methane reforming demonstration problem

Upload: jcmval

Post on 05-Apr-2018

229 views

Category:

Documents


1 download

TRANSCRIPT

  • 7/31/2019 Methane Reforming Demonstration Problem

    1/19

    Methane ReformingDemonstration Problem

    Revision0;October14 ,2008 Reference: WangShuyanet.al.,Simulationofeffectofcatalyticparticleclusteringonmethane

    steamreforminginacirculatingfluidizedbedreformer,ChemicalEngineeringJournal139(2008)136

    146.

    PurposeandOverview

    Steamreformingofmethane(conversionofCH4intosyngas:COandH2)isofinterestforapplications

    suchasfuelcells,conversionofnaturalgastoliquidfuels,etc. Similarreactorsareofinterestforcoal

    toliquid(CTL)andbiofuelsynthesisapplications.

    A

    simple

    demonstration

    problem

    has

    been

    developed

    in

    SINDA/FLUINT,

    using

    both

    the

    Sinaps

    nongeometric(sketchpad)GUIandtheThermalDesktopwithFloCADgeometric(CADbased)GUI.

    Sincebothsetsofmodelsareavailableforinspectionandforuseasastartingpointortemplate,only

    briefdescriptionsareincludedinthisdocumentasgeneralguidance.Abasicunderstandingof

    SINDA/FLUINTmodelingisassumed.

    ReactionKinetics

    Twosimultaneousreactionsoffivespeciesareconsidered:

    CH4+H2O CO+3H2 (reaction1)

    CO+H2O CO2+H2 (reaction2)

    Reaction2isthewatergasshift(WGS)reaction.

    Thereferencedpaper(Shuyan,et.al.)containsthereactionkineticsusedinthisproblem,basedonthe

    presenceofHaldorTopsoeNi/MgAl2O4spinelcalatyticparticles.Becauseofthedemonstrativenature

    ofthisproblem,thedetailsofthecatalystareneglected:thecatalystisassumedtooperateatfull

    activity.

    Shuyanlistsformulaeforforwardreactionratesforthesereactions,1usingequilibriumconstantsasthe

    basisforestimatingreverserates.Assumingfullcatalyticactivity,thecurrentreactionratecanbe

    calculatedasafunctionoftemperature,pressure,andpartialpressures.Thetemperaturedependenciesforboththereactionrateconstantsandtheequilibriumconstantsareexponentialfunctionsofan

    Arrheniusform.

    1 Specifically,equations8,9,11,12,13,and1520.

  • 7/31/2019 Methane Reforming Demonstration Problem

    2/19

    NotesonFluidProperties

    InSINDA/FLUINT,eachspeciesisassignedaletteridentifier:

    Species Symbol LetterIDMethane CH4 M

    Water H2O WCarbonmonoxide(gas) CO G

    Hydrogen H2 H

    Carbondioxide CO2 C

    Watermaybeacondensable(twophase)fluid.However,inthiscasethetemperatureswillalwaysbe

    highenoughsuchthattheliquidphasewillneveroccur.2Therefore,eitheraperfectgas(8000series

    fluid)orarealgas(NEVERLIQ6000seriesfluid)couldhavebeenused.Versionsofsuchfilesare

    available(http://www.crtech.com/properties.html)thatwerebuiltfromtheNISTprogramREFPROP,

    perhapssubsequentlysimplifiedtoaperfectgasusingtheSINDA/FLUINTPR8000utility.

    Unfortunately,theNISTdatabasedoesnotextendtosufficientlyhightemperaturesforCO,CH4,andH2.

    Forexample,theuppertemperaturelimitforCOisonly500KinthecurrentversionofREFPROP.

    Therefore,propertiesforallfivegasesweregeneratedusingNASAsfreeCEAchemicalequilibrium

    programcombinedwithaC&RutilityforconvertingoutputstoFPROPDATAformat.3 Whilethis

    interfaceisintendedtopreparefluidsrepresentingequilibriumreactingmixtureswithvariable

    molecularweights,theonlymodeinCEAcanbeusedtoproducepropertiesforsingleconstituents

    withconstantmolecularweights.Therefore,despitetheresultinggeneralized6000seriesrealgas

    fluidtables,itshouldbenotedthattheactualpropertiesreflectCEAsintrinsicassumptionofperfect

    gases. Heatofformation(HFORM)anddiffusionvolume(DIFV)informationwerethenaddedtothese

    fluidfiles,notingthatCEAusestheheatofformationasthebasisoftheenthalpyat25C.Theheatofformationofwateristhereforediminishedbytheheatofvaporizationtoreflecttheallgasnatureof

    thisCEAderivedfluidfile.(Forafulltwophasewaterdescription,theuncorrectedheatofformation

    shouldinsteadbeapplied,sincealiquidstateexistsatstandardtemperatureandpressure.)

    ReactorDesign

    Anintentionallygenericandsimplifiedplugflowreactor(PFR)isassumed.

    Inclusionofthesolidcatalystparticlescouldtakeseveralformsasapackedbedofstationaryparticles

    heldinplacebystructuredcatalystssupports,freeflowingwiththegastransport,orother

    configurationsbutinclusionofthecatalystsinanyformovershadowstheillustrativeintentofthis

    demonstrationproblem.Thechosenscenariofocusesonhowtoincorporatethereactionkineticsinto

    SINDA/FLUINT.

    2 Iftemperaturesbelowthedewpointareincluded,afulltwophasedescriptionofwatershouldbeusedinstead.

    Indeed,suchadescription(http://www.crtech.com/properties.html)wasusedaspartofthevalidationeffort

    whentemperaturesaslowas25Cwereusedasafeedtemperature.3 http://www.crtech.com/EQfluids.html

  • 7/31/2019 Methane Reforming Demonstration Problem

    3/19

    Temperatures,pressures,inletflows,anddimensionsmaybechangedparametricallyinthemodel,with

    asubsetofpossibleparametricsettingsusedforthebaselinerunsshowninthetablebelow.

    Parameter RegisterName Value Units Comment

    Reactor

    length

    length

    1

    m

    (Arbitrary

    choice.)

