• Aucun résultat trouvé

Effects of reducing the reactor diameter on thedense gas–solid fluidization of very heavyparticles: 3D numerical simulations

N/A
N/A
Protected

Academic year: 2021

Partager "Effects of reducing the reactor diameter on thedense gas–solid fluidization of very heavyparticles: 3D numerical simulations"

Copied!
11
0
0

Texte intégral

(1)

HAL Id: hal-01413039

https://hal.archives-ouvertes.fr/hal-01413039

Submitted on 9 Dec 2016

HAL is a multi-disciplinary open access

archive for the deposit and dissemination of

sci-entific research documents, whether they are

pub-lished or not. The documents may come from

teaching and research institutions in France or

abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est

destinée au dépôt et à la diffusion de documents

scientifiques de niveau recherche, publiés ou non,

émanant des établissements d’enseignement et de

recherche français ou étrangers, des laboratoires

publics ou privés.

Effects of reducing the reactor diameter on thedense

gas–solid fluidization of very heavyparticles: 3D

numerical simulations

Renaud Ansart, Florence Vanni, Brigitte Caussat, Carine Ablitzer, Méryl

Brothier

To cite this version:

Renaud Ansart, Florence Vanni, Brigitte Caussat, Carine Ablitzer, Méryl Brothier. Effects of

re-ducing the reactor diameter on thedense gas–solid fluidization of very heavyparticles: 3D numerical

simulations. Chemical Engineering Research and Design, Elsevier, 2016, vol. 117, pp. 575-583.

�10.1016/j.cherd.2016.11.008�. �hal-01413039�

(2)

O

pen

A

rchive

T

OULOUSE

A

rchive

O

uverte (

OATAO

)

OATAO is an open access repository that collects the work of Toulouse researchers and

makes it freely available over the web where possible.

This is an author-deposited version published in :

http://oatao.univ-toulouse.fr/

Eprints ID : 16724

To link to this article : DOI:10.1016/j.cherd.2016.11.008

URL :

http://dx.doi.org/10.1016/j.cherd.2016.11.008

To cite this version : Ansart, Renaud and Vanni, Florence and

Caussat, Brigitte and Ablitzer, Carine and Brothier, Méryl Effects of

reducing the reactor diameter on thedense gas–solid fluidization of

very heavyparticles: 3D numerical simulations. (2016) Chemical

Engineering Research and Design, vol. 117. pp. 575-583. ISSN

0263-8762

Any correspondence concerning this service should be sent to the repository

administrator:

staff-oatao@listes-diff.inp-toulouse.fr

(3)

Effects

of

reducing

the

reactor

diameter

on

the

dense

gas–solid

fluidization

of

very

heavy

particles:

3D

numerical

simulations

Renaud

Ansart

a,∗

,

Florence

Vanni

a,b

,

Brigitte

Caussat

a

,

Carine

Ablitzer

b

,

Méryl

Brothier

c

aLaboratoiredeGénieChimique,UniversitédeToulouse,CNRS,INPT,UPS,Toulouse,France

bCEACadaracheDEN,DEC/SFER,LaboratoiredesCombustiblesUranium,F-13108Saint-PaulLezDurance,France cCEAMarcouleDEN,DTEC/SECAServiced’EtudedesCombustiblesetmatériauxàbased’Actinides,F-30207 BagnolssurCèze,France

Keywords:

Fluidization CFDsimulation

Verydenseparticlessuspension Euler–Eulermodeling

Microfluidizedbed

Inthisstudy,3DnumericalsimulationsusinganEuleriann-fluidapproachofagas–solid fluidizedbedcomposedofverydenseparticlesoftungsten(19,300kgm−3)werecarried outtoexaminethebehaviorofthissuspension,especiallytheeffectsofthereductionof thefluidizationcolumndiameteronthefluidizationquality.Tungstenwasselectedasa surrogatematerialofU(Mo)(Uraniummolybdene)whichisofinterestfornewnuclearfuels withlimitedenrichment.Comparisonsbetweenexperimentsandcomputationsfortheaxial pressureprofileofa5cmdiametercolumndemonstratethecapabilityofthemathematical modelsoftheNEPTUNECFDcodetosimulatethefluidizationofthispowderlocated out-sidetheclassificationofGeldart.Thenumericalresultsshowthatthemobilityintothebed oftheseverydenseparticlesisverylow.Thereductionofthefluidizationcolumn diam-eterfrom5cmto2cmdoesnothavesignificanteffectonthelocalsolidcirculationbut stronglydecreasestheaxialandradialmixingoftheparticlesduetowall-particlesfriction effects.Theseresultsconfirmandallowtobetterunderstandthewalleffectsexperimentally evidenced.

1.

Introduction

Gas–solidfluidized beds (FB) are widely used in industrial applicationssuchasdrying,coalcombustion(Basu,1999)and gasification (Hofbauer et al., 2002), oilrefining and nuclear application(KuniiandLevenspiel,1991).Thisisduetothefact thatFBreactorshaveexcellentheatandmasstransfer capa-bilities,highthroughputrates,andcanoperatecontinuously, thusreducingoperatingcosts(YdstieandDu,2011;Liuand Xiao,2014).

