm.
2)"+"Endothermic(absorbingheat),"-"Exothermic(releasingheat)8Inthisstudy,themethanationreactionsarenotconsideredsincetheproductionofmethaneisnegligibleunderthestudiedoperatingconditions.
Thevolatilesaremodeledtogothroughatwo-stepthermalcrackingprocess(R1.
7)andgasificationprocesses(R1.
8andR1.
9)withCH4andC2H2astheintermediateproducts.
ThecoalusedinthisstudyisJapaneseBlackCoal,whosecompositionsaregiveninTable7.
Thecompositionsofvolatilesarederivedfromthecoal'sheatingvalue,proximateanalysis,andultimateanalysis.
Theoxidantisconsideredtobeacontinuousflowandthecoalparticlesareconsideredtobediscrete.
Thediscretephaseonlyincludesthefixedcarbonandliquidwaterdropletsfromthemoisturecontentofcoal(5.
3%wt).
Othercomponentsofthecoal,suchasN,H,S,O,andash,areinjectedasgas,togetherwiththeoxidantinthecontinuousflow.
NistreatedasN2,HasH2,andOasO2.
SandasharenotmodeledandtheirmassesarelumpedintoN2.
Table7CompositionsofJapaneseBlackCoalVolatile46.
8C68.
2FixedCarbon35.
8H5.
71Moisture5.
3O12.
26Ash12.
1N0.
99100S0.
19HHV(kJ/kg)2.
74x104Ash12.
65100.
00ProximateAnalysis(MF),wt%UltimateAnalysis(MF),wt%2.
0COMPUTATIONALMODEL2.
1GoverningEquationsThetime-averaged,steady-stateNavier-Stokesequations,aswellasthemassandenergyconservationequations,aresolved.
Thegoverningequationsfortheconservationsofmass,momentum,andenergyaregivenas:()mjiiSρux=(1)()()jjiijijjjiiSuuρτxxPgρuρux+′′+=(2)()hipiiipiSμΦTuρcxTλxTuρcx++′′=(3)wherethesymmetricstresstensor,τij,isgivenby:+=kkijjiijijxuδ32xuxuμτ.
(4)Theequationforspeciestransportisgivenby:()jjiijiijiiSCuρxCρDxCρux+′′=.
(5)92.
2TurbulenceModelsThedetailedanalysisoftheeffectofvariousturbulencemodelsonthegasificationprocesshasbeendocumentedinapreviouspaperbySilaenandWang(2010).
Theycomparedtheresultsoffiveturbulencemodels,includingStandardk-ε,RNG(Re-NormalizedGroup)k-ε,Standardk-ωModel,ShearStressTransport(SST)k-ωModel,andReynoldsStressModel(RSM).
TheyreportedthatthestandardRSMmodelachievedthemostconsistentresultsandthek-εturbulencemodelwasfoundtoyieldreasonableresultsnexttotheRSMmodel,buttheRSMmodelusedalmostseventimesmorecomputationaltimethanthek-εmodel.
Followingtheirconclusionswithoutrepeatingthesameprocessagain,thestandardk-εturbulencemodelwithenhancedwallfunctionandvariablematerialpropertiesisusedinthisstudytoreducethecomputationaltime.
2.
3RadiationModelTheP-1radiationmodelisusedtocalculatethefluxoftheradiationattheinsidewallsofthegasifier.
TheP-1radiationmodelisthesimplestcaseofthemoregeneralP-NradiationmodelthatisbasedontheexpansionoftheradiationintensityI.
TheP-1modelrequiresrelativelylittleCPUdemandandcaneasilybeappliedtovariouscomplicatedgeometries.
ItissuitableforapplicationswheretheopticalthicknessaLislarge,where"a"istheabsorptioncoefficient,andListhelengthscaleofthedomain.
Inagasifier,theopticalthicknessisthickduetothepresencecoalparticles,soot,andashes.
Theheatsourcesorsinksduetoradiationarecalculatedusingtheequation:-qr=aG–4aGσa4(6)where()GCσσa31qssr+=(7)andqristheradiationheatflux,σsisthescatteringcoefficient,Gistheincidentradiation,Cisthelinear-anisotropicphasefunctioncoefficient,andσistheStefan-Boltzmannconstant.
Thefluxoftheradiation,qr,w,atthewalls,causedbytheincidentradiation,Gw,isgivenas()()www4wwwr,ρ12Gρ1πσT4ππq+=(8)whereεwistheemissivityandisdefinedasεw=1-ρwandρwisthewallreflectivity.
2.
4DiscretePhases(CoalParticlesorLiquidDroplets)Discretephasesincludecoalparticlesandliquiddroplets.
TheLagrangianmethodisadoptedinthisstudytotrackeachparticle.
Particlesintheairflowcanencounterinertiaandhydrodynamicdrag.
Becauseoftheforcesexperiencedbyadropletinaflowfield,theparticlescanbeeitheracceleratedordecelerated.
Thevelocitychangecanbeformulatedbympdvp/dt=Fd+Fg+Fo(9)whereFdisthedragofthefluidontheparticleandFgisthegravity.
Forepresentstheotherbodyforces,whichtypicallyincludethe"virtualmass"force(suchascentrifugalforce,coriolisforce,magneticforce,etc.
),thermophoreticforce,Brownianforce,Saffman'sliftforce,etc.
Vpistheparticlevelocity(vector).
Inthisstudy,Saffman'sliftforcereachesabout30%ofFg,soitisincludedinthisstudy.
10Whenthecoalisinjectedthroughtheinjectors,thewatercontentinthecoalistreatedasbeinginthecondensedphase(i.
e.
liquidwater),whichcan'tbelumpedintothecontinuousphase,sotheliquidwaterisatomizedintosmalldroplets.