    Longer

    lengths

    mightbeconsideredasneededto

    bringtheoutletconditionscloser

    totheequilibriumcondition

    Reactorhydraulicdiameter diam 0.008 m (Arbitrarychoice)

    Reactorflowarea area 4.e4 m2 Undefinedfinsorslitshapeis

    assumed,asneededtoprovide

    adequateheattransferarea

    Inlettemperature temp_in 800 C (Arbitrarychoice,but

    representativeoftypical

    applications.)

    Walltemperature Assumed

    equaltothe

    inlet

    800 C Inletandwalltemperaturesare

    assumedtobeequal,butcanbe

    independentlyadjustedifneedbe

    Inletflowrate moles_in 0.03e3 kmol/s (Arbitrarychoice)

    InletmolefractionsofCH4,H20

    andotherinletstream

    constituents,plusexcesssteam

    fractionbeyondstoichiometric

    mol_in_m,

    mol_in_w,

    SteamX

    0.5,

    0.5,

    3.0

    Stoichiometric(equalmole

    fractions)plus2timesmore

    steam.UseSteamX=1.0for

    stoichiometricfeedrates.

    Outletpressure pres 101325. Pa Thisindirectlysetsthereactor

    pressureduetothesmall

    pressuredropthroughthereactor

    ARefresher:ModelingReactionsinSINDA/FLUINT

    TherateofcreationofmassofanyspeciesiisdesignatedXMDOTiinSINDA/FLUINT,withthecaveat

    thatiXMDOTi=0:thereisnonetgenerationorlossoftotalmassasaresultofachemicalreaction.A

    speciesthatisbeingconsumedwillhaveanegativeXMDOT,andaspeciesthatisbeingproducedwill

    haveapositiveXMDOT.Sincethismodelusesmetricunits,XMDOThasunitsofkg/s(lbm/hrwouldbe

    theunitsintheEnglishsystem).

    WhenXMDOTsarenonzero,thecodewillautomaticallycalculateandapplyacorrectiveheatofreaction

    (QCHEM)basedonthefluidproperties(heatsofformationandenthalpybasis).Ifeachfluidusesthe

    heatofformationastheenthalpybasisatSTP,thiscorrectiveQCHEMtermiszero.

    Whenthedesiredconcentrationofoneormorespeciesisknown(e.g.,wheneithertheequilibrium

    concentrationorthereaction/combustionefficiencyisknown),theEQRATEutilitycanbeusedto

    generateXMDOTvalues,perhapsobeyingoneormorechemicalreactions(suppliedasstoichiometric

    numbers). Forexample,preliminaryCEArunscouldhavegeneratedequilibriumstatepredictions

    (perhapsasfunctionsoftemperature),whichcouldhavebeenusedasinputstotheEQRATEutility.

  • 7/31/2019 Methane Reforming Demonstration Problem

    4/19

    Ifinsteadonlytherateofreactionisknown(asisthecaseinthisexample),thenEQRATEisnot

    applicableandtheXMDOTvaluesshouldinsteadbecalculatedexplicitlybytheuserinuserlogicblock

    FLOGIC0atthestartofeachsolutioninterval(i.e.,timesteporsteadystaterelaxationstep).The

    XMDOTvalueforeachspeciesiwillbethesumofallthereactionsinwhichitparticipates.

    TheresultingsetofXMDOTvaluesisassumedtobeconstant(alongwiththecorrespondingQCHEM)overthenextsolutioninterval.Inotherwords,thespeciesproductionandextinctionratesarebasedon

    thetemperaturesandpressuresatthestartoftheinterval,andarenotimplicitlyadjustedfor

    temperatureorpressurechangesthatareabouttooccur.4

    Asignificantrestrictiononchemicalreactionsisthattheycanonlyapplytotanks(controlvolumes),and

    nottojunctions(massless/volumelessstatepoints).NotingthatthemainsolutionroutineSTEADY(aka

    FASTIC)treatstankstemporarilyasjunctions,thismeansthatsteadystatesolutionscanonlybe

    achievedusingSTDSTL(pseudotransientintegrationtowardatimeindependentstate).Thisinturn

    meansthat,unlikemoststeadystateanalyses,chemicalreactionsteadystatesolutionsaredependent

    oninitialguesses(especiallyspeciesfractions).

    ReactorModel

    Thesourceofgasesisrepresentedbytwoplena:one(reactor.1000)representingastoichiometric

    mixtureofmethaneandsteam,andanother(reactor.1100)representingexcesssteamthatcanbe

    addedtothereactor.Themixtureratiowithinplenumreactor.1000issetusingregisters(e.g.,mol_in_w

    forsteam,mol_in_mformethane)basedonmolefractions,whicharethenconvertedintogasmass

    fractions(XGiforeachspeciesiasrepresentedbytheregisterxin_m,xin_w,xin_h,xin_c,andxin_g.

    Notethatwhilexin_candxin_garebothzero(noinflowingCO2orCO),xin_hisnonzero:itissettoa

    verysmallvalue.Anextremelysmallamountofhydrogenisaddedsimplytoavoidtheneedtodealwith

    themathematicalsingularitycausedbythepresenceofthepartialpressureofhydrogeninthe

    denominatoroftherateequations.

    Therearemanywaystorepresenttheinletconditions.Forexample,themassfractionsonasingle

    plenumcouldbeset.Alternatively,NplenacouldbeusedforeachoftheNspeciesinvolved,settingthe

    massorvolumetricflowrateofeachspeciesbeinginjected.

    Inthiscase,asingleSetFlow(SINDA/FLUINTMFRSETconnector)isappliedfromeachplenum,withthe

    ratiooftheflowratesbeingcontrolledviatheSteamXregister:SteamX=1forstoichiometricflow,2for

    twicethesteamrequired,etc.Thesteadystatecase(namedPFR)isrunatSteamX=1:stoichiometric

    inletconditions,orequalmolarflowsofmethaneandsteam,withzeroexcesssteamflowaddedfrom

    plenumreactor.1100.

    4 Bydefault,SINDA/FLUINTdoesimplicitlyreducereactionsforreactantsthatarepresentinlowconcentrations

    andthatvanishduringthesolutioninterval,asdescribedintheUsersManual.Inotherwords,XMDOTscanbean

    implicitfunctionofconcentration,buttheyareassumedindependentoftemperatureandpressureduringeach

    solutioninterval.Normally,theyareadjustedinastairstepfashionbetweensolutionintervalsaccordingto

    calculationsinuserlogic.

