• Aucun résultat trouvé

Numerical simulation of spreading drops

N/A
N/A
Protected

Academic year: 2021

Partager "Numerical simulation of spreading drops"

Copied!
10
0
0

Texte intégral

(1)

To link to this article

:

DOI

:

10.1016/j.colsurfa.2013.04.046

URL

: http://dx.doi.org/10.1016/j.colsurfa.2013.04.046

To cite this version

:

Legendre, Dominique and Maglio, Marco

Numerical simulation of

spreading drops.

(2013) Colloids and Surfaces A: Physicochemical

and Engineering Aspects, vol. 432 . pp. 29-37. ISSN 0927-7757

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 : 10539

Any correspondance concerning this service should be sent to the repository

administrator:

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

(2)

Numerical

simulation

of

spreading

drops

Dominique

Legendre

,

Marco

Maglio

InstitutdeMécaniquedesFluidesdeToulouse(IMFT),UniversitédeToulouse,CNRS-INPT-UPS,France

g

r

a

p

h

i

c

a

l

a

b

s

t

r

a

c

t

•Numerialsimulationofspreadingdropsonawettablesurface.

•Wediscusstheeffectofthemainparameters(density,viscosity,surfacetension,dropradiusandwettability).

•Powerlawevolutionsobservedinexperiments(t1/2,t1/10)arerecoveredanddiscussed.

•Thecombinedeffectofviscosityandwettabilityhasasignificanteffectonthespreading.

Keywords: Dropspreading Numericalsimulation

a

b

s

t

r

a

c

t

Weconsideraliquiddropthatspreadsonawettablesurface.Differenttimeevolutionshavebeen observedforthebaseradiusrdependingoftherelativeroleplayedbyinertia,viscosity,surface ten-sionandthewettingcondition.Numericalsimulationswereperformedtodiscusstherelativeeffectof theseparametersonthespreadingdescribedbytheevolutionofthebaseradiusr(t)andthespreading timetS.Differentpowerlawevolutionsr(t)∝tnhavebeenobservedwhenvaryingtheparameters.Atthe

earlystageofthespreading,thepowerlawt1/2(n=1/2)isobservedaslongascapillarityisbalancedby

inertiaatthecontactline.Whenincreasingtheviscositycontribution,theexponentnisfoundtoincrease despitetheincreaseofthespreadingtime.Theeffectofthesurfacewettabilityisobservedforliquids moreviscousthanwater.Forasmallcontactangle,thepowerlawt1/2isthenfollowedbythefamous

Tannerlawt1/10oncethedropshapehasreachedasphericalcap.

1. Introduction

When a liquid drop contacts a wettable surface, the liquid spreadsoverthesolidtominimizethetotalsurfaceenergy.The firstmomentsofspreadingtendtoberapid(somemilli-seconds for millimetersized waterdrops) whileit takes a muchlonger timeforviscousliquidstocompletelyspread.Fromtheliterature

[22,17,9,1,6,3,2,24,7],itappearsthatseveraltimeevolutionscanbe observedforthebaseradiusdependingifinertia,viscosityand/or

∗ Correspondingauthors.Tel.:+33534322818;fax:+33534322991. E-mailaddress:legendre@imft.fr(D.Legendre).

gravityareinvolved.Inaddition,theeffectofthesubstrate wett-abilityondropspreadingisnotyetclearlyunderstood.

Theevolutionofthewettedsurfaceisreportedusingeitherthe baseradiusr(t)orthewettedsurfaceA(t)=r2(t).The

experimen-talresultsareusuallypresentedinordertoprovideapowerlaw evolutionundertheform:

r(t)∼tn,orA(t)∼t2n, (1)

Themostpopularlawfordropspreadingiscertainlytheso-called Tanner law[23,22] withn=1/10. Thisevolution can be simply obtainedbyconsideringthattheshapeofthedropisathin spheri-calcapandthatthecontactanglefollowstheCox-Voinovrelation. Suchevolution hasbeenfoundtobeinagreementwithseveral

(3)

experiments(see[3]).TheTannerlawconcernsspreading situa-tionsinwhichsurfacetensionisbalancedbyviscosity.Varyingthe relativedifferentcontributions(inertia,gravity,viscosity,surface tension)differentevolutionshavebeenreported:n=0.1,n=0.14, n=0.16,n=1/8,n=1/7,n=1/2[17,9,3].Thet1/2powerlawresults

fromabalancebetweenthecapillarypressureandtheinertial pres-sure.Such evolution hasbeenobserved in severalexperiments

[1,2,24].LaviandMarmur[15]performedexperimentsfor differ-entliquidsonacoatedsiliconwaferandreportedthefollowing exponentialevolution: A(t) Af = 1−exp



