ebook img

The effect of turbulence on mass and heat transfer rates of small inertial particles PDF

0.48 MB·
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview The effect of turbulence on mass and heat transfer rates of small inertial particles

UnderconsiderationforpublicationinJ.FluidMech. 1 ff The e ect of turbulence on mass and heat transfer rates of small inertial particles 7 1 0 NilsErlandL.Haugen1,2,JonasKru¨ger1,DhrubadityaMitra3 andTereseLøvås1 2 n 1DepartmentofEnergyandProcessEngineering,NorwegianUniversityofScienceandTechnology, KolbjørnHejesvei1B,NO-7491Trondheim,Norway a J 2SINTEFEnergyResearch,N-7465Trondheim,Norway 3Nordita,KTHRoyalInstituteofTechnologyandStockholmUniversity,Roslagstullsbacken23, 7 SE-10691Stockholm,Sweden 1 (Received) ] n y The effect of turbulence on the mass and heat transfer between small heavy inertial particles d (HIP) and an embedding fluid is studied. Two effects are identified. The first effect is due to - therelativevelocitybetweenthefluidandtheparticles,andamodelfortherelativevelocityis u presented.Thesecondeffectisduetotheclusteringofparticles,wherethemasstransferrateis l f inhibited due to the rapiddepletion of the consumedspecies inside the dense particle clusters. . s This last effect is relevant for large Damko¨hler numbers and it may totally control the mass c i transfer rate for Damko¨hler numbers larger than unity. A model that describes how this effect s y shouldbeincorporatedintoexistingparticlesimulationtoolsispresented. h p Keywords:Reactingmultiphaseflow,Particle/fluidflow,Combustion,Turbulentreactingflows, [ Turbulencesimulations 1 v 7 6 1. Introduction 5 4 Both in nature and in industrial applications,one regularlyfinds small inertial particles em- 0 beddedinturbulentflows.Bysmallinertialparticles,wemeanparticlesthataresmallerthanthe . 1 smallestscalesoftheturbulenceandhavesignificantlyhighermaterialdensitythanthefluid.For 0 suchparticles,therewillbemomentumexchangebetweentheparticlesandtheturbulentfluid, 7 and,dependingontheconditions,theremayalsobeheatandmasstransfer.Thisisparticularly 1 soforchemicallyreactingparticles,buttherearealsoalargenumberofotherapplicationswhere : v heatandmasstransferbetweenparticlesandfluidareimportant.Here,themainfocuswillbeon i X reactingparticlesthatconsumeoneormoreofthespeciesintheembeddinggasthroughsurface reactions.Relevantexamplesare;chemicalreactionsonthe surfaceof a catalytic particle,fuel r a oxidationonthesurfaceofaoxygencarryingparticleinaChemicalLoopingCombustion(CLC) reactor,condensationofwatervaporonclouddropletsandcombustionorgasificationofchar. The presence of turbulence in a fluid will enhance the transportpropertiesof the flow. This means that the mean-field viscosity, diffusivity and conductivity may be drastically increased from their laminar values. This effect has been studied for many years, and a large numberof different models exist in the literature, such as the k-ǫ model (Jones&Launder (1972)) and differentversionsoftheReynoldsStressModels(e.g.Pope (2003)).Turbulencemayalsomodify gasphasecombustion,andeventhoughthisissomewhatmorecomplicated,asignificantnumber ofmodelshavebeendevelopedduringthelastdecades.SomeexamplesaretheEddyDissipation