Farouk 2010 Advances in Heat Transfer

ADVANCESINHEATTRANSFERVOL.
42
AcousticWaveInducedFlowsandHeatTransfer
inGasesandSupercriticalFluids
BAKHTIERFAROUK
*
,YIQIANGLIN,andZHIHENGLEI
DepartmentofMechanicalEngineeringandMechanics,DrexelUniversity,
Philadelphia,PA19104
I. Introduction
Acoustic waves can be generated in a compressible media by various
sources. Usually acoustic waves are generated by mechanical vibrations,
such as wall/surface vibration or shaking of a chamber. They can also be
createdbythermalperturbations,suchasrapidheating/cooling,combustion,
byspecialarrangementofheatsourcesandsinks,andsoon.Inthischapter,
wefocusontheacousticwavesinfluidsthatareinducedeitherbywall/surface
vibrationsorbytherapidheatingofaconfiningsurface,andtheassociated
convectionandtransportphenomenaunderdifferentgravityconditions.
It is well known that sound sources may generate an acoustic field in
which the particle velocities are not simply sinusoidal, and a pattern of
time-independent vortical flows or steady circulations is often found in the
body of compressible media. These second-order steady flow patterns are
known as acoustic streaming, which always have rotational characteristics.
Thestreamingvelocityincreaseswiththesoundintensity.Acousticstream-
ing may be effective in accelerating certain kinds of rate processes and
has applications in localized micro-mixing, convective cooling, acoustic
compressors, micro-fluidic devices, acoustic levitation, ultrasonic medical
diagnostic devices and ultrasonic cleaning of contaminated surfaces.
Acoustic streaming also attracted a lot of attention in the past due to its
effectontheheattransferenhancement.
Whenaconfiningboundarywall/surfaceissubjectedtorapidtemperature
increase,thefluidintheimmediatevicinityoftheboundarygetsheatedby
conductionandtendstoexpand.Thesuddenexpansionofthefluidduetothe
energyinput,constrainedbytheinertiaoftheunperturbedmediacreatesa
localpressuredisturbance,whichthenleadstotheproductionofapressure
*Current address: Department of Mechanical Engineering and Mechanics, Drexel University, Philadelphia, PA 19104.
AdvancesinHeatTransfer 1 Copyright2010ElsevierInc.
Volume42ISSN0065-2717 Allrightsreserved
DOI:10.1016/S0065-2717(10)42001-8
2 B.FAROUKETAL.
wave.Thepressurewavegeneratedfromthehotwallimpingesontheoppo-
site wall, and is reflected back. The wave repeatedly traverses between the
walls,anditsamplitudeeventuallydampsoutduetotheviscousandthermal
losseswithinthefluidandwallboundary[1].Thefluidvelocitiesinducedby
theacousticwavemaybesufficienttocauseaconvectiveheattransfermode.
Consequently, the fluid is heated more rapidly than it would have been by
pureconduction.Innormalgravityconditions,thevelocitiesgeneratedbythe
acousticwavesmayalsoaffectthedevelopmentofbuoyancy-inducedflow,
because the velocities in both modes are expected to be of the same order.
Hence,theinteractionbetweenathermallyinducedacousticwaveandgravity
createsaninterestingproblem,particularlyatreducedgravityenvironments.
Previousstudiesshowthatthoughtheheattransfereffectsofsuchwaves
are usually not appreciable under standard conditions, they may be very
significant when other modes of convection are weak or absent, such as
under zero-gravity conditions, or when the fluid is close to its thermody-
namic critical point [2]. Because of the high density and compressibility
character of the high-pressure fluid, strong thermally induced acoustic
waves can be produced and heat transfer effects of these waves become
significant even when the system undergoes a small temperature variation.
The thermally induced acoustic waves may also cause unwanted distur-
bances in static processes like cryogenic storage systems, which involve
rather weak diffusive and convective transport of heat, especially in a
reduced-gravityenvironment.
Thethermal-acousticinteractionsarefoundinmanynaturalphenomena
and industrial applications, such as thermoacoustic refrigerators and
engines. All of these and other applications involve mechanically or ther-
mallydrivenacousticfieldsinchamberswithvariousmedia.Asoundwave
inacompressiblefluidisusuallyregardedasconsistingofcoupledpressure
and motion oscillations, but temperature oscillations are also present. For
instance,whenthesoundtravelsinsmallchannels,oscillating heattransfer
occurs along the channel walls. For intense sound waves in pressurized
gases, the acoustic energy can be harnessed to produce powerful engines,
pulsatingcombustion,heatpumps,refrigerators,andmixtureseparators.
The interactions between a standing wave and a stack (thin low-
conductivity parallel plates) or between atraveling wave and a regenerator
(conductive wire-mesh screens) introduces one of the important industrial
applications: thermoacoustic engines and refrigerators [3]. A thermoacous-
tic engine absorbs heat at a high temperature and exhausts heat at a low
temperaturewhileproducingacousticpowerasanoutput.Athermoacous-
tic refrigerator works in the opposite way. It absorbs heat at a low
temperature, and requires the input of acoustic power to exhaust heat to
a high temperature.
3 ACOUSTICWAVEINDUCEDFLOWSANDHEATTRANSFER
In this chapter, the important transport phenomena resulting from
mechanically driven acoustic waves and thermally induced acoustic waves
under the microgravity and normal gravity conditions are also considered.
The fundamental characteristics of these problems are investigated and the
results show interesting features. The chapter first focuses on the flow
patterns and energy transport in high-intensity sound fields driven by pure
mechanical vibrations in gas-filled resonators. Thermally induced pressure
waves and associated convective flows for different kinds of fluids under
differentpressuresareconsiderednext.Theinteractionsbetweenflowfields
generated by acoustic waves and the buoyancy induced flows in different
heating enclosures are studied. Finally, the generation and propagation of
thermally induced acoustic waves in supercritical carbon dioxide are con-
sidered. Recent results on the generation, propagation and decay of ther-
mallyinducedacousticwavesinsupercriticalcarbondioxidearepresented.
A. MECHANICALLY DRIVEN ACOUSTIC WAVES
Mechanically driven sound wavesare sometimes associated with second-
order, steady, circulatory fluid motion, viz. acoustic streaming flow struc-
tures. Acoustic streaming is a net mean flow generated by sound. Acoustic
streamingcanbeclassifiedonthebasisofthedifferentmechanismsbywhich
the streaming is generated [4]. Rayleigh streaming [5] is referred to as
boundary layer driven streaming inside a standing wave resonator, as
the shear viscosity close to a solid boundary is responsible for the induced
flow. Rayleigh streaming also describes the mean fluid motion outside the
boundarylayer,whereitisoftenreferredtoouterstreaming.Schlichting
streaming(alsoknownasinnerstreaming)isthemeanflowmotioninside
theboundarylayerofastandingwavefield.ThesizeofSchlichtingstream-
ing vortices is estimated to be l/4D. Here, D is approximately equal to
(2i/.)
1/2
1.9c
i
where c
i
= is the viscous penetration depth. Inner and outer
streaming can be formed inside a channel where the acoustic wave propa-
gates longitudinallyor there isoscillationofsolidboundary wall insidethe
bulk medium. Eckert streaming is driven by dissipation within the fluid
volume and is predominant in the high frequency range. This kind of
streaming has been used in ultrasonic applications and can be referred to
asbulkdissipationdrivenstreaming.
Andres and Ingard [6,7] analytically investigated acoustic streaming
aroundacylinderanddiscussedthedistortionofthestreamingflowpatterns
asafunctionofsoundintensity,underdifferentReynoldsnumbers.Leeand
Wang[8]studiedtheeffectofcompressibilityonthestreamingpattern.Lee
andWangconcludedthatcompressibilitycouldaffecttheinnerbutnotthe
outerstreamingflowfortheflowbetweenparallelplates,butthatfortwoor
4 B.FAROUKETAL.
three-dimensionalobjects,suchasacylinderorasphere,thecompressibility
also affects the outer streaming pattern. They used the limiting velocity at
the edge of the inner streaming layer as a slip boundary condition to solve
forthelargeouterstreamingforthedifferentgeometries.
Theeffectofcompressibilityonacousticstreamingneararigidboundary
wasinvestigatedbyQi[9].Qitriedtoresolveanexistinginconsistency:while
thecompressibility isanecessarycondition forthepropagationofacoustic
waves, previous analyses of acoustic streaming were limited to incompres-
sible fluids. This inconsistent description is due to adapting the incompres-
sible boundary layer solutions obtained by Schlichting to solve acoustic
streamingnearrigidboundaries.Qisresultsshowedthattheconsideration
ofcompressibilityleadstoalargerstreamingvelocityoutsidetheboundary
layer.Theeffectwasfoundtobesignificantingases,butnotinliquids.The
studywasextendedtoinvestigateacousticstreaminginacirculartube[10].
Vainshtein [11] combined the problems of Rayleigh streaming and Cou-
etteflowandinvestigatedtheeffectofstreamingonshearflow.Menguyand
Gilbert[12]studiednonlinearacousticstreaminginaguidewithaperturba-
tioncalculationusingasymptoticexpansions.Anoticeabledistortionofthe
acousticfieldduetothefluidinertiawasdemonstratedandacomparisonof
slow and nonlinear acoustic streaming was presented. Andrade [13] experi-
mentallystudiedthestreamingflowsoccurringoutsidetheboundarylayers
inatubeandaroundcylindricalandsphericalobstacles.Theouterstream-
ingstructureswerevisualizedinairaroundacylindrical rodimmersedina
standing sound wave. The fluid particles flow towards the cylinder from
above and from below, and move away in both directions parallel to the
acousticoscillations(horizontal).
The phenomenon of flow reversal in the circulation occurring in the
vicinity of oscillating cylinders was experimentally observed by West [14].
Theresultsofthisinvestigationshowedthat,assoonastheamplitudeofthe
excited sound wave exceeds a critical value, the direction of the circulation
near the cylinderreverses and streaming outsidetheboundarylayer forms.
Holtsmark et al. [15] studied the vortices in air near a cylindrical rod in a
Kundts tube. The streaming in and outside the boundary layer was
observed. The boundary layer streaming structures around a cylinder were
visualized by Raney et al. [16] in a water-glycerin mixture. Coppens and
Sanders[17]measuredtheacousticwaveinaresonator.Theyfoundthatthe
microphone output waveforms become irregular when the sound intensity
increases. Asimilar measurementwas carriedout byCruikshank[18]. This
kindofnon-harmonicwaveshapeisbelievedtobethenon-linearsourceof
theacousticstreaming.
Trinh and Robey [19] investigated the streaming flows associated with
ultrasoniclevitators.Inthisexperimentalstudy,thestreamingflowfieldwas
caused by an ultrasonic vibrating wall of an enclosure. The circulation
inducedbythepresenceofasampleintheenclosurewasstudied.Asinitially
describedbyKempton[20]andMorfey[21],themixingofseverallumps
offluidwithdifferenttemperaturescanactasasoundsourcemainlydueto
the fact that the physical properties of the medium are not uniform. As
another example, a heated surface can maintain large amplitude acoustic
oscillations.Kempton[20]analyticallyexaminedtheroleofheatdiffusionas
an internal noise source. He considered a number of problems: sound
induced byunsteady heattransfer from asolid body;fluctuation flow over
a hot body, and unsteady heat transfer between gas streams. Morfey [21]
analytically examinedtheradiation ofsoundfrom a free turbulentjet. The
soundwasgeneratedduetothermaldissipationandcompositiongradient.
Acousticstreaminghasreceivedattentionasaheattransferenhancement
mechanism. Fand and Kave [22] used photography to study the streaming
around a circular cylinder in standing wave tube. The experimental results
showed that the streaming around the heated cylinder was much stronger
than the isothermal streaming for the same geometry and sound intensity.
The mean flow motion due to streaming not only contributes to the con-
vectiveheattransferbutalsopromotesthetransitiontoturbulence.Richard-
son[23]analyticallystudiedtheeffectofsoundonnaturalconvectionalong
a horizontal cylinder whichis immersed in soundfields. Later, Richardson
[24]presentedshadowgraphs foraheatedhorizontalcircular cylinder,sub-
jected to transverse horizontal and vertical sound fields. The shadowgraph
demonstratedthelocalchangesinboundarylayerthicknessandheattrans-
fercoefficient.
Engelbrecht and Pretorius [25] experimentally studied the influence of
soundwavesonthetransitionfromlaminartoturbulentflowinthebound-
ary layer associated with natural convection from a vertical flat plate with
uniform heat flux. The transition has been shown to occur at a Grashof
number lower than the normally cited value. Gopinath and Mills [26] esti-
mated the convective heat transfer from an isolated sphere in a standing
soundfieldduetoacousticstreaminginlargestreamingReynoldsnumbers.
TheNusseltnumbercorrelationsforawiderangeofPrandtlnumberswere
obtained. Later, Gopinath and Mills [27] extended their study to the air-
filledtube,modelinganacousticlevitationchamber.Theflowwastreatedas
incompressible. Analytical techniques were used along with a commercial
numerical solver for the solution of the complete elliptic form of the equa-
tionsgoverningthesteadytransportduetothestreamingmotion.
Vainshteinetal.[28]performedatheoreticalanalysisontheheattransfer
between two horizontal parallel plates in presence of a steady sonic wave
propagatedinthelongitudinaldirection.AnacousticPecletnumber,which
represents the interaction between heat conduction and forced convection,
6 B.FAROUKETAL.
wasintroduced,andthemeanNusseltnumberintermsofthePecletnumber
wasderived.Mozurkewich[29]placedaheatedcylindricalwireatavelocity
antinodeofastandingwaveandmeasuredtherateofheattransferfromthe
wire to the acoustic medium. The Nusselt number showed a distinctive
variation with acoustic amplitude. A similar experiment was reported by
GopinathandHarder[30].
Recently Kawahashi and Arakawa [31] numerically studied acoustic
streaming induced by finite-amplitude oscillation in a closed duct driven
by a piston. The results showed velocity distributions in the oscillatory
boundary layer and the change ofthe streaming profiledue tothe increase
intheamplitudeofoscillation,andtheexistenceofadoublelayernearthe
duct wall. Kawahashi et al. [32] performed experimental studies on the
interactionofacousticstreamingwithnaturalconvectioninaclosedrectan-
gularduct,andobservedintensifiedsteadystreaming.Lohetal.[33]inves-
tigated the acoustic streaming induced by ultrasonic vibrations in an open
spaceandtheassociatedconvectionenhancementusingboththeoreticaland
numerical methods. Later, Loh and Lee [34] measured the heat transfer
enhancementcapabilityofacousticstreaminggeneratedbyultrasonicvibra-
tion.Thecoolingeffectwasfoundtohaveastrongcorrelationwiththegap
sizebetweentheultrasonicdeviceandtheheatsource.Thecoolingeffectwas
maximizedwhenthegapcorrespondedtoamultipleofthehalf-wavelength
oftheultrasonicwave.
Mozurkewich [35] measured the heat transfer within a cylindrical reso-
nance tube, mediated by acoustic streaming. For an empty resonator with
heated wall segment, the radial heat flux varied with position in a manner
consistent with the global streaming pattern within the tube. Wan and
Kuznetsov [36] numerically studied the fluid flow and heat transfer due to
acousticstreaminginthegapbetweentwohorizontalbeams,thepartofthe
lowerofwhichisvibrating.Byutilizingtheperturbation method,thecom-
pressible Navier-Stokes equations were decomposed into the first-order
acousticequationsandsecond-orderstreamingequations.Asimilarproblem
was experimentally studied by Wan and co-workers [37]. The visualization
clearly showed that vertical streaming could be induced by bimorph vibra-
tion, which enhanced the heat transfer between the heated surface and
surrounding air. Hamilton et al. [38] derived an analytic solution for the
average mass transport velocity generated by a standing wave which is
excited by shaking the system harmonically between parallel plates. In
their study, both the streaming structure and streaming velocity were
described for various channel widths. They found that the inner vortices
increaseinsizerelativetotheoutervorticesaschannelwidthisreduced,and
theoutervorticesdisappearwhenthewidthofchannelislessthan5.7.The
analysis method was extended to a gas in which heat conduction and
dependenceofviscosityontemperatureweretakenintoaccount[39].Itwas
revealedthatforchannelwidths1020timestheviscousvelocitypenetration
depth, thermal effects may alter the streaming velocity substantially. For
significantly wider or narrower channels, thermal effects influence the
streamingvelocitybyonlyafewpercent.
WangandKuznetsov[40]analyticallysolvedfortheacousticstreamingin
achannelboundedbyonebeamatrestandonebeamvibratingatultrasonic
frequency. The flow field was found to be highly dependent on the aspect
ratio.Theclassicaltheoryofacousticstreamingisrestrictedtosloworlinear
streaming. It is assumed that in the second-order governing equations, any
nonlinear term that involves a second-order quantity is negligible. As a
result, the second-order equations are linear in the dependent variables.
Thisiswhyitusuallyreferredtoaslinearstreaming.Thenumericalmethod
that directly solves the Navier-Stokes equations provides a new method to
overcomethislimit.
Yano [41] used an upwind TVD scheme to solve the full Navier-Stokes
equation to investigate the turbulent acoustic streaming in a resonator.
When M<< 1(and hence Re 1),shockwavesare formedandthe gas
oscillationattainsaquasi-steadystate,wherethemagnitudeofthestreaming
velocity is linearly proportional to the acoustic Mach number. Aktas and
Farouk[42]directlysolvedthecompressibleNavier-Stokesequationsbyan
FCTscheme.Theyclassifiedacousticstreamingintoclassicalstreamingand
irregular streaming. The transfer line between the two streaming patterns
wasdeterminedbythetwofollowingnon-dimensionalvariables:wallvibra-
tion amplitude and width of channel. Aktas et al. [43] numerically investi-
gated thermal convection in a two-dimensional resonator due to acoustic
excitationsinducedbythevibrationoftheleftsidewall.Inthiscase,theleft
and right walls were held at different temperatures. The mechanically
inducedperiodicoscillationsinthefluidwerefoundtohaveaninsignificant
effectontheheattransfercharacteristicsofthesystem.
B. THERMALLY INDUCED ACOUSTIC WAVES IN GASES
Trilling [1] treated the problem of thermally induced sound fields in a
semi-infinitebodyofaperfectgas,subjectedtoastepchangeintemperature
at the solid wall. The one-dimensional compressible flow equations were
linearized and a closed-form asymptotic solution was obtained using a
Laplace transform technique. He also determined how sound intensity
dependedonwalltemperaturehistorybydevelopinganalyticalsolutions.
Larkin was apparently the first to use the numerical method to study
the thermally induced fluid motion in a confined medium as a result
of step increase in temperature on one wall [44]. In his numerical model,
8 B.FAROUKETAL.
one-dimensional equations of motion, continuity and energy were consid-
ered, while the kinetic energy and dissipation terms were omitted. He used
theexplicitfinite-differenceschemeforthemotionandenergyequations,but
an implicit scheme for the continuity equation. The acoustic nature of the
fluidmotionwasobserved,butthebehaviorofpressureoscillationwasnot
completelycapturedduetohighnumericaldiffusion.Thefluidmotionwas
observed to greatly increase the rates of both heat flow and pressure rise.
Solutions to the problem of the response of a perfect gas in a slot to a
monotonically varying temperature disturbance at the boundaries were
developed by Kassoy [45] employing a variety of perturbation methods.
Radhwan and Kassoy [46] studied the response of a gas confined between
infiniteparallelplanarwallssubjectedtosignificantheataddition.Solutions
weredevelopedintermsofasymptoticexpansions,validonlywhentheratio
ofacoustictoconductiontimescalesissmall.Numericalstudiesofone- and
two-dimensionalthermoacousticconvectioninaconfinedregionhavebeen
carriedoutbyOzoeandco-workers[47,48].ThecompressibleNavier-Stokes
equations for a gas with constant viscosity and conductivity and negligible
viscous dissipation were non-dimensionalized. Numerical solutions were
obtained by employing a first-order upwind scheme. The pressure waves
were captured, but it showed effects of substantial numerical diffusion.
The effects of various parameters, such as gravity, viscosity, wall tempera-
ture, andfluidproperties onthedevelopmentof convectionwerediscussed
intheirpaper.
A simplified model (the hyperbolic equation of conduction) for thermo-
acoustic motion was compared with the one-dimensional Navier-Stokes
equations model of the phenomena, and limitations of the simplified
approach were discussed by Brown and Churchill [49]. By using finer dis-
cretization, Brown and Churchills numerical results of one-dimensional
compressible Navier-Stroke equations showed that rapid heating of a solid
surface bounding a region of gas generates a slightly supersonic wave with
positiveamplitude in pressure, temperature,density andmass velocity[50].
Huang and Bau [51] solved the linearized equations for thermally induced
acousticwavesinasemi-infinitemedium,sufferingstepandgradualchanges
in the boundary temperature, by using a numerical inverse Laplace trans-
form method. They also solved equations of the nonlinear wave model by
using a finite differences scheme modified with a Galerkin finite element
interpolation in space. A similar theoretical analysis for thermally induced
acousticwavesinaconfinedmediumwasrepeatedmorerecentlybyHuang
andBau[52].Theirresultsillustratedthatthermallyinducedacousticwaves
haveacharacteristicshapeconsistingofasharpfrontandalongtail,which
is different from the round symmetric wave shape predicted by the earlier
numericalmethod.
Farouk et al. [53] used a control-volume-based flux-corrected transport
algorithm to predict the early time behavior of thermally induced acoustic
waves in a compressible-fluid filled cavity. In their numerical model, the
temperature dependent fluid properties were used. Aktas and Farouk [54]
recently studied the effect of gravity on the fluid motion generated by the
thermally induced acoustic waves in a rectangular enclosure. The gravita-
tional accelerationwas foundtohave anegligible effectonthebehavior of
thermal induced acoustic wave for early times. Only a uniformly heated
sidewall was considered and the flow development of thermally induced
acousticwavesunderzerogravityconditionswasnotstudied.
The generation of thermally induced acoustic waves in gases has been
studiedexperimentallybyonlyafewinvestigators.ParangandSalah-Eddine
[55]investigatedthethermoacousticconvectionphenomenainacylindercon-
tainingairinbothnormalandreducedgravityenvironments.Intheirresulting
measurementsofairtemperature,nooscillationswasrecordedduetothesmall
oscillation amplitude and low sample rate of measurement, but the air
temperatureswerefoundtorisemuchfasterthaninthecomputationalresults
forthecaseofpureconduction.Nopressuremeasurementwasreported.
Experimentalmeasurementsofpressurewavesgeneratedbyrapidheating
ofasurfacewerereportedbyBrownandChurchill[56].Intheirexperiments,
therapidheatingprocedurewasachievedbyanR-Ccircuit.Thesepressure
measurements in the wall of the closed chamber clearly demonstrated the
generationofacousticwavesbyrapidheatingofawall.
C. THERMOACOUSTIC WAVES IN SUPERCRITICAL FLUIDS
Afluidiscalledsupercritical,whenitstemperatureandpressureareabove
itscriticaltemperatureandpressure.Supercriticalfluidsexhibitanumberof
specific properties, such as large density, high conductivity, high specific
heat,lowviscosityandlowthermaldiffusivity,whichmakethemintermedi-
ate between gases and liquids. For carbon dioxide, the critical temperature
andpressureare304.13Kand7.3773MParespectively.Atthecriticalpoint,
thethermalconductivityofcarbondioxideachievesitsmaximumvalue.At
anygivenpressure,thethermalconductivityalsohasapseudomaximumat
the corresponding critical temperature. The thermal conductivity in super-
critical region, especially near the critical point, is higher than that in the
idealgasregion.
The thermal diffusivity D
T
vanishes near the liquid-vapor critical point.
Untilrecentlythisfactplustheusualequationofheatconduction
0T
=D
T
H
2
T (I1)
0t
10 B.FAROUKETAL.
were seen by most researchers as the complete picture for the very slow
relaxationtimesoftenobservedinone-phasecriticalpointexperiments.In
theone-phaseregion,thesetimesareoftheorderofT~L
2
,D
T
, where Lis
the characteristic length of the fluids container. However Straub and his
coworkers [57,58] measured very short temperature equilibration times
close to the critical point even when low gravity was used to suppress
convection. This and related puzzles led to the realization that the diver-
gence of the fluids compressibility near its critical point significantly
affects thermal equilibration [5961]. The heat transport in a pure fluid
near its critical point is thus strongly enhanced due to the very high
compressibility, even though the thermal diffusivity near the critical
point tends to zero. Beysens and co-workers [62] carried out quantitative
interferometry experiments under microgravity conditions in sample cells
filledwithsupercriticalCO
2
atdifferentdensitiesandtemperaturesaround
the critical point. They showed two interrelated mechanisms when the
fluid was heated: a diffusing thermal boundary layer and fast adiabatic
density increase in the bulk fluid via propagation of pressure waves.
The border of the expanding thermal layer acts as a piston, and hence
the phenomenon has been termed as the piston effect [62]. Further details
of thermally induced acoustic waves and their characteristics are given in
Section VI.
The fundamental process associated with the piston effect is due to the
propagationofacousticwavesinthenear-criticalfluidmedia.Itisinterest-
ingtonoteherethatacousticspeedofafluidclosetoitscriticalpointisnot
small, even though the compressibility diverges at the critical point. For
example,thesoundvelocityinSF
6
closetoitscriticalpoint(TT
c
=0.01K)
is 50m/s [63]. The high compressibility of the near-critical fluids with
moderateacousticspeedgiverisetothefastheattransportduetothepiston
effect.
Recentexperimentsintheabsenceofeffectsduetogravity,andespecially
thoseintheNASASpaceShuttleandtheInternationalSpaceStationhave
ushered in a new era in our understanding of the property behavior
and transport modes in fluids near their critical point. In order to investi-
gate the different aspects of the transport of heat in the absence of gravity
[64], performed a thermal cycle close to and around the critical point of
CO
2
atcriticaldensity.Reducedgravitywasprovidedduringa6-minflight
of a sounding rocket (TEXUS 25). A cell was designed which allowed
surface and bulk phenomena to be distinguished. The thermal cycle of the
experiments in weightlessness started at T
c
2.5mK where the authors
studied the relaxation of the perturbations caused by liftoff in the fluid
flow and density and temperature gradients. They also investigated the
effect of a quench from T
c
2.3mK to T
c
v1.3mK and the expected
mechanism of heat transport by the piston effect was experimentally evi-
denced. In particular, they observed homogeneous thermalization, with a
time of thermalization less than a few seconds in a cell with dimensions
10.1mm(diameter)and 9.025mm(length). Regardingotherrelatedstudies
abroadtheSpaceShuttle[65],usedanovel,overdampedoscillatortomeasure
the viscosity of xenon near its critical density ,
c
and temperature T
c
. In
microgravity, usefuldatawereobtained within 0.1mK of T
c
corresponding
310
7
to a reduced temperature T*=(T T
c
)/T
c
= . The data directly
revealtheexpectedpowerlawbehavior.Thesamegroupofresearchers[66]
reportedviscoelasticbehaviorofxenonwhenthereducedtemperatureT*was
lessthan10
5
.
II. MathematicalModelandNumericalMethods
A. OVERVIEW
A sound field is formed by a series of compressions and expansions of a
substance,whichobeythelawsofthermodynamics,heattransfer,andfluid
mechanics. It is essentially characterized by the pressure p, corresponding
density ,, temperature T, and velocities. Hence, an accurate mathematical
model must be able to describe the compressible behavior of the substance
in question. In this study, the compressible form of the Navier-Stokes
equations, including the conservation of mass, momentum and energy,
alongwithastateequation,areusedasthegoverningequations.
B. MATHEMATICAL MODEL
1. ConservationEquationsinTwo-DimensionalCylindricalCoordinates
In two-dimensional cylindrical coordinates (r, z), these equations can be
expressedintheconservativeformas
0, 10(,ru
r
) 0(,u
z
)
= 0 (II 1)
0t r 0r 0z
0(,u
r
) 10(r,u
r
2
) 0(,u
r
u
z
) 0p 10(rt
rr
) 0t
rz
t
00
= ,f
r