  • 7/31/2019 Methane Reforming Demonstration Problem

    5/19

    A10segmentSinapsorFloCADPipe(SINDA/FLUINTHXmacropluswallmodel)isusedtorepresentthe

    reactor.Whileincludingthedetailsoftheheatexchange(structural/thermalmodel)isastrengthofthe

    C&Rtoolsuite,toavoiddistractingfromtheintentofthisdemonstrationproblem,aninfinitesourceof

    energyat800Cisassumed.ThenodesinthePipearethereforechosentobeconstanttemperature

    boundarynodes.

    Thechoiceof10segmentscanbeadjustedbychangingboththeintegerregisterresolpluseditingthe

    Pipemacro.5Higherresolutions(atleastresol=20)areneededtocapturethestronggradientsatthe

    inletandtoguaranteethatequilibriumisachievedattheexit.6However,fordemonstrationpurposes

    thespatialresolutionisleftintentionallylow.TheSinapsdiagramforthereactorisshowbelow.

    ThecorrespondingFloCADdiagramisasfollows:

    TanksmustbeusedwithinthePipetoallowspeciessource/sinkrates(XMDOTs)tobespecified:

    reactionsarenotpossibleinjunctions.NotethatSTUBEconnectors(inertialessductelements)areused

    insteadoftubessincetheflowsconsistoflowdensitygases.Eventhoughthesubsequenttransient

    operatesoveranextremelyshorttimescale,tubeswouldnotberequiredunlesssignificant

    condensationweretooccur.

    TheSTDSTLsolutionmustbeusedtoarriveatasteadystateanswersincechemicalreactionsarenot

    possibleinSTEADY,whichtreatstanksasiftheywerejunctions.NormallyaSTEADYsteadystate

    solutionisnotverydependentoninitialconditions,andoftenonlyrequires1050iterationstoresolve.

    However,anSTDSTLsolutionusesthefulltransienthydrodynamicsolutionalongwithasimultaneous

    5 Therestartrecordnumber(integerregisterrecord)willalsohavetobeadjustedbasedontheoutputsofthefirst

    case(pfr)sincethesizeoftheSAVEandRSI/RSOfileswillchangeifthemodelsizechanges.6 Assumingtemperaturesarehighenoughtoensureequilibrium.Also,thereactorlengthmayneedtobe

    increasedaswell(thiscanbechangedindependentlyofthenumberofsegmentsusedtoresolvethatlength).

  • 7/31/2019 Methane Reforming Demonstration Problem

    6/19

    thermalsolution.STDSTLissimilartorunningatransientsolutionuntileverysystemresponseis

    constant.Itthereforeissensitivetoinitialconditions(assetviathemolefractionregistersminit_h,

    minit_g,etc.whichsetmassfractionsxinit_h,xinit_g,etc.).Thismeansthatsomespeedsavingscanbe

    realizedbysettingreasonableinitialconditions,perhapsbasedonpriorruns.Otherwise,poorlyguessed

    initialconditionsinvoketheevaluationofasevereyetspurioustransient,wastingcomputational

    resourcesandrequiringmorestepNum(aregisterusedtosetthecontrolconstantNLOOPS)to

    converge.

    Normally,steadystatesolutionsaresoinexpensivethattheyareoftenrecalculatedbeforeeach

    transientcaseasinitialconditions.WhilethespeedoftheSTDSTLsolutionisgoodinthiscase,amore

    detailed(perhaps2Dor3D)modelmightnotenjoythesameresult.Therefore,theabilitytosavethe

    resultsofapriorsteadystatecase(casepfr)andreusetheminafuturetransientcase(case

    pfrTransient)isdemonstratedinthismodel(seeRESAVE/RESTARoperationsintheSINDA/FLUINT

    manual).BothcasescanbeselectedintheSinapsCaseManagerortheThermalDesktopCaseSet

    Managerandexecutedsequentially.Ifthisusagewerefrequent,thenthetwosolutionsshouldbe

    combinedintoasinglecasetoavoidtheinefficienciesoftheextrapreprocessandcompilesteps.

    Foreachtankinthesystem,atthestartofeachsolutionstep(namely,inFLOGIC0),thefollowing

    calculationsshouldbemade:

    1. Theratefactors(equilibriumKconstantsandpreexponentialkterms)shouldbeupdatedasafunctionofcurrenttemperature(TL),pressure(PL),andpartialpressures(e.g.,PPGHisthe

    partialpressurefractionforspeciesH).

    Thetemperaturesandpressuresshouldbeconvertedintoabsoluteunits(e.g.degreesK

    insteadofC),whichmeanssubtractingABSZROfromtemperaturesandPATMOSfrom

    pressures.Forexample:TLabs=TLrelABSZRO.Inthelogicblocks,theregisteratemp

    containstheabsolutetemperatureandaprescontainstheabsolutepressure.

    2. Themolarrateforreaction1shouldbecalculated,andusedtosettheXMDOTforeachspeciesinthatequationusingthestoichiometricnumber,notingthatXMDOThasunitsof

    kg/s(orlbm/hrforUID=ENG)andthereforethemolecularweightofeachspeciesmustbe

    usedasamultiplier.ThesemolecularweightscanbecalculatedviaroutinessuchasVMOLW

    orVMOLWPT,butinthisdemonstrationproblemtheyhavebeensetusingregisters(e.g.,

    mw_h2o,mw_co2,etc.).Ifaspeciesisnotparticipatinginreaction1(e.g.,CO2inthiscase),

    itsXMDOTshouldbezeroed.

    3. Themolarrateforreaction2shouldbecalculated,andusedtoupdatetheXMDOTforeachspeciesinthatequation.SinceXMDOTshavealreadybeenupdatedforreaction1,thisstep

    involvessummingintothesevalues,whichexplainswhyXMDOTforCO2(XMDOTC)was

    zeroedinthepriorstep.

