• Nenhum resultado encontrado

Braz. J. Phys. vol.31 número3

N/A
N/A
Protected

Academic year: 2018

Share "Braz. J. Phys. vol.31 número3"

Copied!
10
0
0

Texto

(1)

Surfae Waves and Wave Resistane

in Magneti Fluids

J. Browaeys y z

, R.Perzynski

,J.-C. Bari

, and M. I. Shliomis x

Reeivedon15February,2001

We present here the experimentaldetermination of the dispersion relation of surfae waves of a

ferrouidsubmittedtoasteadyvertialmagnetield. Theresultsareomparedtoalineartheory

whereall themagnetiharateristisoftheuidhavebeentakenintoaount. Theknowledgeof

thedispersionequationisthenusedtoanalysetheapillary-waveresistane,thatisthedragfore

assoiatedtotheemissionofwavesbyamovingdisturbaneatafreeuidsurfae. Itundergoesa

transitionfromzerotoanitevalueasthespeedofthedisturbanereahesaertainritialvalue.

Theeetof visosity is explored, and amagnetiuidis shownto allow ontrolling theritial

speeds. Contrary to the theoretialmodel, however, themeasured wave resistane reveals a non

monotonispeeddependeneafterthethreshold.

I Introdution

Magneti uids, sometimes alled ferrouids, are

ol-loidal suspension of magneti partile whih have a

giant magneti suseptibility, usually in the order of

unity. When submittedto he inuene of anexternal

DCvertialmagnetield,oneisstrikenbythe

stru-turation of its surfae into peaks, usually distributed

on an quasi-hexagonal array. The linear analysis of

this phenomenon [1℄, alled after its disoverer the

\Rosensweiginstability"hasbeenthefoundationstone

offerrohydrodynamis[2℄,thethermo-hydrodynamis

of non-onduting magneti liquids. The uid peaks

arisebeauseapillary-gravitywaves,eitherinduedby

thermalutuationsorambientvibrations,are

ampli-edwhenasuÆientlystrongvertialmagnetieldis

present.

Thepreiseknowledgeofthedispersionequationis

ofparamountimportaneforanyproessthatis

sensi-tive tosurfaeinstabilities. Infat,everylassial

hy-drodynamial instability suh as the Rayleigh-Taylor,

or Benard-Marangoni instabilities [3℄, has a magneti

ounterpart with ferrouids. The additionalmagneti

eld,atingasanewontrolparameter,usuallyallows

toontrol thethresholdoftheinstability[4℄.

Therstpartofthispaperdealswiththe

measure-mentofthedispersionequationofapillary-gravity

sur-fae wavesofamagnetiuidunder theinueneof a

vertialandsteadymagnetield. Theresultsare

on-frontedtoalineartheorythattakesintoaountallthe

magneti harateristis ofthe ferrouidsample used.

The resultsare then used to analyse athreshold

phe-nomenon: theapillary-gravitywaveresistane.

When an objet is moved at the free surfae of a

uid,itexperienesadragforewhihphysially

orig-inatesfrom: (a)bulkdissipationinavisousboundary

layerfor lowReynolds numbers, and in the turbulent

wake for high Reynolds numbers; (b) the emission of

apillary-gravitysurfaewaves.Suhwavesremove

mo-mentum from theperturbatingobjetto innity. The

assoiated fore that the objet experienes is alled

wave resistane.

For the onvenient moderate speeds on whih we

fous in this paperit may overome the bulk

dissipa-tion type drag. Wave resistane hasbeen studied for

morethanaentury in theaseofpure gravitywaves

[5℄, mainly beausethis topi hasobviousnaval

appli-ations[6℄. Howevertheaseofapillary-gravitywaves

hasonlybeentheoretiallytreatedin areentworkof

Raphaeland de Gennes[7℄. The dispersiveproperties

ofsuhwavesaresuhasthereexistsaminimumphase

speedV

=(4g=) 1=4

atwhihwavesareableto

prop-agate(here isthesurfaetensionof thefreeair-uid

interfaeandthedensityoftheuid). Sinethewave

patternisstationaryintherefereneframeofthe

mov-ingobjet,nowaveanbeemittedforV <V

[8℄,and

therefore there is no wave resistanein that ase. As

modeledin[7℄,thewaveresistaneshouldexperienea

GroupeFerrouideassoieal'UniversiteDenisDiderot-Laboratoiredesmilieuxdesordonnesetheterogenes,UMRCNRS7603

UniversitePierreetMarieCurie,ase86,4plaeJussieu75252ParisCedex05,Frane.

y

LaboratoiredeBio-rheologieetd'HydrodynamiquePhysio-himique,ESACNRS7057UniversiteDenisDiderot,ase7056,2plae

Jussieu75251ParisCedex05,Frane.

z

E-mail:browaeysr.jussieu.fr

(2)

nite jump R

at V =V

andinreasesaboveV

. The

systemisthussupposedtoundergoadisontinuous

bi-furation.