Model(Magnussen&Hjertager(1976)),theEddyDissipationConcept(Ertesvåg&Magnussen (2000))andvariationsofProbabilityDensityFunction(e.g.Dopazo (1994))models. Withtheaboveknowledgeinmind,itisinterestingtorealizethat,exceptfortherecentwork 2 NilsErlandL.Haugen1,2,JonasKru¨ger1,DhrubadityaMitra3andTereseLøvås1 ofKrugeretal. (2016),thereiscurrentlynomodeldescribingtheeffectofturbulenceontheheat andmasstransferofsmallinertialparticles.Whenareactingparticleisembeddedinaturbulent flow, the turbulencecan potentiallyinfluence the mass transfer, and hence the surface reaction ratesintwoways.Thefirstwayisthroughparticleclustering,whereparticlesformdenseclus- tersduetoturbulence,andwherethegasphasereactantswithintheclusterarequicklyconsumed whiletherearenoparticlesthatcanconsumethereactantsintheparticlevoidsoutsidetheclus- ters. The maineffectofthe clusteringisto decrease theoverallmasstransferrate.The second way turbulence influence the mass transfer rate is by increasing the mean velocity difference betweentheparticleandthegas.Thiseffectwillincreasethemasstransferrate. Thesametwoeffectsarealsoactivefortheheattransfer.Thesimilaritybetweenheatandmass transfercanbeseenbyconsideringtheexpressionsforthetransfercoefficientsofmass ShD κ= (1.1) d p andheat NuD κ = th, (1.2) th d p where d is the particle radius, Sh and Nu are the Sherwood and Nusselt numbers and D and p D are the mass and thermaldiffusivities. For single spherical particles in flows with low and th mediumparticleReynoldsnumbers,theSherwoodandNusseltnumberscanbeapproximatedby theempiricalexpressionsofRanz&Marshall (1952) Sh =2+0.69Re1/2Sc1/3 (1.3) RM p 1/3 Nu =2+0.69Re1/2Pr. RM p Awellknownexamplewherereactingparticlesareconsumedinaturbulentfluidisthecase ofpulverizedcoalcombustion,whereturbulenceinfluencestheprocessinseveralwaysthatare understood to varying degrees. The combustion of coal can be divided into four separate pro- cesses;1)drying,2)devolatilization,3)combustionofvolatilesand4)burnoutoftheremaining char.Processes1and2involvetheevaporationoffluidsandthermalcrackingofhydrocarbons, whileprocess3involveshomogeneousreactions.Inprocess4,gasphasespeciesdiffusetothe particlesurface andreactwith the solidcarbon.Thishappensvia adsorptionofe.g.an oxygen radicaltoacarbonsiteontheparticlesurfaceandasubsequentdesorptionofcarbonmonoxide intothegasphase.Thismakesprocess4dominatedbyheterogeneouschemicalreactions.Many published studies utilize RANS based simulation tools that describe simulations of pulverized coal conversionin the form of combustion or gasification with an Eulerian-Eulerianapproach (Gaoetal. (2004)andZhangetal. (2005))oraLagrangian-Eulerianapproach(Silaen&Wang (2010);Vascellarietal. (2014,2015);Klimaneketal. (2015);Chenetal. (2012,2000)).How- ever,noneofthesepaperstaketheeffectofturbulenceontheheterogeneouscharreactionsinto account. To the knowledgeof the authors, the only studies where accountis made for this ef- fectarethepapersofLuoetal. (2012);Brosh&Chakraborty (2014);Broshetal. (2015)and Haraetal. (2015)wheretheDirectNumericalSimulations(DNS)approachisused.InaDNS, all turbulence scales are explicitly resolved on the computational grid, such that the effect of turbulenceisimplicitlyaccountedfor.However,theDNSapproachisextremelycostlyandcan