Cet ouvrage fait partie de la bibliothèque YouScribe
Obtenez un accès à la bibliothèque pour le lire en ligne
En savoir plus

CFD simulation of biomass gasification using detailed chemistry [Elektronische Ressource] / vorgelegt von Arash Rashidi

De
123 pages
Inaugural - DissertationzurErlangung der Doktorwu¨rdederNaturwissenschaftlich-Mathematischen Gesamtfakulta¨tderRuprecht-Karls-Universita¨tHeidelbergvorgelegt vonArash Rashidi, MSc. Energy SystemsausShiraz, IranTag der mundlichen Prufung: 4. Februar 2011¨ ¨CFD Simulation of BiomassGasification Using DetailedChemistryGutachter: Prof. Dr. Uwe RiedelPD Dr. Nikolaus DahmenAbstractThe use of biomass as a CO -neutral renewable fuel and the only carbon containing2renewableenergysourceisbecomingmoreimportantduetothedecreasingresourcesof fossil fuels and their effect on global warming. The projections made for theRenewable Energy Road Map [1] suggested that in the EU, the use of biomass canbe expected to double, to contribute around half of the total effort for reaching the20 % renewable energy target in 2020 [2]. To achieve this goal, efficient processes toconvert biomass are required.AttheKarlsruheInstituteofTechnology(KIT),Germany,atwo-stageprocesscalledRbioliq [3], for the conversion of biomass into synthetic fuel, is being developed. Inthis process, straw or other abundant lignocellulosic agricultural by-products areconverted to syngas through fast pyrolysis and subsequent entrained flow gasifica-tion. After gas cleaning and conditioning, the syngas is converted into differentchemicals via known processes such as direct methanol synthesis or Fischer-Tropschsynthesis.
Voir plus Voir moins

Inaugural-Dissertation

rzuErlangungderDoktorwu¨rde
erdNaturwissenschaftlich-MathematischenGesamtfakulta¨t
erdRuprecht-Karls-Universita¨t
ergbeldHei

vorgelegtvon
ArashRashidi,MSc.EnergySystems
suaShiraz,Iran

Tagdermu¨ndlichenPru¨fung:4.Februar2011

CFD

onatiulmSi

of

sasomBi

GasificationUsingDetailed

Gutachter:

trysihemC

f.oPr

PD

.rD

eUw

elediR

Dr.NikolausDahmen

Acttrabs

TheuseofbiomassasaCO2-neutralrenewablefuelandtheonlycarboncontaining
renewableenergysourceisbecomingmoreimportantduetothedecreasingresources
offossilfuelsandtheireffectonglobalwarming.Theprojectionsmadeforthe
RenewableEnergyRoadMap[1]suggestedthatintheEU,theuseofbiomasscan
beexpectedtodouble,tocontributearoundhalfofthetotaleffortforreachingthe
20%renewableenergytargetin2020[2].Toachievethisgoal,efficientprocessesto
convertbiomassarerequired.

AttheKarlsruheInstituteofTechnology(KIT),Germany,atwo-stageprocesscalled
bioliqR[3],fortheconversionofbiomassintosyntheticfuel,isbeingdeveloped.In
thisprocess,straworotherabundantlignocellulosicagriculturalby-productsare
convertedtosyngasthroughfastpyrolysisandsubsequententrainedflowgasifica-
tion.Aftergascleaningandconditioning,thesyngasisconvertedintodifferent
chemicalsviaknownprocessessuchasdirectmethanolsynthesisorFischer-Tropsch
synthesis.

Theprimegoalofthisthesiswasthemodelingandsimulationofthegasification
ofbiomass-basedpyrolysisoil-charslurriesinanentrainedflowgasifier,whichisan
importantstepofthebioliqRprocess.ComputationalFluidDynamics(CFD),as
apowerfultoolformodelingandsimulationoffluidflowprocesses,wasutilizedin
thisthesis.

Alabscaleentrainedflowgasifier,locatedatKIT,wassimulatedusingtheCFD
codeANSYSFLUENT12.0.Duetotheturbulentnatureoftheflow,therealizable
k-εmodelwasusedtomodeltheturbulence.Thediscretephasemodel(DPM)
wasemployedtodescribethefluidphase,consistingofcharparticlessuspendedin

ii

ethyleneglycol.Ethyleneglycolservedasnon-toxicmodelfuelforpyrolysisoil,
mainlybecauseofitssimilarC/H/O-ratioanditssimilarphysicalpropertiesto
biomassderivedliquidpyrolysisproducts.
Adetailedreactionmechanismforhightemperatureoxidationofethyleneglycolwas
implementedintheCFDcode.Themechanismcomprisedof43chemicalspecies
and629elementaryreactions.Theuseofdetailedchemistryenablesonetohavea
deeperinsightintothegasificationprocess.Turbulence-chemistryinteractionswere
modeledwiththeeddydissipationconcept(EDC).Thein-situadaptivetabulation
(ISAT)procedurewasemployedtodynamicallytabulatethechemistrymappings
andreducecomputertimeforthesimulation.Theeffectofthethermalradiation
wastakenintoaccountbyusingthediscreteordinatesmodel(DOM).Theradiative
propertiesofthegasweredescribedwiththeweightedsumofgraygasesmodel
.GGM)WS(Thesimulationresultswerecomparedwiththeexperimentalmeasurementswherever
possible,withgoodagreement.Thesimulationsdepictedtheimportanceofthe
recirculationzoneinentrainedflowgasification.Furthermore,themainreaction
pathofethyleneglycolgasificationcouldbeobservedandanalyzed.
Inordertostudytheeffectofboundaryconditionsonthegasificationprocess,a
seriesofsimulationsweredonetoperformsensitivityanalysis.Fourparameters
werevaried,namely:oxidizerandfuelinlettemperatures,theoxidizercomposition,
theair-fuelratioandtheoperatingpressureofthegasifier.Effectsoftheparame-
tervariationsonthegasificationefficiencyandthecompositionoftheproductgas
werestudied.Threedifferentchemistrymodels(i.e.equilibriumchemistry,flamelet
modelandEDC)werestudiedinthisthesis.Theirrelativeadvantagesanddis-
advantagesforthesimulationofgasificationprocesseswereexamined.TheEDC
modelprovedtobethebetterchoiceforentrainedflowgasifierswithrecirculation
es.nzoTheslurrygasificationsimulationswereperformedtostudytheeffectsofthemass
fractionsofthecharparticlesontheprocess.Withtheaidofthedetailedchemistry
model,sub-processescouldbeanalyzedandsuggestionsfortheimprovementcould
bemade.
Thesimulationsperformedinthisworkhelptobetterunderstandthegasification
processinsideentrainedflowgasifiersandconsiderablyreducethenumberofex-
perimentsneededtocharacterizethesystem.Thesimulationsproducedspatialand
temporalprofilesofdifferentsystemvariablesthataresometimesimpossibletomea-
sureorareaccessibleonlybyexpensiveexperiments.However,moreexperimental

measurementshelptovalidateandoptimize

ysesperformed

conditions

dna

inthisstudyareconsidered

stssia

teh

lccessfusu

pe-ulsca

iii

theCFDmodel.Thesensitivityanal-

sa

fo

abasistofindoptimized

entrained

wofl

gasifiers.

operating

Zusammenfassung

DieNutzungvonBiomassealsCO2-neutralerEnergietra¨gerundeinzigekohlen-
stoffhaltigeerneubareEnergiequelle,gewinntwegenschwindendenfossilenEnergi-
etra¨gernundderenEinflussaufdenKlimawandelzunehmendanBedeutung.Der
Fahrplanfu¨rerneubareEnergien[1]entha¨ltdasZiel,biszumJahr2020denAn-
teilerneuerbarerEnergienamGesamtenergieverbrauchinderEUauf20%zu
steigern.DerAnteilvonBiomasseamEnergiemixsollsichimRahmendiesesPlans
verdoppeln[2].Hierfu¨rsindeffizienteProzessenzurUmwandlungvonBiomasseer-
forderlich.
AmKarlsruherInstitutfu¨rTechnologie(KIT)wurdedaszweistufigebioliqR-Ver-
fahrenkonzipiert[3],indemaustrockenerRestbiomassedurchdezenteraleSchnellpy-
rolyseundzenteraleFlugstrom-DruckvergasungSynthesegaserzeugtwird.Ausdem
gereinigtenundkonditioniertenSynthesegasko¨nnenz.B.durchFischer-Tropsch-
oderMethanol-SynthesenebenSynthesekraftstoffenaucheineVielzahlvonchemis-
chenGrundstoffenerzeugtwerden.
SchwerpunktdieserArbeitistdieModellierungundnumerischeSimulationdesVer-
gasungsprozessesvonbiomassesta¨mmigenO¨l-Koks-SlurrysineinemFlugstromver-
gasernachdembioliqR-Verfahren.NumerischeStro¨mungssimulation(Computa-
tionalFluidDynamics-CFD)wurdeimRahmendieserArbeitalseineeffiziente
MethodezurModellierungreaktiverStro¨mungenverwendet.
DieSimulationeinesVersuchsreaktorsdesKITwirdindervorliegenderArbeit
mitderCFD-SimulationssoftwareANSYSFLUENT12.0durchgefu¨hrt.Fu¨rdie
BeschreibungdesturbulentenStro¨mungsfeldeswirddas”realizable”k-ε-Modellver-
wendet.ZurModellierungderdiskretenPhase(flu¨ssigesEthylenglykolundKokspar-

v

tikel)wurdedasDiscrete-Phase-Modell(DPM)verwendet.Ethylenglykoldienteauf-
grundvergleichbarerphysikalischerEigenschaftensowiea¨hnlichemC/H/O-Verha¨ltnis
alsModellsubstanzfu¨rPyrolyseo¨l.
ImRahmendieserArbeitwurdeeindetaillierterReaktionsmechanismuszurBeschrei-
bungderOxidationsreaktionenvonEthylenglykolverwendet,deraus43Speziesund
629Elementarreaktionenbesteht.DieVerwendungeinesdetailliertenReaktions-
mechanismusermo¨glichtdieEinsichtinchemischeVorga¨ngederVergasung.Fu¨rdie
KopplungvondetaillierterReaktionskinetikundTurbulenzeffektenwirddasEddy
DissipationConcept(EDC)Modellverwendet.DieVerwendungdesIn-SituAdap-
tiveTabulation(ISAT)AnsatzeszurTabellierungdesReaktionsforschrittsreduziert
dieRechenzeitdeutlich.DasModellderDiskretenOrdinaten(DOM)wurdealsther-
mischesStrahlungsmodellverwendet.DieStrahlungseigenschaftendesGaseswerden
mittelsdemWeightedSumofGrayGasesModell(WSGGM)berechnet.
VergleichevonsimuliertenundexperimentellenWerten(sofernmo¨glich)zeigten
akzeptable¨Ubereinstimmungen.DieSimulationenhabenzusa¨tzlichdieWichtigkeit
vonRezirkulationszoneninderFlugstromvergasungdargestellt.Daru¨berhinaus
wurdendiehauptReaktionspfadederEthylenglykolvergasungveranschaulicht.
UmdieAuswirkungenderRandbedingungenaufdieZusammensetzungdesSyn-
thesegasesunddenVergasungswirkungsgradzuuntersuchen,wurdenParameter-
studienmitverschiedenenRandbedingungendurchgef¨uhrt.VierParameterna¨m-
lichdieLuft-undBrennstoffeintrittstemperaturen,derSauerstoffgehaltdesZer-
sta¨ubungsmediums,dieLuftzahlundderVergaserdruckwurdenvariiert.Außerdem
wurdendreiunterschiedlicheChemie-Modelle(Gleichgewichts-Modell,Flamelet-Mo-
dellundEDC-Modell)untersuchtundderenVor-undNachteilemiteinanderver-
glichen.DasEDCModellerwiessichfu¨rdieModellierungvonVergasungsvorga¨ngen
inFlugstromvergasernmitRezirkulationszonenalsgutgeignet.
DieCFD-SimulationenderSlurryvergasungwurdedurchgefu¨hrt,umdenEinfluss
vonMassenanteilenderKokspartikelindemSlurrygemischzuuntersuchen.Mit
HilfederverwendetendetailliertenChemiekonnteneinigeTeilprozesseanalysiert
werdenundVerbesserungsvorschla¨gegemachtwerden.
DurchdieComputersimulationenla¨sstsichdieZahlvonzeit-undkostenintensiven
Experimentenreduzieren.Zudemerha¨ltmaneinezeitlicheund/odero¨rtlicheAuflo¨-
sungderTeilprozessedesGesamtsystemsundkannProzessgro¨ßencharakterisieren,
diemitexperimentellenMethodennicht,odernuruntererheblichenAufwand,zu
bestimmensind.UmjedochdieVerla¨sslichkeitderCFD-Simulationenzugewa¨hrleis-
ten,mu¨ssenjedochauchmehrValidierungsexperimentedurchgefu¨hrtwerden.Die

indieserArbeitdurchgefu¨hrtenSensitivita¨tsanalysenko¨nnenalseineBasisfu¨r

Festlegungvon

erd

Skalierung

optimiertenBetriebsbedingungenverstandenwerdenund

esd

Flugstromvergasers

unterstu¨tzend

gneizteset

.enderw

ko¨nnen

iv

eid

eib

Acknowledgement

Adissertationdoesnotjustappearoutofnowhere,andalthoughitissupposedto
beacontributionbyonepersonforaPhD,therearestillalotofpeoplebehindit
whohavehelpedandinspiredmeduringmydoctoralstudy.

IwouldliketoexpressmyheartfeltgratitudetomysupervisorProf.Dr.UweRiedel
forlettingmedomyPhDinhisgroup,forcreatingtheresearchenvironment,and
forsupprotingmebothfinanciallyandscientificallyduringthisperiod.Iwouldlike
tothankmyco-advisorDr.NicolausDahmenforhisguidanceandeffortsinthis
.jectorp

ThecurrentthesisisapartofthecollaborativeprojectSynthesegaserzeugungdurch
Flugstromvergasungvon¨Ol/Koks-SlurrysausBiomasse-GrundlagenzurProzes-
soptimierungsupportedbytheGermanFederalMinistryofEducationandRe-
search(BMBF).IwouldliketoacknowledgethemfortheirsupportunderGrant
03SF0320D.IalsothanktheprojectpartnersinKarlsruheInstituteofTechnology
(KIT)forprovidingrelevantdata.

ManythanksgotomyfriendsandcolleaguesinreactiveflowsgroupinIWRwho
mademytimeinHeidelbergjoyful.SpecialthankgoestoDr.SimonHafnerforall
thetechnicalandnontechnicaldiscussionswehad,histeamworkaswellashishelps
ineditingmyGermantexts.IwouldalsoliketothankIngridHellwigforherperfect
administrationhelpsandJoachimSimonandJu¨rgenMoldenhauerforprovidingIT
assistancewhereverrequired.

Mydeepestgratitudegoestomyfamilyfortheirunflaggingloveandsupport
throughoutmylife.Myparents,SoghraGholamiandAhmadRashidi,raisedmeup
withaloveforscienceandsupportedmeunconditionallyinallmypursuits.They

iiiv

deservefarmorecreditthanIcanevergivethem.Mysister,AfsoonRashidi,was

alwaystherewithherencouragementandcare.Thankyouall!

Lastbutnotleast,averyspecialthanktomylovelyMareikeHeitnerforherunlim-

itedsupportanddeliciouscakes,

work.Danke!

ArashRashidi

Heidelberg,December2010

without

Ihciwh

lucod

notfind

guoenh

erwop

ot

Contsten

cttrabsA

Zusammenfassung

Acknowledgement

i

iv

vii

1Introduction1
1.1BiomassGasification...........................2
R1.2ThebioliqProcess...........................6
1.3ComputationalFluidDynamicsModeling................7
1.4ScopeofthisThesis............................8

2GoverningEquations10
2.1MassConservationEquation.......................10
2.2MomentumConservationEquation...................11
2.3EnergyConservationEquation......................11
2.4SpeciesConservationEquation......................12
2.5Reynolds-andFavre-Averaging.....................13

3PhysicalModels16
3.1Turbulence.................................16
3.1.1Standardk-εModel........................17
3.1.2Realizablek-εModel.......................18
3.2ThermalRadiation............................20
3.2.1RadiativeTransferEquation...................21
3.2.2DiscreteOrdinatesModel....................21
3.2.3RadiationinReactiveFlow...................24
3.3DiscretePhaseModel...........................24
3.3.1ParticleMotionTheory......................25
3.3.2HeatandMassExchange.....................27
3.3.3CouplingwiththeContinuousPhase..............31

Contents

x

4ChemistryModels33
4.1ChemicalReactionMechanism......................34
4.1.1ReactionMechanismDevelopment...............37
4.1.2ReactionMechanismSimplification...............37
4.2NonpremixedCombustionwithEquilibriumChemistry........38
4.3FlameletModel..............................40
4.4EddyDissipationConcept........................43

5NumericalModels45
5.1FiniteVolumeMethod..........................45
5.1.1PressureBasedSolver......................46
5.1.2DiscretizationofEquations....................47
5.1.3PressureVelocityCoupling....................49
5.2IntegrationofParticleEquationofMotion...............51
5.3InSituAdaptiveTabulation.......................52

6ResultsandDiscussion55
6.1GasifierModelandSimulationConditions...............56
6.2FlowPattern...............................58
6.3TemperatureandSpeciesConcentrations................60
6.4EffectofOperatingConditions......................65
6.4.1InletTemperatures........................65
6.4.2OxidizerComposition......................67
6.4.3Air-FuelRatio...........................68
6.4.4Pressure..............................69
6.5EffectofChemistry............................71
6.5.1ReactionMechanism.......................71
6.5.2ChemistryModel.........................73
6.6SlurryGasificationSimulation......................77

7ConclusionsandPerspective

phibliograBy

83

87

97endicesppAAA.1Nomenclature...............................97
A.2ReactionMechanism...........................101

I1.ductiontron

TheuseofbiomassasaCO2-neutralrenewablefuelandtheonlycarboncontain-
ingrenewableenergysourceisbecomingmoreimportantduetothedecreasing
resourcesoffossilfuelsandtheireffectonglobalwarming.Hence,itisexpected
thatbiomasssubstitutesgraduallyafractionoffossilfuels.IntheEU,around5
%offinalenergyconsumptionisfrombio-energy.Theprojectionsmadeforthe
RenewableEnergyRoadMap[1]suggestedthattheuseofbiomasscanbeexpected
todouble,tocontributearoundhalfofthetotaleffortforreachingthe20%renew-
ableenergytargetin2020[2].Efficientbiomassconversionprocessesarerequired
toachievethisgoal.Biomassasasourceofenergyandthedifferentconversion
methodsarediscussedbrieflyinsection1.1.Thesemethodsaredividedintwomain
groups,i.e.thebiochemicalandthethermochemicalconversionmethods.One
advantageofthermochemicalconversionisthatitistypicallyfasterthanthebio-
chemicalconversion[4].AttheKarlsruheInstituteofTechnology(KIT),Germany
atwo-stageprocess-bioliqR[3]-forbiomassconversionintosyntheticfuelisbe-
ingdevelopedbasedonathermochemicalconversionpath.Inthisprocess,straw
orotherabundantlignocellulosicagriculturalby-productsareconvertedtosyngas
throughfastpyrolysisandsubsequententrainedflowgasification.Aftergascleaning
andconditioning,thesyngasisconvertedintodifferentchemicalsviaknownpro-
cessessuchasdirectmethanolsynthesisorFischer-Tropschsynthesis.ThebioliqR
processwillbediscussedinmoredetailinsection1.2.
ThroughtheentrainedflowgasificationasthesecondstepinthebioliqRprocess,
thebiomass-basedliquidslurryisconvertedintosyngas.Thegoalofthisstudyisthe
modelingandsimulationofthegasificationprocessinsideanentrainedflowgasifier
andparametricstudiestoimprovetheknowledgeonthebiomassgasificationprocess

1.1.BiomassGasification

2

aswellastodesignentrainedflowgasifiers.ComputationalFluidDynamics(CFD),
asapowerfultoolformodelingandsimulationoffluidflowprocesses,isutilizedin
thiswork.CFDsimulationsarenowadayseasytoperform,andwhensufficiently
validatedthroughexperimentaldata,becomeapowerfuldesignandoptimization
tool.Asfarasindustrialapplicationsareconcerned,accuratesimulationtoolscan
beutilizedforthescale-upofdevices.

1.1BiomassGasification
Notlongago,humans’basicsurvivaldependedinwholeorinpartontheavailability
ofbiomassasasourceoffoodandasanenergysourceforheatingandcooking.
Civilizationbeganitsenergyusebyburningbiomass.Theuseoffossilfuelsasa
majorenergysourcedatesbacktotherecentcenturies.AccordingtoUnitedNations
FrameworkConventiononClimateChange(UNFCCC)[5],biomassisdefinedas

non-fossilizedandbiodegradableorganicmaterialoriginatingfromplants,
animalsandmicro-organisms.Thisshallalsoincludeproducts,by-
products,residuesandwastefromagriculture,forestryandrelatedin-
dustriesaswellasthenon-fossilizedandbiodegradableorganicfractions
ofindustrialandmunicipalwastes.

Basedonthisdefinition,onecandividebiomassintotwobasicgroups:

•Virginbiomasswhichincludeswood,plantsandleaves(ligno-cellulose);and
cropsandvegetables(carbohydrates).

•Wastewhichincludesmunicipalsolidwastes,sewage,animalandhumanwastes
andgasesderivedfromlandfilling,etc.

ThesetwogroupsofbiomasswiththeirsubclassificationsarelistedinTable1.1[6].
Amajorpartofthevirginbiomassandforestryandindustrialwastesareligno-
cellulosicmaterialwhosemajorconstituentsarecellulose,hemicelluloseandlignin.
Smalleramountsofpectin,protein,extractivesandasharealsopartofbiomass[7].
Table1.2showsthecompositionofsomeselectedbiomasssources.Thecomposition
canvaryfromonetypetoanotherandalsoinsidethetypeduetodifferentfactors
suchasclimateandtimeofharvest.
Theligno-cellulosicbiomassisnotapartofthehumanfoodchainandhasagood
potentialforbeingusedinbioenergyproduction.Therearetwomajorpathways

1.1.BiomassGasification

VirginTerrestrialbiomassForestbiomass
ssesaGrEnergycrops
Cultivatedcrops
AquaticbiomassAlgae
Waterplants
WasteMunicipalwasteSewage
Biosolids
Landfillgas
AgriculturalsolidwasteLivestockandmanures
Agriculturalcropresidue
ForestryresiduesBark,leaves,floorresidues
IndustrialwasteDemolitionwood,sawdust
Wasteoilorfat

Table1.1:Twomajorgroupsofbiomass

3

forbiomassconversion(Figure1.1);i.e.biochemicalconversionandthermochemi-
calconversion.Thebiochemicalconversionissubdividedtoaerobicandanaerobic
digestionandfermentation.Detailsaboutthebiochemicalconversionpathcanbe
foundin[6]and[8].Thesecondpath,thethermochemicalconversion,hasthree
importantsubdivisions,i.e.combustion,pyrolysis,andgasification.
Combustioninvolveshigh-temperatureconversionofbiomassinexcessairintocar-
bondioxideandsteam.Largeamountsofheatareproducedinthisprocess.Com-
bustionprocessesareavailableforconversionofvirginandwastebiomassfeedstocks
toheat,steam,andelectricpowerinadvancedcombustionsystems.
Biomasspyrolysiscanbedescribedasthedirectthermaldecompositionoftheor-
ganiccomponentsinbiomass,intheabsenceofoxygen,toyieldliquidandsolid
derivativesandfuelgases.Inthisprocess,largehydrocarbonmoleculesofbiomass
arebrokenintosmallerhydrocarbonmolecules.Theamountofliquid,solidand
gasproductsdependsonpyrolysisoperatingconditionsliketemperature,pressure,
heatingrateandresidencetimeinsidethereactor.Unlikecombustion,pyrolysisis
notanexothermicprocess.

1.1.BiomassGasification

Cellulose(%)Hemicellulose(%)Lignin(%)
Willow501925
Larch262735
Switchgrass453212
Corncobs453515
Wheatstraw305015

Table1.2:Compositionofsomebiomasssources[6,7]

Figure1.1:Pathwaysofbiomassconversion(adaptedfrom[6])

4

Gasification,whichisthefocusofthisstudy,convertsfossilornonfossilfuelsinto
usefulgases.Itrequiresamediumforreaction,whichcanbegasorsupercritical
water[6].Gaseousmediaincludeair,oxygen,steam,oramixtureofthese.The
gasifyingagentsreactwithsolidcarbonand(heavier)hydrocarbonstoconvertthem
intolow-molecular-weightgaseslikeCOandH2.Theproductgases(syngas)can
thenbeconvertedoncatalystsintovariousproductslikeFischer-Tropsch(FT)diesel,
olefins,methanol,ethanol,dimethylether(DME),hydrogenorotherchemicalsas
canbeseeninfigure1.2.
Therearesomemajordifferencesbetweengasificationandcombustion.Gasification
packsenergyintochemicalbondswhilecombustionreleasesit.Thegasification
processtakesplaceinreducing(oxygen-deficient)environmentsrequiringheat,but
incaseofcombustiontheenvironmentisoxidizing.Incontrarytoproductgases
fromgasification,combustionproductgasesdonothaveausefulheatingvalue.

1.1.BiomassGasification

5

Gasifiersareclassifiedmainlyonthebasisoftheirgas-solidcontact-modeandgasi-
fyingmedium.Basedonthegas-solidcontact-mode,therearethreemajortypesof
gasifiersusedinindustry[9]:

1.MovingorFixedBedGasifiers
•Updraftgasifiers
•Downdraftgasifiers
•Crossdraftgasifiers
2.FluidizedBedGasifiers(FBG)
•BubblingFBG
•CirculatingFBG
3.EntrainedFlowGasifiers(EFG)
•DownFlowEFG
•UpFlowEFG
Inthefixedbedgasifiers,thesolidfuelissupportedinafixedbed(grate)through
whichthegasifyingmediumflowsincounter-current(updraft),co-current(down-
draft)orcross-current(crossdraft)configurations.Inthefluidizedbedgasifiers,the
fuelisfedtoasuspended(bubbling)orcirculatinghotsolidbed.Thebediskept
inafluidizedconditionbypassingthegasifyingmediumatappropriatevelocities
throughit.Theseandothertypesofgasifierssuchasindirectgasifiers,cyclone
gasifiersandheatpipegasifiersarediscussedin[9]indetail.
Thegasifierusedforthisstudyisofentrainedflowtype.Entrainedflowgasifiers
normallyusefuelintheformofgas,powder,orslurryandarewidelyusedforlarge
scalegasificationofcoal,petroleumcoke,andrefineryresiduesandarealsoagood
choiceforbiomass-basedslurrygasification.Thesegasifiersarecharacterizedbyfuel
particlesdraggedalongwiththegasstream.Thisgenerallymeansshortresidence
times,hightemperatures,andsmallfuelparticles[10].Thefuelismixedwith
theoxidizingagentandgasifiedinapowderflameathightemperaturesgenerally
exceeding1200◦C.ThisallowsproductionofagasrichinCOandH2thatisnearly
tar-freeandhasaverylowmethanecontent.Aproperlydesignedandoperated
entrainedflowgasifiercanhaveacarbonconversionratecloseto100%[6].Another
advantageofthistypeofgasifieristhatthehightemperaturesandpressuresresult
inaleachresistancemoltenslag.