    Iftherewereany3rdor4threactions,thesecouldsimilarlyhavebeensummedintothe

    XMDOTvalues.Toavoidanyconfusionwhenthenumberofreactionsislarge,theXMDOTs

    shouldbezeroedatthestartofthelogicblock,thenalwayssummedinto(ratherthan

    replaced)soasnottooverwritetheeffectsofreactionscalculatedearlierinthesequence.

  • 7/31/2019 Methane Reforming Demonstration Problem

    7/19

    Therearetwoplaceswheresuchreactionupdatelogiccanbeplaced:

    1. HEADERSUBROUTINES:userdefinedsubroutines,whichcanthenbecalledfromFLOGIC0viaaDOloopincrementingloopID.SincetheIDofeachtankwillchangeforeachcalltothe

    updateroutine,dynamictranslationroutines(e.g.,INTLMP,INTSPE)willberequiredto

    converttheuseridentifierintotheinternalarraylocation.2. Networklogic:7logicassociatedwitheachnetworkelement(tanks,inthiscase).Expression

    styleindirectreferencingcanbeusedintheseblocks.Forexample,VOL#thisreferstothe

    volumeofthecurrenttank,PPGH#thisreferstothepartialpressurefractionofhydrogen

    (speciesH)inthecurrenttank,andXMDOTH#thisreferstotheXMDOTofthesamespecies

    H.

    FortheSUBROUTINESapproach(whichisusedinthemodelnotbecauseitisrecommendedbutbecause

    itismorewidelyaccessibletoolderversionsandexperiencedusers),theFLOGIC0blockappearsas:

    do itest = 1,resol

    call reformenddo

    whereitest,thevalueiteratedintheDOloop,referstotheuserlumpID(from1toresol=10).Itestisa

    globalvaluethatisaccessiblewithintheusersubroutineREFORMaspartoftheCALLCOMMON

    commandinthatroutine(whichisinsertedasafile):

    subroutine reform

    call common

    fstart

    integer methane,water,co,co2,h2,lump

    c

    c reactions 1 and 2, from "Simulation of effect of catalytic

    c particle clustering on methane steam reforming in ac circulating fluidized bed reformer" Shuyan et al,

    c Chem Eng Jnl 139 (2008) p.136-146

    c

    c H=H2, W=H20, G=CO, M=CH4, C=CO2

    c

    c itest is assigned in FLOGIC

    lump = intlmp('reactor',itest)

    methane = intspe('reactor','m')

    co2 = intspe('reactor','c')

    co = intspe('reactor','g')

    h2 = intspe('reactor','h')

    water = intspe('reactor','w')

    c

    if(ppg(h2,lump) .le. 0.0) call abnorm(' reform ',itest,

    & ' User routine REFORM cannot handle zero H2 partial pressure ')

    c convert to absolute units (if not)

    atemp = tl(lump)-abszro

    apres = pl(lump)-patmos

    c 1/Pa, except for KH20 which is unitless

    KCH4 = 6.65e-9*EXP(4604.28/atemp)

    7 AvailableinSinapsandFloCADVersions5.2andlater.

  • 7/31/2019 Methane Reforming Demonstration Problem

    8/19

    KH2 = 6.12e-14*EXP(9971.13/atemp)

    KCO = 8.23e-10*EXP(8497.71/atemp)

    KH2O = 1.77e5*EXP(-10666.35/atemp)

    DEN = 1.0 + apres*(KCH4*PPG(methane,lump)

    & + KH2*PPG(h2,lump) + KCO*PPG(co,lump))

    & + KH2O*PPG(water,lump)/PPG(h2,lump)

    c reaction 1

    c mol-sqrt(Pa)/kg-s

    klc1 = 8.336e17*EXP(-28879./atemp)

    c Pa^2

    Kuc1 = 10266.76e6*EXP(-26830./atemp+30.11)

    rxn1 = klc1*(ppg(methane,lump)*ppg(water,lump)/ppg(h2,lump)**2.5

    & - apres**2*ppg(co,lump)*sqrt(ppg(h2,lump))/Kuc1)

    & /sqrt(apres)/DEN**2

    rate1 = rxn1*dl(lump)*vol(lump)

    rate1 = (1.0-rateLag)*rate1 + rateLag*cx(lump)

    xmdot(methane,lump) = -rate1*mw_ch4

    xmdot(water,lump) = -rate1*mw_h2o

    xmdot(co,lump) = rate1*mw_co

    xmdot(h2,lump) = 3.0*rate1*mw_h2

    xmdot(co2,lump) = 0.0

    c reaction 2

    c mol/kg-Pa-s

    klc2 = 12.19*EXP(-8074.3/atemp)

    c unitless

    Kuc2 = EXP(4400.0/atemp-4.063)

    rxn2 = klc2*(ppg(co,lump)*ppg(water,lump)/ppg(h2,lump)

    & - ppg(co2,lump)/Kuc2)*apres/DEN**2

    rate2 = rxn2*dl(lump)*vol(lump)

    rate2 = (1.0-rateLag)*rate2 + rateLag*cy(lump)

    xmdot(co,lump) = xmdot(co,lump) - rate2*mw_co

    xmdot(water,lump) = xmdot(water,lump) - rate2*mw_h2o

    xmdot(co2,lump) = xmdot(co2,lump) + rate2*mw_co2

    xmdot(h2,lump) = xmdot(h2,lump) + rate2*mw_h2

    c store key data away in otherwise unused CX, CY, CZ cells

    c for ease of postprocessing, damping

    c

    cx(lump) = rate1

    cy(lump) = rate2

    c a little out of step (uses last QCHEM), and QCHEM is for both reactions,

    c but ...

    if(rate1 .ne. 0.0) cz(lump) = 1.e-6*qchem(lump)/rate1c

    return

    end

    fstop

    Theequivalenttotheabove,usingthenetworklogicapproach,istoplacethefollowingintheFLOGIC0

    formforalltanks(whichcanbeeditedsimultaneously):