0t r 0r 0z 0r r 0r 0z r
(II 2)
2
0(,u
z
) 10(r,u
r
u
z
) 0(,u
z
) 0p 10(rt
rz
) 0t
zz
= ,f
z
(II 3)
0t r 0r 0z 0z r 0r 0z
_ _
_ _
_ _
_ _
12 B.FAROUKETAL.
= ,f vV
0E 10(ru
r
Eru
r
p) 0(u
z
Eu
z
p) 10(rq
r
) 0q
z
0t r 0r 0z r 0r 0z
(II4)
Here, E is the total energy, and c is the viscous dissipation rate. The
componentsofstresstensort are
10(ru
r
) 0u
z
0u
r
T
rr
=l 2j (II5)
r 0r 0z 0r
10(ru
r
) 0u
z
0u
z
T
zz
=l 2j (II6)
r 0r 0z 0z
10(ru
r
) 0u
z
u
r
t
00
=l 2j (II7)
r 0r 0z r
0u
z
0u
r
t
rz
=t
zr
=j (II8)
0r 0z
wherethebulkviscosityj
/
=l (2,3)jisassumedtobenegligibleandthe
secondcoefficientofviscositylissetequalto(2,3)j(zerobulkviscosity).
Thecomponentsoftheheatfluxvectorarewrittenas:
0T
q
r
= k (II9)
0r
0T
q
z
= k (II10)
0z
Theviscousdissipationtermisgivenas:
0[(t
rr
u
r
t
rz
u
z
)r[ 0[(t
zr
u
r
t
zz
u
z
)r[
c= (II11)
0r 0z
Anequationofstateforperfectgasisusedtorelatethetemperaturetothe
otherthermodynamiccharacteristics:
p=,RT (II12)
Here, R is the specific gas constant of the medium. For supercritical
carbondioxide,thep-,-TrelationsareobtainedfromtheNISTReference
Database12[67].
_ _ __
_ _ __
The definition of the speed of sound waves in compressible fluid is
expressedas

c=
c
p
0p
_
(II13)
c
v
0,
T
Here c
p
and c
v
are the specific heats at constant pressure and constant
volume, respectively. For an ideal gas, this equation can be expressed as a
functionoftemperatureonly:
_
c
p
c= RT (II14)
c
v
Theinternalenergyofidealgascanbeexpressedusingthefollowingsimple
equation:
P
i= (II15)
1
Here,istheratioofspecificheat.Forsupercriticalfluids(SectionVI),the
internalenergy
i=f ( p, ,)andtherelations areobtainedfromtheNISTReferenceData-
base12.
Inthree-dimensional Cartesian coordinates (x, y,and z), these equations
canbeexpressedintheconservativeformas:
0, 0(,u) 0(,v) 0(,w)
=0 (II16)
0t 0x 0y 0z
0(,u) 0(,uu) 0(,uv) 0(,uw) 0p 0 2 0u 0v 0w
=,f
x
j
0t 0x 0y 0z 0x 0x 3 0x 0y 0z
_ _ _ _ __ _ _ __
0 0u 0 0v 0u 0 0w 0u
2j j j (II17)
0x 0x 0y 0x 0y 0z 0x 0z
0(,v) 0(,uv) 0(,vv) 0(,vw) 0p 0 2 0u 0v 0w
=,f
y
j
0t 0x 0y 0z 0y 0y 3 0x 0y 0z
_ _ __ _ _ _ _ __
0 0u 0v 0 0v 0 0w 0v
j 2j j (II18)
0x 0y 0x 0y 0y 0z 0y 0z
_ _ __
14 B.FAROUKETAL.
0(,w) 0(,uw) 0(,vw) 0(,ww) 0p 0 2 0u 0v 0w
= ,f
z
j
0t 0x 0y 0z 0z 0z 3 0x 0y 0z
_ _ __ _ _ __ _ _
0 0u 0w 0 0v 0w 0 0w
j j 2j (II 19)
0x 0z 0x 0y 0z 0y 0z 0z
andconsideringzerobulkviscosity,j
/
= l (2,3)jandsettingl= (2,3)j,
weget
_ _
0E 0(Eu) 0(Ev) 0(Ew) 0(pu) 0(pv) 0(pw)
_ _
= ,(f
x
u f
y
v f
z
w)
0t 0x 0y 0z 0x 0y 0z
_ _
_ _ _ _
_ _ _ _
_ _
0
0u 1 0v 0u 1 0u 0w
2 0u 0v 0w
2j
_
u
_
_
v
_
_
w
_
j
_

_
u
0x
_ 0x 2 0x 0y 2 0z 0x
3 0x 0y 0z _
_ _
_ _ _ _
_ _ _ _
_ _
0
1 0v 0u 0v 1 0w 0v
2 0u 0v 0w
2j
_ _
_
u v
_
_
w
_
j
_

_
v
0y
_ 2 0x 0y 0y 2 0y 0z
3 0x 0y 0z _
_ _
_ _ _ _
_ _ _ _
_ _
0
1 0u 0w 1 0w 0v 0w 2 0u 0v 0w
_ _
w
0z
_ 2 0z 0x 2 0y 0z 0z 3 0x 0y 0z _
2j
_ _
_
u
_
_
v w j
_

_ _ _ _ _ _
0 0T 0 0T 0 0T
_
k
_
_
k
_
_
k
_
(II 20)
0x 0x 0y 0y 0z 0z
Here,Eisthetotalenergyandiistheinternalenergyandu,v,andwarethe
velocitycomponentsinthex,yandzdirectionsrespectively:
1
2
) E= i ,(u
2
v
2
w (II 21)
2
C. NUMERICAL METHODS
Fortheresultspresentedinthischapter,thegoverningequations(except
for the diffusion terms) are discretized using a finite-volume method based
on the flux-corrected transport (FCT) algorithm [68]. FCT is a high-order,
nonlinear, monotone, conservative and positivity-preserving method
designedtosolveageneralone-dimensionalcontinuityequationwithappro-
priatesourceterms.Thisschemehasfourth-orderphaseaccuracyandisable
_ _ _ _
to resolve step gradients with minimum numerical diffusion. The diffusion
terms (the viscous term in the momentum equations and the conduction
terms in the energy equation) were discretized using a second-order central
difference approach. Time-step splitting was also used to couple all of the
representativephysicaleffects.
No-slip boundary conditions were used at all solid walls. In the present
computationalmethod,thetreatmentproposedbyPoinsotandLele[69]was
followedforimplementingtheboundaryconditionsforthedensity.
1. EvaluationofBoundaryConditions
A high-order non-dissipative algorithm such as FCT requires rigorous
formulation of boundary conditions [70]. Otherwise, numerical solutions
may show spurious wave reflections at the regions close to boundaries and
nonphysical oscillations arising from instabilities. In the present computa-
tional method, the treatment proposed by Poinsot and Lele [69] was
followed for implementing the boundary conditions for density. This
methodbasedonthetheoryofwavecharacteristics,avoidsincorrectextra-
polationsandoverspecifiedboundaryconditions.
Alonganysolidwall,thedensityiscalculatedfrom
0, 1 0p 0u
n
,c =0 (II22)
0t c
w
0n 0n
w w
wherethesubscriptwsignifiesthelocationofthewall,andnisthedirection
normaltothewall.
FollowingtheproceduredescribedbyPoinsotandLele,thedensityalong
avibratingwall(foridealgases)iscalculatedby
0, , 0u
w
,(u
w
c
w
)0u (u
w
c
w
)0p
= (II23)
0t c
w
0t c
w
0x c
2
0x
w
III. MechanicallyDrivenAcousticWavesinGas-FilledEnclosures
A. OVERVIEW
Interactionofacousticwavesincompressiblefluidsandsolidboundaries
createschallengingproblems.Itisawell-knownfactthatsoundsourcesmay
generateaflowfieldinwhichtheparticlevelocitiesarenotsimplysinusoidal
and a pattern of time-independent vortical flows or steady circulations is
16 B.FAROUKETAL.
oftenfoundinthebodyofcompressiblemedia[71].Soundathighintensity
levelsingasesandliquidsisaccompaniedbythesesecondordersteadyflow
patterns known as acoustic streaming. These steady flows always have
rotational character and their velocity increases with the sound intensity.
However,evenatthehighintensitylevels,thesecondarystreamingvelocity
magnitude remains smaller than the primary oscillatory particle velocity
magnitude.
While various analytical models are available for describing acoustic
streamingphenomena,theyareusuallybasedonsubstantialapproximations
and the solutions are often limited to idealized conditions. The study of
acousticstreaming startedwiththe theoretical work ofLord Rayleigh [72].
Heconsideredvortexflowsoccurringinalongpipe(Kundtstube)because
of the presence of a longitudinal standing wave. Westervelt [73] obtained a
generalvorticityequationanddevelopedageneralprocedureforevaluating
thestreamingvelocityinducedbyacousticaldisturbancesthatcanbespeci-
fiedinafirst-ordermannerbyawaveequationlinearinthefieldvariables.
Thetreatmentpresentedintheabovestudywasnotapplicabletoproblems
wherethewaveamplitudesarelargeandthefieldvaluescannotbedescribed
byalinearwaveequationbecauseofthefactthatnofirst-orderspecification
ofthefieldexists.Thevorticityequationwasthussubjecttotherestrictions:
solenoidalfirstordermotionandirrotationalfirstordermotion.Andresand
Ingard investigated low [7] and high Reynolds number acoustic streaming
analyticallyanddiscussedthedistortionofthestreamingflowpatternsasa
function of sound intensity. Nyborg reviewed the theories for calculating
steady streaming associated with sound fields. He worked out two illustra-
tiveproblems,bothforrectilinearflowduetoirrotationalsoundfields:the
firstdealswithasingleattenuatedplanewavetravelingdownatubeandthe
second one deals with a pair of crossed plane waves. An approximate
solution was also developed by Nyborg [74] for sonically-induced steady
flow near a fluid-solid interface subjected to the condition of known
irrotational oscillatory velocity distribution in the vicinity of the interface.
The previously developed theoretical approaches are based on successive
approximations to solutions of nonlinear hydrodynamic equations, where
thefirstandsecondordersolutionsineachapproximationmustsatisfythe
boundaryconditions.
B. FLOWS IN AN ACOUSTICALLY DRIVEN RECTANGULAR ENCLOSURE
A rigorous method for predicting acoustically excited flow fields in a
resonator is described in this section [75]. A perfectly sinusoidal primary
standing wave field in the domain of interest is assumed and the investiga-
tion isstartedfrom this point.The study[75]isspecifically concerned with

N
2
y
0
y
0
x
y
L = /2
H
the modeling of the actual physical processes, i.e. the formation of the
sound field and streaming processes in the resonator. No pre-defined
sound field in the computational domain is assumed for the simulation of
acoustic streaming. Instead, we directly compute and fully describe the
formationoftheacousticfieldinthedomain.Arectangularenclosurefilled
with a compressible fluid is considered. The left wall of the enclosure is
modeled as a rigid boundary, which vibrates harmonically in time repre-
senting the motion of a loudspeaker diaphragm or vibration of a commer-
cial ultrasonic mixer probe. The vibrating boundary is the acoustic source
in this geometry and sound field in the enclosure is created by this source.
We are able to model the physical processes including the compression of
the fluid and the generation of the wave, acoustic boundary layer develop-
ment and finally the interaction of wave field with viscous effects and
formation of streaming structures. We consider five different enclosure
heightswithvariouswalldisplacementvaluesforthegeometriesconsidered
inthisstudy.Thispresentmodelisaflexibletoolthatcananalyzeacoustic
streaminginvariousgeometries,andwithdifferentfluids.Theresultsofthe
current study will also be a good reference and starting point for more
advanced studies involving heat transfer analysis in acoustically induced
flow fields with streaming.
The pressure amplitude in the enclosure (see Fig. 1) is determined by
adjusting the maximum displacement of the wall (X
MAX
). The calculations
werestartedwiththevibrationoftheleftwallatx=0andwithuniformvalues
ofpressure(101,325Pa),temperature(300K)anddensity(1.138kg/m
3
)within
thequiescentmedium.Forthedurationofeachcycleofthevibrationofthe
wall, about 25,000 time steps were used for the explicit numerical scheme
describedearlier.
The cases considered [75] are summarized in Table I. The dimensionless
channel width (y
0
/c
i
), the maximum displacement of the left wall (X
MAX
),
and the aspect ratio of the enclosure (H/L=2y
0
/L) are included in Table I
foreachcase.
FIG.1. Schematicoftheproblem[75].
18 B.FAROUKETAL.
TABLE I
SUMMARY OF CASES CONSIDERED FOR ACOUSTIC STREAMING ANALYSIS [75]
Case y
0
/c
i
X
MAX
(mm) H/L
A 10 10 0.0358
B-1 20 10 0.0716
B-2 20 3 0.0716
C-1 30 10 0.1074
C-2 30 2 0.1074
C-3 30 1 0.1074
D-1 40 10 0.1432
D-2 40 0.5 0.1432
E-1 50 10 0.1790
E-2 50 0.5 0.1790
E-3 50 0.2 0.1790
In the first case considered (Case-A), the enclosure half-width is the
smallest (y
0
/c
i
=10) and the maximum displacement of the wall was set to
X
MAX
=10mm. Figures 2(a) and (b) show the pressure and velocity distri-
butions respectively along the symmetry axis (y=y
o
) of the enclosure at
.t=0,/2,, 3/2forCaseA(duringcycle#100).Pressuredistributionfor
.t=2(notshownhere)isidenticaltothecurvegivenfor.t=0.Perfectly
sinusoidalprofileoftheemittedwavebytheoscillatingwallisdistorteddue
toviscous andnonlinear effects.At.t=0and.t=,theamplitudeofthe
pressurewavesreachamaximumvalueintheenclosure.Atthebeginningof
the cycle (.t=0), the pressure is maximum on the vibrating wall of the
enclosureanddecreaseswithdistancefromthewallandreachesaminimum
attheoppositewall(x=L).Inthemiddleofthecycle(.t=)thepressure
profile is symmetric respect to vertical mid-plane (x=L/2) to the profile
givenfor.t=0andreachesamaximumattherightwall.
The pressure profiles given for different time levels intersect at approxi-
matelyx=L/2andcreatesapressurenode.Correspondingu-velocityprofile
for this case is given in Fig. 2 (b). Unlike the pressure field, velocity
maximums and minimums are observed at .t=/2 and .t=3/2. The left
wall isstationaryat.t=/2and.t=3/2.Theprimary oscillatoryflowis
periodicandthemaximumvalue ofthevelocityispredictedtobeapproxi-
mately 12m/s. Due to attenuation caused by viscous and nonlinear effects,
both pressure and velocity profiles slightly differ from a perfect sinusoidal
wavefield.
Predicted secondary flow field for Case-A is shown in Fig. 3 for the
enclosure. This flow field is based on the average mass transport velocity
valuesintheenclosure.Theaveragemasstransportvelocitywasgivenby

(a)
p

(
P
a
)
99,000
100,000
101,000
102,000
103,000
104,000
105,000
106,000
107,000
108,000
109,000
0
/2

3/2
0 2 4 6 8
x (mm)
(b)
10
u

(
m
/
s
)
5
0
5
10
0
/2

3/2
0 2 4 6 8
x (mm)
FIG.2. Variationof(a)pressureand(b)u-velocityalongthesymmetryaxisoftheenclosure
atfourdifferentinstant(.t=0,/2,, 3/2)duringtheacousticcycle#100(CaseA)[75].
,u) ,v)
u
st
= v
st
= (III 1)
,) ,)
whichdescribesthex- and y-componentsofthestreamingvelocities,respec-
tively.Here ) indicatesthetimeaveragedquantities.
0.3
y
(
m
m
)
0.2
0.1
2 4 6 8
x (mm)
20 B.FAROUKETAL.
FIG.3. Meanflowfieldintheenclosureatt=0.005s(CaseA)[75].
Thetimeaveragingwasappliedduringthe100thvibrationcycle(between
cycle#99and#100)andgivesthestreamingvelocityvaluesneart=5ms.
Thestreamingvelocitiescalculatedbasedonthetimeaveragingduring150th
(t=7.5ms) and 200th (t=10ms) cycles do not differ significantly from the
values computed at t=5ms. Hence, the average mass transport velocities
wereassumedtobecycle-independentbythistime(t=5ms).Themaximum
streaming velocity value is found to be approximately 0.06m/s while the
maximum instantaneous velocities reach 12m/s in the primary oscillatory
flow field in the enclosure. Four clockwise and four counter clockwise
circulationsare observed inthe enclosure.Twoofthese vortical structures,
namelyinnerstreamingstructures,formedatthevicinityofthebottomwall.
Theheightofthecirculatoryflowstructures(innerstreaming)observednear
thehorizontalwallsischaracterizedbythethicknessoftheacousticbound-
ary layer. The streaming structures seen in the middle section of the enclo-
sure (outer streaming) have larger sizes. The horizontal length of both the
inner and the outer streaming vortices is characterized by quarter wave-
length(l/4).Predictedstreamingstructuresareingoodagreementwiththe
resultsreportedbyHamiltonetal.[38,39]intheirrecentstudies.
Figure4(a)showsthevariationofthex-componentofstreamingvelocity
along the enclosure semi-height at x=3L/4 for case A. In this figure, the
vertical axis is the x-component of the dimensionless streaming velocity
(u
st
/u
R
) and u
R
is the reference velocity given by u
R
=3u
0
2
/16c
0
where u
0
is
the maximum oscillatory velocity. This reference velocity value represents
the maximum streaming velocity in case of a perfect sinusoidal waveform
obtained by Rayleigh. Results from Hamilton et al. [39] are also included
in the same figure (dashed curve). The predictions of the current study for
CaseAcomparewellwiththeresultsfromHamiltonetal. [38].Figure4(b)

1
0.8
0.6
0.4
0.2
0
0.2
0.4
0.6
0.8
1
1 0.2 0.4 0.6 0.8
y /y
0
Presentstudy
Hamiltonet al.
(a)
u
s
t

/
u
R

0.5
0.4
0.3
0.2
0.1
0
0.1
0.2
0.3
0 0.2 0.4 0.6 0.8 1
y /y
0
v
s
t

x
0
/
u
R
y
0
Presentstudy
Hamiltonet al.
(b)
FIG.4. Variationofthestreamingvelocityalongthesemi-heightoftheenclosure(CaseA)
[75],(a)u-componentofthestreamingvelocityatx=3L/4and(b)v-componentofthestreaming
velocityatx=L/2.
shows the variation of the y-component of streaming velocity along
the enclosure semi-height at x=L/2. The vertical axis represents the
y-component of the non-dimensional streaming velocity (v
st
x
0
/u
R
y
0
).
The maximum difference between the predictions of the current study and
thereferencesolution(dashedcurve)[38]isapproximately10%.Theresults
giveninthereferencesolutionareforaresonatorinwhichthesoundfieldis
assumed to be formed by shaking the system with a harmonic excitation.
Since the present study considers a resonator with a vibrating boundary,
theresultingvelocityfieldsslightlydiffer.
Inthenextcaseconsidered(CaseB-1),theenclosureheight(y
0
/c
i
=20)was
doubledcomparedtoCaseA.Themaximumwalldisplacementwaskeptthe
same(X
MAX
=10mm).Figure5showsthepressuredistributionintheenclo-
sureforthiscaseduringacousticcycle#100.Pressurewavesemanatingfrom
thesinusoidaldisplacementoftheoscillatingwallisstronglydistortedbythe
p

(
P
a
)
1,12,000
1,10,000
1,08,000
1,06,000
1,04,000
1,02,000
1,00,000
98,000
96,000
0 2
0
/2
3/2

4 6 8
x (mm)
22 B.FAROUKETAL.
FIG.5. Variationofthepressurealongthesymmetryaxisoftheenclosureatfourdifferent
instant(.t=0,/2,, 3/2)duringtheacousticcycle#100(CaseB-1)[75].
nonlineareffects.Sincewehaveawidersystem(y
0
/c
i
=20),theviscouseffects
are weaker and shear forces along the top and the bottom walls have less
effect on the bulk of the gas. This leads to higher-pressure amplitudes and
more shock wave type profiles in the enclosure due to less attenuation.
Similar wave profiles were reported by Chester [76] and several other
researchers for resonant gas oscillations in closed tubes. Corresponding
flowstructuresbasedontheaveragemasstransportvelocityintheenclosure
isshowninFig.6.Thefiguredepictstheflowfieldinthebottomhalfofthe
enclosure.Forthiscase,thesecondaryflowpatternsareobservedintheshape
ofirregularvorticalstructures.Severalsecondordervorticesformthestream-
ing flow field. Circulation loops apparently become unstable and non-
symmetricvortexpatternsareobserved.Theinnerstreamingstructuresare
destructed but visible. The maximum primary oscillatory flow velocity is
predictedas20m/swhilethemaximumflowspeedis0.54m/sinthesecond-
aryquasi-steadystreamingflowfield.Forthiscase,wequalitativelycompare
ourresultswiththepredictionsofYano[41].Theaspectratiooftheenclosure
0.07 is quite close to that of 0.1, considered by Yano. Yanos definition of
streamingReynoldsnumbercanbeexpressedasX
MAX
c
0
/iusingthepresent
nomenclature.Basedonthisdefinition,streamingReynoldsnumberforCase
B-1is699.ThisvalueisclosetotheminimumstreamingReynoldsnumber
(560)consideredbyYano.However,thepressureamplitudesandtheoscilla-
toryflowsare much strongerin Case B-1,compared to those addressedby

y

(
m
m
)
0.1
0.2
0.3
2 4 6 8
x (mm)
p

(
P
a
)
110,000
108,000
106,000
104,000
102,000
100,000
98,000
96,000
10m
3m
0.00495 0.005 0.00505
t (s)
FIG.6. Meanflowfieldintheenclosureatt=0.005s(CaseB-1)[75].
Yano.Yanoscomputationsdemonstratestreamingstructureslocalizednear
thetubewall.Anumberofvorticesarevisibleinthebottomhalfofthetube
distributednon-homogeneouslyandinskewedform.Thestreamingcurrents
arenotstrongenoughtopositionthevortexneartwoendsofthetube.The
present computations showmore homogeneousstreamingpatterndistribu-
tionintheenclosureduetolargerstreamingvelocities.
In the next case considered (Case B-2), we decreased the maximum wall
displacementvaluetoX
MAX
=3mmtoreducethesharppressureandvelocity
gradientsobservedforCaseB-1intheenclosure.Theenclosureheightiskept
thesame(y
0
/c
i
=20)asinCaseB-1.Figure7comparesthetemporalvaria-
tionofthepressureforCaseB-1andB-2atthemid-pointoftheleftwallatthe
endof100thvibrationcycle.Thepressureamplitudereachesapproximately
112,000
FIG.7. Timeevolutionofleftwallpressureatthemid-point(CaseB-1,B-2)[75].
y

(
m
m
)
0.3
0.2
0.1
2 4 6 8
x (mm)
24 B.FAROUKETAL.
FIG.8. Meanflowfieldintheenclosureatt=0.005s(CaseB-2)[75].
16kPawith10mmmaximumwalldisplacement(CaseB-1)whileitis6kPa
with3mmmaximumwalldisplacement(CaseB-2).Inaddition,withhigher
walldisplacementvalue(10mm),thepressurewaveformismuchsharperand
shockwavetypeprofileisobserved.Forsmallwalldisplacement(3mm),the
pressureprofiledepictsslightlyattenuatedsinusoidalwaveinthesystem.The
temporalvariationofthepressureattheleftwallischaracterizedbysudden
amplitudeincreaseforCaseB-1whileamoregradualincreaseisobservedfor
CaseB-2.Correspondingquasi-steady(time-averaged)flowstructureinthe
enclosureforCaseB-2isshowninFig.8att=5ms.Forthiscase,theflow
patternsareobservedintheshapeofregularstructures.Theinnerandouter
acoustic streaming structures are clearly visible. The maximum primary
oscillatory flow velocity is computed as 7m/s while the maximum flow
speedis0.03m/sinthesecondarysteadystreamingflowfield.
For the next three cases the enclosure height was increased further
(y
0
/c
i
=30).Inthisgeometry,X
MAX
=10mm,X
MAX
=2mm and X
MAX
=1mm
wereconsideredasmaximumwalldisplacementvaluesforCaseC-1,C-2and
C-3, respectively. Thepressure profilesobtainedforthesethreecasesare pre-
sentedinFig.9.Largerwalldisplacementcreateshigher-pressureamplitudes.
Inaddition,thepressuregradientsinthesoundfieldincreaseswithlargerwall
displacement.Sharpwaveprofileindicatesthepresenceofhigherharmonicsin
the wave field in addition to fundamental component, which describes a per-
fectly sinusoidal wave. The maximum pressure amplitudes are approximately
19.2kPa,5.2kPaand3.1kPaforCaseC-1,C-2andC-3,respectively.Figures10
(a),(b)and(c)showtheinstantaneousflowfieldintheenclosureforthesecases
att=5mswhichcorrespondstothebeginningofacousticcycle#101(.t=0).
Thesearerepresentativefiguresandalthoughtheflowdirectionchangesduring
theacousticcycleduetoperiodicoscillations,thenatureoftheflowstructureis
preserved. The maximum oscillatory flow velocities are 25m/s, 6.8m/s and
3.6m/sforcasesC-1,C-2andC-3,respectively.InFig.10(a),astronglytwo-
dimensionalflowpatternisobserved.Nearthebottomwall,flowisuniform.
p

(
P
a
)
96,000
98,000
100,000
102,000
104,000
0
0
0
106,000
108,000
110,000
112,000

10m
2m
1m
0 2 4 6 8
x (mm)
FIG.9. Variationofthepressurealongthesymmetryaxisoftheenclosureattwodifferent
instant (.t=0, ) during the acoustic cycle # 100 for Case C-1 (X
MAX
=10mm), Case C-2
(X
MAX
=2mm)andCaseC-3(X
MAX
=1mm)[75].
Asthewalldisplacementamplitudedecreases(caseC-2),two-dimensional
form of the transient flow field approaches to a one-dimensional form
(Fig.10(b))andacirculationpatternresultingfrom viscousinteractionsis
presentatx4mm.ForcaseC-3,theflowfieldismostlyone-dimensional
exceptregionsclosetothebottomwall(Fig.10(c)).
Thetime-averagedflowfieldspredicted forthese three cases aregiven in
Fig.11(a),(b)and(c).InFig.11(a),theflowpatternintheenclosureinthe
form of number of vortical structures and five streaming structures are
observedperquarterwavelength.Thisirregularstreamingflowpatternisa
result of two-dimensional transient (periodic) flow field (Fig. 10 (a)). The
maximumstreamingflowspeedis2.1m/s.InFig.11(b),threeouterstream-
ingstructuresareobservedperquarterwavelengthwhileoneinnerstreaming
structure exists per quarter wavelength. The transform of oscillatory flow
fieldfromatwo-dimensionalformtoone-dimensionalformisreducingthe
number of streaming structures and the irregularity of the streaming field.
The streaming pattern is steady but quite different from that of classical
Rayleighstreaming.

(a)
0.4
2 4 6 8
y

(
m
m
)
0.2
x (mm)
(b)
0.4
2 4 6 8
y

(
m
m
)
0.2
x (mm)
(c)
0.4
2 4 6 8
y

(
m
m
)
0.2
x (mm)
26 B.FAROUKETAL.
FIG.10. Oscillatoryflowfieldintheenclosureatt=0.005sfor(a)CaseC-1;(b)CaseC-2;
(c)CaseC-3[75].

(a)
0.4
y

(
m
m
)
0.2
(b)
0.4
y

(
m
m
)
0.2
(c)
0.4
y

(
m
m
)
0.2
x (mm)
x (mm)
x (mm)
2 4 6 8
2 4 6 8
2 4 6 8
FIG. 11. Mean flow field in the enclosure at t =0.005 s for (a) Case C-1; (b) Case C-2;
(c)CaseC-3[75].