Inthe nuclearfield,newfuels,withlimited enrichment in235U,areunderdevelopmentforresearchreactors.U(Mo)

Correspondingauthor.

E-mailaddress:renaud.ansart@ensiacet.fr(R.Ansart).

(Uraniummolybdene)fuelpowdersdispersedinanaluminum matrix are among the most promising materials. But the coating ofU(Mo)byabarriermaterialisnecessarytolimit interfacialinteractionsbetweenthefuelanditsmatrixunder irradiation(Mazaudieretal.,2008).Siliconseemstobeagood candidate (Zweifel et al., 2013). Amongother technics, the FluidizedBedChemicalVaporDepositionprocess(FB-CVD)is understudytodeposituniformsiliconlayersonU(Mo) parti-cles(Vannietal.,2015a).FB-CVDisanefficienttechnologyto uniformlycoatpowdersbyagreatvarietyofmaterials(Vahlas etal.,2006).Thereisaninterestinbeingabletotreatweights ofU(Mo)aslowaspossiblebytheFB-CVDprocess.Itisto

(4)

Nomenclature

Romansymbols

Ar Archimedenumber(−)

dp particlediameter(m)

ec particle–particlenormalrestitutioncoefficient

(–)

g gravitationalconstant(ms−2)

mp particlemass(kgm−3)

P meangaspressure(Nm−2) q2

p randomkineticenergy(m2s−2) Uk,i meanvelocityofphasek(ms−1) Umf minimumfluidizationvelocity(ms−1)

uk,i fluctuatingvelocityofphasek(ms−1)

Vf superficialgasvelocity(ms−1) Greeksymbols

˛k volumefractionofphasek(–)

g gasviscosity(kgm−1s−1) k densityofphasek(kgm−3) p granulartemperature(kgm−2s−2) Subscripts g gas p particle

benotedthatforpreliminaryexperimentsatungstenpowder (19,300kgm−3,75␮mofmediandiameter)whoseproperties areveryclosetothoseofU(Mo)(17,500kgm−3,50␮mmedian diameter)hasbeenusedasasurrogatepowder.Both pow-dersareofveryhighdensity.Themaximumdimensionless densityis10,000intheGeldart’sclassification,butthe dimen-sionlessdensityofthetungstenparticlesclearlyexceedsthe upperlimit(Fig.1).Firstexperimentshavedemonstratedthat theFB-CVDtechnologycanuniformlycoat1,500goftungsten powderwithsiliconinareactorof3.8cmindiameter.Other experimentshaveshownthatthisweightcannotbedecreased inthereactorof3.8cmbecausethetargetdeposition temper-ature(650◦C)cannotbereachedsinceheatexchangebetween thereactorwallsand thebed ofparticles becomestoolow (Vannietal.,2015a).

Anexperimentalstudyhasbeenconductedinglassand steelcolumnstospecificallyanalyzetheimpactofdecreasing

Fig.1–Representationofthetungstenparticleswitha diameterof75␮minGeldart’sclassificationmodifiedby

Yang(2007).

thereactordiameterfrom5to2cmonthefluidization hydro-dynamicsofthetungstenpowder(Vannietal.,2015b).Itwas foundthatuniformsilicondepositioncouldonlybeachieved ifthefluidizedbedhydrodynamicswasofgoodquality,i.e.if itinvolveshighthermalandmasstransferratesbetweenthe gasandthesolidphases.Thisexperimentalstudyhasshown thatwalleffectsdecreasingthequalityofbedhydrodynamics appearinthereactorof2cmfor100gand 180gofpowder. Thiswasevidencedbyanincreaseofthehystereticbehavior ofthepressuredropcurves,anincreaseoftheminimum flu-idizationvelocityandadecreaseofthebedvoidage.Thiswas especiallyevidentinaglasscolumnwheretheelectrostatic effectscannotbeneglected(Vannietal.,2015b).

Theaimofthepresentstudyistocomplementthe exper-imentalworkbyusing3Dnumericalsimulationstoanalyze more precisely the influence of a decrease of the reactor diameteronsomekeylocalparametersofthefluidizedbed hydrodynamics.Theresultsareexpectedtobehelpfulinthe FB-CVDprocessdesignandinthechoiceofitsoptimalgas flowparameters.

Very few studies are available in the literature about the numerical simulation of fluidized beds involving such denseparticles.For UO2 particles of2mmindiameter,Liu

et al. (2015a) studied the impact of particle density up to 10,800kgm−3 on the fluidization behavior of particles in spoutingbedusing2DCFD-DEMcoupling.Theparticlecycle time,spoutbehavior,dominantspoutfrequencyandgas–solid contact efficiency was discussed and a flow pattern map under different densities and different gas velocities was obtained.Pannalaet al.(2007)alsostudiedconicalspouted bedwithheavyparticles(upto6,050kgm−3zirconiaparticles fornuclearfuelcoaters)by2-DEulerian–Euleriannumerical simulations.

Miniaturizationofchemicalreactors,whichisoneofthe most popular research areas inchemical engineering, has led to the concept of micro-fluidized bed reactors (Wang et al., 2011). Wang et al. (2011) haveshown byCFD simu-lations that for Geldart A particles, the onset of turbulent fluidizationisadvancedsignificantlyinmicrofluidizedbeds.