1.2.ThebioliqRProcess

Figure1.2:Pathwaysofsyngastochemicals[11]

6

R1.2ThebioliqProcess
About30milliontonsofdrystrawandwoodresiduesfromforestryareavailablein
Germanyperyearforenergeticusages[12].Thisisabout43%ofthetotalamount
ofbiomassproducedinGermany.Butthebulkyandinconvenientformofbiomass
isamajorbarriertoindustrialutilizationofthisenergysource.Unlikegasorliquid,
biomasscannotbehandled,stored,ortransportedeasily.Thisprovidesamajor
motivationfortheconversionofsolidbiomassintoliquidandgaseousfuels,which
canbeachievedthroughoneoftwomajorpaths:(1)biochemical(fermentation)
and(2)thermochemical(pyrolysis,gasification).
Basedonthethermochemicalconversionpath,atwosteppyrolysis/gasificationpro-
cesscalledbioliqR(biomassliquefaction)isdevelopedattheKarlsruheInstituteof
Technology(KIT),Germany[13].Inthefirststageoftheprocess,straworother
abundantlignocellulosicagriculturalby-productsareliquefiedbyfastpyrolysisat
500◦Candatmosphericpressureinaninertatmosphereusingatwinscrewreactor
[3].Thepurposeoffastpyrolysisistogainasmuchliquidpyrolysisoilaspossible
withlowyieldofcharandgas.Thepyrolysisoilisthenmixedwiththecharto
prepareaslurrywithupto90%oftheoriginalbiomassenergycontent[3].The
slurriespreparedindecentralfacilitiesaretransportedbyrailtoalargecentralplant
forgasificationandfuelproduction.
Thegasificationprocessisperformedbyutilizinganentrainedflowgasifierusing
technicalO2asoxidizingagent.Thegasificationtemperatureandpressurearehigh,
usuallyabove1200◦Cand30bar[13],toachieveatar-freesyngaswithlowmethane

1.3.ComputationalFluidDynamicsModeling

7

content.AftergascleaningandadjustmentoftheH2/COratiowiththewatergas
shiftreactionandCO2removal,thepuresyngascanbeconvertedoncatalystsinto
variousproductslikeFischer-Tropsch(FT)diesel,methanol,dimethylether(DME),
olefins,hydrogenorotherchemicals[14].
Figure1.3showstheprocessstepsofbioliqR.Afterfastpyrolysisofbiomassand
slurrypreparationinregionalplants,theslurryistransportedtolargecentralplants
forgasification,gascleaningandfuelsynthesissteps.

3.1

Figure1.3:ThebioliqRprocesssteps(adaptedfrom[15])

ComputationalFluidDynamicsModeling

Combustionandgasificationhavebeenaveryimportantpartoftheenergycon-
versionprocessesandremainkeytechnologiesfortheforeseeablefuture.Effective
andeconomicusageofenergyresourcesaswellasprotectingtheenvironmentby
producinglessCO2andotherpollutantsduringthermochemicalprocessesneedthe
employmentofefficientconversionprocesses.Significanteffortshavebeenfocused

1.4.ScopeofthisThesis

8

onthedevelopmentofnumericalmodelsofthermochemicalreactors(suchascom-
bustors,boilersandgasifiers).Duetoavailabilityofefficientcomputersystems
nowadays,thenumericalmodelingtechniquessuchasComputationalFluidDynam-
ics(CFD)methodsareusedinindustryaswellasinacademia.CFDsimulations
helptooptimizethesystemdesignandoperationandunderstandthephysicaland
chemicalprocessesinsideareactor.Theygivethenecessarypredictivecapacityfor
designingsuchsystems.CFDmodelinghasestablisheditselfasapowerfultoolfor
thedevelopmentofnewideasandtechnologies.Agoodmathematicalmodelcan
findoptimumoperatingconditions,identifyareasofconcernordangerinopera-
tion,provideinformationonextremeoperatingconditions(hightemperature,high
pressure)whereexperimentsaredifficulttoperformandhelpstobetterinterpret
experimentalresults.Lastbutnotleast,modelingcanaddressscale-upproblems
fromonesuccessfullyoperatingsizetoanotherandfromonefeedstocktoanother.
CommercialsoftwaresuchasANSYSCFX,ANSYSFluentandCFD2000,aswellas
non-commercialcodes,areavailableforCFDsimulations.Areviewandcomparison
ofthesecodesisgivenin[16]and[17].ANSYSFLUENT12.0softwareisusedin
theframeworkofthisstudy.
Inthefieldofbiomasscombustionandgasificationmodeling,onestillfacessignifi-
cantchallengesduetocomplexityofthebiomasscomposition.ManyCFDstudies
havebeenperformedfordifferentbiomasstoenergyconversionsystemssuchas
combustion[18,19]orgasificationsystems[20,21].Throughsuchsimulations,the
numberofexperimentsneededtocharacterizethesystemcanbeconsiderablyre-
duced.Thesimulationsproducespatialandtemporalprofilesofdifferentsystem
variablesthatareeitherimpossibletomeasureorareaccessibleonlybyexpensive
experiments.

1.4ScopeofthisThesis
Thisthesisisapartoftheproject”SynthesegaserzeugungdurchFlugstromvergasung
vonOel/Koks-SlurrysausBiomasse-GrundlagenzurProzessoptimierung”(Syngas
generationbyentrainedflowgasificationofbiomass-basedslurry-basicsofprocess
optimization)utilizingthebioliqRprocess.Withintheframeworkoftheproject,
thegasificationprocessinsidealabscaleentrainedflowgasifier[22]ismodeled
andtheCFDsimulationisdoneusingthecommerciallyavailablesoftwareANSYS
FLUENT12.0.
InordertoperformaCFDsimulationofaflow,thegoverningpartialdifferential
equationsneedtobesolvednumerically.Theseequationscomprisethemasscon-

1.4.ScopeofthisThesis

9

servationequation,themomentumconservationequation,theenergyconservation
equationand-forreactiveflows-thechemicalspeciesconservationequations,which
willbediscussedinChapter2.

Duetotheturbulentnatureoftheflowandthehightemperaturesinsideanentrained
flowgasifier,propertechniquesarerequiredtomodelturbulenceandthermalradi-
ation,which,inadditiontothemethodsforsimulatingfuelparticlemotion,willbe
discussedinChapter3.

Chapter4concentratesonthechemistryofgasificationandthemodelsforturbulence-
chemistryinteractions.Thedetailedchemicalreactionmechanismusedinthiswork
willbeintroducedhere.Modelsbasedonequilibriumchemistry,flamelet,andthe
eddydissipationconceptapproacheswillalsobediscussedinthischapter.

Thegoverningequationsandtheadditionalphysicalandchemistrymodelsaregen-
erallyintheformofpartialdifferentialequations,thesolutionofwhichrequires
numericalmethods.Thisthesisusesthefinitevolumemethodwhichwillbedis-
cussedinChapter5inmoredetail.

TheCFDsimulationresultsarepresentedinChapter6alongwiththeparameter
studiesconductedtoidentifytheeffectofdifferentoperatingparameterssuchas
inlettemperatures,oxidizercomposition,air-fuelratio,andpressureonthegasifi-
cationprocess.Furthermore,theeffectoftheturbulence-chemistrycouplingmodel
-discussedinchapter4-arestudiedhere.Attheendofthischapter,theresultsof
theslurrygasificationsimulationarepresented.

InChapter7,theimportantresultsandfindingsaresummarizedandtheperspective
forfutureworkandafewimprovementpropositionsaregiven.

2.GoverningEquations

Thebasicsetofequationsforcontinuousphaseflowcomputationcomprisesthe
massconservationequation,themomentumconservationequationandtheenergy
conservationequation.Thegoverningequationssetforageneral3-dimensionalfluid
flowisknownasNavier-Stokesequations.Theequationsdescribebothlaminarand
turbulentflows.Incaseofachemicallyreactingflow,thesystemateachpoint
canbecompletelydescribedbyspecifyingtemperature,pressure,densityandthe
velocityoftheflowaswellastheconcentrationofeachspecies[23].Thelatteris
computedfromcorrespondingchemicalspeciesconservationequations.

Inthischaptertheconservationequationsarepresented.Additionalequationsfor
modelingturbulence,chemistry,thediscretephase,andradiationarediscussedin
thefollowingchapters.

2.1MassConservationEquation

Thegeneralformofthemassconservationequation,alsoknownasthecontinuity
equation,iswrittenasfollows:

ρ∂∂t+∙(ρv)=Sm(2.1)
ThesourceSmisthemassaddedtothecontinuousphasefromthedispersedsecond
phase(e.g.,duetovaporizationofliquiddroplets).

2.2.MomentumConservationEquation

11

2.2MomentumConservationEquation
Themomentumequation,basedontheNewton’slawsofmotion,relatesthesumof
theforcesactingonafluidelementtoitsaccelerationwhichistherateofchangeof
momentuminthedirectionoftheresultantforce[24].
Themomentumconservationequationcanbewritteninthefollowingform[25]:

∂∂t(ρv)+∙(ρvv)=−p+∙(τ¯)+ρg+F(2.2)
wherepisthestaticpressure,ρgandFarethegravitationalbodyforceandex-
ternalbodyforces(e.g.,thatarisefrominteractionwiththedispersedphase[26]),
respectively.
Thestresstensorτ¯inEquation2.2isdefinedby:

τ¯=µ(v+vT)−2(∙vI)
3whereIistheunitymatrixandvTisthetransposeofv.

)3.2(

2.3EnergyConservationEquation
Basedonthefirstlawofthermodynamics,statingthattheinternalenergygained
byasystemmustbeequaltotheheatabsorbedbythesystemminusworkdone
bythesystem,onecanobtaintheequationofconservationofenergyinthegeneral
formasfollows:

N∂t∂(ρE)+∙(v(ρE+p))=∙λeffT−hjJj+(¯τ∙v)+Sh(2.4)
=1jwhereλeffistheeffectivethermalconductivitywhichwillbedefinedinChapter3.
ThefirstthreetermsoftherighthandsideoftheEquation2.4representheat
fluxduetoconductionaccordingtotheFourierlawofconduction,speciesdiffusion
andviscousdissipationduetonormalshearstresses,respectively.Thesourceterm
accountsforheatofchemicalreactions,radiationandinteractionwiththedispersed
phase[26](seechapters3and4).
Intheaboveequation,

E=h−ρp+v2/2

)5.2(

2.4.SpeciesConservationEquation

wheretheenthalpyhisdefinedas

Nh=Yjhj,
=1jwithYjbeingthemassfractionofspeciesjand

21

)6.2(

Thj=cp,jdT.(2.7)
TfreInthecaseofnonpremixedcombustion(Section4.2),assumingunityLewisnumber
(Le=1),thefollowingequationforthetotalenthalpyissolved
∂(ρH)+∙(ρvH)=∙λtH+SH(2.8)
ct∂pwherethetotalenthalpyHisdefinedas

nad

H=YjHj

)9.2(

T0Hj=Trefcp,jdT+hj(Tref)(2.10)
withhj0(Tref)beingtheenthalpyofformationofspeciesjatthereferencetempera-
e.rutTheLewisnumberwhichquantifiestheratioofthermaldiffusivitytomassdiffusivity
isdefinedas

(2.11)

λLe=.(2.11)
Dρcmi,p2.4SpeciesConservationEquation
Foreachchemicalspeciesi,aconvection-diffusionconservationequationistobe
solvedtocalculatethecorrespondingspeciesmassfraction(Yi).Thisequation
knownasspeciesmassconservationequationhasthefollowinggeneralform

∂∂t(ρYi)+∙(ρvYi)+∙Ji=Ri+Si(2.12)

2.5.Reynolds-andFavre-Averaging

31

whereRiistherateofproductionofspeciesiduetochemicalreactions(see
chapter4)andSiisanyothersourceterm.ThediffusionfluxJiofthespecies
iisgivenby

µtT
Ji=−ρDi,m+Yi−DT,i.(2.13)
TcStThediffusionfluxtermconsistsoftheregularmassdiffusiontermaccordingtothe
Fick’slawandathermaldiffusiontermaccordingtotheSoreteffect[23].Sctandµt
aretheturbulentSchmidtnumberandviscosity(discussedinchapter3)respectively.
Inturbulentflowsitisnotgenerallyrequiredtospecifydetailedlaminardiffusion
propertiesastheturbulentpropertiesoverwhelmthelaminarones[26].DT,iis
calledcoefficientofthermaldiffusionwhichisonlyimportantforlightspeciesand
lowtemperature[23].
WhenthesystemconsistsofNspecies,theequation2.12needstobesolvedfor
N−1speciesasaccordingtothedefinitionofYithesumofmassfractionsofall
speciesisunity.Thereforeforthelastspeciesthemassfractioniscalculatedasone
minusthesumofN−1solvedmassfractions[26].

2.5Reynolds-andFavre-Averaging
FullnumericalsolutionofNavier-Stokesequationisaverydifficulttaskformost
engineeringapplications.Insuchflowproblemswithturbulentnature,theinfor-
mationofinterestislimitedusuallytodeterminethemeanvaluesofquantitiesof
interest,somemeasuresfortheextendoffluctuationandsomemeasuretocorrelate
thesevariousquantities.
Theideaofaveragingconsistsinneglectingthewholesetofflowdetailsandconsider
thattheflowcanbedescribedasthesuperpositionofthemeanfieldandafluctuating
fielddefinedasthedifferencebetweentheinstantaneousandthemeanfield[25].
InReynoldaveraging(alsocalledtimeaveraging),eachquantityφiscomposedofa
meanandfluctuatingvalue.

φ=φ¯+φ,
tφφ¯=1φ(t)dt,
tφ0φ¯=0,

(2.14)

2.5.Reynolds-andFavre-Averaging

14

wheretφisalargeenoughtimetoaverageoutthefluctuationsinφ.
Substitutingexpressionsofthisformfortheflowvariablesintotheinstantaneous
Navier-Stokesequationsandtakingatime(orensemble)average,onegetstheso
calledReynolds-averagedNavier-Stokesequations(RANS).
AnotherformofaveragingtheequationsistheFavreaveraging(density-weighted
average),inwhichallfluidmechanicalquantitiesexceptpressurearemassaveraged.
Thusforaquantityφ,

φ=˜φ+φ,
ρφ˜,=φρ¯.0=ρφ

(2.15)

Inequations2.14and2.15,thebarindicatestheReynoldstimeaveragingwhereas
thetildedenotesmassaveraging.Adoubleprimeindicatesthefluctuationaboutthe
mass-averagedmeanandprimesignshowsfluctuationsfortimeaveragingmethod.
Favreaveraginghasconsiderableadvantagesinsimplifyingtheformulationofthe
averagedNavier-Stokesequationsinvariabledensityflows.Inthemomentumequa-
tions,butalsointhebalanceequationsforenergyandthechemicalspecies,the
convectivetermsaredominantinhighReynoldsnumberflows.Sincethesecontain
productsofthedependentvariablesandthedensity,Favreaveragingisthemethod
ofchoice[27].
ConservationequationsobtainedbyFavreaveragingareidenticalinformtothe
RANSequationsforconstantdensityflow,makingFavreaveragingonlyamathe-
maticalformalism[28,29].
Reynoldsaveragedequationsformassandmomentumconservation(equations2.1
and2.2)areasfollows:

∂∂tρ+∙(ρv¯)=Sm,

(2.16)

∂(ρv¯)+∙(ρv¯v¯)=−p¯+∙(τ¯)−∙(ρvv)+ρg+F.(2.17)
t∂ApplyingFavreaveragingtoequations2.1and2.2oneobtains

t∂∂ρ¯+∙(ρ¯v˜)=Sm,

(2.18)

2.5.Reynolds-andFavre-Averaging

51

∂(ρ¯v˜)+∙(ρ¯v˜v˜)=−p¯+∙(τ¯)−∙(ρ¯vv)+ρ¯g+F.(2.19)
t∂Equations2.17and2.19aresimilartoEquation2.2exceptforthethirdtermon
therighthandside(∙(ρvv)and∙(ρ¯vv))duetothefluctuationinturbulent
flows.Theseunknowncorrelationtermsneedtobemodeledtoclosetheequation

system(seeSection3.1).

3.PhysicalModels

Inthischapter,physicalmodelsaredescribedthat,inadditiontothegoverning
equationsdiscussedinchapter2,arerequiredforthesimulationofthefluidflow
insideanentrainedflowgasifier.Thesemodelstakeintoconsiderationtheeffectof
turbulence,thermalradiationandtheinteractionofthegasphasewiththeliquid
fuelphase.Thestandardk-εturbulencemodelwillbeintroducedtogetherwithan
improvedversionofit,calledrealizablek-εmodel,whichisusedinthisthesisfor
themodelingofturbulentfluidflowinthegasifier.Thediscreteordinatesmodel
willbediscussedinsection3.2whichisusedtosolvethethermalradiationtransfer
equation.Insection3.3,thediscretephasemodel(DPM),whichutilizestheEuler-
Lagrangeapproachtomodeltheliquidfuelphase,willbediscussedindetail.

3.1Turbulence

Hinze[29]hasdefinedaturbulentfluidmotionasanirregularconditionofflowwith
randomspatiotemporalvariationofvariousquantitiessothatstatisticallydistinct
averagevaluescanbediscerned.Turbulencecausesanenhancementinmixingand
accountsfortheflowregimeinmostofthecombustionapplications.
Asalreadydiscussedinsection2.5,eachquantityisdefinedasasumofanaveraged
valueandafluctuatingpart.Theaveragingmethodintroducesadditionalunknown
termsinthemomentumequation.ThesetermsarecalledReynoldsStresses,defined
sa

Reij=ρuiuj

3()1.

3.1.Turbulence

71

whichstemfromthemomentumtransferbyfluctuatingvelocityfield.Tomodel
thesestresses,theBoussinesqassumption[29]isusedstatingthat
−ρuiuj=µt∂ui+∂uj−2ρk+µt∂ukδij(3.2)
∂xj∂xi3∂xk
whereµtiscalledturbulent(eddy)viscositywhich,unlikethemolecularviscosity,is
notapropertyofthefluid.Theclosureproblemissolvedbyexpressingtheturbulent
viscosityintermsofknownorcalculablequantities.Therearedifferentmethodsto
modelturbulence,basedonthenumberoftransportequationssolvedtocalculate
µt,suchaszero-equationmodels,one-equationmodelsandtwo-equationsmodels.
Thezero-equationmodelisnowadaysobsolete[23].Intheone-equationmodels,as
thenamesuggests,onlyoneadditionaldifferentialequationissolvedtocalculatethe
turbulentviscosity.k-εandk-ωmodelsarenowadayscommontypesofturbulence
models,whichbelongtothetwo-equationmodelscategory.Thek-εmodel,which
definestheturbulentviscosityasafunctionofturbulentkineticenergykandits
dissipationrateε,isusedinthecurrentwork.Thek-ωmodelsolvestwotransport
equationsforturbulentkineticenergykandspecificdissipationω,whichcanbe
thoughtastheratioofεtok[30].

3.1.1Standardk-εModel
Thestandardk-εmodelwasfirstproposedbyLaunderandSpalding[31]andis
oneofthemostusedturbulencemodelsincomputationalfluiddynamicsduetoits
robustnessandreasonableaccuracyforawiderangeofflows.Itisasemiempirical
modelbasedontransportequationsforturbulentkineticenergykanditsdissipation
rateε.Inthederivationofthemodelitisassumedthattheflowisfullyturbulent,
andtheeffectsofmolecularviscosityarenegligible[26].Thereforeitisvalidonly
forfullyturbulentflows.
Thetransportequationsforturbulentkineticenergyanditsdissipationrateare
definedas

∂(ρk)+∂(ρkui)=∂(µ+µt)∂k+Gk−ρε+Sk(3.3)
∂t∂xi∂xjσk∂xj
∂∂∂µt∂εεε2
∂t(ρε)+∂xi(ρεui)=∂xj(µ+σε)∂xj+C1εkGk−C2ερk+Sε(3.4)
whereSkandSεarethesourcetermsforkandε,respectivelyandGkisthetermfor
theproductionofturbulentkineticenergyduetothemeanvelocitygradients.The

3.1.Turbulence

C1εC2εC1C2σkσεCµ
Standardk-ε1.441.92——1.01.30.09
Realizablek-ε——Eq.3.91.91.01.2Eq.3.12

Table3.1:Valuesofconstantsfork-εmodels([26,32])

81

empiricalvaluesforkandεPrandtlnumbers(σkandσε)aswellastheconstants
C1εandC2εarelistedinTable3.1.
ThetermGkisdefinedas

Gk=µtS2

withSbeingthemodulusofthemeanrate-of-straintensordefinedas

S=2SijSij.
Theturbulentviscosityiscomputedfrom

2kµt=ρCµε
whereCµhasaconstantvalueaslistedinTable3.1.

)5.3(

6.3()

)7.3(

3.1.2Realizablek-εModel
Thestandardk-εmodelhassomedeficienciessuchasanomalyaboutthespreading
rateofplanarandroundjets[32].Toovercometheseproblems,arecentdevelopment
ofk-εmodelisproposedbyShihetal.[32]calledrealizablek-εmodel.Itdiffers
fromthestandardk-εmodelintwoways.First,itcontainsanewformulationfor
theturbulentviscosity.Second,anewtransportequationforthedissipationratehas
beenderivedfromanexactequationforthetransportofthemean-squarevorticity
fluctuation.Thetermrealizablemeansthatthemodelsatisfiescertainmathematical
constraintsontheReynoldsstresses,consistentwiththephysicsofturbulentflows.
Theequationforturbulentkineticenergyisthesameasthatofthestandardk-ε
model(equation3.3).
Fortherateofdissipation(ε),thefollowingequationisproposed

3.1.Turbulence

91

∂∂∂µt∂εε2
∂t(ρε)+∂xj(ρεuj)=∂xj(µ+σε)∂xj+ρC1Sε−ρC2k+√νε+Sε(3.8)
whereC1isfoundtobeasimplefunctionofthetimescaleratiooftheturbulence
tothemeanstrain,η

htwi

ηC1=max0.43,η+5

()9.3

kη=Sε(3.10)
andSdefinedbyEquation3.6.Fortheproductionofturbulentkineticenergy(Gk),
thesameequation(Equation3.5)isusedasinthecaseofthestandardk-εmodel.
Asinthestandardk-εmodel,theturbulentviscosityisdefinedbyEquation3.7
2kµt=ρCµ(3.11)
εwherethevalueofCµisnotaconstantanymoreandisdefinedas

eerwh

nad

1Cµ=A0+AskU∗
ε

U∗=SijSij+Ω˜ijΩ˜ij

(3.12)

(3.13)

˜Ωij=Ωij−2εijkωk,(3.14)
Ωij=Ω¯ij−εijkωk,(3.15)
withΩ¯ijbeingthemeanrate-of-rotationtensorviewedinarotatingreferenceframe
withtheangularvelocityωk.ThemodelconstantA0=4.04andAsisgivenby[32]:
√As=6cosφ(3.16)

3.2.ThermalRadiation

eerwh

dna

φ=31cos−1(√6W),

W=SijSjkSki,
3˜SS˜=SijSij,

20

(3.17)

(3.18)
(3.19)

S=1∂uj+∂ui.(3.20)
ij2∂xi∂xj
ItcanbeseenthatCµisafunctionofthemeanstrainandrotationrates,theangular
velocityofthesystemrotationandtheturbulencefield.Themodelcoefficientsare
summarizedinTable3.1.
Therealizablek-εmodelhasbeenvalidatedforawiderangeofflowtypes.Its
performancehasbeenfoundtobesubstantiallybetterthanthatofthestandardk-ε
model[32].
Forbothstandardandrealizablek-εmodels,theeffectivethermalconductivityused
inenergyequation(Equation2.4)isdefinedas

λeff=λ+cPprµt(3.21)
twhere,fortheturbulentPrandtlnumberthevalueofPrt=0.85isused[26].

3.2ThermalRadiation
Allsubstancesemitandabsorbelectromagneticradiationcontinuously.Theemitted
radiationis,duetothemolecularandatomicagitation,associatedwiththeinternal
energyofthematerial[33].Thoseradiationsoccupyingintermediatewavelength
range(approximatelybetween10−1µmand103µm)arecalledthermalradiation
.]33[Theimportanceofthermalradiationforgasificationandcombustionpropertiesisits
dependenceontemperature.Generallyforconductionandconvectionheattransfer,

3.2.ThermalRadiation

12

theheatfluxdependsonthefirstpoweroftemperaturedifference.Thethermal
radiativeheatflux

Qrad=σ(T4max−Tmi4n)(3.22)
showsthatthetransferofenergydependsonthedifferencebetweentheabsolute
temperatureseachraisedtoapoweroffour.Thisdepictsthattheradiationheat
transferwillbeimportantathighabsolutetemperaturedifferencelevels.Conse-
quently,onehastoconsiderthermalradiationincombustionprocesses.

3.2.1RadiativeTransferEquation
Radiationalongacertainpathisenhancedbyemissionandbyscatteringfromother
directionsandisattenuatedbyabsorptionandscattering.Employingtheseconcepts
(see[33]),anequationgoverningtheradiationintensityalongapaththrougha
mediumisdeveloped.ThisequationiscalledRadiativeTransferEquation(RTE)
havingtheform

dI(r,s)σT4σ4π
=−(a+σs)I(r,s)+a+sI(r,s)Φ(s∙s)dΩ.(3.23)
π4πds0Inthisequation,aistheabsorptioncoefficient,σsisthescatteringcoefficient,the
sumofwhich(a+σs)iscalledextinctioncoefficientand(a+σs)sisdefinedasoptical
thickness(opacity)ofthemedium.Furthermore,I,ΦandΩaretheradiation
intensity,phasefunctionandsolidangle,respectively.Thephasefunctionhasthe
physicalinterpretationofbeingthescatteredintensityinadirection,dividedbythe
intensitythatwouldbescatteredinthatdirectionifthescatteringwereisotropic
[33].Forthecurrentstudy,isotropicscattering,havingΦ=1,isconsidered.
Theintensitygivenbyequation3.23isthelocalradiationtravelinginasingle
directionperunitsolidangleandwavelength.Thefirsttermontherighthand
sideoftheequationindicatesthelossbyabsorption(includingthecontributionby
inducedemission)andscattering.Thesecondtermindicatesthegainbyemission
(notincludinginducedemission)andthelasttermontherighthandsideshowsthe
gainbyscattering(seeFigure3.1).