The seond part of this work deals with the

mea-surementofthewaveresistane. Inordertohekthese

theoretialpreditionsitisneessarytovaryV

.

Vary-ingandisveryineetivewaytohangeV

(itwould

merely vary from 17 m/s to 23 m/sfor any kindof

uid). Wehereshowtheoretially andexperimentally

that theationofamagnetield onamagneti uid

providesameans totune theritial veloity from its

maximumvalueV H=0

=(4g=) 1=4

downto 0.

II Surfae waves

1.1 Dispersionequation

Cowley&Rosensweig[1℄werethersttostudythe

peakinstabilityatthefreesurfaeofamagnetiuid.

Although they didn't atually express the dispersion

equation for waves at the free surfae of a ferrouid

(theywereinterestedinthethresholdvalueofthe

mag-netieldatwhihthepeaksform),theydevelopedthe

linear theory whih allowsto diretlyderive the

mag-netiterminthedispersionequation. Theyonsidered

thattheferrouidwasinvisid,thatitsdepthwas

in-niteandthatitwasinnitelyextendedinthehorizontal

diretion.

Sine then, other authors rened this analysis

in-ludingmorerealistionditions: Zelazo&Melher[9℄

expressedthedispersionequationofsurfaewavesofa

ferrouidlayer restingbetweenthe poles ofamagnet.

More reently, Abou & al. [10℄ hanged this

bound-aryondition(thelayerisgenerallysurroundedbyfree

spae,for visualizationpurposes),andinludedthe

ef-fet of visosity. However, they omitted to take into

aount the demagnetization oeÆient and also

sup-posed that the permeability of the material was

on-stant,afatthatisnotsupportedexperimentally,sine

amagnetizationsaturationexistsinthematerial. This

shortoming hasbeenremovedin anartile ofthe

o-authors[11℄.

Inordertosimplify,onlytheirmainresultsare

pre-sentedhere. Intheaseofonstantpermeabilityforan

invisid uid of innite depth, the dispersion relation

writes:

! 2

=gk+ k

3

0 H

e2

k 2

2

(1+)(2+)

; (1)

where!isthepulsation,kthewavevetor,gthe

grav-itational aeleration, the surfaetensionofthe

fer-rouidlayer,itsdensity,

0

thevauumpermeability,

H e

theexternalvertialmagnetieldandthe

mag-Figure 1. Dispersion relation of surfae waves in a

fer-rouid. Here!

=2(g

3

=) 1=4

is introdued tonormalize

the urves. It orresponds to the pulsation when k = k

withoutmagnetield.

Therst twotermsof theright sideof eq. (1) are

lassial and respetively aount for the gravity and

apillarity eet. The last term inludes the eet of

themagneti eld. It is a destabilizingterm : at the

ritialeld H

, the dispersionurve(Fig. 1) adjoins

the zero frequeny axis at a non zero wavevetor k

.

Analytialderivationyields:

H

=

s

2(1+)(2+)

2

p

g

0

(2)

k

=

r

g

(3)

It is noteworthy that the wavevetor at the

thresh-oldoftheRosensweiginstabilityisindeedtheapillary

wavevetor.

Inthemorerealisti asewherethe uidlayerhas

nite depth, is visous andits magneti suseptibility

is eld dependent, the general piturepreviously

pre-sentedisonlyslightlymodied.

Inluding visosity in the previous equations

ren-dersthemveryomplex,andasweshowfurtheron,the

termsaounting for visosityin the dispersion

equa-tionanmostoftimebeomitted. Itispossibletodene

twoReynoldsnumbers[10℄, either basedonthedepth

hoftheferrouidlayer,orbasedonthewavelength:

R e= !h

2

; or R e= !

k 2

: (4)

Theinueneofvisosityontheowgeneratedby

sur-faewavesisnegligibleprovidedtheReynoldsnumber

isgreaterthan10[12℄. Thisleadstotwoonditionsfor

thepulsation! forshortandlongwavelengths:

!>sup

10

h 2

;10k 2

: (5)

InFigs. 5and6,theareaoftheparameterswherethe

visosityannotbenegleted(aordingtotheprevious

equation)has been shaded. Most of theexperimental

pointslieoutsidethisregion.

Finally,itisimportanttonotethat theexperiment

isrealizedinylindrialgeometry: theproduedwaves

(3)

equa-observed wavelength is smaller than theradius of the

ferrouidontainer, it is possibleto neglet the eet

ofgeometry. Tobemorepreise,theBesselJ

0

funtion

is the basis for the ylindrial apillary-gravity wave

dispersionequation. Ifthe distane between two

on-seutiverests is assumed to be theplane-wave

wave-length, then the error is 11 % for the rst tworests

(startingfromenter),and thenonly4%for thenext

two(anddereasingastheradiusinreases). Sinethe

radius of theontaineris 9.6 m, thewavelengths are

measured in at least within a 11 % approximation if

they are smaller than 4.8 m, whih orrespond to a

limitingwavevetorof 1.3 m 1

. Theomplete

alu-lationofthedispersionequationinludingthemagneti

terminaylindrialgeometryisatheoretialproblem

that has yet to be solved; it might yield a better

un-derstandingoftheeetofnon-periodimagnetield

perturbations.