ZivkovicandBiggs(2015)pointedouttheimportanceofwall surfaceforcesrelativetovolumetricforces,suchasgravity, onmicro-scalefluidizedbeds(thecross-sectionaldimensions ofthemicrochannelswere 400×175␮m2). Liuetal.(2015b)

performed several 2D Eulerian–Eulerian numerical simula-tionsforagas–solidmicro-fluidizedbed(channelwidth3mm) for Geldart A particles (53␮m and 1,400kgm−3) and com-pared the predicted minimum bubbling velocity and bed voidagetoexperimentalmeasurements.Theyusedveryfine meshes(ofapproximately oneparticlediameter).Using the Gidaspow drag model, their simulations have shown that thepredictedminimumbubblingvelocitiesweresignificantly lowerthan theirexperimentaldata(Liuetal.,2015b).Wang and Fan (2011) have established a flow regime map from experimental results obtained in column diameters start-ing from 0.7mm to 5mm in size for FCC particles. Their results revealed an increase in the minimum fluidization andbubblingvelocitiescomparedtothoseinlarge-scale flu-idized beds. To the best of our knowledge, no numerical work has been reported concerning reduced diameter flu-idizedbedsofheavyparticles,whicharestudiedinthepresent article.

Theexperimentalfluidizationset-upwillbefirstpresented beforedetailingthenumericalmodelandpresentingand dis-cussingtheresultsofthesimulation.

(5)

Table1–Particlepropertiesandparticlesizedistribution characteristicsofthetungstenparticlesobtainedwitha MalvernMasterSizerSirocco2000indrymode.

Meandiameter Value(␮m)

d3/2 70 d10 50 d50 75 d90 105 Density Value(kgm−3) Particle 19,300 Bulk 9,600 Umf(Thonglimpetal.,1984) 3.2cms−1

Fig.2–Schematicoftheexperimentalsetup.

2.

Experimental

set

up

Theexperimentalsetupiscomposedofthreesteelcolumns of1mheightand5,3and2cmofinnerdiameter(Vannietal., 2015b).AllareequippedwithasimilarInoxporousplate dis-tributor(PoralTM40).Thetungstenpowderused(CERAC,Inc.

T-1220)issuppliedbyNeyco.Itsgraindensity isclaimedto be19,300kgm−3.TheParticleSizeDistributioncharacteristics ofthepowderaresummarizedinTable1withitsbulk den-sity.Theexperimentswereconductedatambienttemperature usingargonasthefluidizinggas.Itsflowratewascontrolled byamassflowcontroller(AeraFC-7710C0,0-20slm).A dif-ferentialfastresponsepressuresensor(GEDruckLPX5380), withtapsunder the distributor andon top ofthecolumn, wasusedtomeasurethetotalpressuredropacrossthebed anddeterminetheminimumfluidizationvelocity.Threeother pressuresensorsweresetonthe5cmcolumnat0cm,3cm and4.5cmabovethedistributortomeasuretheaxialpressure profile(Fig.2).Thesamplingrateforpressuremeasurementis 2Hzandthetime-averagesarerealizedduring2min.

3.

Numerical

simulation

description:

Euler

n-fluid

approach

Three-dimensional numerical simulations are carried out using the code NEPTUNECFD. This Eulerian n-fluid

unstructuredparallelizedmultiphaseflowsoftwarehasbeen developedintheframeworkoftheNEPTUNEproject finan-ciallysupportedbyCEA(Commissariatàl’EnergieAtomique et aux énergies alternatives), EDF (Electricite de France), IRSN(InstitutdeRadioprotectionetdeSuretéNucléaire),and AREVA-NP (Méchitouaet al., 2003). Themodeling approach forpoly-dispersedfluid-particleflowsisimplementedbythe InstitutdeMécaniquedesFluidesdeToulouse(IMFT)inthe NEPTUNE CFDV1.08version.Thenumericalsolverhasbeen developedforHighPerformanceComputing(Neauetal.,2010, 2013).

3.1. Mathematicalmodels

The Eulerian n-fluid approach used is a hybrid method (MoriokaandNakajima,1987)inwhichthetransport equa-tions are derived by phase ensemble averaging for the continuousphaseandbyuseofthekinetictheoryofgranular flowssupplementedbyfluideffectsforthedispersedphase. Themomentumtransferbetweengasandparticlephasesis modeled using the draglaw ofWenand Yu (Wenand Yu, 1965),limitedbytheErgun(Ergun,1952)equationfordense flows (Gobinet al., 2003;Fede et al., 2016). The collisional particle stresstensor isderivedinthe frameofthe kinetic theoryofgranularmedia(Boëlleetal.,1995).Inthepresent studythegasflowequationsaretreatedaslaminarbecause thegasReynoldsstresstensorinthemomentumequationis negligiblecomparedtothedragterm.Forthesolidphase,a transportequationforthe particlefluctuantkineticenergy, q2

p,issolved. Accordingtothe largeinertiaoftheparticles

and to thelow inelasticityofparticles (ec=0.9), the spatial

correlationbetweenneighboringparticlefluctuantvelocities willremainnegligible(Fox,2014;Fevrieretal.,2005;Simonin etal.,2002).Theparticlefluctuatingmotionmaybeassumed spatiallyuncorrelatedobeyingthesocalledmolecularchaos assumptionsothattheproductoftheparticlemassand par-ticle fluctuantkinetic energympq2p canbe identifiedto the

granulartemperaturepdefinedinspatialgranularflow

the-ories(LunandSavage,1986).