3.2.2DiscreteOrdinatesModel
Foranaccuratemodelingofhigh-temperaturesystems(suchascombustingsys-
tems),oneneedstosolvetheradiativetransferequationsimultaneouslywiththe

3.2.ThermalRadiation

Figure3.1:RadiativeHeatTransfer(adaptedfrom[26])

22

Navier-Stokesequations.ThismeansthatthemodeltosolvetheRTEmustbe
computationallyefficientenoughtopermititsinclusionintheothersubmodelsand
thenumericalprocedureusedfortheRTEmustbecompatiblewiththetransport
equationsfortheotherprocesses[34].Simplesolutionsareusuallynotpossiblefor
TE.RehtManydifferentmethodshavebeendevelopedforsolvingtheequationofradiative
transfer.TheyincludemethodssuchasP-Nmethod,MonteCarlomethod,discrete
transfermethodanddiscreteordinatesmethod(DOM).Eachofthesemethodshas
itsownrelativeadvantagesanddisadvantages,andnoneofthemissuperiortoothers
inallaspects.However,theDOMhasbeenwidelyrecognizedtobeoneofthemost
appropriatemethodsinhigh-temperatureapplicationsbecauseitsharesthesame
philosophyandcomputationalgridasthefluiddynamicsapproach[34]andcanbe
employedintheentirerangeofopticalthicknesses.
TheP-Nmethod[35]usesasetofmomentequationsoftheRTE(bymultiplyingthe
RTEbyvariouspowersofthedirectioncosinesoftheintensity)andanexpansion
oftheintensityintermsofthesphericalharmonics(denotedbyP)truncatedafter
aselectednumberofterms(N)[33].TheMonteCarloisamethodofstatistical
simulationandconsistsoffollowinganumberofindividualbundlesofenergyas
theytravelwithinthegeometryandareabsorbedorscattered[33,36].Withthe
discretetransfermethod[37],thetotalradiativefluxiscalculatedbyintegratingthe

3.2.ThermalRadiation

32

energycontributionalongraysemanatingfromtheradiativesourceandpointingto
anyselecteddirection[38].
Thediscreteordinatesmethod(DOM)isanextensionofamethodcalledtwo-flux
method[33]forstudyingradiativetransferinstellaratmosphereandlaterimple-
mentedbyFiveland[39,40]fortheanalysisofheattransferincoal-firedfurnaces.
Inmanystudies,DOMhasprovedtoproducegoodresultsinpredictingradiation
heattransfer[34,40,41,42,43]andisusedinthecontextofthisthesisastheRTE
solutionmethod.
Thediscreteordinatesmodelsolvestheradiativetransferequationforafinitenum-
berofdiscretesolidangles,eachassociatedwithavectordirectionsi(i=1,2,...,n)
fixedintheglobalCartesiansystem.Theintegralsoverthesedirectionsarereplaced
bynumericalquadratures.ThemodelconsiderstheRTEinthesdirectionasafield
equation[26].Thus,equation3.23iswrittenas

σT4σs4π
∙(I(r,s)s)=−(a+σs)I(r,s)+a+I(r,s)Φ(s∙s)dΩ.(3.24)
π4π0Incaseofthepresenceofaseconddiscretephaseintheflow,equation3.24is
modifiedasfollows

σT4σ4π
∙(I(r,s)s)=−(a+ap+σp)I(r,s)+a+Ep+sI(r,s)Φ(s∙s)dΩ.(3.25)
π4π0Intheaboveequationothersourcesofscatteringinthegasphaseareneglected.
ap,Epandσpareequivalentabsorptioncoefficientofparticles,equivalentparticle
emissionandequivalentparticlescatteringfactor,respectively,definedby[26]

NApnap=Vlim→0pn,(3.26)
V=1nEp=limpnApn,(3.27)
NσTpn4
V→0n=1πV
NApnσp=Vlim→0(1−fpn)(1−pn),(3.28)
V=1nwherepn,Apn,Tpnandfpnareemissivity,projectedarea,temperatureandscatter-
ingfactorofparticlen.

3.3.DiscretePhaseModel

42

Thesolidangleof4πaroundapointatanyspatiallocationisdividedintoseveral
sectors.Thesizeofeachsectorisdeterminedfromquadratureschemes.Thetrans-
portequationisthensolvedforthissetofdiscretedirections,representedbyits
directioncosines,spanningthetotalsolidangle[42].

3.2.3RadiationinReactiveFlow
Flamescanbeclassifiedasluminousandnonluminous.Theradiationfromthe
nonluminousfractionofthecombustionproductsisfairlywellunderstood[33].In
thecaseofhydrocarboncombustion,theradiationisfromtheCO2andH2Obands
intheinfrared.Luminousradiation,whichismostlyduetosootparticles,isnot
thesubjectofthepresentwork.
Therearedifferentmethodsforspecifyingabsorption-emittanceoftheradiating
gases(see[33]and[44]formoredetailabouttheavailablemethods).Oneacceptable
compromisebetweentheverysimplemethodofGrayGasesandcompletemodels,
takingintoaccounttheparticularabsorptionbands,isthesocalledWeightedSum
ofGrayGasesModel(WSGGM).Inthismodelthegasisassumedtobehavelikea
mixtureofgraygasesandatransparentmediumtoaccountforthewindowsbetween
theabsorptionbands[33].Inthismodelthetotalemissivityoverthedistancesis
calculatedas

I=a,i1−e−κips.(3.29)
=0iTheweightingfactora,idependsontemperatureandisdefinedin[45]as

Ja,i=b,i,jTj−1(3.30)
=1jwhereb,i,jaretheemissivitygastemperaturepolynomialcoefficients,whichtogether
withκi,aredeterminedbycurvefittingoftheexperimentalvaluesofemittanceof
CO2,H2Oandamixtureofthesetwogases(see[33]and[44]).
TheWSGGMisexaminedindetailin[46].Goodresultsareobtainedwithasub-
stantialreductionincomputationtime.

3.3DiscretePhaseModel
Therearebasicallythreenumericalmethodstosolvethedispersedmultiphaseflows
[47].ThesemethodsareknownasEuler-Lagrange,Euler-EulerandPDFmethods.

3.3.DiscretePhaseModel

52

TheDiscretePhaseModel(DPM),whichisappliedinthiswork,usestheEuler-
Lagrangeapproach[26].Thisapproachtreatsthecontinuousphase(fluidphase)as
acontinuumandtheparticlesasdiscreteentities.Forthecontinuum,theNavier-
Stokesequations,discussedinChapter2,aresolved,whilethedispersedphaseis
solvedbytrackingalargenumberofparticlesordropletsthroughthecalculated
flowfield[48].Thedispersedphasecanexchangemomentum,mass,andenergy
withthefluidphase[26].TheEuler-Lagrangeapproachisthemostpopularmodel
tosolvemultiphaseflows[47].
InthePDFmethods,thestateoftheflowateachpositionandtimeisdescribedby
aProbabilityDensityFunction(PDF),whichcanbeaonevariableorajointmulti-
variablePDF.ThetransportequationofthePDFisdeducedfromtheNavier-Stokes
equations[49,50].

3.3.1ParticleMotionTheory
Newton’ssecondlawofmotionisthegoverningequationofmotionoftheparticles
intheDPM.Accordingtothislaw,thesumoftheforcesactingonaparticleis
responsibleforitsacceleration.Theequationofmotioncanbewrittenas

dupgx(ρp−ρ)
=FD(u−up)++Fx(3.31)
ρdtpwhereFD(u−up)isthedragforcebasicallyduetothefrictionaleffectsasdefined
belowandFxisanyotherforceactingontheparticle,bothperunitparticlemass.Fx
canbethevirtualmassforcenegligiblewhenρ<ρp,forceduetopressuregradient
inthefluid,thermophoreticforceorBrownianforce,whichareallneglectedinthe
currentstudy.Thedragforceisoftendominatingthemotionoftheparticle[47].
Thesecondtermontherighthandsideofequation3.31isduetothebuoyancy
(basedonArchimedes’sprinciple)andgravitationalforce.
FDisdefinedas

eRµC3DFD=4∙ρpd2(3.32)
pwhereReistheparticleReynoldsNumbertocharacterizetheeffectofdispersed
phaseontheturbulencevariationofthecarriergas,definedby

Re=ρdp|up−u|.
µ

(3.33)

3.3.DiscretePhaseModel

62

ThedragcoefficientCDisusedtomodelthedependencybetweenparticleandflow
condition.Thesphericaldraglawisconsideredinthisstudy,statingthatthedrag
coefficientcanbedefinedas

aa32CD=a1++2.(3.34)
eReRThecoefficientsa1,a2anda3aredefinedintheworkofMorsiandAlexander[51]
forseveralrangesofRe.
Inordertotakeintoaccounttheeffectofturbulenceonthedispersionoftheparti-
cles,thestochastictrackingmodelhasbeenused,whichemploystheinstantaneous
gasvelocity,u=u¯+u(t)alongtheparticlepathduringthecalculations.
Thetime,aparticlespendsinturbulentmotionalongitspathds,isknownas
integraltimescaleτTpresentedby

∞up(t)up(t+s)
τT=2ds(3.35)
u0pForsmalltracerparticlesthatmovewiththefluid(zerodriftvelocity),theintegral
timebecomesthefluidLagrangianintegraltimeτL,whichcanbeapproximatedfor
thek-εturbulencemodel[26],[52]as

kτL≈0.15.(3.36)
εAstochasticmethod(randomwalkmodel)isusedtodeterminetheinstantaneous
gasvelocity.Inthediscreterandomwalk(DRW)model,alsoknownaseddylifetime
model,thefluctuatingvelocitycomponentsarediscretepiecewiseconstantfunctions
oftime.Theirrandomvalueiskeptconstantoveranintervaloftimegivenbythe
characteristiclifetimeoftheeddies[26].
Inthismodel,thefluctuatingvelocitycomponentsui,thatprevailduringthelife-
timeoftheturbulenteddyaresampledbyassumingthattheyobeyaGaussian
probabilitydistribution,sothat

u=ζu2(3.37)
whereζisanormallydistributedrandomnumber.TheRMSvalueofthevelocity
fluctuationiscomputed(assumingisotropy)as

3.3.DiscretePhaseModel

72

ui2=23k.(3.38)
Theinteractiontimebetweenparticlesandeddiesisthesmalleroftheeddylifetime
τe,andtheparticleeddycrossingtimetcross.Thecharacteristiclifetimeoftheeddy
isdefinedas

τe=2τL

(3.39)

whereτLisgivenbyequation3.36.
Theparticleeddycrossingtimeisdefinedas
Ltcross=−τln1−e(3.40)
τ|u−up|
2whereτistheparticlerelaxationtimeτ=ρ18pdρνp,ameasureforhowaparticle
reactsafterasuddenflowvelocitychange,andLeistheeddylengthscale.The
particleinteractswiththefluideddyovertheinteractiontime.Whentheeddy
lifetimeisreached,anewvalueoftheinstantaneousvelocityisobtainedbyapplying
anewvalueofζinequation3.37[26].
Integrationoftheequation3.31yieldsthevelocityoftheparticleateachpointalong
thetrajectory.Thetrajectoryitselfcanbecalculatedbysolving

dxdt=up.(3.41)
Equations3.31and3.41areasetofcoupledordinarydifferentialequations,the
numericalsolutionofwhichwillbediscussedinchapter5indetail.

3.3.2HeatandMassExchange
Theparticlescanexchangeheatandmasswiththecontinuousphase.Basedon
theparticletype,theseexchangesmightbedifferentresultingindifferentheatand
masstransferrelationships,alsocalledlaws.Theparticlesconsideredinthisstudy
areliquiddroplets.Thedropletscanundergodifferentlaws(e.g.inertheating,
evaporationorboiling)accordingtothephysicalconditionofthecontinuousphase.
Thesecontaininertheatingoftheparticle,evaporationandboiling.
UnlesstheparticletemperatureTpislessthantheevaporationtemperatureTvap,
theparticleexchangesheataccordingtothefollowinglaw

3.3.DiscretePhaseModel

82

dTmpcpp=hAp(T∞−Tp)+pApσ(θR4−Tp4)(3.42)
dtwheremp,cp,Apandparethemass,heatcapacity,surfaceareaandemissivityof
theparticle,correspondingly,andhistheconvectiveheattransfercoefficient.
Thisequationisderivedbyasimpleenergybalanceoftheparticle,assumingthat
theparticleisatauniformtemperaturethroughout.Thefirsttermontheright
handsideoftheequationdenotestheconvectiveheattransferandthesecondterm
4/1indicatesradiationheattransferwithθR=4Gσbeingtheradiationtemperature
andG=Ω=4πIdΩbeingtheincidentradiation(seesection3.2formoredetails).
Equation3.42isintegratedintimeusinganapproximate,linearizedformthatas-
sumesthattheparticletemperaturechangesslowlyfromonetimevaluetothenext
:]62[

dT
mpcpp=Ap−h+pσTp3Tp+hT∞+pσθR4.
dtIntegratingtheaboveequationyields

Tp(t+Δt)=αp+[Tp(t)−αp]e−βpΔt
withΔtbeingthetimestepand

dna

hT∞+pσθR4
=αph+pσTp3(t)

(3.43)

(3.44)

(3.45)

Ap(h+pσTp3(t))
βp=mc.(3.46)
ppRanzandMarshall[53,54]proposedthefollowingequationforcalculatingthecon-
vectivetransfercoefficient

hdNu=p=2.0+0.6Red1/2Pr1/3(3.47)
k∞wherek∞isthethermalconductivityofthecontinuousphaseandPr=kcpµisthe
∞Prandtlnumberofthecontinuousphase.TheparticleReynoldsnumberisdefined
inequation3.33.

3.3.DiscretePhaseModel

92

Theheattransferbytheparticleasittraverseseachcomputationalcellappearsas
asourceorsinkofheatinsubsequentcalculationsofthecontinuousphaseenergy
equation.Inthiscase,dropletsdonotexchangemasswiththecontinuousphase
anddonotparticipateinanychemicalreaction[26].
Whenthetemperatureoftheparticleisbetweentheevaporationtemperatureand
boilingtemperatureTvap≤Tp<Tbp,thedropletexchangesheatandmassaccord-
ingtotheevaporationlaw.Duringthislaw,therateofvaporizationisgovernedby
gradientdiffusion,withthefluxofdropletvaporintothegasphasedefinedas:

Ni=kc(Ci,s−Ci,∞)(3.48)
wherekcisthemasstransfercoefficientandthetermintheparenthesesindicates
thedifferenceinvaporconcentrationbetweenthedropletsurfaceandthebulkgas.
Themasstransfercoefficientkciscalculatedusingthefollowingcorrelationforthe
Sherwoodnumber[53,54]:

ShAB=kcdp=2.0+0.6Red1/2Sc1/3(3.49)
Dmi,µwiththeSchmidtnumberdefinedasSc=ρDi,mandDi,mbeingthediffusioncoeffi-
cientofvaporinthebulk.
Forthecalculationofthevaporconcentrationatthedropletsurface,assumingthat
thepartialpressureofvaporattheinterfaceisequaltothesaturatedvaporpressure
atthedroplettemperature,psat(Tp),thefollowingequationcanbeused.

C=psat(Tp).
si,TRpThevaporconcentrationinthebulkgasisdefinedby

(3.50)

pCi,∞=XiRT(3.51)
∞whereXiisthelocalbulkmolefractionofspeciesiandRistheuniversalgas
constant.
Thevaporfluxcalculatedbyequation3.48isusedasasourcetermforspeciesiin
thespeciestransportequation(seeequation2.12).
Themassofthedropletreducesaccordingto

3.3.DiscretePhaseModel

03

(3.52)

mp(t+Δt)=mp(t)−NiApMw,iΔt(3.52)
whereMw,iisthemolecularweightofthespeciesi.
Theheattransferequationaccordingtothislawissimilartoequation3.42withan
additionaltermduetodropletevaporation

dmdTmpcpp=hAp(T∞−Tp)+pApσ(θR4−Tp4)+phfg(3.53)
dtdtwherehfgisthelatentheatofevaporation.Theheattransferredaccordingtothis
lawbecomesasourceofenergyduringsubsequentcalculationsofthecontinuous
phaseenergyequation.
WhenthetemperatureofthedropletreachestheboilingtemperatureTbp,the
dropletstartstoboilandthedropletboilinglawisusedtopredicttheheatand
massexchangewiththecontinuousphase.Accordingtothislawandassumingthat
thedroplettemperatureremainsconstantduringboiling,equation3.53ismodified
tocalculatetheboilingrate:

ro

−dmphfg=hAp(T∞−Tp)+pApσ(θR4−Tp4)(3.54)
dt

−d(dp)=2k∞Nu(T∞−Tp)+pσ(θR4−Tp4).(3.55)
dtρphfgdp
Usingequation3.47fortheNusseltnumberandanempiricalvalueforthePrandtl
numberintheaboveequation[26],itbecomes

√d(d)2k∞1+0.23Red
dtρphfgdp
−p=(T∞−Tp)+pσ(θR4−Tp4).(3.56)
Aslongasthedropletboilinglawgoverns,theenergyrequiredforvaporization
appearsasa(negative)sourcetermintheenergyequationforthegasphaseand
theevaporatedliquidentersthegasphase.

3.3.DiscretePhaseModel

13

Figure3.2:Mass,momentumandheatexchangebetweendiscreteandcontinuous
phases(adaptedfrom[26])

3.3.3CouplingwiththeContinuousPhase
Basedontheparticlevolumefraction,Elgobashi[55]hasdefineddifferentclassesof
interactionsbetweenthedifferentphases.Whenthediscretephasehasanegligible
effectontheturbulenceofthecontinuousphase,onetalksaboutone-waycoupling.
Whentheparticlevolumefractionincreases,feedbackofthedispersedphaseonthe
propertiesofthecontinuousphasefluiddynamicsmustalsobetakenintoaccount,
whichisknownastwo-waycoupling.Inthecaseofdenseflows,particle-particle
interactionshavetobeconsideredaswell.Thisclassofinteractionsisknownas
four-waycoupling.Forthisstudyatwo-waycouplingistakenintoconsideration
basedonthecriteriaof[55].Inthiswaybothphasesexchangemass,momentumand
heatwitheachother.Thisinterphaseexchangefromtheparticletothecontinuous
phaseisdepictedqualitativelyinFigure3.2.
Thetransferofmass,momentumandheatfromthecontinuousphasetotheparticle
iscomputedbydeterminingthechangeincorrespondingvariablesoftheparticleas
itpassesthrougheachcontrolvolume.
Themasschangeiscomputedas

mΔpM=Δmm˙p,0(3.57)
0,pThismassexchangeappearsasasourceterminthecontinuityequationandalsoin
thecorrespondingspeciesconservationequation.

3.3.DiscretePhaseModel32
Themomentumexchangeiscomputedas
3µCDRe
F=∙2(up−u)+Fotherm˙pΔt(3.58)
dρ4ppwherethesummationisoverallthecontrolvolumes(seesection5.1)thattheparticle
passesthroughandFotherisanyforceotherthandragforce.Thismomentumforce
isusedasasourceterminthecontinuousphasemomentumequation.
Intheabsenceofchemicalreactionsofparticles,thefollowingequationisusedto
computetheheatexchangewithcontinuousphase:
m˙p,0Tp,outTp,in
Q=Hlat,ref(mp,out−mp,in)−mp,outcppdt+mp,incppdt(3.59)
mp,0TrefTref
wherethesubscriptsinandoutareforcellentryandexitcorrespondinglyand
subscript0indicatestheinitialvalueofthevariable.Thelatentheatatreference
conditionsHlat,ref,isdefinedasthedifferencebetweenliquidandgasstandardfor-
mationenthalpiesandcanberelatedtothelatentheatattheboilingpointHlat,
usingthefollowingequation:
TbpTbp
Hlat,ref=Hlat−cpgdt+cppdt(3.60)
TrefTref
withTbpandTrefbeingtheboilingpointandreferencetemperaturescorrespondingly.
Inthecaseofchemicalreactionsofparticles,afractionoftheenergyproducedby
thereactionsisusedadditionallyasaheatsourceforthecontinuousphaseandthe
restofitisabsorbedbytheparticledirectly[56].

delsMotryChemis4.

Theuseofglobalreactionsinreactiveflowproblemsdoesnotcompletelytakeinto
accounttheeffectsofthechemicalintermediates.Ontheotherhand,adetailed
descriptionofchemistrygivesadeeperinsightintoreactiveflowprocessessuchas
combustionandgasificationbutrequiresoftenaprohibitiveamountofcalculation
time.Thedetailedchemicalmechanismrequiredtodescribesuchprocessescontains
typicallyhundredsofchemicalspeciesinthousandsofelementaryreactions.Inthis
chapterthebasicsofchemicalkineticsandreactionmechanismdevelopmentare
.ssedscuid

InordertoperformaCFDsimulationofareactiveflow,chemistrymodelsshouldbe
usedtogetherwithotherfluidmechanicalsubmodels.Incaseofveryfastchemistry,
thechemistrycanbedecoupledfromtheflowandtheChemicalEquilibriummodel
canbeused.Forsuchcasesthemolecularspeciesconcentrationsandtemperature
arefunctionsofonlyoneprogressvariable,i.e.themixturefraction.IntheFlamelet
model,twoprogressvariablesarerequiredtofullydescribethesystem.Thesetwo
variablesaremixturefractionandscalardissipationrate.Theflameletmodeluses
alsotheprincipleofdecouplingchemistryfromthefluidflowbuttakesthenonequi-
libriumeffectsintoconsiderationbyusingthesecondprogressvariable.Thesetwo
modelswillbediscussedlaterinthischapter.

Attheendofthechapter,theEddyDissipationConcept(EDC)willbeintroduced
whichisachemistry-turbulenceinteractionmodelthatconsidersdetailedchemical
reactionmechanismsinturbulentreactiveflows.Here,thechemistryandfluidflow
calculationswillnotbedecoupled.EDCisacomputationallyexpensivemodeland
shouldbeusedwheretheassumptionoffastchemistryisnotvalid.Theassumption
offastchemistryandaccordinglydecouplingofchemistryfromthefluidflowis

4.1.ChemicalReactionMechanism

43

basedonthecomparisonofthetimescalesofchemistryandphysicalprocesses.The
chemicalreactionstypicallycoveratimerangefrom10−10stomorethan1s[23,27].
Thephysicalprocesseslikemoleculartransport,ontheotherhand,coveramuch
smallerrangeascanbeseeninFigure4.1.

Figure4.1:TimeScalesinaReactiveFlow[23]

4.1ChemicalReactionMechanism
AchemicalreactionwithNchemicalspeciescanbedescribedby:

NkfN
νiAi−→νiAi(4.1)
=1i=1iwhereνiandνiarethestoichiometriccoefficientsforreactantsandproducts,Ai
denotesthechemicalspeciesiandkfistheratecoefficient.Thereactionrateof
creation/destructionofspeciesicanbewrittenas

Ndci=(νi−νi)kfcini,f(4.2)
dt=1iwithcibeingtheconcentrationofspeciesiandni,fbeingthereactionorderwith
respecttothisspecies.
Forthereversereactionof4.1,havingaratecoefficientofkr,thereactionrateis
definedas

Ndci=−(νi−νi)krcini,r
dt=1i

()3.4

4.1.ChemicalReactionMechanism

53

whereni,risthereactionorderoftheithproductspecies.
Atchemicalequilibrium,theforwardandreversereactionshavethesamerate,i.e.

NN
(νi−νi)kfcini,f=(νi−νi)krcini,r(4.4)
=1i=1imeaningthatnonetreactioncanbeobservedonamacroscopiclevel.Theequi-
libriumconstantKc,whichrepresentstherelationbetweentheforwardandreverse
reactionsisobtainedas

NKc=kf=ci(νi−νi)(4.5)
kr=1iTheequilibriumconstantforthejthreactioniscomputed,basedonthechangein
Gibbsfreeenergy,from

NΔS0ΔH0p(νi,j−νi,j)
Kc,j=expj−j0i=1(4.6)
TRTRRwherep0istheatmosphericpressureandRistheuniversalgasconstant.Thevalues
ofΔSj0andΔHj0beingtheentropyandenthalpyofreactionatstandardconditions,
respectively,arecalculatedfromthermodynamicaldatabases(forexamplefrom[57]
or[58]).Thereversereactionratecoefficientkrcanthenbedeterminedfromkfand
theequilibriumconstantcalculatedfromEquation4.6.
Thereactionratecoefficientkdependsstronglyontemperatureinanonlinearman-
ner[23].Arrheniusgaveanimpiricalexpressionfortheformofthisdependencein
1889as[59]

Eak=A∙e−RT.(4.7)
Thepre-exponentialfactorAintheaboveequationcanbeafunctionoftemper-
atureaswell[23].Therefore,thefollowingexpressionisusedtocalculatetherate
coefficient:

k=ATb∙e(−RETa),(4.8)
whereAandbarethepre-exponentialfactorandtemperatureexponent,respectively.
TheactivationenergyEacorrespondstoanenergybarriertobeovercomeduring

4.1.ChemicalReactionMechanism

63

thereaction.Itsmaximumvaluecorrespondstothebondenergiesinthemolecule,
butcanbemuchsmallerifnewbondsareformedsimultaneouslyastheoldbonds
break[23].
Undercertainconditionsforsomedissociation/recombinationreactions,thereac-
tionratesdependstronglyonpressureaswellastemperature[23].Thepressure
dependenceofthesesocalledfall-offreactionsisdescribedbytwolimitingsitua-
tions;highpressureandlowpressurelimits.Forbothlowpressurelimit(k0)and
highpressurelimit(k∞),theratecoefficientsareintheformofEquation4.8.The
ratecoefficientsforthesetwolimitsarethenblendedtoproduceasmoothpressure
dependencerateexpression.AnoftenusedformalismistheF-Centertreatment
ofTroe[60,61].Inthismethodthescaledratecoefficientkkisexpressedasthe
∞productoftheLindemann-Hinshelwoodformula[62]andafactorF:
k=prF(4.9)
k∞1+pr

whereprisdefinedas

pr=k0[M](4.10)
k∞with[M]beingtheconcentrationofthecollisionpartner.FiscalledtheLorentzian
broadeningfactorwhichisusedtoreducethesystematicerrorsassociatedwiththe
Lindemann-Hinshelwoodformulainthepressurefall-offrange[63]andisgivenby
logp+c2−1
logF=logFcent1+n−d(lorgpr+c)(4.11)
eerwh

eerwh

c=−0.4−0.67logFcent
n=0.75−1.27logFcent
14.0=d

(4.12)

andFcentdescribingthecenterofthefall-offrangeasafunctionoftemperature

∗∗TTTFcent=(1−a)exp(−∗∗∗)+aexp(−∗)+exp(−).(4.13)
TTTTheparametersa,T∗∗∗,T∗andT∗∗aswellastheArrheniusparametersforthelow
andhighpressurelimitsarespecifiedforeachpressuredependentreaction.

4.1.ChemicalReactionMechanism

4.1.1ReactionMechanismDevelopment

73

Thechemistryofcombustionandgasificationisdescribedbychemicalreactionmech-
anismscontainingtensofspeciesinhundredsofreactions.Areactionmechanism
isdefinedasacompletesetofelementaryreactionstogetherwiththeirratecoeffi-
cients.Theinteractionoftheseelementaryreactionsproducestheoverallbalanced
stoichiometricchemicalequationoftheglobalreaction.Onthecontrarytoglobal
reactions,elementaryreactionsoccuronamolecularlevelexactlyinthewaywhich
isdescribedbythereactionequation[23].
Forthecurrentstudy,areactionmechanismisdevelopedforthehightemperature
gasificationofethyleneglycol[64].Ethyleneglycol(HOCH2CH2OH)isusedasnon-
toxicmodelfuelforthepyrolysisoilbyKarlsruheInstituteofTechnology(KIT),
whereexperimentalmeasurementswereperformed.Thebasemechanismisanim-
provedversionofapreviouslyvalidatedC1-C4mechanism[65],consistingof61
speciesand778elementaryreactions.Thismechanismisenhancedbyreactionsfor
ethanol[66].Reactionconstantsofreactionswithethyleneglycolanditsproducts
areimplementedbasedonexperiments,similarreactionschemesorestimatedusing
analogymethods.Themodifiedreactionschemeconsistsof80speciesand1243
elementaryreactions[64].Fordetailsondevelopmentandvalidationoftheethylene
glycolreactionmechanism,thereaderisreferredto[64].