thereforeonlybeusedforsmallsimulationdomains.Forsimulationsoflargescaleapplications, theRANSorLESbasedsimulationtoolswillthereforebetheonlyapplicabletoolsforthefore- seeablefuture. Inthecurrentpaper,thesameframeworkaswasdevelopedbyKrugeretal. (2016)hasbeen usedandextended.Theaimofthepaperistoidentifytheeffectofturbulenceonthemassand Theeffectofturbulenceonmassandheattransferratesofsmallinertialparticles 3 heattransferofsolidparticles,andtodevelopmodelsthatdescribethiseffectforallDamko¨hler numbers. 2. Mathematicalmodelandimplementation In the current work, the so called point-particle direct numerical simulation (PP-DNS) ap- proach is used. Here, the turbulent fluid itself is solved with the direct numerical simulation (DNS) methodology,where all turbulent scales are resolved and no modelling is needed. The particles are howevernotresolved,but rathertreated as pointparticles where the fluid-particle momentum,massandheatinteractionsaremodelled.Thepointparticleapproachisasimplifica- tionthatreliesheavilyonthequalityofthemodels.Thealternativeapproach,whichistoresolve theparticlesandtheirboundarylayer,isextremelyCPUintensiveandcancurrentlynotbedone formorethanafewhundredparticles,evenonthelargestcomputers(Deen&Kuipers (2014)). Anumberofsimplificationsaremadeinthispaper.Thishasbeendoneinordertomakethe simulationslessCPU intensive,and,evenmoreimportantly,toisolate thedominatingphysical mechanisms. The particles are considered to be ever lasting, i.e. they are not consumed. The reactionontheparticlesurfaceisconvertingreactantAtoproductB; A→ B (2.1) isothermally,i.e.;thereisnoproductionorconsumptionofheat,suchthatonlythemasstransfer effectisconsidered.Asexplainedabove,theeffectontheheattransferratewillbesimilartothe effectonthemasstransferrate.AsreactantAisconvertedproductB,thethermodynamicaland transportpropertiesarenotchanged. 2.1. Fluidequations Theequationsdeterminingthemotionofthecarrierfluidisgivebythecontinuityequation ∂ρ +∇·(ρu)=0, (2.2) ∂t andtheNavier–Stokesequation Du ρ =−∇P+∇·(2µS)+ρf +F. (2.3) Dt Here,ρ,u,µ = ρνandνarethedensity,velocityanddynamicandkinematicviscositiesofthe carrierfluid,respectively.Thepressure Pandthedensityρarerelatedbytheisothermalsound speedc ,i.e., s P=c2ρ, (2.4) s whilethetrace-lessrateofstraintensorisgivenby 1 1 S= ∇u+(∇u)T − ∇·u. (2.5) 2 3 (cid:16) (cid:17) Kinetic energy is injected into the simulation box through the forcing function f, which is solenoidaland non-helicaland injectsenergyandmomentumperpendicularto a randomwave vectorwhosedirectionchangeseverytime-step(Haugenetal. 2012;Krugeretal. 2016).Sim- ilar kinds of forcing has also previously been used for particle laden flows by other groups (Becetal. 2007). The energy injection rate is maintained at a level such that the maximum Mach numberis always below 0.5. The domain is cubic with periodicboundariesin all direc- tions.Themomentumexchangeterm,F,ischosentoconservemomentumbetweenthefluidand thesolidparticles,i.e., 1 F=− mkak (2.6) V cell Xk 4 NilsErlandL.Haugen1,2,JonasKru¨ger1,DhrubadityaMitra3andTereseLøvås1 whenV isthevolumeofthegridcellofinterestandmk and ak arethemassandacceleration cell (duetofluiddrag)ofthek’thparticlewithinthegridcell. Theequationofmotionofthereactanthasthewell-knownadvection-reaction-diffusionform: ∂X +∇·(Xu)= DM¯ ∇·(∇X)+R˜, (2.7) ∂t c whereX,M¯ andDarethemolefraction,themeanmolarmassandthediffusivityofthereactant, c respectively.ThelastterminEq.