The gas–particle turbulent correlation is negligible.The effectsoftheparticle–particlecontactforceintheverydense zoneoftheflowaretakenintoaccountintheparticlestress tensor by the additionalfrictionalstress tensor (Srivastava and Sundaresan,2003).AlltheequationsoftheEuler–Euler approacharedetailedinFotovatetal.(2015).

3.2. Numericalparameters

3.2.1. Geometry

Thereferencefluidizedbedisacolumnof5cmindiameter and40cminheightwithaconicalportionatitstopend.The influenceofthe columndiameterreductionisanalyzedby comparisonwithcolumnsof3and2cmindiameterandof similarheight.

3.2.2. Mesh

The3Dreferencemesh(Fig.3)iscomposedof277,635 hexa-hedra,basedonO-gridtechniquewith50cellsonthecolumn diameterandapproximatelyx=y≈1mmandz=1.9mm.

Itisnoteworthythatafinermesh,refinedof1.5ineachspace direction,hasbeentestedbutnomeshrefinementeffectwas observedonthebedheightandonthetime-averagedradial profilesofsolidvolumefraction,solidvelocityandsolidtime variancevelocity.

(6)

Fig.3–3Dmeshforthenumericalsimulationwith277,635 cells.

Table2–Phases’propertiesusedforthesimulation.

Parameters Value

Gasdensity(Argon) 1.6kgm−3 Gasviscosity 2.23×10−5Pas Particlediameter 70␮m Particledensity 19,300kgm−3

The meshes for the 3 and 2cm column diameter have exactlythesamenumberofcellsandwereobtainedby apply-inghomothetictransformation.

3.2.3. Phaseproperties

Theflowiscomposedoftwophases:gas(argon)andtungsten particles.Thephases’propertiesaresummarizedinTable2. The particles are assumed to be spherical with a mono-dispersediameterequaltotheSauterdiameter.Theauthors havetestedsimulationswiththemediandiameterbut dis-crepancywithexperimentalresultswereobservedespecially forthepredictionoftheminimumfluidizationvelocity.

3.2.4. Boundaryconditions

Atthebottom(z=0),thefluidizationgridisaninletforthegas withanimposeduniformsuperficialvelocity(Vf)

correspond-ingtothefluidizationvelocityandawallfortheparticles.At thetopofthefluidizedbed,afreeoutletforbothgasand par-ticlesisdefined.Thewall-typeboundaryconditionisno-slip forthegasandfortheparticles(Fedeetal.,2016).

3.2.5. Initialconditions

ThesolidmassesusedineachcolumnaredetailedinTable3. Theycorrespondtoaratiobetweenthefixedbedheightand thecolumndiameterH0/Dcloseto3.

3.2.6. Simulationprogress

Thenumericalsimulationshavebeenperformedonparallel computerswith20cores.Anumericalsimulationisdivided

Table3–Solidmasses.

Column 5cm 3cm 2cm Massofsolid(g) 2,827 741 181

Table4–Experimentalandpredictedminimum fluidizationvelocities.

Umf(cms−1) D=5cm D=3cm D=2cm

Exp.glasscolumn 2.92 3.82 4.11 Exp.steelcolumn 3.14 3.02 3.19 Simulation 3.41 3.50 3.82

intotwosteps:atransitorystepof10scorrespondingtothe establishmentofthehydrodynamicsofthefluidizedbedand anestablishedregimeduringwhichthe statisticsare com-putedfor100s.Thenumericalradialprofilespresentedinthis studyarefromtheO–xdirectionbuttheyaresymmetricalwith theprofilesoftheO–ydirection.Theauthorshavecheckedthat alltime-averagedbedvaluesareaxisymmetric.

4.

Results

and

discussion

Asshown inFig.4,Vanniet al.(2015b)haveobservedwall effectsfora2cmglasscolumn,consistinginanincreaseofthe hystereticbehaviorbetweenincreasinganddecreasing pres-suredropcurvesandoftheminimumfluidizationvelocityin theglasscolumn.Thisobservationislessmarkedinthesteel columnduetothereductionofelectrostaticforces.

Defluidization curves in Fig. 5 present a comparison betweennumericalresultsandexperimentalmeasurements carriedoutofthetotalbedpressure dropatdecreasinggas velocityinthesteelcolumnVannietal.(2015b).The horizon-tal dashedlines correspondtothetheoreticalbed pressure drop(equaltothebedweightpercolumnsurfacearea).A rela-tivegoodagreementcanbeobserved.However,thenumerical resultsslightlyunderestimatethepressureinthefixedbed part.Thisleadstoanoverestimationoftheminimum fluidiza-tionvelocityasdetailedinTable4.Thisslightoverestimation oftheminimumfluidizationvelocitycanbeattributedtothe nonregular shapeofparticles (Fig.6).Itisnoteworthythat thenumericalresultspredictanincreaseoftheminimum flu-idizationvelocityforthecolumnof2cmastheexperimental ones,confirmingtheexistenceofwalleffects.TheThonglimp etal.(1984)correlationwasusedtocalculateUmf.Thevalue

of3.2cms−1wasfoundlogicallyclosetotheexperimentalUmf

valueinthesteel5cmcolumnwheretheelectrostaticandwall effectsareminimum.