Themaximumstreamingflowspeedis0.044m/s.InFig.11(c),twoinner
and two outer streaming structures are present in the system and the max-
imumstreamingflowspeedis0.008m/s.
In thelast three cases considered (Case E-1, E-2, E-3), thedimensionless
enclosure height was chosen as y
0
/c
i
=50. In this geometry we consider
X
MAX
=10mm, X
MAX
=0.5mm and X
MAX
=0.2mm as maximum wall dis-
placementsforCaseE-1,E-2andE-3,respectively.Figure12comparesthe
temporal variation of the pressurefor these three cases at the mid-point of

115,000
10m
110,000
105,000
p

(
P
a
)
100,000
95,000
90,000
t (s)
0.2m
0.5m
0.004975 0.005 0.005025 0.00505
28 B.FAROUKETAL.
FIG.12. Timeevolutionofleftwallpressureatthemid-pointforCaseE-1(X
MAX
=10mm),
CaseE-2(X
MAX
=0.5mm)andCaseE-3(X
MAX
=0.2mm)[75].
theleftwallattheendofthe100thvibrationcycle.Thepressureamplitude
reaches a value of approximately 21.2kPa for the 10mm wall displacement
(CaseE-1),2.5kPafor0.5mm(CaseE-2)and1.1kPafor0.2mm(CaseE-3)
wall displacements. In addition, with higher wall displacement value, the
pressurewaveformismuchsharperanddistorted.Thetemporalvariationof
pressure at the left wall is characterized by sudden amplitude increase for
Case E-1 while amoregradualincreaseis observed forCasesE-2 and E-3.
Corresponding streaming structures in the enclosure for all three cases are
shown in Figs. 13(a), (b), and (c). All three figures demonstrate highly
irregular flow structure for the nonlinear sound fields. For Case E-1,
approximatelyfourouterstreamingrollsareobservedperquarterwavelength.
In Fig. 12 the pressure profiles seem linear for Case E-2 (X
MAX
=0.5mm)
andforCaseE-3(X
MAX
=0.2mm)thoughthe(weak)attenuationeffectsare
present. However, the corresponding flow structures do not contain the
classical streaming field having two inner and two outer vortices per half
wavelength. Approximately four and half streaming rolls are observed per
quarterwavelengthforthecasesE-2andE-3.Thepressurevariations(Fig.12)
areweakerinthedomainforcasesE-2(2.5kPa)andE-3(1.1kPa)compared
totheclassicalstreamingcasesA(Fig.2,9.6kPa),B-2(Fig.7,6kPa)),andC-3
(Fig. 9, 3.1kPa) though weakly attenuated pressure fields predicted for all
five cases. Similar behavior is observed also for the maximum velocity
x (mm)
2 4 6 8
x (mm)
2 4 6 8
(a)
0.6
y

(
m
m
)
0.4
0.2
(b)
0.6
y

(
m
m
)
0.4
0.2
(c)
0.6
y

(
m
m
)
0.4
0.2
x (mm)
2 4 6 8
FIG. 13. Mean flow field in the enclosure at t=0.005s for (a) Case E-1; (b) Case E-2;
(c)CaseE-3[75].
30 B.FAROUKETAL.
TABLE II
CHARACTERISTIC RESULTS FOR THE CASES STUDIED
Case u
MAX
v
stMAX
(m/s) Re
s
Streamingstructure
A 12 0.06 73 Classical
B-1 20 0.51 203 Irregular
B-2 7 0.03 24.8 Classical
C-1 25 2.1 317 Irregular
C-2 6.8 0.044 23.4 Irregular
C-3 3.6 0.008 6.6 Classical
D-1 26.5 3.17 356 Irregular
D-2 2.5 0.005 3.2 Irregular
E-1 28 4.25 397 Irregular
E-2 3 0.047 4.5 Irregular
E-3 1.3 0.02 0.86 Irregular
values(u
MAX
)intheoscillatoryflowfields(TableII).Thesetwopointsappear
to be the major factor resulting in irregular streaming motion. When the
enclosure is sufficiently high, (H/L 0.14) Rayleigh streaming patterns
arenot observedeven for small vibration amplitudes.Comparedto narrow
enclosures having oscillatory velocity amplitudes of the same order, wide
systems are associated with larger streaming velocities and irregular flow
structures. The maximum oscillatory flow velocities are 28m/s, 3m/s, and
1.3m/sforcasesE-1,E-2andE-3,respectively,whilethemaximumstreaming
flow speeds are 4.5m/s, 0.047m/s and 0.02m/s. Since the thickness of the
acoustic boundary layer is small compared to the enclosure height, inner
streamingstructuresdiminish.
Figure14showsthevariationofnormalizedpressureamplitude(Dp/Dp
1
)
calculated at resonance with the modified acoustic Reynolds number
(Re
a
1/3
/). Here Dp indicates the amplitude of the pressure variation
at the right wall (Dp=p
max
p
min
) and Dp
1
=8p
0
/[(1)]
1/2
, where
=(X
MAX
/L)
1/2
and p
0
is the initial pressure. Acoustic Reynolds number
isdefinedasRe
a
=c
0
X
MAX
/i and=L/Histhedimensionlesslengthofthe
enclosure. In this figure dashed curve indicates the values computed based
onthecorrelationsuggestedbyAlexeevandGutfinger[77].Thiscorrelation
is given for 0.05 < < 0.25. The data from present study (symbols in the
figure) are in the range of 0.03 < < 0.06. The pressure amplitudepredic-
tionsofthepresentstudycomparewellwiththeabovecorrelation[77].The
authors also reported recirculating flow patterns at each half of the tube
resulting from acoustic streaming for =0.137 and H/L=0.016. In their
predictions,thedirectionofstreamingatresonancewasoppositetothatat
non-resonance frequencies, when gas oscillations are continuous. In the
0
0.2
0.4
0.6
0.8
1
p
1
AlexeevandGutfinger[77]
Presentstudy
O
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1
1/3
/ Re
a
FIG. 14. Normalized pressure amplitude at resonance versus dimensionless parameter
Re
1/3
a
/ [75].
present computations, we did not observe flow reversal for the circulatory
structureshavingwidthofquarterwavelengthfor0.06andH/L _ 0.036.
Merkli and Thomann observed similar irregular circulations resulting
from the transition to turbulence in their experiments. They considered
gas oscillations excited with non-resonant frequencies (the flow properties
varied sinusoidally in time) in a tube and reported a number of vortices
non-uniformly distributed in the tube for H/L 0.01. In this study, the
criticalReynoldsnumberfortransitionwasfoundasA
c
=2u
MAX
/(i.)
1/2
400. The predictions of the present study differ from the observations of
Merkli and Thomann especially for the distribution of streaming patterns.
For the cases we considered, irregular streaming structures appear in
uniform form in the enclosure. However, for the present computations,
the maximum value of A
c
is 40 and the time variations of flow properties
are not sinusoidal.
Some characteristic results obtained for each case and the forms of
streaming fields are summarized in Table II. The maximum velocity in the
oscillatoryflowfield(u
MAX
),themaximumacousticstreamingvelocityand
streamingReynoldsnumber(Re
s
=u
2
MAX
/i.)areincludedinthetable.An
observation from this data is that the relative magnitude of the streaming
velocityrespecttotheoscillatoryflowvelocity(v
stMAX
/u
MAX
)ismuchlarger
forirregularstreamingconditionscomparedtoclassicalstreaming.
The regions (defined by the aspect ratio of the enclosure and the max-
imum displacement of the wall amplitude) for which classical streaming
patterns are shown in Fig. 15. The symbols in Fig. 15 indicate the cases
32 B.FAROUKETAL.
0
0.0002
0.0004
0.0006
0.0008
0.001
0.0012
X
m
a
x
/
L
Irregularstreaming
Classical
streaming
0 0.05 0.1 0.15 0.2
H /L
FIG. 15. Map of the cases considered as a function of the normalized maximum wall
displacement and the enclosure aspect ratio. The symbols () indicate the cases considered in
termsofthegeometricalconditionandthenormalizedmaximumwalldisplacement[75].
considered in terms of the enclosure aspect ratio and the normalized max-
imumwalldisplacement.Thedashedlineindicatesthecombinationsofthe
aspectratiosandwallvibrationalamplitudesthatresultintheformationof
the classical acoustic streaming flow patterns characterized by two outer
streaming vortices per half wavelength. Based on our results, the line
defines the upper bound of the geometrical and vibrational amplitude
conditions where the classical Rayleigh streaming patterns form. Clearly,
both the vibrational amplitude and the height of the enclosure (wall
inducedviscouseffects)playrolesintheformationoftheclassicalRayleigh
streaming structures. On the right hand side of the line, the acoustic
streaming structures appear in complex and irregular forms. The number
of streaming rolls increases as the enclosure height and the wall displace-
ment increases.
The formation of pressure (acoustic) waves, acoustic-viscous boundary
layerinteractionsandassociatedflowsinarectangularenclosurearestudied
by solving the unsteady compressible Navier-Stokes equations in 2-D Car-
tesiancoordinatesystem.Theacousticfieldintheenclosureiscreateddueto
theharmonicvibrationoftheleftwall.Theeffectsofthemaximumamount
of wall displacement on the wave field and the formed flow structures are
determined by utilizing a highly accurate flux corrected transport (FCT)
algorithm. Observed primary oscillatory and secondary steady flow fields
demonstratethesignificanteffectsofthewalldisplacementandtheenclosure
height.Flowstructureisstronglydependentonthepressurewaveforminthe
enclosure. For a given enclosure height, increasing pressure amplitude
induces stronger streaming motion (large u
st
/u
R
) and greatly changes the
flow structures. Up to a certain enclosure height, the vibrational motion
causes steady streaming flows, which usually appear as two streaming, roll
per half wavelength as reported in previous studies. However, when the
enclosure height is increased beyond this limit, the streaming structures
becomeirregularandcomplex.
C. FLOWS IN AN ACOUSTICALLY DRIVEN CYLINDRICAL ENCLOSURE
Acoustic streaming motion generated by finite-amplitude resonant oscil-
lationsinanair-filledtwo-dimensionalcylindricalenclosurewasexperimen-
tallystudiedandnumericallysimulated[78].Theoscillatoryflowfieldinthe
enclosurewascreatedbythevibrationofoneendofthecylindricalresonator
(L=325mm,R=12.5mm).Thefrequencyofthewallvibrationwaschosen
as f=1062Hz, such that the corresponding wavelength was equal to the
length of the resonator. A standing wave was then generated in the closed
tube.Thepressurewaveintheaxialpointswasmeasuredbyapiezoresistive
pressure transducer (Endevco #4428A). To simulate the flow field, the full
compressibleformoftheNavier-Stokesequationsincylindricalcoordinates
was considered and solved by a highly accurate flux-corrected transport
algorithm for convection terms and a central differencing scheme for the
viscous and diffusive terms. In both of the experimental and numerical
studies, outer acoustic streaming due to interaction of acoustic waves with
viscous boundarylayerswas observed, andthe effects ofsoundfieldinten-
sityontheformationofstreamingstructureswerestudied.
1. ExperimentalApparatus
Anexperimental systemwas designedandconstructed[79]for theinves-
tigationofthestreamingmotioninair.Theresonatorwasacylindricaltube
madeofPlexiglas.Theinsidediameterofthetubewas25mmandthelength
was 295mm. The acoustic chamber was mounted horizontally with the
sound driver onthe left end andan aluminum plug, which closed the right
end.Thesourceofsoundwavesinthissystemwasacompressiondrivertype
loudspeaker (JBL 2426H). The sinusoidal driving signal of the acoustic
driverwasprovidedbyaGoldStarFG-8002functiongeneratorandampli-
fiedbyaCrownCE1000typepoweramplifier.
AnEndevco8507C-1seriespiezoresistivepressuretransducerwasusedto
detectandquantifytheacousticfield.Thetransducerwasinstalledina1/8
steeltube, whichcouldbemovedalongtheaxis ofthetubetomeasurethe
pressure at different locations. The cross-sectional area of the microphone
was approximately one percent of the resonator area, therefore the error
introduced by the presence of the probe in the sound field was assumed
negligible. An Endevco 4428A pressure signal conditioner was used to
Compression
Driver
Diaphragm
Resonator
295mm
=
2
5
76mm
34 B.FAROUKETAL.
FIG.16. Thecompressiondriverwiththemountedresonator[79].
processthesignalfromthetransducerandprovideexcitedvoltage.Finally,
thepressuresignalwascollectedbyanNI6052EDAQsystem.
Flowvisualizationexperimentsintheresonatorwereperformedbyutiliz-
ingalasersystemandasmokeinjectiontechnique.Thelaserbeamgenerated
byaSpectra-Physics120SHelium-Neon15mWlasersystempassesthrough
acylindricallens.Anexpandedbeamwasthenobtainedtoilluminateathin
horizontallayeroftheresonator.AnAristo-CraftART-29308U-25Bsmoke
unitwasusedtoproducesmokeinthesystem.Thesmokewasinjectedinto
thechamberfromthedriversidethroughaninlethole(seeFig.16).During
thefillingprocedure,asmalldischargingholeontheothersideoftubewas
opened to allow smoke to freely flow into the chamber. After the filling
procedurewascompleted,boththefillinganddischargingholeswereclosed.
A Canon ZR80 digital camera recorder was used to record and store the
visualizedflowpatternsintheexperiments.
2. ExperimentalResults
Figure 17 shows the sketch of theacoustic driver and theresonator with
critical dimensions of the system. As the figure illustrates, the vibrating
diaphragm of the driver is located 76mm behind the driver end of the
cylindricalchamber.
The series of experiments were run under the following ambient condi-
tions: P=0.1017MPa, T=297.15K. First, a number of experiments were
conducted to find a resonant frequency for the cylindrical chamber and
drivergeometry.Figure17indicatesthevariationsofthepressureamplitude
in the right wall of the tube with the driving frequency of the loudspeaker.
The maximum pressure amplitude was received at approximately 1062Hz.
At this frequency, the pressure fluctuations in the cylindrical chamber
1200
1150
1050
1100
1000
950
900
P

(
P
a
)
850
750
800
700
850 900 950 1000 1050 1100 1150 1200 1250
f (Hz)
FIG.17. Variationofpressureamplitudeintherightwallwithfrequencyofloudspeakerfor
drivingpower22.4mW(experimental)[79].
reachedresonance.Sincethespeedofsoundcouldbecalculatedfromtheair
_
temperature,c= RT =345.54m,s,thewavelengthoftheoscillationswas
givenbyl=c/f.Figure18showsthepeak-to-peakpressurevaluesalongthe
axial direction for a loudspeaker power of 91.5mW at 1062Hz. The max-
imum pressure antinode (AN) was located in the middle of the resonator,
with two minimum nodes (N) symmetrically located on the two sides.
According to this rule, we were able to calculate the effective length of
resonator to be about 325mm (indicated by the vertical dashed line in
Fig.16),whichissameasthelengththatwecalculatedfromthewavelength.
The pressure wave was not very symmetrical because the boundary condi-
tionsandgeometriesofthetwoendplatesarenotthesame.Figure19shows
the root-mean-square pressure values along the axial direction for three
different power values with the frequency held at 1062Hz. While the
root-mean-square pressure values decrease with decreasing loudspeaker
power,thepositionsofthenodesandantinodesremainthesame.
Theinstantaneouspressurefluctuationonthenode(z=3L/4)isshownin
Fig. 20 for three different power levels. A sinusoidal pressure wave was
recorded when the power of loudspeaker was low. When the power was
increased, the pressure wave couldnot keep itsregularform andit became
more and more skewed. This signifies the existence of acoustic streaming
because the generation of steady vertical structures destroys the regular
4000
3500
3000
2500
Power:2.133mW
Power:1127.1mW
Power:91.5mW
30 25 0 5 10 20
P

(
P
a
)
2000
1500
1000
500
0
15
Z (cm)
36 B.FAROUKETAL.
1500
1500
500
500
0
P

(
P
a
)
1000
1000
0 5 10 15 20 25 30
Z (cm)
FIG. 18. Peak-to-peak value of the standing pressure wave along the axis of the tube for
loudspeakerpower91.5mWandf=1062Hz(experimental)[79].
FIG.19. RMSvalueofthestandingpressurewavealongtheaxisofthetubefordifferent
loudspeakerpowersandf=1062Hz(experimental).
P

(
P
a
)
300
250
200
150
100
50
0
50
100
150
200
Power:2.133mw
Power:91.5mw
Power:1127.1mw
0.1 0.1005 0.101 0.1015 0.102
t (s)
FIG.20. Instantaneouspressurefluctuationsatz=21.5cm(3L/4)fordifferentloudspeaker
powersandf=1062Hz[79].
harmonicpressuredistribution.Figure21showsthefrequencyspectraofthe
pressurefluctuationsattherightwallofthetubeforadrivingfrequencyof
1062Hz and a loudspeaker power of 91.5mW. A strong first harmonic
frequency was found at 1062Hz, the driving frequency. Weaker higher
harmonicsarealsopresent.
Fortheflowvisualizationexperiments,visualizationresultswererecorded
under driving frequency f=1062Hz and a loudspeaker power of 81.6mW.
The time evolution of the structures is presented here. Figure 22 (a) shows
theseparatedflowfieldstructureintheresonator.Asuddenflowseparation
was observed in the middle section of the chamber once the standing wave
was formed by the loudspeaker. Since the pressure is highest in this zone
(lowestvelocity),thesmokeparticleswerepushedinbothhorizontaldirec-
tions and formed the separated flow regions. A few seconds later, the
formationofthestreamingvorticeswasobservedintheresonator.Figure22
(b) depicts this flow pattern in the half portion of the tube near the right
fixed end. The length of clockwise and counterclockwise circulation was
approximately a quarter wavelength of the standing sound field (as
expected). The four vortices shown in Fig. 22 (b) were found to be nearly
symmetricwithrespecttotheresonatoraxis.
0 1000 2000 3000 4000 5000 6000 7000 8000 900010,000

f (Hz)
10
20
30
40
50
60
70
80
S
P
L
(
d
B
)
90
100
110
120
130
140
150
160

r
Computationaldomain
z
L = =32.5mm,D =2.5mm
38 B.FAROUKETAL.
FIG. 21. Frequency spectra of the pressure fluctuations at the right wall of the tube for
loudspeakerpower91.5mWandf=1062Hz.
FIG.22. Schematicofthecomputationaldomain.
3. NumericalStudy
The time-varying primary flow field in the resonator was numerically
simulated. To simplify the computations, the resonator was modeled as a
cylindricaldomainwhoselengthisfromthedriverendtothefixedend.To
maintain high grid density (as required for the high resolution calculations
here), the scale of the computational model is reduced to 1/10 of the
experimental resonator (L=32.5mm and D=2.5mm). The schematic of
the computational domain is shown in Fig. 23. Due to symmetry, only the
39
0
0
2000
1500
1000
500
0
500
1000
1500
2000
2500
3000
P

(
P
a
)
/2
/2
3/2
3/2

3500
1200
1000
800
600
P

(
P
a
)
400
200
0
200
400
600
800
0 10 20 30
z (mm)
0 10 20 30
z (mm)
ACOUSTICWAVEINDUCEDFLOWSANDHEATTRANSFER
FIG.23. Variationofpressurealongtheaxisoftheenclosureatfourdifferentinstantswith
f=10,620Hz(computational):(a)X
max
=4mm;(b)X
max
=10mm.
tophalfofthedomain wasconsideredforthesimulations.Initially,theair
inthechamberisconsideredasquiescenteverywhere.Att0,theleftwall
starts to vibrate sinusoidally. The displacement of the vibrating wall is
given as
X =X
max
sin(.t) (III2)
Here,X
max
isthemaximumdisplacement;andwistheangularfrequencyof
the vibration, .=2f. Since the computation domain is scaled down by a
factorof10comparedtotheexperimentalsystem,thedrivingfrequencyfor
thecomputationswasscaledupbyafactorof10(f=10,620Hz)sothatthe
wavelengthl=Lforboththemeasurementsandthecomputations.
The input power of the speaker is simulated by adjusting the maximum
displacementoftheleftwall,X
max
.Inthepresentedpaper,thecalculations
arerunfortwovaluesofX
max:
10mmand4mm.Theinitialconditionsofthe
airaresameasthoseintheexperiment:P=0.1017MPa,T=297.15K.For
everycycleofleftwallvibration,about28,778timestepswereused;theflow
isfoundtobequasi-steadyafter100cycles.
Figures 23 and 24 show the pressure and axial velocity distributions,
respectively, along the axis of the tube at wt=0, /2, , and 3/2 for two
driver amplitudes at f=10,620Hz. Both pressure and velocity profiles
repeat as a cycle from .t=0 to 2, and the values at .t=2 are identical
to those at .t=0. The perfect sinusoidal profile of the input sound source
is distorted by the viscous effects between the fluid and solid walls. Since
the lengthof the enclosure is equal to the wavelength, the pressure profiles
are symmetrical with respect to the vertical mid plane. At the beginning of
0
3/2
/2

7
6
5
4
3
2
1
1
2
3
4
5
6
0
v
x
a
x
i
a
l
(
m
/
s
)
0
3/2
/2

7
3
2
v
x
a
x
i
a
l
(
m
/
s
)
1
0
1
2
3
0 10 20 30
z (mm)
0 10 20
z (mm)
40 B.FAROUKETAL.
30
FIG. 24. Variation of axial velocity along the axis of enclosure at four different instants
withf=10,620Hz(computational)[79]:(a)X
max
=4mm;(b)X
max
=10mm.
the cycle (.t=0), the amplitude of the pressure wave is maximum at the
two walls of the enclosure and minimum at the middle of the enclosure.
When.t=,themaximumpressureappearsatthemiddleoftheenclosure
anddecreasestobothofthewalls.Fortheaxialvelocityprofiles,thevalue
changes with time at the left wall due to the vibration of the wall, and is
stationary at the right wall. It has a zero point near the mid zone of the
enclosure.
Figure 25 shows the root-mean-square pressure values along the axial
directionfordifferentvibrationamplitudesandf=10,620Hz.Theseprofiles
are similar to those for the experiments: there is one maximum pressure
antinode(AN)inthemiddleoftheresonator,withtwominimumnodes(N)
symmetricallylocatedonthetwosides.Inaddition,theabsolutevaluesare
foundtoincreasewithincreasingleftwallvibrationamplitude.
Figure 26 compares the temporal pressure distribution at the location
3L/4fortwoleftwallvibrationamplitudeswithf=10,620Hz.Thetimeinter-
valshowninthefigureisonecycle.Thestandingwaveisfoundtobeirregular.
Thelargerthewallvibrationamplitudeis,themoreirregularitbecomes.
The time-averaged (steady acoustic streaming) flow fields are shown in
Fig. 27 (for the top half of the cylindrical enclosure) for two different
vibrational amplitudes and f=10,620Hz. The mean velocity is based on
the average mass transport velocity in the enclosure. The average mass
transportvelocityisgivenby
u
z.mean
=
,u
z
)
,)
. u
r. mean
=
,u
r
)
,)
(III3)
2000
1500
4m
10m
P

(
P
a
)
1000
500
0
0 10 20 30
Z (mm)
1000
800
600
400
4m
10m
0.01805 0.0181 0.01815
t (s)
P

(
P
a
)
200
0
200
400
600
800
0.018
FIG.25. ComputedRMSvalueofthestandingpressurewavealongtheaxisofthetubefor
differentleftwallvibrationamplitudes.
FIG. 26. Computed instantaneous pressure fluctuations at 3L/4 for different left wall
vibrationamplitudes.
(a)
1
r

(
m
m
)
0.5
0
0 10 30 20
z (mm)
10 20 30
z (mm)
(b)
1
r

(
m
m
)
0.5
0
0
42 B.FAROUKETAL.
FIG. 27. Mean flow field in the resonator in the symmetric computational domain,
f=10,620Hz(computational):(a)X
max
=4mm;(b)X
max
=10mm.
Here, ) indicates time-averaged quantities. The time average is applied
duringthe100thcycleofthepressurewave.Asintheexperimentalresults,
fourcircularflowpatternsareobservedineach(toporbottom)halfofthe
resonator.Thefourvorticalstructuresareequallyarrangedalongtheaxial
direction,andtherotationaldirectionsoftheneighborvorticesareopposite.
Due to the large diameter of the resonator, R,c
i
= 57.6, only the outer
streamingisgenerated,andnoinnerstreamingisobserved.
Thecomparison oftheroot-mean-squarevalues ofthestandingpressure
wave along the axis of the tube for the computational and experimental
resultsisshowninFig.28wherel=Lforbothcases.Fortheexperiments,
loudspeaker power was 66.1mW, andfor the computations, wall vibration
amplitude was X
max
=4mm. Since direct measurement of the loudspeaker
vibrational amplitude was not possible, the applied power was varied until
reasonable agreements were obtained for the computations done with the
above vibrational amplitude. The experimental and computational results
arefoundtoagreewell.
4. Conclusions
Theformationofacousticstandingwaves,acoustic-boundarylayerinterac-
tions,andassociatedflowsinacylindricalresonatorarestudiedbyexperimental
0
100
200
300
P

(
P
a
)
400
500
600
700
800
900
1000
Experiment
Computation
0 0.25 0.5 0.75 1
z /L
FIG.28. ComparisonofRMSvalueofthestandingpressurewavealongtheaxisofthetube
forcomputationalandexperimentalresults.
andnumericalmethods.Thedimensionofthecomputationalmodelisreduced
to1/10of thatoftheexperimentalsetupandthecomputationalfrequencyis
increasedto10timestheexperimentalvalue,sothatl=Lforboththeexperi-
ments and the computations. The computational and experimental results
demonstratethesamecharacteristicsofthestandingwavecharacteristicsand
acousticstreamingandmatcheachotherverywell.Forthestudiedresonator
characteristics, the pressure wave is found to be distorted from the perfect
sinusoidalprofileandtheouterstreamingstructuresareobserved.
D. INTERACTIONS OF MECHANICALLY DRIVEN ACOUSTIC WAVES WITH HEAT
TRANSFER IN A RECTANGULAR CHAMBER
Heat transfer in differentially heated enclosures has been studied exten-
sively in the past both experimentally and computationally. However, the
effects of acoustically driven oscillatory flow fields on the convective heat
transfer processes have been primarily studied experimentally. A detailed
understandingofthermalenergytransportintheseproblemsischallenging.
Sound sources whose elements move essentially sinusoidally may generate
a flow field in which the particle velocities are not simply sinusoidal but
44 B.FAROUKETAL.
a pattern of steady vortices (streaming) is often found in the body of the
irradiated fluid. Sound at high intensity levels in gases and liquids are
accompaniedbythesesecondordersteadyflowpatterns.
Acousticstreamingisoftenobservedwhereastandingwaveisformedina
resonator driven by vibrating surface. With suitable design, this steady
vortex flow can be employed for cooling of electronic systems in micro-
gravityenvironmentwherefreeconvectiveflowsinfluidsaregreatlyreduced
or eliminated. Such flows can also aid mixing processes in containers and
augment heat and mass transfer from resonator walls. Acoustically
enhancedconvectioncanbesignificantzero- ormicro-gravityenvironments
whereconductionistheonlyheattransfermode.
In this chapter, acoustic streaming generated by standing waves and the
associatedheattransferingas-filledclosedenclosuresisinvestigatednumeri-
cally. To simulate the flow field, the full compressible form of the Navier-
Stokesequationsisconsideredandsolvedbyahighlyaccurateflux-corrected
transport algorithm. In the numerical model, temperature-dependent heat
conductivity and viscosity are taken into account. The study is focused on
thechannelshavingintermediatewidths(3050timestheviscouspenetration
depth). The simulations are run for different wall temperature conditions:
T
t
=T
b
=T
0
andT
t
=T
b
DT.
1. ProblemGeometry
Atwo-dimensionalrectangularenclosuresfilledwithnitrogenisconsidered
(Fig.29).Theleftwalloftheenclosurevibratesharmonically,eithercomple-
telyorpartially.Thevibratingwallboundaryconditionisthustheacoustic
energy source in this geometry. The displacement of vibration is given as
x=X
max
sin(.t), where .=2f is the angular frequency. The length of the
enclosureischosenasL=2x
0
=8.825mmforallcasesstudied.Thewidthof
the enclosure (H) is varied in the cases studied. The frequency of the wall
vibrationissetasf=20kHz,andthecorrespondingwavelengthofthesound
waves is l=17.65mm based on the undisturbed enclosure temperature
T
0
=300KandpressureP
0
=1atm.Henceforthecasesstudied,l=2L.
Oneof the important parameters inthe resonator isthe viscous penetra-
tiondepthc
i
= (2i,2f)
1,2
=0.0158mm.Herefisthefrequencyofthewall
vibration and i is the kinematic viscosity of the fluid. Previous studies
already showed that the ratio of the resonator width H over c
i
play an
important role in the acoustic-fluid dynamic interactions. In the present
study,theresonatorwidthsarelimitedtotherange:H(=2y
0
):30c
i
to50c
i
(TableIII).Inthiswidthrange,thebehaviorsofacousticstreamingaremost
significant. The width of resonator was kept small such that the viscous
penetrationdepthc
i
canbewellresolvedinthesimulation.
T
top
H
1
y
Nitrogen
x
T
bott
L =/2=2x
0
(a)
T
top
Nitrogen
y
x
T
bott
L =/2=2x
0
(b)
H

=
2
y
0
H

=
2
y
0
FIG.29. Schematicofthecomputationaldomain[80].(a)Entireleftwallvibration;(b)Part
leftwallvibration.
TABLE III
LIST OF CASES STUDIED
No. y
0
/c
i
x
0
/c
i
X
max
(m) H
1
/H DT(K)
1A 279 0
1B 20 1.010
5
1.0 20
1C 60
2 20 0.510
5
1.0 20
3 20 1.510
5
1.0 20
4 15 1.010
5
1.0 20
5 25 1.010
5
1.0 20
6A 20 1.010
5
0.4 0
6B 20
7 20 1.010
5
0.1 0
Inthepresentstudy[80],theaspectratiosofresonatorsL/Hvariedfrom
11to18.Thehighaspectratioisduetotheorderofwavelengthlismuch
higherthanthatofviscouspenetrationdepthc
i
.Toinvestigatetheinteraction
between acoustic waves (propagating primarily in the x-direction) and heat
transfer (in the y-direction), the top and bottom walls are set at different

2
1.5
1
0.5
0
y

/
y
0
0 0.5 1 1.5 2
x/x
0
46 B.FAROUKETAL.
temperatures. As mentioned earlier, the vibrating left sidewall and the right
sidewallsarethermallyinsulated.Typically,agridsizeof15098isusedfor
the rectangular domain. Resolving the boundary layer in the computational
methodisessentialforaccuratelysimulatingtheformationofacousticstream-
ingstructuresandheattransferalongthewalls.Forthisreason,weemploya
non-uniformgridstructure.Thisstructurehasfinestgridsnearthewallsand
thegridqualitygraduallydecreasesasthedistancefromthehorizontalwalls
increases. For the present calculations, there are always 35 grids inside the
boundary layer. To model the wall vibration, moving cell boundaries were
consideredforafewcellsnexttotheleftwall.Thenumberofmovingcellis
chosensuchthatthemaximumvariationofacellsizeislessthan25%.
2. ResultsandDiscussion
a. ValidationResults.Toverifythenumericalmodelandgridsize,thenumer-
icalmodelisfirstusedtosimulatetheanalyticalresultsgivenbyHamilton[39]
under the following conditions: P
r
=0.67, =5/3, y
0
=20c
i
. The predicted
steady streaming flow field (at the end of the 120th cycle) obtained by the
present numerical method (where f=20kHz) for the conditions given in
Hamilton et al. [39] is shown in Fig. 30. This cycle-averaged flow field is
FIG. 30. End wall oscillation induced acoustic streaming (cycle averaged values for the
120thcycle)[80]forconditionsgivenbyHamiltonetal. [39].