4.1.2ReactionMechanismSimplification

ThemainproblemintheuseofdetailedreactionmechanismsinCFDsimulations
isgivenbythefactthatforeachspeciesofthemechanism,onespeciesconservation
equation(Equation2.12)needstobesolved.Thisisbondedwithagreatamountof
computerresourcesandcomputationtime.Dependingontheactualconditionsina
numericalstudy,manyofthereactionsandcorrespondingspeciescanbeneglected.
Here,analysismethodstoeliminatethesereactions,areofinterest.
Inordertosimplifythedevelopedreactionmechanism,twomethodsareused;sensi-
tivityanalysisandreactionflowanalysis[23].Thesensitivityanalysisidentifiesthe
rate-limitingreactionsteps.Forthedeterminationofcharacteristicreactionpaths,
reactionflowanalysisisperformed.
ThesimulationprogramHOMREA[67]isusedforthecomputationoftimede-
pendenthomogeneousreactionsystems.Thegoverningequationsarederivedfrom
theNavier-StokesequationsdiscussedinChapter2andaresolvednumericallywith
eitheramodifiedDASSL[68]oramodifiedLIMEX[69]solver,neglectingtheradia-
tiveheatfluxes.WiththecomputationalpackageMIXFLA[70],thesimulationof

4.2.NonpremixedCombustionwithEquilibriumChemistry38

flamespeedsandstructureofstationarypremixed1-dimensionallaminarflatflames
canbeperformed.
Thesensitivityanalysisofthereactionmechanismisdonefortheignitiondelay
timesandspeciesconcentrationsusingHOMREAandfortheflamevelocityusing
MIXFLA.Theratelawsforthereactionmechanismcanbewrittenasasystem
offirstorderordinarydifferentialequationswithratecoefficientsasparametersof
thesystem.TheprogramHOMREAcanalsobeusedforreactionflowanalysis,
wherethecontributionsofdifferentreactionstotheformationorconsumptionofa
chemicalspeciesareconsidered.

4.2NonpremixedCombustionwithEquilibrium
Chemistry

Thenonpremixedcombustionoccurswhencombustionandmixingoffuelandoxi-
dizeroccursimultaneously[23].Thistypeofcombustionisusedmostlyinindustrial
furnacesandburnersduetosafetyissues.Incaseofveryfastchemistry,themixed-
is-burntmodelcanbeusedassumingthatthecombustionoccursassoonasthe
fuelandoxidantmixwitheachother.Forsuchcases,themolecularspeciescon-
centrationsandtemperaturearefunctionsofonlyoneconservedscalar[71].There
areanumberofconservedscalarsthatcanbeusedtodescribethemixinginsuch
flows.Undertheassumptionofequaldiffusivity(meaningallspeciesdiffusealike),
themixturefractionfissuchavariable.Itisdefinedas

Zi−Zi2
f=Zi1−Zi2(4.14)
withZibeingthemassfractionofelementi.Thesubscripts1and2refertothe
twofeedstreams(fuelandoxidantstreamsrespectively).Mixturefractionvaries
between1forthefuelstreamand0theforoxidantstreamandcanbeinterpreted
asthemassfractionoriginatedfromthefuelstream.
Tocalculatethemixturefraction,twoequationsforitsmeanandvariancearesolved.
Thetransportequationformeanmixturefractionisasfollows

µ∂(ρf¯)+∙(ρvf¯)=∙(tf¯)+Sm,(4.15)
σt∂twhereSmisthesourcetermformasstransferfromtheliquidfueltothegasphase.
Forthemixturefractionvariance,f2,thefollowingconservationequationissolved

4.2.NonpremixedCombustionwithEquilibriumChemistry39

∂∂t(ρf2)+∙(ρvf2)=∙(σµttf2)+Cgµt(f¯)2−Cdρkεf2,(4.16)
wheretheconstantsσt,CgandCdtakethevaluesof0.85,2.86and2.0,respectively
.]62[Undertheassumptionofequilibriumchemistryforanonadiabaticcombustioncase
(e.g.liquidfuelcombustion)

φi=φi(f,H)

(4.17)

whereφirepresentsinstantaneousmassfraction,densityortemperatureandHis
theinstantaneousenthalpydefinedinChapter2.
Inturbulentflows,theaveragevaluesofvariablesarecalculated(Equations4.15and
4.16).Tocorrelatethesevaluestoinstantaneousvalues,thepresumedprobability
densityfunction(PDF)approachisselectedinthisworkbecauseofitssimplicity
[23,71].
UsingmixturefractionPDFp(f),themeanvalueofspeciesmassfraction,density
andtemperaturecanbecalculatedby

1¯φi=0p(f)φi(f)df.(4.18)
Aβ−functionPDFisemployedherebecauseofitsflexibility(seeFigure4.2)and
theabilitytocloselyrepresentexperimentalPDFs[23,27,72].
ThebetaPDFshapeisgivenbythefollowingfunctionoff¯andf2

eerwh

nad

α−1β−1
p(f)=ffα−1(1(1−−ff))β−1df,

¯¯α=f¯f(1−f)−1
2f

β=(1−f¯)

¯f(1−2¯f)−1f

.(4.19)

(4.20)

4(2.)1

4.3.FlameletModel

Figure4.2:ShapeofbetaPDFfordifferentsetsofparametersαandβ

04

4.3FlameletModel
Inturbulentflows,thelocalmicromixingrate(i.e.theinstantaneousscalardissipa-
tionrateχinEquation4.22)isarandomvariable.Thus,whilethechemistrymaybe
fastrelativetothemeanmicromixingrate,atsomepointsintheflowtheinstanta-
neousmicromixingratemaybefastcomparedwiththechemistry[72].Theeffects
causedbythefluctuationinmicromixingratecanbemodeledusingthelaminar
flameletconcept.
Theflameletconceptviewstheturbulentdiffusionflameasanensembleofthin,
laminar,locallyone-dimensionalflameletstructuresembeddedwithintheturbulent
flowfield[27,73].Commonlyacounterflowlaminardiffusionflameisusedtorepre-
senttheflameletinaturbulentflow.Asthenamesuggests,thecounterflowdiffusion
flameconsistsofopposed,axisymmetricoxidizerandfueljets(Figure4.3).
Thescalardissipationrate,characterizingthedeparturefromequilibrium,isdefined
sa

χ=2D|f|2(4.22)
withDbeingthediffusivity.Thescalardissipationrateaccountsfornon-equilibrium
effectscausedbybothconvectionanddiffusion.Itsrelationwiththestrainratea
ispresentedin[74].

4.3.FlameletModel

Figure4.3:Laminarcounterflownonpremixedflame(adaptedfrom[23])

14

Forcounterflowdiffusionflames,thecharacteristicstrainrateisdefinedasa=v/2d,
wherevistherelativevelocityofthefuelandoxidizerjetanddisthedistance
betweenjetnozzles[26].
Atthepositionwherethemixturefractionfisstoichiometric,thescalardissipation
canbecalculatedby[75]:

aχst=exp−2[erfc−1(2fst)]2(4.23)
πwhereerfc−1istheinversecomplementaryerrorfunctionandfstisthestoichiometric
mixturefraction.
Theinstantaneousstoichiometricscalardissipationχst,hasadimensionofs−1and
maybeinterpretedastheinversevalueofacharacteristicdiffusiontime.Inthe
limitwhereχst→0,thechemistrytendstoequilibrium.Theincreaseinχstdueto
aerodynamicstrainingincreasesnon-equilibrium.
Toaccountfortheeffectofvariabledensityacrosstheflamelet,thefollowingequa-
tionissolvedasanextensionoftheaboveequation[76]:

2χ(f)=a3(ρ∞/ρ+1)exp−2[erfc−1(2f)]2
4π2ρ∞/ρ+1
whereρ∞isthedensityoftheoxidizerstream.

(4.24)

4.3.FlameletModel

24

Inordertomodelthelaminarcounterflowdiffusionflame,theequationsaretrans-
formedfromthephysicalspacetothemixturefractionspace[26].Hereasimplified
setofequationsissolved[75]:

2ρ∂Yi=1ρχ∂Yi+Si(4.25)
∂t2∂f2
whereYiandSiarethemassfractionandreactionrateofspeciesi,respectively.
Thefirsttermintherighthandsideoftheequationtakestheeffectofinstantaneous
micromixingintoaccount.Generally,nearthestoichiometricsurface,bothtermsin
therighthandsideoftheaboveequationarelargeinmagnitudeandoppositeinsign
[27].Aquasi-stationarystateisthenquicklyreached,whereintheaccumulationterm
onthelefthandsideisnegligible.Thestationarylaminarflamelet(SLF)modelis
foundbysimplyneglectingtheaccumulationterminEquation4.25.TheSLFmodel
canbeused,yieldinggoodresults,forthepredictionofheatrelease,concentration
ofmajorchemicalcomponentsandevenOHconcentrations[77].
Thefollowingequationissolvedforthetemperature

∂T1∂2T11∂cp∂Yi∂T
ρ∂t=2ρχ∂f2−cpHiSi+2cpρχ∂f+cp,i∂f∂f(4.26)
iiwherecp,iandcparetheithspeciesspecificheatandmixtureaveragedspecificheat,
respectively.
Theturbulentflameismodeledasanensembleofdiscretelaminarflamelets.Since
inadiabaticcasesthespeciesmassfractionsandtemperaturearefunctionsofonly
mixturefractionfandscalardissipationχst,themeanvaluesoftheseparameters
canbedeterminedfromthePDFoffandχstas

1∞φ¯i=00p(f,χst)φi(f,χst)dfdχst.(4.27)
Inthisstudy,fandχstareassumedtobestatisticallyindependent,sothejoint
PDFp(f,χst)canbesimplifiedaspf(f)pχ(χst).Aβ−functionPDFisconsidered
formixturefractionasdiscussedinsection4.2.Forthesakeofsimplicity,the
fluctuationsinχstareignoredandadeltafunctionisused[26].

pχ=δ(χst−χst)
wherethemeanscalardissipation,χst,ismodeledas[75]

(4.28)

4.4.EddyDissipationConcept

34

εχst=2f2.(4.29)
kTheintegrationofEquation4.27ispreprocessedandstoredinlook-uptables.
Fornon-adiabaticsteadylaminarflamelets,theadditionalparameterofenthalpyis
required.However,thecomputationalcostofmodelingsteadyflameletsoverarange
ofenthalpiesareprohibitive,sosomeapproximationsaremade[26].Heatgain/loss
tothesystemisassumedtohaveanegligibleeffectonthespeciesmassfractions,
andadiabaticmassfractionsareused[78].Thisapproximationisnotappliedfor
thecasecorrespondingtoascalardissipationofzero.Suchacaseismodeledby
equilibriumchemistryassumptiondiscussedinsection4.2.

4.4EddyDissipationConcept
TheEddyDissipationConcept(EDC)model[79]canconsiderdetailedchemical
reactionmechanismsinturbulentreactiveflowsimulations.Itisanextensionof
theEddyDissipationModel[80].ThebasicideabehindEDCisthatthereactions
occurinregionswherethedissipationofturbulenceenergytakesplace.These
regionsoccupyasmallfractionoftheflow.Thesmallturbulentstructures(theso
calledfinestructures)haveacharacteristicdimensionsintheKolmogorovlength
scaleorderinoneandtwodimensions[81].TheKolmogorovscalesarethesmallest
scalesofturbulentmotionwithalengthscaleof[82]

andatimescaleof

ν31/4
=η,kε

.=τν1/2
ηεThelengthfractionofthefinestructuresisdefinedby

(4.30)

(4.31)

νε1/4
ξ∗=2.13772.(4.32)
kThefractionoftheflowoccupiedbythefinestructuresismodeledas(ξ∗)3.The
timescaleoverwhichthereactionstakeplace,iscalculatedas

ν1/2
τ∗=0.4082.
ε

(4.33)

4.4.EddyDissipationConcept

44

Thecombustionatfinescalesisassumedtooccurasaconstantpressurehomoge-
neousreactor,withinitialconditionstakenasthecurrentspeciesandtemperature
inthecell[26]andaresidencetimeτ∗.Thereactions,governedbyArrheniusrates,
aresolvednumericallyusingISATalgorithm(seesection5.3).ThesourcetermRi
fortheithspeciesconservationequation(Equation2.12)isthencalculatedby

ρ(ξ∗)2∗
Ri=τ∗[1−(ξ∗)3](Yi−Yi)(4.34)
wheresuperscript∗denotesfinescalequantities.Intheaboveequation,themass
exchangebetweenthefinestructuresandthesurroundingsismodeledas(ξ∗)2/τ∗
[83].Thefactor1/[1−(ξ∗)3]comesfromthecorrespondingequationforthemass
averagedmeanstateasdiscussedindetailin[81].

Typicalchemicalreactionmechanismscontaintensofspeciesinhundredsofre-
actions.Theordinarydifferentialequationsystemgoverningthecombustionpro-
cessisnormallystiffanditsnumericalsolutioniscomputationallycostlyandoften
unstable[23].Therefore,simulatingdetailedchemicalreactionschemesusingthe
EDCmodelneedsmorecomputationalresourcesthanequilibriumchemistryorthe
flameletmodel.Efficientnumericalproceduresarehencerequiredtodecreasethe
computationalresourcesrequiredtotreatthedetailedchemistryusingEDC.Inthis
thesis,theEDCmodelisusedinconjunctionwithISATprocedure.

5.NumericalModels

InordertoperformaCFDsimulation,thegoverningequationsandmodelsdiscussed
earlierneedtobetransferredtoanumericaldomain.Inthenumericaldomain,the
governingequationsarediscretizedandsolvedbycomputerprograms.Appropriate
numericalalgorithmsarerequired.Themethodusedinthisstudyisbasedonthe
FiniteVolumemethodwhichwillbediscussedinthenextsection.Attheendof
thechaptertheISATalgorithmusedtoacceleratechemistrycalculationswillbe
introduced.

5.1FiniteVolumeMethod

Fluiddynamicsisgovernedbypartialdifferentialequationsalreadydiscussedin
Chapter2.Thereareanumberofdifferentmethodstosolvethemnumerically.
Finitedifference,finitevolumeandfiniteelementmethodsareamongthoseusedin
theliterature.Inthefinitedifferenceapproach,thederivativesarewritteninfinite
differenceformusingtruncatedTaylorseriesexpansionsresultingincoupledalge-
braicequations[84].Themeshconfigurationsforthismethodmustbestructured
[85].Inthefiniteelementmethod,someformofweightedresidualofthegoverning
equationsisminimizedovereachfiniteelement[84].Theunderlyingprinciplesand
formulationsinfiniteelementmethodsrequiremathematicalrigorandaredescribed
indetailin[86].Bothfinitedifferenceandfiniteelementmethodsdonotexplicitly
enforcetheconservationprincipleintheiroriginalforms.Hence,themeshshould
befineenoughforacorrectnumericalsolutionoftheCFDproblem[84].Thefinite
volumemethod,ontheotherhand,usesanintegralrepresentationoftheconser-
vationequationstodevelopthealgebraicequations.Thismethodguaranteesthe

5.1.FiniteVolumeMethod

46

conservationpropertiesthroughoutthedomainandneedsnocoordinatetransfor-
mationsforunstructuredmeshesandcomplexgeometries[85].Finitevolumeand
finitedifferencemethodsareshowntobethepredominantmethodsinengineering
applications[87]althoughafaircomparisonbetweenthemethodsisdifficult.
TheFiniteVolumeMethodisusedintheframeofthisworktosolvethegoverning
equationsdescribingthecontinuousphase.Theintegralformoftheseequationsare
discretized.Inthismethod,thewholecomputationaldomainissubdividedintoa
setofnon-overlappingcellscalledcontrolvolumes(CV).Thegoverningequations
arethenappliedtoeachofthecontrolvolumestodeterminetheflowvariablesin
thecells.
Thevelocityfieldisobtainedfromthemomentumequation.Apressurebased
approachisusedherewhichcalculatesthepressurebysolvingapressureorapressure
correctionequationobtainedbymanipulatingcontinuityandmomentumequations
[26].Thistypeofsolverandthediscretizationoftheequationsarediscussedinthe
subsectionsthatfollow.

5.1.1PressureBasedSolver

Thepressurebasedsolutionmethodisaparticularformofamoregeneralmethod
calledProjectionMethod[88].Inthismethodapressureequationisderivedfrom
thecontinuityandthemomentumequationsinsuchawaythatthevelocityfield
satisfiesthecontinuity[26].Thismethodhasoftenbeenusedintheliteratureto
solvethegoverningNavier-Stokesequations([89,90,91]).Duetothenonlinearity
oftheequations,aniterativesolutionmethodisrequired.
Thepressurebasedmethodsforsolvingincompressibleflowshavebeenthemethod
ofchoiceinthelastdecades[92].Thesegregatedpressurebasedsolverisusedin
thisstudyinwhichtheequationsaresolvedsequentially.Thesolutionprocedure
foreachiteration,outlinedinFigure5.1,isasfollows:

1.Thefluidpropertiesareupdatedbasedoncurrentavailablesolution.

2.Themomentumequationsaresolvedsequentiallyusingupdatedvaluesofpres-
sureandmassfluxes.

3.Thepressure-correctionequationissolvedusingtheresultsofstep2.

4.Thevaluesofpressure,velocityfieldandmassfluxesareupdated.

5.Theequationforadditionalscalarssuchasenergy,species,turbulenceand
radiationaresolved.

5.1.FiniteVolumeMethod

Figure5.1:Pressurebasedsegregatedsolver

74

6.Thesourcetermsarisingfrominteractionswithdiscretephaseareupdated.
7.Convergenceischecked.

Forthefirstiteration,theinitialconditionsdefinedforthefluidpropertiesareused
andtheabovementionedprocedurestartsfromthestep2.Theiterationcontinues
untiltheconvergenceisobtained.
Thesegregatedalgorithm(inwhichtheequationsaresolvedinadecoupledmanner)
ismemoryefficient,becauseitstoresthediscretizedequationsinthememoryone
atatime.However,thesolutionconvergenceisrelativelyslow,inasmuchasthe
equationsaresolvedinadecoupledmanner[26,92].

5.1.2DiscretizationofEquations
Considerthedifferentialequationforthetransportofthescalarquantityφ.This
equationcanbewritteninintegralformforacontrolvolumeVas

∂ρφdV+ρφv∙dA=Γφφ∙dA+SφdV(5.1)
t∂VVwhereΓφisthediffusioncoefficientforφandSφindicatesthesourceofφperunit
volume.Discretizationandintegrationoftheaboveequationonthecontrolvolume
resultsinthefollowingequation:

NfacesNfaces
∂ρφV+ρfvfφf∙Af=Γφφf∙Af+SφV(5.2)
t∂ff

5.1.FiniteVolumeMethod

84

whereNfacesisthenumberoffacesenclosingthecellandφfistheamountofφ
convectedthroughtheface.
Forthesteadystatecaseconsideredinthisstudy,∂∂tρφV=0,andnotemporal
discretizationisrequired.Equation5.2canthenbewrittenintheform

Jfφf=Df+SφV(5.3)
ffwhereJfisthemassflowrateandDfshowsthetransportduetothediffusion
throughthefacef.Themassflowrateisdefinedfromthesolutionofcontinuity
andmomentumequations.
ThefacevalueofthescalarφiscalculatedusingaFirst-OrderUpwindscheme
indicatingthatthefacevalueφfisequaltothecellvalueofthescalarofthe
upstreamcell.Hence,

φf=φupwind

4.5()

Oneneedstodeterminethegradientφofthescalarφnotonlytocalculatevelocity
derivatives,butalsothesecondarydiffusionterms.Calculationofthegradientsis
basedonthedivergencetheoremstatingthatthegradientofφatthecellcenteris
definedas

(φ)0=1φ¯fAf(5.5)
Vfwherethesummationisoverallthefacesofthecellandthefacevalueofφis
obtainedbyarithmeticaveragingattheneighboringcell

φ¯f=φ0+φ1(5.6)
2ThediscretizationprocedureyieldsalinearizedformoftheEquation5.2forφat
thecellcenterintheform

apφp=anbφnb+bp(5.7)
nbwherethesubscriptnbindicatestheneighborcellsandaisthelinearizedcoefficient
forφ.Herethesummationisoveralltheneighborsnbofcellp.Similarequations

5.1.FiniteVolumeMethod

94

intheformoftheaboveequationhavetobewrittenforallthecellsinthedomain.
ThesystemofequationsissolvedusingaGauss-Seidellinearequationsolver[87]in
conjunctionwithanalgebraicmultigrid(AMG)methoddiscussedindetailin[87],
[93]and[94].
Bysettingφ=u,onecanobtainthediscretizedequationformomentuminthe
samemannerasdiscussedabove.Theequationhastheform

apu=anbunb+pfA∙iˆ+S
nbandthediscretecontinuityequationiswrittenas

)8.5(

NacesfJfAf=0(5.9)
fBothvelocityandpressurecomponentsarestoredatcellcenters.ComputingJf
byaveragingthecellvelocitiescausescheckerboarding[92].Thiscanbeavoided
byusingaschemesimilartothatproposedin[95,96].Amomentum-weighted
averagingisusedwithweightingfactorsbasedontheapcoefficientfromEquation
5.8.Themassflowratecanthenbewrittenas

ap,0vn,0+ap,1vn,1
Jf=ρf+df[(p0+(p)0.r0)−(p1+(p)1.r1)]=Jˆf+df(p0−p1)
ap,0+ap,1
(5.10)
wherep0,p1,vn,0andvn,1arethepressureandnormalvelocity,respectively,ofthe
cellsatbothsidesofeachface.Thetermdfisafunctionofa¯p,theaverageofthe
momentumequationcoefficientsforthecellsoneithersideoftheface.

5.1.3PressureVelocityCoupling
PressurevelocitycouplingisaccomplishedbyusingtheEquation5.10toachieve
aformulationforpressurethroughmanipulatingthecontinuityequation.Thisis
achievedbyusinganalgorithmcalledSemi-ImplicitMethodforPressure-Linked
Equations(SIMPLE)[95].Thealgorithmisbasedonthefinitevolumediscretization
onthestaggeredgridsemployedbythepresentwork.Itdescribestheiterative
procedurebywhichthesolutionsofthediscretizedequationsareobtained.
Foranarbitrarypressurep∗,themassflowrateobtainedfromEquation5.10is
writtenas

5.1.FiniteVolumeMethod

05

Jf∗=Jˆf∗+df(p0∗−p1∗)(5.11)
which,ingeneraldoesnotsatisfythecontinuityequation.Tosolvethis,atermJ
isaddedintheform

sothattheresultingmassflux

Jf=df(p0−p1)

(5.12)

Jf=Jf∗+Jf(5.13)
satisfiesthecontinuity.piscalledcellpressurecorrection.TheSIMPLEalgorithm
thensubstitutesthecorrectionequationinthediscretizedcontinuityequationto
obtainanequationforpressurecorrectionintheform

app=anbpnb+b
nbwherebisthenetflowratetothecellforthestarredcondition,

(5.14)

b=Jf∗Af.(5.15)
fIfb=0,thestarredconditionsatisfiesthecontinuityandnopressurecorrectionis
needed.Thereforethetermbrepresentsamasssourcewhichthepressurecorrections
mustannihilate[95].
Whenthepressurecorrectionequationissolved,thevaluesofcorrectedpressure
andmassfluxare

p=p∗+αpp

(5.16)

Jf=Jf∗+df(p0−p1)(5.17)
whereαpistherelaxationfactorforpressurehavingavaluebetween0and1.
Arelaxationmethodisusedtoacceleratetheconvergence.Largechangeinthe
variablescouldcausenumericalinstability.Therefore,thevariableφischangedas

5.2.IntegrationofParticleEquationofMotion

φ=φold+αφΔφ.

TheSIMPLEprocedurecanbesummarizedinthefollowingsteps:

51

(5.18)

1.Guessthepressurep∗.
2.SolveEquation5.11forstarredvelocities.
3.Solvepequation(Equation5.14).
4.CalculatepfromEquation5.16.
5.CalculatethevelocityfieldfromEquation5.17.
6.Treatthecorrectedpressurepasanewguessp∗,returntostep2andrepeat
untilconvergence.

Onecanusethemasssourcebasausefulindicatoroftheflowsolutionconvergence.
Theiterationsshouldbecontinueduntilthevalueofbbecomessufficientlysmall
everywhere.

5.2IntegrationofParticleEquationofMotion
Asalreadydiscussedinsection3.3,theparticlevelocityiscalculatedbyintegrating
Equation3.31andthetrajectoryiscalculatedbysolvingEquation3.41.Rearranging
Equation3.31toageneralform,oneobtains:

1dup=(u−up)+a(5.19)
τdtpwhereaistheaccelerationduetoallforcesotherthanthedragforce.
TwonumericaldiscretizationschemesareusedheretosolveEquation5.19numeri-
cally.UsingEulerimplicitdiscretization,oneobtains
nupn+Δta+τu
upn+1=Δtp.(5.20)
+1τpWhenapplyingatrapezoidaldiscretizationtoEquation5.19,thevariablesuandup
ontherighthandsidearetakenasaverages,andaasaconstant.Thesolutionwill
thenbeasfollows:

5.3.InSituAdaptiveTabulation

eerwh

un+1−un1
pp=(u∗−up∗)+a
τtΔp

25

(5.21)

u∗=1(un+1+un)
2up∗=1(upn+1+upn)(5.22)
2un+1=un+Δtupn.un
Combiningequations5.21and5.22,theparticlevelocityatthenewlocationn+1
iscomputedby

n+1upn(1−21τΔpt)+τΔpt(un+21Δtunp.un)+aΔt
up=1Δt.(5.23)
+1τ2pUsingtrapezoidaldiscretizationoftheEquation3.41,thetrajectoryoftheparticle
iscalculatedas

1xpn+1=xpn+Δt(upn+1+upn).(5.24)
2AcombinationoftheimplicitandtrapezoidalschemesisusedinANSYSFLUENT
andhenceinthisstudyaswell.Insituationswheretheparticleisfarfromhydro-
dynamicequilibrium,atrapezoidalschemeproducesbettersolution,whereaswhen
theparticlereacheshydrodynamicequilibrium,thehigherordertrapezoidalschemes
becomeinefficientandthemechanismswitchestoastableimplicitscheme[26].