Theoretically,evaporationoccursattwostages:(a)whenthetemperatureishigherthanthesaturationtemperature(basedonthelocalwatervaporconcentration,)waterevaporatesfromthedroplet'ssurface,andtheevaporationiscontrolledbythewatervaporpartialpressureuntil100%relativehumidityisachieved;and(b)whentheboilingtemperature(determinedbythegas-watermixturepressure)isreached,watercontinuestoevaporateeventhoughtherelativehumidityreaches100%.
Afterthemoistureisevaporatedduetoeitherhightemperatureorlowmoisturepartialpressure,thevapordiffusesintothemainflowandistransportedaway.
Therateofvaporizationisgovernedbytheconcentrationdifferencebetweenthesurfaceandthegasstream,andthecorrespondingmasschangerateofthedropletcanbegivenby:)C(Ckπddtdmsc2d∞=(10)wherekcisthemasstransfercoefficientandCsistheconcentrationofthevaporattheparticle'ssurface,whichisevaluatedbyassumingthattheflowoverthesurfaceissaturated.
C∞isthevaporconcentrationofthebulkflow,obtainedbysolvingthetransportequations.
ThevaluesofkccanbecalculatedfromempiricalcorrelationsbyRanzandMarshall(1952),0.
330.
5dcdSc0.
6Re2.
0DdkSh+==.
(11)whereShistheSherwoodnumber,ScistheSchmidtnumber(definedasν/D),Disthediffusioncoefficientofvaporinthebulkflow.
RedistheReynoldsnumber,definedasuν/d,uistheslipvelocitybetweentheparticleandthegas,anddistheparticlediameter.
Whentheparticletemperaturereachestheboilingpoint,thefollowingequationcanbeusedtoevaluateitsevaporationrate:()pfgp0.
5d2dc/h/)TT(c1ln)0.
46Re(2.
0dλπddtdm++=∞(12)whereλistheheatconductivityofthegas/air,andhfgisthedropletlatentheat.
cpisthespecificheatofthebulkflow.
Theparticletemperaturecanalsobechangedduetoheattransferbetweenparticlesandthecontinuousphase.
Theparticle'ssensibleheatchangesdependingontheconvectiveheattransfer,latentheat(hfg),speciesreactionheat(Hreac),andradiation,asshowninthefollowingequation:()44pfgpppdtdmhdtdmT)-h(TdtdTcmTAHfARppreachp+++=∞θσε(13)wheretheconvectiveheattransfercoefficient(h)canbeobtainedwithasimilarempiricalcorrelationtoEq.
14:33.
05.
0ddPrRe6.
00.
2λhdNu+==(14)whereNuistheNusseltnumber,andPristhePrandtlnumber.
Eq.
(13)isusedforbothwaterdropletsandcoalparticles.
11StochasticTrackingofParticles--Thevariousturbulencemodelsarebasedonthetime-averagedequations.
Usingthisflowvelocitytotracethedropletwillresultinanaveragedtrajectory.
Intherealflow,theinstantaneousvelocityfluctuationwouldmaketheparticledancearoundthisaveragetrack.
However,theinstantaneousvelocityisnotcalculatedinthecurrentapproachasthetimeaveragedNavier-Stokesequationsaresolved.
Onewaytosimulatetheeffectofinstantaneousturbulenceondropletdispersionistousethestochastictrackingscheme.
Basically,theparticletrajectoriesarecalculatedbyusingtheinstantaneousflowvelocity(u'u+)ratherthantheaveragevelocity(u).
Thevelocityfluctuationisthengivenas:(15)whereζisanormallydistributedrandomnumber.
Thisvelocitywillapplyduringacharacteristiclifetimeoftheeddy(te),givenfromtheturbulencekineticenergyanddissipationrate.
Afterthistimeperiod,theinstantaneousvelocitywillbeupdatedwithanewζvalueuntilafulltrajectoryisobtained.
Whenthestochastictrackingisapplied,thebasicinteractionbetweentheparticlesandthecontinuousphaseremainsthesame,andisaccountedforbythesourcetermsintheconservationequations.
Thesourcetermsarenotdirectly,butratherindirectlyaffectedbythestochasticmethod.
Forexample,thedragforcebetweenparticleandtheairflowdependsontheslipvelocitycalculatedbytheaveragedNavier-Stokesequationsifwithoutthestochastictracking.
Withstochastictracking,arandomvelocityfluctuationisimposedataninstantoftime,andthedragforceandadditionalconvectiveheattransferwillbecalculatedbasedonthisinstantaneousslipvelocity.
Thesourcetermsassociatedwiththisinstantaneousdragforceandconvectiveheattransferenterthemomentumandenergyequationswithoutanyadditionalformulation.
Forasteady-statecomputation,the"instantoftime"means"eachiterationstep.
"Therefore,theaveragedmomentumequationwillnotbeaffectedbythestochastictrackingscheme;ratherthetrajectoryoftheparticlewillreflecttheeffectoftheimposedinstantaneousperturbation.
2.
5DevolatilizationModelsAfterallthemoisturecontainedinthecoalparticlehasevaporated,theparticleundergoesdevolatilization.
SilaenandWang(2010)comparedtheeffectoffourdifferentdevolatilizationmodelsongasificationprocess:namelytheKobayashimodel,thesingleratemodel,theconstantratemodel,andtheCPD(ChemicalPercolationDevolatilization)model[FletcherandKerstein(1992),Fletcheret.
al(1990),andGrantet.
al(1989)].
TheanalysisconcludedthattheratecalculatedbytheKobayashitwo-competingratesdevolatilizationmodel[H.
Kobayashietal.