2
Hamiltionet al. [39]
1.5
Presentstudy
1
0.5
0
0.5
1
1.5
0.8 1
y /y
0
U
s
t

/
U
R

0 0.2 0.4 0.6
FIG. 31. Comparison of the transverse distribution of x-component streaming velocity
atx/x
0
=1.5[80]withtheresultsbyHamiltonetal. [39].
based on the average mass transport velocity values in the enclosure. The
predictedstreamingpatternandsizearesimilartothosegivenbyHamilton
et al. (not shown here), except that, the left and right vortices are not
symmetricalongthemiddleverticalplaneofresonatorforthepresentcalcula-
tions.Inthepresentcase,onlytheleftwallvibrates,howeverfortheresults
given in Hamilton et al. the entire resonator vibrates, albeit with the same
frequencyandamplitudeconsideredinthepresentsimulations.
Figure31showsthecomparisonofthevariationofthex-componentofthe
cycle-averagedvelocityalongtheenclosureheightatx/(L/2)=x/x
0
=1.5of
the present model predictions with the results given by Hamilton et al.
Figure32givessimilarcomparisonsforthevariationofthey-componentof
thecycle-averagedvelocityalongtheenclosureheightatx/(L/2)=x/x
0
=1.0.
Inthesetwofigures,thecycle-averagedvelocitiesarenondimensionalizedby,
whichisRayleighsresultforthemaximumstreamingvelocityinthecenterof
awidechannelcontainingapureviscous fluid.Themaximum error that is
foundinFig.31is6%,whichcanbeperhapsexplainedbythenon-symmetric
streamingstructureduetothedifferentwallconditionsinthetwostudies.
b. Parametric Case Studies. After the verification of the mathematical for-
mulationandthesolutionprocedure,weinvestigatetheeffectsoftransverse
acousticwavesontheconvectiveheattransferintheenclosurewithimposed
longitudinaltemperaturedifference.TableIIIliststhecasesreportedinthis

1.2
Hamiltonet al.
Presentstudy
1
0.8
0.6
0.4
0.2
0
0.2
0
(
V
s
t

x
0
)
/
(
U
R
y
0
)
0.2
0.4 0.6 0.8 1
y /y
0
48 B.FAROUKETAL.
FIG.32. Comparisonofthetransversedistributionofthey-componentstreamingvelocity
atx/x
0
=1.0withtheresultsbyHamiltonetal. [39].
paper. For each case, the calculations were started by keeping the top and
bottom walls at the same temperatures. After the system achieves steady
state,thenthetopwalltemperatureischangedtoinvestigatetheinteraction
betweenheattransferandacousticallydrivenflows.
Inthefirstcaseconsidered(Case1A),themaximumdisplacementofthe
entire left wall vibration is set to x=X
max
sin(.t) with X
max
= 10mm, and
f=./2=f=20kHz. The width of channel is set to y
0
=20c
i
. At first, the
temperatureofthetopandbottomwallsarekeptattheinitialtemperature
T
0
=300K. The cycle-averaged steady state flow condition is achieved at
aboutthe120thacousticcycle.
Figure33showsthepressuredistributionalongthehorizontalmid-plane
of the enclosure at .t=0, /2, , 3/2 (during cycle # 120) for Case 1A.
These pressure profiles remain essentially the same at any other horizontal
planeawayfrom thebottom andtopwalls andthisindicatesthenear-one-
dimensional character of the acoustic field in the bulk medium. The pres-
suredistributionfor.t=2(notshowninFig.33)isidenticaltothecurve
given for .t=0. At .t=0 and .t=, the amplitude of the pressure waves
reachamaximumandaminimumvalueattheendsoftheenclosure.Atthe
beginning of the cycle (.t=0), the pressure is maximum on the vibrating
(left) wall of the enclosure and decreases with increasing distance from the
wallandreachesaminimumvalueatxL.Thebehaviorisquiteopposite
at .t= where the pressure is minimum on the vibrating (left) wall of the
enclosureandincreaseswithincreasingdistancefromthewallandreachesa
0 2 4 6 8
0
3/2
/2

6000
4000
2000
0
P

P

0
(
P
a
)
2000
4000
6000
8000
x (mm)
FIG. 33. Variation of pressure along the horizontal mid-plane of the enclosure at four
different instants (.t=0, /2, , 3/2) at the 120th cycle, isothermal top and bottom walls,
y /c =20,X =1.010
5
0 v max
(CASE1A).
maximumvalueatxL.TheuvelocityprofilesgiveninFig.34for.t=0,
/2,and3/2,alongthehorizontalmidplaneshowthatthevelocitynode
is formed at approximately x=L/2. Cycle-averaged (120th cycle) solutions
from the present simulations predict steady streaming flows in the enclo-
sure. The predicted steady streaming flow field isshown in Fig. 35. At this
stage, the predicted cycle-averaged velocity fields were found to be cycle-
independent.
Afterthesteadystateisachieved(120thcycle),thetopwalltemperatureis
thenincreasedto320K(DT=20K),whilethebottomwallisstillkeptatthe
initial temperature (Case 1B). Figure 36 shows that cycle-averaged velocity
field (121st cycle) changes drastically immediately after the heating is
initiated.Thesteadystreamingstructureisdestroyedduetothetemperature
changeofthetopwall.Instead,thegasflowsfromtoptobottom,asheating
ofthetopwallincreasesthelocalpressureintheupperpart.Theflowfieldis
foundtoattainsteady-stateconditionsagain(inacycle-averagedsense) by
the 360th cycle. The new steady streaming flow field at the 360th cycle is
showninFig.38,whereonlytwovortices(insteadofthefourshowninFig.36
earlier)arefound.Thus,thetemperaturedifferencebetweenthetopandthe
bottom walls affect the acoustically driven flow field in a significant way.

5
15
10
5
0
3/2

/2
V
x

(
m
/
s
)
0
10
15
0 2 4 6 8
x (mm)
0.6
0.4
0.2
0
y

(
m
m
)
0 1 2 3 4 5 6 7 8 9
x (mm)
50 B.FAROUKETAL.
FIG. 34. Variation of the x-component velocity along the horizontal mid-plane of the
enclosure at four different instants (.t=0, /2, , 3/2) in 120th cycle, isothermal top and
bottomwalls,
y
0
/c
5
v
=20,X
max
=1.010
(Case1A).
FIG. 35. Cycle averaged flow field showing acoustic streaming at the 120th cycle,
isothermaltop
andbottomwalls,y
0
/c
v
=20,X
max
=1.010
5
(Case1A).
Sincethedensity,viscosityandconductivityvaluesofthegasaretemperature
dependent,thesymmetricstreamingstructureinFig.37isdistortedandwe
essentiallyhavearatherasymmetric(cycle-averaged)flowfield.
To further study the effect of temperature difference on streaming velo-
city, we increase the top wall temperature to 360K from 320K (Case 1C).
For the three different heating cases (1A, 1B and 1C), x-component of the
0.6
5 6 7 8 9
y

(
m
m
)
0.4
0.2
0
0 1 2 3 4
x (mm)
FIG. 36. Streamlines (cycle-averaged) immediately after differential heating along the top
andthebottomwalls,
at121stcycle,y
0
/c
v
=20,X
5
max
=1.010
(Case1B).
0.6
5 6 7 8 9
y

(
m
m
)
0.4
0.2
0
0 1 2 3 4
x (mm)
FIG. 37. Acoustic streaming in 360th cycle, differential heating along the top and the
bottomwalls,y
0
/c
v
=20,X
max
=1.0 10
5
(Case1B).
cycle-averagedstreamingvelocityprofilesalongthevertical-planeatx=3L/4
are given in Fig. 38, and the y-component of the cycle-averaged streaming
velocity profiles along the vertical-plane at x=L/2 are shown in Fig. 39.
Comparedwiththeunheatedcase(1A),velocityprofileschangedramatically
duetothestreamingstructurechangesforcases1Band1C.Thestreaming-
velocity components also increase when the top wall temperature increase.
Consequently,differentialheatingincreasesthestrengthofacousticstream-
ing,thoughthesymmetricalstructurefoundinFig.35isdestroyed.
The cycle-averaged temperature contours are shown in Fig. 40 for
T
top
=320K (Case 1B). The effect of the cycle-averaged flow field (Fig. 41)
onthetemperaturefieldisevident.Thevariationofcycle-averagetempera-
ture along the vertical plane of the enclosure at three different locations
(x=L/4,L/2,3L/4)aregiveninFig.41(Case1B)alongwiththeconduction
profilefornoacousticperturbation.Theheattransfertothehorizontalwalls
is found at the middle of resonator, where the mean streaming flow is

U
s
t
(
m
/
s
)
0.2
0
0.2
0.4
Isothermaltopandbottomwallsat
initialtemperature
Heatedtopwall,T =20k
Heatedtopwall,T =60k
0 0.0001 0.0002 0.0003 0.0004 0.0005 0.0006
y (m)
0.6
0.4

V
s
t
(
m
/
s
)
0.06
Heatedtopwall,T=60k
Heatedtopwall,T=20k
Isothermaltopandbottomwallsat
initialtemperature
0.05
0.04
0.03
0.02
0.01
0
0.01
0.0001 0.0002 0.0003
y (m)
0 0.0004 0.0005 0.0006
52 B.FAROUKETAL.
FIG.38. x-componentofthecycle-averagedvelocityalongthevertical-planeatx=3L/4for
differentheatingconditionsalongthetopandbottomwalls(Cases1A,1Band1C).
FIG. 39. y-component streaming velocity along the vertical-plane at x=L/2 for different
fordifferentheatingconditionsalongthetopandbottomwalls(Cases1A,1Band1C).

320
315
Heatconduction
L /4
3L /4
L /2
T

-
a
v
310
305
300
0 0.1 0.2 0.3 0.4 0.5 0.6
y
0.6
y
(
m
m
)
308
318
316
308
306
304
3
1
0
3
1
2
3
1
4
3
1
6
302
318
308
306
316
304
302
306
310
312
314
5 6 7 8 9
0.4
0.2
0
0 1 2 3 4
x(mm)
FIG. 40. Cycle-average temperature contours,at the360thcycle,differentiallyheatedtop
and
bottomwalls,y
0
/c
v
=20,X
max
=1.0 10
5
(Case1B).
FIG.41. Variationofcycle-averagedtemperaturealongtheverticalplaneoftheenclosure
atthreedifferentlocations(x=L/4,L/2,3L/4),atthe360thcycle,differentiallyheatedtopand
bottomwalls,y
5
0
/c
v
=20,X
max
=1.0 10
(Case1B).
strongest.Thecycle-averagedNuvaluesalongthetopandbottomwallsare
found to be 0.998 and 1.231, respectively. The asymmetry is due to the
acoustic power input from the left wall vibration. The acoustic power
inputtothesystemisconvertedtothermalenergybytheviscousdissipation,
which increases the bottom wall heat load, compared to the top wall.
Considering the temperature dependent thermal conductivity, Nu number
is0.975forthecaseofsteadyheatconduction(nowallvibration).Hence,the
0.6
318
318
316
316 316
314
314 314
y

(
m
m
)
0.4 312 312
310
310
308
308
0.2 306 306
304
304
304
302 302
0
0 1 2 3 4 5 6 7 8 9
x (mm)
3
1
4
3
1
6
3
1
8
3
1
0
3
1
2
3
1
0
3
1
4
308
308
0.6 318
316
318
312
y

(
m
m
)
0.4
312 306
304
0.2
306
302
306
304
0
302
0 1 2 3 4 5 6 7 8 9
x (mm)
54 B.FAROUKETAL.
acoustic streaming increases the Nu number by 2.4% in the top wall and
by26.3%forthebottomwallforthegeometryconsideredhere.
To further study the effects of wall vibration amplitude on the heat
transfer, simulations are run for different amplitudes: X
max
=0.510
5
m
(Case2)andX
max
=1.510
5
m(Case3)withDT=20K.Forthedifferent
wallvibrationamplitudes,thereisnosignificantdifferenceonthestreaming
structurecomparedtotheresultsobtainedforcase1B(X
max
=1.010
5
m).
The corresponding cycle-averaged (360th cycle) temperature contours are
shown in Figs. 42 and 43. For lower acoustic power input (Case 2), the
isothermsaresimilartothoseobtainedfortheheatconductioncase.Forhigher
acoustic power input (Case 3) compared to Case 1B, the temperature field
undergoesfurtherdistortiontoaccommodatethelargeracousticpowerinput.
Cases 4 and 5 are run for different channel widths y
0
=15c
i
, and 25c
i
respectivelywithDT=20K.Inthisrange,thepredictedstreamingstructures
donotchangemuchcomparedtoresultsobtainedforCase1B.Howeverthe
FIG. 42. Cycle-average temperature contour, in 360th cycle, differentially heated top and
bottomwalls,y
0
/c
v
=20,X
max
=0.510
5
m(Case2).
FIG. 43. Cycle-average temperature contour in 360th cycle, differentially heated top and
bottom
walls,y
0
/c
v
=20,X
max
=1.510
5
m(Case3).
3
1
0
318
318
0.4
316
316
316
314
314
312
312
310
310
0.2 308 308
306
306
304
304
y

(
m
m
)
304
302 302
0
0 1 2 3 4 5 6 7 8 9
x (mm)
3
1
6
3
1
2
314
3
1
6
3
1
4
3
1
0
3
0
8
3
0
8
3
0
8
3
1
0
3
1
2
3
0
4
3
0
2
3
0
4
304
306
306
306
3
0
2
0.8
318
318
316
314
312
0.6
0.4
0.2
0
y
(
m
m
)
0 1 2 3 4 5 6 7 8 9
x (mm)
FIG. 44. Cycle-average temperature contour, at the 360th cycle, differentially heated top
and
bottomwalls,y
0
/c
v
=15,X
max
=1.010
5
(Case4).
FIG. 45. Cycle-average temperature contour, in 360th cycle, differentially heated top and
bottom
walls,y
0
/c
v
=25,X
max
=1.010
5
(Case5).
temperature contours for y
0
=15c
i
(Fig. 44) and y
0
=25c
i
(Fig. 45) show
significant differences. For a larger enclosure height, the acoustic power
input is higher and the temperature field shows more distortion. The effect
of enclosure height on Nu number is given in the Table IV. Nu along the
bottomwallincreasessignificantlywiththeincreasingchannelwidth,butthe
Nu along top wall does not change much. To compensate for the energy
addition by the vibrating wall, the heat addition from the hot wall to the
fluidislesserthantheheatremovalfromthebottomwall.
Next,wediscussamorerealisticsystemwherethevibrationoftheleftwall
islimitedtothecentralpartofthewall.Toinvestigatethiskindofproblem,
wevibratethepartofleftwallvibration:H
1
/H=0.4and0.1,inthecases6
and7.TheschematicisgiveninFig.29(b).
Figure46(a)showsthecycle-averagedflowfield(360thcycle)forCase6,
where 40% of the left wall vibrates, under unheated (isothermal top and
bottom walls) condition. In the bulk zone, four outer streaming and four
56 B.FAROUKETAL.
Case
1B
4
5
TABLE IV
EFFECT OF ENCLOSURE WIDTH ON NU
y
0
/c
i
Topwall
20 0.998
15 1.016
25 1.098
Bottom
1.231
1.114
1.395
(a)
0.6
0.4
0.2
0
y
(
m
m
)
0 1 2 3 4 5 6 7 8 9
x(mm)
(b)
0.6
0.5
y
(
m
m
)
0.4
0.3
0.2
0.1
0
0 0.1 0.2 0.3 0.4 0.5 0.6
x(mm)
FIG. 46. Acoustic streaming, at 120th cycle, isothermal top and bottom walls, y
0
/c
v
=20,
X
5
max
=1.010
,differentiallyheatedtopandbottomwalls,partial(40%)leftwallvibration,
(Case6A).
innerstreamingrollsareobserved,similartothecaseofthefullwallvibra-
tion(seeFig.35).However,twootheradditionalvorticesarefoundnearthe
centralvibratingpartoftheleftwall.Thesteamingpatternsneartheleftwall
areenlargedintheFig.46(b),whichclearlyshowstheeffectsofthepartial
vibrationontheflowfield.
Followingearliersimulationswithfullwallvibration(Cases1Aand1B),
thetopwalltemperatureisnowincreasedto320K,whilethebottomwallis
held at 300K (Case 6B). The cycle-averaged flow field (after an additional
0.6
318 318
316 316
316
y

(
m
m
)
0.4
314
312
314
312
310
310
310
0.2
306
308
306
308
306
304 304 304
302 302
0
0 1 2 3 4 5 6 7 8 9
x (mm)
0.6
0.4
0.2
0
y

(
m
m
)
(a)
0 1 2 3 4 5 6 7 8 9
x (mm)
(b)
y

(
m
m
)
0.6
0.5
0.4
0.3
0.2
0.1
0
0 0.1 0.2 0.3 0.4 0.5 0.6
x (mm)
FIG.47. Acousticstreaming,at360thcycle,y
0
/c
v
=20,X
5
max
=1.010
,40%ofleftwall
vibrating(Case6B).
360cycles)isgiveninFig.47.Theresultsaresimilartothosereportedearlier
inFig.37(forthefullwallvibration).Theeffectofthepartialwallvibration
is only limited to the vicinity to the left wall, considering. Figure 48 shows
thetemperaturecontourinthesteadystate.Itisclosedtotheresultsofheat
conduction,becauseoftheweakstreamingvelocity.
FIG. 48. Cycle-average temperature contour, in 360th cycle, heating, y
0
/c
v
=20,
X
max
=1.010
5
,40%ofleftwallvibrates(Case6B).
58 B.FAROUKETAL.
0.6
0.4
0.2
0
0 1 2 3 4 5 6 7 8 9
y

(
m
m
)
x (mm)
FIG. 49. Acoustic streaming, in 120th cycle, isothermal top and bottom walls, y
0
/c
v
=20,
X
max
=1.0 10
5
,10%ofleftwallvibrating(Case7).
InCase7,thevibratingpartoftheleftwallisfurtherdeceasedto10%.The
acoustic streaming when the top and bottom walls are kept at the initial
temperatureisshowninFig.49.Thestreamingstructurebecomesirregular
from the case of whole wall vibration, because of stronger nonuniform
boundaryconditionintheleftwallandweakeracousticstreaminginthefield.
3. Conclusions
Thestructureandstrengthofacousticstreamingexcitedbywallvibration,
theeffectofheatingonacousticstreaming,andacousticstreaminginduced
thermalconvectionprocessinanitrogen-filledclosedenclosurewerestudied
numerically. The formation of pressure (acoustic) waves and associated
thermalandflowsphenomenaaredominatedbytheunsteadycompressible
Navier-Stokesequationswithtemperaturedependentviscosityandconduc-
tivity.Thesteadysecond-orderacousticstreamingispredictedbymechani-
callyinducedperiodicoscillationsinthefluid.Afteronewalltemperatureof
theclosedsystemisincreased,thestructureofthestreamingischangeddueto
the nonuniform fluid properties. Furthermore, the streaming velocity
increaseswhenthetemperatureoftheheatedwallisincreased.Hence,heating
isanotherwaytoincreasetheacousticstreaming,otherthanincreasingthe
amplitudeofwallvibration.Toaclosedsystem,thewallvibrationhasmixed
effectsontheheattransferalongthewalls.First,themeanvelocityinducedby
acousticstreamingcanenhancetheheattransfer.Second,theacousticpower
input to the system adds a new heat load to the cooling plate. Therefore,
acousticstreamingisnotalwaysconducivetocoolingdownthehotsource.
Finally,theeffect ofpartial wallvibrationisstudied.Thenonuniform wall
conditionisfoundtobethesignificanteffectonstreamingstructure.Acoustic
streaming introduces an additional convective heat transfer mode into
systemsinazerogravityenvironmentwhereitisassumedthatconductionis
theonlyheattransfermode.Themodeldevelopedcanbeavaluabletoolin
theoptimized designofheatexchangerswithacousticallyenhancedperfor-
manceandotherapplicationsofacousticstreaming.
IV. NumericalStudyofThermallyInducedAcousticWavesinGases
A. INTRODUCTION
Thephysicsofthermallyinducedacousticwavesandtheeffectsofgravity
on the flows induced are considered in this section. First, the early time
behaviorofacousticwavesproducedbyrapidheatingincarbondioxidegas
arestudiedforinitial pressuresranging from 0.1MPa to8.0MPa.Here we
trytounderstandtheeffectsofthevariationoffluidpropertieswithpressure
onthegeneration,propagationanddampingofthermallyinducedacoustic
waves. Then the flow fields generated by thermally induced waves under
different gravity conditions are studied. The effects of gravity are investi-
gatedundertwocategories:sidewallheatingandbottomwallheating.
B. THERMALLY INDUCED ACOUSTIC WAVES IN ATMOSPHERIC AND HIGH
PRESSURE GASES
1. PropertyVariations
An earlier study [54] showed that the behaviors of thermally induced
acousticwavesaresomewhatdifferentfordifferentgases,duetothevaria-
tion of their properties. It is known that the effect of pressure on other
thermodynamicpropertiesisverysmallwhenthegaspressureislow.How-
ever, for a high-pressure fluid such as a real gas or super-critical fluid, the
effect of pressure cannot be neglected. In this section, we investigate the
behaviorofthermallyinducedacousticwavesforcarbondioxidegasunder
differentinitialpressures.Thepressurerangeisfrom0.1MPato8MPa.
Forhigh-pressuregasandsuper-criticalfluid,theP-,-Trelationsarequite
differentfromtheidealgaslaw,andtheinternalenergyandsoundspeedare
functionsoftemperatureanddensity(orpressure).Insteadofusingpublished
correlationsforhighpressuregases,whicharelimitedinapplicability,weuse
theNISTStandardReferenceDatabase12[81] to evaluate the P-,-Trelations
andotherthermodynamicpropertiesforrealgasesandsupercriticalfluids.
Figure50givesthevariationofdensityforcarbondioxideasafunctionof
pressure (from 0.1 to 8.0MPa) for different temperatures. The density is
strongly nonlinear with respect to pressure. Figure 51 shows the variation
350
300
250
T =310K
T =320K
T =330K
6000 8000

(
k
g
/
m
3
)
200
150
100
50
0
0 2000 4000
P (kPa)
60 B.FAROUKETAL.
FIG. 50. Variation of density for carbon dioxide as a function of pressure at constant
temperatures.
180,000
160,000
140,000
T =310K
T =320K
T =330K
i

(
J
/
k
g
)
120,000
100,000
80,000
60,000
0 2000 4000 6000 8000
P (kPa)
FIG. 51. Variation of internal energy for carbon dioxide as a function of pressure at
constanttemperatures.

300
T =310K
T =320K
280
T =330K
260
240
220
200
180
0 2000 4000
P (kPa)
c
(
m
/
s
)
6000 8000
FIG.52. Variationofsoundspeedforcarbondioxideasafunctionofpressureatconstant
temperatures.
of internal energy as a function of pressure at different temperatures. It is
interestingtoobservethatathighpressure,theinternalenergyisnotonlya
functionoftemperature,butalsoafunctionofpressure(density).Theother
thermodynamicproperties,suchassoundspeed,viscosityandheatconduc-
tivityarealsoshowninFigs.5254.Allfluidpropertiesshowstrongdepen-
dence on pressure, and these relationships become more dramatic when
pressureapproachesthecriticalpoint,P
crit
=7.377MPa,T
crit
=304.128K.
TousethedataprovidedbytheNISTStandardReferenceDatabase12in
ourcomputationalmodel,polynomialfitswereobtainedtobuildtheprop-
erty equations in the specific data regions. The six relation equations are
usedinthenumericalmodels:P=f(T,,),i=f(T,,),c=f(T,,),j=f(T,,),
k=f(T,,),T=f(i,,).
2. ProblemStatement
The effects of pressure on the thermally induced acoustic waves are
reportedforatwo-dimensionalcavity(L=0.001m)containingcarbondiox-
ide. As shown in Fig. 55, the sidewalls of the cavity are held at specified
temperatures and the horizontal surfaces are thermally insulated. Initially,

26
24
22
T =310K
T =320K
T =330K
6000 8000
P
a
-
s
)
20
18
16
0 2000 4000
P (kPa)

60
50
T =310K
T =320K
T =330K
6000 8000
K
(
m
W
/
m
-
K
)
40
30
20
10
0 2000 4000
P (KPa)
62 B.FAROUKETAL.
FIG. 53. Variation of viscosity for carbon dioxide as a function of pressure at constant
temperatures.
FIG. 54. Variation of heat conductivity for carbon dioxide as a function of pressure at
constanttemperatures.
X
L
0
y
T
L
(t )
T
R L

Co
2
_ _ _ _
_ _
FIG.55. Geometryandboundaryconditionsoftheproblem.
the fluid is quiescent (u=v= 0 everywhere) at a constant temperature and
pressure (P
0
and T
0
). At the given time, the temperature along the left
vertical wall is raised impulsively or gradually, while the right wall is kept
attheinitialtemperature.Thistriggersathermallyinducedacousticwavein
the medium. The temperature history of the left wall is governed by the
followingequation:
T
max
T
0
t
t
h
T
L
(t) = T
0
1 1e (IV.1)
T
0
InEq.(IV.1),T
max
isthemaximumtemperaturethattheheatedleftwallcan
attain.Theincreaserateofthewalltemperaturecanbecontrolledbyvarying
the time constant t
h
value in Eq. (IV.1). All the cases presented in this
sectionwerecarriedoutforazerogravitycondition.
3. ResultsandDiscussions
Tostudythebehaviorofthethermallyinducedacousticwavefordifferent
fluidpressures,eightcasesaresimulated,asshowninTableV.Forallcases,
theinitialtemperatureintheenclosureisassumedtobe310Kandtheright
walltemperaturewasimpulsivelyraisedto320K.ForCase1,P
0
=0.1MPa
where ideal gas assumptions can be used. From Case 2 to 7, the initial
enclosure pressure is increased gradually and P
0
=8MPa for Case 8. For
this case (T
0
=310K, P
0
=8MPa) carbon dioxide exists as a supercritical
80
NISTdatabase
Idealgaslaw
60
40
20
0
0 5E-06
Time(s)
P

P

0
(
P
a
)
1E-05 1.5E-05
64 B.FAROUKETAL.
TABLE V
LIST OF COMPUTATIONAL CASES
Case T
0
(K) P
0
(MPa) Propertiestable T
max
(K) t
h
(s)
,(kg/m
3
) T(K)
1
2
3
4
5
6
7
8
310
310
310
310
310
310
0.1
1
3
5
7
8
1.621.90
16.918.4
5561
78120
170221
210450
308324
308324
308324
308324
308324
308324
320
320
320
320
320
320
3.5710
7
3.5710
7
7.1410
7
3.5710
7
3.5710
7
3.5710
7
fluid.Thedensityandtemperaturerangesshowedinthetablewereusedto
constructthepropertyrelationsfromtheNISTReferenceDatabase12.
Thetimeconstantt
h
=3.5710
7
sforCase1.ForthisCase,theidealgas
law applies. Figure 56 shows the computed pressure variation at the mid
pointoftheenclosureforCase1.Resultsareshownforthecaseswherethe
FIG. 56. Variation of pressure at the mid point of the enclosure: (a) ideal gas law and
(b)NISTdatabase,case1(P
0
=0.1MPa,t
h
=3.5710
7
s).
800
600
400
200
0
0 5E-06 1E-05 1.5E-05
Time(s)
P

P
0
(
P
a
)
2E-05 2.5E-05 3E-05

P-,-TrelationiseitherobtainedfromtheidealgaslawortheNISTRefer-
ence Database 12.The two relations predict thesame wave speed and very
similar wave shape. The pressure in the middle point remains unchanged
until the acoustic wave reaches it. The local pressure rises almost instanta-
neously, and then decreases gradually. After the pressure wave passes the
middle point, it hits the right wall and then is reflected. Eventually, these
repeated oscillations are damped due to viscous dissipation. The overall
pressureinthechamberincreasesbecauseoftheenergyinputtothesystem
bytheheatedleftwall.
Fortheremainingcases(highpressure),weobtainedtheP-,-Tandother
propertyrelationsfrom the NISTReference Database 12.Figure 57shows
the variation of pressure with time at the mid point of the enclosure for
Case 3 (P
0
=3MPa, T
0
=310K). The corresponding of time variation of
x-velocitycomponentatthemiddlepointisgiveninFig.58.
The effect of the rapidity of the wall heating process on the thermally
inducedwavebehaviorwasinvestigatedandtheresultsareshowninFig.59.
Theheatingrateiscontrolledbyselectingthevalueoft
h
inEq.(IV.1).The
rapidity of the wall heating process has a very significant effect on the
strengthofthepressurewave.
FIG. 57. Variation of pressure with time at the mid point of the enclosure, P
0
=3 MPa,
t
h
=3.5710
7
s(Case3).

600
500
400
h
=3.5710
7
s
h
=7.1410
7
s
h
=7.1410
6
s
P

P
0
(
P
a
)
300
200
100
0
0 5E-06 1E-05 1.5E-05
Time(s)
66 B.FAROUKETAL.
10
5
0
5
10
15
20
V
x
(
m
m
/
s
)
0 5E-06 1E-05 1.5E-05 2E-05 2.5E-05 3E-05

Time(s)
FIG. 58. Variation of velocity with time at the mid point of the enclosure, P
0
=3 MPa,
t
h
=3.5710
7
s(Case3).
FIG.59. Variationofpressureatthemidpointoftheenclosurefordifferentheatingrates,
P
0
=3MPa(Cases3,4and5).