AK f m



, (2)

whereAfisthefinalvalueofthewettedareaandisthe

dimen-sionlesstime=t/V1/3.LaviandMarmur[15]reportsvaluesfor

mintherangem=0.618−0.964.Theinterestofthisrelationisto proposeapowerlawoftheformtminthelimitoflargefinalwetted

surfaceAf.Asitwillbeshowninthesimulationsreportedinthis

paper,theviscosityhasanimportanteffectonthedrop spread-ing.Forlowviscousliquids(aswater),apowerlawoftheform(1)

canbeobservedduringalmostalltheevolution.Whenincreasing theviscosity,theslopeoftheevolutionchangesovertime, show-inganexponentialdependencyoftheform(2).Thewettabilityhas alsoanimpactonthedropspreading.Thedistinctionbetweenthe differentwettingstatesisusuallymadebyconsideringthe equi-libriumspreadingcoefficientSeq=(cosS−1),whichrepresents

thesurfacefreeenergyrelativetoitsvalueforcompletewetting. Whenadropisdepositedona planesurface,theinitialcontact anglei≈180◦ madebytheinterfacewiththewallislargerthan

equilibriumS.AninitialspreadingcoefficientSi=(cosi−1)can

beintroducedtocharacterizetheinitialenergyofthesystem.The differenceS=Seq−Si=cosS−cosiindicatesthenatureofthe

spreading.SinceS<0,thedropspreadsuntilthecontactangle stabilizesattheequilibrium angleS.Neitherdifferent

theoret-icalmodelsnorexperimentalobservationshave showna direct effectofS=cosS−cosionthedropspreading.Furthermore,the

impactofthesurfacewettabilityonthedropspreadingisstillunder controversy.Bianceetal.[1]observedthescalinglawr(t)t1/2in

theirexperimentsofwaterdropsspreadingonacompletely wet-tingsurface.Birdetal.[2]performedexperimentswithdroplets ofdifferentwater-glycerolmixtures(1.0,3.7,and10.7cP)on sili-conwafersurfacesofvariablewettability(fromS=0◦toS=180◦).

Thespreadingradiusisfoundtofollowapower-lawevolutionof theform(1)wheretheexponentndependsontheequilibrium contactangle.Typicallyitvariesfromn=1/2forperfectwetting surfaceston=1/4fornonwettingsurfaces.Chenetal.[7]observed spreadingofwaterdroponhydrophobicglasswithn=0.3.Onthe contrary,therecentexperimentsofWinkelsetal.[24]pointtoward aspreadingregimethatisindependentofwettabilitywithn=1/2. Theexperiments weremade withwater dropsonthree differ-entsubstrateswithdifferentequilibriumcontactanglesS:clean

glass(almostperfectlywetting,S∼0◦),coatedglass(S∼65◦),and

Teflon-coatedglass(S∼115◦).

Theaimofthisworkistodiscusstheeffectofthemain param-eters controllingthe drop spreading:viscosity, density, surface tension,dropvolumeandsurfacewettability.Thetimeevolution r(t)ofthebaseradiusisanalyzedaswellasthetimetSforthe

com-pletespreadingofthedrop.Eachparameterhasbeenvariedina largerangewhilekeepingconstanttheotherparameters.

2. Numericalprocedure

2.1. NavierStokessolver

Thenumericalsimulationsreportedinthisworkareperformed withthevolumeoffluid(VoF)solverdevelopedintheJADIMcode

[4,12].Theone-fluidsystemofequationsisobtainedbyintroducing theone-fluidfunctionCusedtolocalizeoneofthetwophases.In thisstudy,wedefineCasC=1inthedrop(phase1),andC=0inthe surroundingair(phase2).Theone-fluidfunctionCmakespossible thedefinitionoftheonefluidvariablesU=CU1+(1−C)U2forthe

velocity,P=CP1+(1−C)P2 forthepressure,=C1+(1−C)2 for

thedensityand=C1+(1−C)2fortheviscosity.Theposition

oftheinterfaceisthengivenbythetransportequation:

C

t +U.

C=0 (3)

The two fluids areassumed tobe Newtonian and incompress-iblewithnophasechange.Underisothermalconditionandinthe absenceofanysurfactant,thesurfacetensionisconstantand uni-format theinterfacebetweenthetwo fluids.In suchcase,the velocityfieldUandthepressurePsatisfytheclassicalone-fluid formulationoftheNavier–Stokesequations:

.U=0 (4) 



U

t +U.

U



=−

P+

.+g+F (5)

where istheviscous stresstensor, gis thegravity.F isthe

capillarycontribution:

F=

.nnıI (6)

whereisthesurfacetension,ndenotesbyarbitrarychoicethe unitnormaloftheinterfacegoingoutfromthedropandıIisthe

Diracdistributionassociatedtotheinterface.

ThesystemofEqs.(3)–(6)isdiscretizedusingthefinite vol-umemethod.Timeadvancementisachievedthroughathird-order Runge-Kuttamethodfortheviscousstresses.Incompressibilityis satisfiedattheendofeachtimestepthoughaprojectionmethod. Theoverallalgorithmissecond-orderaccurateinbothtimeand space. The volume fraction C and the pressure P are volume-centeredandthevelocitycomponentsareface-centered.Dueto thediscretizationofC,itresultsanumericalthicknessofthe inter-face,thecellscutbytheinterfacecorrespondingto0<C<1.The specificaspectofourapproachcomparedtotheclassicalVoFor LevelSetmethods[19–21]concernsthetechniqueusedtocontrol thestiffnessoftheinterface.Inourapproachnointerface recon-structionorredistancingtechniquesareintroduced.Theinterface locationandstiffnessarebothcontrolledbyanaccuratetransport algorithmbasedonFCT(Flux-Corrected-Transport)schemes[25]. Thismethodleadstoaninterfacethicknessofaboutthreegridcells duetotheimplementationofaspecificprocedureforthevelocity usedtotransportC,inflowregionofstrongstrainandshear[4].The numericaldescriptionofthesurfacetensionisoneofthecrucial pointswhenconsideringsystemswherecapillaryeffectscontrol theinterfaceshape.Thisinterfacialforceissolvedusingtheclassical CSF(continuumsurfaceforce)modeldevelopedby[5].Aclassical problemofthisformulationisthegenerationofspuriouscurrents