5.3InSituAdaptiveTabulation
Asdiscussedinchapter4,detailedchemicalmodelstypicallyincludereactants,
productsandreactionintermediatesthatsumuptotensofspeciesresultingfrom
hundredsofreactions.Thecorrespondingreactiontimescalescanrangefrom10−10s
tomorethan1s[23].Reactionschemeswithawiderangeoftimescalesproducea
stiffnumericalsystemthatisdifficulttointegrate.
ForthereactionfractionalstepintheEDCmodel(seesection4.4),eachparticle
evolvesaccordingtothechemicalsourceterm:
)i(dφ=S(φ(i))(5.25)
dt

(5.25)

5.3.InSituAdaptiveTabulation

whereφistheparticlecompositionvector

φ=(Y1,Y2,...,YN,T)

35

(5.26)

withYkbeingthekthspeciesmassfraction.
DirectIntegration(DI)oftheabovedifferentialequationsiscomputationallyex-
pensivefordetailedreactionschemes.TocircumventthecostofDI,theInSitu
AdaptiveTabulation(ISAT)algorithm[97]isusedinthisstudy.
ISATisapowerfultoolthatenablesrealisticchemistrytobeincorporatedinmulti-
dimensionalflowsimulationsbyacceleratingthechemistrycalculations.Infull,the
methodis:insitu,unstructured,adaptivetabulationoftheaccessedregionwith
controlofretrievalerrors[97].
Inordertouseatabulationmethodforaparticularflow,itissufficienttotabulate
theaccesseddomain,ratherthanthewholeoftherealizabledomainwhichismuch
larger.Sincetheaccesseddomaindependsonmanyaspectsoftheflowincluding
thekinetics,thetransportprocessesandtheboundaryconditions,itisnotknown
beforeperformingthecalculation.Hence,thetableisbuiltupduringthereactive
flowcalculation.Eachentryinthetablecorrespondstoacompositionthatoccurs
inacellduringthecalculationandthecorrespondingS(φ(i)).Thisisreferredtoas
insitutabulation.
ThebasicideabehindISATmethodistointegratethegoverningequationusingDI
andthenstorethereactionmappingaswellassensitivityinformationinabinary
treedatastructureforlateruse[98].Forsubsequentcalculations,DIisavoidedfor
thepointsthatarewithinasmalldefineddistancefrompreviouslycalculatedpoints.
Here,thereactionmappingwillbeestimatedusingmultilinearinterpolation[98].
However,DIwillbeperformedwherethereactionmappingcannotbeinterpolated
withsufficientaccuracy.ThisideaisdepictedinFigure5.2andsummarizedas
ws:ollfo

•Onsubsequentcallthetableisqueried.
•CheckiftheinitialstatefallsinsideEllipsoidOfAccuracy(EOA).
•Ifyes,interpolateandretrievethemapping.
•Ifnot,aDirectIntegrationisperformed.
•CheckifthemappingfallswithinISATerrortolerance.

5.3.InSituAdaptiveTabulation

Figure5.2:KeystepsinvolvedinISATalgorithm[98]

•Ifyes,theEOAisgrown.
•Ifnot,anewtableentryisadded.

45

Atthestartofthesimulation,mostoperationsare“Addition”and“Growth”which
areslowduetoperformingDI.Later,asmorepointsinthecompositionspace
aretabulated,“Retrieve”becomesfrequentandhencetheCFDcalculationwillbe
accelerated.Typicalspeed-upfactorof100-1000isobtainedcomparedtoDI[97],
.]89[

6.ResultsandDiscussion

ThischapterfocusesontheCFDsimulationresultsofthegasificationofethylene
glycolwhichwasusedasamodelfuelforpyrolysisoilinthelab-scaleentrainedflow
gasifierREGA.Thegasifierandthegasificationconditionsaredescribedinsection
6.1ofthischapter.Thephysical,chemicalandnumericalmodelsdiscussedinpre-
viouschaptersareutilizedtoperformCFDsimulationsusingtheANSYSFLUENT
12.0code.Theresultedflowpatterns,temperatureprofilesandproductgascom-
positionsarepresentedandcomparedwithexperimentalmeasurementswherever
possible.Theresultsarepresentedinsections6.2and6.3.

Insection6.4,aseriesofsimulationsisperformedtostudytheeffectoftheboundary
andoperatingconditionsonthegasificationefficiencyandtheproductgascomposi-
tions.Oxidizerandfuelinlettemperatures,oxidizercomposition,air-fuelratio,and
thegasifieroperatingpressurearethefourvariablesusedforsensitivityanalyses.

Section6.5isdedicatedtostudytheeffectofchemistryonthegasificationprocess.
Inthefirstpart,threeversionsoftheethyleneglycolreactionmechanismareused
tostudytheeffectofreactionkineticsonthegasification.Inthesecondpart,the
chemistrymodelsdiscussedinchapter4,namelyequilibriumchemistry,flamelet
modelandeddydissipationconceptarecomparedwitheachotherandtheireffects
onthesimulationresultsarediscussed.

Attheendofthischaptersimulationresultsoftheslurrygasificationarepresented
withafocusontheeffectofchargasificationonthewholeprocess.

6.1.GasifierModelandSimulationConditions

65

6.1GasifierModelandSimulationConditions

ThemodeledgasifierinthisthesisisapilotscaleResearchEntrainedFlowGasifier
(REGA)whichisoperatedattheKarlsruheInstituteofTechnology(KIT).Itis
a60kWatmosphericentrainedflowgasifierhavingatotallengthof3mandan
innerdiameterof28cm.Itisequippedwithanexternalmixingburnernozzlefor
atomizationofslurrieswithair[99].Fuelandoxidizerenterthegasifieratthetop
throughtheburnerandthehotproductgasesexitatthebottomofthegasifieras
depictedinFigure6.1.Theelectricalheatingofthereactorwallsupto1200◦C
allowsadiabaticoperatingconditions[22].
Inthisstudy,ethyleneglycol(HOCH2CH2OH)servesasnon-toxicmodelfuelfor
pyrolysisoil,mainlybecauseofitssimilarC/H/O-ratioanditssimilarphysical
propertiestobiomassderivedliquidpyrolysisproducts[99,100].
Theethyleneglycoloxidizationreactionmechanism,developedbyHafner[64,66],
wassimplifiedusingthemethodsandsoftwaresdiscussedinsection4.1.Thecurrent
reducedversionofthereactionmechanismcomprisesof43chemicalspeciesand
629elementaryreactions(seeAppendixA.2).Theanalysisofthemechanismfor
stoichiometric,fuel-richandfuel-leancases,usingHOMREAandMIXFLApackages,
showedthattheconcentrationsofmajorspeciesinthereducedmechanismdeviate
bylessthan2%fromthecorrespondingvaluesoftheoriginalmechanism.Hence,
usingthesimplifiedreactionmechanismdoesnotintroducesignificanterrorsinthe
reactiveflowCFDcalculation.Thereducedmechanismwillthusbeusedinthe
contextofthisthesis.
A2DaxisymmetricgeometrywasusedduetotheavailablesymmetryoftheREGA.
AstructuredquadraticelementgridwithSuccessiveRatioschemewasgenerated.
Thisschemeisanon-symmetricscheme,inwhichthecellsizeincreasesinboth
radialandaxialdirectionsfromtheburner.Thegridnodesgeneratedforthetop
capoftheREGAusingthesuccessiveratioschemecanbeseeninFigure6.2.The
gasifiermeshwasgeneratedusingGAMBITsoftwareandconsistedof17612cells.
TheCFDsimulationsareperformedusingANSYSFLUENT12.0software.Tocheck
thegeneratedmesh,acoldflowsimulationwasperformedinwhichairatTa=300K
andavolumeflowrateof17.41m3/hwasinjectedandtheaxialvelocityalongthe
axisofsymmetryofthegasifierwascomparedwithmeasuredvalues.Duetothetur-
bulentnatureoftheflow,therealizablek-εmodelwasused.Theexperimentaldata
werederivedbyKITusingapropelleranemometerandLaserDopplerAnemometer
(LDA)[100,101].Figure6.3showsacomparisonofthesimulatedaxialvelocityof

6.1.GasifierModelandSimulationConditions

Figure6.1:Thesimulatedentrainedflowgasifier

75

theairalongthesymmetryaxisoftheREGAwiththetwomeasurementmethods.
Thesimulationresultsshowedagoodagreementwithexperimentalvalues.
Themeshwasalsousedforthehotreactiveflowsimulations.Theturbulence-
chemistryinteractionsweretakenintoaccountbyusingtheEDCmodel(section
4.4).TheEDCmodelwasemployedtogetherwiththeISATalgorithm(section
5.3)todynamicallytabulatethechemistrymappingandtoreducethetimetosolu-
tion.TheDiscretePhaseModel(section3.3)togetherwiththeDiscreteOrdinates
model(section3.2)wereusedtomodeltheliquidphaseandradiationheattransfer,
respectively.
ANSYSFLUENTappliesthefinitevolumemethod(section5.1)tosolvethegov-
erningequationsnumerically.Here,afirst-order-upwindschemewasappliedfor
interpolationwithinapressure-basedimplicitsolver.TheSIMPLEprocedurewas
employedforpressurevelocitycoupling.
Forthereactiveflowsimulation,acasewasconsideredinwhichethyleneglycolwas
injectedataflowrateof9.5kg/handgasifiedunderfuelrichcondition(λ=0.43).
Theoxidizingagentwasamixtureofairandpureoxygen.Theenrichedaircontained
40%voloxygen(CaseC1inTable6.2).Thegasifierwallwaskeptataconstant
temperatureof1373K.

6.2.FlowPattern

85

Figure6.2:Gridnodesgeneratedonthetopcapusing’SuccessiveRatioScheme’

Figure6.3:Axialvelocityalongthesymmetryaxisofthegasifier
LDAandPropellerareexperimentalvalues[101]

6.2FlowPattern

Figure6.4showsthecontoursofthegasvelocity,thestreamlinesandthedroplet
trajectoriesforthetop1mofthegasifier.Inthemiddleplot,onecanseethe
recirculationzonethatisformedaroundthecenterlineofthegasifier.Fromthe
middlepartofthegasifiertotheoutlet,theflowpatternturnstoauniformturbulent
plugflowprofile.Ethyleneglycoldropletsarevaporizedquicklyduetothehigh
temperatureinsidethegasifieranddonotentertherecirculationzone,ascanbe
seeninFigure6.4.c.Therandomshapeofthedroplettrajectoriesisduetoatracking
modelofANSYSFLUENT(theDRWmodelwasusedinthisthesis)thatwasused
tobetterdescribetheirturbulentandstochasticnature.Inessence,theparticlesare
notexpectedtofollowthesamegeometricalrouteseverytimetheyareinjectedinto
theflowfield,theyratherfollowascattered(aroundatime-meanpath)routewhich

6.2.FlowPattern

95

Figure6.4:Gasvelocity(left),streamlines(middle),anddroplettemperature
(right)forthetop1meterofthegasifier

issetbyarandomnumbergeneratordeterminedbythelocalturbulencelevelsas
discussedinsection3.3.
ThemolecularviscosityofthegaseswascalculatedusingtheSutherlandviscosity
law[102],basedonkinetictheoryofidealgasesandanidealizedintermolecular-force
potential,as

T+CT3/2
µ=µ00,(6.1)
TC+T0whereµ0andT0arethereferencevaluesofviscosityandtemperature,respectively,
andCistheSutherlandconstant.Thevaluesofµ0,T0andCfordifferentrelevant
chemicalspeciesarederivedfrom[103].
InFigure6.5,thecontoursofviscosityandgastemperatureonthetopthirdofthe
gasifierarepresented.Ascanbeseen,themolecularviscosityandthetemperature
haveasimilarprofileduetothetemperaturedependenceofviscosityaccordingto
theSutherlandlaw.

6.3.TemperatureandSpeciesConcentrations

06

Figure6.5:Molecularmass,molecularviscosityandtemperatureofthegasforthe
top1meterofthegasifier

6.3TemperatureandSpeciesConcentrations
DuetotherecirculationpatternshowninFigure6.4,hotgasfromthelowerpart
oftheflamewillmoveupalongthewalltothetopofthegasifier.Therecirculated
gasisrichinreactivespecies(i.e.COandH2)andhasahightemperature.Hence,
itwilleasilybeoxidizedwhenbroughtincontactwiththeoxygeninjectedfromthe
burner.Thisassiststheflametoholditshightemperatureandalsotheformation
ofregionswithhightemperaturesclosetotheburner,whereoxygenmixeswith
recirculationgas,ascanbeseenfromthetemperaturecontoursinFigure6.5.The
maximumtemperatureachievedinthisregionisabout2310K.Becauseoftheplug
flownatureoftheflow,mostofthereactorhasahomogeneoustemperatureofabout
K.5731Hafneretal.[66]performedareactionflowanalysisoftheethyleneglycoloxidiza-
tionunderfuel-richconditionsinajetstirredreactor.Themainreactionpathunder
thisconditionwasthedecompositionofethyleneglycoltoacetaldehyde(CH3CHO)
withsubsequentH-abstractiontoacetaldehyderadicalCH3COandfinallythede-
compositiontoCH3andCO.ThistrendcanbeobservedintheREGAsimulation
aswell.Aftertheinjection,liquidethyleneglycolvaporizesandentersthegasphase
atanaxialdistancebetween120-450mmfromtheburnerwithamaximummole
fractionoccurringaroundx=200mmfromtheinjectionpoint(seeFigure6.6).At

6.3.TemperatureandSpeciesConcentrations

16

Figure6.6:Molefractionsofethyleneglycol(HOCH2CH2OH),O2,H2O,CO2,
CO,andH2forthetop1meterofthegasifier

anaxialdistanceofabout155mmacetaldehydeisformedandthekineticrateof
theethyleneglycoldecompositionreaction

HOCH2CH2OHCH3CHO+H2O

)2.6(

reachesitsmaximumvalueatthesamedistance.Thisvalueisoneorderofmag-
nitudehigherthanotherethyleneglycoldecompositionreactions.Thecontoursof
molefractionsofacetaldehydeanditsradicalCH3COcanbeseeninFigure6.7.

ItisobservedfromFigure6.6thatthewholeamountofoxygenisconsumedas
itentersthegasifierwhichistheresultoftheethyleneglycolintermediatespecies

6.3.TemperatureandSpeciesConcentrations

26

Figure6.7:MolefractionsofCH3CHO,CH3CO,CH4,C2H2,C2H4,andOHfor
thetop1meterofthegasifier

oxidationaswellastherecirculationzoneeffectalreadydiscussed.Theproductsof
oxygenreactionswiththereactivecomponentsCOandH2reachhighvaluesatthe
hotregionsascanbeseeninthecontoursofmolefractionsofCO2andH2O.The
recirculationzonecausessomegastobetrappedinthetopcornerofthegasifier
whichcanbeseenfromforexampletheH2molefractioncontours(Figure6.6).

Themolefractionsofmajorgasificationproductsatthegasifierexitarelistedin
Table6.1,forthegasificationcaseC1.ThesyngascomponentsCOandH2have
nearlythesamemolefractions,18.69mole%and18.28mole%,respectively.The
listedspeciesaccountfor99.15mol%oftheproductgas.Acetylene(0.7mol%)and
ethylene(0.13mol%)aretwominorspeciespresentintheproductgas.

6.3.TemperatureandSpeciesConcentrations

SpeciesCOH2CO2H2ON2CH4
mol%18.6918.2810.2825.6424.082.18

Table6.1:ProductgascompositionatthegasifierexitforCaseC1

36

Figure6.8:Experimental(E)andsimulated(S)molefractionsofCO2andCOvs.
distancefromtheburner

Figure6.8showsthemolefractionsofCO2andCOinpercentageofgasvolume
fordifferentdistancesfromtheburnerhead.Theexperimentaldataweretaken
fromthemeasurementsperformedatKIT[104].Thegassamplesareextracted
fromthegasifierthroughasamplingprobe,cooledto160◦C,filtered,andcooled
furtherviaacoolertocondensewatervapor.Thesampleisthenanalyzedina
gasanalyzer[100].Inthisway,theCH4,H2,COandCO2molefractionscanbe
measuredindrycondition(%voldry).AscanbeobservedfromFigure6.8,the
CO2concentrationisslightlyunder-predictedandtheCOconcentrationfarfrom
theburnerisover-predictedbythemodel.
TheradialprofilesofmolefractionsofCO2andH2atanaxialdistanceofx=200mm
fromtheburneraredepictedtogetherwiththeexperimentalvaluesinFigure6.9.
Outsidetheflameregion,theH2concentrationishigherthantheexperimentalval-
uesbuttheCO2concentrationshowsgoodagreementwiththemeasurements.In
general,thesimulationresultsshowedacceptableagreementwiththeexperimental
values.Areasonforthedifferencebetweennumericalandexperimentalvaluescould
betheadditionoferrorsduetothedifferencesinthemolefractionswhencalculating

6.3.TemperatureandSpeciesConcentrations

46

Figure6.9:Experimental(E)andsimulated(S)molefractionsofCO2andH2vs.
radialdistanceatanaxialdistanceof200mmfromtheburner

themindrycondition.Numericalerrorsandmeasurementerrortolerancesarealso
anothersourceofdiscrepancybetweenthesimulationandtheexperimentalresults.
Theerrortoleranceofthemeasurementswasaround2.2%forH2and1.2%
forH2O,CO,andCO2[101].Furthermore,theeffectofleakageairwasneglected
inthesimulations.Leakageairfromthesealsofthegasifierflanges[22]hassome
influenceonthetemperaturefieldandthegasconcentrationsduetoavailabilityof
excessoxygen.
Thek-εturbulencemodelisknowntoover-predictthestrengthofthevortexstruc-
ture(andconsequentlytherecirculationzoneeffect)[105].Thiscausesareduction
ofthetemperatureinthesymmetryplaneandhencethechemicalspeciesconcen-
trationschangeaswell.
Todevelopthedetailedchemicalreactionmechanism,someestimatesaremade[64],
whichduetotheunavailabilityofenoughkineticdataofethyleneglycoloxidation
tovalidatethem,areanothersourceofuncertaintyoftheCFDresults.

6.4.EffectofOperatingConditions

56

6.4EffectofOperatingConditions
Inordertostudytheeffectofboundaryandoperatingconditionsonthegasifica-
tionprocess,aseriesofsimulationsweredonetoperformsensitivityanalysis.Four
parameterswerevaried,namely:oxidizerandfuelinlettemperatures,theoxidizer
compositionwhichistheenrichmentofairwithO2,theair-fuelratioandtheoper-
atingpressureofthegasifier.Table6.2showsanoverviewofthedifferentsimulation
casestakenintoaccountfortheparameterstudies.
Theobjectiveofgasificationprocessistheproductionofhighqualitysynthesisgas.
Tochecktheeffectivenessofgasification,aparametercalledgasificationefficiency
isdefinedas[106]:

ηG=m˙gLHVg(6.3)
LHVm˙ffwherem˙andLHVarethemassflowrate(kg/s)andlowerheatingvalue(MJ/kg),
respectively.Thesubscriptgstandsfortheproductgasandfforthefuelwhich
inourcaseisethyleneglycol.ThegasificationefficiencyηGisthentheratioofthe
heatcontentoftheproductgasesgeneratedbygasificationtotheheatcontentof
thefuelwhenitistotallyburnt.
Thelowerheatingvalueofethyleneglycoliscalculatedbasedonitschemicalel-
ementsandhaveavalueofabout17.94MJ/kg.Fortheproductgas,theLHV
iscalculatedbasedontheamountofavailableburnablechemicalspecies(CO,H2,
CH4,C2H2,andC2H4).

6.4.1InletTemperatures
Fourcasesareconsideredforstudyingthechangesininlettemperatures.These
cases(caseC1-C4)arelistedinTable6.2.Inthebasiccase(C1),bothoxidizerand
fuelhadaninlettemperatureof300K.Theoxidizertemperaturewasthenincreased
to330K(C2)and350K(C3)keepingthefuelinlettemperatureconstant.Forthe
caseC4,thefuelandtheoxidizerbothenteredthegasifierat350K.
Figure6.10showsthegasificationefficiencyasafunctionoftheinletoxidizertemper-
ature.Withincreasingtemperaturefrom300Kto350K,thegasificationefficiency
increasedsignificantlyfrom68.58%to74.48%.Atthesametime,theresultsshow
thatthecompositionofproductgasvariedanditsLHVincreasedduetoanincrease
inCOandH2molefractions,asshowninFigure6.10.Thegasificationefficiency
forthecaseC4didnotshowasignificantdifferencewiththatofcaseC3,meaning
thatpreheatingthefuelfrom300Kto350Kdoesnotaffectthesyngascomposition

6.4.EffectofOperatingConditions

Caseλ[-]m˙f[kg/h]Tf[K]Tox[K]xO2[%]Tw[K]p[atm]
C10.439.53003004013731
C20.439.53003304013731
C30.439.53003504013731
C40.439.53503504013731
C50.4311.93003005013731
C60.435.123003002113731
C70.609.53003004013731
C80.759.53003004013731
C90.309.53003004013731

Table6.2:OverviewoftheboundaryconditionsfortheCasesC1-C9

66

Figure6.10:Gasificationefficiencyandmolefractionsatthegasifieroutletfor
differentoxidizerinlettemperatures

andhencethegasificationefficiency.Thisisduetothehightemperaturesinsidethe
gasifiercausingthefueltoevaporateveryfastasalreadyshowninFigure6.4.
Thehighesttemperatureinsidethegasifierhasincreasedfromaround2310Kforthe
caseC1toaround2370KforthecaseC3.Thepositionofthehighesttemperature
areamovedtowardtheburnerwithpreheatingtheoxidizer.Thistrendisvisualized
inFigure6.11.Thetemperatureoutsidetheflamezonedidnotshowsignificant
differenceandwasabout1375Kforallthecases.

6.4.EffectofOperatingConditions

76

Figure6.11:ContoursofgastemperaturesintheflamezoneforCasesC1,C2,C3

6.4.2OxidizerComposition
Tostudytheeffectoftheoxygencontentoftheoxidizeronthegasificationefficiency,
threecaseswereconsideredthatareshowninTable6.2.Inthefirstcase(C1)the
oxidizercontained40%oxygenandtheother60%ofthegasvolumeisN2.In
thesecondcase(C5),theoxidizerwasenrichedwithevenmoreoxygentoreach
xO2=50%.Thesetwocaseswerecomparedwithacase(C6),inwhichthegasifying
agentwasair(xO2=21%).

Figure6.12:Gasificationefficiencyandmolefractionsatthegasifieroutletfor
differentoxidizercompositions

Themolefractionsofsyngascomponents(H2,CO)atthegasifierexitaswellasthe
gasificationefficienciesareplottedinFigure6.12.Theplotshowsanincreaseinthe
gasificationefficiencyastheoxygenmolefractionincreases,althoughtheincrease

6.4.EffectofOperatingConditions

86

isnotconsiderablebetweenthecaseswithxO2=40%andxO2=50%.AsxO2was
increased,themolefractionofH2increasedfasterthanthatofCO.
Themaximumtemperatureinsidethegasifierincreasedfromabout1820Kfor
gasificationwithairtoabout2490Kwhenthegasifyingagentcontained50%
oxygen.ThiswasmainlyduetothedecreaseinthethermalballastN2.

6.4.3Air-FuelRatio
Inordertostudytheeffectoftheair-fuelratio(λ)onthegasificationefficiencyand
thecompositionoftheproductgas,thebasiccaseC1(λ=0.43)wasconsideredwith
threeothercasesC7,C8,andC9withλ=0.60,λ=0.75,andλ=0.30,respectively.
TheresultsofthecomparisonareshowninFigure6.13.
Withincreasingλ,onemovesfromgasificationtocombustion.Thisimpliesthatthe
gasificationefficiencyshoulddecreaseandlesssyngasshouldbeproduced(solidlines
inFigure6.13).Ontheotherhand,thecombustionproductsCO2andH2Oincrease.
Furthermore,theheatreleasefromtheprocessincreasesandlargeamountofheat
transferedfromthewallsofthegasifiercausingtheprocessnottobeadiabatic
anymore.Anadiabaticboundaryconditionforthegasifierwallwasselectedto
studyitseffectonthegasification.ThedashedlinesinFigure6.13indicatethe
resultsofthesimulationsofthethreecasesC1,C7andC8withadiabaticboundary
conditions.

Figure6.13:Gasificationefficiencyandmolefractionsatthegasifieroutletfor
differentair-fuelratiosforconstantwalltemperature(solidlines)and
adiabaticwalls(dashedlines)

6.4.EffectofOperatingConditions

96

Aslightincreaseofthegasificationefficiencywasobservedfortheadiabaticcases
incomparisonwiththenonadiabaticcases.Thiswasduetotheincreaseinthe
molefractionsofH2andCOandhencetheLHVoftheproductgas.However,
thesimulationsindicatedthatthemolefractionsofCH4,C2H2andC2H4were
aroundzerofortheadiabaticcases.Thiswasduetotheincreaseinthereactor
temperature.Thereactortemperatureincreasedfromaround1373Kforthenon
adiabaticcasesto1485K,1954Kand2300Kfortheadiabaticcaseswithλ=0.43,
λ=0.60andλ=0.75,respectively.Athigherreactortemperatures,thereactions
proceededfaster,resultinginthefasterdecompositionofCH4,C2H2andC2H4to
theendproductspeciesCO,CO2,H2andH2O.Forthenonadiabaticcases,the
molefractionsoftheminorspeciesdecreasedwithincreasingtheair-fuelratio.For
example,themethanemolefractiondecreasedfrom3.47%forλ=0.30to0.73%
forλ=0.75.
Themaximumflametemperaturehasincreasedfromaround2310Kforλ=0.43to
around2480Kforλ=0.75.Theincreaseintheheatreleasewithincreasingair-fuel
ratiocausedthehotzonetobebigger.

uresresP6.4.4Operatingagasifierunderhighpressuresleadstoareductionofthespecificvolume
ofthegases,whichinreturndecreasesthedimensionsoftheequipment[107].Onthe
otherhand,increasingtheoperatingpressurecausesanincreaseinmanufacturing
costs.IncaseofthebioliqRprocess,thehighpressureoperationisdesirableasit
obviatesintermediatesyngascompressionpriortothefuelsynthesisstep[3].
ThecaseC6(seeTable6.2)wasconsideredasthebasiscaseforstudyingtheeffect
ofthereactoroperatingpressureonthegasificationefficiency.Twomorecases
wereconsideredinwhichtheoperatingpressurewasincreasedto2and5bars,
respectively.AlltheotherboundaryconditionswerekeptconstantasthoseofC6.
ThegasificationefficiencyincreasedfromηG=66.72%foratmosphericgasification
to71.55%whentheoperatingpressurewas5bars.Thiswasduetotheincreasing
oftheLHVoftheproductgasduetohigherfractionsofCOandH2ascanbeseenin
Figure6.14.Agradualincreaseintheratioofhydrogentocarbonmonoxide,from
about0.83atatmosphericpressuretoover1.11at5barswasobservedwhichshows
thecapabilityofadjustingtheratioofsyngasconstituentsbychangingtheoperating
pressure.However,thismaybeoflimitedvaluesincetheoperatingpressureis
usuallydeterminedbyotherprocessbasedconsiderations.
TheREGAisdesignedfortheoperationunderatmosphericpressures.Dueto
thisfact,highpressuregasificationsimulationswerenotperformedbasedonthe

6.4.EffectofOperatingConditions

07

Figure6.14:Gasificationefficiencyandmolefractionsatthegasifieroutletfor
differentpressures

Figure6.15:MolefractionsofCOandH2fordifferenttemperaturesandpressures

geometryofthegasifierandtheburner.Tochecktheeffectofhighpressuresonthe
compositionoftheproductgas,aseriesofsimulationsusingtheHOMREAsoftware
werecarriedout.Theinitialreactionpressurewasvariedfrom1barupto50bars
andtheinitialtemperaturewasvariedbetween900Kand2100K.Theresultsof
twocaseswithinitialtemperaturesof1500Kand2000KaredepictedinFigure
6.15.Atinitialreactiontemperatureof2000K,anincreaseinthemolefractions
ofH2andCOwasseenuptoreactionpressureof15bar.Inthecaseofhigher
pressures,thegascompositionremainedalmostconstant.Thesametrendcanbe
observedinthelowerreactiontemperature.