P
0
(
P
a
)
5000
4000
3000
2000
1000
0
P
0
=8MPa
P
0
=7MPa
P
0
=5MPa
P
0
=3MPa
P
0
=1MPa
P
0
=0.1MPa
0 5E-06 1E-05 1.5E-05
Time(s)
FIG. 60. Variation of pressure at the mid point of the enclosure for different initial
pressures,t
h
=3.5710
7
s(cases1,2,3,6,7and8).
The effect of the initial pressure on the thermally induced acoustic
wave is shown in Fig. 60 for cases 13 and 68 (Table V). All six cases
have the same initial temperature, maximum wall temperature and heat-
ing rate. Carbon dioxide is at supercritical conditions for Case 8.
Although the shapes of pressure wave are very similar for all cases,
stronger acoustic waves are observed when the initial pressure is higher.
The pressure also increases faster in cases with higher initial pressure.
It is also observed that the waves damp out much faster for higher
initial pressure conditions, because of higher viscosity. No qualitative
difference is observed for Case 8 (supercritical fluid conditions). It is
noted that initial conditions for Case 8 (P
0
=8 MPa, T
0
=310K) is not
near the critical point (P
crit
=7. 377MPa, T
crit
=304.128K). The beha-
vior of thermoacoustic waves near the critical point are further explored
in Section VI.
Figure 61 shows the variation of temperature at the mid point of the
enclosure for the above cases as shown in Fig. 60. It is interesting to
note that the strongest temperature wave occurs when P
0
=0.1MPa, then
that the wave amplitude decreases with initial pressure increase until
the weakest temperature wave is observed in the case with P
0
=3 MPa.
After that, the temperature waves become stronger with initial pressure
increase.

T

(
K
)
0 5E-06 1E-05 1.5E-05

Time(s)
310.06
310.05
310.04
310.03
310.02
310.01
310
P
0
=0.1MPa
P
0
=8MPa
P
0
=7MPa
P
0
=1MPa
P
0
=5MPa
P
0
=3MPa
68 B.FAROUKETAL.
FIG. 61. Variation of temperature at the mid point of the enclosure for different initial
pressures,t
h
=3.5710
7
s(Cases1,2,3,6,7and8).
C. INTERACTIONS OF THERMALLY INDUCED ACOUSTIC WAVES WITH BUOYANCY
INDUCED FLOWS: SIDE-WALL HEATED ENCLOSURES
1. ProblemStatement
The flow field generated by a thermally induced acoustic wave and the
effect of gravity on this field are studied in a square enclosure filled with
nitrogengasat1atm.and300K. Theheatingisappliedontheleftwallof
the square enclosure, which has a side of length L=13mm (Fig. 62). The
horizontal walls of the square enclosure are adiabatic. Initially the gas and
all of the walls are in thermal equilibrium (T=T
R
everywhere). At later
times t 0, the left-wall temperature is increased to T
L
(T
L
T
R
), either
suddenly or gradually. We examine the effects of increases in temperature
thatareeitherspatiallyuniformornonuniform.
Numericalsimulationsoftheflowsgeneratedbythermallyinducedacous-
tic waves were performed in a square enclosure filled with nitrogen gas at
1 atm. and 300K [82]. Results were obtained for four cases: uniform

N
2
T
L
(t,y)
T
R
y
0
x
L =13mm
L

=
1
3
m
m
FIG.62. Geometryandboundaryconditionsoftheproblem.
impulsiveheatingwithoutgravity,uniformgradualheatingwithoutgravity,
spatiallynonuniformgradualheatingwithoutgravity,andspatiallynonuni-
formgradualheatingwithgravity.
Resultsofapriorinvestigation[53]ontheveryshorttimebehaviorofthe
thermallyinducedacousticwavesgeneratedbyimpulsiveandgradualheat-
ing of a wall were in very good agreement with the results given in the
literature. In the present study, longer time behavior of the velocity fields,
producedbyeitherstepimpulsiveheatingorgradualheatingoftheleftwall
was investigated under zero- and normal-gravity conditions. In the numer-
icalscheme,impulsiveheatingcanbeapproximatedbychangingthevalue
of the temperature in the first time step. In the present computations, the
firsttimestepissmall,2.8110
8
s,comparedtothecharacteristicacoustic
anddiffusiontimesofthesystem.
a. Uniform Impulsive Heating at Zero Gravity. For impulsive heating, the
temperatureoftheleftwallisgivenas
T
L
(t) = T
0
[1 A[ (IV.2)
wheretheoverheatratioAisgivenby
T
L
T
0
A= (IV.3)
T
0
andisequalto1/3forallcasespresentedforspatiallyuniformwallheating.
The variation of pressure at the mid point of the enclosure is shown in
Fig.63forearlytimes;adistinctivepeakisobservedwheneverthethermally
inducedacousticwavecrossesthemidpoint.Thiswaveperiodicallyreflects
101,800
201 201
241 241
281 281
101,700
101,600
101,500
P

(
P
a
)
101,400
101,300
101,200
0 5E-05 0.0001 0.00015 0.0002
Time(s)
70 B.FAROUKETAL.
FIG.63. Variationofpressureatthemidpointoftheenclosureforthreegridsizes,under
spatiallyuniformimpulsiveheating,zerogravity[82].
from both sidewalls as seen by the multiple peaks. We have shown earlier
thatthestrengthofthepressurewaveisstronglycorrelatedtotheoverheat
ratio and that the pressure oscillations damp out with increasing time.
Computationalcellswerenonuniformlydistributedwithfinegridsnearthe
boundary,andtheratioofthemaximumtominimumgridsizewas5.0.The
results for a resolution study using three mesh sizes, 201201, 241241
and281281,arealsoshowninFig.64;theinitialtimestepwas,however,
thesameforallthreemeshsizestorepresentidenticalimpulsiveheatingfor
allcases.Fromthisfigure,ameshsizeof241241wasfoundtobeadequate
for the present computations. The time variation of the pressure and the
x-componentofthevelocityatthemidpointoftheenclosureareshownin
Fig. 65 a short time after the impulsive heating. The x-component of the
velocity fluctuates between positive and negative values, depending on
thedirectionoftheoncomingpressurewave.
ThevelocityvectorsinFig.65showtheflowfieldthatdevelopedfromthe
wave motion created by impulsive heating at two different instants (25.78
and 100ms).Dueto spatiallyuniform heatingand zero-gravity conditions,
theflowfieldisessentiallysymmetricalongthehorizontalmidplane.Flow
structures develop along the top and bottom left corners due to viscous
0.8 101,800
0.6
101,700
0.4
0.2
101,600
P
u
0.0004 0.0005
P

(
P
a
)
(
m
/
s
)
0
101,500
0.2
0.4
101,400
0.6
101,300
0.8
0 0.0001 0.0002 0.0003
t (s)
ACOUSTICWAVEINDUCEDFLOWSANDHEATTRANSFER 71
FIG. 64. Variation of pressure and the x-component of velocity at the mid point of the
enclosureforearlytimes,underspatiallyuniformimpulsiveheating,zerogravity.
interactions between the solid surfaces and bulk fluid. The strength of the
flowfielddecreasesconsiderablyat100mscomparedtoitsvalueat25.78ms.
In Fig. 66, we show the temporal variation of the x-component of the
velocity at the mid point for 100ms. The thermally induced acoustic wave
undergoesmanyreflectionsfromthesidewallsduringthisperiod.Thedark
regionsreflectthehighlyoscillatorynatureoftheflowfield.Thex-velocity
component decays with every reflection and by 100ms, the value becomes
quite small; however, a flow field still exists within the enclosure as shown
earlierinFig.65(b).
b. Spatially Uniform Gradual Heating under Zero Gravity. The thermal
inertia of walls and heating systems as well as unavoidable heat losses to
the environment make it difficult in practice to generate a step change
(impulsive heating) in the wall temperature. Here we study the effect of
more gradual heating by using an exponential expression for the wall
heating:
_ _ __
t
T
L
(t) = T
0
1 A 1e
t
h
(IV.4)

4
5
6
7
8
(a) (b)
13 13
12 12
11 11
10 10
9 9
8 0.005m/s
Y

(
c
m
)
Y

(
c
m
)
7
6
0.01m/s
5
4
3 3
2 2
1 1
0
1 2 3 4 5 6 7 8 9 10 11 12 13
0
1 2 3 4 5 6 7 8 9 10 11 12 13
X(cm)
t = 25.78ms
X (cm)
t =100ms
0.6
0.4
0.2
0.08 0.1
u
(
m
/
s
)
0
0.2
0.4
0.6
0 0.02 0.04 0.06
t(s)
72 B.FAROUKETAL.
FIG.65. Velocityvectorsattwodifferenttimes,underspatiallyuniformimpulsiveheating,
zerogravity[82]:(a)27.58ms;(b)100ms.
FIG. 66. Variation of the x-component of velocity with time at the mid point of the
enclosure,underspatiallyuniformimpulsiveheating,zerogravity.
Here,t
h
isatimeconstant.Forimpulsiveheating,t
h
tendstozero.Figure67
shows the variation of the pressureandthe x-component of the velocity at
themidpointforearlytimeswithgradualwallheatingandatimeconstant
t
h
=5t
c
, where t
c
is the travel time of sound waves over the length of the
enclosure.Inthesamefigure,thevariationoftheleftwalltemperaturewith
101,800
Pressureatthemidpoint
0.06 420
uatthemid-point
Temperatureattheleftwall
0.05 400
101,700
0.04 380
101,600
P

(
P
a
)
0.03
u

(
m
/
s
)
360
T

(
K
)
101,500
0.02 340
101,400
0.01 320
101,300
0 0.0002 0.0004 0.0006 0.0008
0 300
T (s)
_ _
_ _
FIG. 67. Variation of pressure and the x-component of velocity at the mid point of the
enclosureforearlytimes,underspatiallyuniformgradualheatingwitht
h
=5t
c
, 0 g.
timeisshown.ComparingFig.67withFig.62wenotethatforgradualwall
heating,thethermally inducedacoustic waveismuch weaker.Asshownin
Fig.67,thepressurevaluesatthemidpointincreasecontinuously;however,
discretejumpsinitsvalueareobservedeachtimethewavecrossesthecenter.
Theeffectofgradualheatingonthex-componentofthevelocityisinterest-
ing. The velocity component still fluctuates with time, but does not reverse
signasinFig.62,andhencethedampingrateisslowerinthiscase.
Figure68showsvelocityvectorsattwodifferenttimes(25.78and100ms)
forthecasedescribedabove.Thevelocityfieldisfoundtobesimilartothat
for the case of impulsive heating; it is symmetric along the horizontal mid
plane,anddecayswithtime.
c. Spatially Nonuniform Gradual Heating at Zero Gravity. Forthespatially
nonuniformgradualheatingsituation,theleftwalltemperatureisnowgivenby
Ly
t
t
h
T
L
(t) = T
0
1 2A
1e (IV.5)
L
wherealinearvariationofthespatialtemperaturedistributionisconsidered.
6
5
4
3
2
1
0
Y 0.005m/s 0.005m/s
0
1
2
3
4
5
6
7
8
9
10
11
12
13
0 1 2 3 4 5 6 7 8 9 10 11 12 13 0 1 2 3 4 5 6 7 8 9 10 11 12 13
t =25.78ms t =100ms
X X
(a) (b)
13
12
11
10
9
8
7
Y
74 B.FAROUKETAL.
FIG. 68. Velocity vectors at two different times, under spatially uniform gradual heating
witht
h
=5t
c
,zerogravity:(a)27.58ms;(b)100ms.
Figures69(a)(d)showthedevelopmentofvelocityvectorsfrom0.074ms
to 100ms. At 0.074ms (Fig. 69 (a)), all fluid flows in the direction nearly
normal to the vertical wall because the pressure disturbance is generated in
thehorizontaldirection.Figure69(a)alsoshowsthatthehorizontalvelocityin
thelowerregionoftheenclosure(withhigherwalltemperature)ishigherthan
thatintheupperregion,whichdemonstratesthattheacousticwave-induced
flowisproportionaltothetemperatureincreaseonthewall.Later,7.37ms,
thereisarecirculatingflowpatternatthelefttopcorner,asshowninFig.69
(b).Withincreasingtime,thisvorticalflowfieldgrowsandcoverstheentire
enclosure.Evenat100ms,theflowfieldisquitestrong.Wesawearlierthat
spatiallyuniformgradualheatingcangeneratealongerlastingflowfieldthan
impulsiveheatingwiththesameoverheatratio.Hereweseethatthestrengthof
thisflowfieldcanfurtherbeenhancedwithspatiallynonuniformheating.
d. Spatially Nonuniform Gradual Heating under Normal Gravity. Now we
consider the effect of gravity on the flows induced by acoustic waves. The
conditionsconsideredhereareidenticaltothoseconsideredabove,albeitat
normal gravity conditions. Figures 70 (a)(d) show the development of
velocity vectors in the enclosure from 0.074ms to 100ms. At t=0.074ms,
Fig.70(a),andt=7.37ms,Fig.70(b),thepredictedflowfieldsaresimilar
to the results shown in Figs. 69 (a) and (b) under zero-gravity conditions.
Thebuoyancy-inducedflowtakescomparativelylargertimetodevelopthan
thethermoacousticwave-inducedflows.TheflowfieldshowninFig.70(c)
att=25.78msshowsthesignsofadevelopingbuoyancy-inducedflowalong
the left lower corner of the enclosure. With increasing time (t=100ms,

7
(a) (b)
13 13
12 12
11 11
10 10
9 9
8 8
0.05m/s 0.01m/s
YY
7
6 6
5 5
4 4
3 3
2 2
1 1
0
0 5 10
0
0 5 10
X t =0.074ms X t =7.37ms
(c) (d)
13
13
12
12
11
11
10
10
9
0.01m/s 9
8
8
Y
7
Y
7
0.01m/s
6
6
5
5
4
4
3
3
2
2
1
1
0
0
0 5 10
0 5 10
X t =25.78ms
X t =100ms
FIG. 69. Velocity vectors at four different times, under spatially non-uniform gradual
heatingwitht
h
=5t
c
,zerogravity:(a)0.074ms(b)7.37ms(c)25.78ms(d)100ms.
Fig.70(d)),thisvorticalflowfieldenlargesandcoverstheentireenclosure.
The overall strength of the flow field in the enclosure at t=100ms under
zero-, Fig. 69 (d), and normal-, Fig. 70 (d), gravity conditions are quite
similar.Thedirectionsoftheverticalflows,however,areinoppositedirec-
tions. With increasing time, the flow field under normal-gravity conditions
willrepresentapurelybuoyancyinducedflowfield,andtheflowfieldunder
zero-gravityconditionswilldampout.
3. Conclusions
TableVIsummarizestheresultsforthecasesstudied.Impulsiveheatingis
foundtogeneratestrongthermoacousticwaves;however,thedampingratefor
theresultingvelocityfieldishigh.Gradualheatingconditionsstudiedproduce

7
6
(a) (b)
13
13
12
12
11
11
10
10
9
9
8
8
5
0.05m/s
5
4
4
3
3
2
2
1
1
0
0
0 5 10 0 5 10
x t =0.074ms x t =7.37ms
(c) (d)
13 13
12 12
11 11
10 10
9 9
0.01m/s
8 8
0.01m/s
7
YY

6
7 7
YY
0.01m/s
6 6
5 5
4 4
3 3
2 2
1 1
0 0
0 5 10 0 5 10
x t =25.78ms x t =100ms
76 B.FAROUKETAL.
FIG. 70. Velocity vectors at four different times, under spatially non-uniform gradual
heatingwitht
h
=5t
c
,normalgravity:(a)0.074ms;(b)7.37ms;(c)25.78ms;and(d)100ms.
weakeracousticwavesbuttheytendtoproduceflowfieldsthatcansustainfor
longerperiods.Spatiallynonuniformheatingisfoundtoproducefasterflows
inthe region with higher heating and eventually results inrecirculating flow
fieldsintheenclosure.Fortheconditionsstudied,theeffectofgravityonthe
flow fields is only evident after about 25ms from the start of the heating
process. Buoyancy induced flow field strength at 100ms was similar to that
obtainedbythermallyinducedacousticwave-inducedflowatzerogravity.
D. INTERACTION OF THERMALLY INDUCED ACOUSTIC WAVES WITH BUOYANCY
INDUCED FLOWS: BOTTOM-WALL HEATED ENCLOSURE
1. ProblemStatement
Inthissection,theinteractionsofthermallyinducedacousticwaveswith
the buoyancy induced flow are reported for bottom heated rectangular
TABLE VI
SUMMARY OF THE MAIN OBSERVATIONS [82]
Case Description Observationsfromresultsfor Observationsfromresultsat
shorttimes longertimes
1 Uniformimpulsive
heatingatzerogravity
2 Spatiallyuniform
gradualheatingunder
zerogravity
3 Spatiallynonuniform
gradualheatingat
zerogravity
4 Spatiallynonuniform
gradualheatingunder
normalgravity
1.Strongthermallyinduced
pressurewaves
2.Horizontalvelocity
componentreversessign
afterreflectionfroma
sidewall
1.Weakthermallyinduced
pressurewaves
componentfluctuatesin
valueafterreflectionfroma
sidewallbutdoesnot
reversesign
1.Pressurewavesare
insensitivetothevariation
oftheheatingratealongthe
verticaldirection
componentvaluesare
higheratthelowerregion
oftheenclosure(higher
heatingrate)
1.Effectofgravityisnot
observed
2.Pressurewavesare
insensitivetothevariation
oftheheatingratealongthe
verticaldirection
componentvaluesare
higheratthelowerregion
oftheenclosure(higher
heatingrate)
1.Essentiallyone-dimensional
flowfieldisdevelopedin
theenclosure
2.Thedampingrateforthe
flowfieldishigh
1.Two-dimensionalflowfield
isdevelopedinthe
enclosure,nearthecorners
oftheheatedwall
flowfieldisslowandthe
flowfieldappearstobe
quasi-steady
1.Two-dimensionalcounter-
clockwisevorticalflowfield
isestablished
flowfieldislow
1.Effectofgravity(buoyancy)
becomesevident
2.Thecounter-clockwiseflow
fieldcreatedbyspatially
nonuniformheatingis
replacedbyastrong
clockwisevorticalflow
3.Overallstrengthoftheflow
fieldat100msissimilarto
thatobservedinCase3.
enclosures [83]. We consider enclosures with varying aspect ratios:
H/L=0.5, 1.0 and 5.0. The vertical walls of the rectangular enclosure are
considered to be insulated. Initially the gas and all walls are in thermal
_ _
_ _
78 B.FAROUKETAL.
equilibrium (T=T
0
everywhere). At t 0, the bottom wall temperature is
increasedrapidlytoavalueT
b
(T
b
T
0
)inthefollowingmanner:
T
max
T
0
t
t
h
T
b
(t) = T
0
1 1e (IV.6)
T
0
Meanwhile,thetopwallisheldattheinitialtemperature(T
t
=T
0
).InEq.
(IV.6), T
max
is the maximum temperature that the heated bottom wall can
attain.Therateofincreaseofthebottomwalltemperaturecanbecontrolled
byvaryingthetimeconstantt valueinEq.(IV.6).Thisisimportantbecause
the temperature increase rate has a significant effect on the strength of a
thermallyinducedacousticwave,accordingtopreviousstudies[53].Forall
calculationsreportedinthispaper,thevalueof(T
max
T
0
),T
0
issetto1/3.
To study the effect of gravity on the flow field generated by a thermally
induced acoustic wave, the computations were carried out for different
gravityconditions,viz.,1.0 g,0 gand1.0 g.
Numericalsimulationsofthermallyinducedwavesandbuoyancyinduced
flow were performed in a rectangular enclosure filled with air. The initial
pressure of air is 1 atm., and the initial temperature (T
0
) is 300 K.
The bottom wall is heated to 400K (T
max
) at different heating rates
(varying t
h
), while the top wall is kept at the initial temperature. Results
wereobtainedforthefollowingsevencases(TableVII)tostudytheeffectsof
gravity, heating rate, and aspect ratio (H/L) on the flow field and the heat
transfercharacteristics.Forallcasesstudied,thewidthoftheenclosurewas
consideredtobeL=0.02m.
Foramesh-resolutionstudy,thecomputationalresultsforthevariationof
temperature with y at the middle plane of the enclosure at t=1.27s, using
three different mesh sizes, 8080, 100100 and 120120, are shown in
Fig.71forCase1(seeTableVII).Thetimestep(2.8110
8
s)wasthesame
forallthreemeshsizes.Fromthisfigure,ameshsizeof100100wasfoundto
beadequateforthepresentcomputationsforcaseswithH/L=1.0.Forcases
withnon-unityaspectratios(TableVII),themeshsizeswereadjustedsuchthat
thegriddensityisapproximatelythesameasfoundinthebasecase.
a. Effects of Gravity. Simulations were carried out for identical conditions
withthreedifferentvaluesof,viz.,1.0g,0g,and1.0g(Cases1,2and3,
respectively). The aspect ratio of the enclosure is fixed at 1.0 for all of
the above three cases, and the time constant of the heating process is also
keptat5.7610
5
s.Thisvalueisgivenbythetimeforanacousticwaveto

400
380
360
T

(
K
)
340
320
300
10 15
80 80
120 120
100 100
0 5 20
y (mm)
TABLE VII
LIST OF COMPUTATIONAL CASES
L(m) H/L Heatingratet
h
(s) Gr
H
Gravity (g)
1 0.02 1.0 5.7610
5
5.2710
4
1.0
2 0.02 1.0 5.7610
5
0
3 0.02 1.0 5.7610
5
5.2710
4
1.0
4 0.02 1.0 2.8810
4
5.2710
4
1.0
5 0.02 1.0 1.1510
2
5.2710
4
1.0
6 0.02 0.5 5.7610
5
6586 1.0
7 0.02 5.0 5.7610
5
6.5910
6
1.0
FIG.71. Variationoftemperaturewithyalongthehorizontalmidplaneoftheenclosureat
t=
1.27 s(H/L=1,t
h
=5.76 10
5
s, 1.0g),bydifferentgridsizes.
travelfromthebottomoftheenclosuretothetopundernormalatmospheric
conditions.
The predicted variations of pressure atthe midpoint ofthe enclosurefor
allthreecasesareshowninFigs.72(a)and(b).Adistinctivepeakisobserved
whenever the thermally induced pressure wave crosses the mid point.
This wave, traveling at slightly supersonic speed, periodically reflects from
both sidewalls as seen by the multiple peaks. The shapes of the pressure
wavesarealsofoundtobenearlysameforthedifferentgravityconditionsat
early times (Fig. 72 (a)). The pressure variations are plotted over a larger
(a) (b)
500
+1g,0g,and1g
12000
10000
400
0g, 1g
+1g
8000
P

(
P
a
)
300
P

(
P
a
)
6000
200
4000
100
2000
0
0 0.0005 0.001 0.0015 0.002
0
0 0.2 0.4 0.6 0.8 1 1.2
t (s) t (s)
0.05
20
15
15 20
y

(
m
m
)
10
5
0
0 5 10
x (mm)
80 B.FAROUKETAL.
FIG.72. Variationofpressure(pp
i
)withtimeatthemidpointoftheenclosure(H/L=1,
t
h
=5.7610
5
s),underdifferentgravityconditions(1.0g,0gand1.0g).
time-scale in Fig. 72 (b) and the oscillations shown in Fig. 72 (a) are no
longernoticeableinthefigureduetothechangeofscaleinthepressureaxis.
For all three cases, the pressure rise behavior is identical. Minor differences
onlyappearwhentheslowerresponsebuoyancyinducedflowstartsdeveloping
atlargertimesforCase1(Fig.72(b)).ForCase1,thecenterpointisatthe
middleofadevelopingvorticalflow,thuscausingtheincipientpressurerise.
ThevelocityvectorsinFig.73showtheflowfielddevelopedbythewave
motionduetotherapidheatingofthebottomwallatanearlytime(t=0.058s)
FIG.73. Velocityvectorsatt=0.058s(H/L
=1,t
h
=5.7610
5
s),1.0g.

(a) 0.001 (b) 0.01
20
15
5
0
0 5 10 15 20 0 5 10 15 20
x (mm) x (mm)
20
15
10
5
0
y

(
m
m
)
0.001
(c)
20
15
y

(
m
m
)
10
15 20
y

(
m
m
)
10
5
0
0 5 10
x (mm)
for Case 1. Due to spatially uniform heating, the flow field is essentially
symmetricalongtheverticalmiddle plane.Flow structures developalongthe
topandbottomcornersduetoviscousinteractionsbetweenthesolidsurfaces
and bulk fluid. Results for Cases 2 and 3 (at t=0.058s) show essentially no
difference from those shown in Fig. 73. The flow fields at a larger time
(t=0.58s), however, start to display the distinct effects of gravity on the
acoustic wave induced flow field, as shown in Figs. 74(a), (b) and (c) for
Cases 1, 2 and 3, respectively. For the case of 1.0g two characteristic eddies
areformedduetothebottomheatedbuoyancyinducedflow.Forthecasesof0g
FIG.74. Velocityvectorsatt=0.58s(H/L=1,t
h
=5.7610
5
s):(a)1.0g;(b)0g(c)1.0g.

400
380
360
T

(
K
)
340
320
300
0 5 15 20
0g,1.0g
+ 1.0g
10
y (mm)
82 B.FAROUKETAL.
FIG. 75. Variation of temperature with y at the mid point of the enclosure at t=1.27s
(H/L=1,t
h
=
5.7610
5
s),underdifferentgravityconditions.
(Fig. 74 (b))and1.0g,theflow fields generatedbythe acoustic wavesare
considerably damped, due to viscous diffusion. The dampening is further
enhancedbythestratifieddensityfieldthatiscreatedintheopposedgravity
(1.0g)case.
Figure75showstheinstantaneoustemperatureprofilesalongthevertical
mid plane for Cases 1, 2 and 3 at t=1.27s. In the cases of 0g (Case 2)
and 1.0g (Case 3), the temperature profiles tend to be nearly linear,
because heat conduction and acoustic compression and rarefaction influ-
ence the temperature distribution (no recirculating flow development).
However, for the case of 1.0g (Case 1), sharp temperature gradients
areobservednearthetwohorizontalwallsandaninversiontothetempera-
ture profile occurs near the middle of the enclosure due to a developing
vorticalflow.
b. Effects of Heating Rate. Next, we studied the effect of heating rate (as
characterizedbythevalueoft inEq.(IV.12)).Simulationswerecarriedout
forvarying values of t
h
:5.7610
5
, 2.8810
4
and1.1510
2
sas shown
inTableVII(Cases1,4and5,respectively).Theaspectratiooftheenclosure
isfixedat1.0andthevalueof=1.0gforallthreecases.
The variation of pressure at the mid point of the enclosure is shown in
Fig. 76 for Cases 1, 4 and 5. The strength of the pressure wave and the
500
h
=5.76 10
6
s
h
=2.88 10
6
s
h
=1.15 3 10
3
s
400
300
200
100
0
0 0.0005 0.001
(
P
a
)
0.0015 0.002
t (s)
FIG.76. Variationofpressurewithtimeatthemidpointoftheenclosure(H/L=1.0,1.0g),
underdifferentheatingrates.
overall rate of pressure increase are found to be strongly correlated to the
heatingrate.Theacousticwavesproducedareratherweakwhentheheating
rateisslower,asinCase5(t
h
=1.1510
2
s).
For t
h
=1.1510
2
s, the development of the flow field at t=0.058s is
shown in Fig. 78. At this early time, the fluid motion is limited to the
verticaldirectionasdictatedbytheacousticwaves.Comparingtheresults
shown in Fig. 77 to those shown in Fig. 73 (t
h
=5.7610
5
, Case 1), the
strengthoftheflowfieldisfoundtodecreaseconsiderably whenthetime
constant of the heating rate is increased from 5.7610
5
s (Case 1) to
1.1510
2
s (Case 5). Thus at early times, for both values of t, the flow
field is dominated by the thermally induced pressure field and no buoy-
ancy effects are evident.
Figure 78 shows the development of velocity vectors at t=0.58s for
Case 5. The flowfieldbeginstoshowthe signs of adeveloping buoyancy
induced flow. Two eddies are found to form near the bottom wall.
Compared to similar results for Case 1 (Fig. 74 (b)), the velocity values
are lower in Case 5 (Fig. 79). The strong acoustically induced flow field
in case 1 excites the development of the buoyancy induced flow field in
Case1.TheflowfieldsforCases1and5willeventuallybecomeidentical,
as the acoustic waves damp out (as shown in Figs. 81(a) and (b) at
t=1.27s).
0.005
20
15
15 20
y

(
m
m
)
10
5
0
0 5 10
x (mm)
0.01
20
15
15 20
y

(
m
m
)
10
5
0
0 5 10
x (mm)
84 B.FAROUKETAL.
FIG. 77. Velocity vectors at t=0.058s (H/L=1.0,
2
1.0g), under heating rate
t
h
=1.1510

s.
FIG.78. Velocityvectorsatt=0.58s(H/L
=1.0,1.0g),underheatingratet
h
=1.1510
2
s.
c. EffectsofAspectRatio.Simulationswerecarriedoutforvaryingvaluesof
aspectratioH/L=1.0,0.5and5.0asshowninTableVII(Cases1,6and7,
respectively).The timeconstant oftheheatingrateist
h
=5.7610
5
s and
thevalueof=1.0gforallthreecases.Theearlytimevariationofpressure
at the mid point of the enclosure is shown in Fig. 80 for Cases 1, 6 and 7.