I.2 Experimentalsetup

I.2.1 Charateristis of the ferrouid sample

Weuseanioniferrouidsynthesizedaordingto

Massart's method [13℄. It is omposed of a olloidal

suspensionofobaltferritepartilesinwater. Its

den-sityis1560kg/m 3

andthededuedvolumefrationof

partiles is 14 %. Its surfae tension with air,

are-fullymeasuredwithaKrussK10Tringtensiometer,is

71.4mN/mat20C.Thisvalueisveryloseto the

sur-fae tensionofwater(73.2mN/m), in agreementwith

the fat that our ioni ferrouidis free of surfatant.

The dynami visosity of the sample, measured with

a Poiseuille visometer, is 20 mPa.s whih leads to a

kinemativisosityof1:2810 5

m 2

/s.

Figure2.Magnetizationurveoftheferrouidsample. The

bestleastsquaretwithequaton(6)givesA0=26.15kA/m

and A1 =3.63 kA/m. inset : same urvesfor a more

ex-tendedmagnetield range. TheLangevinurvedoesnot

tthewhole magnetizaionurve.

Themagnetization urvehasbeenobtainedbythe

use of a alibrated uxmeter (Fig. 2). Beause our

ferrouid is highly onentrated and polydisperse, it

isunlikelytofollowalassialLangevinparamagneti

rangeofmagnetield amplitudeofourinterest(from

0toabout12kA/m, theeld aroundwhih thepeaks

instabilitydevelops) :

M A

0

(oth(H i

=A

1 ) A

1 =H

i

): (6)

Here M is the magnetization and H i

is themagneti

eldinsidetheferrouid: H e

=H i

+M. Itshouldbe

noted that the A

0

value obtained is a priori dierent

fromthemagnetizationsaturationoftheferrouid

sam-ple. TheknowledgeoftheparametersofthetA

0 and

A

1

allows us to ompute all the magneti parameters

oftheproblem.

Figure 3. Eetive magneti suseptibility

e

as a

fun-tion of the external magneti eld H e

for diernet

non-dimensinalwavevetorskh.

Eq. (1) represents the dispersion equation when

magnetipermeabilityisaonstant,whihisobviously

not thease here. Following [11℄, it is possibleto

de-ne an eetive magneti suseptibility

e

, funtion

ofthemagnetield andthenon-dimensionalprodut

kh, andomputed sothat eq.(1) isorret. It isalso

ameasurementof the disrepany when themagneti

termofeq.(1) ismisused. Thefollowingplot(Fig. 3)

showsthat the value of

e

is almost independent of

the magneti eld thus justifying a posteriori the use

of a simplied magneti term (i.e. onstant) ; on

the other hand it shows that the dependene of the

magneti termon thethiknessof the layerhasto be

aountedfor.

I.2.2 Experimentdesign

A irular Teon dish of 20 m in diameter that

ontains the ferrouidis plaed between two

horizon-taloils;theyproduethestativertialmagnetield

and are arranged suh as to ensure a horizontal

spa-tialhomogeneitymoreauratethan99%. Itshouldbe

(4)

Figure4. Shadowgraphmethod.

The shadowgraph method [14℄ is a very sensitive

optial method to detet the wave amplitude(60 m

amplitude ripples arestill visible). Theferrouid

sur-faeisilluminatedbyaparallellightbeamomingfrom

apoint-likesoure(anoptialberilluminator)at the

foaldistaneoflensL1. TheameralensL2isplaed

in order to provide a parallel light beam to the CCD

detetor(seeFig. 4). ThedistanedbetweentheCCD

andtheameralensissetsuhastofousjustbeneath

orabovetheferrouidsurfae. Theurvedpartsofthe

surfae at asvirtual lenses, and appear lightened or

darkenedonthevideosreen whethertheurvatureis

positiveornegative.

To produesurfaewaves, asmall sinusoidal

mod-ulationinthevertialeld isapplied. Propagative

ir-ularwavesareemitted by theedges due to magneti

foresinthemenisus. Beauseofthevisousdamping

[15℄, no stationary wavesare observed, exept in the

enteroftheferrouidontainer.

This method is very reliable for the prodution

of waves in the interesting frequeny range (3 Hz

-25 Hz). The undulating omponent is merely 0.5 %

of thethresholdeld H

. It ishoweverdiÆultto

ex-trapolatetheresultsat eldslowerthan0.1H

where

theundulationratiois already5%.