6.5.EffectofChemistry

17

Itshouldbenotedthattheethyleneglycolreactionmechanismisvalidatedfor
pressuresuptoabout40bars[64].Forsimulationswithhigherpressures,care
shouldbetakenabouttheusageofthereactionmechanism.
WhencomparingFigures6.14and6.15,thepredictedgascompositionsshowsignif-
icantdifferenceswhichisduetothedifferencesinthemodelassumptionsusedin
thesimulationsoftwares.InHOMREA,anidealjetstirredreactorisconsideredand
theeffectofthermalradiationisneglected.WhereasinANSYSFLUENT,effects
ofturbulenceandthermalradiationaretakenintoaccount.

6.5EffectofChemistry
Inthissection,theeffectofchemistryonthesimulationresultsofthegasification
processisstudied.Inthefirstpart,threeversionsofthechemicalreactionmech-
anism,developedforethyleneglycoloxidation,areusedforCFDsimulationsand
theresultsarecompared.Inthesecondpartofthissection,thethreechemistry
modelsalreadydiscussedinchapter4(equilibriumchemistrymodel,flameletmodel
andeddydissipationconcept)willbecomparedtogether.Thesemodelsaredevel-
opedbasedondifferentunderlyingassumptionswhichoffercertainadvantagesand
disadvantagesforthesimulationofaselectedreactiveflowproblem.

6.5.1ReactionMechanism
Duetothelackofexperimentalkineticratedataforhightemperatureethylene
glycoloxidation,someoftherateconstantshadtobeestimatedusingstatistical
correlationsandanalogiestootherreactions[64].Fortheestimationoftheactivation
energiesfromanalogreactions,theBell-Evans-Polanyiequation[108,109]wasused

Ea,1=Ea,0+αΔHR0(6.4)

whereEa,0istheactivationenergyoftheanalogreaction,αisafactorbetween0
and1andΔHR0isthedifferencebetweenthestandardreactionenthalpiesofthe
reactions.
Thereactionmechanismusedinthisthesiswasdevelopedbasedonα=0.5[64].To
studytheeffectofthechangesinthereactionschemeonthegasification,twoversions
ofthereactionmechanism,createdwithα=0.0andα=1.0,wereconsidered
togetherwiththeoriginalmechanism.
ThecalculationsbasedonthepackageHOMREAdidnotshowsignificantdifferences
inthemolefractionsofthemajorgasificationproductspeciesCO,CO2,H2andH2O.

6.5.EffectofChemistry

27

Figure6.16:ContoursofmolefractionsofH2(top)andOH(bottom)atthetop
partofthegasifierforsimulationsusingdifferentversionsofthere-
actionmechanismwithα=0.0(right)andα=0.5(left)

FortheCFDsimulationsinANSYSFLUENT,thegasificationcaseC6wasstudied
usingthethreechemicalreactionmechanisms.Nosignificantdifferencehasbeen
observedinthemolefractionsofmajorspecies(<0.5%)forthecaseswithα=0.5
andα=1.0.However,inthesimulationusingthemechanismwithα=0.0,themole
fractionofH2wasincreasedabout2%atthegasifieroutlet.Thisincreasecaused
adecreaseofabout1%inthemolefractionofH2O,asexpected.Thechangesin
themolefractionsofotherspecieswerenotsignificant(<0.3%forCOandCO2).
ThecontoursofthemolefractionsofH2andOHonthetoppartofthegasifieris
showninFigure6.16.TheshapeoftheflamebasedontheOHconcentrationcan
beseeninthisfigureaswell.NoOHproductionwasobservedintheveryvicinity
oftheburnerwhenα=0.0wasused.

6.5.EffectofChemistry

37

Theresultsofthesimulationsemphasizethatthechoiceofreactionkineticsplays
aroleifoneisinterestedintheflameshapeandproperties,however,theproduct
gascompositionatthegasifieroutletdidnotshowgreatsensitivitytothechoiceof
reactionkinetics(αvalues)forthestudiedchemicalreactionmechanism.Forthe
effectsofthefactorαontheignitiondelaytimesandflamevelocities,thereaderis
referredto[64].

delMotryChemis6.5.2Withintheframeworkofthisthesis,acomparisonofthechemistrymodelsavail-
ableintheCFDcodeANSYSFLUENThasbeendone.Thesemodelshavebeen
discussedinchapter4indetail.Theboundaryconditionsused,werebasedonthe
gasificationcaseC1(seeTable6.2).Thesimulationswereperformedusingnon-
premixedcombustionwithequilibriumchemistry(EQ),thesteadylaminarflamelet
model(SLF)andtheeddydissipationconcept(EDC).
ThecomputationalcostsoftheEQandtheSLFmodelsaremuchlowerthanthat
oftheEDCmodel.Thisisduetothefactthatthepreprocessingofchemistryin
look-uptablesispossiblefortheEQandSLFmodels.Therefore,onlytwoand
threetransportequationsfortheEQandSLFmodels,respectively,arerequired
tobesolved.TheEDCmodel,ontheotherhand,solvesonetransportequation
foreachchemicalspecies.ThemodelutilizestheISATproceduretodecreasethe
computationaltimeforchemistrycalculations,butisstillaverytimeconsuming
modelwhendetailedreactionmechanismsareused.
Theresultingmolefractionsofmajorproductspeciesatthegasifieroutletareshown
inFigure6.17.Asseeninthisfigure,themolefractionsofthespeciesforsimulations
withtheflameletmodel(SLF)andtheequilibriumchemistrymodel(EQ)have
almostthesamevaluesattheoutlet.Asthestoichiometricscalardissipation(χst
inequation4.23)tendstozerointheSLF,thechemistrytendstoequilibrium.This
happensoutsideoftheflamezoneandisthereasonfortheequalityofthespecies
molefractionsatthegasifieroutlet.Themolefractionsofmethane,acetyleneand
ethylenearepredictedbybothEQandSLFmodelstobezeroatthegasifieroutlet,
whereastheEDCmodelpredictedmethanetobe2.18%andacetyleneandethylene
togetheraround0.83%oftheproductgasattheoutlet.
Intheflamezone,thetemperaturesarepredictedhigherbytheEQmodelthanby
theSLFmodel,whichisduetotheequilibriumcalculations[110].However,the
temperaturespredictedbybothmodelsarequalitativelysimilar,ascanbeseenin
Figure6.18.Furthermore,theSLFmodelunderpredictsthetemperatureincom-

6.5.EffectofChemistry

47

Figure6.17:MolefractionsofCO,CO2,H2andH2Oatthegasifierout-
letResultingfromsimulationswithEddyDissipationConcept
(EDC),SteadyLaminarFlameletModel(SLF)andEquilibriumNon-
premixedChemistryModel(EQ)

parisonwiththeEDCmodel.Thistrendintemperaturepredictionofbothmodels
ismentionedintheliteratureaswell[27,111].
AnoverpredictionofH2isobservedbytheEQmodelincomparisonwiththeSLF
modelwhichisduetotheassumptionoffastchemistryinEQmodel[112].Onthe
symmetryaxisofthegasifier,COandH2molefractionspeaknearertoburnerin
SLF(andEQ)modelcomparingwithEDCasseeninFigure6.18.
ThemassfractionsofH2O,CO2andOHfortheSLFandEDCmodelsaredepictedin
Figure6.19.ItisevidentthattheSLFmodelpredictshigherH2Omassfractionsand
significantlylowerCO2massfractionsthantheEDCmodel.Liuetal.[113]reported
thesametrendwhencomparingtheSLFmodelwithdirectnumericalsolutions.The
maximumOHmassfractionpredictedbybothmodelsdidnotdiffersignificantly.
TheOHmassfractionintheSLFmodel,however,spreadsmuchfurtherdownstream
thanthatoftheEDCmodelasobservedinFigure6.19.Generally,theSLFmodel
predictsafasterconversionofthefuelspeciesintoproducts.Thisisinaccordance
withtheresultsofotherstudies[114].
Oneshouldnotetheimportanceoftheturbulencemodelonthepredictionsoftem-
peratureandchemicalspeciesbychemistrymodels.Thepredictedprofilesofthe
mixturefraction,itsvarianceandthescalardissipationrateintheSLFmodelare
sensitivetotheturbulencemodel[115].Inaccuratedescriptionofmixingcausesdis-
crepanciesbetweenpredictionsandmeasurements.ThequalityoftheEDCmodel
predictionsdependsalsoontheperformanceoftheturbulencemodel.Inthiscase,

.5.6

foEffect

Figure

6.18:

Chemistry

Contoursof

temperature,

,CO

dna

H2

omel

fractions

simulationswithEddyDissipationConcept(EDC),

resulting

57

mofr

SteadyLami-

narFlameletModel(SLF),andEquilibriumNonpremixedChemistry

Model(EQ)atthetoppartofthegasifier

6.5.EffectofChemistry

67

Figure6.19:ContoursofmassfractionsofCO2,H2O,andOH,resultingfromsim-
ulationswithEddyDissipationConcept(EDC),andSteadyLaminar
FlameletModel(SLF)atTopPartoftheGasifier

thelengthfractionandtimescaleofthefinestructures(equations4.32and4.33,
respectively)directlydependontheturbulenceproperties(kandε).Aninaccurate
predictionofthesetwoquantitiesleadstoerroneouscalculationofξ∗andτ∗and
hencethethermochemicalfield.

Thescalardissipationrate(usedintheSLFmodel)isinsufficienttoquantifythe
non-equilibriumstructureofadiffusionflameinanaxisymmetriccoflowconfigu-
ration[113].Inaddition,thestudiedgasifier(REGA)withrecirculationzonesis
problematicforflameletmodels.Forthesereactors,partiallyreactedfluidisrecir-
culatedtomixwiththefeedstreamssothatthesimplenon-premixedflowmodel

6.6.SlurryGasificationSimulation

SlurryOxidizer
Caseλ[-]weg[%]wchar[%]xO2[%]
C10.43100040
C100.43901040
C110.43802040

Table6.3:Slurrygasificationcases

77

nolongerapplies[72].TheEDCmodelprovestobeabetterchoicewhenmodeling
gasificationinentrainedflowgasifierswithrecirculationzones.
However,careshouldbetakennottomakeadefinitestatementaboutonemodel’s
superiorityoveranother,asthemodelpredictionsdependstronglyontheprocess
andboundaryconditionsfortheprobleminquestion.

6.6SlurryGasificationSimulation
Asdiscussedinsection1.2,amixtureofpyrolysisoilandcharwasusedasthefeedfor
thegasificationinthebioliqRprocess.Ethyleneglycolwasusedasamodelfuelfor
pyrolysisoil.Inordertosimulatethegasificationprocessusingslurry,asubmodel,
developedbyHafner[64],wasutilizedinANSYSFLUENT,whichmodeledthechar
particlegasificationandcombustion.Forthedetaileddescriptionofthemodel,the
readerisreferredto[64].
Themodelisbasedontheheterogeneousreactionsofcarbonwithgasifyingagents
CO2,H2O,andO2whichtakesintoaccounttheinhibitioneffectofCOandH2.Each
charparticleiscomposedofaporouscarbonsphere.Theslurryisthenamixture
ofcharparticlesandethyleneglycoldroplets.Itisassumedthatatthebeginning,
theporouspartofacharparticleisfilledwithethyleneglycol.
Aftertheslurryentersthegasifier,theparticlesareheatedandtheethyleneglycol
vaporizesandentersthegasphase.Atthesametime,thecharparticlesareheated
andreactwiththegasifyingagents.Thechargasificationproductiscomposedof
CO,CO2,andH2.
Inordertosimulatetheslurrygasificationandtostudytheeffectofcharparticles
ontheproductgasandthegasificationefficiency,twocaseswereconsideredtogether
withthecaseC1.TheconsideredcasesaresummarizedinTable6.3.

6.6.SlurryGasificationSimulation

87

Figure6.20:ContoursofcharparticlestemperatureandcharconversionforCase
C10atthetop1meterofthegasifier

ThecaseC1hasalreadybeendiscussedinsection6.1.ForthecasesC10andC11,
themassfractionofcharparticlesintheslurry(wchar)waschosentobe10%and
20%,respectively.Allotherboundaryconditionswerekeptconstantasthoseofthe
caseC1inTable6.2.
Figure6.20showsthecharparticletemperaturesandthecharconversionforthe
caseC10.Forclarityreasonsonly20%ofthesimulatedparticlesareshowninthis
figure.Thehighestparticletemperatureoccuredintheflamezoneandwasmore
than2200K.Thiscausedtheparticlestoreactveryfastwiththeavailableoxygen.
Therecirculationzoneplayedanimportantroleinthecharconversionbyincreasing
theresidencetimeoftheparticlesinsidethegasifier.Thoseparticlesthatwerenot
trappedinthiszonehavenotbeencompletelyconvertedandexitedthegasifier,
whichinturnresultsinmoreeffortinthegascleaningsteps.
Figure6.21showsthecontoursofthechemicalspeciesCOandH2producedthrough
charparticlereactionsforbothconsideredcasesC10andC11.Hydrogenwaspro-
ducedbythereactionofcharparticles(Cf)withH2Oas

Cf+H2OC(O)+H2.

()5.6

6.6.SlurryGasificationSimulation

97

Figure6.21:ContoursofproducedCOandH2fromcharparticlesforCasesC10
andC11atthetoppartofthegasifier

BycomparingFigures6.7and6.21,itcanbeobservedthattheH2productionwas
highintheareaswhereH2Ohadahighconcentration.Atthetopcornerofthe
gasifier,wheresomewatervaporwastrapped,thehydrogenproductionwasalso
ofimportance.Atthesecondhalfofthegasifier(notshowninFigure6.21),the
charparticlesnottrappedintherecirculationzonereactedwithH2Omoleculesand
producedmorehydrogen.

COwasproducedneartheburneroutletduetotheavailabilityofoxygenaccording
tothefollowingchemicalreaction:

Cf+O2−→CO+CO2

.6()6

6.6.SlurryGasificationSimulation

80

Figure6.22:Molefractionofethyleneglycolinthegasphasealongthesymmetry
axisofthegasifierfordifferentcharcontents

ThekineticsoftheabovereactionindicatesthatmoreCOisproducedathigh
temperatures(above1000K).Atlowtemperatureregions,ontheotherhand,more
CO2isproduced.ThiscanbeseenaswellinFigure6.21.
AsinthecaseofH2,atthesecondhalfofthegasifiersomecharparticles,nottrapped
intherecirculationzone,reactedwiththeavailableCO2moleculesandproduced
COaccordingtothereaction

Cf+CO2CO+C(O).

)7.6(

TherateofCOproductioninthiszonewasnotveryhighincomparisontothatof
theareaneartheburner.
Theevaporationofethyleneglycoldropstrappedintheporesofcharparticlescaused
achangeinthedistributionofethyleneglycolinthegasphaseascanbeobserved
fromFigure6.22.ForthecaseC1,wherenocharparticleswerepresent,theethylene
glycoldropletsstartedtoevaporateatanaxialdistanceofaboutx=120mmfrom
theburnerandreachedtheirmaximumataboutx=220mmacrossthesymmetry
axis.InthecasesC10andC11someethyleneglycolenteredthegasphaseata
distanceofaboutx=60mmfromtheburnerwhichwasduetosomeevaporation
fromtheporesofthecharparticles.Themaximumvaluesofethyleneglycolcon-

6.6.SlurryGasificationSimulation

18

Figure6.23:Gasificationefficiencyandmolefractionsatthegasifieroutletfor
differentcharcontents

centrationingasphasedecreasedforcasesC1,C10andC11duetothedecreasein
theinitialmassfractionandthedistributionofdropletsinthefuel.
ThemolefractionsofCOandH2showedadecreasewithincreasingthemassfraction
ofcharasshowninFigure6.23.AnexplanationforthedecreaseinCOandH2is
thatthecharparticlesenterthegasifierattheburnerpositionandentertheflame
zonewhereveryreactivechemicalradicalssuchasOHandOarepresent.The
producedspeciesCOandH2,resultingfromthereactionofcarbonparticleswith
oxygenandwatervapor(reactions6.5and6.6),reactwiththeradicals,forexample
OH,throughthefollowingreactions

H2+OHH2O+H

CO+OHCO2+H

)8.6(

)9.6(

ThiscausestheproductionofCO2andH2O,whichinturndonotreactbacktoCO
andH2veryeasily.Onewaytodealwiththisproblemisalaterinjectionofchar
particlessothattheydonotcomeintocontactwithreactivechemicalradicals.The
newinjectionshouldagaincreatearecirculationzonesothatthecharparticlesare
presentinthegasifierlongenoughforthechargasificationreactionstotakeplace.
Figure6.23furthershowsadecreaseinthegasificationefficiencywhenusingchar
particlesinthefuel.OnereasonisthementioneddecreaseinCOandH2andthe

6.6.SlurryGasificationSimulation

28

Figure6.24:ContoursoftemperatureforCasesC1andC10atthetoppartofthe
erfisiag

otherreasonistheincreaseinthelowerheatingvalue(LHV)ofthefueldueto
thepresenceofchar.Thismeansanincreaseintheheatcontentofthefuel(the
denominatoroftherighthandsideofequation6.3)ofabout10%and20%forthe
casesC10andC11,respectively.
Figure6.24showsthecontoursoftemperatureforcasesC1andC10atthetopofthe
gasifier.Themaximumtemperatureintheflamezoneinsidethegasifierdecreased
fromaround2310KinthecaseC1toaround2285KforthecaseC10.Thisdecrease
wasduetothegasificationofcharparticles.Thereactionsofcharparticleswith
H2OandCO2areendothermicreactionswhichoccuredinthehotpartoftheflame
andcausedthetemperaturetodrop.Furthermore,theexothermicreactionofchar
particleswithoxygencausedthelocationofthemaximumtemperaturetomove
furtherupstreamtowardtheburnerascanbeobservedinFigure6.24.

7.ConclusionsandPerspective

Theprimegoalofthisworkwasthemodelingandsimulationofthegasification
ofbiomass-basedpyrolysisoil-charslurryinanentrainedflowgasifierasapartof
thebioliqRprocess.Inthistwo-stageprocess,straworotherabundantlignocel-
lulosicagriculturalby-productsareconvertedtosyngasthroughfastpyrolysisand
subsequententrainedflowgasification.

Theentrainedflowgasificationbelongstotheclassofreactiveturbulentflowprob-
lemswhich,duetothecomplexinteractionsbetweenchemistryandturbulence,
needsspecialattention.Thechoiceofthechemistry-turbulenceinteractionmodel
aswellasotherrelatedphysicalandnumericalsubmodelsplayanimportantrolein
theCFDsimulationresults.

Themodelsdiscussedinchapters2to5aspartoftheCFDsoftwarepackageANSYS
FLUENTwereusedtoperformthesimulations.Ethyleneglycolservedasanon-
toxicmodelsubstanceforthebiomass-basedpyrolysisoilintheexperimentsat
KIT.Ithasalsobeenusedinthisworkasthemodelsubstanceinordertoallowa
comparisonbetweensimulationresultsandtheexperimentalresults.

A2-Daxisymmetricgeometricalmodelofthepilotscaleentrainedflowgasifier
REGAwasusedforthemeshgeneration.Thesimulationresultsarepresentedand
discussedinchapter6.

Thecoldflowsimulationresultsshowedacceptableagreementwiththeexperimental
measurements.However,moreexperimentalvalueswouldhelptooptimizethemodel
constantsofthek-εturbulencemodelshowninTable3.1.Inthiswork,thevalues
suggestedby[26]and[32]havebeenused.

84

Thereactiveflowsimulationswerealsocomparedwithexperimentalmeasurements
whereverpossible.Thesecomparisonsagainshowedacceptableagreement.The
simulationsdepictedtheimportanceoftherecirculationzoneinentrainedflowgasi-
fication.Itbringsthehotreactivegasintocontactwiththeinjectedoxidizer,helping
theflametoholditshightemperature.Therecirculationzoneplaysanotherimpor-
tantroleintheentrainedflowgasificationofslurry.Duetotherecirculation,the
charparticleshavelongerresidencetimeinthegasifiertoreactwiththegasphase.
Thechoiceoftheturbulencemodelisofessentialimportanceforthemodelingof
therecirculationzone.Moreworkisrequiredtostudytheeffectofotherturbulence
modelsonthepredictionofflowpatternsinsideanentrainedflowgasifier.
Theuseofeddydissipationconcept(EDC)enabledustoemploydetailedchem-
icalreactionschemesintheturbulentflow.Thereactionmechanismutilizedin
thisworkisbasedonthesimplificationofthemechanismdevelopedbyHafner[64].
Thesimplificationwasperformedusingsensitivityandreactionflowanalysis.With
detailedchemistrythereactionpathoftheoxidationofethyleneglycolcouldbe
observedinthesimulations.Thedetailedchemistryenablesonetostudythechem-
icalprocessesandcompositionofthechemicalintermediateswhichisnotpossible
whenusingglobalreactions.Withregardtocalculationtime,theEDCisavery
expensivemodelandshouldthusbeusedwherethefastchemistryassumptioncan
notbeassumed.
Inordertostudytheeffectofboundaryconditionsonthegasificationprocess,a
seriesofsimulationsweredonetoperformsensitivityanalysis.Fourparameters
werevaried,namely:oxidizerandfuelinlettemperatures,theoxidizercomposition,
theair-fuelratioandtheoperatingpressureofthegasifier.
Anincreaseintheoxidizerinlettemperaturecausedanincreaseinthegasification
efficiencyaswellasanincreaseinthemolefractionsofH2andCO.Theincrease
intheinlettemperatureofthefueldidnotshowasignificanteffectongasification
efficiencynorontheproductgascomposition.Here,theheatintheproductgas
canbeusedtopreheattheoxidizertoachieveamoreefficientgasification.
Enrichmentoftheairwithoxygenhasapositiveeffectonthegasificationprocess.
Astheairisenriched,theamountofN2decreaseswhichinreturncauseshigher
temperaturesintheflameandahigheramountofsyngas.Inthiswaythegasification
temperaturecanberegulatedasdesired.DecreaseintheoxidizerN2isalsoin
favorofdecreasingthepollutantproduction(NOx,NH3,etc.).TheNOxformation
chemistry,whichisofinterestwhenusingairastheoxidizer,hasnotbeenconsidered
inthisthesis.

58

Withincreasingair-fuelratio,conditionsshiftfromgasificationtocombustion.This
impliesthatthegasificationefficiencyshoulddecreaseandlesssyngas(COandH2)
andmoreCO2andH2Oshouldbeproduced.Largervaluesoftheair-fuelratiocause
higherheatreleaserates.Operatingthegasifierunderadiabaticboundarycondition
increasesthereactortemperatureandpromotessyngasproduction.Otherburnable
gasessuchasCH4andC2H2werenotpresentintheproductgas,whichwould
facilitatethegascleaningandconditioningsteps.
Operatingthegasifiersunderhighpressuresisdesirableasitdecreasesthesizeofthe
gasifierandinthecaseofthebioliqRprocess,obviatesintermediatesyngascom-
pressionpriortothefuelsynthesis.Anincreaseinthegasificationpressureshowed
anincreaseintheefficiencyoftheprocess.Furthermore,theratioofhydrogento
carbonmonoxidechangedbychangingthepressure,whichcanbeofinterestfor
differentdownstreamsyngasutilizations.
Variationsinotheroperatingandboundaryconditionsarenotconsideredinthis
thesis.Moresimulationsforthevariedparameterstocoverabroaderrangemay
helptobetterunderstandtheeffectoftheseparametersontheprocess.
Threedifferentchemistrymodelswerestudiedinthisthesis.Theirrelativeadvan-
tagesanddisadvantageswereexamined.TheEDCmodelprovedtobethebetter
choicewhenmodelinggasificationinentrainedflowgasifierswithrecirculationzones.
However,careshouldbetakennottomakeadefinitestatementaboutonemodel’s
superiorityoveranother,asthemodelpredictionsdependstronglyontheprocess
andboundaryconditionsfortheprobleminquestion.
Thegasificationofslurrywassimulatedusingcharparticlessuspendedinethylene
glycol.ThecharreactionmodelwasdevelopedbyHafner[64].Thesimulations
showedadecreaseintheflametemperaturewithincreasingthemassfractionofchar
particlesintheslurry.Thisisduetotheendothermicreactionsofparticleswith
watervaporandCO2.ThemolefractionsofCOandH2decreasedtoo,whichcaused
adecreaseinthegasificationefficiency.ThisisbecausetheCOandH2producedby
thereactionsofcharparticleswithoxygenandwatervaporreactwithchemically
reactiveradicalsintheflameregiontoCO2andH2O.Onewaytodealwiththis
problemisalaterinjectionofthecharparticlessothattheydonotcomeintocontact
withtheseradicals.Thenewinjectionshouldagaincreatearecirculationzoneso
thatthecharparticlesarepresentlongenoughforthechargasificationreactionsto
takeplace.
Thesimulationsperformedinthisworkhelptobetterunderstandthegasification
processinsideentrainedflowgasifiersandconsiderablyreducethenumberofex-

68

perimentsneededtocharacterizethesystem.Thesimulationsproducedspatial

andtemporalprofilesofdifferentsystemvariablesthatarehardorsometimeseven

impossibletomeasureorwouldrequireexpensiveexperiments.However,moreex-

perimentalmeasurementswouldhelptovalidateandoptimizetheCFDmodel.The

sensitivityanalysesperformedinthisstudyareconsideredas

mizedoperating

gasifiers.

conditions

nad

stssia

ni

eht

uslccessfu

pe-ulsca

abasistofindopti-

fo

eht

entrained

wofl

yphibliograB

[1]EuropeanCommission.RenewableEnergyRoadMap,renewableenergies
inthe21stcentury:buildingamoresustainablefuture.Technicalreport,
COM(2006)848,2007.

[2]EuropeanCommission.Onsustainabilityrequirementsfortheuseofsolidand
gaseousbiomasssourcesinelectricity,heatingandcooling.Technicalreport,
COM(2010)11,2010.

[3]E.HenrichandF.Weirich.Pressurizedentrainedflowgasifiersforbiomass.
EnvironmentalEngineeringScience,21:53–64,2004.

[4]T.Bridgwater.Biomassforenergy.JournaloftheScienceofFoodandAgri-
culture,86:1755–1768,2006.

[5]UNFCCC.Clarificationsondefinitionofbiomassandconsiderationofchanges
incarbonpoolsduetoaCDMprojectactivity.Technicalreport,EB-20,Annex
8,2005.

[6]P.Basu.Biomassgasificationandpyrolysis-practicaldesignandtheory.
AcademicPress,2010.

[7]P.Kumar,D.M.Barrett,M.J.Delwiche,andP.Stroeve.Methodsforpre-
treatmentoflignocellulosicbiomassforefficienthydrolysisandbiofuelproduc-
tion.Ind.Eng.Chem.Res.,48(8):3713–3729,2009.