[13,18].Inordertodecreasespuriouscurrentsintensity,aclassical methodintroducedby[5]wasemployed,whichconsistsin cal-culatingthesurfacecurvaturefromasmootheddensitygradient, whilethediscretizationofthedeltafunctionusesanun-smoothed density.ThespuriouscurrentsinourcodeJADIMhavebeen char-acterizedby[12].Theirmaximummagnitudeevolvesas0.004/, inagreementwithothercodesusingtheBrackbill’sformulation. 2.2. Numericalmodelingofthecontactangle

Themodeling usedinthis workaims torespectthephysics atthecontactline.Itisbasedontwomainconsiderationsatthe macroscopicscale:

(4)

Fig.1.Statementoftheproblem.Initialandfinaldropshapecharacterizedbythe baseradiusr0andrf,respectively.TheradialandverticalsizeareLr=Lz=3R0.his

theinitialdistancebetweenthedropcenterandthewall.

-Thewallconditionseenbythefluidisanonslipcondition. -There is a direct relation between the interfacevelocity and

theshapeoftheinterfacecharacterizedbytheapparentcontact angle.

TheobjectiveistosolvethephysicattheapparentscaleLwhere thecontactangleistheapparentordynamiccontactangle.Atthis scalethefluidobeysanonslipconditionfortheresolutionofthe momentumequation:

UW =0 (7)

HydrodynamicsatasmallerscalethanLhasanimportantrolesince itlinkstheapparentangletothestaticangle.Asub-gridmodeling ofsucheffectneedstobeintroducedinthesimulation.Thisisdone byusingtherelationderivedbyCox[8]inordertolinktheapparent angletothestaticangle:

g(W)=g(S)+CaLn



L



(8) whereCa=Ucl/isthecontactlinecapillarynumber.Thefirst

Vol-umeofFluidnodebeinglocatedat/2fromthewall,itfollowsthat L=/2.Consideringthevalueoftheobservedsliplength[14],we haveusedafixedvalue =10−9mforthesliplengthin(4)inthe simulationsreportedinthiswork.Notethatexperimental observa-tionsgivealsoarecommendationforthechoiceofthemacroscopic lengthLconsistentwiththedynamicmodelusedinournumerical modeling.Indeedoneshouldhave/2=L≈10−4−10−5m.

Thecomparisonbetweenourmodelingwithothervariant mod-elings,i.e.staticcontactangleversusdynamiccontactangle,noslip conditionversusslipcondition,ispresentedin[16].Agridandtime convergencehasbeenperformedanddiscussedforthedifferent possiblemodels.Thesecomparisonsclearlystressthestrongeffect ofthenumericalmodelingusedforthesimulationofdynamic con-tactlinesasillustratedinthenextsection,wherethesimulations arecomparedwithexperimentsforwaterandmoreviscousliquids. 2.3. Comparisonwithexperimentsforwaterandviscousdrops

We consider a spherical droplet of volume V=4R3 0/3

depositedonanhorizontalwallthatentersincontactwithan hori-zontalwettablesurfaceattimet=0.Allthenumericalexperiments reportedinthisworkaredescribedbyFig.1.Thecomputational domainisasquaredomainofequalradialandverticalextension

10−2 10−1 100 101 102 103 0 0.2 0.4 0.6 0.8 1 τ r / r f

Fig.2.Comparisonwiththeexperimentsforviscousdrop[15].Thenormalized radiusr/rfisreportedversusthenormalizedtime=t/V1/3.Contactanglemodels

usedforthesimulations:—Dynamicmodel,−−−staticmodel

Lr=Lz=3R0.Theaxisofsymmetrycorrespondstothewest

bound-ary,thewettingconditionsgivenbyeq.(7-4)areimposedonthe southboundaryandclassicalwallconditions(noslip)areimposed onthetwootherboundaries(eastandnorth).Theinitial condi-tioncorrespondstoadropinitiallylocatedatthedistanceh≤R0

fromthewall. Theeffectofthis initialconditionas wellasthe gridandtimeconvergenceshavebeenperformed[16]by consid-eringdifferentinitiallocationsofthedropcenter(h=0.95,0.98, 0.99,0.995,0.998R0),differentgrids(correspondingto32,64,128

and167nodesperradius)anddifferenttimesteps(t=10−5 to t=10−8s).

Squalaneandwaterdropsareconsideredforcomparisonwith theexperimentsofLavi&Marmur [15]Winkelsetal.and[24], respectively.