(1976)]isveryslow,whilethatoftheCPDmodelgivesamorereasonableresult.
Therefore,theChemicalPercolationDevolatilization(CPD)modelwaschosenforthisstudy.
TheCPDmodelconsidersthechemicaltransformationofthecoalstructureduringdevolatilization.
Itmodelsthecoalstructuretransformationasatransformationofachemicalbridgenetwork,whichresultsinthereleaseoflightgases,char,andtar.
Inthisstudy,thevolatilescontainedinthecoalarebackcalculatedasCH2.
2538O0.
3015fromthecoalheatingvalueandcoalcompositioninTable7.
Theinitialfractionofthebridgesinthecoallatticeis1,andtheinitialfractionofcharis0.
Thelatticecoordinationnumberis5.
Theclustermolecularweightis400,andthesidechainmolecularweightis50.
()0.
50.
522k/3ζu'ζu'==122.
6ReactionModels2.
6.
1Gasphase(homogeneous)reactionsForthegasphasereactions,boththeeddy-dissipationandfiniteratesareusedtocalculatethereactionrate,andthesmallerofthetworatesisusedinfurthercalculation.
TheEddy-dissipationmodeltakesintoaccounttheturbulentmixingofthegases.
Itassumesthatthechemicalreactionisfasterthanthetimescaleoftheturbulenceeddies.
Thus,thereactionrateisdeterminedbytheturbulencemixingofthespecies.
Thereactionisassumedtooccurinstantaneouslywhenthereactantsmeet.
Thenetrateofproductionordestructionofaspeciesisgivenbythesmallerofthetwoexpressionsbelow:′′=MνYminkεMAρνR(16)and′ΣΣ′=MνYminkεMBρνR(17)wherevisthestoichiometriccoefficientofthereactantandvisthestoichiometriccoefficientoftheproduct.
Thesmallerofthetwoexpressionsisusedbecauseitisthelimitingvaluethatdeterminesthereactionrate.
Thefiniteratemodeldoesnottakeintoaccounttheturbulentmixingofthespecies.
Instead,thereactionrateisexpressedinanArrheniusform.
ReactionratesinArrheniusformforallofthegasphasereactionsaregiveninTable6.
2.
6.
2Heterogeneousreactions(coalparticles)Therateofdepletionofthesolid,duetosurfacereactions,isexpressedasafunctionofthekineticrate,thesolidspeciesmassfractiononthesurface,andparticlesurfacearea.
Thereactionratesareallglobalnetrates.
ReactionrateconstantsusedinthisstudyaresummarizedinTable6.
Gasificationandcombustionofcoalparticlesaredictatedbythefollowingglobalprocesses:(i)evaporationofmoisture,(ii)devolatilization,(iii)gasificationtoCOand(iv)combustionofvolatiles,CO,andchar.
Therateofdepletionofsolidduetoasurfacereactionisexpressedas:(18)and(19)whereR=rateofparticlesurfacespeciesdepletion(kg/s)Ap=particlesurfacearea(m2)Y=massfractionofsurfacethesolidspeciesintheparticleη=effectivenessfactor(dimensionless)R=rateofparticlesurfacespeciesreactionperunitarea(kg/m2-s)pn=bulkpartialpressureofthegasphasespecies(Pa)D=diffusionratecoefficientforreactionk=kineticrateofreaction(unitsvary)N=apparentorderofreaction.
Thekineticrateofreactionisdefinedas:.
(20)TherateofparticlesurfacespeciesdepletionforreactionorderN=1isgivenby:RηARpΥ=NnDRpkR=()RTEneATk=13.
(21)ForreactionorderN=0,.
(22)Theeffectivenessfactor(η)issetunity(i.
e.
,notbeingused)foranapparentreactionratemodel.
2.
7PhysicalCharacteristicsoftheModelandAssumptionsThegeometryoftheCRIEPIgasifierdescribedbyWatanabeandOtaka(2006)isshowninFig1.
AsimplifiedgeometryshowninFigure2isemployedinthispresentstudywithoutincludingthecontractionsectionconnectingtheoxidationandreductionsections.
Thecoal(fuel)injectiondesignfollowsthatofCRIEPI,consistingofatwo-stageinjectionmethodwithfourtangentialinjectionsatthefirststageandtwooppositeinjectionsatthesecondstage.
Theresidencetimeisaround3-4seconds.
TherecycledcharisinjectedfromtwooppositecharinjectionlocationsatthefirststageintheCFDmodel.
Thegridconsistsof1,106,588unstructuredtetrahedralcells.
Inthesimulations,thebuoyancyforceisconsidered,varyingfluidpropertiesarecalculatedforeachspeciesandthegasmixture,andthewallsareassumedimpermeableandadiabatic.
Theflowissteadyandtheno-slipcondition(zerovelocity)isimposedonthewallsurfaces.
Fig.
1CRIEPIresearchscalecoalgasifierFig.
2Boundaryconditionsofthesimulatedgasifier3.
0BOUNDARYANDINLETCONDITIONSJapaneseBlackcoalisusedasthefeedstockinthisstudy;itscompositionisgiveninTable7.
TheCRIEPIgasifierisanair-blown,dry-fedgasifierandisoperatedat20atm.
Theinlet,boundary,andkDkDpAηRn+Υ=kAηRΥ=Topviewof1ststageinjectorsAir:0.
095kg/s,400KCoal:0.
0139kg/s,300KRecycledChar:0.
009kg/s,300KTopviewof2ndstageinjectorsAir:0.
0172kg/s,400KCoal:0.
0145kg/s,300KPressure:2MPaNoslipconditionatwallAdiabaticwallsInletturbulenceintensity10%Coal&AirCoal&AirRecycledChar&AirCoal&AirRecycledchar&AirCoal&Air9m1.