Forthefollowingsimulations,thegasfluidizationvelocity hasbeenfixedat12cms−1.Thefluidizationratioisthenclose to3forthecolumnsof5and3cmandto2.8forthecolumn of2cm.Theminimumfluidizationvelocitiesusedto calcu-latethefluidizationratioaretheexperimentalones.Theyare extractedfromFig.5usingtheDavidsonandHarrisonmethod (Davidsonetal.,1963).

Thepredictedtime-averaged gaspressure atthe wallas afunctionofheightforthethreesteelcolumnsispresented inFig.7.Thenumericaldatawere extractedfromthe cells directlyatthewalltoplotthesecurves.Forallcases,thegas pressuredropislinearinsidethebedwithalmostthesame slope. Thepressure gradient isthen constant and isnota functionofthecolumndiameter.Moreover,the experimen-talmeasurementsforthecolumnof5cmshowaverygood agreementwiththenumericalvalues.

Table 5shows the time-averagedpressure gradient pre-dictedatthewall.Thevaluesofthepressuregradientinside thebedforthedifferentcolumnsarealmostthesameas pre-viouslyobserved.Usingthispressuregradientatthewall,the voidfractionoftheoverallfluidizedbedhasbeenestimated

(7)

Fig.4–ExperimentalbedpressuredropcurvesintheglassandsteelcolumnsforH0/D=3.

Fig.5–Comparisonbetweenexperimentalmeasurements forsteelcolumnandnumericalresultsofthebedpressure dropcurvesobtainedatdecreasinggasflowrate.

Fig.6–SEMviewoftungstenparticles.

Table5–Time-averagedgaspressuredrop,bedheight andsolidvolumefraction,Vf=12cms−1.

Column 5cm 3cm 2cm P/L(mbarm−1) 1,006 1,012 1,017 Bedheight(cm) 13.79 10.05 5.55 ¯ ˛g 0.468 0.465 0.462 ˛g 0.497 0.512 0.517 error(˛¯g,˛g) 6.1% 10.1% 11.9%

byneglectingthefrictionforcesoftheparticlesatthewalland assuminganequilibriumbetweenthebuoyancyandthedrag forces: ¯ ˛g=1− 1 (p−g)g P L (1)

Thisestimationofthebedvoidfractioncanbecompared to the mean volume fraction obtained by integrating the localvoidfractionovertheentirecellsofthecomputational domain:

˛g=







Vbed

˛gdV (2)

Thiscomparison(lastline,Table5)showsthattherelative errorincreaseswiththereductionofcolumndiameter.Hence, fortheconditionstested, thefrictionforcesatthewallcan notbeneglectedanymorefora2cmcolumnoffluidization. Thesmallerthecolumndiameteris,thehigherarethe wall-particlefrictionforces.Whenthecolumndiameterdecreases, theratiobetweenthewallsurfaceandtheparticlesvolume increaseswhichgeneratesaliftingeffectofthebed.Thus,the reductionofthecolumndiameterincreasesthevoidfraction causingthebedexpansion.Noexperimentalmeasurements

P-P

ref

(mbar)

0 50 100 150

Height (cm)

0 5 10 15 20 Simulations column D=5cm Simulations column D=3cm Simulations column D=2cm Experiments column D=5cm

Fig.7–Time-averagedaxialprofilesofgaspressureatthe wallfromnumericalsimulationsandexperimental measurements(Prefisavalueabovethedensefluidized

(8)

Table6–Phasespropertiesusedforthesimulationof sandparticle.

Parameters Value

Gasdensity(Argon) 1.6kgm−3 Gasviscosity 2.23×10−5Pas Sandparticlediameter 206␮m Sandparticledensity 2,600kgm−3

Umf(Thonglimpetal.,1984) 3.2cms−1

havebeenperformedinthesteelcolumntobecomparedwith thesepredictedvalues.

Sometime-averagedradialprofileshavebeencalculated at2/3ofthebedheight.Thebedheighthasbeendetermined (Table5)fromthechangingslopeoftheaxialprofilesofthe time-averagedgaspressure(Fig.7).Consequently,theprofiles wereextractedononecellatthepositionofz=9.1cm,6.6cm and3.6cmrespectivelyforthecolumnsof5cm,3cmand2cm.

Fig.8(a)showsthetime-averagedradialprofileofthesolids volumefraction.Numericalpredictionspresentaminimum at the center of the tube and a maximum near the wall. Thistrendisthesameasthatnormallyobservedfor classi-calGeldartgroupBparticles.However,byperformingspecific simulationswithsandparticles(Table6)ofsimilarminimum fluidizationvelocity,wehaveverifiedthatthefluidizationof tungstenpowdershowshighsolidvolumefractionin compar-isontoresultsobtainedformoreconventionalsandparticles. Thus,theexpansionofthispowderislowerthanforsand par-ticles.Moreover,itisnoteworthythatthereductionofcolumn

diameter inducesa reduction ofthe extremaof the time-averagedradialprofileofthesolidsvolumefractionandsoan increaseofthemeanbedvoidfraction.Thiseffectconfirms thepreviousobservationsontheincreaseofthevoidfraction.