We first consider the experiments of [15]. They have con-sidered the spreading of squalane drops with R0=1mm. The

fluidpropertiesare=809kg/m3,=0.034Pasand=0.032N/m.

Thecomparisonbetweenoursimulationsandtheexperimentsis showninFig.2.Thenormalizedradiusr/rfisreportedversusthe

normalizedtime=t/V1/3.Thedynamicmodeliscomparedwith

thestaticmodel(aconstantcontactangleSbeingimposed

dur-ingallthesimulation).Asignificantdifferenceisobservedbetween thedynamicmodel(4)andthestaticmodel(W=S).Thedynamic

modelsgiveasatisfactoryagreementwiththeexperimentsof[15]. Wenowconsiderthespreadingofwaterdropsandwe com-pare our simulations with the experiments performed by [24]

formillimeter-sizedwaterdropletsinair(R0=0.5mm).Wehave

selectedthecasescorrespondingtoverydifferentcontactangles S=115◦andS≈0◦.

Basedonpreliminarytests[16],thecomparisonwiththe exper-imentsperformedby[24]areperformedwithagridcorresponding to128nodesperradiusR0andatimestept=10−7s.Theinitial

positionish=0.998R0.ThecomparisonisshowninFig.3wherethe

normalizedcontactlineradialpositionr/R0isshownasfunctionof

thetimenormalizedbythecapillary-inertialtimet=(R30/)1/2.

Theagreementisverysatisfactoryforthetwocasesconsidered.In particular,thepowerlawt1/2isclearlyreproducedbythe

simu-lations.Consequently,oursimulationsdonotshowanoticeable effect ofthesurface wettabilityonthe powerlawevolution in agreementwith[24]butinoppositionwith[2].Below,we con-siderindetailthispointwithadditionalsimulationsperformedfor afluidtentimesmoreviscousthanwater.Notealsothatour sim-ulationrevealstheoscillationofthecontactlineattheendofthe

(5)

32 D.Legendre,M.Maglio/ColloidsandSurfacesA:Physicochem.Eng.Aspects432 (2013) 29–37 10−3 10−2 10−1 100 101 10−2 10−1 100 101 1 2 t / tρ r / R0

Fig.3.Evolutionofthecontactlineradialpositionr/R0asfunctionofthe

nor-malizedtimet/(R3 0/)

1/2

.Experiments[24]:()S=115o,(♦)S≈0◦.Numerical

simulations:−−−S=115◦,—S=5◦.

spreading.Suchoscillationarenotreportedintheexperimentsof

[24]andwillbediscussedinthefollowing.

3. Results

Weconsidertheeffectofthemainparameterscontrollingdrop spreading: fluid properties (liquid density, liquid viscosity and surfacetension),thedropsizeandthewettingpropertiesofthe sur-face.Eachparameterisvariedindependentlyoverawiderangeof variation.Allthesimulationshavebeenperformedconsideringair asthesurroundingfluid.Asaconsequencethedensityandviscosity ratiovalues(i.e.2/1and2/1)arenotconstantbuttheyalways

remainmuchsmallerthanunitysothatthereisnoexpectedeffect ofboththeviscosityandthedensityoftheaironthespreading. 3.1. Inertial-capillaryregimeofspreading

Wefirstconsiderlow-viscousspreading.Thedensity,the sur-facetensionandthedropradiusarevariedindependentlywhile theviscosityis setfixedtotheviscosityofwater =0.001Pas.

Fig. 4 shows the evolution of the contact line radius when varying the density from =100kg/m3 to =4000kg/m3 with

theotherparameterskeptconstant:=0.072N/m,=0.001Pas, R0=0.5mmandS=65◦.Whenthedensityisincreasedthe

spread-ingtimegrows(thecurvesaretranslatedtotheright)sincethe inertiaoftheliquidtobeacceleratedbythemovingcontactline ismore important. We notice that thesame powerlaw evolu-tiont1/2isobservedforalltherange ofdensityconsidered.The

t1/2 powerlaw resultsfrom acapillary-inertial balance.Indeed,

whenthesphericaldropofradiusR0contactsthesolid,thebalance

betweenthecapillarypressure∼R0/r2andtheinertialpressure

∼(dr/dt)2givesafterintegrationthefollowingevolution:

r(t)∼



R0





1/4

t1/2 (9)

Fig.4(bottom)confirmsthatthedropspreadingiscontrolled byabalancebetweencapillary andinertia,sincetheuseofthe timenormalizedusingthecapillary-inertiacharacteristictimet=

(R3 0/)

1/2

makespossibleagoodcollapseofthedifferentcurves.

Fig.4alsorevealscontactlineoscillationsatboththeearlyand thefinalstageofspreading.IndeedasshowninFig.5,theinterface issignificantlydeformedbythepropagationof capillarywaves.

Fig. 4.Effect of the liquid densityon the dropspreading for=0.072N/m, =0.001Pas, S=65◦ and R0=0.5mm. The normalized contact line radius

r*=(rr

0)/(rf−r0)isreportedasafunctionof(top)thedimensionaltimetand

(bottom)thenormalizedtimet/(R3 0/)

1/2

.