5m0.
75m2.
25mRawSyngas0.
75m14operatingconditionsforthebaselinecaseareshowninFigure2.
Atthefirststage,coalisinjectedtangentiallywithamassflowrateof0.
00695kg/sateachinjectionlocation.
Therecycledcharisinjectedoppositelywithamassflowrateof0.
0045kg/sateachinjector.
Thetotalmassflowrateofairisdistributedintofourinjectorsequallyat0.
095kg/s.
Atthesecondstage,coalisinjectedthroughapairofoppositeinjectorswithamassflowrate0.
00725kg/sateachinjectionlocation.
Thetotalmassflowrateofairis0.
0172kg/s.
Alloftheparametersstatedabovearethesameastheexperimentdata.
Thewallsareallsettobeadiabaticandwiththeno-slipcondition(i.
e.
zerovelocity).
Theboundaryconditionofthediscretephaseatthewallsisassignedas"reflect",whichmeansthediscretephaseelasticallyreboundsoffoncereachingthewall.
Theoperatingpressureinsidethegasifierissetat2MPa.
Theoutletissetataconstantpressureof1bar.
Thesyngasisconsideredtobeacontinuousflow,andthecoalandcharfromtheinjectionlocationsareconsideredtobediscreteparticles.
Theparticlesizeisuniformlygivenassphericaldropletswithauniformarithmeticdiameterof40μm.
Althoughtheactualsizedistributionofthecoalparticleswillbenon-uniform,asimulationusinguniformparticlesizeprovidesamoreconvenientwaytotrackthedevolatilizationprocessofcoalparticlesthananon-uniformsizedistribution.
Thecomputationisperformedusingthefinite-volume-basedcommercialCFDsoftware,FLUENT12.
0,fromANSYS,Inc.
Thesimulationissteady-stateandusesthepressure-basedsolver,whichemploysanimplicitpressure-correctionschemeanddecouplesthemomentumandenergyequations.
SIMPLEalgorithmisusedtocouplethepressureandvelocity.
Thesecond-orderupwindschemeisselectedforspatialdiscretizationoftheconvectiveterms.
Forthefiniteratemodel,wheretheEulerian-Lagrangianapproachisused,theiterationsareconductedbyalternatingbetweenthecontinuousandthediscretephases.
Initially,oneiterationinthecontinuousphaseisconductedfollowedbyoneiterationinthediscretephasetoavoidhavingtheflamedieout.
Theiterationnumberinthecontinuousphasegraduallyincreasesastheflamebecomesmorestable.
Oncetheflameisstablyestablished,fifteeniterationsareperformedinthecontinuousphasefollowedbyoneiterationinthediscretephase.
Thedrag,particlesurfacereaction,andmasstransferbetweenthediscreteandthecontinuousphasesarecalculated.
Basedonthediscretephasecalculationresults,thecontinuousphaseisupdatedinthenextiteration,andtheprocessisrepeated.
Convergedresultsareobtainedwhentheresidualssatisfyamassresidualof10-3,anenergyresidualof10-5,andmomentumandturbulencekineticenergyresidualsof10-4.
Theseresidualsarethesummationoftheimbalanceineachcell,scaledbyarepresentativefortheflowrate.
ThecomputationisperformedinaPC-clusterof20nodes.
4.
0RESULTSANDDISCUSSIONS4.
1ResultsofUsingThreeOriginalExperimentalWGSReactionRates.
ThethreeoriginalexperimentalWGSreactionratesshowninTable5areusedfirstforcomparison.
InWatanabeandOtaka'spaper(2006),thesumofthemolefractionsofCO,CO2,H2O,andH2inthesyngascompositionisonly45%.
Theother55%ofthegasesarenotstated,althoughthemajorcomponentisN2.
FortheconvenienceofcomparisonbetweentheexperimentaldataandtheCFDresults,themolefractionsofCO,CO2,H2O,andH2arerenormalizedto100%asshowninTable8.
TheCFDresultsshowthatallthreeoftheoriginallypublishedratesaretoofast,ascanbeseenbythemuchhighermolefractionofH2(product)andmuchlowermolefractionofremainingCO(reactant).
DuetotheexothermiccharacteroftheWGSreaction,therapidWGSreactionrateresultsintheexit15temperaturesofallthreecasesbeing100–130Khigherthantheexperimentaldata.
Basedonthisresult,itisnecessarytoslowdownthefiniterateofthewatershiftreactioninordertomatchtheexperimentalresultsbetter.
Theadoptedapproachistokeeptheactivationenergyintactandsubsequentlyreducethepre-exponentialconstantvalue(A)untiltheCFDresultsmatchtheexperimentalresults.
ForJones'srate,whichwasobtainedunderacatalyticcondition,itcouldbeexplainedthattheWGSreactionrateislowbecausenocatalystisusedinthegasifier.
ForWade'srateandSato'srate,therearetworeasonsthatmightpartiallycontributetothefasterexperimentalreactionrates:(a)Theexperimentswereconductedinanenvironmentdeprivedofconcentrationsofproducts(H2andCO2)andothergases,sotheforwardexperimentalreactionratescouldbefaster.
(b)Theexperimentalpressureandtemperatureconditionsaredifferentfromthoseinthegasifier.
Inthegasifier,thetemperatureishigherthanintheexperimentalconditions,thustheWGSreactionratecouldbelowerinthegasifierthanintheexperimentaltestcondition.
ThetemperatureandspeciesdistributionsinthegasifierareshowinFig.
3Table8Comparisonofexittemperatureandsyngascompositionbetweentheexperimentaldataandthesimulatedcasesusingthe3originalWGSratesMolefractionExperimentalDataJones'sRateA=2.