Fig.8(b)and(c)showsrespectivelythetime-averaged verti-calcomponentofsolidphasevelocityandtheassociatedsolid massflux.Theradialprofilesaresimilarwithapositive maxi-mumvalueatthecenterandanegativeminimumonecloseto thewallduetobubbleriseinthecenterofthetube.Because thesizeandthenthevelocityofthebubblesare limitedby thereductionofthecolumndiameter,themaximumofthe particlevelocitydecreases.

Theparticlesriseatthecenterofthecolumnisobserved inFig.9whichshowsthesolidvelocityvectorsforthethree columns.Anupwardsolidflowinthecenterofthetubeand adownwardmoreconcentratedsolidflownearthewallare clearlyobservedwiththeformationoftwo2Dloops(resulting intorusin3Dcylindricalgeometry)atthebottomandatthe topofthefluidizedbed.Ascanbenoticed,thereductionofthe columndiameterdoesnotinfluencethelocalbehaviorofthe particlesinthefluidizedbed.Suchaconicalcirculation pat-ternhasalreadybeenobservedindensefluidizedbed(Fotovat etal.,2015;WertherandMolerus,1973).

Fig. 8(d) presents the time-averaged granular temper-ature of particles, which represents the frequency of particle–particlecollisions.Forthethreecolumns,itpresents amaximumatthecenterofthetubeandaminimumnear the wall.Itisnoteworthythatthe granulartemperatureof tungstenparticlesismorethanfifteentimeslowerthanfor the sand particles fora fluidization ratio of4.This means

(9)

Fig.9–Crosssectionalviewatthecenterofthecolumnofthetime-averagedvolumefractionsandofthestreamlinesofthe tungstenparticlesvelocity,Vf=12cms−1.

Fig.10–Effectofcolumndiameteronthetime-averagedflowvariablesradialprofilesatz=2/3×bedheight,Vf=12cms−1.

thatthehighdensityofthetungstenparticlestendsto drasti-callyreducethefrequencyofparticle–particlecollisions.The granulartemperatureincreaseswiththereductionofthe col-umndiameterdue toanincrease ofthe solidshearstress atthewall.Themeangranulartemperatureforthecolumn of2cmistwicehigherthanforthecolumnof5cm.So,the reductionofthecolumndiameterincreasesthefrequencyof particle–particlecollisionsbutthisremainsnotablylowerthan forconventionalpowders.

Fig.10(a)and(b)presentsthetime-varianceofthe verti-calandhorizontalsolidphasevelocities.Thetime-variance oftheverticalcomponentofsolidvelocityismaximumatthe centerandminimumnearthewallandthestandard devia-tionisapproximativelyofthesameorderasthetime-averaged velocity.Thetime-varianceofverticalvelocityoftungsten par-ticlesisofthesameorderasthatofsandparticles butthe time-varianceofhorizontalvelocityisthreetimeslower.The time-varianceoftheverticalandthehorizontalcomponents ofsolidvelocitiesdecreaseswiththereductionofthecolumn

diameterduetotheconfinementeffect.Thismeansthatthe solidmixinginthefluidizedbeddecreasesconfirmingthewall effectsexperimentallyobserved.Thisresultcanalsoexplain that the heat transfer ratesbetween the particles and the reactorwallsdecreasewiththecolumndiameter,asobserved experimentally(Vannietal.,2015b).

FirstCVDexperimentshaveshownthattheconversionrate ofsilaneislowerthanforconventionalpowders.Thiscanbe explainedbythesimulationresults,indicatingalower parti-clemobilityandalowerparticlemixingfortheseverydense tungstenparticles(Vannietal.,2015a).

5.

Conclusion

3Dnumericalsimulations,usinganEulerian2-fluidmethod, wereperformedinthisworktoanalyzethehydrodynamicsof veryheavyparticlesoftungsten,locatedoutsidetheclassical classificationofGeldart,inadensegas–solidfluidizedbed.

(10)

Theaimwastostudytheinfluenceofthereductionofcolumn diameter.

Thesimulationsaresuccessfulinreproducingthe exper-imentalmeasurements ofaxial gas pressure drop and the increaseofminimumfluidizationvelocitybydecreasingthe columndiameterfrom 5cm to2cm. Thenumericalresults pointedoutthatthereductionofcolumndiameterincreases thevoidfractioninthedensefluidizedbedduetowall-particle frictioneffects.From 5cmto2cm ofcolumndiameter,the particleshaveaconventionallocalcirculationwiththesolids risinginanannularzoneofincreasedbubbledevelopmentin thecolumncenterandthendescendingnearthewall.

However,theveryhighdensityoftungstenparticlestends todrasticallyreducetheirmobilityinsidethebed.Moreover, thereductionofcolumndiameterdecreasesthesolidmixing, confirmingthewalleffectsexperimentallyobserved.