Theappearanceofacapillarywavethattravelsfromthecontact linetothetopofthedroplethasalsobeenreportedinthe simula-tionsreportedby[11]usingadiffuseinterfacemethod.Thisfigure reportstheevolutionoftheshapeofawaterdropforS=65◦.The

sequenceispresentedlinebylineforthetimesgiveninthelegend. Thetimestepsarenonregularinordertoshowparticularevents observedduringthespreading.Atthebeginningofthespreading, onlythecontactlineismoving.Thetopofthedropstartstomove duetothecapillarywaveinducedbytheinitialimpulsegenerated atthecontactline.Surprisinglythiscapillarywaveinducesthe ele-vationofthetopofthedrop.Asaconsequence,combinedwiththe displacementofthecontactline,thedropiselongatedintotwo orthogonaldirectionsresultinginthegenerationofapinchevent (see2ndimage(t=2.52ms),line3).Inthiscase,thepinch-offdoes notoccursresultinginthepropagationofarelativelylarge ampli-tudewave.Thiscapillarywavedisplacesagainthetopofthedrop inthedirectionoppositetothewall.Thisgeneratesasecondpinch thatresultsintheejectionofasmallsatellitedroplet(see2ndimage (t=3.34ms),line5).Then, thesatellitedropletimpactsand coa-lesceswiththedropgeneratingagainwaveoscillations,explaining thefinaloscillationsobservedforthecontactlineinFig.4.Such regimeofdropletejectionfollowingthedepositionofdropsonto

(6)

Fig.5. Shapeevolutionofawaterdrop(radialsection)forS=65◦.Thesequenceisshownlinebylineforthetimest=0.06,0.36,0.56,1.00,1.40,1.66,1.86,2.52,2.70,2.86,

2.98,3.12,3.28,3.34,3.52,3.60,4.80and5.80ms.

a solid substratehasbeen identified inexperiments and direct numericalsimulationsby[10]forsmallercontactangles(S≤65◦).

Weshowherethatthedropletejectioncanalsobeobservedfor largercontactangles.

WeillustrateinFig.6theeffectofthedensityonthedroplet ejec-tion.Thesimulationsareshownforthreedensities=700kg/m3,

=1000kg/m3 and=4000kg/m3.Theimagescorrespondingto

thetwosuccessivepinch-offsandtothemaximumheightofthe satellitedropletareshown.For=700kg/m3twopincheventsare

observedbutnodropletejectionisgeneratedbythesurface oscil-lation,whilefor=4000kg/m3thedropletisejectedatthefirst

pinchevent.

The effect of thesurface tensionis now reported in Fig. 7. The surfacetensionis variedfrom=0.007N/m to=0.2N/m. The other parameters are fixed to =103kg/m3, =0.001Pas,

R0=0.5mmandS=65◦.Thespreading timeis decreasedwhen

increasingthesurfacetension,sincetheinitialimpulse transmit-tedtothedropisincreased.Wealsoobservethatthepowerlaw

(7)

Fig.6.Shapeevolutionofawaterdrop(radialsection)showingthepincheventsforS=65◦.Firstline:=700kg/m3–fromlefttoright:t=2.16,2.64ms.Secondline:

=1000kg/m3fromlefttoright:t=2.52,3.28,3.34ms.Thirdline:=4000kg/m3fromlefttoright:t=4.56,4.94ms.

evolutiont1/2isnotsignificantlyaffectedbythechangeofthe

sur-facetension. Againtheuseofthenormalizedtime t/(R3 0/)

1/2

makespossiblethecollapseofthecurvesandclearlyrevealsthe capillaryoscillationsthatimpactthecontactlinespreading.

WereportinFig.8thespreadingtimetSforthecontactline

evo-lutionshowninFigs.4and7.Thespreadingtimesdeducedfrom simulationsforwaterdropswithradiiR0=0.5mmandR0=5mm

arealsoreported.tSisshownasafunctionofthecharacteristic

capillary-inertialtimet=(R30/)1/2.tSisdefinedasthetime

nec-essarytoreach98%ofthefinalequilibriumradiusrfofthecontact

line.Theplotconfirmsthatspreadingisclearlycontrolledbythe balancebetweeninertiaandcapillarity.Thespreadingtimecanbe estimatedas: tS≈1.7



R3 0 



1/2 (10) Thespreadingtimeisthusincreasedwhenreducingthesurface tensionandincreasingthedropdensityandthedropsize. 3.2. Viscosityeffect

WereportinFig.9theeffectoftheviscosityonthedrop spread-ing.Theviscosityisvariedfrom=0.001Pasto=0.07Pasfor =103kg/m3,=0.072N/m,R

0=0.5mmandS=65◦.The

spread-ingtimeisenlargedwhenincreasingtheviscosity(thecurvesbeing translatedtotheright)andtheoscillationsobservedattheendof thespreadingduetothepropagationofcapillarywavearedamped. Despitetheincreaseofthespreadingtime,theslopeofthe evolu-tionattheearlystageofthespreadingincreasesfromt1/2(water)

tot2/3forthelargestviscosityconsidered.Butthedurationofthe

powerlawregimeisreducedbecausetheendofthespreadingis characterizedbyanexponentialevolutionoftheformgivenby(2). Forcomparison,theTannerlawt1/10isalsoshowninthefigure.