I.3 Results

Inordertorepresentexperimentalresultsandtot

parametersofthetheoretialwavedispersionequation,

wehose twoutsinthe(k;!;

0 H

2

)spaeof

parame-ters (Fig. 5and Fig. 6). Theonlyparameterallowed

to vary is the surfaetension, beause of its

sensitiv-ity to ontaminants. All other parameters (the

mag-ferrouid layer) are set to their previously measured

values. Wehavedrawn onbothFig. 5andFig. 6the

parameterdomain where the eets of visosity ould

notbenegleted: ourexperimentalpointslieoutsideof

thiszone.

0

2

4

6

8

10

0

5000

10000

15000

20000

25000

0

1

2

3

0

1000

2000

3000

k (cm

-1

)

ω

2

(s

-2

)

H/H

c

= 1

H/H

c

= 0.79

H/H

c

= 0.63

H/H

c

= 0.48

H/H

c

= 0.31

H/H

c

= 0.16

Figure5. Experimentaldispersion equation inthe (k;! 2

)

planefordierentmagnetields.Theblaklinesrepresent

the theoretial dispersion equations with = 60 mN/m.

The dotted line is the theoretial dispersion equation at

the threshold eld. The shaded area represents the

fre-quenydomainwherevisosityeetsbeomenonnegligible

(R e<10).

0

1

2

3

4

5

6

7

8

9

0

50

100

150

200

k (cm

-1

)

µ

0

H

2

(J.m

-3

)

3 Hz

4 Hz

5 Hz

6 Hz

7 Hz

10 Hz

15 Hz

20 Hz

Figure 6. Dispersion equation in the (k;

0 H

2

) plane for

dierent frequenies. The blak lines represent the

theo-retial dispersionequation with =60mN/m;the thik

dotted line is the theoretial marginal urve of stability

(null frequeny) ; the thin dotted lines loate the

mini-mum of the marginal urve and read k = 5:1 m 1

and

HC = 11.2 kA/m. The shaded area represents the

fre-quenydomainwherevisosityeetsbeomenonnegligible

R e<10). Forlarity purpose, onlyafew error barshave

beenplotted.

Thetforanyvalueof!andH leadstoanaverage

surfaetensionof60mN/m,whih is16% lowerthan

thearefullymeasuredvalueobtainedfrom aring

ten-siometer. Thenaturalontaminationof thesurfaeby

atmospheridustbehavingasasurfatantmayexplain

this disrepany. We thus perform another

tensiome-ter measurement by letting the ferrouid surfae rest

in anunpuried atmosphere: after afew minutes, the

(5)

of any ontaminant, the tted valuehas to be

under-stoodasanin situ value,ompatiblewith tensiometer

measurements.

Inarstexperimentthemagnetieldissetto

dif-ferentvaluesandthewavevetormeasuredasafuntion

offrequeny. Theexperimental dispersionispresented

in Fig. 5. Thets are satisfatory,although it is

dif-ult to make any assertion when the wavevetor is

smaller than 3 m 1

. The ritial eld is estimated

from the ts: as we will see later on, it is extremely

diÆulttodeneapreisevalueofthethreshold

mag-netieldH

inourpartiularsystem. Themaximum

experimentedvalueoftheratioH=H

isvoluntarily

lim-itedto0.79beauseabovethiseld,peaksformatthe

boundary of the ferrouid vessel; during the time of

the experiment, the uid in thepeaks would dry and

beomelumpy.

Thisiswhyit isbetterto setthefrequenyto

dif-ferent xed values, and measure the wavelength as a

funtion of themagneti eld. The orresponding

re-sults are presented in Fig. 6. This plot is more

dis-riminating, espeially in the low wavelength domain

(k <3m 1

). Here it an beseen that the t is not

ompatiblewiththeexperimentaldataespeiallyinthe

regionwherethedispersionurveisnonmonotoni.

Figure7. Spontaneousstatideformationbelowtheritial

eld(H=H=0.79).

Inthis region, where the magneti eld is lose to

H

,itseemsthatanunexpetedstatiirularpattern

develops(Fig. 7). It atually beomes visible

(ampli-tudeofthedeformationexeeds60m) assoonasthe

magnetieldexeedshalfofthethresholdvalue. Only

thepreisionoftheshadowgraphmethodallowstosee

suhadeformationofthesurfae. Thepatternis

inde-pendentofthemagnetield,onlytheamplitudeofthe

deformationinreaseswiththemagnetieldintensity.

It ould be understood as apre-transitionnal eet of

theRosensweig instability, perhapsdue to the limited

horizontalextensionofthelayer.