10
20
15
10
5
0
y

(
m
m
)
(a) 0.2 (b) 0.2

20
15
y

(
m
m
)
5
0
0 5 10 15 20 0 5 10
x (mm) x (mm)
15 20
1000
800
600
P

(
P
a
)
400
200
0
0
t (s)
H/L =5
H/L =1
H/L =0.5
0.0005 0.001 0.0015 0.002
ACOUSTICWAVEINDUCEDFLOWSANDHEATTRANSFER 85
FIG. 79. Velocity vectors at t=1.27 s (H/L=1.0, 1.0g), under varying heating rates:
(a)t
h
=5.7610
5
s(b)t
h
=1.1510
2
s.
FIG. 80. Variation of pressure with time at the mid point of the enclosure
(t
h
=5.7610
5
s,1.0g),fordifferentaspectratios.

10
(b) 0.01
(a)
y

(
m
m
)
5
0
0.01
100
80
60
40
20
0
y

(
m
m
)
0 5 10 15 20
0 5 10 15 20
x (mm)
x (mm)
86 B.FAROUKETAL.
FIG. 81. Velocity vectors at t=1.27 s (t

h
=5.7610
5
s, 1.0g), under different aspect
ratios:(a)H/L=0.5(b)H/L=5.0.
For an enclosure with the smallest aspect ratio (H/L=0.5), the acoustic
wavesreverberateintheenclosurewithahigherfrequencyandthepressure
increase at the mid point is rapid. For larger aspect ratios, the above fre-
quencydecreases,asdoestherateofincreaseofpressureatthemidpointof
theenclosure.
Theflowfieldsatt=1.27sforCases6and7(whichhavenon-unityaspect
ratios)areshowninFigs.81(a)and(b).ThecorrespondingplotforCase1
was shown earlier in Fig. 81(a). In all cases, two eddies created by the
buoyancyinducedflowformnearthebottomwall.InthecaseofH/L=5.0
(Case7)theeddiesonlyaffectthelowerpartofthedomain,withtherestof
the domain unaffected. For the tall enclosure (Case 7), the flow field was
computedforanextendedtime.Thevelocityfieldsatt=3.8sand7.02sare
showninFigs.82(a)and(b)forCase7.Amulticellular flowfielddevelops
forthiscase.
d. Conclusions. Interaction of the flow fields in an enclosure generated
by acoustic waves (created by rapid heating of the bottom wall) and the
buoyancy effects were studied by solving the unsteady compressible
Navier-Stokesequations.Therapidityofthewallheatingprocessisdirectly
proportional to the strength of the pressure waves induced. The thermally
y

(
m
m
)
(a) 0.1 (b)

0.03
100
80
60
40
20
0
0
100
80
60
40
20
0
y

(
m
m
)
5 10 15 20 0 5 10 15
20
x (mm) x (mm)
F .82. VelocityvectorsforH/L=5.0,t =5.7610
5
IG
h
s,1.0g,at(a)t=3.8s;(b)t=7.02s.
induced waves create an essentially one-dimensional oscillatory flow in the
enclosure, whichdamps outwith time. Thebuoyancyinducedflow isdevel-
oped slowly compared to the acoustic wave-induced flows. In addition, the
strengthofthethermallyinducedwaveshassignificanteffectsonthetemporal
development of the buoyancy induced flow. The stronger acoustic wave
induced by faster heating will delay the development of buoyancy induced
flow.Itisalsoobservedthatthethermallygeneratedflowfieldandthebuoy-
ancyinducedflowfieldarebothaffectedbytheaspectratiooftheenclosure.
Acousticwavescanbegeneratedinanenclosurebyrapidheatingofawall
foridealgas,realgasorsupercriticalfluid.Similarpressurewaveshapesare
observedforallcompressiblefluids.However,thestrengthofpressurewave
increases when the initial pressure increases, because of the variation of
thermodynamic properties. The rapidity of the wall heating process is
directly proportional to thestrength ofthe pressure wavesinduced.Impul-
siveheatingisfoundtogenerateastrongthermallyinducedacousticwave;
however, the damping rate for the resulting velocity field is high. Immedi-
ately after the heating, the thermally induced wave creates an oscillatory
flow in the enclosures flow field, which will then damp out with time.
The buoyancy induced flow is developed slowly in comparison with the
acousticwave-inducedflows,butitwillreachatthesteadystateeventually.
88 B.FAROUKETAL.
In addition, the strength of the thermally induced wave has a significant
effectonthetemporaldevelopmentofthebuoyancyinducedflow.Itisalso
observed thatthethermallygeneratedflowfieldandthebuoyancyinduced
flowfieldarebothaffectedbytheaspectratiooftheenclosure.
V. ExperimentalStudyofThermallyInducedAcousticWavesinGases
A. INTRODUCTION
Thebehaviorsofthermallyinducedacousticwavesgeneratedbytherapid
heatingofaboundingsolidwallinaclosedcylindricaltubeareinvestigated
experimentally.Intheexperiments,aresistance-capacitance(R-C)circuitis
usedtogeneratearapidtemperatureincreaseinathinnickelfoillocatedat
one end of the closed cylindrical tube. The thermally generated pressure
(acoustic)wavesaregeneratedinsidethetube,andundergorepeatedreflec-
tions at the two ends of the tube and gradually decay. The time-dependent
gaspressurevariationinthetubeandthevoltageandtemperaturehistories
ofthefoilarerecordedbyafast-responsemeasurementsystem.Finally,the
measurementresultsarecomparedwiththenumericalresults.
B. EXPERIMENTAL APPARATUS AND PROCEDURE
The realization of a rapidincrease of a wall temperature isof significant
interest in the experimental design. The temperature increase rate does not
onlyinfluencethecharacterofthermallyinducedacousticwaves,butitalso
determineswhetherthewavegeneratedcanbedetectedbyasensor.Withthe
knowledgegainedfrompastwork[56],weutilizeadirect-currentheatingof
ametal-foilcoveredend-pieceofaplastictubebymeansofanR-Ccircuit.
AschematicoftheexperimentalsetupisshowninFig.83.
Aplastictubewithaninsidediameterof38mmandalengthof201mmis
used for studying the thermally generated acoustic waves. A nickel foil
completely covers one end of the tube and is attached with a mica end-
piece.Asimilarmicaplate isusedattheotherendofthetube. Thesemica
plates ensure rigid reflecting surfaces for the pressure waves. In addition,
flexibleEPDM(ethylenepropylenedienemonomer)rubberplatesandthick
plasticplatesareaddedatthebackofthemicaplateandanchoredwiththe
tube to ensure a good seal for the tube. The foil is anchored by two long
copper bars that serve as connection wires and supports for the foil.
The cross-sectional dimensions of the copper bars are 70mm70mm.
This minimizes the voltage drop through the copper bar, and guarantees
thattheelectriccurrentflowsthroughthefoilevenly.
Battery(fortriggering)
Relay
DAQcard
NI6052E
ThermocoupleAmplifier
OmniAmpIIB-E
Thermocouple
type:E,0.0005
Pressureprobe
BK4192 Gasfilling
Endevco4428A
signalconditioner
Endevco8507C-1
pressuretransducer
Battery(forcharging)
SCR
+
Capacitor
Foil
Vent
Powersupply
Nexus2690
Computer
Terminal
block
(SCG-68)
FIG.83. Schematicoftheexperimentalsetup[79].
A very thin nickel foil (thickness is 6.0mm, electrical resistivity is
6.8410
8
-m)isusedinthepresentstudyforrapidheatingofitssurface.
Asilicon-control-rectifier(SCR),(Littelfuse,model:TO-218X)witha2.5ms
triggering time is used to fully discharge the capacitor. This arrangement
provides a rapid rise of the foil temperature for generating the acoustic
wavesinthesurroundinggasalongthetube.Someotheradvantagesofthe
R-C heating system include good repeatability and easy control of heating
rate and maximum temperature increase. The details of the apparatus and
circuitelementsarelistedinTableVIII.
In the experimental setup, the dimensions of the foil (attached around a
circular mica end-piece) are 42mm50mm6.0mm. The electrical resis-
tance of the foil is calculated to be 0.014 . By carefully calculating every
componentintheR-Ccircuit,thecircuitefficiencyE=R
foil
,R
total
isfoundto
beabout53%.Severalcapacitorsratedat18.0mF,27.0mFand56.0mFare
used for the measurements. The circuit time constant, t
RC
=R
total
C is
hence 0.0005s, 0.00071s, and 0.0015s for capacitances of C=18.0mF,
27.0mF,and56.0mFusedinthecircuit,respectively.
Two different types of pressure probe (see Fig. 84) are used to measure
andmonitorthethermallygeneratedacousticwavesinthetube.Aconden-
ser microphone (12.7mm Bruel & Kjr model 4193) and a piezoresistive
pressure transducer (Endevco model 8507C-1) are used. The probes are
90 B.FAROUKETAL.
TABLE VIII
SPECIFICATIONS FOR THE EXPERIMENTAL SYSTEM [79]
No. Part Makeandmodel Specification
1 Capacitors SpragueAntexA8 18,27and56mF
2 SCR LittelfuseTO-218X Maxtriggertime:2.5ms
3 Foil Nickel,0.006mmthick,99.95%,
resistivity6.8410
8
-m
4 Thermocouple OmegaBare-8-E-12 12.7mmdia.E-type
5 Pressureprobes B&K4193withUC0211 2mV/Pa
adaptor
Endevco8507C-1 2.1mV/KPa
6 Dataacquisitionboard NI6052E 333kHz
(a) Brel&Kjrmicrophone (b) Endevcopressuretransducer
4Conductorcable
Housing
Diaphragm
Backplate
Sound
Sound
field
Field
0.020(.51)DIA
Vent
venttube
Wallof
Wallof
enclosure
enclosure
FIG. 84. Details of the pressure probes [79]. (a) Bruel & Kjr microphone. (b) Endevco
pressuretransducer.
mounted transversely on the plastic tube (see Fig. 93), with the sensing
surfacesflushwiththetubeinnerwall.Theprobesarethusdirectlyexposed
tothegasandminimallyinterferedwiththepressurewavepropagation.The
gapsbetweentheprobeboundariesandtheholesaresealedbyDowCorning
832 non-corrosive sealant. The Bruel & Kjr condenser microphone was
chosen because of its high sensitivity, fast response, and wide frequency-
response band.For theBruel&Kjr condenser microphone,thedetection
system also included a low-frequency adaptor (UC0211), a microphone
preamplifier (model 2669) and a conditioning amplifier (ZN 2690). The
piezoresistive pressure transducer, Endevco 8507C-1, is used to measure
the variation of the total pressure (sum of the static and dynamic
components) in the tube. The Endevco 8507C-1 piezoresistive pressure
transducer is powered and conditioned by a 4428A signal conditioner.
The sensitivity of the Endevco transducer is, however, not sufficient to
allow the measurement of the small dynamic pressure fluctuations in the
tube due to the sudden heating of the foil. Use of these two pressure
transducers allow us to probe the thermally induced acoustic waves in
greaterdetailthanreportedearlier.
TheBruel&Kjr4193condensermicrophone(seeFig.84(a)),consistsof
ametalhousing,insideofwhichadelicateandhighlytensioneddiaphragm
isplacedaheadofaback-plate.Thedistancebetweenthediaphragmandthe
back-plate changes if there is any pressure difference between the micro-
phone housing and the tube volume exposed to the diaphragm. The corre-
sponding capacitance variation is converted to pressure signal by the
microphone cartridge (not shown). To eliminate the influence of static
pressure variation and protect the diaphragm, the microphone housing is
connected to the tube medium by a static pressure equalization vent hole
(Fig. 84 (a)). The narrow vent hole ensures that the static pressure of the
microphonehousingfollowsthepressurevariationinthetube.Theventhole
wasdesigned[84]toequalizethestaticpressurevariationswithoutsuppres-
sing the low-frequency components of the dynamical acoustic pressure
which are to be measured. The time constant of the microphones pressure
equalization system is about 0.1s; therefore, frequencies below 10Hz are
affectedbytheventhole.TheBruel&Kjr4193microphoneisthussuitable
formeasuringthedynamicpressurevariationwhenthestaticpressureinthe
tuberemainsconstantorvariesslowly.
TheEndevco8507C-1piezoresistivepressuretransducer,showninFig.84(b),
ontheotherhand,measuresthevariationofresistanceduetothepressure
difference between the two sides of the piezoresistive element. One side is
exposed to the gas in the tube and the other side communicates with the
outsideambientviaaventtube.
Otherthanthepressuremeasurements,thetemperatureandvoltagedrop
historiesofthefoil(afterthecapacitorisdischarged)arealsorecordedinour
measurements.Previousnumericalstudies[54]ofthermallygeneratedacous-
tic waves by a rapidly heated surface show that the early temperature
behavior of the heated surface (during the first several micro-seconds) has
a significant influence on the behavior of the acoustic wave generated.
Hence,wepayspecialattentiontothemeasurementofthefoiltemperature
atearlytimesafterthecapacitorisdischarged.Chromel-constantantypeE
thermocouples with a diameter of 12.7mm are used for their fast response
time. The attachment of thermocouple on the foil is also critical. Silver
adhesive (503, Electron Microscope Science) was chosen after evaluating
many other brands of adhesives, mainly for its high thermal conductivity
92 B.FAROUKETAL.
and ease of applicability. Due to its heat capacitance, the adhesive slows
downtheresponsetimeofthethermocoupleatveryearlytimes.Thesignals
fromthethermocoupleprobeareconditionedbyanOmegaOmniAmpIIB-
E conditioning amplifier. The amplifier not only amplifies the weak signal
from the thermocouple, but also isolates the high-frequency electrical dis-
turbance by its interior signal conditioning circuit. Although the amplifier
eliminates the electrical disturbance at early times, it does not accurately
record the peak value of the temperature rise of the foil (due to signal
conditioningbytheOmegaOmniAmpIIB-Eamplifier).
The analog temperature, pressure and voltage measurements were
recorded,digitizedandsavedthroughaNationalInstrumentSCB-68term-
inal block and a 6052E data acquisition (DAQ) board. High sample rate
(333kHz)ofthe6052EDAQboardguaranteesthatthesignalsarerecorded
withhighfidelity.Thedataacquisitionsystemalsoprovidesavoltageoutput
toarelaytocontrolthetriggeringtimeoftheSCR.
Beforeeachexperiment,alldeviceswerepoweredupandrunsuchthatthe
warm-uptimerequirementsweremet.Allmodulesanddeviceswerechecked
for the reliability of the experiment. The capacitor was charged up to a
desired voltage V
0
, which was measured by an HP 34401A multi-meter.
The LabView 7.0 software was used to record the signals of temperature,
voltageandpressureintheexperiments,andtoprovidecontrollingsignals.
The SCR was triggered to initiate the experiment. At this point in the
experiment, the capacitor is discharged causing rapid heating of the foil.
The foil temperature gradually falls after the initial rapid rise. For most
experiments, data were collected for about 5.1s. After one experiment was
carried out, the experimental conditions were initialized before the next set
ofmeasurementswasmade.Eachexperimentwasrepeatedseveraltimesin
ordertoconfirmthemeasurements.
C. EXPERIMENTAL RESULTS AND DISCUSSION
Approximatelyonehundredexperimentalrunswerecarriedoutwiththe
testrigusingairatambientconditions(p
0
=1atm.,T
0
=299K).Thenickel
foilusedfortherapidheatingwas6.0mmthick.Severalcapacitorsratedat
18.0mF,27.0mFand56.0mFwereusedforthemeasurements.Foragiven
capacitor,anumberofdifferentchargingvoltagesV
0
wereapplied.
Figure85exhibitsthetransientvoltagedropacrossthefoilwithC=27.0mF
andV
0
=30V.TheresponseshowsatypicalR-Ccircuitdischargebehavior.
Thefoilvoltagereachesthepeakvalueatabout0.00003s,andthenitgradually
decreasestozerobyabout0.005s.Thefoilvoltageisabout37%ofthepeak
valueatapproximatelyt=0.0011s.Thisisclosetothecalculatedcircuittime
constant,t
RC
=R
total
C=0.00071sforC=27.0mF.
0
5
10
V
o
l
t
a
g
e
o
f
f
o
i
l
(
V
)
15
20
0.005 0 0.005 0.01 0.015 0.02
Time(s)
FIG.85. TransientvoltagedropacrossthefoilatC=27mFandV
0
=30V.
The corresponding temperatures measured by the thermocouple are
showninFigures86(a)and(b).Atearlytimes(fort<0.02s)themeasured
temperature rise rate is found to be much slower than the corresponding
voltagedischargerate.Thetemperatureriseinthefoildependsontheheat
(a) Earlytime (b) Latertime
350
340
330
T
(
K
)
350
340
330
320
310
300
T
(
K
)
320
310
300
0.005 0 0.005 0.01 0.015 0.02 0 1 2 3 4 5
Time(s) Time(s)
FIG. 86. Transient temperature of the foil at C=27 mF and V
0
=30 V. (a) Early time;
(b)Latertime.
40
60
100
80
0
20
20
40
P
(
P
a
)
60
80
100
P
(
P
a
)
20
0
0 0.005 0.01 0.015 0.02 0 1 2 3 4 5
Time(s) Time(s)
94 B.FAROUKETAL.
FIG.87. Experimentaltraceofpressure(pp
atm
)wavemeasuredbytheB&Kmicrophone,
C=27mF,V
0
=30V[79].(a)Earlytime;(b)Latertime.
gainandlosstotheair.Itisalsonotedthatduringthecapacitordischarge
period, the electromagnetic disturbance is high, andthe thermocouple can-
notrespondduringtheinitialdischargeperiod(foraboutt<0.005s).
Figures 87(a) and (b) show the trace of the pressure wave measured by
the Bruel & Kjr microphone under the following conditions: C=27mF,
V
0
=30V. The probe records a peak when the acoustic wave sweeps past
the probe diaphragm during its motion from the heated side to the
unheated side. The next peak indicates the reflected acoustic wave that
now travels from the unheated side to the heated side. During the first
several acoustic cycles (Fig. 87(a)) the wave shape shows steep fronts with
longtails,whichisconsistentwiththepreviousstudies[53].Duetoviscous
dissipation and energy losses, the characteristics of the acoustic wave pro-
file, i.e., sharp peaks and steep fronts, gradually disappear. The measured
Mach number of the acoustic wave is about M=1.03. The pressure varia-
tionrecordedbytheBruel&Kjr4193microphoneoveralongerperiodof
time(about5.1s)ispresentedinFig.87(b).Itisinterestingtoobservethat
the probe records pressure values lower than the ambient value (that is,
non-physical values) beginning at about t=0.6s. The anomalous result is
tracedtotheinherent construction andoperationalfeaturesoftheBruel&
Kjr condenser microphone. The existence of the narrow vent hole
(see Fig. 84 (a)) causes the non-physical results at longer times. While the
Bruel & Kjr microphone is suitable for the measurement of dynamic
pressure where static pressure remains constant, it is not designed for the
measurement with varying static pressure. When the foil is first heated, it
heatsthegasinsidethetube,sothestaticpressureincreases.Therefore,the
staticpressureinsidethetubeincreasestoamaximumvalue,thendecreases
to the initial value in several seconds as the heat is lost to the outside
environment from the wall. As the diameter of the vent hole connecting
the microphone housing to the pressure field in the tube is very small, the
static pressure inside the microphone housing varies more slowly than that
in the enclosed pressure field. Within some time period, the static pressure
inside the microphone housing becomes greater than that of the air in the
tube, causing the pressure readings to fall below zero (from t=0.6 s to
t=4.2s in Fig. 87 (b)). Finally, the static pressure difference between the
microphone housing and the tube is eliminated (due to the cooling of the
system), and the pressure measurement will become equal to zero, as dis-
played in Fig. 87(b) as time approaches t=5.0s.
As mentioned earlier, a second pressure probe, the Endevco 8507C-1
piezoresistive pressure transducer, was also used in the experiments. In
contrasttotheBruel&Kjr4193microphone,theventholeoftheEndevco
8507C-1pressuretransducerisexposedtooutsidethetestsection(seeFig.84
(b)),sothevariationofstaticpressureinsidethetestsectiondoesnotcause
any anomalous results. The unfilteredmeasurements madebythe Endevco
probeareshowninFigs.88(a)and(b),forthesameexperimentalconditions
used to produce Figs. 87(a) and (b). At early times, the pressure waves
measured by both probes are quite similar, except the noise generated by
theEndevcoprobeismuchhigherandcausesdistortioninthesignal.More
interestingly,thetwomeasuredpressureshapesareverydifferentforlonger
times(compareFigs.87(b)and88(b)),forreasonsdiscussedabove.
100
80
60
P
(
P
a
)
40
0
20
P
(
P
a
)
40
60
80
100
20
0.015 0.02
0
0 0.005 0.01
0 1 2 3 4 5
Time(s) Time(s)
20
FIG.88. Experimentaltraceofpressurewave(pp
atm
)bytheEndevcopressuretransducer
(unfiltered)atC=27mF,V
0
=30V[79].(a)Earlytime;(b)Latertime.
96 B.FAROUKETAL.
Brel&Kjr
Endevco
100
80
60
40
P
(
P
a
)
20
0
20
0 1 2 3 4 5
Time(s)
FIG.89. Comparisonofpressure(pp
atm
)measurementbytheBruel&Kjrmicrophone
(unfiltered)andEndevcopressuretransducer(filtered)atC=27mF,V
0
=30V.
To decrease the noise in the Endevco probe results, the signal is digitally
filteredbyusingathird-orderButterworthmethod.Thecomparisonbetween
themeasurementsbythetwopressureprobes,usingthefilteredresultsforthe
Endevcoprobe,isgiveninFig.89.Noapparentnegative(lowerthanambient)
pressure signal is found for the Endevco probe, due to its construction and
measurement principle. From our experiments, we find the Bruel & Kjr
microphonetobeusefulforthemeasurementsofthedynamicpressurewave
(providedthatthereisnovariationinthebackgroundstaticpressure),while
theEndevcopressuretransducerisusefulforthemeasurementoftheoverall
pressurevariationswherethereislessfluctuationinthepressurefield.
A series of measurements were carried out where the capacitor charge
voltagewasvaried.Figure90showstheeffectsofthechargevoltageonthe
temporaldecayofvoltageacrossthefoilwhereC=27mF.Thecorrespond-
ingtemperaturemeasurementsaregiveninFig.91.Anychangeofthecharge
voltageofthecapacitoronlychangestheelectricenergystoredinthecapa-
citor;itdoesnotaffectthecharacteristicsofthecircuit.Thevariationofthe
maximumvoltagedropacrossthefoilwithchargevoltageisshowninFig.91
withC=27mF.Therelationshipisfoundtobealmostlinear(Fig.92).
Figure93exhibitstheeffectsofthechargevoltageofthecapacitoronthe
acoustic wave (early time) as measured by the Bruel & Kjr microphone
30
25
20
25V
30V
35V
40V
45V
0 0.005 0.01 0.015
V
o
l
t
a
g
e
o
f
f
o
i
l
15
10
5
0
Time(s)
FIG. 90. Effects of the charge voltage on the temporal decay of voltage drop across foil,
C=27mF.
420
400
380
25V
30V
35V
40V
45V
T
(
K
)
360
340
320
300
0 1 2 3 4 5
Time(s)
FIG.91. Effectsofthechargingvoltageonthetemperaturehistoryoffoil(extendedtime),
C=27mF.
98
25V
250
30V
35V
40V
200 45V
150
P

(
P
a
)
100
50
0
0 0.002 0.004 0.006 0.008 0.01
Time(s)
B.FAROUKETAL.
16
18
20
M
a
x
i
m
u
m
v
o
l
t
a
g
e
o
f
f
o
i
l
(
V
)
22
24
26
28
30
25 30 35 40 45
Chargevoltage(V)
FIG. 92. Variation of the maximum voltage drop across foil with charge voltage,
C=27mF.
FIG.93. Effectsofchargevoltageonthepressure(pp
atm
)wave(earlytime)asmeasured
bytheBruel&Kjrmicrophone,C=27mF.
P

(
P
a
)
60
55
50
45
40
35
30
25
20
15
10
25 30 35 40 45
Chargevoltage(V)
FIG.94. Variationoftheunattenuatedamplitudeoftheacousticwavewithchargevoltage,
C=27mF.
undersimilarconditions.Strongeracousticwavesaregeneratedforhigher
charge voltages, because of the higher temperature increases rate. In
addition, the static pressure increase is higher for higher charge voltage.
The variation of the unattenuated amplitude of the acoustic wave with
chargevoltageisgiveninFig.94.Theamplitudeofthefirstpressurepeak
asmeasuredinthemiddleoftubeisplotted.Physically,theacousticwave
originatesfromthelocalpressuredisturbancenearthewallintroducedby
sudden gas temperature increase. The sudden gas temperature increase is
duetotheheatconductionfromthefoilthatisrapidlyheated.Therefore,
the strength of the generated acoustic wave is found to vary almost
linearly with the charge voltage.
Figure95showstheeffectsofthecapacitorvalue,orcapacitance,onthe
voltagedecayrateacrossthefoilatV
0
=30Vasthecapacitorisdischarged.
The corresponding temperature measurements are given in Fig. 96. The
capacitor value not only changes the amount of electric energy stored in
the capacitor according to the relationship Q=CV, but also changes the
dischargecharacterofthecircuit.
Figure 97 exhibits the effects of capacitor value on the acoustic wave
(early time) as measured by the Bruel & Kjr microphone under similar
conditions. The higher capacitor values result in high energy input from
the foil, so the acoustic wave generated is stronger, with higher static
25
20
C =18mF
C =56mF
C =27mF
V
o
l
t
a
g
e
o
f
f
o
i
l
15
10
5
0
0 0.005 0.01 0.015 0.02
Time(s)
100 B.FAROUKETAL.
FIG.95. Effectsofcapacitanceonthevoltagedropacrossfoilatearlytimes,V
0
=30V.
440
T
(
K
)
420
400
380
360
340
320
300
C =18mF
C =56mF
C =27mF
0 1 2 3 4 5
Time(s)
FIG.96. Effectsofthecapacitanceonthetemperatureresponseofthefoil,V
0
=30V.
pressure increases. The variation of the first peak of the acoustic wave
with capacitance (as measured at the midpoint of the tube) is given
in Fig. 98.
200
150
C =18mF
C =27mF
C =56mF
0.004 0.005
P
(
P
a
)
100
50
0
0 0.001 0.002 0.003
Time(s)
FIG. 97. Effects of the capacitance on the acoustic wave (early time) measured by the
Bruel&KjrmicrophoneatV
0
=30V.
20
P

(
P
a
)
35
30
25
40
45
20 25 30 35 40 45 50 55 60
Capacitance(mF)
FIG. 98. Variation of the unattenuated amplitude of the pressure (p p
atm
) wave with
capacitanceatV
0
=30V.
102 B.FAROUKETAL.
1. EstimationofExperimentalUncertainty
Severalindependentmeasurementswereperformedforthecompletionof
theexperimentalstudy.Thesemeasurementsareasfollows:gaspressurein
the tube (as measured by the Bruel & Kjr 4193 microphone and the
Endevco 8507C-1 pressure transducer), foil temperature as measured by
thefinethermocouple,andvoltagedecayacrossthefoil.Allmeasurements
arerecordedbytheNI6052EDAQboard.TheaccuracyoftheNI6052Eis
+4.747mVfortherangeofvoltagemeasurementsmadebythe4193micro-
phone,+0.479mVforthe8507C-1pressuretransducermeasurements,and
+0.242mVforthethermocouplemeasurements.
There are two sources of error in the Bruel & Kjr 4193 microphone
system: +0.2 dB from the 4193 microphone itself and +0.1 dB from the
2690 signal conditioner. With the error introduced by the DAQ board, the
total maximum error for the pressure measurements made by the Bruel &
Kjr 4193 microphone is estimated to be +0.15Pa. There are also two
sources of error in the Endevco 8507C-1 pressure transducer system.
The uncertainty of the Endevco 8507C-1 pressure transducer is 0.25% of
the full scale output (6895Pa); also, the 4428A signal conditioner has an
uncertainty of0.2%of thefull scale output. The total uncertainty ofthe
8507C-1pressuretransducersystemis+31.03Pa.
The errors in the thermocouple measurements include +0.045C error
fromthecoldjunctioncompensation,0.01%/Cinstabilityfromtheampli-
fier, +0.02C error from conversion relation, and +0.08C DAQ board
error.Consequently,thetotaltemperatureerroris0.145Cplus0.01%ofthe
readingvalue.TheonlyerrorinthevoltagemeasurementisfromtheDAQ
board,whichis+7.47mV.
The uncertainty associated with the experimental data is estimated by
using the method recommended by Kline [85]. Both the transient pressure
measurementsbytheBruel&Kjr4193microphoneandvoltagemeasure-
ments exhibit very high degrees of consistency and very low uncertainties.
For the 4193 microphone measurement, the uncertainty, with a 95% con-
fidence level, is 0.75% for the experimental case with the lowest charging
voltage and capacitance. The corresponding uncertainty is 0.07% for the
voltagemeasurements.ThetransientpressuremeasurementsbytheEndevco
8507C-1 pressure transducer show significant noise, due to the uncertainty
being close to the amplitude of the pressure fluctuation. The maximum
uncertaintyis+1.0%forthetemperaturemeasurements.
Theambienttemperatureforallrunsiswithin+0.6%of299K,andthe
ambientpressureiswithin+1%ofthestandardatmosphericpressure.The
errorassociatedwiththevariationoftheambientconditionsisconsideredto
beminor.
Leakage
Centerline
Heat
Air
Z
r