  • 7/31/2019 Methane Reforming Demonstration Problem

    9/19

    c

    c reactions 1 and 2, from "Simulation of effect of catalytic

    c particle clustering on methane steam reforming in a

    c circulating fluidized bed reformer" Shuyan et al,

    c Chem Eng Jnl 139 (2008) p.136-146

    c

    c H=H2, W=H20, G=CO, M=CH4, C=CO2

    c

    if(ppgh#this .le. 0.0) call abnorm(' reform ',#this,

    & ' User logic cannot handle zero H2 partial pressure ')

    c convert to absolute units (if not)

    atemp = tl#this-abszro

    apres = pl#this-patmos

    c 1/Pa, except for KH20 which is unitless

    KCH4 = 6.65e-9*EXP(4604.28/atemp)

    KH2 = 6.12e-14*EXP(9971.13/atemp)

    KCO = 8.23e-10*EXP(8497.71/atemp)

    KH2O = 1.77e5*EXP(-10666.35/atemp)

    DEN = 1.0 + apres*(KCH4*PPGM#this + KH2*PPGH#this + KCO*PPGG#this)

    & + KH2O*PPGW#this/PPGH#this

    c reaction 1

    c mol-sqrt(Pa)/kg-s

    klc1 = 8.336e17*EXP(-28879./atemp)

    c Pa^2

    Kuc1 = 10266.76e6*EXP(-26830./atemp+30.11)

    rxn1 = klc1*(ppgm#this*ppgw#this/ppgh#this**2.5

    & - apres**2*ppgg#this*sqrt(ppgh#this)/Kuc1)

    & /sqrt(apres)/DEN**2

    rate1 = rxn1*dl#this*vol#this

    rate1 = (1.0-rateLag)*rate1 + rateLag*cx#this

    xmdotm#this = -rate1*mw_ch4

    xmdotw#this = -rate1*mw_h2o

    xmdotg#this = rate1*mw_co

    xmdoth#this = 3.0*rate1*mw_h2

    xmdotc#this = 0.0

    c reaction 2

    c mol/kg-Pa-s

    klc2 = 12.19*EXP(-8074.3/atemp)

    c unitless

    Kuc2 = EXP(4400.0/atemp-4.063)

    rxn2 = klc2*(ppgg#thisppgw#this)/ppgh#this

    & - ppgc#this/Kuc2)*apres/DEN**2

    rate2 = rxn2*dl#this*vol#this

    rate2 = (1.0-rateLag)*rate2 + rateLag*cy#this

    xmdotg#this = xmdotg#this - rate2*mw_co

    xmdotw#this = xmdotw#this - rate2*mw_h2o

    xmdotc#this = xmdotc#this + rate2*mw_co2

    xmdoth#this = xmdoth#this + rate2*mw_h2

    c store key data away in otherwise unused CX, CY, CZ cells for ease of postprocessing,

    damping

  • 7/31/2019 Methane Reforming Demonstration Problem

    10/19

    c

    cx#this = rate1

    cy#this = rate2

    c a little out of step (uses last QCHEM), and QCHEM is for both reactions,

    c but ...

    if(rate1 .ne. 0.0) cz#this = 1.e-6*qchem#this/rate1

    TheaboveisthebasisoftheSinapsmodelpfr_nl.smdlandtheFloCADmodelpfr_nl.dwg,wherethenlinthenamesreferstotheNetworkLogicoption.

    Thedifferencetothetwoapproachesislargelyamatterofuserpreference.Networklogicis

    expanded/translatedbySinapsorFloCADandinsertedintothespecifiedlocation(FLOGIC0inthiscase).

    Notethatthekeydifferenceishowtoreferenceeachproperty:

    dl(lump)inHEADERSUBROUTINES,wherelumphasbeencalculatedusingthedynamic

    translationfunctionINTLMP,isequivalenttodl#thisinnetworklogic.

    ppg(methane,lump)inHEADERSUBROUTINES,wheremethaneisanintegercalculatedusing

    thedynamicfunctionINTSPE,isequivalenttoppgm#thisinnetworklogic.

    DampingandTimeStepControl:DealingwithTemperatureSensitivities

    Inspectionofthekineticratecalculationsaboverevealsthatsomeextrastepswerenecessarytoassure

    steadystateconvergence:thepriormolarreactionrates(rate1andrate2)arestoredfromtheprevious

    iterationinunused8lumpcoordinatesCXandCY,respectively.(Thisapproachhasaconvenientside

    effect:itallowsthosevaluestobepostprocessed.)Thesepriorratesarethenusedtoslowtherateof

    changeofXMDOTvaluestoencourageconvergenceinsteadystatesaccordingtotheregisterrateLag,

    whichvariesfromaninitialvalueof0.5(midrangedampingviaarunningaveragebetweenthecurrent

    andlastvalue)to1.0(heavydampingbytheendofthesteadystate):

    rateLag=0.5+0.5*(loopct/stepNum)

    wheresstepNumisthenumberofpseudotimestepsallows(usedtosetNLOOPS)andloopctisthe

    currentstepnumber.ForreactionN,thisdampingtakesthegeneralformat:

    rateNnew=(1.0rateLag)*rateNnew+rateLag*rateNold

    Intransients,rateLagissettozero(meaningnodamping)viacasedependentregisters.Inother

    words,nodampingisrequiredintransients.However,thedefaulttimesteperrortolerancefactor

    DTSIZF,whichdefaultsto0.1(10%max.changepertimestepinanykeyparametersuchaspressure,

    speciesfraction,etc.),mustbereducedtoatmost0.02(2%max.change)topreventoscillationsinthetimeplots.9

    8 Ifbodyforceshadbeenimportant,thesevariableswouldnotbeavailableforthisusage.Eitherotheropen

    variablesmustbesought,orextraregistersmustbedeclaredtoholdthememoryforeachtanksreactionhistory.9 TheexperiencedusermightbewonderingwhysimilarlyreducingRSSIZFisntavalidapproachforSTDSTL

    solutions,avoidingtheneedforrunningaveragedamping.ThereasonisthatsmallRSSIZFvalueswouldrequire

    excessiveiterations(highLOOPCT)unlesslargevaluesareusedinitially.Moreimportantly,theconvergence

  • 7/31/2019 Methane Reforming Demonstration Problem

    11/19

    Whiletheapproachdiffersbetweensteadystate(STDSTL)andtransient(TRANSIENT)solutions,the

    causeforthisunusualstepisthesame:theextremetemperaturedependenceofthereactionrates,

    whichistobeexpectedbecauseofthepresenceofexponentialintheArrheniusrateequation.

    Specifically,theXMDOTvaluesarenotadjustedimplicitlybythecodeasafunctionoftemperature.

    Rather,XMDOTvalues

    are

    assumed

    constant

    over

    each

    solution

    step.Iftoolargeofastepistaken,the

    temperatureinthetankwillchangetoomuchandtheassumptionofconstantreactionratesisinvalid:

    theabsolutevalueofthepartialderivativeofXMDOTiwithrespecttoTListoohigh.Therefore,either

    thetimestepmustbereducedinanticipationofasignificantchangeinXMDOT(thetransient

    approach),ortheXMDOTvaluemustbedampedtopreventbinaryoscillations(thesteadystate

    approach).

    SteadyStateResults

    With3timesasmuchsteamasneededforstoichiometricconditions(SteamX=3),andgivenafeed

    temperatureof800Candwalltemperaturesof800Caswell,theresultingtemperaturegradientsat

    steadystate

    are

    shown

    below.

    Inthediagramabove,thefluidspeciesaremixedinjunctionreactor.1001(withoutreactionsyet

    permitted),theninjectedattheleftofthediagram,exitingattheright.Thetemperaturesdroptoabout

    550Cattheinlet,beforerisingtoabout780Cattheoutlet.Ifalongerreactorhadbeenused,theexit

    temperatureswouldhavebeenclosertothewalltemperatureof800C.Iffinerspatialresolutionhad

    beenused(forexample,20segmentsinsteadof10),theresultswouldhaverevealedthatthecoldest

    spotinthereactorisnotexactlyattheinlet,butisinsteadlocatedafewcentimetersdownstream

    whereendothermicreaction#1reachesitspeakrate.

    Thecorrespondingpartialpressurefraction(molefraction)forhydrogenisshownnext.

    checkerwouldstillpreventconvergencefrombeingdeclaredattheendoftherunsinceitalsocheckstheabsolute

    sizeofthepseudotimestepasanindicationofahavingachievedatimeindependentstate.

  • 7/31/2019 Methane Reforming Demonstration Problem

    12/19

    Recallingthepresenceofexcesssteam,themolefractionofhydrogenproducedisabout56%inthe

    outletstream.

    Steadystateanalysesareakeytoolforsizing,sensitivitystudies,etc.Recallthat,unlikemost

    SINDA/FLUINTmodels,thetypeofsteadystatesolutionusedforchemicalreactionsisnotinsensitiveto

    guessedinitialconditions.Therefore,eithertheruntimeallowed(stepNum)mayhavetobeenlarged,or

    moreaccurateinitialconditionsmayneedtobeguessed(registersminit_h,minit_c,etc.),orboth.ForparametricsweepsorSolverruns(sizing,correlationtotest,etc.),theguessedinitialconditionsshould

    reflectthefirstcasetobesolved,withonlymodestincreasesinstepNumtoaccommodatevariations

    thatmightbeexperiencedduringwhilesolvingforvariedconditions.Moresignificantincreasein

    stepNummayberequiredforstatisticaldesignruns(usingtheReliabilityEngineeringModule)sincethe

    variationsbetweensamplingsarenotnecessarilygradual:thepriorsteadystateanswersmightnotbe

    goodinitialconditionsforthenextsamplingruntobemade.

    TransientDemonstrationEvent

    AkeyfeatureofSINDA/FLUINTreactingflowmodelingistheabilitytosimulatetransients,includingfast

    transienteventssuchaspressurewaves,flowinstabilities,controlsystemactions,andvalvedynamics.

    Asademonstrationofthesecapabilities,thesteadystatesolutiondescribedinthelastsectionwillbe

    usedastheinitialconditionforanarbitraryevent:thereductionofsteaminjectiontostoichiometric

    conditions:athreefolddecreaseinsteamfeedattimezero,withanapproximatelytwofoldreductionin

    overallflowrate.

    Normally,atransientcaseisrunsimplybyaddingacalltotheTRANSIENTsolutionroutinefollowingthe

    calltothesteadystatesolution(STDSTL)withinOPERATIONS.Inthesamplemodelsassociatedwiththis

    document,thesteadystateandtransientanalysesareperformedastwoseparatecases:pfrand

    pfrTransient,respectively.ThepfrTransientcasestartsbyreadingbacktheinitialconditionsleftforitby

    thepfrcase,whichmusthavebeenexecutedbeforehand.Ifsomedimensionorboundaryconditionis

    changedthataffectsbothcases,theyshouldbothbeselectedandreruntogether.

    Theplotbelowshowsthetransienttemperatureresponseofthereactortoastepreductioninsteam

    injection.Astheflowratediminishes,thetemperaturesattheexit(e.g.,TL10,thetemperatureoflump

    #10)risereflectingtheirtemporarystagnationnexttoahotwall.Neartheinlet(TL1)thetemperature

    actuallygoesdownbeforerisingsincethereislesssteamtocoolfrom800Cbuttheendothermic

  • 7/31/2019 Methane Reforming Demonstration Problem

    13/19

    reactioncontinuesataboutthesamerate.Asanewsteadyconditionisestablished,thenetchangeisa

    warmingofthemiddleofthereactorandthereforemorecompletereformingoftheinflowingmethane.

    Analternativeviewofthissameeventisdepictedbelowintermsofthepartialpressurefraction(mole

    fraction)ofhydrogengas.Astheexcesssteamiswithdrawn,thefractionofhydrogenjumpsfirstattheinlet,andthisfrontcanbeseenprogressingthroughthereactorforthefirst0.16seconds.Thereactor

    shiftsfromproducing55%H2tomoreoptimum72%(themaximumpossibleis75%).

  • 7/31/2019 Methane Reforming Demonstration Problem

    14/19

    ValidationwithAspenPlus

    LikeNASAsCEA,AspenTechsAspenPluscanbeusedtopredictequilibriumconstituentfractionsbased

    ontheminimizationofGibbsfreeenergy.Inaddition,theenergyinputsneededtosustainan

    isothermalprocesscanbecalculated,ascantheenergyrequiredwhenthefeedtemperatureis

    substantiallycolderthantheproducttemperature.Finally,thetemperaturedropexperiencedbyhot

    steamandmethaneinjectedintoanadiabaticreactorcanalsobecalculated.