Unfortunately it renders the measurement of the

wavelength more unertain, espeially lose to the

threshold: that is the drawbak of the shadowgraph

method. It is diÆult to preisely assess the value

of thethreshold magneti eld H

. Besides, a

experi-mentwithdiretvisualizationofthesurfaeshowsthat

thepeaksdon'tappearsimultaneouslyonthesurfae:

when theeld is inreased to thethreshold value, the

peaksgrowinwardfrom theedgestotheenterof the

ell(they start appearingon theedgesat lowerelds,

beauseoftheeldgradientreatedbythepreseneof

an edge). A preise estimation of the threshold

mag-neti eld is dependent on the hoie of a threshold

riteria. Wehooseto keepthe valueof themagneti

eld above whih no image is visible anymore. This

method providesameasurement ofthe thresholdeld

of11.5kA/m whih ismerely3%abovetheexpeted

valueobtainedfromthetsin Fig. 5andFig. 6.

Thenowpreise knowledge ofthedispersion

equa-tion of surfae waves in a ferrouid submitted to a

steadyvertialmagnetield allowsusto foreastthe

inuene ofthis magnetield in otherhydrodynami

proessesthatinvolvesurfaewaves.

III Wave resistane

II.1 Ordinary uids

II.1.1 Theory

Inaregularuid,thewaveemissionproessis

on-trolled bythedispersionequationfor apillary-gravity

surfaewaves,

! 2

=gk+k 3

=; (7)

where!isthepulsationandkthemodulusofthewave

vetor. AsalreadypointedoutbyKelvin[5℄,thewave

pattern is stationary in the frame of referene of the

movingobjet.

Thedispersionrelationof surfaewavesis valid in

the frame ofreferene K in whih theuid is at rest,

andallphysialvariables(speed,pressure,...) are

pro-portionnaltoexp(i(k:r !t)). Intheframeofreferene

K 0

whihtranslatesatspeedV ,thosevariablesarethen

proportionalto exp(i(k:r 0

! 0

t))wherer 0

=r V tis

the position vetorin K 0

and ! 0

= ! kV the

pul-sation in K 0

(one reognizes the Doppler eet). The

stationarityonditionimplies! 0

=0and thus:

!=kV os; (8)

where is the angle between the speed and the wave

vetor.

Combining(7)and(8)weobtainthefollowing

equa-tion(V

=(4g=) 1=4

theminimumphasespeed):

k

k

2

2

V

V

os

2

k

k

+1=0; (9)

whihhasnosolutionforV <V

. Thusnowavepattern

istodevelopifthespeedofthemovingperturbationis

(6)

ForaDiraDeltapressuredistributionP(x;y;t)=

pÆ(x Vt;y)movingalongaxisx,thewaveresistane

is[7℄:

R= p

2

Z

aros V

V

0

os k

+ ()

2

+k () 2

k

+

() k ()

d; (10)

where k

+

() and k () are thetwo roots of eq. (9).

Thisformularemainsvalidaslongastheharateristi

sizeofthepressuredistributioninexperimentsismuh

smallerthan theapillarywavelength. Inthose

ondi-tions, aording to eq.(10), the waveresistanetakes

a nite value R

=p

2

k

=2

p

2 just abovethe

thresh-oldandinreasesmonotoniallywithspeed(seethefull

line inFig. 12andtheuppermosturvein Fig.15).

II.1.2 Experiments

In order to measure R as a funtion of speed V

for variousuids, weuseairularhannelduginto a

Teon overedaluminum dish. The latter is xed to

ashaftandrotatedatonstantrate,thussimulatinga

steady owforthe uid. Theradiusof the hannelis

20m, itswidth is2m.

Thedisturbing objetonsists of avertialbronze

wire (diameter d=0.2mm) moving over theuid

sur-fae. Thewireiswettedbyafewtenthsofmillimeters

ofuid. Thedeetionofthewireisproportionaltothe

horizontal fore exerted onits free end (whih is

typ-ially in theorder of amironewton). It is measured

withaninfraredoptialsensor(Fig.9). Thealibration

of the sensor is obtained by tilting the base to whih

thewireisattahed.

Figure 8. Waveresistane measurementsetup : uidow

Displacement screw

Wire

Optical sensor

Fluid

(top view)

Wire

(side view)

Infrared LED

Phototransistor

Figure9. Waveresistane measurement setup: fore

mea-surement. ThelightemittedbytheLEDisreetedbythe

wireandolletedintothephototransistor.

Beausenotheoryinludes3Dvisouseetssofar,

wemeasurethewaveresistanefordierentvisosities.

Tothispurposeseveralmixturesofwaterandglyerol

areused: the surfaetension and thedensities of

themixturesareverylosetooneanother(seetable1)

sothattheimpatofvisosityalonemaybemonitored

inourexperiments. Thevisositiesaremeasuredwith

astandardPoiseuillevisometer.

Figure10. ExperimentaldragRexp as afuntionofspeed

V fortwowater-glyerolmixtures.

Fig. 10 displays the variationof theexperimental

drag R

exp

as afuntion of speed for twotypial

mix-tures. Allthemeasurementsareobtainedbyinreasing

and then dereasingthespeed: there is no hysteresis.