[8]Y.ShinyaandM.Yukihiko,editors.TheAsianBiomassHandbook.TheJapan
InstituteofEnergy,2008.

[9]I.Olofsson,A.Nordin,andU.Soderlind.Initialreviewandevaluationof
processtechnologiesandsystemssuitableforcost-efficientmedium-scalegasi-
ficationforbiomasstoliquidfuels.Technicalreport,ETPCReport05-02,
.5002

Bibliography

88

[10]A.vanderDrift,H.Boerrigter,B.Coda,M.K.Cieplik,andK.Hemmes.
Entrainedflowgasificationofbiomass:Ashbehaviour,feedingissues,and
systemanalyses.Technicalreport,ECNandShellGlobalSolutions,ECN-C-
04-039,2004.
[11]R.L.Bain.Biomassgasificationoverview.Technicalreport,NationalRenew-
ableEnergyLaboratory(NREL),USDepartmentofEnergy(DoE),2004.
[12]L.Leible,S.Kaelber,G.Kappler,S.Lange,E.Nieke,P.Proplesch,D.Wintzer,
andB.Fuerniss.Kraftstoff,StromundWaermeausStrohundWaldrestholz
-einesystemanalytischeUntersuchung.Technikfolgenabschaetzung-Theorie
undPraxis,15(1):61–72,2006.
[13]E.Henrich,N.Dahmen,K.Raffelt,R.Stahl,andF.Weirich.TheKarlsruhe
”bioliq”processforbiomassgasification.In2ndEuropeanSummerSchoolon
RenewableMotorFuels,2007.
[14]E.Henrich,N.Dahmen,andE.Dinjus.Costestimateforbiosynfuelpro-
ductionviabiosyncrudegasification.Biofuels,BioproductsandBiorefining,
3:28–41,2009.
[15]N.Dahmen,E.Dinjus,andE.Henrich.DasKarlsruherbioliq-Verfahren-
StandundEntwicklung.InMobilmitBiomasse,2006.
[16]Y.WangandL.Yan.CFDstudiesonbiomassthermochemicalconversion.
Int.J.Mol.Sci.,9:1108–1130,2008.
[17]T.Norton,D.W.Sun,J.Grant,R.Fallon,andV.Dodda.Applicationsof
computationalfluiddynamics(CFD)inthemodellinganddesignofventilation
systemsintheagriculturalindustry:Areview.BioresourceTechnology,98
(12):2386–2414,2007.
[18]T.F.Dixon,A.P.Mann,F.Plaza,andW.N.Gilfillan.Developmentof
advancedtechnologyforbiomasscombustion-CFDasanessentialtool.Fuel,
84(10):1303–1311,2005.
[19]H.A.J.A.vanKuijk,R.J.M.Bastiaans,J.A.VanOijen,andL.P.H.De
Goey.ModellingNOx-formationforapplicationinabiomasscombustionfur-
nace.InProceedingsoftheEuropeanCombustionMeeting,2005.
[20]D.F.Fletcher,B.S.Haynes,F.C.Christo,andS.D.Joseph.ACFDbased
combustionmodelofanentrainedflowbiomassgasifier.AppliedMathematical
Modelling,24(3):165–182,2000.

Bibliography

98

[21]A.K.Sharma.Modelingfluidandheattransportinthereactive,porousbed
ofdowndraft(biomass)gasifier.InternationalJournalofHeatandFluidFlow,
28:1518–1530,2007.
[22]U.Santo,D.Kuhn,H.J.Wiemer,E.Pantouflas,N.Zarzalis,H.Seifert,and
T.Kolb.ErzeugungvonSynthesegasausbiomassestaemmigenSlurriesim
Flugstromvergaser.ChemieIngenieurTechnik,79No.5:651–656,2007.
[23]J.Warnatz,U.Mass,andR.W.Dibble.Combustion.Heidelberg:Springer
Verlag,2006.
[24]J.F.Douglas,J.M.Gasiorek,andJ.A.Swaffield.FluidMechanics.Longman
ScientificandTechnical,1995.
[25]J.Piquet.TurbulentFlows-ModelsandPhysics.SpringerVerlag,1999.
[26]ANSYSFLUENT12.0TheoryGuide2009.
[27]N.Peters.TurbulentCombustion.CambridgeUniversityPress,2000.
[28]D.VeynanteandL.Vervisch.TurbulentCombustionModelling.VonKarman
Institute,1999.
[29]J.O.Hinze.Turbulence.McGraw-HillPublishingCompany,1975.
[30]D.C.Wilcox.TurbulencemodelingforCFD.DCWIndustries,1998.
[31]B.E.LaunderandD.B.Spalding.Lecturesinmathematicalmodelsofturbu-
lence.AcademicPress,1972.
[32]T.H.Shih,W.W.Liou,A.Shabbir,Z.Yang,andJ.Zhu.Anewk-εeddy
viscositymodelforhighReynoldsnumberturbulentflows.ComputersFluids,
24(3):227–238,1995.
[33]R.SiegelandJ.R.Howell.ThermalRadiationHeatTransfer.Hemisphere
PublishingCorporation,1992.
[34]J.Liu,H.M.Shang,andY.S.Chen.Developmentofanunstructuredradi-
ationmodelapplicablefortwo-dimensionalplanar,axisymmetric,andthree-
dimensionalgeometries.JournalofQuantitativeSpectroscopyandRadiative
Transfer,66:17–33,2000.
[35]A.G.Ratzel.P-NDifferentialApproximationforSolutionofOneandTwo
DimensionalRadiationandConductionEnergyTransferinGrayParticipating
Media.PhDthesis,UniversityofTexas,1981.

Bibliography

09

[36]J.R.HowellandM.Perlmutter.MonteCarlosolutionofthermaltransfer
throughradiantmediabetweengraywalls.J.HeatTransfer,86:116–122,
.4691

[37]N.G.Shah.AnewMethodofComputationofRadiantHeatTransferinCom-
bustionChambers.PhDthesis,ImperialCollegeofScienceandTechnology,
London,1979.
[38]E.P.Keramida,H.H.Liakos,M.A.Founti,A.G.Boudouvis,andN.C.
Markatos.Thediscretetransferradiationmodelinanaturalgasfiredfurnace.
Int.J.Numer.Meth.Fluids,34:449–462,2000.
[39]W.A.Fiveland.Threedimensionalradiativeheattransfersolutionsbythe
discreteordinatemethod.J.Thermophys.HeatTransfer,2:309–316,1988.
[40]W.A.Fiveland.Discreteordinatemethodsforradiativeheattransferin
isotropicallyandanisotropicallyscatteringmedia.JournalofHeatTransfer,
109:807–812,1987.
[41]D.Joseph,P.J.Coelho,B.Cuenot,andM.ElHafi.Applicationofthediscrete
ordinatesmethodtogreymediaincomplexgeometriesusingunstructured
meshes.InProceedingsofEurotherm73onComputationalThermalRadiation
inParticipatingMedia,2003.

[42]B.R.AdamsandP.J.Smith.Threedimensionaldiscreteordinatesmodelling
ofradiativetransferinageometricallycomplexfurnace.CombustionScience
andTechnology,88:293–308,1993.

[43]A.S.JamaluddinandP.J.Smith.Predictingradiativetransferinaxisym-
metriccylindricalenclosuresusingthediscreteordinatesmethod.Combustion
ScienceandTechnology,62:173–186,1988.
[44]C.B.Ludwig,W.Malkmus,J.G.Reardon,andJ.A.Thomson.Handbookof
InfraredRadiationfromCombustionGases.NASASP-3080,1973.
[45]T.F.Smith,Z.F.Shen,andJ.N.Friedman.Evaluationofcoefficientsforthe
weightedsumofgraygasesmodel.J.HeatTransfer,104:602–608,1982.
[46]M.F.Modest.Theweighted-sum-of-gray-gasesmodelforarbitrarysolution
methodsinradiativetransfer.J.HeatTransfer,113:650–656,1991.

[47]M.Chrigui.Eulerian-LagrangianApproachforModelingandSimulationsof
TurbulentReactiveMulti-PhaseFlowsunderGasTurbineCombustorCondi-
tions.PhDthesis,TechnischeUniversitaetDarmstadt,2005.

Bibliography

19

[48]M.Chrigui,A.Sadiki,J.Janicka,andM.Maneshkarimi.CFDanalysisof
conjugateeffectsofturbulenceandswirlintensityonspraycombustionina
singlegasturbinecombustor.InASMETURBOEXPO,2004.

[49]S.B.Pope.PDFmethodsforturbulentreactiveflows.Prog.EnergyCombust.
Sci.,11:119–192,1985.

[50]H.W.Ge.ProbabilityDensityFunctionModelingofTurbulentNon-reactive
andReactiveSprayFlows.PhDthesis,HeidelbergUniversity,2006.

[51]S.A.MorsiandJ.A.Alexander.Aninvestigationofparticletrajectoriesin
two-phaseflowsystems.J.FluidMech.,55(2):193–208,1972.

[52]Z.F.Tian,J.Y.Tu,andG.H.Yeoh.Numericalsimulationandvalidation
ofdilutegas-particleflowoverabackward-facingstep.AerosolScienceand
Technology,39:319–332,2005.

[53]W.E.RanzandW.J.Marshall.Evaporationfromdrops,partI.Chem.Eng.
Prog.,48:141–146,1952.

[54]W.E.RanzandW.J.Marshall.Evaporationfromdrops,partII.Chem.Eng.
Prog.,48:173–180,1952.

[55]S.Elgobashi.Particle-ladenturbulentflows:directsimulationandclosure
models.AppliedScientificResearch,48:301–314,1991.

[56]R.K.BoydandJ.H.Kent.Three-dimensionalfurnacecomputermodelling.In
21st.Symposium(International)onCombustion,TheCombustionInstitute,
.6891

[57]A.Burcat.CombustionChemistry,chapterThermochemicalDataforCom-
bustion.SpringerVerlag,1984.

[58]R.J.Kee,F.M.Rupley,andJ.A.Miller.Chemicalthermodynamicdata
base.Technicalreport,SandiaReport87-82153,1987.

[59]S.Arrhenius.UeberdieReaktionsgeschwindigkeitbeiderInversionvon
RohrzuckerdurchSaeuren.ZPhys.Chem.,4:226,1889.

[60]R.G.Gilbert,K.Luther,andJ.Troe.Theoryofthermalunimolecularreac-
tionsinthefall-offrange.Ber.Bunsenges.Phys.Chem.,87:169–177,1983.

[61]J.Troe.Predictivepossibilitiesofunimolecularratetheory.J.Phys.Chem.,
83:114–126,1979.

Bibliography

29

[62]P.W.AtkinsandJ.DePaula.PhysicalChemistry.OxfordUniversityPress,
.6002

[63]P.ZhangandC.K.Law.Afittingformulaforthefalloffcurvesofunimolecular
reactions.InternationalJournalofChemicalKinetics,41:727–734,2009.

[64]S.Hafner.ModellentwicklungzurnumerischenSimulationeinesFlugstromver-
gasersfuerBiomasse.PhDthesis,HeidelbergUniversity,2010.

[65]C.Heghes.C1-C4HydrocarbonOxidationMechanism.PhDthesis,Heidel-
bergUniversity,2006.

[66]S.Hafner,A.Rashidi,G.Baldea,andU.Riedel.Adetailedchemicalkinetic
modelofhightemperatureethyleneglycolgasification.CombustionTheory
andModelling,InPress,2010.

[67]U.Maas.MathematischeModellierunginstationaererVerbrennungsprozesse
unterVerwendungdetaillierterReaktionsmechanismen.PhDthesis,Heidel-
bergUniversity,1988.

[68]L.R.Petzold.AdescriptionofDASSL:Adifferential/algebraicsystemsolver.
Technicalreport,SandiaNationalLaboratories,SAND82-8637,1982.

[69]P.Deuflhard,E.Hairer,andJ.Zugck.Onestepandextrapolationmethods
fordifferential-algebraicsystems.Numer.Math.,51:501–516,1987.

[70]J.Warnatz.CalculationofthestructureoflaminarflatflamesI:Flameveloc-
ityoffreelypropagatingozonedecompositionflames.Ber.Bunsenges.Phys.
Chem.,82:193–200,1978.

[71]P.A.LibbyandF.A.Williams.TopicsinAppliedPhysics:TurbulentReacting
Flows.SpringerVerlag,1980.

[72]R.O.Fox.ComputationalModelsforTurbulentReactingFlows.Cambridge
UniversityPress,2003.

[73]N.Peters.Fifteenlecturesonlaminarandturbulentcombustion.InErcoftac
SummerSchool,1992.

[74]E.S.BishandW.J.A.Dahm.Straineddissipationandreactionlayeranalysis
fornon-equilibriumchemistryinturbulentreactingflows.Combustionand
Flame,100:457,1995.

Bibliography

39

[75]N.Peters.Laminardiffusionflameletmodelsinnon-premixedturbulentcom-
bustion.Prog.EnergyCombust.Sci.,10:319–339,1984.

[76]J.S.KimandF.A.Williams.Extinctionofdiffusionflameswithnonunity
Lewisnumbers.JournalofEngineeringMathematics,31:101–118,1997.

[77]H.Pitsch,M.Chen,andN.Peters.Unsteadyflameletmodellingofturbu-
lenthydrogen-airdiffusionflames.In27thSymposiumonCombustion,The
CombustionInstitute,1998.

[78]C.M.Muller,H.Breitbach,andN.Peters.Partiallypremixedturbulent
flamepropagationinjetflames.In25thSymposium(Int)onCombustion,
TheCombustionInstitute,1994.

[79]B.F.Magnussen.Onthestructureofturbulenceandageneralizededdy
dissipationconceptforchemicalreactioninturbulentflow.In19thAIAA
Meeting,1981.

[80]B.F.MagnussenandB.H.Hjertager.Onmathematicalmodelsofturbulent
combustionwithspecialemphasisonsootformationandcombustion.In16th
Symp.(International)onCombustion.TheCombustionInstitute,1976.

[81]I.R.GranandB.F.Magnussen.Anumericalstudyofabluff-bodystabi-
lizeddiffusionflame.part2.influenceofcombustionmodelingandfinite-rate
chemistry.CombustionScienceandTechnology,119:191–217,1996.

[82]M.T.LandahlandE.Mollo-Christensen.TurbulenceandRandomProcesses
inFluidMechanics.CambridgeUniversityPress,1992.

[83]B.F.Magnussen.Modelingofpollutantformationingasturbinecombustors
basedontheeddydissipationconcept.In18thInternationalCongresson
CombustionEngines,1989.

[84]W.J.Minkowycz,E.M.Sparrow,andJ.Y.Murthy,editors.Handbookof
NumericalHeatTransfer.JohnWileyandSons,2006.

[85]T.J.Chung.ComputationalFluidDynamics.CambridgeUniversityPress,
02.20

[86]J.N.Reddy.AnintroductiontoFiniteElementMethod.Mc-Graw-Hill,1993.

[87]P.Wesseling.PrinciplesofComputationalFluidDynamics.SpringerVerlag,
.0002

Bibliography

49

[88]A.J.Chorin.Numericalsolutionofnavier-stokesequations.Mathematicsof
Computation,22:745–762,1968.

[89]M.H.Kobayashia,J.M.C.Pereira,andJ.C.F.Pereira.Aconservativefinite-
volumesecond-order-accurateprojectionmethodonhybridunstructuredgrids.
JournalofComputationalPhysics,150:40–75,1999.

[90]A.S.Almgren,J.B.Bell,P.Colella,L.H.Howell,andM.L.Welcome.Acon-
servativeadaptiveprojectionmethodforthevariabledensityincompressible
Navier-Stokesequations.JournalofComputationalPhysics,142:1–46,1998.

[91]J.B.Bell,P.Colella,andH.M.Glaz.Asecond-orderprojectionmethod
fortheincompressibleNavier-Stokesequations.JournalofComputational
Physics,85:257–283,1989.

[92]S.R.MathurandJ.Y.Murthy.Apressure-basedmethodforunstructured
meshes.NumericalHeatTransfer,PartB:Fundamentals,31:195–215,1997.

[93]B.R.HutchinsonandG.D.Raithby.Amultigridmethodbasedonadditive
correctionstrategy.Numer.HeatTransfer,9:511–537,1986.

[94]W.Hackbusch.Multi-GridMethodsandApplications.SpringerVerlag,1985.

[95]S.V.Patankar.NumericalHeatTransferandFluidFlow.HemispherePub-
lishingCorporation,1980.

[96]C.M.RhieandW.L.Chow.Numericalstudyoftheturbulentflowpastan
airfoilwithtrailingedgeseparation.AIAAJournal,21:1525–1532,1983.

[97]S.B.Pope.Computationallyefficientimplementationofcombustionchemistry
usinginsituadaptivetabulation.Combust.TheoryModelling,1:41–63,1997.

[98]J.J.ShahandR.O.Fox.Computationalfluiddynamicssimulationofchemical
reactors:Applicationofinsituadaptivetabulationtomethanethermochlori-
nationchemistry.Ind.Eng.Chem.Res.,38:4200–4212,1999.

[99]T.Kolb,T.Jakobs,andN.Zarzalis.Syngasfrombiomass-basedslurry
entrained-flowgasification.In10thConferenceonEnergyforaCleanEn-
vironment,2009.

[100]H.J.Wiemer,D.Kuhn,U.Santo,M.Eberhard,T.Kolb,E.Pantouflas,
andN.Zarzalis.ThermochemischeUmsetzungvonMatrixfluessigkeitin

Bibliography

59

einemFlugstromvergaserbeiderVariationdesLuftbedarfs.InVDI-GET-
FachtagungVerbrennungundFeuerung-23.DeutscherFlammentag,vol.VDI-
Bericht,2007.

[101]PersonalcorrespondanceswithKIT,2007-2009.

[102]W.Sutherland.LII.theviscosityofgasesandmolecularforce.Philosophical
Magazine,Series5,36:507–531,1893.

[103]Crane.Flowoffluidsthroughvalves,fittingsandpipe.Technicalreport,No.
410M,CraneCo.,1988.

[104]A.Rashidi,M.Eberhard,andU.Riedel.CFDsimulationofgasificationpro-
cesswithethyleneglycolasmodelsubstanceforbiomassbasedpyrolysisoil.
InternationalJournalofEnergyforaCleanEnvironment,2010.

[105]A.Bennardo,F.Podenzani,andL.Borello.Aflarecase:comparisonof
theresultsoftwoCFDcodeswithexperimentaldata.In10thConferenceon
EnergyforaCleanEnvironment,2009.

[106]P.MathieuandR.Dubuisson.Performanceanalysisofabiomassgasifier.
EnergyConversionandManagement,43:1291–1299,2002.

[107]H.R.Weir.HighpressuregasificationofcoalinGermany.Industrialand
EngineeringChemistry,39:48–54,1947.

[108]R.P.Bell.Thetheoryofreactionsinvolvingprotontransfers.Proceedingsof
theRoyalSocietyofLondon.SeriesA,MathematicalandPhysicalSciences,
154:414–429,1936.

[109]M.G.EvansandM.Polanyi.Inertialanddrivingforceofchemicalreactions.
TransactionsofFaradaySociety,34:11–24,1938.

[110]T.PoinsotandD.Veynante.TheoreticalandNumericalCombustion.Edwards,
.5002

[111]F.Tabet-Helal,B.Sarh,M.Birouk,andI.Goekalp.Numericalinvestigation
ofthenear-fieldregionofaturbulentnon-premixed(CH4-H2-N2)/airflame.
InProceedingsof3rdEuropeanCombustionMeeting,2007.

[112]K.ClaramuntAltimira.NumericalSimulationofNon-PremixedLaminarand
TurbulentFlamesbymeansofFlameletModellingApproaches.PhDthesis,
UniversitatPolitecnicaCatalunya,2005.

Bibliography

]311[

]411[

[]511

96

F.Liu,H.Guo,andG.J.Smallwood.Evaluationofthelaminardiffusion
flameletmodelinthecalculationofanaxisymmetriccoflowlaminarethylene-
airdiffusionflame.CombustionandFlame,144:605–618,2006.

B.A.Albrecht,S.Zahirovic,R.J.M.Bastiaans,J.A.vanOijen,andL.P.H.
deGoey.Apremixedflamelet-PDFmodelforbiomasscombustioninagrate
furnace.EnergyandFuels,22:1570–1580,2008.

F.Tabet-Helal,B.Sarh,andI.Goekalp.Acomparativestudyofturbulence
modellingindilutedhydrogennon-premixedflames.IFRFIndustrialCom-
bustion,Nr.200803:1–41,2008.

endicesppA.A

A.1Nomenclature

aaapAAApnbcicpC1ε,C2ε
C,C21CiCDddpDDmi,D,iTEEaEpffpn

StrainRate
AbsorptionCoefficient
EquivalentAbsorptionCoefficientofParticles
Pre-exponentialFactorinArrheniusFormula
SurfaceArea
ProjectedAreaofParticlen
TemperatureExponentinArrheniusFormula
ConcentrationofSpeciesi
SpecificHeatCapacityatConstantPressure
ConstantsofStandardk-εModel
ConstantsofRealizablek-εModel
VaporConcentration
DragCoefficient
DistanceBetweenJetNozzles
ParticleDiameter
Diffusivity
MassDiffusionCoefficient
ThermalDiffusionCoefficient
TotalEnergy
ActivationEnergy
EquivalentParticleEmission
MixtureFraction
ScatteringFactorofParticlen

A.1.Nomenclature

FFFDgGGkhhhfgHHlatHref,latIJikkck,krfk∞KcLeLeHVLmm˙M,iwni,f,ni,r
NNiuNpp0psatrPQradReReRij

LorentzianBroadeningFactor
ForceVector
DragForce
GravitationalAcceleration
IncidentRadiation
ProductionofTurbulentKineticEnergy
SpeciesEnthalpy
ConvectiveHeatTransferCoefficient
LatentHeatofEvaporation
TotalEnthalpy
LatentHeatatBoilingPoint
LatentHeatatReferenceCondition
RadiationIntensity
DiffusionFlux
TurbulentKineticEnergy
MassTransferCoefficient
RateConstantforForward/ReverseReactions
ThermalConductivityofContinuousPhase
EquilibriumConstant
EddyLengthScale
LewisNumber
LowerHeatingValue
ssMaMassFlowRate
MolecularWeightofSpeciesi
ReactionOrdersofForward/ReverseReactions
NumberofChemicalSpecies
FluxofDropletVaporintoGasPhase
NusseltNumber
eressuPrAtmosphericPressure
SaturatedVaporPressure
PrandtlNumber
RadiativeHeatFlux
UniversalGasConstant
ReynoldsNumber
ReynoldsStresses

89

A.1.Nomenclature

sSSijcSthSttsscroTTbpTapvTwuiupvVwharcxiXiYi∗YiZi

αδpnεζηηGηkθRλλλeffλtµ

Direction(Radiation)
SourceTerm
MeanRate-of-StrainTensor
TurbulentSchmidtNumber
SherwoodNumber
emTiParticleEddyCrossingTime
Temperature
BoilingTemperature
EvaporationTemperature
GasifierWallTemperature
VelocityMagnitude
ParticleVelocity
OverallVelocityVector
Volume
CharMassFractioninSlurry
Direction
MoleFractionofSpeciesi
MassFractionofSpeciesi
FineScaleMassFractionofSpeciesi
MassFractionofElementi

Bell-Evans-PolanyiFactor
DeltaFunction
TotalEmissivity
EmissivityofParticlen
TurbulentDissipationRate
NormallyDistributedRandomNumber
MeanStrain
GasificationEfficiency
KolmogorovLengthScale
RadiationTemperature
Air-FuelRatio
ThermalConductivity
EffectiveThermalConductivity
TurbulentThermalConductivity
DynamicViscosity

99

A.1.Nomenclature

µtνν,ν∗ξρρ∞σσ,σεkσpσsτ∗ττ¯χωk0HΔj0SΔjtΔΦΩΩij

TurbulentViscosity
KinematicViscosity
StoichiometricCoefficientforReactants,Products
LengthFractionofFineStructures
ytsienDDensityoftheOxidizerStream
Stefan-BoltzmannConstant
kandεPrandtlNumber
EquivalentParticleScatteringCoefficient
ScatteringCoefficient
TimeScale
TimeScaleofFineStructures
StressTensor
ScalarDissipationRate
AngularVelocity
EnthalpyofReactionj
EntropyofReactionj
TimeStep
PhaseFunction
SolidAngle
MeanRate-of-RotationTensor

001

A.2.ReactionMechanism

110

A.2ReactionMechanism
Thereactionmechanismforhightemperatureoxidationofethyleneglycol,usedin
thisthesis,isbasedonthechemicalreactionmechanismdevelopedbyHafner[64].
TheoriginalmechanismissimplifiedusingthemethodsdiscussedinChapter4.
ThereactionratesaredefinedinmodifiedArrheniusform(Equation4.8)
Ekf=ATb∙e(−RaT).Theunitsofthepre-exponentialfactorAandactivationen-
ergyEaare(cm3mol−1)n−1s−1andkJ/mol,respectively.Theratecoefficientsof
reversereactionsarecalculatedasdiscussedinChapter4.
Thecollisionefficienciesusedareasfollows:
M(1)=[H2]+6,5[H2O]+0,4[O2]+0,35[AR]+0,4[N2]+0,75[CO]+1,5[CO2]+3,0[EthGly]
M(2)=[H2]+2,55[H2O]+0,4[O2]+0,15[AR]+0,4[N2]+0,75[CO]+1,5[CO2]+3,0[EthGly]
M(3)=[H2]+6,5[H2O]+0,4[O2]+0,29[AR]+0,4[N2]+0,75[CO]+1,5[CO2]+3,0[EthGly]
M(4)=2,0[H2]+5,0[H2O]+2,0[CO]+3,0[CO2]+3,0[EthGly].