75x1010E=8.
38x107J/kmolWade'sRateA=2.
512x105E=1.
325x105J/kmolSato'sRateA=105.
58±1.
38E=1.
16x105J/kmolT1250K1356K1382K1378KH20.
200.
480.
480.
48CO0.
420.
270.
260.
27CO20.
200.
250.
260.
25H2O0.
18<0.
01<0.
01<0.
01Fig.
3GastemperatureandspeciesmolefractiondistributionsforusingtheJones'srate(A=2.
75x1010E=8.
38x107J/kmol)(Note:themolefractionsinthisfigurearebasedonallgasesandarenotthesameasthoseshowninTable8.
)164.
1.
1BackwardWGSreactionrateThebackwardWGSreactionrateisaddedtotheJones'sratewithA=2.
65x10-2andE=3.
96x103J/kmol(Gururajanetal.
,1992).
TheresultshowsthataddingthisbackwardWGSreactionratedoesn'tslowdownthereactionrateduringcomputation,resultinginthesamegascompositionandtemperatureatthegasifierexitasthosedatawithoutaddingthebackwardWGSreactioninTable8.
ThebackwardWGSreactionrateisthereforenotaddedtotherestofcases.
4.
2CalibrationoftheCatalyticWGSRates(Jones's)AgainsttheExperimentalDataThecalibrationagainsttheexperimentaldataisperformedbyconsecutivelychangingthepre-exponentialrateconstant,A,from2.
75x1010inJones'srateto2.
75x10-2,whiletheactivationenergyiskeptthesameastheoriginalvalue(E=8.
38x107J/kmol.
)TheCFDresultsofsevencasesareshowninTable9togetherwiththeexperimentaldata.
TheresultclearlyshowsthegradualchangeofsyngascompositionandtemperatureattheexitwhentheAvalueisreducedfrom2.
75x1010to2.
75x10-2.
ThereislittlechangeinsyngascompositionandtemperatureattheexitwhentherateconstantAisreducedfrom2.
512x1010to2.
512x104becausethewatervaporcontentisalmostcompletelyconsumedinbothcases.
ItdemonstratesthattherateA=2.
512x104isstilltoofastcomparedtotheexperimentaldata.
ThereisarelativelylargechangeofthegascompositionwhentherateconstantAisreducedfrom2.
75x102to2.
75.
WhentheA-valueisreducedbelow2.
75,theresultappearsstabilizedandfluctuatesslightly.
ThecasewithA=2.
75seemstoresultinthebestmatchwiththeexperimentaldata.
Table9Comparisonofthetemperatureandsyngascompositionatexitwiththeexperimentaldatabyconsecutivelyreducingthepre-exponentialvalue,A,oftheJones'srateExitExp.
Data2.
75x10102.
75x1042.
75x1022.
75x1012.
752.
75x10-12.
75x10-2T1250K1356K1296K1282K1278K1267K1246K1236KH20.
200.
480.
440.
410.
310.
270.
290.
27CO0.
420.
270.
310.
320.
410.
450.
440.
45CO20.
200.
250.
250.
230.
140.
120.
120.
12H2O0.
18<0.
01<0.
010.
050.
140.
160.
150.
16Figure4showsthegastemperatureandspeciesmolefractiondistributionsfortheCasewithA=2.
75andE=8.
38x107J/kmol.
Thegastemperatureishigherintheregionabovethesecondstageinjectionlocationthanitisintheregionbetweenthefirstandsecondstages.
Themaximumgastemperatureinthefirststagereaches1650Knearthefuelinjectionlocations,and,inthesecondstage,itreaches1570K.
Thisphenomenonisdifferentfromthewell-knownE-gasgasifierinwhichnooxygenisprovidedatthesecondstage,sothetemperatureafterthesecond-stageinjectionismuchlowerthaninthefirststagebecausetheendothermicChar-CO2(R2)andChar-steam(R3)gasificationprocessesareveryactiveafterthesecondstageinjection.
Inthisgasifier,itisveryinterestingtoseethatthehighestproductionofCO2occursnearthefirststageinjectionlocationsandthelowestproductionoccursnearthesecondstage.
TheCO2molefractionislowinmostpartsoftheproductionuntilthesyngasreachesthetopquarterofthegasifierwheretheCO2molefractionincreasesagain.
ThishistoryoftheCO2molefractionchangesindicatesthatcompletecharcombustion(R4)occursnearthefirst-stageinjection,butCO2iseffectivelyconsumedviaChar-CO2gasificationinmostparts17ofthegasifiertoproduceH2andCO,asalsoisevidencedbyincreasingH2andCOmolefractions.
TheWGSseemstobecomemoreactiveintheupperquarterofthegasifierasthetemperatureincreasesduetotheexothermiceffectoftheWGSprocess.
Inthisupperquarterregion,theactiveWGSreactioncanbeevidencedbytheincreasedH2andCO2anddecreasedCO.
Fig.
4GastemperatureandspeciesmolefractiondistributionsfortheCasewithmodifiedJones'srate(A=2.
75andE=8.
38x107J/kmol).
(Note:themolefractionsinthisfigurearebasedonallgasesandarenotthesameasthoseshowninTable9.
)4.
3CalibrationoftheNon-catalyticWGSRates(Wade'sandSato's)AgainsttheExperimentalDataThesamecalibrationprocessisperformedforWade'sandSato'srates,whichwereobtainedwithoutusingcatalysts,butinthesupercriticalrange.
Withoutshowingalltheincrementalcases,onlythreeselectedcasesareshowninTable10.
TheresultsofA=2.
512x10-3forWade'srateandA=1x10-3forSato'srateshowthebestmatcheswiththeexperimentaldata.