(2.7),R˜,isthesinktermduetothegas-solidreactionsonthe surfaceofthesolidparticles. 2.2. Particleequations The N particlesthatareembeddedintheflow aretreatedaspointparticles,whichmeansthat p theyareassumedtobesignificantlysmallerthantheviscousscaleofthefluidandthediffusive scaleofthereactant.Themotionofthek’thparticleisdescribedbytheequationsforposition dXk =Vk (2.8) dt andvelocity dVk = ak (2.9) dt whentheparticleaccelerationduetofluiddragisgivenbyak = 1 u(Xk)−Vk .Notethatgravity τ isneglectedinthiswork.Theparticleresponsetimeisgivenby(Shchiller&Niaumann (1933)) τ τ= St (2.10) 1+ f c whenτ =Sd2/18νistheStokestime, f =0.15Re0.687isaReynoldsnumbercorrectiontermto St p c p theclassicalStokestime,S =ρ /ρisthedensityratio,ρ isthematerialdensityoftheparticles, p p |u(Xk)−Vk|d u d p rel p Re = = (2.11) p ν ν istheparticleReynoldsnumberandd istheparticlediameter. p 2.3. Surfacereactions Letus nowmodelthe reactiveterm. We assume thatthereactionsarelimited to the surfaceof theparticlesandthatthereactionsarediffusioncontrolled,i.e.thatallreactantthatreachesthe particlesurfaceisconsumedimmediately†.Thereactivetermcanthenbewrittenas 1 R˜ = AkκXK (2.12) V p ∞ cell Xk whereA =4πr2 istheexternalsurfaceareaoftheparticle,themasstransfercoefficientisgiven p p by DSh κ= (2.13) d p andShistheSherwoodnumber. Tocouplethereactiveparticlewiththecontinuumequationsweusethefollowingprescription; forthek-thparticle,whichisatpositionXk,weset Xk = X(Xk), (2.14) ∞ † Itispossibletorelaxtheassumptionofdiffusioncontrolledreactionsbyalsoaccountingforchemical kineticsattheparticlesurface,seeKrugeretal. (2016). Theeffectofturbulenceonmassandheattransferratesofsmallinertialparticles 5 i.e.;thefarfieldreactantmolefractionissetequaltothereactantmolefractionofthefluidcell where the particle is. In the currentwork, the particle Sherwood number is determined by the expressionofRanz&Marshall (1952)(seeEq.(1.3)intheintroduction),whichisincontrastto theworkofKrugeretal. (2016)wheretheSherwoodnumberwassettoaconstantvalueof2, whichcorrespondstotheSherwoodnumberinaquiescentflow.TheparticleReynoldsnumber isgivenbyEq.(2.11)andtheSchmidtnumber,Sc = ν/D,istheratioofthefluidviscosityand themassdiffusivity. 2.4. Thereactantconsumptionrate Itisusefultodefineareactantconsumptionrateas R˜ α=− =n A κ, (2.15) p p X∞! when O represents the volume average of flow property O and n is the particle number den- p sity. If everything is assumed to be homogeneously distributed over the volume, the reactant consumptionrateisgivenby ShD α =n A κ=n A (2.16) hom p p p p d p foragivenparticlesizeandnumberdensity. In many RANS based simulation tools, where the local fluid velocity is not resolved, it is commontoneglecttherelativevelocitydifferencebetweentheturbulenteddiesandtheparticles. ThisimpliesthatSh=2.Sincetheeffectofparticleclusteringisalsoneglectedinsuchmodels, themodelledreactantconsumptionratebecomes; 2D α = lim α=n A . (2.17) Sh,Da p p Sh→2,Da→0 dp Inthefollowing,α willbeusedfornormalization. Sh,Da