As the measurements have shown that the tungsten particlescaneasilybecomeelectrostaticallycharged,future numericalsimulationswilltakethisfactintoaccount,inorder to improve the predictive capability of the model. Finally, futureexperimentalresearchopportunitieswouldbetocarry outexperimentsonthelocalbehaviorofdensegas–tungsten particle suspension. It would be interesting to access to thebubbles sizeand velocityusingappropriateprobesand accordingtotheparticlestrajectoryfromRadioactiveParticle Tracking(RPT)ortoPositronEmissionParticleTracking(PEPT) measurementstodeterminetheparticlevelocity.

Acknowledgements

ThisworkwasgrantedaccesstotheHPCresourcesofCALMIP undertheallocationP11032andCINESundertheallocation gct6938 made byGENCI.The authors acknowledge helpful commentsprovidedbyProfessorOlivierSimonin.

References

Basu,P.,1999.Combustionofcoalincirculatingfluidized-bed boilers:areview.Chem.Eng.Sci.54(22),5547–5557.

Boëlle,A.,Balzer,G.,Simonin,O.,1995.Second-orderprediction ofthepredictionoftheparticle-phasestresstensorof inelasticspheresinsimplesheardensesuspensions. Gas–ParticleFlows,ASMEFED,vol.28.,pp.9–18.

Davidson,J.,Harrison,D.,Jackson,R.,1963.FluidizedParticles. CambridgeUniversityPress,155pp.35s,1964.

Ergun,S.,1952.Fluidflowthroughpackedcolumns.Chem.Eng. Prog.48,89–94.

Fede,P.,Simonin,O.,Ingram,A.,2016.3Dnumericalsimulation ofalab-scalepressurizeddensefluidizedbedfocussingon theeffectoftheparticle–particlerestitutioncoefficientand particle-wallboundaryconditions.Chem.Eng.Sci.142, 215–235.

Fevrier,P.,Simonin,O.,Squires,K.D.,2005.Partitioningofparticle velocitiesingas–solidturbulentflowsintoacontinuousfield andaspatiallyuncorrelatedrandomdistribution:theoretical formalismandnumericalstudy.J.FluidMech.533,1–46.

Fotovat,F.,Ansart,R.,Hemati,M.,Simonin,O.,Chaouki,J.,2015.

Sand-assistedfluidizationoflargecylindricalandspherical biomassparticles:experimentsandsimulation.Chem.Eng. Sci.126,543–559.

Fox,R.O.,2014.Onmultiphaseturbulencemodelsforcollisional fluid-particleflows.J.FluidMech.742,368–424.

Gobin,A.,Neau,H.,Simonin,O.,Llinas,J.,Reiling,V.,Sélo,J.,2003.

Fluiddynamicnumericalsimulationofagasphase polymerizationreactor.Int.J.Numer.MethodsFluids43, 1199–1220.

Hofbauer,H.,Rauch,R.,Loeffler,G.,Kaiser,S.,Fercher,E., Tremmel,H.,2002.Sixyearsexperiencewiththe

FICFB-GasificationProcess.In:12thEuropeanConferenceand TechnologyExhibitiononBiomassforEnergy,Industryand ClimateProtection.

Kunii,D.,Levenspiel,O.,1991.FluidizationEngineering. ButterworthHeinemann.

Liu,S.-S.,Xiao,W.-D.,2014.Numericalsimulationsofparticle growthinasilicon-CVDfluidizedbedreactorviaaCFD-PBM coupledmodel.Chem.Eng.Sci.111,112–125.

Liu,M.,Wen,Y.,Liu,R.,Liu,B.,Shao,Y.,2015a.Investigationof fluidizationbehaviorofhighdensityparticleinspoutedbed usingCFD-DEMcouplingmethod.PowderTechnol.280,72–82.

Liu,X.,Zhu,C.,Geng,S.,Yao,M.,Zhan,J.,Xu,G.,2015b.Two-fluid modelingofGeldartAparticlesingas–solidmicro-fluidized beds.Particuology21,118–127.

Lun,C.,Savage,S.,1986.Theeffectsofanimpactvelocity dependantdevelopedbyshearedgranularmaterial.Acta Mech.63,15–44.

Méchitoua,N.,Boucker,M.,Laviéville,J.,Hérard,J.,Pigny,S.,Serre, G.,2003.Anunstructuredfinitevolumesolverfortwo-phase water/vapourflowsmodellingbasedonellipticoriented fractionalstepmethod.In:NURETH10,Seoul,SouthKorea.

Mazaudier,F.,Proye,C.,Hodaj,F.,2008.Furtherinsightinto mechanismsofsolid-stateinteractionsinUMo/Alsystem.J. Nucl.Mater.377(3),476–485.

Morioka,S.,Nakajima,T.,1987.Modelingofgasandsolid particlestwo-phaseflowandapplicationtofluidizedbed.J. Theor.Appl.Mech.6(1),77–88.

Neau,H.,Laviéville,J.,Simonin,O.,2010.NEPTUNECFDhigh parallelcomputingperformancesforparticle-ladenreactive flows.In:7thInternationalConferenceonMultiphaseFlow, ICMF2010,Tampa,FL,May30–June4.