Thereisnoclearevidencethatthistypeofevolutionisobservedin oursimulationsduringthefinalstageofthespreading.Infact,the staticcontactangleS=65◦istoolargetoobservesucharegime.

Forliquidsmoreviscousthanwater,weexpectthattheevolutionis controlledbyaviscous–capillarybalance.Theviscousstressnear thecontact linecanbeestimatedas∼(dr/dt)/ where isthe characteristiclengthcontrollingtheshearstressatthecontactline. Assuminganevolutionoftheform =R1−p0 rp,thebalancebetween

theviscousstressandthecapillarypressure∼R0/r2 gives,after

integration,theobservedt2/3evolution:

r(t)∼



R1/20 

2/3 t2/3 (11)

for p=3/2 corresponding to ∼R−1/20 r3/2. The corresponding

viscous–capillarycharacteristictime t=R0/ is usedinFig.4

(bottom)tonormalizethetime.Foraviscositylargerthan0.01Pas, asatisfactorycollapseofthecurvesisobservedindicatingthe tran-sitiontoaspreading regimecontrolledbythebalancebetween capillarityandviscosity.

TheevolutionofthespreadingtimetSisreportedinFig.10as

afunctionoftheviscous–capillarycharacteristictimet=R0/.

Twodifferentbehaviorsare clearlyobserved. For low viscosity, tSisfoundtoevolveast1/5(i.e.1/5sincealltheother

param-etersarenotvarying).We havenoexplanationoftheparticular evolution tS∝1/5 revealed by the simulation for low-viscous

drops. Forlargerviscositiesan expectedlinearevolution tS∝t

isobserved.Thelinearbehaviorresultsfromthelinearityofthe governingequations(Stokesequation)atlowReynoldsnumber. Theviscous–capillaryspreadingtimecanbeestimatedusingthe relation

tS≈55

R0

(8)

Fig.7. Effectofthe surfacetension onthedropspreading for =103kg/m3,

=0.001Pas, R0=0.5mm and S=65◦. The normalized contact line radius

r*=(rr

0)/(rf−r0)isreportedasafunctionof(top)thedimensionaltimetand

(bottom)thenormalizedtimet/(R3 0/)

1/2

.

3.3. Contactangleeffect

WefinallydiscusstheeffectofthestaticcontactangleSon

dropspreading.Figs.3,4and7showthatthet1/2powerlawisnot

affectedbythesurfacewettabilityatthebeginningofthe spread-ing,whenconsideringafluidhavingthesameviscosityaswater foralargerangeofcontactangles(S=5◦,65◦and115◦),in

agree-mentwiththeexperimentsreportedby[24]forwaterdrops.The wettingconditionsconsideredby[24](S=5◦,65◦ and115◦)are

nowsimulatedforafluidwithaviscositytentimeslargerthanthat ofwater(=0.01Pas),theotherparametersbeingkeptthesame asforwater(=0.072N/m,=103kg/m3)andR

0=0.5mm.

Addi-tionalwettingconditions(S=140◦andS=160◦)havealsobeen

considered.

Fig.11clearlyrevealsthatthewettingconditionhasnowa sig-nificanteffectonthespreadingrate.Whenthecontactangleis increasedtheslopeseemstodecreaseinagreementwiththe exper-imentsof[2].However,iftheradiusrisnormalizedusingtheradius reachedattheendoftheinertial-capillaryregime,allthecurves collapseonasinglecurveshowingat1/2evolutionindependenton

thecontactangleSinagreementwiththeexperimentsof[24].

Fig.8.SpreadingtimetSversusthecapillary-inertialtimet=(R30/) 1/2when

varyingtheliquiddensity,thesurfacetensionandthedropradius.(◦)is vary-ingfrom100kg/m3to4000kg/m3for=0.001Pas,=0.072N/m,R

0=0.5mmand

S=65◦.()isvaryingfrom0.007N/mto0.2N/mfor=103kg/m3,=0.001Pas,

R0=0.5mmandS=65◦.( )R0isvaryingfrom0.5mmto5mmfor=103kg/m3,

=0.001Pas,=0.072N/mandS=65◦.Thecontinuouslinerepresentstherelation

tS=1.7t

Fig.9. Effectof theliquidviscosityonthe dropspreadingfor =0.072N/m, =103kg/m3,R

0=0.5mmandS=65◦.Thenormalizedradiusr*=(r−r0)/(rf−r0)is

(9)

Fig.10.SpreadingtimetSversustheviscous–capillarycharacteristictimet=R0/

for=0.072N/m,=103kg/m3,R

0=0.5mmandS=65◦.(—)tS∝t1/5,(−−−)

tS∝t.

Fig.11.EffectofthecontactanglesSonthetimeevolutionofthedropbaseradius

rfor=0.01Pas,=0.072N/m,=103kg/m3andR

0=0.5mm.Thedropradial

sec-tionscorrespondingtoa,bandcareshowninFig.12.