Table10TemperatureandsyngascompositionatexitforsixcasesandexperimentdatabasedonWade'srateandSato'srateMFA-valueofWades'Rate(E=1.
325x105J/kmol)A-valueofSato'sRate(E=1.
16x105J/kmol)(%)Exp.
Data2.
512x1052.
512x10-12.
512x10-3105.
58±1.
381x10-11x10-3T1250K1382K1280K1273K1378K1276K1238KH20.
200.
480.
430.
290.
480.
400.
28CO0.
420.
260.
290.
430.
270.
330.
46CO20.
200.
260.
240.
120.
250.
210.
11H2O0.
1800.
040.
1600.
060.
1518InSummary,allthreeWGSreactionratescanmatchtheexperimentaldatareasonablywellbyreducingthevalueofthepre-exponentialrateconstant,A.
Theexittemperaturecanbematchedwithin2%(20K)oftheexperimentalvalue.
ThemolefractionsofCOandH2Ocanbematchedfairlywellwithin4percentagepoints(or10%);however,thesimulatedH2molefractionsarealways7-9percentagepoints(orabout40%)higherthantheexperimentaldata.
5.
0CONCLUSIONSTheWGSreactionratehasbeendiscoveredtonotablyaffecttheresultofCFDmodelingofcoalgasificationprocessesinagasifier.
AlmostallofthepublishedWGSrateswereconductedwithcatalystsunderlimitedtemperaturerangesandatacertainfixedpressurecondition.
OnlyafewWGSrateswereobtainedwithoutinvolvingcatalysts,buttheywereperformedundersupercritical(water)conditions.
Therefore,employinganyofthepublishedWGSreactionratestosimulatethecoalgasificationprocessinagasifier,whichusuallydoesn'tusecatalystsanddoesn'toperateatthesametemperatureorpressureconditionsasinthelaboratoryconditions,isdestinedtoresultinmisleadingoruncertainresults.
TohelpcalibratetheglobalWGSreactionrate,threepublishedWGSreactionratesareusedinthisstudy.
TheyareJones'srate(A=2.
75x1010,E=8.
38x107J/kmol,)Wade'srate(A=2.
512x105,E=1.
325x105J/kmol,)andSato'srate(A=105.
58,±1.
38E=1.
16x105J/kmol.
)Theexperimentaldataarefromtheair-blown,dryfed,2-stage,research-scalegasifieroperatedbytheCentralResearchInstituteofElectricPowerIndustry(CRIEPI).
Theconclusionsare:(1)Alloftheoriginallypublishedratesaretoofast.
AddingabackwardWGSreactionratedoesn'tslowdownthereactionratetoomuch,resultinginthesamegascompositionandtemperatureatthegasifierexitasthecasewithoutaddingthebackwardWGSreactionrate.
(2)Eachofthethreeratesaresloweddownbyconsecutivelyreducingthepre-exponentialrateconstant,A,whiletheactivationenergyiskeptthesameastheoriginalvalue.
TheresultsshowthatallthreeWGSreactionratescanmatchtheexperimentaldatareasonablywellbyreducingthevalueofthepre-exponentialrateconstant,A.
Theexittemperaturecanbematchedwithin2%(20K).
ThemolefractionsofCOandH2Ocanbematchedfairlywellwithin4percentagepoints(or10%);however,thesimulatedH2molefractionsarealways7-9percentagepoints(orabout40%)higherthantheexperimentaldata.
(3)ThecalibratedglobalWGSreactionratesthatbestmatchtheexperimentaldataare:ModifiedJones'srate:A=2.
75,E=8.
38x107J/kmolModifiedWade'srate:A=2.
512x10-3,E=1.
325x105J/kmolandModifiedSato'srate:A=1x10-3,E=1.
16x105J/kmolItneedstobeemphasizedthatthecalibratedreactionratesareobtainedunderair-blownanddry-fedoperatingconditions.
Thesecalibratedratesmaynotbeapplicabletoslurry-fedoroxygen-blowngasifiersbecausethehigherwatervaporconcentrationinslurry-fedgasifiersandhigheroperatingtemperaturesinoxygen-blowngasifiersmayaffecttheglobalWGSrate.
Morestudiesareneededinthefuture,andmoreadequateexperimentaldataisneeded.
Theadequatedataarethosedatatakenimmediatelyattheendofgasificationsection,butrightbeforethesyngascoolingsectioninthegasifier,andthewatervaporconcentrationmustbeincluded.
Analternativeapproachistoincludethe19kineticsofelementaryreactionsinsteadoftakingaglobaldatamatchontheconditionthattheadequateelementaryreactionsareknown.
6.
0ACKNOWLEDGMENTThisstudywaspartiallysupportedbytheLouisianaGovernor'sEnergyInitiativeviatheCleanPowerandEnergyResearchConsortium(CPERC)andwasadministeredbytheLouisianaBoardofRegentsandpartiallysupportedbytheU.
S.
DepartmentofEnergy.
7.
0REFERENCESBenyonP.
,2002,"ComputationalModellingofEntrainedFlowSlaggingGasifiers,"PhDthesis,SchoolofAerospace,Mechanical&MechatronicEngineering,UniversityofSydney,Australia,2002.
Bockelie,M.
J.
,Denison,K.
K.
,Chen,Z.
,Linjewile,T.
,Senior,C.
L.
,andSarofim,A.
F.
,2002,"CFDModelingforEntrainedFlowGasifiersinVision21Systems,"Proceedingsofthe19thAnnualInternationalPittsburghCoalConference,Pittsburgh,PA.
September24-26,2002Chen,C.
,Horio,M.
,andKojima,T.
,2000,"NumericalSimulationofEntrainedFlowCoalGasifiers.