Wemaynotethat:

(a)Thereisaritialveloityatwhihthemeasured

dragdrastially inreases. Cameraimagingoftheuid

surfaeshowsthatthesharpdraginreaseoursatthe

samespeedat whih thewavepattern develops. The

(7)

wave-veloity V

is 23 0.5 m/sfor pure water. It

orre-spondstoasurfaetensionintervalof65.1-77.7mN/m

into whih lies the tabulated value of pure water

sur-fae tension72.75 mN/mat 20 Æ

C.Forwater/glyerol

mixturesweobtainV

22.5m/s,thatisompatible

withinexperimentalerrorbarswiththesurfaetension

ofthemixtures(around70mN/m ).

Figure11. Experimental dragRexp asafuntionof speed

V fordierentwater-glyerolmixtures. Thespeedremains

subritial (V <V) and only the visousontribution to

the total drag is measured. The plain line represents the

phenomenologialtheory(eg. eq.11).

Figure12. WaveresistaneR=R

exp R

vis

asafuntion

ofV. Fullline: theoretialexpressionfromeq.10.

(b)Theexperimentaldragisnotnullbelowthe

riti-alveloity,allthemoresinethevisosityishigh. The

visous drag R

vis

that is exerted overthe immersed

wire must be added to the wave resistane R to

a-ountforthe measureddrag R

exp

. Sinethelengthof

thewettedpartofthe wireis omparableto its

diam-proportionaltothatofasphereofthesamediameterd.

ForthemoderateReynoldsnumbersofourexperiment

(0:5<R e<80)thevisous dragmaybedesribed by

anempirialformula[16℄:

R

vis

=3Vd 1+0:15

Vd

0:687 !

; (11)

based on numerous sphere drag experiments. In the

previousexpression representsthevisosityofthe

so-lution, itsdensityand theaspetratio introdued

by usto makeallowaneforadistintion of ourshort

ylinder from a sphere(for thelatter, =1[16℄). A

simultaneousttingproedure onthefourR

exp urves

for subritial speeds(V <18 m/s) yields = 0:77

(seeFig.11). Fig.12presentstheR (V)variationsafter

subtration of thevisous drag R

vis

for eah sample.

It is this quantity that has to be ompared with the

theoretial expression (10)(full line). It seems that a

pretransitionaleettakesplae,asthemeasureddrag

inreasesjust below thethreshold(the higherthe

vis-osity,thestrongertheeet). Areentmodel[17℄for

2Dvisouswaveresistanepreditssuhafeature.

() The amplitude of the wave resistane inrease

at V

isomparable to the theory. Assumingaperfet

wettingof thewireby theuid, thetotal foreating

on theuid is p= 2r (r is the radiusof the wire).

Thusanestimateofthewaveresistaneinreaseatthe

thresholdis givenbyR

=

2

r 2

p

2g. Aomparison

betweenexpetedvaluesandwhatis observedisgiven

in table 1. Thedisrepanyispartially dueto the

im-perfetwetting oftheuidonthewire,whih leadsto

overestimate the applied vertial fore. On the other

hand thedrag valueslosetothethresholdutuatea

lot.

(d) The wave resistane is a non-monotoni

fun-tion of speed for V > V

. In fat, it an be seen in

Fig.12thatfor V >V

thewaveresistaneR rst

de-reasesasthespeedinreases,andtheninreasesagain

forhighenoughspeeds. Thisfeatureisnotpreditedby

theurrenttheory,whihanywayoverestimatesthe

a-tualdrag. Visositydoesnotseemtoinuenethewave

resistanederease. Oneouldalsoimaginethatasthe

wavepatterndevelops,theimmersiondepthofthewire

oulddrop,thusreduingthevisousdrag. Thisis

un-likelytooursinewedonotobserveanysalingofthe

dragredutionwithvisosity. Suhanon-monotoniity

ispossiblyageneralfeatureofapillary-gravitywave

re-sistane, andin this asethetheoryshould berevised

(8)

Table1: Experimentaldragdisontinuityatthethresholdomparedtothetheoretialpreditionsof[4℄,forvarious

waterglyerolmixturesandanaqueousmagnetiuid(MF).

Glyerolmassfration(%) 60 44.5 30 0 MF

Visosity(mPa.s) 12.5 5.1 2.6 1.0 7.0

Density(g/m 3

) 1.16 1.13 1.09 1.00 1.56

Theory(N) 3.9 3.8 3.8 3.7 4.2

Experiene(N) 2.9 2.6 4.0 3.6 4.0

Unertainty(N) 0.3 0.3 0.4 1.8 1.0

IV Magneti uids

Inamagnetiuidthedispersionequationof

apillary-gravitysurfaewavesis modied with allowanefor a

vertial uniform magneti eld, as shown in eq. (1).