0
FIG.99. Computationaldomain.
2. ComparisonofNumericalandExperimentalResults
Tosimulatethepresenceofpossibleleakageintheexperimentaltube,part
ofthesideboundarywasalsoconsideredasbeingopen,asshowninFig.99,
in some of the calculations. A 501000 (rz) grid size is used for all
calculationsastheproblemisessentiallyone-dimensional(inthez-direction)
for the short time-scale considered. A very fine mesh in the axial direction
allows us to capture the characteristics of the thermally induced acoustic
waves. In the simulations, we provide the transient temperature history of
thefoilasthetemperatureboundaryconditionfortheheatedwall.Defining
the transient temperature history of the foil is rather challenging. We used
the experimental measurements of both voltage and temperature decay
to develop the thermal boundary condition for the heated wall for the
numericalsimulations.
a. Temporal Foil Temperature. The experimental measurements of both
voltageandtemperaturedecay were usedto develop thethermalboundary
condition of the heated wall for the numerical simulations. Limited by the
propertiesoftheusedthermocouple,itscontactwiththefoilandtheelectric
noiseofthedischargingcircuit,thethermocoupleinourexperimentscannot
accuratelymeasurethefoiltemperatureatveryearlytimes.Thefoilvoltage
measurementsandtheelectricenergyinputintothefoilwereusedtodevelop
thetemperaturedecayrateatveryearlytimes.
Figure100givesthefoilvoltagemeasurementsandthefoiltemperatures
calculatedfromthefoilvoltagemeasurementsforthecaseofV
0
=30 Vand
C=27mF. In the calculation of temperature (from the voltage measure-
ments),anenergybalanceequationforthefoilisconsidered:
0T V(t)
2
,ccA = q
loss
(V.1)
0t R
foil

Voltagemeasurement
Foiltemperaturecalculation
withradiationheatloss
15
Foiltemperaturecalculationwithoutheatloss
20
500
F
o
i
l
t
e
m
p
e
r
a
t
u
r
e
c
a
l
c
u
l
a
t
i
o
n
(
K
)
450
400
10
350
5
300
0
0 0.001 0.002 0.003 0.004
t (s)
F
o
i
l
v
o
l
t
a
g
e
m
e
a
s
u
r
e
m
e
n
t
(
V
)
104 B.FAROUKETAL.
FIG.100. Measuredfoilvoltageandcalculatedfoiltemperatureatearlytimes,C=27mF
andV
0
=30V.
Here,cisthespecificheatofthefoil,Aisthefoilarea,q
loss
isthetotalheat
loss ofthefoil totheambientenvironment, and V(t)isvoltageresponseof
thefoil.Here,weconsidertwocases:(a)noheatloss,q
loss
=0,and(b)heat
lossbyradiation,q
loss
=q
rad
=co
b
(T
foil
T
ambient
)
wherecistheabsorptivityofthefoil(0.36)ando
b
istheStefan-Boltzmann
constant.
Theaboveequationwasnumerically solved byanexplicitscheme. From
the calculated temperature profiles (with and without heat loss), we find
from Fig. 100 that the heat loss term has only a slight effect on the foil
temperatureduringthetemperaturerisetime(withintheinitial0.001s).
Figure 101 shows the comparison of the calculated foil temperature
(withradiationheatloss)fromthevoltagemeasurementsandthethermo-
couple temperature measurements. As stated earlier, the thermocouple
data were conditioned by an Omega Omni Amp IIB-E conditioning
amplifier. The adhesive (silver paint) used also slows down the response
of the thermocouple. These effects perhaps cause the measured tempera-
ture data at early times to be under-predicted. For larger values of time,
somewhat different reasons cause the discrepancy between the measured
andpredictedvaluesofthetemperature.Thereareseveralsourcesofheat
loss, such as convective heat loss to air and conduction losses to the
T
(
K
)
500
Thermocouplemeasurement
Foiltemperaturecalculationfromvoltagemeasurement
450
400
350
300
0 0.1 0.2 0.3 0.4 0.5 0.6
t (s)
FIG. 101. Calculated foil temperature (with radiation heat loss) and thermocouple
measurements,C=27mFandV
0
=30V.
copper bars and the mica plate, that were not considered in Eq. (V.1).
Independent experiments in simple systems confirm that the temperature
measurements at larger values of time (t 0.1s) are accurate.
A polynomial fit of the measured data was obtained for t 0.1s. We
extend the polynomial fit for the measured temperature decay from
t=0.1 s to t=0.0001s, as shown in Fig. 102. We show the temperature
profilesshowninFig.101onemoretimeinFig.103albeitforearlytimes
(0 t 0.01s). For the experiments, the foil temperature increases
rapidly and then gradually decays. The predicted temperature increase
of the foil (from the voltage measurement) meets the extrapolated curve
of the temperature decay at about t=0.0007s. This is close to the calcu-
lated circuit time constant, t
RC
= R
total
C=0.00071s for C=27.0mF as
discussed earlier.
The constructed temperature profile for the foil is shown in Fig. 104 for
0t4.4s.Theinitialtemperatureriseforthefoilwhen0t0.0007sis
givenas
8 _ _
T = a
0
a
i
+ cos(i+ 1000+ t+ w) b
i
+ sin(i+ 1000+ t+ w) 297.636
i=1
(V.2)
Foiltemperaturecalculatedfrom
Voltagemeasurement
Extendedpolynomialfitofthe
thermocouplemeasurement
voltagemeasurement
20 500
450
15
F
o
i
l
t
e
m
p
e
r
a
t
u
r
e
(
K
)
400
10
350
5
300
0
0 0.002 0.004 0.006 0.008 0.01
t (s)
F
o
i
l
v
o
l
t
a
g
e
(
K
)
106 B.FAROUKETAL.
500
450
Extrapolationofthermocouplemeasurement
Foiltemperaturecalculationfromvoltagemeasurement
T
(
K
)
400
350
300
0 0.1 0.2 0.3 0.4 0.5 0.6
t(s)
FIG.102. Extrapolationofmeasuredfoiltemperature,C=27mFandV
0
=30V.
FIG.103. Extrapolationofmeasuredfoiltemperature,C=27mFandV
0
=30V.
440
420
400
380
360 T

(
k
)
340
320
300
Corrected
0 1 2 3 4
t (s)
FIG.104. MeasuredandcalculatedfoiltemperaturehistoryC=27mFandV
0
=30V.
wherea
0
=140.3,a
1
=60.57,a
2
=34.98,a
3
=22.27,a
4
=13.28,a
5
=6.817,
a
6
=2.757, a
7
=0.7334, a
8
=0.05938, b
1
=30.39, b
2
=13.74, b
3
=2.497,
b
4
=2.769,b
5
=4.091,b
6
=3.337,b
7
=1.943,b
8
=0.7773,w=2.393
When0.0007st4.2s,
a
T = 297.636. (V.3)
10tb
wherea=50.92,b=0.3704
Theabovecorrectedtemperaturehistory(Eqs.(V.2)and(V.3))for0t
4.2swasusedasthetemperatureinputinournumericalsimulationsforthe
experimentalcasewhereV
0
=30 VandC=27mF.Themethodwasusedto
obtain foil temperature histories for other cases with different charge vol-
tagesandcapacitances.
Thenumericalmodeldescribedearlierwasusedtoobtainthepredictions
oftheflowandpressurefieldsinthetubedrivenbyrapidheatingofthefoil.
For the pressure at the mid point of the tube wall, the comparison of the
numerical and experimental results is given in Fig. 105(a), where V
0
=30V
andC=27mF.Inthefirstseveralacousticcycles(t<0.002s)thenumerical
and experimental results have an excellent match. The thermal boundary
conditionforthefoilwasswitchedfromEq.(V.2)toEq.(V.3)att=0.0007s.
30
(a) (b)
60
200
50
150
40
Computation
Measurement
0 0.0005 0.001 0.0015 0.002
Computationwithoutleakage
Computationwithleakage
MeasurementbyBrel&Kjr4193
0 0.005 0.01 0.015 0.02 0.025 0.03
P
(
P
a
)
(
P
a
)
100
20
50
10
0 0
t(s) t (s)
108 B.FAROUKETAL.
FIG.105. Comparisonofexperimentalandcomputationalpressurewaves,C=27mFand
V
0
=30V.(a)Earlytime.(b)Latertime.
Thegoodagreementbetweenthecomputationalandexperimentalpressure
results demonstrates that the method for considering the transient foil
temperatureisreasonable.
The comparison for the long time scale is shown in Fig. 105 (b). In the
numerical model, all walls are considered smooth and rigid, and also the
system is assumed to be sealed perfectly. The numerical results correctly
predictthepressurewavespeedandamplitudeofdynamicpressurefluctua-
tionstobethesameastheexperimentalresults.Thepressureinthecompu-
tational study is, however, found to be larger than the experimental
measurements.Intheexperiments,theacousticenergyispartiallyabsorbed
by the walls when the wave hits one. There is also the possibility of the
existenceofleakageintheexperimentalsetup.Werepeatedthesimulations
with a small opening at the side boundary (0.1% of the tube length). The
computational pressure results considering the leakage are also shown in
Fig.105(b).Comparedtothesealedcase,thepressureresponseinthiscase
ismuchclosertotheexperimentalresults.
Figure 106 shows the comparison of experimental data and numerical
predictions for three cases: C=27mF and V
0
=25V, C=27mF and
V
0
=35V, and C=56mF and V
0
=30V. The computational and experi-
mentalresultsarefoundtohavethesameacousticspeedsandsimilarwave
shapes. The comparisons between the computational and experimental
results on the unattenuated amplitude of the acoustic wave (the first peak
measuredinthemiddleofthetube)fordifferentchargingvoltagesaregiven
inFig.107.Inallthedifferentcases,thecomputationalresultsarefoundto
be close to the experimental results. The maximum deviation between the

(
P
a
)
120
C = 27mF, V
0
= 25V(computaion)
C = 27mF, V
0
= 35V(computaion)
C = 56mF, V
0
= 30V(computaion)
C = 27mF, V
0
= 25V(measurement)
100
C = 27mF, V
0
= 35V(measurement)
C = 56mF, V
0
= 30V(measurement)
80
60
40
20
0
0 0.0005 0.001 0.0015 0.002
t (s)
FIG.106. Comparisonsofexperimentalandcomputationalpressurewavesatthemiddleof
thetubefordifferentchargingvoltagesandcapacitances.
experimentalmeasurementsandnumericalpredictionsisabout7%,forthe
case of V
0
=45 V and C=27mF. The computational results are always
found to slightly under predict the experimental results. The reason the
computational results are always a little bit lower than the experimental
results may be that the parameters that used in the numerical model are
slightlydifferentfromtherealvalue.Forexample,thethicknessofthemetal
foilplaysasignificantroleinthenumericalmodel.However,thevaluegiven
by manufacturer has +10% toleration. Therefore, it is possible that the
inputofournumericalmodelisnottotallysameastheexperimentalsetup.
3. Conclusions
Experimental measurements were carried out to characterize the genera-
tion,developmentanddecayofthermallyinducedacousticwavesinacylind-
rical tube. In the experiments, the development and the decay of pressure
wavesinsidethetubewereinvestigatedbytwodifferentpressureprobes,and
thedifferentresultsareexplainedbythephysicsoftheprobes.Thestrengthof
110 B.FAROUKETAL.
60
Computation
55 Experiment
50
45
40
35
30
25
20
15
10
25 30 35
Chargingvoltage(V)
P
(
P
a
)
40 45
FIG. 107. Comparison of experimental and computational results on the unattenuated
amplitudeofacousticwaveatthemiddleofthetubeatC=27mF.
theunattenuatedamplitude(thefirstpeak)oftheacousticwaveproducedis
foundtobelinearlydependentonthetemperatureincreaserateofthefoilat
earlytimes.Atearlytimes,thecomputationalandexperimentalresultsarean
excellent match. The experimental measurements and numerical results for
the pressure evolution, however, do not match well at larger times. This is
perhapsduetothepresenceofleaksinthechamberanddampingduetothe
walls(consideredrigidinthesimulations).However,thephysicalfeaturesof
thecomputationalandexperimentalresultsareingoodagreement.
VI. ThermallyInducedAcousticWavesinSupercriticalFluids
A. INTRODUCTION
In general, near-critical fluids have higher thermal conductivities and
densities than gases, and the compressibilities are much higher. The use of
the van der Waals equation of state for the numerical simulation of the
piston effect was recently studied by Wagner et al. [86,87] for near-critical
fluids by using the SIMPLE algorithm [88]. Unfortunately, the authors do
notindicatetheheatingrateattheleftwall.Thethermallyinducedacoustic

t > 0
t > 0
T1
Tr
SupercritcalCO
2
X
Ti,Pi
t =0
L
wave development, propagation, and reflection etc. were also not explicitly
predicted, although they found that the van der Waals equation of state
under-predicted the magnitude of the pressure wave significantly, as com-
paredtothepredictionsusingtheNISTdatabase12[67].
The generation and propagation of thermally induced acoustic waves in
supercriticalcarbondioxideareinvestigated[89]viaahigh-ordernumerical
scheme. The generation, propagation and dissipation of the thermally
induced acoustic waves in a supercritical carbon dioxide filled layer are
predicted. A one-dimensional problem is considered where supercritical
carbon dioxide is contained between two infinite parallel plates. The NIST
database 12 was used to obtain the ,-p-T relations for supercritical carbon
dioxideaswellastheinternalenergyi=f(,,T)andthermalconductivity,and
viscosityrelationsk=k(,,T),j=j(,,T)
Aconfinedlayer(0.01m)ofsupercriticalcarbondioxide(pp
cr
;TT
cr
where p
cr
=7.377MPa and T
cr
=304.1K) is considered. The left wall is
impulsively heated/cooled to initiate the thermally induced acoustic waves
(seeFig.108).Therightwallisheldattheinitialtemperature.Thethermally
inducedacousticwavegeneratedisstudiedasafunctionofspaceandtime.
Theone-dimensionalcontinuityandthecompressibleformoftheNavier-
Stokes equations are used to describe the generation and propagation of
thermallyinducedacousticwavesinsupercriticalcarbondioxide:
0, 0(,u)
=0 (VI.1)
0t 0x
FIG.108. Schematicoftheproblemgeometry.
_ _ _ _
112 B.FAROUKETAL.
0(,u) 0(,u
2
) 0p 0t
= (VI.2)
0t 0x 0x 0x
0(,e) 0 0 0q
[(,e p)u[ = [ut[ (VI.3)
0t 0x 0x 0x
Wheretistime,,isdensity,uisthevelocitycomponent,eisthetotalenergy,
pispressure,andqistheheatflux.
2
Here e= i u ,2, and i is the internal energy. The equation of state is
representedasp=f(,, T).
The component of the heat-flux is written as follows, where k is thermal
conductivityandTistemperature:
0T
q
x
= k (VI.4)
0x
Thestresscomponentt isgivenas:
0u 0u 4 0u
t = l 2j = j
/
j (VI.5)
0x 0x 3 0x
wheremisthedynamicviscosity,listhesecondcoefficientofviscosityand
j
/
= l (2,3)jisthebulkviscosity.
B. EQUATION OF STATE AND THERMODYNAMIC PROPERTIES OF SUPERCRITICAL
CARBON DIOXIDE
Theequationofstatedescribingthep-,-Trelationofsupercriticalfluidsis
complicated.Ithasbeenshownearlier[90]thatthevanderWaalsequation
do not represent the properties of supercritical carbon dioxide accurately
nearthecriticalpoint.Inthisstudy,weusedtheNISTStandardReference
Database12forthep=f(,,T)relationsandforevaluationofotherthermo-
dynamicpropertiesofsupercriticalcarbondioxide.Variationofpressurefor
carbon dioxide as a function of temperature and density near the critical
point is shown in Fig. 109, where the p-,-T relation is strongly nonlinear,
especially near the critical point. Polynomial fits were used in the present
study to represent, the p=f(,,T), i=f(,,T), k=k(,,T) and j=j(,,T) data
provided by the NIST Standard Reference Database 12. The temperature
range considered for the developed polynomials is 304.15312.5K, and the
pressure range is 7.388.2MPa for all the polynomial fits. The pressure
deviation for the fitted data form the original NIST data ranges from
0.00002% to 0.01%. The viscosity deviation for the fitted data from the
8.2
8
7.8
P
r
e
s
s
u
r
e
(
M
P
a
)
7.6
7.4
P
C
T
c
:304.13K
306K
307.5K
312K 310.5K 309K
200 300 400 500 600 700
c
Density(kg/m
3
)
FIG.109. Variationofpressureforcarbondioxideasafunctionoftemperatureanddensity
nearthecriticalpoint.
originalNISTdatarangesfrom0.00001%to0.001%,thethermalconduc-
tivitydeviationrangesfrom0.01%to9%,andtheacousticspeeddeviation
ranges from 0.0004% to 3%. The fitted internal energy deviation from the
originalNISTdatarangesfrom0.0001%to0.6%.
Figure 110 shows the variation of thermal diffusivity of supercritical
carbondioxideasafunctionoftemperatureandpressurenearthecritical
point. It is noted that the thermal diffusivity is very low near the critical
point. Figure 111 shows the variation of internal energy i=f(,,T) and
Figure 112 shows the variation of the bulk viscosity of carbon dioxide
near the critical point.
C. NUMERICAL RESULTS FOR SUPERCRITICAL CARBON DIOXIDE
Numerical simulations for thermally induced acoustic waves in super-
critical carbon dioxide were carried out for 8 cases, which are summarized
in Table IX. The supercritical carbon dioxide media confined within two
parallelplatesisinitiallyconsideredtobequiescent.
A thermally induced acoustic wave is generated by a step increase/
decrease in the temperature at the left wall dictated by the initial time
114 B.FAROUKETAL.
0.0005
0.0004
P
c
:7.38MPa
7.6MPa
7.8MPa
8.0MPa
8.2MPa
T
h
e
r
m
a
l
d
i
f
f
u
s
i
v
i
t
y
(
c
m
2
/
s
)
0.0003
0.0002
0.0001
304 306 308 310 312
T
c
T(K)
FIG.110. Variationofthermaldiffusivityforcarbondioxideasafunctionoftemperature
andpressurenearthecriticalpoint.
1,00,000
80,000
P
c
:7.38MPa
7.6MPa
7.8MPa
8.0MPa
8.2MPa
I
n
t
e
r
n
a
l
e
n
e
r
g
y
(
J
/
k
g
)
60,000
40,000
20,000
0
20,000
304 306 308 310 312
T
c
T (K)
FIG.111. Variationofinternalenergyforcarbondioxideasafunctionoftemperatureand
pressurenearthecriticalpoint.
_
304 306 308 310 312
Temperature(K)
10
0
10
1
10
2
10
3
10
4
B
u
l
k
v
i
s
c
o
s
i
t
y
(
P
a
.
s
)
10
5
7.38MPa(P
c
)
7.60MPa
7.80MPa
8.00MPa
8.20MPa
FIG. 112. Variation of the bulk viscosity of carbon dioxide as a function of temperature
andpressurenearthecriticalpoint.
TABLE IX
DESCRIPTION OF CASES
Case T
i
[K] T
r
[K] T
l
[K] p
i
[MPa] Bulkviscosity
1 310 310 312 7.9 Zero
2 310 310 308 7.9 Zero
3 308 308 309 7.7 Zero
4 308 308 307 7.7 Zero
5 306 306 306.5 7.5 Zero
6 306 306 305.5 7.5 Zero
7 310 310 312 7.9 Non-zero
8 310 310 308 7.9 Non-zero
step of the computations. The temporal variation of the left wall
temperature is:
T
i
fort= 0
T
1
(t) =
T
1
fort0
The results presented are for the early behavior of thermally induced
acousticwavesintermsofpressureandothervariables.Thecomputational
domain(1.0cm)wasdiscretizedwith300gridpoints.Wesimulatedmultiple
wavereflectionsandwavedampingwithinashortcomputationaltime.

0 0.002 0.004 0.006 0.008 0.01
x(m)
40,000
20,000
0
20,000
40,000
0.1cycle 0.2cycle 0.7cycle
Case1
Case2
0.6cycle
p
p

i
(
P
a
)
116 B.FAROUKETAL.
1. ParametricResults
Figure 113 shows thermally induced acoustic (pressure) wave profiles at
four early different times for Cases 1(sudden heating) and 2 (suddencool-
ing). The sudden heating causes a pressure peak while the sudden cooling
producesapressurevalley.Asonegetsclosertothecriticalpoint,thenon-
linearity of p-,-T relation is more prominent. So finer meshes and longer
computingtimeareneededforcasesclosertocriticalpoint.Asshowninthe
figure, the pressure waves were generated along the left heated/cooled wall
traveled towards the right wall with a wave speed of about 197m/s. This
predictedwavespeedisclosetothevalueobtainedfromtheNISTdatabase
12. Unlike the characteristic long tails of the thermally induced acoustic
waves in ideal gases, the pressure wave signatures in Fig. 113 show a
symmetric profile, due to the high density of the supercritical fluid.
InFig.113,thearrowsdepictthedirectioninwhichthewavesaremoving.
The four wave profiles (both for heating and cooling) at 0.1 cycle and 0.2
cycle on the left are traveling towards the right wall. The other four wave
profilesat0.6cycle and 0.7cycle traveling left areafterreflectionfrom the
rightwall.Asthewavetravelswithinthetwowalls,thewaveshapesbecome
broaderduetodissipation.
FIG.113. Thermallyinducedacoustic(pressure)waveprofilesatfourearlydifferenttimes
within1cycleforCases1and2.
310.4
310.2
310.0
309.8
309.6
0 0.002
0.1cycle 0.2cycle 0.7cycle
Case1
Case2
0.6cycle
0.004 0.006 0.008 0.01
T
(
K
)
x(m)
FIG.114. SpatialvariationsoftemperaturewavesatfourearlydifferenttimesforCases1
and2.
The spatial variation of the temperature wave at four early times for
Cases 1 and 2 are shown in Fig.114. Comparing Figs. 113 and 114, we see
thattemperaturewavesandpressurewavesexhibitsimilarshapes,andtravel
under the same local sound speed. Thin thermal boundary layers start to
formattheleftwall,whichexpandslowlytowardtherightwallwithtime.
The temporal variations of pressure at the center point at early time for
Cases 1 and 2 are shown in Fig. 115. Rapid temporal variations of the
pressure are predicted as the thermally induced acoustic wave arrives at
the midpoint after being generated from the left wall and then reflected
frombothwalls.Twoobservationscanbemadefromthisfigure.First,the
amplitude of the pressure waves are progressively damped as the energy is
dissipated in the media. Secondly, the pressure in the bulk fluid gradually
increases for Case 1 and decreases for Case 2 as the waves travel back and
forth and undergo dissipative losses. These two observations can be more
clearlyshowninFig.116,whichshowsthetemporalvariationsofpressureat
the center point for a longer period (0.1s) for Cases 1 and 2. Another
interesting observation from Fig. 116 is that the temporal variations of
pressure at the center for Cases 1 and 2 are not symmetric, although these
two cases undergo symmetric temperature changes at the left wall. For the
sameinitialstate,acoolingcasegeneratesstrongerthermallyinducedacous-
tic waves than a heating case, as the wave amplitude is determined by the
118 B.FAROUKETAL.
7,920,000
7,915,000
7,910,000
7,905,000
7,900,000
7,895,000
7,890,000
7,885,000
7,880,000
7,875,000
Case2
Case1
P
(
P
a
)
0 0.0005 0.001 0.0015 0.002

t(s)
FIG.115. TemporalvariationsofpressureatthecenterpointatearlytimeforCases1and2.
valueof(0p0T)
,
whichdecreaseswithincreasingtemperature.Forthecases
studied, a temperature decrease (at constant density) produces a larger
changeinpressurethanacorrespondingtemperatureincrease.
The spatial variations of temperature at t=0.025s, t=0.05s, t=0.075s
and t=0.1s for Cases 1 and 2 are shown in Fig. 117. At these times, the
thermallyinducedacousticwaveshavereverberatedwithinthedomaintens
ofthousands oftimesandthewaveamplitudeshavedampedoutconsider-
ably, heating/cooling the bulk fluid homogeneously. This demonstrates the
existence of the so called piston effect in supercritical fluid due to rapid
heating/cooling. The thermal boundary layers keep expanding as the bulk
fluidtemperaturesincrease/decrease.Figure118showsthespatialtempera-
turevariationsforcases1and2,andalsoforconductionsolutionsofCases1
and 2 at t=0.1s. The effects of thermally induced acoustic waves in tem-
peratureequilibrationareevident.Fortheconductionsolutions,thethermal
boundarylayersexpandslowlyandthebulktemperaturesremainessentially
unchangedevenafter0.1s.Comparingbetweenthermallyinducedacoustic
waveresultsandconductionresultsafterlongertimesshowsthepistoneffect
clearly.
7,920,000
7,910,000
Case1
7,900,000
Case2
7,890,000
7,880,000
7,870,000
0 0.02 0.06 0.08 0.1
P
(
P
a
)
0.04
t (s)
FIG.116. Temporalvariationsofpressureatthecenterpointuntil0.1sforCases1and2.
Simulations for other cases with initial states closer to the critical point
(p
cr
=7.377MPa and T
cr
=304.1K) were also carried out. We show the
temporal variations of pressure at the center point until 0.1s for Cases 3
and 4 (P
i
=7.7 MPa and T
i
=308K) in Fig. 119, and for Cases 5 and 6
(P
i
=7.5MPa and T
i
=306K) in Fig. 120. These two figures demonstrate
that as the thermally induced acoustic waves progressively damp out, they
heat/coolthebulkfluidhomogenously.Thesimulationsfortheabovecases
provetheexistenceofpistoneffect[91]inthesupercriticalregion.
Temporalvariationsofpressureatthecenteratearlytimeforcases5and
6areshowninFig.121.Figure122showsthespatialvariationsoftempera-
tureatt=0.025s,t=0.05s,t=0.075sandt=0.1sshowingthepistoneffect
for Cases 5 and 6. As the acoustic waves are dissipated, the bulk fluid is
heated/cooledhomogeneously.
In cases 16, simulations were carried out for zero bulk viscosity. Cases
fornon-zerobulkviscositywerealsosimulatedinthisstudy.Theconditions
for Cases 7 and 8 are the same as in Cases 1 and 2 respectively, except
non-zero bulk viscosity values are considered. Figure 123 shows temporal
variations of pressure at the center for Cases 7 and 8. Figure 124 shows
spatial variations of temperature at t=0.025s, t=0.05s, t=0.075s and
(
K
)
310.2
310.1
t =0.010s
t =0.075s Case1
t =0.050s
t =0.025s
Case2
310.0
t =0.025s
t =0.050s
t =0.075s
309.9
t =0.010s
309.8
0 0.002 0.004 0.006 0.008 0.01
x (m)
120 B.FAROUKETAL.
FIG.117. Spatialvariationsoftemperatureatt=0.025s,t=0.05s,t=0.075sandt=0.1s
forCases1and2.
t=0.1s for Cases 7 and 8. Through comparing Figs. 123 and 115,
Figs. 124 and 116, we see that the effect of non-zero bulk viscosity can be
neglectedinthesetwocases.Therearetworeasonsforthis.Thefirstisthat
inthisstudyallcasesareone-dimensional.Theeffectofbulkviscositymay
be more evident in a two or three dimensional flow problem. The second
reason is that the computed non-zero bulk viscosity value is not very large
compared to the molecular viscosity in Cases 7 and 8. The bulk viscosity
effectswillbestrongerastheinitialstateofthefluidisclosertothecritical
point.
2. Conclusions
The generation and propagation of thermally induced acoustic waves in
supercritical carbon dioxide are investigated by numerically solving a fully
compressible form of the Navier-Stokes equations. The NIST Standard
Reference Database 12 [67] was used to generate the polynomial fits to
represent the equation of state and other thermodynamic properties for
supercritical carbon dioxide. The numerical algorithm used here based on
309.4
309.6
309.8
310.0
310.2
310.4
310.6
T

(
K
)
Case1
Case2
ConductionsolutionofCase1
ConductionsolutionofCase2
0 0.002 0.004 0.006 0.008 0.01
x (m)
FIG.118. SpatialtemperaturevariationsforCases1and2,andforconductionsolutionsof
Cases1and2att=0.1s.
7,712,000
7,708,000
7,704,000
Case3
Case4
P
(
P
a
)
7,700,000
7,696,000
7,692,000
7,688,000
0 0.02 0.04 0.06 0.08 0.1
t(s)
122 B.FAROUKETAL.
7,506,000
7,504,000
7,502,000
Case5
Case6
P
(
P
a
)
7,500,000
7,498,000
7,496,000
7,,494,000
7,492,000
0 0.02 0.04 0.06 0.08 0.1
t(s)
7,506,000
7,504,000
7,502,000
Case5
Case6
P
(
P
a
)
7,500,000
7,498,000
7,496,000
7,494,000
7,492,000
0 0.0005 0.001 0.0015 0.002
t(s)
FIG. 121. Temporal variations of pressure at the center point at early time for Cases 5
and6.
T