It is useful to define the Damko¨hler number, which is the ratio of the typical turbulent and chemicaltimescales,as τ Da= L (2.18) τ c whereτ = L/u is the integraltime scale of theturbulence, L is the turbulentforcingscale, L rms u istheroot-mean-squareturbulentvelocityandthechemicaltimescaleis rms τ =1/α . (2.19) c Sh,Da Particlesinaturbulentflowfieldwilltendtoformclusterswithhigherparticlenumberdensity thantheaverage(Squires&Eaton 1991;Eaton&Fessler 1994;Toschi&Bodenschatz 2009; Woodetal. 2005).Ifthechemicaltimescaleisshortcomparedtothelife-timeoftheclusters, the reactantconcentrationwithin the clusterswill be much lower than outside the clusters. On the otherhand,if the particle numberdensityis low, the particleclusterswill nothaveenough time to consume a significant fraction of the reactant during the life-time of the cluster, and hence,thereactantconcentrationwillberoughlythesameinsideasitisoutsidetheclusters.By assuming that the life-time of the clusters is of the orderof the turbulenttime scale, it is clear thatthereactantconcentrationofparticleflowswithlowDamko¨hlernumberwillbehaveasifthe particles were homogeneouslydistributed overthe volume, i.e.; for small Damko¨hlernumbers thereisnoeffectofparticleclusteringonthereactantconsumption. FromEqs.(2.16)-(2.19)itcanbededucedthatforthehomogeneouscase,andthenalsofor allcaseswithlowDamko¨hlernumbers,thereactantconsumptionratewillscalelinearlywiththe 6 NilsErlandL.Haugen1,2,JonasKru¨ger1,DhrubadityaMitra3andTereseLøvås1 Damko¨hlernumberforagiventurbulentflowfield,suchthat DaSh α = . (2.20) hom τ 2 L When relaxing the restriction to small Damko¨hler numbers, the effect of particle clustering eventuallycomesintoplay.Krugeretal. (2016)haveshownthatthereactantconsumptionrate isgivenby α α α= c hom (2.21) α +α c hom whenα isaclusterdependentdecayrate.(NotethatsinceKrugeretal.assumedtheSherwood c number to be 2, their α equals our α = Da/τ .) From this expression, the following hom Sh,Da L normalizedreactantconsumptionrateisfound α α τ Sh α˜ = Sh = c L . (2.22) Sh α α τ +DaSh/2 2 Sh,Da c L whenSh isgivenbyEq.(1.3) andthecorrespondingrelativevelocitybetweenthe particleand thefluidisdeterminedbyamodel(whichwillbeobtainedinthenextsubsection).Fordiffusion controlledreactions,themodifiedreactiondecayrate,asgivenbyEq.(2.22),isameasureofthe relativemodificationtothemasstransferrateduetotheeffectofturbulence.Thismeansthata modifiedSherwoodnumbercannowbedefinedthataccountsfortheeffectofturbulence; Sh =2α˜. (2.23) mod InthelimitofsmallDamko¨hlernumbers,thisexpressionreducestoSh =Sh,asexpected. mod ByemployingthemodifiedSherwoodnumbergivenbyEq.(2.23),onecannowusethecom- monexpressionforthereactantconsumptionrate,asgivenbyEq.(2.16),tofindtherealreactant consumptionrate.In mostcases, however,oneneedstheparticleconversionraten˙ forindi- reac vidualparticles,which is closely connectedto the reactantdecay rate. For diffusioncontrolled masstransfer,theparticleconversionrateisgivenbyn˙ =−κX C ,whereC isthemolarcon- reac ∞ g g centrationofthegasphaseandthemasstransfercoefficientisnowfoundbyusingthemodified Sherwoodnumber(asgivenbyEq.(2.23))in;Eq.