For agivenwavevetor,aninreaseoftheeld

inten-sitylowersthefrequeny ofthe waves. Thefrequeny

dropsto zerowhen H reahestheritialvalueH

for

whihtheRosensweiginstabilityistriggered.

Theondition for stationarityimplies that k must

beasolutionof:

k

k

2

2

V

V

os

2

+

H

H

2 !

k

k

+1=0:

(12)

Realsolutionsexists ifandonlyif[18℄:

V >V H

with V H

=V

s

1

H

H

2

; (13)

thereforeasteady vertial magnetield should allow

thetuningof theritialveloityat whihwaves(and

waveresistane)appear.

Thewaveresistane,followingeq.(10)andeq.(12),

isgivenbytheintegral:

R H

(V)= p

2

k

Z

aros(V H

=V)

0

os

2B(V;) 1

2

+B(V;) 1

2

d; (14)

where B(V;)=

V

V

os

2

+

H

H

2 !

2

1: (15)

d

Just abovethe threshold, the wave resistanehas the

nite value:

R H

=

p 2

k

2 p

2 1

H

H

2 !

1

2

: (16)

An experiment is onduted using a water based

magneti uid synthesized aording to the Massart

method [13℄. Its ritial eld H

is 9.15 kA/m and

its surfaetensionof 60mN/mdoesn'tdepend onthe

magnetield. Otherarateristisaregivenintable1.

The ritial value V H

is experimentallyestimated

astheveloityforwhih asuddeninreaseofthedrag

is observed. R H

is the amplitudeof suh an inrease.

Both are plotted versusthenormalized magnetield

H=H

inFig.13andFig.14,andareomparedto

the-Figure13. ReduedritialspeedV H

=V

H=0

atwhihwave

resistaneappearsinfuntionoftheappliedredued

mag-netieldH=H. Thestraightlinerepresentsthe

(9)

Figure 14. Dragat threshold R H

as a funtionof the

re-dued magneti eld H=H. The full line represents the

theoretiallawasgivenbyeq.(16). Thereisnoadjustable

parameter.

Figure15. WaveresistaneR H

=R H

exp R

H

vis

asafuntion

ofreduedspeedV=V H

fordierentreduedmagnetields

H=H.Thetheoretialurvesarederivedfromeq.(14,15).

Theuppermosturvedesribesthewaveresistaneofa

reg-ularnon-magnetiuid.

Thetheoretialexpression (13)of V H

(Fig. 13)

re-markably ts the data points | note that there are

noadjustableparameters. Adatapointliesoutsidethe

urve,butthisisprobablyrelatedtoanimperfet

mag-netiwettingphenomenon. As themagnetield gets

loser to the peak instability threshold value H

, the

uid\limbs"onto thewire,produing amuhhigher

visousdrag, asituationwhihgets awayfrom our

in-visidlineartheoretialanalysis. Thisalsoexplainsthe

disrepanyinFig.14betweenexperimentaland

theo-retialR H

values. Wedonotaountfortheforethat

themagneti eld is exertingat the menisuslose to

the wire. Indeed,the very shapeof the menisus

re-atesanonhomogeneousmagnetieldwhihresultsin

aforethatsuksthemagnetiuidupandhangesthe

shapeofthemenisus. Onlyadvanednumerial

simu-lationswouldallowtoomputethenetforeaddedand

[19℄.

Inorder to ompute the wave resistane from the

visous drag R drag

usingeq. (11). This time, beause

theimmerged lengthof thewiredependson the

mag-netield intensity,eahurveatsubritialspeedare

separatelytted. Fig. 15presentstheresultsobtained

for dierent magneti elds in a redued

representa-tionR H

=R H

=f(V=V H

) withR H

=R

exp R

vis . It

alsogivesaomparisontothetheoretialpreditionsof

eq. (14). As it waspointedout about regularvisous

uids,thetheoretialvariationsofR =R H

lieabovethe

data points, exept for H H

. Then the

experi-mental dataand thetheoryareveryomparable. The

presenttheoretialdesriptionthusgivesaorret

gen-eral trend for the inuene of the eld on the wave

resistane.

V Conlusion

We havedesignedashadowgraph experimentin order

tomeasurethedispersionequationofwavesatthefree

surfaeofamagnetiuidsubmittedtoavertial

mag-neti eld. The model inludes the spei magneti

behaviorof theferrouidused, sensibleboundary

on-ditionsandtheeets oflimitedthikness. The

agree-ment between theory and experiene is quite

satisfa-toryexeptintheregionwherethedispersionequation

isnon-monotoni.Wehaveusedthisknowledgeinwave

resistanemeasurements.