    ItisimportanttorememberthatalloftheAspenPluscalculationsassumedequilibriumconditionsto

    exist,whereastheSINDA/FLUINTpredictionsweremadebysimulatingfiniteratereactionsasaprocess.

    Inotherwords,tobeabletopredicttheheataddedtoastoichiometricstreaminjectedat25Cand

    heatedto(say)1200C,thereactorwallsweresetto1200Candsufficientreactorlengthwasaddedas

    neededtoprovidetheheattransferrequiredtoaccomplishthistask.Theexitstatewasthencompared

    withAspenPluspredictions,andtheheataddedtothestreamviaheattransferwastalliedtogeneratea

    totalnetpower.

    Comparedwiththemodelspresentedabove(andincludedwiththissample),uptothreedifferences

    weremadeinthevariationofthedemonstrationmodelthatwasusedtogeneratecomparisonresults:

    1) Thelengthwasincreased(upto40m)

  • 7/31/2019 Methane Reforming Demonstration Problem

    15/19

    2) Theresolutionwasincreased(from10segmentsto20)3) Atwophasedescriptionofwater(REFPROPderivedf6070_water.inc)wassubstitutedwhen

    25Cinletconditionswereused,suchthatthefactthatwaterisliquidatSTPcouldbe

    included.

    Theresultsofthisvalidationexercisearesummarizedbelow.

    Sixdiabaticcaseswerecompared,threewithbothfeedandproductsatthesametemperature(500C,

    800C,and1000C),andthreewithfeedat25Cwhileproducttemperaturesremainedthesame.The

    predictedspeciesmolefractionsaretabulatedbelow,alongwiththetwopowerrates(QfpforTf=Tp,

    andQ25forTf=25C).Thedifferencesareexpressedaspercentages,whicharecalculatedbasedontotal

    massforspeciesfractions.

    500C 800C 1000C

    Aspen S/F diff. Aspen S/F diff. Aspen S/F diff.

    CH4 0.3203 0.3188 0.14% 0.0269 0.0262 0.07% 0.0009 0.0009 0.00%

    H2O 0.2493 0.2475 0.18% 0.0199 0.0196 0.03% 0.0008 0.0008 0.00%

    CO 0.0189 0.0193 0.04% 0.2296 0.2304 0.07% 0.2495 0.2495 0.00%

    CO2 0.0710 0.0713 0.03% 0.0069 0.0066 0.04% 0.0001 0.0001 0.00%

    H2 0.3406 0.3431 0.25% 0.7166 0.7173 0.06% 0.7488 0.7487 0.00%

    Qfp(kJ/mol) 126.99 128.13 0.89% 319.45 318.64 0.26% 393.94 393.49 0.11%

    Q25(kJ/mol) 42.18 42.70 1.22% 201.77 202.14 0.19% 225.40 224.58 0.36%

    Notethattheagreementimprovesasthetemperaturerises(fasterreactionratesmakingequilibriuma

    morelikelyresult),butisverygoodevenforthe500Ccaseconsideringthatthetwoprogramsdonot

    usethesameapproach(theAspendataassumesequilibrium,andthisparticularSINDA/FLUINTrunuses

    reactionkinetics),nordotheyusetheexactsameequilibriumconstants(thoughapparentlytheyare

    veryclose),andthattheenthalpyfunctions(fluidproperties)aredifferentaswell.

    Threemorecaseswerecomparedbasedonisothermalprocesses,withinlettemperaturesbeing500C,

    800C,and1000C.Equalmolefractionsofmethaneandsteamwereused:stoichiometricconditions.

    Theresultsareshownbelow.Again,thespeciesfractiondifferenceswerecalculatedonthebasisoftotal

    mass.Thetemperaturedifferenceswerecalculatedonthebasisoftemperaturedropfromtheinlet.

    500C 800C 1000C

    Aspen S/F diff. Aspen S/F diff. Aspen S/F diff.

    CH4 0.4342 0.4346 0.04% 0.3438 0.3430 0.08% 0.2842 0.2828 0.14%

    H2O 0.4020 0.4026 0.06% 0.2775 0.2765 0.10% 0.2103 0.2087 0.16%

    CO 0.0006 0.0006 0.00% 0.0118 0.0120 0.02% 0.0340 0.0345 0.06%

    CO2 0.0323 0.0320 0.02% 0.0663 0.0665 0.02% 0.0739 0.0741 0.02%

    H2 0.1309 0.1301 0.08% 0.3006 0.3020 0.14% 0.3977 0.3999 0.23%

    Tprod(C) 367.5 368.3 0.61% 477.6 476.9 0.22% 531.3 530.3 0.21%

  • 7/31/2019 Methane Reforming Demonstration Problem

    16/19

    DemonstrationUseofaCEA-derivedEquilibriumFluid

    ItisnoteworthythattheAspenPlusequilibriumcalculationyieldedresultssimilartoafiniterate

    approach,evenattemperaturesaslowas370C.Thisobservationmeansthatavariablemolecular

    weightequilibriumfluidmaybeappropriate,atleastwhenthefeedratiosarenotalteredduringthe

    courseofananalysis.10

    Todemonstratethismodelingapproach,avariationoftheabovemodelwasmadeandisavailablein

    Sinapsform.

    First,CEAwasrunforastoichiometricmixtureofwaterandmethane,restrictingtheresultingmixture

    toonlythefivespeciesusedintheoriginalanalysis(i.e.,excludingcarbon,otherminorspecies,ions,

    etc.).ThistableofCEAequilibriumdatawasconvertedintoasingleSINDA/FLUINTFPROPblockusing

    thecea2fproputility,11 andtheheatofformationwassettobethesameastheenthalpyofthisnew

    fluidatSTP.

    ThisnewfluidwasthenusedinthePFRmodelasspeciesE,suchthatwhenM(methane)andW

    (water)arereacted,theyresultinthisvariableweightequilibriumfluidasfollows:

    CH4+H2O E (equivalentreaction)

    ThespeciesH(hydrogen),G(carbonmonoxide),andC(carbondioxide)arethereforeeliminatedfrom

    themodel.NotethatmethaneandwaterconstitutesignificantportionsoffluidE,especiallybelow

    600C.