ForthesmallestcontactangleS=5◦,weobserveafterthet1/2

evolutionacleart1/10evolutioncharacteristicoftheTannerLaw.A

similarbehaviorhasbeenreportedinthesimulationsof[11]where aregimeclosetotheTanner’slawisobservedwhenincreasing theviscositywhilearegimeclosetot1/2isobservedattheearly

stageofthespreading.Fig.12showstheshapeofthedropduring

thet1/2evolution,duringthet1/10evolutionandatthetransition

betweenthesetworegimes.Thecorrespondingtimesaredepicted bybulletsinFig.11.Asphericalcapisclearlyidentifiedduringthe t1/10evolution.

Inthelimitofaverythinsphericalcap,thecontactangleW,

thedropvolumeVandthecontactlineradiusrarelinkedbythe geometricalrelationW≈4V/r3.Inthelimitofasmalldynamic

contactanglewithS≈0,theCoxrelation(4)canbeexpressedas:

W3 =9 

dr dtln(L/ )

Combiningthesetworelationsandintegratingintime,theradius ofthecontactlineisfoundtofollowthefamousTannerlaw:

r(t)∼R0



t t



1/10 (13)

Forspreadingdrops,suchevolutionisthusobservedforsmall contactanglesoncethedrophasreachedasphericalcapshape asshowninFig.12c.InthelimitofS→0,thebaseradiusofthe

sphericalcapcanbeexpressedasrf≈R0−1/3S anditfollowsan

estimationofthespreadingtime:

tS≈t−10/3S (14)

For the simulations reported in Fig. 11, relation (14) gives tS≈0.2swhichisacorrectorderofmagnitudeforS=5◦.The

corre-spondingviscous–capillaryspreadingtimegivenbyrelation(12)is twoordersofmagnitudesmaller≈0.0038s,butinagreementwith theevolutionreportedforthelargercontactangles(S=65◦,115◦,

140◦and160◦).

4. Conclusion

Wehaveconsideredthespreadingofadroponahorizontal wet-tablesurfacebynumericalsimulation.Inournumericalmodelingof movingcontactlines,theapparentcontactangleisgivenbythe[8]

relationwhileazerosliplengthisimposedforthemomentum cal-culation.Thesimulationshavebeensatisfactorilycomparedwith theexperimentsof[15]forsqualaneandwiththeexperimentsof

[24]forwater.Theeffectsofthemainparameterscontrollingthe dropspreadinghavebeenstudiedseparately.Forviscosityofthe orderofthatofwater,thet1/2powerlawhasbeenrecoveredat

theearlystageofthespreadingforallthecontactangles consid-eredinagreementwiththeexperimentsof[24].Whenincreasing theviscosity,theearlystageofthespreading is still character-izedbya powerlawevolutionbutwithanincreasingexponent inagreementwiththeexperimentsof[2].Boththeviscosityand thecontactanglearethenshowntoimpactthelineexpansion.In particular,consideringthespreadingofaviscousdroponahighly wettablesubstrate(S=5◦),weobservetwosuccessivepowerlaw

evolutions.Thecapillary-inertialregimer∝t1/2isfollowedbythe

Fig.12.Dropradialsectioncorrespondingtothetimesa,bandcshowninFig.11(S=5◦).(a)inthemiddleofthet1/2evolution(t=4×10−4s),(b)thetransitionbetween

(10)

Tannerr∝t1/10evolutiononcethedroplethasreacheda

spheri-calcapshape.Oursimulationsshowthatthecombinedeffectof theviscosityandthecontactangleplaysanimportantroleinthe spreadingofadrop.Theymakeconsistentthedifferent experimen-talobservations(aprioricontradictory)concerningtheeffectofthe wettingconditions.

enddocument

References

[1]A.-L.Biance,C.Clanet,D.Quéré,Firststepsinthespreadingofaliquiddroplet, Phys.Rev.E69(016301)(2004).

[2]J.Bird,S.Mandre,H.Stone,Short-timedynamicsofpartialwetting,Phys.Rev. Lett.100(234501)(2008).

[3]D.Bonn,J.Eggers,J.Indekeu,J.Meunier,E.Rolley,Wettingaandspreading,Rev. Mod.Phys.81(739)(2009).

[4]T.Bonometti,J.Magnaudet,Aninterfacecapturingmethodfor incompress-ibletwo-phaseflows.validationandapplicationtobubbledynamics,Int.J. MultiphaseFlow33(2007)109–133.

[5]J.Brackbill,D.Kothe,C.Zemach,Acontinuummethodformodelingsurface tension,J.Comput.Phys.100(1992)335–354.

[6]A.Cazabat,M.CohenStuartt,Dynamicsofwetting:effectsofsurfaceroughness, J.Phys.Chem.90(1986)5845–5849.

[7]L.Chen,E.Bonaccurso,M.E.R.Shanahan,Inertialtoviscoelastictransitionin earlydropspreadingonsoftsurfaces,Langmuir29(6)(2013)1893–1898.

[8]R.Cox,Thedynamicsofthespreadingofliquidsonasolidsurfaces.Part1: viscousflow,J.FluidMech.168(169)(1986).