PartI:ModelingofCoalGasificationinanEntrainedFlowGasifier",ChemicalEngineeringScience,Vol.
55,3861-38740.
Chen,C.
,Horio,M.
,andKojima,T.
,2010,"NumericalSimulationofEntrainedFlowCoalGasifiers,"ChemicalEngineeringScience,Vol.
55,pp.
3861-3883Chen,W.
H.
,Hsieh,T.
C.
,andJiang,T.
L.
,2008,"AnExperimentalStudyonCarbonMonoxideConversionandHydrogenGenerationfromWaterGasShiftReaction,"EnergyConversionandManagement,Vol.
49,pp.
2801-2808Choi,Y.
,andStenger,H.
G.
,2003,"WaterGasShiftReactionKineticsandReactorModelingforFuelCellGradeHydrogen,"JournalofPowerSources,Vol.
124,pp.
432–439Fletcher,T.
H.
,Kerstein,A.
R.
,Pugmire,R.
J.
,andGrant,D.
M.
,1990,"ChemicalPercolationModelforDevolatilization:2.
TemperatureandHeatingRateEffectsonProductYields,"EnergyandFuelsVol.
4,pp.
54-60Fletcher,T.
H.
,andKerstein,A.
R.
,1992,"ChemicalPercolationModelforDevolatilization:3DirectUseof13CNMRDatetoPredictEffectsofCoalType,"EnergyandFuels,Vol.
6,pp.
414-431Grant,D.
M.
,Pugmire,R.
J.
,Fletcher,T.
H.
,andKerstein,A.
R.
,1989,"ChemicalPercolationofCoalDevolatilizationUsingPercolationLatticeStatistics,"EnergyandFuels,Vol.
3pp.
175-186Grenoble,D.
C.
andEstadt,M.
M,1981,"TheChemistryandCatalysisoftheWaterGasShiftReaction1,theKineticsoverSupportedMetalCatalysts,"JournalofCatalysis,Vol.
67,pp.
90-10220Gururajan,V.
S.
,Agarwal,P.
K.
,andAgnew,J.
B.
,1992,"Mathematicalmodellingoffluidizedbedcoalgasifiers,"Chemicalengineeringresearch&design,vol.
70,No.
A3,p.
p.
211-238,ISSN:0263-8762Hirth,T.
,andFranck,E.
U.
,1991,"OxidationandHydrothermolysisofHydrocarbonsinSupercriticalWateratHighPressures,"Ber.
Bunsenges.
Phys.
Chem.
,Vol.
97,pp.
1091-1097Hla,S.
S.
,Park,D.
,Duffy,G.
J.
,Edwards,J.
H.
,Roberts,D.
G.
,Ilyushechkin,A.
,Morpeth,L.
D.
,andNguyen,T.
,2009,"KineticsofHigh-temperatureWater-GasShiftReactionoverTwoIron-basedCommercialCatalystsUsingSimulatedCoalderivedSyngases,"ChemicalEngineeringJournal,Vol.
146,pp.
148–154Holgate,H.
R.
,Webley,P.
A.
,andTester,J.
W.
,1992,"CarbonMonoxideOxidationinSupercriticalWater:theEffectsofHeatTransferandtheWater-Gas-ShiftReactiononObservedKinetics,"Energy&Fuels,Vol.
6,pp.
586-597Hughes,R.
,Lu,D.
,Majeski,A.
,Corber,A.
,andAnthony,B.
,2010,"EntrainedFlowSlaggingSlurryGasificationandtheDevelopmentofComputationalFluidDynamicsModelsatCanmetENERGY",2010InternationalPittsburghCoalConference,Istanbul,TurkeyOctober11–14,2010.
Jones,W.
P.
,andLindstedt,R.
P.
,1998,"GlobalReactionSchemesforHydrocarbonCombustion,"CombustionandFlame,Vol.
73,pp.
233Keiski,R.
L.
,Salmi,T.
,Niemisto,P.
,Ainassaari,J.
,andPohjola,V.
J.
,1996,"StationaryandTransientKineticsoftheHighTemperatureWaterGasShiftReaction,"AppliedCatalysisA:General,Vol.
137,pp.
349-370Kobayashi,H.
,Howard,J.
B.
,andSarofim,A.
F.
,1976,"CoalDevolatilizationatHighTemperatures,"16thSymp.
(Int'l.
)onCombustion,TheCombustionInstituteKoryabkina,N.
A.
,Phatak,A.
A.
,Ruettinger,W.
F.
,Farrauto,R.
J.
,andRibeiro,F.
H.
,2003,"DeterminationofKineticParametersfortheWater–GasShiftReactiononCopperCatalystsunderRealisticConditionsforFuelCellApplications,"JournalofCatalysis,Vol.
217,pp.
233-239Kusar,H.
,Hocevar,S.
,andLevec,J.
,2006,"KineticsoftheWater-Gas-ShiftReactionoverNanostructuredCopper–CeriaCatalysts,"AppliedCatalysisB:Environmental,Vol.
63,pp.
194–200.
Ranz,W.
E.
,andMarshall,W.
R.
,1952,"EvaporationfromDropsPart1andPart2,"ChemicalEngineeringProgress,Vol.
48,pp.
173-180Rase,H.
F.
,ChemicalReactorDesignforProcessPlants:Volumetwo–CaseStudiesandDesignData,JohnWileyandSons,1977.
Rhodes,C.
,Hutchings,G.
J.
,andWard,A.
M.
,1995,"WGSR:FindingtheMechanisticBoundary,"CatalysisToday,Vol.
23,pp.
43-48.
21Rhodes,C.
,andHutchings,G.
J.