(1.1) DSh κ= mod. (2.24) d p In many applications, the mass transfer rate is not purely diffusion controlled.This can be ac- countedforbyincludingtheeffectofreactionkineticsattheparticlesurface.Thecorresponding particleconversionratecanthenbeexpressedas(Krugeretal. 2016) λκ n˙ =− X C , (2.25) reac λ+κ ∞ g whereλisthesurfacespecificmolarconversionrate.Sincethereactionkineticsisonlydepen- dent on the conditionsat the particle surface, the surface specific molar conversion rate is not affected by the turbulence.This is, as we have alreadyseen, notthe case for the mass transfer coefficient,whichisnowgivenbyEq.(2.24).Inthisway,allthecommonmachineryforcalcu- latingparticlereactionratescanstillbeusedsincetheeffectsoftheturbulenceareincorporated intothemodifiedSherwoodnumber. 3. Results In all of the following,statistically stationaryhomogeneousand isotropic turbulenceis con- sidered.TheReynoldsnumberisvariedbychangingthedomainsizewhilemaintainingconstant Theeffectofturbulenceonmassandheattransferratesofsmallinertialparticles 7 Table1.Summaryofthesimulations.ThefluiddensityisunitywhiletheSchmidtnumberis0.2andthe viscosity is 2×10−4 m2/s for all the simulations. For every simulation listed here, a range of identical simulationswithdifferentDamko¨hlernumbershavebeenperformed. Label L(m) N d ρ Re Sh St τ α α τ St/Sh grid p p L c c L 1A π/2 643 3.4×10−3 50 80 2.5 1.0 1.6 0.9 0.63 2A 2π 1283 19×10−3 50 400 2.8 1.0 5 0.23 0.43 3A 8π 2563 11×10−3 500 2200 2.8 1.0 15 0.07 0.41 2AB 2π 1283 19×10−3 25 400 2.7 0.5 5 0.26 0.25 3AB 8π 2563 11×10−3 250 2200 2.6 0.5 15 0.09 0.26 2B 2π 1283 11×10−3 50 400 2.5 0.3 5 0.21 0.13 3B 8π 2563 11×10−3 150 2200 2.6 0.3 15 0.09 0.18 2C 2π 1283 19×10−3 5 400 2.4 0.1 5 0.55 0.12 3C 8π 2563 11×10−3 50 2200 2.4 0.1 15 0.20 0.13 2D 2π 1283 19×10−3 1.5 400 2.3 0.03 5 1.20 0.08 3D 8π 2563 11×10−3 16 2200 2.3 0.03 15 0.45 0.10 2E 2π 1283 19×10−3 0.5 400 2.2 0.001 5 4.10 0.10 viscosityandturbulentintensity.TheDamko¨hlernumberisvariedbychangingthenumberden- sity of particles, while keeping everythingelse the same. All relevantsimulations are listed in table1. 3.1. Themeanrelativeparticlevelocity In order to predicta representative value of the particle Sherwoodnumber from Eq. (1.3), the particleReynoldsnumberRe isrequired.FromEq.(2.11)itisclearthatthisalsorequiresthe p relativeparticlevelocityu ,whichwillbefoundinthissubsection. rel GivenaparticlewitharesponsetimethatequalstheStokestime; Sd2 τ = p, (3.1) p 18ν such that τ < τ < τ , where τ is the Kolmogorov time scale and τ is the integral time k p L k L scale.Withrespecttotheparticle-turbulenceinteractions,theturbulentpowerspectrummaybe dividedintothreedistinctregimes,basedontherelationbetweentheparticleresponsetimeand theturbulenteddyturnovertimeτ .Thefirstregimeisdefinedasthesectionoftheturbulent eddy power spectrum where the turbulent eddies have turnover times that are much larger than the response time of the particles, i.e. where τ ≫ τ . All the turbulent eddies in this regime eddy p will see the particles as passive tracers, which follow the fluid perfectly. I.e., there will be no relativevelocitybetweentheparticlesandtheeddies.Thethirdregimeisdefinedasthepartof thepowerspectrumwheretheturbulenteddieshavemuchshortertimescalesthantheparticles, i.e.whereτ ≪τ .Theeddiesinregimethreewillseetheparticlesasheavybulletsthatmove eddy p instraightlines,withoutbeingaffectedbythemotionoftheeddies.Hence,thevelocityofthese eddies will contribute to the relative particle-fluid velocity. The second regime is now defined as the relatively thin band in-between regimes one and three, where τ ≈ τ . These are the eddy p eddiesthatare responsibleforparticleclustering,sincetheyareabletoacceleratetheparticles 8 NilsErlandL.Haugen1,2,JonasKru¨ger1,DhrubadityaMitra3andTereseLøvås1 Figure1. Theparameterβ,relatingtherelativeparticlevelocitytothesubscalevelocityasdefinedin Eq.(3.9),isshownasafunctionofStokesnumber. Figure2.Leftpanel:kineticenergyspectrumfordifferentReynoldsnumbers.Rightpanel:relativeparticle velocityasafunctionofStokesnumber. to a levelwhere they are thrownout of the eddy due to their inertia. In the following,we will refertoatypicaleddyinregimetwoasaresonanteddy,andwedefinethescaleofthiseddyas ℓ.Theresonanteddiesareidentifiedbytheirtimescale,τ ,whichisoftheorderoftheparticle ℓ responsetime,τ .Forconvenience,wesetthetwotimescalesequal,suchthat p τ =τ . (3.2) ℓ p Basedonthedefinitionsabove,itisclearthatthelargestturbulenteddiesthatyieldarelative velocitybetweenthe fluidandtheparticles,aretheresonanteddies.By assumingKolmogorov scaling, the velocity of the resonant eddies is known to be u = u (ℓ/L)1/3, which can be ℓ rms combinedwiththeaboveexpressionforthetimescalestoyield k =k St−3/2 (3.3) ℓ L whentheparticleStokesnumberisdefinedas τ p St= (3.4) τ L andk =2π/landk =2π/Larethewave-numbersoftheresonanteddiesandtheintegralscale, ℓ L respectively.InobtainingEq.(3.3),ithasalsobeenusedthattheturnovertimeoftheresonant eddiesisτ =l/u ,whilethatoftheintegralscaleeddiesisτ = L/u . ℓ ℓ L rms Since all scales smaller than ℓ will induce a relative velocity between the particles and the fluid,itisreasonabletoassumethattherelativevelocitybetweenthefluidandtheparticleswill beacertainfractionβoftheintegratedturbulentvelocityu˜ ofallscalessmallerthanℓ,suchthat ℓ u =βu˜ (3.5) rel ℓ Theeffectofturbulenceonmassandheattransferratesofsmallinertialparticles 9 whenu˜ isdefinedas ℓ 1 kη u˜2 = E(k)dk (3.6) 2 ℓ Zkℓ and k = 2π/η is the wave-number of the Kolmogorov scale (η = (ν3/ǫ)1/4), where ǫ is the η dissipationrateofturbulentkineticenergy.IntegrationofEq.(3.6)yields k−2/3−k−2/3 u˜ℓ =urmsvtkℓ−2/3−kη−2/3 (3.7) L η forE(k)=cǫ2/3k−5/3whenithasbeenusedthatthetotalturbulentkineticenergyisgivenby 1 kη u2 = E(k)dk, (3.8) 2 rms Z k1 wherek isthewavenumberofthelargestscaleinthesimulation.CombiningEqs.(3.3)and(3.7) 1 withEq.(3.5)finallyyields Stk−2/3−k−2/3 urel =βurmsvt k−L2/3−k−η2/3 . (3.9) L η Theunknownconstantinthisequation,β,canbedeterminednumericallyfromEq.(3.5),i.e.β= u /u˜ .Here,u isfounddirectlyfromDNSsimulations,whileu˜ iscalculatedfromEq.