    Byassumingequilibrium,thekineticratecalculationsareeliminated.Infact,reactionsinallbutthefirst

    tankareeliminatedaswell:downstreamreactionsaretakencareofbytemperature andpressure

    dependentshiftswithintheequilibriumfluiditself.WhiletheutilityEQRATEcouldhavebeenusedtoset

    XMDOTvalues,insteadthereactionratesinthefirsttank(tank1)weresetinFLOGIC0applyinga

    reactionefficiencybasedoninflowingmethane,whichwaspresumedtobethelimitingreagent.A

    newregisterconvert,setto0.999,declaresthat99.9%ofincomingmethanewillbecombinedwithan

    equalmolaramountofwaterandreplacedbyanequalmassofequilibriumfluidE. Inflowingmethane

    massflowiscalculatedastheproductofthemassfractionupstreaminjunction1000,xgM1001and

    theincomingmassflowrateinpath3,fr3:

    xmdotM1 = -xgM1001*fr3*convert

    xmdotW1 = xmdotM1*mw_h2o/mw_ch4

    xmdotE1 = -(xmdotM1+xmdotM1)

    Becauseexcesssteamaffectstheequilibriumpoint,theequilibriumfluidEshouldonlybeusedto

    replaceastoichiometricmixtureofmethaneandwater.Indeed,whensuchastoichiometric(SteamX=1)

    10Excessoxidizer(e.g.,air)canusuallybeaddedorsubtractedfromanequilibriumfluidrepresentingthehot

    productsofcombustion.Thiswouldonlybetrueforamethanereformerinthespecialcaseoftheadditionor

    subtractionofachemicallyinertsubstance. Seewww.crtech.comandtheSINDA/FLUINTUsersManualforfurther

    details.11 http://www.crtech.com/EQfluids.html

  • 7/31/2019 Methane Reforming Demonstration Problem

    17/19

    mixturewastested,theresultingtemperaturesinadiabaticcasesmatchedverywellagainsttheAspen

    PlusanalysisandfullreactingflowmixtureSINDA/FLUINTanalysispresentedabove.

    Asafirstapproximation,iftheexcesssteamcouldbeassumedtobeindependentoftherestofthe

    mixture,

    CH4+3H2O E+2H2O (equivalentreaction,excesssteam:SteamX=3)

    thentheSteamX=3steadystatecasepresentedinapriorsectioncouldberepeatedwiththe

    equilibriumfluidapproach,yieldingsimilar(butnotcompletelyequivalent)results:

    Ifthemolarratioofsteamtomethaneweretoremainat3:1,amoreaccurateapproachwouldbeto

    createanewequilibriumfluidEforthatscenariousinganotherCEArunandFPROPconversion(whichis

    atriviallyeasyprocess):

    CH4+3H2O E (equivalentreaction,SteamX=3,newequilibriumfluid)

    Usinganequilibriumfluidapproachreducesthenumberofspeciestrackedfrom5to3,andalso

    eliminatestheextremesensitivityofthereactionratestotemperaturebyassumingtheyreinfinitely

    large.Theresultisasignificantincreaseincomputationalspeed,andmorerobustconvergencethatis

    muchlesssensitivetocoarselyguessedinitialconditions.

    However,theabilitytoresolvetheinternaldetailsoftheequilibriumfluidarelostasacompromise.For

    example,onecannolongerplotthefractionofhydrogensincetheactualcurrentamountofhydrogenis

    hiddenwithinthechemicalcontrolvolumethattheequilibriumfluidrepresents.Toovercomethis

    limitation,aseparateCEAruncouldbemadeusingtheSINDA/FLUINTpredictedtemperaturesand

    pressurestocalculatethefractionalconstituentsoftheequilibriumfluidatoneormorepoints.12

    Whenusedwithappropriatecautions,validatingassumptionsasneeded,theequilibriumfluidmethodcanbeapowerfulcompanionapproachtoperformpreliminarysizingandsensitivityanalyses,for

    example.Ifkineticratesarenotknown,itmightbetheonlyapproachavailableforfirstorder

    estimationsandbracketinganalyses.

    12FutureplanscallfortheinclusionofallCEAcapabilitieswithinSINDA/FLUINT,whichwouldeliminatethis

    additionalpostsimulationstep.

  • 7/31/2019 Methane Reforming Demonstration Problem

    18/19

    SideNote:Carbon(coke)Formation

    Atonebarpressure,withequalmolarinputsofmethaneandsteam(stoichiometricconditions),CEA

    predictsthefollowingmolefractionsversustemperature(indegreesC).

    Asisevidencedbytheaboveequilibrium(Gibbsfreeenergyminimization)analysis,thefractionof

    carbon(coke,graphite)peaksatabout600C,atwhichpointtherearemoremolesofcarboninthe

    mixturethaneitherCOorCO2.Whilethepresenceofcarbonhasbeenneglectedinthisdemonstration

    case,itclearlymustbeincludedforamorecompletetreatment.Therefore,thissectionprovidesbrief

    notesonhowsuchamoredetailedmodelcouldbeconstructed.

    Toincludecarboninthemethanereformingmodel,anotherspeciesmustbeintroduced:onethatdoes

    notcontributetothegaspressure.EventhoughtherearenosolidspeciesperseinSINDA/FLUINT,a

    9000seriesnonvolatileliquidisanadequatesubstitute,withfakeviscosityandsurfacetensionvaluesas

    neededtocomplywithinputrequirements.Bydefault,thisfluidwillbeassumedtomixhomogeneously

    withtheremainingspecies.

    Threemorechemicalreactionswouldneedtobeincluded:

    CH4 C+2H2 (reaction3)

    2CO C+CO2 (reaction4)

    CO+H2 C+H2O (reaction5)

    Equations2530ofthereferencedpaper(Shuyan,et.al.)providetherelevantequilibriumKconstants

    andpreexponentialktermsforthesereactions.

  • 7/31/2019 Methane Reforming Demonstration Problem

    19/19