#

nsioctReaHCOOHM(1)+OHHCO1M(1)+OHHCO2OH+OHHCO3OH+OHHCO4H+OHHCO5H+OHHCO67HCOOH+CH3
8HCOOH+HO2
9O+OHHCO

CHOCHOReactions
CHO+HOC10HOCHOC1213CHOCHO+OH
O+HOCHOC14H+HOCHOC15M(4)+HOCHOC1719CHOCHO+HO2
20CHOCHO+CH3
21CHOCHO+O2

HOCHCOReactions
H+CHCOHO22

EbAa

→H2O+CO+M(1)2.090∙10140.0169.026
→H2+CO2+M(1)1.350∙10150.0253.54
→CO2+H2O+H2.620∙10062.0563.832
→CO+H2O+OH1.850∙10071.5−4.025
→CO2+H2+H4.240∙10062.120.367
→CO+H2+OH6.060∙1013−0.3512.501
→CO+CH4+OH3.90∙10−075.809.204
→CO+H2O2+OH2.40∙1019−2.2058.699
→CO+OH+OH1.770∙1018−1.9012.447

CHOCHO1.00∙10130.00.0
→CO+CO+H24.070∙1042−8.5289.847
→CHO+CO+H2O1.00∙10130.00.0
→CHO+CO+OH7.240∙10120.08.242
CH2O+CHO1.00∙10120.00.0
CHO+CHO+M(4)4.270∙10120.0211.72
→CHO+CO+H2O21.70∙10120.044.767
→CHO+CO+CH41.740∙10120.035.311
→CHO+CO+HO21.00∙10140.0154.801

CH2OH+CO2.710∙10042.7504.03

A.2.ReactionMechanism

24HOCHCO+OHCOOH+CO3.613∙10110.06.98
26HOCHCO+OHCH2OH+CO26.239∙10110.05.6
28HOCHCO+OHHCOOH+CHO0.337∙10110.04.19
30HCOOH+CHHOCHCO+H9.460∙10130.0−2.1
32HOCHCO+M(1)CH2O+CO+M(1)3.00∙10140.00298.51
LOW3.60∙10150.00249.48
TROE0.500.00.00.0
34HOCHCO+O2CO2+HCOOH1.00∙10080.0−0.05

HOCHCHOReactions
36HOCHCHO+HHOCHCO+H22.00∙10130.03.19
38HOCHCHO+OHHOCHCO+H2O3.00∙10130.03.19
40HOCH2CHO+O2HOCHCHO+HO22.00∙10130.0217.79
42HOCHCHO+HCHOCHO+H23.00∙10130.08.1
44HOCHCHO+OHCHOCHO+H2O1.510∙10130.08.1
46HOCHCHO+O2CHOCHO+HO28.430∙1015−1.28.1
48HOCHCHO+O2CHOCHO+HO24.820∙10140.028.2
50HOCHCHO+M(4)CHOCHO+H+M(4)1.00∙10140.0112.77
52HOCHCHO+OCHOCHO+OH1.00∙10140.08.1
54HOCHCHO+HHOCH2CO+H5.00∙10120.01.34
56HOCHCHO+HCHO+CH2OH5.00∙10130.04.23
58HOCHCHOCO+CH2OH1.170∙1043−9.83187.31
60HOCHCO+M(1)+HHOCHCHO+M(1)3.30∙1014−0.0632.38
62HOCHCHO+OHOCHCO+OH2.00∙10130.019.93

HOCH2COReactions
64HOCH2CO+HHOCHCO+H22.580∙10071.654.95
66HOCH2CO+OHHOCHCO+H2O4.640∙10110.15−6.88
68HOCH2CO+OHOCHCO+OH1.880∙10071.850.75
70HOCH2CO+HO2HOCHCO+H2O28.20∙10032.5538.1
72HOCH2CO+CH3HOCHCO+CH47.280∙10022.9926.38
74C3H5OH+OCH3+HOCH2CO5.00∙10120.00.12
76HOCH2CO+HCHO+CH2OH9.60∙10130.0−4.87
78HOCH2CO+OCO2+CH2OH1.50∙10140.0−4.87
80HOCH2CO+HO2→CO2+CH2OH+OH3.00∙10130.0−4.87
81HOCH2CO+OH→CO+CH2OH+OH3.00∙10130.0−4.87
82CH2OH+CO+M(1)HOCH2CO+M(1)5.058∙10110.0025.89
LOW3.109∙10140.0013.5
TROE0.500.00.00.0

HOCH2CHOReactions
84HOCH2CHO+M(1)CH2OH+CHO+M(1)2.20∙10150.0348.04
LOW5.10∙10120.0136.64
TROE0.500.00.00.0
86HOCH2CHO+HHOCH2CO+H22.047∙10091.1612.41

201

A.2.ReactionMechanism

88HOCH2CHO+H
90HOCH2CHO+O
92HOCH2CHO+O
94HOCH2CHO+OH
96HOCH2CHO+OH
98HOCH2CHO+OH
100HOCH2CHO+O2
102HOCH2CHO+HO2
104HOCH2CHO+HO2
106HOCH2CHO+CH3
108HOCH2CHO+CH3
110CH2CH2OH+O2
112HOCH2CO+H+M
114HOCH2CHO+CHO

R−CHOHReactions
116R−CHOH+M
118R−CHOH+H
120R−CHOH+H
122R−CHOH+O
124R−CHOH+OH
126R−CHOH+O2
128R−CHOH+O2
130R−CHOH+HO2

R−CH2OReactions
132R−CH2O
134R−CH2O
136R−CH2O+H
138R−CH2O+H
140R−CH2O+O
142R−CH2O+OH
144R−CH2O+O2
146R−CH2O+CO

EthGlynsioctReaM(4)+EthGly148

M(4)+EthGly150

M(4)+EthGly152

HOCHCHO+H2
HOCH2CO+OH
HOCHCHO+OH
HOCH2CO+H2O
HOCHCHO+H2O
CH2OH+HCOOH
HOCH2CO+HO2
HOCH2CO+H2O2
HOCHCHO+H2O2
HOCH2CO+CH4
HOCHCHO+CH4
HOCH2CHO+OH
HOCH2CHO+M
HOCH2CO+CH2O

2.580∙10071.658.01
1.770∙1018−1.9014.81
1.880∙10071.853.81
9.240∙10061.50−1.68
4.640∙10110.15−3.82
3.00∙1015−1.0765.24
2.00∙10130.50175.63
2.40∙1019−2.2061.09
8.20∙10032.5541.16
3.90∙10−075.8011.56
7.280∙10022.9929.44
4.90∙1011−0.4830.13
9.60∙10130.00−2.35
7.80∙10130.0037.67

HOCH2CHO+H+M1.00∙10140.0106.68
HOCH2CHO+H23.00∙10130.02.01
CH2OH+CH2OH3.00∙10130.07.25
HOCH2CHO+OH1.00∙10140.02.01
HOCH2CHO+H2O1.510∙10130.02.01
HOCH2CHO+HO28.432∙1015−1.22.01
HOCH2CHO+HO24.820∙10140.022.11
HOCH2CHO+OH+OH4.00∙10130.02.01

HOCH2CHO+H2.00∙10140.099.58
CH2O+CH2OH1.50∙10150.097.78
HOCH2CHO+H21.00∙10140.02.58
CH2OH+CH2OH3.00∙10130.07.82
HOCH2CHO+OH1.210∙10140.02.58
HOCH2CHO+H2O1.00∙10140.02.58
HOCH2CHO+HO26.00∙10100.09.58
CH2CH2OH+CO24.680∙10023.1629.95

CH2OH+CH2OH+M(4)5.94∙1023−1.68390.66
LOW3.11∙1085−18.84482.21
TROE0.50550.0825.06100.0
CH2CH2OH+OH+M(4)2.50∙1023−1.54410.55
LOW6.50∙1085−18.81489.79
TROE0.50300.0900.05000.0
CH3CHO+H2O+M(4)3.720∙10130.09281.99
LOW3.43∙1083−18.85367.05
TROE0.70350.0800.03800.0

130

A.2.ReactionMechanism

154EthGly+M(4)HOCH2CHO+H2+M(4)1.448∙10120.10381.03
LOW8.92∙1087−19.42493.56
TROE0.90900.01100.03500.0
156EthGly+HR−CHOH+H25.160∙10071.6513.56
158EthGly+HR−CH2O+H23.00∙10071.613.87
160EthGly+OR−CHOH+OH3.760∙10071.859.36
162EthGly+OR−CH2O+OH3.160∙10072.019.77
164EthGly+OHR−CHOH+H2O9.280∙10110.151.73
166EthGly+OHR−CH2O+H2O1.492∙10120.308.0
168EthGly+HO2R−CHOH+H2O21.640∙10042.5546.71
170EthGly+HO2R−CH2O+H2O25.00∙10120.0101.58
172EthGly+CH3R−CHOH+CH41.456∙10032.9934.99
174EthGly+CH3R−CH2O+CH42.90∙10022.9933.16

156EthGly+HR−CHOH+H2
158EthGly+HR−CH2O+H2
160EthGly+OR−CHOH+OH
162EthGly+OR−CH2O+OH
164EthGly+OHR−CHOH+H2O
166EthGly+OHR−CH2O+H2O
168EthGly+HO2R−CHOH+H2O2
170EthGly+HO2R−CH2O+H2O2
172EthGly+CH3R−CHOH+CH4
174EthGly+CH3R−CH2O+CH4

OxyhydrogenandCO/CO2System

H2/O2Reactions
176O2+HOH+O2.650∙1016-0.6771.3
178H2+OOH+H3.818∙10120.033.256
180H2+OOH+H1.025∙10150.080.230
182H2+OHH2O+H2.168∙10081.52014.466
184OH+OHH2O+O3.348∙10042.420-8.064
186H+H+M(1)H2+M(1)1.015∙1017-0.600.0
188O+O+M(1)O2+M(1)5.40∙10130.0-7.4
190H+OH+M(2)H2O+M(2)5.560∙1022-2.00.0

HO2Reactions
192H+O2+M(3)HO2+M(3)1.746∙10170.00.0
LOW2.367∙1019-1.200.0
TROE0.50.00.00.0
194HO2+HOH+OH4.457∙10140.05.819
196HO2+HH2+O21.054∙10140.08.563
198HO2+HH2O+O1.445∙10120.00.0
200HO2+OOH+O21.626∙10130.0-1.862
202HO2+OHH2O+O29.275∙10150.073.246

H2O2Reactions
204HO2+HO2H2O2+O24.220∙10140.050.140
206HO2+HO2H2O2+O21.325∙10110.0-6.820
208OH+OH+M(1)H2O2+M(1)1.566∙10130.00.0
LOW5.980∙1019-0.80.0
TROE0.500.00.00.0
210H2O2+HH2+HO21.686∙10120.015.713
212H2O2+HH2O+OH1.024∙10130.014.970
214H2O2+OOH+HO24.216∙10110.016.628

401

A.2.ReactionMechanism

216H2O2+O
218H2O2+OH
220H2O2+OH

sionctReaCOO+OC222OH+OC224OH+OC226OH+OC228230CO+HO2
232CO+O2

OxidatiCno1

nsioctReaC+HC234HO+C2362

nsoctiReaCHO+HC238OH+HC240242CH+O2
CO+HC244

246CH+CO2
248CH+H2O
250CH+H2O

nsioctReaCHO252CHO+M(1)
H+HOC254O+HOC256O+HOC258OH+HOC260262CHO+O2
CHO+HOC264266CHO+HO2
267CHO+O2

CH2Reactions
2693CH2+H
2713CH2+O
2723CH2+O
2743CH2+O2

OH2OH2OH2

M(1)+CO2CO2CO2CO2CO2

CCO

COCHOCHOM(2)+

CHOOCH23CH2

COCOCOCO2COCOOCH2CO→2CO2

CHCO→COCO

+O24.216∙10110.016.628
+HO21.64∙10180.0123.047
+HO21.92∙10120.01.787

CO2+M(1)1.540∙10150.012.560
+H1.05∙10130.066.927
+H9.034∙10110.019.120
+H1.012∙10110.00.249
+OH1.50∙10140.098.70
+O2.50∙10120.0200.0

H+2O+

5.0∙10140.00.0
6.023∙10130.02.66

+H4.0∙10130.00.0
+H3.0∙10130.00.0
+O1.686∙10130.00.0
HCCO+M(2)1.024∙1015-0.40.0
LOW3.790∙1000-2.50.0
TROE0.600.00.00.0
+CO6.384∙10071.51-2.993
+H4.577∙1016-1.420.0
+OH4.577∙1016-1.420.0

+H+M(1)1.860∙1017-1.071.13
+H29.034∙10130.00.0
+OH3.011∙10130.00.0
+H3.011∙10130.00.0
+H2O1.084∙10140.00.0
+HO22.710∙10100.68-1.962
+CO3.0∙10130.00.0
+H+OH3.0∙10130.00.0
+OH1.510∙10120.00.0

+H21.204∙10140.00.0
+H+H1.228∙10140.02.244
+H28.191∙10130.02.244
+OH+H1.806∙10120.00.0

501

A.2.ReactionMechanism

2763CH2+O2CO2+H21.806∙10120.00.0
2783CH2+3CH2C2H2+H21.806∙10140.049.884
2803CH2+3CH2C2H2+H+H1.626∙10150.049.884
2823CH2+CH3C2H4+H7.227∙10130.00.0
2841CH2+M(1)3CH2+M(1)6.023∙10120.00.0
2861CH2+H2CH3+H1.260∙1016-0.5666.5
2881CH2+O2CO+OH+H3.10∙10130.00.0
2903CH2+OHH+CH2O2.50∙10130.00.0
2923CH2+CO2CO+CH2O1.10∙10110.04.184
2943CH2+O2O+CH2O3.290∙1021-3.311.999
2963CH2+O2→CO2+H+H3.290∙1021-3.311.999
2973CH2+O2CO+H2O7.280∙1019-2.547.569
2993CH2+O2CHO+OH1.290∙1020-3.31.188
3011CH2+CH4CH3+CH34.0∙10130.00.0
3031CH2+C2H6CH3+C2H51.20∙10140.00.0
3051CH2+O→CO+H+H3.0∙10130.00.0
3061CH2+OHCH2O+H3.0∙10130.00.0
3081CH2+HCH+H23.0∙10130.00.0
3101CH2+CO2CH2O+CO3.0∙10120.00.0
3121CH2+CH2COC2H4+CO1.60∙10140.00.0

CH2OReactions
314CH2O+M(1)CHO+H+M(1)4.872∙10150.0316.348
316CH2O+M(1)CO+H2+M(1)2.830∙10150.0266.962
318CH2O+HCHO+H22.190∙10081.7712.560
320CH2O+OCHO+OH4.155∙10110.5711.556
322CH2O+OHCHO+H2O7.20∙10052.46-4.06
324CH2O+HO2CHO+H2O24.095∙10042.542.734
326CH2O+O2CHO+HO22.439∙10052.5152.562
328CH2O+CH3CHO+CH43.192∙10013.3618.041
330CH2O+CHCH2CO+H9.460∙10130.0-2.155

CH2OHReactions
332CH2OH+M(1)CH2O+H+M(1)2.80∙1014-0.73137.306
LOW1.50∙1034-5.39151.456
TROE0.9667.21855.07543.0
334CH2OH+HCH2O+H22.445∙10130.00.0
336CH2OH+HCH3+OH1.048∙10130.00.0
338CH2OH+O2CH2O+HO22.891∙1016-1.50.0
340CH2OH+O2CH2O+HO27.230∙10130.014.97

CH3Reactions
342CH3+M(1)3CH2+H+M(1)2.922∙10160.0379.0
344CH3+M(1)CH+H2+M(1)1.892∙10160.0355.839
346CH3+OCH2O+H6.745∙10130.00.0

601

A.2.ReactionMechanism

348CH3+OH→CH3O+H1.204∙10100.058.114
349CH3+OH1CH2+H2O3.0∙10130.011.640
351CH3+OH+M(1)CH3OH+M(1)4.336∙1015-0.790.0
LOW1.098∙1038-6.215.578
TROE0.252101434.00.0
353CH3+HO2CH3O+OH1.80∙10130.00.0
355CH3+O2→O+CH3O4.20∙10130.0135.851
356CH3+CO+M(1)CH3CO+M(1)5.058∙10110.028.77
LOW3.109∙10140.015.88
TROE0.500.00.00.0
358CH3+1CH2C2H4+H7.227∙10130.00.0
360CH3+CH3+M(1)C2H6+M(1)3.613∙10130.00.0
LOW3.627∙1041-7.011.60
TROE0.6273.01180.00.0
362CH3+O2CH2O+OH2.510∙10110.061.295
364CH3+CHC2H3+H3.0∙10130.00.0

CH3OReactions
366CH3O+M(1)CH2O+H+M(1)6.80∙10130.0109.49
LOW4.660∙1025-3.0101.68
TROE0.450.00.00.0
368CH3O+H→CH3+OH1.626∙10130.02.494
369CH3O+HCH2O+H23.794∙10130.02.494
371CH3O+O→O2+CH31.129∙10130.00.0
372CH3O+OOH+CH2O3.764∙10120.00.0
374CH3O+OHCH2O+H2O1.810∙10130.00.0
376CH3O+O2CH2O+HO22.168∙10100.07.3
378CH3O+CH2OCH3OH+CHO1.150∙10110.05.2
380CH3O+COCH3+CO24.680∙10023.1622.525

CH4Reactions
382CH4+M(1)CH3+H+M(1)2.80∙10160.0439.0
LOW5.50∙1047-8.2492.180
TROE0.013501.07834.0
384CH4+HH2+CH34.143∙10052.540.115
386CH4+OOH+CH34.396∙10052.527.519
388CH4+OHH2O+CH31.050∙10062.1811.223
390CH4+HO2H2O2+CH34.697∙10042.587.879
392CH4+CHC2H4+H1.325∙1016-0.940.241
394CH3+HO2CH4+O23.0∙10120.00.0
396CH4+3CH2CH3+CH34.0∙10130.00.0

384CH4+HH2+CH3
386CH4+OOH+CH3
388CH4+OHH2O+CH3
390CH4+HO2H2O2+CH3
392CH4+CHC2H4+H
394CH3+HO2CH4+O2
396CH4+3CH2CH3+CH3

CH3OHReactions
398CH3OH+HCH2OH+H2
400CH3OH+HCH3O+H2

2.746∙10091.2418.789
6.866∙10081.2418.789

017

A.2.ReactionMechanism

402CH3OH+OCH2OH+OH
404CH3OH+OCH3O+OH
406CH3OH+OHCH2OH+H2O
408CH3OH+OHCH3O+H2O
410CH3OH+HO2CH2OH+H2O2
412CH3OH+O2HO2+CH2OH
414CH3OH+CH3CH4+CH2OH
416CH3OH+CH3CH4+CH3O
418CH3OH+CH3OCH2OH+CH3OH
420CH3OH+CH2O→CH3O+CH3O

noOxidatiC2

1.975∙10130.022.198
4.938∙10120.022.198
5.273∙10061.92-1.197
0.930∙10061.92-1.197
0.620∙10130.081.10
2.050∙10130.0189.10
9.937∙10003.4533.422
2.017∙10013.4533.422
1.50∙10120.029.30
0.153∙10130.0333.20

HCCOReactions
421HCCO+H3CH2+CO1.060∙10130.00.0
423HCCO+OCO+CO+H1.250∙10140.00.0
425HCCO+3CH2C2H3+CO3.0∙10130.00.0
427HCCO+O2CO2+CHO2.40∙10110.0-3.576
429HCCO+OCO2+CH2.950∙10130.04.66

C2H2Reactions
431C2H2+O3CH2+CO1.10∙10081.49.228
433C2H2+OHCCO+H7.0∙10081.49.228
435C2H2+O2HCCO+OH4.0∙10071.5126.0
437C2H2+OHCH2CO+H2.18∙10−044.50-4.187
439C2H2+OHCH2CO+H2.0∙10110.00.0

CH2COReactions
441CH2CO+M(1)3CH2+CO+M(1)3.0∙10140.0297.179
LOW3.60∙10150.0248.152
TROE0.500.00.00.0
443CH2CO+HCH3+CO2.710∙10042.7502.989
445CH2CO+OCH2O+CO3.613∙10110.05.653
447CH2CO+O→CHO+H+CO1.806∙10110.05.653
448CH2CO+OCHO+CHO1.806∙10110.05.653
450CH2CO+OHCH3+CO26.239∙10110.04.240
452CH2CO+OHCH2O+CHO0.337∙10110.04.240
454CH2CO+O3CH2+CO21.750∙10120.05.648
456CH2CO+HHCCO+H22.0∙10140.033.471
458CH2CO+OHCCO+OH1.0∙10130.033.471
460CH2CO+OHHCCO+H2O1.0∙10130.08.368
462CH2CO+OHCH2OH+CO3.730∙10120.0-4.238
464CH2CO+O2CO2+CH2O1.0∙10080.00.0

801

A.2.ReactionMechanism

C2H3Reactions
466C2H3+M(1)

468C2H3+H
470C2H3+O
472C2H3+O
474C2H3+O
476C2H3+OH
478C2H3+O2
480C2H3+O2
482C2H3+O2
484C2H3+O

CH3COReactions
486CH3CO+H
488CH3CO+H
490CH3CO+O
492CH3CO+HO2
493CH3CO+OH
494CH3CO+OH
496CH3CO+O
498CH3CO+CH3

CH2CHOReactions
500CH2CHO+H
502CH2CHO+H
504CH2CHO+H
506CH2CHO+OH
508CH2CHO+O
510CH2CHO+CH
CHOHC5122514CH2CO+M(1)+H
516CH2CHO+O
518CH2CHO+OH
520CH2CHO+HO2

C2H4Reactions
521C2H4+M(1)
523C2H4+M(1)
525C2H4+H

526C2H4+H
528C2H4+O

C2H2+H+M(1)7.80∙10081.62155.056
LOW3.237∙1027-3.40149.818
TROE0.350.00.00.0
C2H2+H24.216∙10130.00.0
C2H2+OH3.011∙10130.00.0
CH3+CO3.011∙10130.00.0
CHO+3CH23.011∙10130.00.0
C2H2+H2O5.0∙10120.00.0
CH2O+CHO9.0∙10120.0-0.997
C2H2+HO24.650∙10110.0-1.039
CH2CHO+O5.50∙1014-0.6122.023
CH2CO+H3.0∙10130.00.0

C2H2+H2
C2H2+OH
CH3+CO
CHO+3CH2
C2H2+H2O
CH2O+CHO
C2H2+HO2
CH2CHO+O
CH2CO+H

CH2CO+H22.0∙10130.00.0
CHO+CH39.60∙10130.00.0
CO2+CH31.50∙10140.00.0
→CO2+CH3+OH3.0∙10130.00.0
→CO+CH3+OH3.0∙10130.00.0
CH2CO+H2O1.20∙10130.00.0
CH2CO+OH2.0∙10130.00.0
CH2CO+CH45.0∙10130.00.0

CH2CO+H22.0∙10130.00.0
CH3CO+H5.0∙10120.00.0
CHO+CH35.0∙10130.00.0
CHO+CH2OH3.010∙10130.00.0
CHO+CH2O1.0∙10140.00.0
CHO+C2H31.0∙10140.00.0
CO+CH31.170∙1043-9.83183.08
CH2CHO+M(1)3.30∙1014-0.0635.57
CH2CO+OH2.0∙10130.016.74
CH2CO+H2O3.0∙10130.00.0
→CH2O+OH+CHO7.0∙10120.00.0

C2H2+H2+M(1)4.50∙10171.0327.488
C2H3+H+M(1)7.399∙10170.0404.060
+M(1)→C2H5+M(1)3.975∙10091.2805.40
LOW1.178∙10190.03.20
TROE0.7640.010250.0
C2H3+H24.0∙10023.6247.140
CH2CHO+H4.743∙10061.880.764

901

A.2.ReactionMechanism

530C2H4+OCHO+CH31.020∙10071.880.764
532C2H4+OCH2CO+H26.770∙10051.880.764
534C2H4+OHC2H3+H2O1.070∙10130.024.9
536C2H4+1CH2C3H67.240∙10130.00.0
538C2H4+CH3C2H3+CH46.023∙10071.5669.60

CH3CHOReactions
540CH3CHO+M(1)CH3+CHO+M(1)2.20∙10150.0342.8
LOW5.10∙10120.0131.4
TROE0.500.00.00.0
542CH3CHO+HCH3CO+H22.047∙10091.1610.059
544CH3CHO+HCH2CHO+H21.70∙10091.1610.059
546CH3CHO+OHCH3CO+H2O9.240∙10061.50-4.028
548CH3CHO+OHCH2CHO+H2O2.023∙10071.35-6.584
550CH3CHO+OHCH3+HCOOH3.70∙1015-1.0760.0
552CH3CHO+OCH3CO+OH1.770∙1018-1.9012.456
554CH3CHO+OCH2CHO+OH3.720∙1013-0.2014.888
556CH3CHO+CH3CH3CO+CH43.90∙10−075.809.211
558CH3CHO+CH3CH2CHO+CH42.450∙10013.15023.978
560CH3CHO+HO2CH3CO+H2O22.40∙1019-2.2058.741
562CH3CHO+HO2CH2CHO+H2O22.320∙10110.4062.233
564CH3CHO+O2CH3CO+HO22.0∙10130.50173.28
566C2H5+O2CH3CHO+OH4.90∙1011-0.48034.989
568CH2CHO+HO2CH3CHO+O23.0∙10120.00.0
570CH3CO+H+MCH3CHO+M9.60∙10130.00.0
572CH3CHO+CHOCH3CO+CH2O7.80∙10130.035.315

542CH3CHO+HCH3CO+H2
544CH3CHO+HCH2CHO+H2
546CH3CHO+OHCH3CO+H2O
548CH3CHO+OHCH2CHO+H2O
550CH3CHO+OHCH3+HCOOH
552CH3CHO+OCH3CO+OH
554CH3CHO+OCH2CHO+OH
556CH3CHO+CH3CH3CO+CH4
558CH3CHO+CH3CH2CHO+CH4
560CH3CHO+HO2CH3CO+H2O2
562CH3CHO+HO2CH2CHO+H2O2
564CH3CHO+O2CH3CO+HO2
566C2H5+O2CH3CHO+OH
568CH2CHO+HO2CH3CHO+O2
570CH3CO+H+MCH3CHO+M
572CH3CHO+CHOCH3CO+CH2O

C2H5Reactions
574C2H5+M(1)→C2H4+H+M(1)4.10∙10130.0166.80
LOW3.654∙10180.0139.68
TROE0.7597.013790.0
575C2H5+HCH3+CH31.0∙10140.00.0
577C2H5+OCH2O+CH33.975∙10130.00.0
579C2H5+O2C2H4+HO22.410∙10100.00.0
581C2H5+CH3C2H4+CH49.034∙10110.00.0
583C2H5+C2H5C2H4+C2H61.40∙10120.00.0
585C2H5+HC2H4+H21.250∙10140.033.471
587C2H5+HC2H63.0∙10130.00.0
589C2H5+OHC2H4+H2O4.0∙10130.00.0
591C2H5+HO2→CH3+CH2O+OH3.0∙10130.00.0

CH2CH2OHReactions
592CH2CH2OHC2H4+OH1.0∙10140.0140.0
594CH2CH2OH+HCH3CHO+H25.0∙10130.00.0

011

A.2.ReactionMechanism

C2H6Reactions
596C2H6+H
598C2H6+O
600C2H6+OH
602C2H6+HO2
604C2H6+O2
606C2H6+3CH2
608C2H6+CH3
610C2H6+CH3
612C2H6+CH
614C2H6+M(1)

noOxidatiC3

C3H6Reactions
HC61663618C3H6+O
620C3H6+O
622C3H6+O
624C3H6+OH
626C3H6+OH
628C3H6+H

C2H5
C2H5
C2H5
C2H5
C2H5
C2H5
C2H5
C2H5
C2H4
C2H5

C2H3
C2H4
C2H5
CH3C2H5
CH3C2H4

+H25.540∙1002
+OH1.0∙1009
+H2O9.154∙1006
+H2O21.102∙1005
+HO27.287∙1005
+CH32.20∙1013
+CH45.601∙1010
+CH48.432∙1014
+CH31.084∙1014
+H+M(1)8.850∙1020
LOW6.920∙1042
TROE0.500.0

HC+3OCH+2CHO+COCH+3OCH+2CHOCH+3HC+3

3.10∙1021
5.90∙1013
3.60∙1012
5.0∙1012
7.90∙1012
5.10∙1012
7.230∙1011

21.623.524.41.54.1572.070.5022.58205.682.536.30.039.4080.093.1160.0-1.10.0427.62-1.22448.55-6.430.00.0

-1.20.00.00.00.00.00.7

408.821.00.02.50.00.05.447

111