Adragdisontinuityisalwaysobservedataritial

veloityV

. Thanksto amagnetiuidtheritial

ve-loityrangeisexperimentallyextended. Inallasesthe

measuredritialveloitiesandtheritialvaluesofthe

resistaneareingoodaordanewiththedevelopped

model. Ifaninvisidtheoryisorretatthethreshold,

therearesomedisrepaniesforV >V

suhasa

non-monotoni behavior of the waveresistane. Visosity

andnonlinearaspetsshouldbetakenintoaountin

further works. Finally, in order to get rid of the

vis-ous drag that is always present in our experiments,

another mode of disturbane is envisaged, suh as a

small magnet plaed just above the free surfae of a

owingmagnetiuid.

Aknowledgments

WewishtothankJ.ServaisandP.Lepertfortheir

tehnial assistane, O. Cardoso and C. Flament for

the help provided in theoneptionand thetuning of

theexperimentalsetup,S.Neveuforprovidinguswith

theferrouidsample,O.Sandreforgrantingusaess

to an aurate tensiometer. Fruitful disussions were

madewithB.Abou,R.E.RosensweigandE.Raphael.

Referenes

(10)

[2℄ R. E. Rosensweig, Ferrohydrodynamis (Dover,

Mine-ola,1997).

[3℄ S. Chandrasekhar, Hydrodynami and hydromagneti

stability(Dover,1981).

[4℄ J.WeileppandR.Brand,J.Phys.II2,419 (1996).

[5℄ LordKelvinPro.R.So.LondonA42,80(1887).

[6℄ A.A.Kostyukov,TheoryofShipWavesandWave

Re-sistane(EetiveCommun.In.,IowaCity,1968).

[7℄ E.RaphaelanddeP.-G.deGennes,Phys.Rev.E53,

3448 (1996).

[8℄ J. Lighthill, Waves ind Fluids, Cambridge University

Press, Cambridge,1996.

[9℄ R. E.Zelazo andJ.R. Melher, J.FluidMeh.39, 1

(1969).

[10℄ B. Abou, G. Nron de Surgy and J. E. Weisfreid, J.

Phys.IIFrane,7,1159 (1997).

[11℄ J.Browaeys,J.-C.Bari,C.Flament,S.NeveuandR.

Perzynski,Eur.Phys.J.B9,335(1999).

[12℄ F.H.Leblond&F.Mainardi, AtaMehania68,203

(1987).

[13℄ R.Massart, IEEETrans.Magn.17,1247(1981).

[14℄ S.Rasenat,G.Hartung,B.L.WinklerandI.Rehberg,

Experimentsinuids7,412(1989).

[15℄ Thewavesaredampedbuttheirfrequenyremains

un-aetedbyvisosityattherstorder.

[16℄ R.W.FoxandA.T.MDonald,Introdution toFluid

Mehanis(JohnWileyandSons,NewYork,1994).

[17℄ D. Rihard and E. Raphael, Europhys. Lett.48, 53

(1999).

[18℄ Stritly speaking, a real solution exists provided

V=V

>

p

1 H 2

=H 2

=josj, where is the angle

between the wave vetor of the emitted plane wave

andthediretionofthespeed.Wavesarethereforenot

emittedisotropially.Theintervalofpossibleanglesis

givenbythepreedingequation,orbytheboundaryof

theintegral ineq. (14). Theabsolute minimum value

isobtainedfor =0.

[19℄ A. G. Boudouvis and L. E. Sriven J. Magn. Magn.

Imagem

Figure 1. Dispersion relation of surfae waves in a fer-
Figure 2. Magnetization urve of the ferrouid sample. The
Figure 4. Shadowgraph method.
Figure 7. Spontaneous stati deformation below the ritial
+5

Referências

Documentos relacionados

O culto ao corpo no Brasil vem crescendo gradativamente após a liberdade corporal feminina alcançar emancipação a partir do processo civilizador da sociedade, que ao

O Pórtico ensina, na figura do sábio Crisipo, que a justiça não deve ser compreendida como uma lei ou convenção, mas sim, como uma vontade da natureza, porém ela também pode

Assim como em outras comunidades atendidas pelo AfroReggae, identificamos que as áreas mais pobres e de maiores privações estão geograficamente localizadas nas áreas de

This project explore the possibility of revising the dominant definition of modernism, based on the following hypothesis: the modernisms of Southern Europe have affirmed

Sistemas de Processa- mento de Transações Sistemas de Controle de Processos Sistemas Colaborativos Sistemas de Apoio às Operações Sistemas de Informação Gerencial Sistemas de Apoio

A empresa vencedora poderá retirar o instrumento equivalente (Nota de Empenho) na Reitoria do IFRJ à Rua Pereira de Almeida, 88 - Praça da Bandeira - Rio de Janeiro/RJ. A

Os autores trazem também que outros estudos demonstraram que a avaliação da mortalidade em relação à população de risco variaram entre 20 e 30 óbitos por

Logo após controlada a hipertensão e o aumento da hemoglobina, e se recomendado o tratamento com alfaepoetina, a sua administração somente deverá ser reestabelecida com baixas