[9]P.-G.deGennes,Wetting:staticsanddynamics,Rev.Mod.Phys.57(827)(1985).

[10]H.Ding,E.Li,F.Zhang,Y.Sui,P.Spelt,S.Thoroddsen,Propagationofcapillary wavesandejectionofsmalldropletsinrapiddropletspreading,J.FluidDyn. 697(2012)92–114.

[11]H.Ding,P.D.M.Spelt,Inertialeffectsindropletsspreading:a comparison betweendiffuse-interfaceandlevel-setsimulations,J.FluidMech.576(2007) 287–296.

[12]J.Dupont,D.Legendre,Numericalsimulationsofstaticandslidingdropwith contactanglehysteresis,J.Comput.Phys.229(2010)2453–2478.

[13]B.Lafaurie,C.Nardone,R.Scardovelli,S.Zaleski,G.Zanetti,Modellingmerging andfragmentationinmultiphaseflowswithsurfer,J.Comput.Phys.113(1994) 134–147.

[14]E.Lauga,M.Brenner,H.Stone,HandbookofExperimentalFluidDynam-ics. Ch.19,in:Microfluidics:TheNo-SlipBoundaryCondition,Springer,New-York, 2007,pp.1219–1240.

[15]B.Lavi,A.Marmur,Theexponentialpowerlaw:partialwettingkineticsand dynamiccontactangles,ColloidsSurf.250(2004).

[16]M.Maglio,Numericalsimulationofspreading,slidingandcoalescingdropson surfaces,InstitutNationalPolytechniquedeToulouse,2012,Ph.D.thesis.

[17]A.Marmur,Equilibriumandspreadingofliquidsonsolidsurfaces,Adv.Colloid InterfaceSci.19(75)(1983).

[18]S.Popinet,S.Zaleski,Afront-trackingalgorithmforaccuraterepresentationof surfacetension,Int.J.Numer.MethodsFluids30(1999)775–793.

[19]R.Scardovelli,S.Zaleski,Directnumericalsimulationoffreesurfaceand inter-facialflow,Ann.Rev.FluidMech.31(567)(1999).

[20]J.Sethian,Levelsetmethodsandfastmarchingmethods,CambridgeUniversity Press,Cambridge,1999.

[21]M.Sussman,E.Fatemi,P.Smereka,S.Osher,Alevelsetapproachfro comput-ingsolutionsinincompressibletwo-phaseflows,J.Comput.Fluids27(1998) 663–680.

[22]L.Tanner,Thespreadingofsiliconeoildropsonhorizontalsurfaces,J.Phys.D: Appl.Phys.12(1979)1473–1484.

[23]O.Voinov,Hydrodynamicsofwetting,J.FluidDyn.11(1976).

[24]K.Winkels,J.Weijs,A.Eddi,J.Snoeijer,Initialspreadingoflow-viscositydrops onpartiallywettingsurfaces,Phys.Rev.E85(055301(R))(2012).

[25]S.Zalesak,Fullymultidimensionalflux-correctedtransportalgorithmsfor flu-ids,J.Comput.Phys.31(1979)335–362.

Figure

Fig. 2. Comparison with the experiments for viscous drop [15]. The normalized radius r/r f is reported versus the normalized time  = t/V 1/3
Fig. 3. Evolution of the contact line radial position r/R 0 as function of the nor- nor-malized time t/(R 3 0 /) 1/2
Fig. 5. Shape evolution of a water drop (radial section) for  S = 65 ◦ . The sequence is shown line by line for the times t = 0.06, 0.36, 0.56, 1.00, 1.40, 1.66, 1.86, 2.52, 2.70, 2.86, 2.98, 3.12, 3.28, 3.34, 3.52, 3.60, 4.80 and 5.80 ms.
Fig. 6. Shape evolution of a water drop (radial section) showing the pinch events for  S = 65 ◦
+3

Références

Documents relatifs

This model leads, after time discretization, to a Stokes system with non standard dissipative boundary conditions that cannot be easily and directly implemented in most FEM software,

(Left) A cathepsin B-green fluorescent protein chimera contain- ing the first 65 amino acids of truncated procathepsin B starting with position 52 (aa numbering of preprocathepsin

On suppose qu'on dispose du répertoire /opt/zimbra qui servira à l'installation de l'application ZIMBRA. On suppose que ce répertoire contient les répertoires bin et lib. 1) Définir

Keywords: Dynamics of discrete mechanical systems; Frictionless unilateral constraints; Microimpacts; Restitution coefficient; Event-driven scheme; Time-stepping

This study focuses on the numerical simulation of roller hemming of an aluminium alloy sample with a curved geometry; this non-planar geometry is obtained by a prestrain

Figure 5 ± Visualization of the (antenna) array pattern obtained by numerical simulation of the BRAMS interferometer in Humain.. The gain is normalized to the

5.6.. Par conséquent, une structure dynamique pour 1es représenter est 1ourde et vo1umineuse. On choisira donc de représenter 1es premiers niveaux sous forme d'un

Jahrhunderts: hatten die Savoyer oder mailand oder der Bischof von Sitten oder die Herrn von turn oder die Sieben Zenden oder die Herrn von Attinghausen oder der Hilfskreis