(3.7). rel ℓ rel ℓ Itisseenfromfigure1thatβiscloseto0.41formostStokesandReynoldsnumbers.Themain exceptionisforlow ReynoldsandStokesnumbers,whereβissignificantlylarger.Thiscanbe understoodbyinspectingtheleftpaneloffigure2,whereitisseenthatforRe=180andSt<0.1, we are alreadyfarinto thedissipativesubrange,whereourmodelis notexpectedtobe correct sinceitreliesonaKolmogorovscaling. It is surprising to see that Eq. (3.9) reproduces the relative particle velocity for such low Stokes numbers, even for the smaller Reynolds numbers. This may be explained by reconsid- ering Eq. (3.2), where we assumed that the resonanteddiescorrespondto the eddies thathave exactly the same turnover time as the response time of the particles. This is just an order of magnitudeestimate,andamorecorrectexpressionwouldprobablybe τ =γτ , (3.10) ℓ p where γ is of the orderof unity.More work should, however,be devotedto understandingthe couplingbetweentheparticlesandtheturbulenteddies.Inparticular,amoreexactdefinitionof theresonanteddiesisneeded.We neverthelessbelievethatβisauniversalpropertyoftheHIP approximationandtheNavier-StokesequationsthatwillhaveaconstantvalueforallReandSt aslongastheresonanteddiesarewithintheinertialrange. In the right panel of figure 2, the average relative particle velocity, as found from the DNS simulations(symbols), is comparedwith the predicted valuesfromEq. (3.9) (solid lines). It is seenthatthefitisrathergoodformostReynoldsandStokesnumbers.Thissupportstheuseof Eq.(3.9)forpredictingtherelativeparticlevelocity. 3.2. Theclustersize Thetypicalsizeoftheclustersℓisassumedtobethesizeoftheresonanteddies.FromEq.(3.3) thisyieldsaclustersizeof l= LSt3/2. (3.11) Itcanbeseenfromfigure3thattheparticlenumberdensitydistributiondoesindeedshowmore smallscalevariationforthesmallerStokesnumbers.Thishasbeenquantifiedinfigure4where 10 NilsErlandL.Haugen1,2,JonasKru¨ger1,DhrubadityaMitra3andTereseLøvås1 Figure3.ParticlenumberdensityforSt=1(upperleft),St=0.3(upperright),St=0.1(lowerleft)and St=0.03(lowerright)(runs3A,3B,3Cand3Dintable1). the power spectrum of the particle number density is shown. Here we see that the spectrum peaks at large scales for St = 1 while the peak is located at much smaller scales for smaller Stokes numbers. The peak in the spectrum does not, however, follow Eq. (3.11) as accurately asexpected.Thereasonforthisismostlikelythatpowerspectraarenottherightdiagnosticsto studythesizeofparticleclusters,butitmayalsobepartlybecauseof:1)poorstatisticsduetotoo fewparticles(thesmallerclustersarenotfilled withparticles),2)theconstantinthedefinition oftheresonanteddiesnotbeingunity(seee.g.Eq.(3.10)),or3)finiteReynoldsnumbereffects. The power spectrum P can be integrated to yield a measure of the strength in the particle numberdensityfluctuations,givenbytheroot-mean-square(rms)particlenumberdensity; n = Pdk. (3.12) rms Z ItisfoundthatthermsparticlenumberdensityisdecreasingwithStokesnumber.Morespecif- ically,n is1.6,1.5,1.2and0.8forStokesnumbersof1,0.3,0.1and0.03,respectively.This rms meansthatthehighdensityregimeshavehigherparticlenumberdensitiesforlargerStokesnum- bers. 3.3. Reactantconsumptionrate Thenormalizedreactantconsumptionrateisshowninfigure5.Thesymbolscorrespondtothe resultsfromtheDNSsimulations,whilethesolidlinesaregivenbyEq.(2.22).Here,theStokes numberisfoundbyusingthemodelfortherelativevelocity,asgivenbyEq.(3.9),intheexpres- sionfortheSherwoodnumber(Eq.(1.3)).Thevalueoftheclusterdecayrate,α ,istheonlyfree c parameteranditischosenbyabestfitapproach.Thevaluesofα arefoundintable1. c The valueof α˜ for small Damko¨hlernumbersequalsthe Sherwoodnumberdividedby two,

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.