Neau,H.,Fede,P.,Laviéville,J.,Simonin,O.,2013.High performancecomputing(HPC)forthefluidizationof particle-ladenreactiveflows.In:The14thInternational ConferenceonFluidization,FromFundamentalstoProducts, Noordwijkerhout,Netherlands.

Pannala,S.,Daw,C.S.,Finney,C.E.,Boyalakuntla,D.,Syamlal,M., O’Brien,T.J.,2007.Simulatingthedynamicsofspouted-bed nuclearfuelcoaters.Chem.VaporDepos.13(9),481–490.

Simonin,O.,Février,P.,Laviéville,J.,2002.Onthespatial distributionofheavy-particlevelocitiesinturbulentflow: fromcontinuousfieldtoparticulatechaos.J.Turbul.3(1),1–40.

Srivastava,A.,Sundaresan,S.,2003.Analysisofafrictional kineticmodelforgas/particleflows.PowderTechnol.129, 72–85.

Thonglimp,V.,Hiquily,N.,Laguerie,C.,1984.Vitesseminimalede fluidisationetexpansiondescouchesfluidiséesparungaz. PowderTechnol.38,233–253.

Vahlas,C.,Caussat,B.,Serp,P.,Angelopoulos,G.N.,2006.

PrinciplesandapplicationsofCVDpowdertechnology.Mater. Sci.Eng.R:Rep.53(1),1–72.

Vanni,F.,Caussat,B.,Ablitzer,C.,Iltis,X.,Bothier,M.,2015a.

Siliconcoatingonverydensetungstenparticlesbyfluidized bedCVDfornuclearapplication.Phys.StatusSolidi(A)212(7), 1599–1606.

Vanni,F.,Caussat,B.,Ablitzer,C.,Brothier,M.,2015b.Effectsof reducingthereactordiameteronthefluidizationofavery densepowder.PowderTechnol.277,268–274.

Wang,F.,Fan,L.-S.,2011.Gas–solidfluidizationinmini-and micro-channels.Ind.Eng.Chem.Res.50(8),4741–4751.

Wang,J.,Tan,L.,VanderHoef,M.,vanSintAnnaland,M., Kuipers,J.,2011.Frombubblingtoturbulentfluidization: advancedonsetofregimetransitioninmicro-fluidizedbeds. Chem.Eng.Sci.66(9),2001–2007.

Wen,C.,Yu,Y.,1965.Mechanicsoffluidization.Chem.Eng.Symp. Ser.62,100–111.

Werther,J.,Molerus,O.,1973.Thelocalstructureofgasfluidized beds—II.Thespatialdistributionofbubbles.Int.J.Multiph. Flow1(1),123–138.

Yang,W.-C.,2007.Modificationandre-interpretationofGeldart’s classificationofpowders.PowderTechnol.171(2),69–74.

(11)

Ydstie,B.E.,Du,J.,2011.ProducingPoly-SiliconfromSilaneina FluidizedBedReactor.INTECHOpenAccessPublisher.

Zivkovic,V.,Biggs,M.,2015.Onimportanceofsurfaceforcesin amicrofluidicfluidizedbed.Chem.Eng.Sci.126,

143–149.

Zweifel,T.,Palancher,H.,Leenaers,A.,Bonnin,A.,Honkimaki,V., Tucoulou,R.,VanDenBerghe,S.,Jungwirth,R.,Charollais,F., Petry,W.,2013.CrystallographicstudyofSiandZrNcoated U–Moatomisedparticlesandoftheirinteractionwithal underthermalannealing.J.Nucl.Mater.442(1),124–132.

Figure

Fig. 1 – Representation of the tungsten particles with a diameter of 75 ␮ m in Geldart’s classification modified by Yang (2007).
Table 1 – Particle properties and particle size distribution characteristics of the tungsten particles obtained with a Malvern Master Sizer Sirocco 2000 in dry mode.
Fig. 3 – 3D mesh for the numerical simulation with 277,635 cells.
Fig. 4 – Experimental bed pressure drop curves in the glass and steel columns for H 0 /D = 3.
+3

Références

Documents relatifs

Open Archive TOULOUSE Archive Ouverte (OATAO) OATAO is an open access repository that collects the work of Toulouse researchers and makes it freely available over

Many factors, such as the initial concentration of carbon source, fermentation time, incubation temperature, inoculum size, shell size, volume and medium composition

The criteria for the safety evaluation of a mechanical recycling process to produce recycled PET intended to be used for the manufacture of materials and articles in contact with

The tool then reads the trace to extract the role information and convert it into a form suitable for interactive graphical display. The tool evaluates the roles of the objects

Genes encoding rhamnose synthase and beta-xylosidase are represented in the GC and TC tissue specific ESTs indicating that the mucilage synthesis and secretion pathway observed

image deblurring, spatially varying blur, integral operator approximation, wavelet compression, windowed convolution.. AMS

Costa, 2010, Simulation and life cycle assessment of process design alternatives for biodiesel production from waste vegetable oils, J. Yao, 2012, Economic and environmental

The Halpin–Tsai adapted model permits to link the mechanical behavior to the aspect ratio of randomly dispersed nanowires that plays the role of entanglements in the rubbery