,2003,"StudiesoftheRoleoftheCopperPromoterintheIronOxide/ChromiaHighTemperatureWaterGasShiftCatalyst,"PhysicalChemistryChemicalPhysics,Vol.
5,pp.
2719–2723Salmi,T.
,andHakkarainen,R.
,1989,"KineticStudyoftheLow-TemperatureWater-Gas-ShiftReactionoveraCu-ZnOCatalyst,"AppliedCatalysis,Vol.
49,pp.
285-306Sato,T.
,Kurosawa,S.
,SmithJr.
,R.
L.
,Adschiri,T.
,andArai,K.
,2004,"WaterGasShiftReactionKineticsunderNoncatalyticConditionsinSupercriticalWater,"JournalofSupercriticalFluids,Vol.
29,pp.
113-119,2004.
Silaen,A.
,andWang,T.
,2005,"SimulationofCoalGasificationProcessInsideaTwo-StageGasifier,"Paper20-1,Proceedingsofthe22ndInternationalPittsburghCoalConference,Pittsburgh,Pennsylvania,September19-22.
Silaen,A.
,andWang,T.
,2006,"EffectofFuelInjectionAnglesonPerformanceofaTwo-StageGasifier,"Paper20-4,presentedatthe23rdInternationalPittsburghCoal-GenConference,Pittsburgh,Pennsylvania,Sept.
25-28,2006.
Silaen,A.
,andWang,T.
,2009,"ComparisonofInstantaneous,EquilibriumandFiniteRateGasificationModelsinanEntrainedFlowCoalGasifier,"Paper14-3,presentedatthe26thInternationalPittsburghCoalConference,Pittsburgh,Pennsylvania,Sept.
21-24,2009.
Silaen,A.
,andWang,T.
,2010,"EffectofTurbulenceandDevolatilizationModelsonGasificationSimulation,"InternationalJournalofHeatandMassTransfer,Vol.
53,pp.
2074-2091,2010.
Silaen,A.
,andWang,T.
,2011,"InvestigationofCoalGasificationProcessunderVariousOperatingConditionsInsideaTwo-StageEntrainedFlowGasifier",acceptedforpublicationinASMEJournalofThermalScienceandEngineeringApplications,2011.
Smith,B.
,Loganathan,M.
,andShanatha,M.
S.
,2010,"AReviewoftheWaterGasShiftReactionKinetics,"InternationalJournalofChemicalReactorEngineering,Vol.
8.
Smoot,D.
L.
,andSmith,P.
J.
,1985,"CoalCombustionandGasification,"PlenumPressSundararajan,A.
T.
,andShet,U.
S.
P.
,2009,"NumericalModelingofaSteam-AssistedTubularCoalGasifier,"InternationalJournalofThermalSciences,Vol.
48,pp.
308-321.
TampaElectricPolkPowerStationIntegratedGasificationCombinedCycleProject,FinalTechnicalReport,August2002.
Twigg,M.
V.
,1989,CatalystHandbook,secondedition,WolfePublishingLtd.
WabashRiverCoalGasificationRepoweringProject,FinalTechnicalReport,August2000.
22Waind,T.
,andWhitty,K.
,2010,"CharacterizationofaSmallScaleSlurry-Fed,Oxygen-BlownEntrainedFlowGasifier:HowInjectorGeometryAffectsFlameStabilityandPerformance,"2010InternationalPittsburghCoalConference,Istanbul,TurkeyOctober11–14,2010.
Wang,T.
,Silaen,A.
,Hsu,H.
W.
,andLo,M.
C.
,2006,"PartialLoadSimulationandExperimentsofaSmallCoalGasifier,"Paper20-3,presentedatthe23rdInternationalPittsburghCoal-GenConference,Pittsburgh,Pennsylvania,Sept.
25-28,2006.
Wang,T.
,Silaen,A.
,Hsu,H.
W.
,andShen,C.
H.
,2007,"EffectofSlagTapSizeonGasificationPerformanceandHeatLossesinaQuench-typeCoal,"Paper37-4,presentedatthe24thInternationalPittsburghCoal-GenConference,Johannesburg,SouthAfrica,Sept.
10-14,2007.
Wang,T.
,Silaen,A.
,Hsu,H.
W.
,andShen,C.
H.
,2010,"InvestigationofHeatTransferandGasificationofTwoDifferentFuelInjectorsinanEntrained-FlowGasifier,"ASMEJournalofThermalScienceandEngineeringApplications,Vol.
2,pp.
011001/1-10,March2010.
Wang,T.
,Silaen,A.
,Hsu,H.
W.
,andShen,C.
H.
,2011,"TopFuelInjectionDesigninanEntrained-FlowCoalGasifierGuidedbyNumericalSimulations,"ASMEJournalofThermalScienceandEngineeringApplications,Vol.
3,pp.
011009/1-8,March2011.
Watanabe,H.
,andOtaka,M.
,2006,"NumericalSimulationofCoalGasificationinEntrainedFlowCoalGasifier,"Fuel,Vol.
85,pp.
1935-1943,2006.
Watanabe,M.
,Mochiduki,M.
,Sawamoto,S.
,Adschiri,T.
,andArai,K.
,2001,"PartialOxidationofN-HexadecaneandPolyethyleneinSupercriticalWater,"J.
Supercrit.
Fluids,Vol.
20,pp.
257Westbrook,C.
K.
,andDryer,F.
L.
,1981,"SimplifiedReactionMechanismsfortheOxidationofHydrocarbonFuelsinFlames,"Vol.
27,pp.
31-43Wheeler,C.
,Jhalani,A.
,Klein,E.
J.
,Tummala,S.
,andSchmidt,L.
D.
,2004,"TheWater-Gas-ShiftReactionatShortContactTimes,"JournalofCatalysis,Vol.
223,pp.
191–199