(
K
)
306.10
306.05
t=0.010s
t=0.075s
t=0.050s
t=0.025s
Case5
306.00
Case6
t=0.025s
t=0.050s
t=0.075s
t=0.010s
305.95
305.90
0 0.002 0.004 0.006 0.008 0.01
x (m)
showingthepistoneffectforCases5and6.
7,920,000
7,910,000
7,900,000
Case1
Case2
P
(
P
a
)
7,890,000
7,880,000
7,870,000
0 0.02 0.04 0.06 0.08 0.1
t(s)

T

(
K
)
310.2
310.1
t =0.010s
t =0.075s
t =0.050s
t =0.025s
Case7
Case8
310.0
t =0.025s
t =0.050s
t =0.075s
309.9
t =0.010s
309.8
0 0.002 0.004 0.006 0.008 0.01
x (m)
124 B.FAROUKETAL.
forCases7and8.
the FCT algorithm, along with the theory of characteristic-based wall
boundaryconditionsfordensity,providedthehigher-orderaccuracyneeded
toresolvethewave.
The details of thermally induced acoustic wave generation, propagation
anddecayinsupercriticalfluidswerepresented.Inthisstudy,non-zerobulk
viscosity was included in the computational model, however no effect of
bulkviscositywasfoundfortheone-dimensionalcasesconsidered.
VII. ExperimentalStudyofThermallyInducedAcousticWaves
inSupercriticalFluids
A. INTRODUCTION
Thebehaviorsofthermallyinducedacousticwaves,inhighpressureand
supercriticalcarbondioxide,generatedbyrapidheatingofaboundingsolid
wall in a closed cylindrical tube were investigated experimentally. In the
experiments, a thin aluminum foil attached to one end of the cylinder is
heated by a resistance-capacitance (R-C) circuit, which is controlled by a
silicon-controlled rectifier (SCR). The rapid heating of the aluminum foil
generatesathermallyinducedacousticwaveintheclosedcylinder,inwhich
thethermallyinducedacousticwavekeepstraversingbetweenthetwoends
and undergoes many reflections before it is fully damped out. The time-
dependent pressure variations in the cylinder ismeasured by amicrophone
located at a specific point on the side of the cylinder, and the temporal
temperaturevariationsofthealuminumfoilaremeasuredbyathermocouple
attachedtoit,alongwithafast-responsedataacquisitionsystem(DAQ).
B. EXPERIMENTAL APPARATUS AND PROCEDURES
Carbondioxideischosenastheworkingfluid,duetoitsgoodcharacter-
istics, such as non-toxicity, non-flammability, inexpensiveness, and benign
critical temperature (304.13K) and critical pressure (7.3773MPa) [92]. The
thermallyinducedacousticwavesaregeneratedandtravelswithinacylinder
filled with supercritical carbon dioxide, while the DAQ system, the sensor
cables and wires must be connected to a data logging system (a personal
computer).
Figure125showstheschematicoftheexperimentalsystemused.Carbon
dioxide is stored in a tank, and in order to increase the carbon dioxide
pressure, a hand pump is employed to pressurize the carbon dioxide in the
supercritical chamber until the destinationpressureis reached (the destina-
tion pressure is above the supercritical pressure). The reason to choose a
handpumpisthatitisabletocompressesafluidwithinasmallvolumeto
develop high pressure. A handpumpalso enables us to accuratelyregulate
thepressureinsidethesupercriticalchambermanually.
1. GeneratingThermallyInducedAcousticWaves
Basedonthepastexperimental work[93,94], anR-Ccircuit isemployed
to rapidly heat a metal foil attached to a end of the cylinder by means of
direct-current heating. The schematic of foil heating is similar to that
described in Section V.B earlier. Lin and Farouk [94] used 6.0mm thick
Nickel foils to generate thermally induced acoustic waves, because with a
thinnerfoil,theheatingisfasterandthemaximumtemperatureofthefoilis
higher,resultinginstrongerthermallyinducedacousticwaves.Inthisway,a
bettermeasuringaccuracycanbeachieved.
In the experimental setup, the dimensions of the foil (attached around a
circular mica end-piece) are 42mm50mm6.0mm. The electrical resis-
tance of the foil is calculated to be 0.014 . By carefully calculating every
componentintheR-Ccircuit,thecircuitefficiency=R
foil
/R
total
isfoundto
beabout53%.Severalcapacitorsratedat18.0mF,27.0mFand56.0mFare
usedforthemeasurements.Thecircuittimeconstant,t
RC
=R
total
Cishence
Thermalcouple
PressureGauge Foil Copperbar
Teflonplate
C
D
B
Rupturedisc
Handpump
A
Thermalcouple
Powerlead
Stopvalve
CO
2
tank
Filter
Microphonecable
Gland
Glandwires
Housingtube
Preamplifier
Microphone
FIG. 125. Schematic of experimental tubing system.
0.0005s, 0.00071s, and 0.0015s for capacitances of C=18.0mF, 27.0mF,
and 56.0mF used in the circuit, respectively. Thermally generated acoustic
wavesaregeneratedandtravelswithinacylinderofa1indiameter.During
an experiment, the circuit connecting the capacitor and the foil is switched
onbytheSCR,whichistriggeredbythetriggeringcircuit.
Thecriticaltemperatureandpressureforcarbondioxideare304.13Kand
7.3773MPa,respectively.Inordertoachievingsupercriticalstate,weneeded
toheatthecarbondioxidefromtheroomtemperature(generally295K)to
supercritical temperature value, and also to pressurize the carbon dioxide
formaCO
2
tanktosupercriticalpressurevalue.AThermolyne.54
/
heat-
ing tape was used to heat the supercritical chamber from the lab room
temperature (generally 295K) to the destination temperature (above the
critical temperature). Rupture discs were employed to make sure that the
pressurewithinthesystemwillneverbeaboveaspecificvalue.
For the experiments of thermally induced acoustic waves in supercritical
carbondioxide,theprobesareinahighpressurezone,whilethedataneeds
to be transmitted to the DAQ card, which is in an atmospheric region. In
order to measure the thermally induced acoustic pressure waves in the
supercritical chamber, the microphone is connected with the preamplifier
throughascrewthreadonthepreamplifier.Acableconnectsthepreampli-
fierandtheamplifier,whichisconnectedwiththeDAQcard.
C. EXPERIMENTAL RESULTS AND DISCUSSION
Theexperimentalinitialconditions(p
i
=7.69MPaandT
i
=307K)werein
the supercritical region (close to the critical point). At the beginning of an
experiment, the supercritical chamberwas slowly heated by aheating tape,
whichiswoundroundthechamber,tothesetvalue(T
i
=307K).Afterthe
desiredtemperaturewasreached,thepoweroftheheatingtapewasadjusted
inordertomaintainthedesiredinitialtemperature.Beforethecapacitorwas
discharged to generate thermally induced acoustic waves, the supercritical
chamber wasmaintainedatthe initial temperaturefor some timeto ensure
steadystatewithinit.
Figure 126 shows the temporal variation of the foil temperature
(V
0
=35V).Thefoiltemperatureincreasesrapidlytothemaximumvalueat
earlytimes(forlessthan0.04safterthebeginningofthecapacitor-discharge).
Thepressurewave(V
0
=35V)inthesupercriticalcarbondioxideisshownin
Fig.127.Duringthefirstseveralacousticcycles(Fig.127(a))thewaveshape
showsthesamesteepfrontsfollowedbysomenoisesasingases(SectionV).
Thenoiseinthedataiscausedbytheunevenfoilsurfaceandunsmoothinternal
surface of the cylinder. The measured acoustic speed is in a good match
withvaluepredictedbytheNISTStandardReferenceDatabase12[92].

0
800
600
p
(
P
a
)
400
200
0
200
p
(
P
a
)
400
600
0 1 2 3 4
0.11 0.115 0.12
200
0.1 0.105
t (s) t (s)
311
317
315
313
T
(
K
)
305
307
309
311
313
315
317
T
(
K
)
309
307
305
0.1 0.11 0.12 0.13 0.14 0.15 0.16 0 1 2 3 4
t(s) t(s)
128 B.FAROUKETAL.
FIG.126. Temporalvariationofthefoiltemperature(V
0
=35V).(a)Earlytime.(b)Longer
timescale.
FIG. 127. Temporal variation of pressure (p p
i
) wave (V
0
=35 V). (a) Early time;
(b)Longertimescale.
The effect of capacitor charging voltage on the foil temperature was
studied in the supercritical carbon dioxide experiments (p
i
=7.69MPa and
T
i
=307K). Figure 128 shows the temporal variations of pressure (p p
i
)
wave for different capacitor charging voltages. In Fig. 128 (a), we see that
the wave speed and the wave shape are the same for different capacitor
charging voltages. At the early time the wave amplitude is found to be
dependent on the foil temperature increase rate, which is determined by
thecapacitorchargingvoltage.

0
p
(
P
a
)
200
15V
15V
20V
25V
35V
20V
25V
35V
p
(
P
a
)
200
0
200
400
0.1 0.105 0.11 0.115 0 1 2 3
t (s) t (s)
600 600
400
200
4
FIG.128. Temporalvariationsofpressure(pp
i
)wave(showingtheeffectofthecapacitor
chargingvoltage).(a)Earlytime;(b)Longertimescale.
1. ComparisonofNumericalandExperimentalResults
Comparisonofnumericalandexperimentalresultsforthermallyinduced
acousticwavesinsupercriticalcarbondioxideatearlytimewascarriedout.
A0.01-m-longone-dimensional problem geometrywas employed forthe
numerical simulation (see Section V), although the length of the cylinder
filled with supercritical fluids in experimental is 0.257m. The problem is
essentially one-dimensional at early time. The mesh size was 500 and the
Courant number ((cDt,Dx)) was 0.4 for the numerical simulation. In the
numerical model, the temperature boundary condition is determined from
theexperimentalmeasurements.
Figure 129 compares the experimental and computational pressure waves
(V
0
=35V). The vertical axis is pressure (pp
i
) and the horizontal axis is the
non-dimensionaltimetc/L, where tistime,cisacousticspeedandLislength.
The pressure wave speed and amplitude predicted in the numerical results
compare well with the experimental results. Furthermore, as the wave travels
betweentheleftandrightwalls,thepredictedpressureriseofbulkfluidcompares
wellwiththemeasurements.Themeasurementsareabitnoisyduetotheuneven
foilsurface.Eachwavepeakintheexperimentalresultsiswiderthanthatinthe
numericalresults.TheB&Kmicrophoneactuallymeasurestheaveragepressure
withintheregioncoveredbythediaphragm,whilethenumericalresultsshowthe
pressurerightatthecentercellofthecomputationaldomain.
2. Conclusions
Experimental measurements were conducted to characterize the genera-
tion, propagation and damping of thermally induced acoustic waves in
0
0
100
200
300
400
Experimentalresults
Numercialresults
500
1 2 3
p

p
i
(
P
a
)
4 5
t
c
/L
130 B.FAROUKETAL.
FIG.129. Comparisonofexperimentalandcomputationaltemporalvariationsofpressure
atthecenterbetweentheleftandrightwallsinsupercriticalcarbondioxide(V
0
=35V).
supercritical carbon dioxide (p
i
=7.69MPa and T
i
=307K). In the experi-
ments, the generation, propagation and damping of pressure waves for
different capacitor charging voltage were investigated, and the strength of
the thermally induced acoustic wave was found to be dependent on the
temperature increase rate of the foil at early time. Low frequency distur-
banceswasobservedinbothearlytimeandlongertimepressureprofiles,due
to the fluid flow within the supercritical chamber induced by the non-
uniform temperature field caused by the heating tape. At early times, the
computationalandexperimentalresultsareingoodagreement.
VIII. SummaryandConclusions
Acoustic waves in compressible media are generated by two kinds of
sources: rapid heating of the media, and mechanical vibration along a
bounding/immersed solid wall. The chapter reviews the recent computa-
tionalandexperimentalworkoninteractionsofacousticwaveswithviscous
andconductingfluids.Theflowfield,andheattransferassociatedwithboth
typesofgenerationmethodarecharacterizedbymeasurementsandsimula-
tions. The effect of gravity on the propagation and decay of the acoustic
wavesarealsoreviewed.
For high-intensity standing waves driven by wall vibration, the second-
ordersteadyacousticstreamingduetoacoustic-boundarylayerinteractions
are investigated both experimentally and numerically. The convection
induced by acoustic streaming is found to have a marked effect on heat
transfer.Inturn,thewalltemperaturevariationalsochangeboththestruc-
ture of acoustic streaming and streaming velocity due to the nonuniform
fluidproperties.
Anexperimentalsetupisdescribedthatallowedustocharacterizeacoustic
waves and visualize the flow field produced. The developed computational
modelsprovidedynamicsimulationsofthetransportphenomenadrivenby
theinteractionsofacousticwaveswithviscousflowsandheattransfer.The
modelshavebeenverifiedbytheexperimentaldata,andvarioussimulation
results in both Cartesian and cylindrical coordinates are presented in the
chapter. The models developed can be used as flexible design tools in
industrialapplications.
Thegeneration,propagationand damping ofthermallyinducedacoustic
wavesinanenclosure(filledwithagasorwithsupercriticalcarbondioxide)
werestudiedbyexperimentalandnumericalmethods.Inbothexperimental
andnumericalinvestigations,thermallyinducedacousticwavesweregener-
ated by rapidly increasing the wall temperature of the enclosure, and the
strengthofthepressurewaveswasfoundtobedirectlyproportionaltothe
temperatureincreaserate.Thecomputationalandexperimentalresultsarein
goodagreementwitheachother.
Thenumericalmodeldevelopedforthermallyinducedacousticwaveswas
applied for a high pressure carbon dioxide. Simulations of thermoacoustic
waveswereobtainedinenclosureswithinitialpressurevaluesvaryingfrom
0.1MPa to 8MPa and an initial temperature of 310K. The pressure and
temperaturevaluesconsideredincludeideal,realgasandsupercriticalstates
ofcarbondioxide.Thepressurewaveshapesobservedinbothrealgasand
supercritical states are quite similar to those observed for an ideal gas.
However,thestrengthofthepressurewavesincreasewithincreasinginitial
pressure, which is explained by the variations of fluid properties with
pressure.
Theeffectsofthethermallyinducedacousticphenomenaonthetransient
natural convection process in an enclosure were studied by solving the
unsteady compressible Navier-Stokes equations. The effect of gravitational
accelerationwasfoundnegligibleforearlytimes.Astheacousticwaveslose
their strength, buoyancy inducednatural convectioncurrents dominate the
flowfield.Thedevelopmentofbuoyancyinducedflowishowever,strongly
affected by the strength of the thermally induced acoustic wave. It is also
observed thatthethermallygeneratedflowfieldandthebuoyancyinduced
flowfieldarebothaffectedbytheaspectratiooftheenclosure.
132 B.FAROUKETAL.
The generation, propagation and decay of thermally induced acoustic
waves in supercritical carbon dioxide gas are also reported. A numerical
studywascarriedouttoinvestigatethecharacteristicsofthesewavesinnear-
critical (super) carbon dioxide gas. The equation of state and property
variations of the fluid were obtained via NIST Reference Database 12.
Experimentalmeasurementsofthermoacousticwavesinsupercriticalcarbon
dioxidearealsoreported.
References
1. Trilling,L.(1955).Onthermallyinducedsoundfields.J.Acoust.Soc.Am.27,425431.
2. Zappoli,B., Amiroudine, S., Carles, P., and Ouazzani, J. (1996). Thermoacoustic and
Buoyancy-driven transport in a square side-heated cavity filled with a near-critical fluid.
J.FluidMech.316,5372.
3. Swift,G.W.(1995).Thermoacousticenginesandrefrigerators.Phys.Today48,22.
4. Boluriaan, S. and Morris, P. J. (2003). Acoustic streaming: from Rayleigh to today. Int.
J.Aeroacoust.2,255292.
5. Schlichting, H. (1932). Calculation of plane periodic boundary-layer streaming. Physik
Zeitschr.33,327335.
6. Andres, J. M. and Ingard, U. (1953). Acoustic streaming at low Reynolds numbers.
J.Acoust.Soc.Am.25,932938.
7. Andres, J. M. and Ingard, U. (1953). Acoustic streaming at high Reynolds numbers.
8. Lee, C. P. and Wang, T. G. (1990). Outer acoustic streaming.J. Acoust. Soc. Am. 88,
23672375.
9. Qi,Q.(1993).Theeffectofcompressibilityonacousticstreamingneararigidboundaryfor
aplanetravelingwave.J.Acoust.Soc.Am.94,10901098.
10. Qi, Q., Johnson, R. E., and Harris, J. G. (1995). Boundary layer attenuation and
acoustic streaming accompanying plane-wave propagation in a tube. J. Acoust. Soc.
Am. 97, 14991509.
11. Vainshtein,P.(1995).RayleighstreamingatlargeReynoldsnumberanditseffectonshear
flow.J.FluidMech.285,249264.
12. Menguy, L. and Gilbert, J. (2000). Non-linear acoustic streaming accompanying a plane
stationarywaveinaguide.Acoustica86,249259.
13. Andrade, E. N. (1931). On the circulations caused by the vibration of air in a tube. Proc.
R.Soc.Lon.A134,445470.
14. West,G.D.(1951).Circulationsoccurringinacousticphenomena.Proc.Phys.Soc.B64,
483487.
15. Holtsmark,J.,Johnsen,I.,Sikkeland,T.,andSkavlem,S.(1954).Boundarylayerflownear
acylindricalobstacleinanoscillatingincompressiblefluid.J.Acoust.Soc.Am.26,2639.
16. Raney,W.P.,Corelli,J.C.,andWestervelt,P.J.(1954).Acousticstreaminginthevicinity
ofacylinder.J.Acoust.Soc.Am.26,10061014.
17. Coppens,A.B.andSanders,J.V.(1968).Finite-amplitudestandingwavesinrigid-walled
tubes.J.Acoust.Soc.Am.43,516529.
18. Cruikshank, D. B. (1972). Experimental investigation of finite-amplitude acoustic oscilla-
tionsinclosedtubes.J.Acoust.Soc.Am.52,10241034.
19. Trinh,E.H.andRobey,J.L.(1994).Experimentalstudyofstreamingflowsassociatedwith
ultrasoniclevitators.Phys.Fluids6,35673579.
20. Kempton,A.J.(1976).Heatdiffusionasasourceofaerodynamicsound.J.FluidMech.78,
131.
21. Morfey, C. L. (1976). Sound radiation due to unsteady dissipation in turbulent flow.
J.SoundVib.48,95111.
22. Fand,R.M.andHaye,J.(1960).Acousticstreamingnearaheatedcylinder.J.Acoust.Soc.
Am.32,579584.
23. Richardson, P. D. (1967). Heat transfer from a circular cylinder by acoustic streaming.
24. Richardson, P. D. (1969). Local effect of horizontal and vertical sound fields on natural
convectionfromahorizontalcylinder.J.SoundVib.10,3241.
25. Engelbrecht,H.andPretorius,L.(1992).Theeffectofsoundonnaturalconvectionfroma
verticalflatplate.J.SoundVib.158,213218.
26. Gopinath,A.andMills,A.F.(1993).Convectiveheattransferfromasphereduetoacoustic
streaming.J.HeatTransf.115,332341.
27. Gopinath, A. and Mills, A. F. (1994). Convective heat transfer due to acoustic streaming
acrosstheendsofaKundttube.J.HeatTransf.116,4753.
28. Vainshtein, P., Fichman, M., and Gutfinger, C. (1995). Acoustic enhancement of heat
transferbetweentwoparallelplates.Int.J.HeatMassTransf.38,18931899.
29. Mozurkewich, G. (1995). Heat transfer from a cylinder in an acoustic standing wave.
30. Gopinath, A. and Harder, D. R. (2000). An experimental study of heat transfer from a
cylinderinlow-amplitudezero-meanoscillatoryflows.Int.J.HeatMassTransf.43,505520.
31. Kawahashi, M. and Arakawa, M. (1996). Nonlinear phenomena induced by finite-
amplitude oscillation of air column in closed duct. JSME Int. J. 39, 280286.
32. Kawahashi,M.,Tanahashi,M.,Arakawa,M.,andHirahara,H.(1995).Visualizationand
measurement of acoustic streaming coupling with natural convection. Presented at ASME/
JSME Fluids Engineering and Laser Anemometry Conference and Exhibition, Hilton
Head,SC,USA.
33. Loh, B., Hyun, S., Ro, P. I., and Kleinstreuer,C. (2002). Acoustic streaming induced
by ultrasonic flexural vibrations and associated enhancement of convective heat transfer.
34. Loh, B. and Lee, D. (2004). Heat transfer characteristics of acoustic streaming by long-
itudinalultrasonicvibration.J.Thermophys.HeatTransf.18,9499.
35. Mozurkewich,G.(2002).Heattransportbyacousticstreamingwithinacylindricalresonator.
Appl.Acoust.63,713735.
36. Wan, Q. and Kuznetsov, A. V. (2003). Numerical study of the efficiency ofacoustic
streamingforenhancingheat transferbetweentwo parallelbeams. FlowTurbul.Combust.
70,89114.
37. Wan, Q., Wu, T., Chastain, J., Roberts, W. L., Kuznetsov, A. V., and Ro, P. I. (2005).
Forced convective cooling via acoustic streaming in a narrow channel established
byavibratingpiezoelectricbimorph.FlowTurbul.Combust.74,195206.
38. Hamilton, M. F., Ilinskii, Y. A., and Zabolotskaya, E. A. (2003). Acoustic streaming
generated by standing waves in two-dimensional channels of arbitrary width. J. Acoust.
Soc.Am.113,153160.
39. Hamilton, M. F., Ilinskii, Y. A., and Zabolotskaya, E. A. (2003). Thermal effectson
acousticstreaminginstandingwaves.J.Acoust.Soc.Am.114,30923101.
134 B.FAROUKETAL.
40. Wan, Q. and Kuznetsov, A. V. (2005). Investigation of Hysteresis in acoustically driven
channelflowatultrasonicfrequency.Numer.HeatTransf.A47,137146.
41. Yano, T. (1999). Turbulent acoustic streaming excited by resonant gas oscillation with
periodicshockwavesinaclosedtube.J.Acoust.Soc.Am.106,L7L12.
42. Aktas, M. K. and Farouk, B. (2004). Numerical simulation of acoustic streaming
generated by finite-amplitude resonant oscillation in an enclosure. J. Acoust. Soc. Am.
116, 28222831.
43. Aktas, M. K., Farouk, B., and Lin, Y. (2005). Heat transfer enhancement by acoustic
streaminginanenclosure.ASMEJ.HeatTransf.127,13131321.
44. Larkin, B. K. (1967). Heat flowto a confined fluid in zero gravity. In Progress in
Astronautic and Aeronautics Thermophysics of Spacecraft and Planetary Bodies
(G.B.Heller,ed.),pp.819832.Academic,NewYork,NY.
45. Kassoy, D. R. (1979). The response ofa confined gas to a thermal disturbance: slow
transients.SIAMJ.Appl.Math.36,624634.
46. Radhwan, A. M. and Kassoy, D. R. (1984). The response of a confined gas to a thermal
disturbance:rapidboundaryheating.J.Eng.Math.18,133156.
47. Ozoe, H., Sato, N., and Churchill, S. W. (1980). The effect of variousparameters on
thermoacousticconvection.Chem.Eng.Commun.5,203221.
48. Ozoe, H., Sato, N., and Churchill, S. W. (1990). Numerical analyses of two and three
dimensionalthermoacousticconvectiongeneratedbyatransientstepinthetemperatureof
onewall.Numer.HeatTransf.A18,115.
49. Churchill,S.W.andBrown,M.A.(1987).Thermoacousticconvectionandthehyperbolic
equationofconduction.Int.Commun.HeatMassTransf.14,647655.
50. Brown, M. A. and Churchill, S. W. (1999). Finite-difference computation of thewave
motion generated in a gas by a rapid increase in the bounding temperature. Comput.
Chem.Eng.23,357376.
51. Huang, Y. and Bau, H. H. (1995). Thermoacoustic waves in a semi-infinite medium. Int.
J.HeatMassTransf.38,13291345.
52. Huang,Y.andBau,H.H.(1997).Thermoacousticwavesinaconfinedmedium.Int.J.Heat
MassTransf.40,407419.
53. Farouk,B.,Oran,E.S.,andFusegi,T.(2000).Numericalstudyofthermoacousticwavesin
anenclosure.Phys.Fluids12,10521061.
54. Aktas,M.K.andFarouk, B.(2003).Numericalsimulationofdevelopingnaturalconvec-
tioninanenclosureduetorapidheating.Int.J.HeatMassTransf.46,22532261.
55. Parang, M. and Salah-Eddine, A. (1984). Thermoacoustic convection heat-transfer
phenomenon.AIAAJ.22,10201022.
56. Brown,M.A.andChurchill,S.W.(1995).Experimentalmeasurementsofpressurewaves
generatedbyimpulsiveheatingofasurface.AIChEJ.41,205213.
57. Nitsche,K.andStraub,J.(1986).Naturewissenschaften73,370376.
58. Nitsche, K. and Straub, J. (1986). The critical hump of Cv under microgravity result
from the D-1 spacelab experiment. Presented at Proceedings of the Sixth European
Symposium on Materials Sciences under Microgravity Conditions, ESA SP-256,
Bordeaux, France.
59. Zappoli,B.,Bailly,D.,Garrabos,Y.,LeNeindre,B.,Guenoun,P.,andBeysens,D.(1990).
Anomalous heat transport by the piston effect in supercritical fluids under zero gravity.
Phys.Rev.A41,22242267.
60. Onuki, A., Hao, H., and Ferrell, R. A. (1990). Fast adiabatic equilibrium ina single
componentfluidneartheliquid-vaporcriticalpoint.Phys.Rev.A41,22562279.
61. Boukari,H.,Shumeyer,J.N.,Briggs,M.E.,andGammon,R.W.(1990).Criticalspeeding
upinpurefluids.Phys.Rev.A41,226022263.
62. Bonetti, M., Perrot, F., Beysens, D., and Garrabos, Y. (1994). Fast thermalization in
supercriticalfluids.Phys.Rev.E49,47794782.
63. Straub, J., Eicher, L., and Haupt, A. (1995).Dynamic temperature propagationin a pure
fluid near its critical point observed under microgravity during the German Spacelab
MissionD-2. Phys.Rev.E51,55565563.
64. Guenoun, P., Khalil, B., Beysens, D., Garrabos, Y., Kammoun, P., and Le Neindre, B.
(1993). Thermal cycle around the critical point of carbon dioxide under reduced gravity.
Phys.Rev.E47,15311540.
65. Berg,R.F.andMoldover,M.R.(1999).Frequency-dependentviscosityofXenonnearthe
criticalpoint.Phys.Rev.E60,40794087.
66. Berg,R.F.,Moldover,M.R.,andZimmereli,G.A.(1999).ViscoelasticityofXenonnear
thecriticalpoint.Phys.Rev.Lett.82,920923.
67. Lemmon, E. W., Peskin, A. P., McLinden, M. O., and Friend, D. G. (2000). NIST
ThermodynamicandTransportPropertiesofPureFluids.NationalInstituteofStandard
andTechnology,Gaithersburg,MD,StandardReferenceDataProgram.
68. Oran,E.S.andBoris,J.P.(2000).NumericalSimulationofReactiveFlows.Cambridge
UniversityPress,Cambridge,UK.
69. Poinsot, T. J. and Lele, S. K. (1992). Boundary conditions for direct simulationsof
compressibleviscousflows.J.Comput.Phys.101,104129.
70. Saint-Martin-Tillet, X. N. and Oran, E. S. (1997). Boundary Conditions for FCT Based
SolutionsoftheNavier-StokesEquations.NavalResearchLaboratory,Washington,DC.
71. Nyborg,W.L.(1998).Acousticstreaming.In NonlinearAcoustics(M.F.Hamiltonand
D.T.Blackstock,eds.),pp.207231,AcademicPress,NewYork,NY.
72. Rayleigh., L. (1884). On the circulation of air observed in Kundts tubes.Philos. Trans.
A175,121.
73. Westervelt, P. J. (1953). The theory of steady rotational flow generated by a sound field.
74. Nyborg, W. L. (1958). Acoustic streamingnear a boundary. J. Acoust. Soc. Am. 30,
329339.
75. Aktas,M.andFarouk,B.(2004).Numericalsimulationofacousticstreaminggeneratedby
finiteamplitudestandingwavesinanenclosure.J.Acoust.Soc.Am.116,28222831.
76. Chester,W.(1964).Resonantoscillationsinclosedtubes.J.FluidMech.18,4464.
77. Alexeev,A.andGutfinger,C.(2003).Resonancegasoscillationsinclosedtubes:numerical
studyandexperiments.Phys.Fluids15,33973408.
78. Lin, Y. (2007). Acoustic wave induced convection and transport in gases under normal
andmicro-gravity conditions,Ph.D.thesis,DrexelUniversity,Philadelphia,PA.
79. Lin, Y. and Farouk, B. (2008). Experimental and numerical studies of thermally induced
acousticwaveinaclosedenclosure.AIAAJ.Thermophys.HeatTransf.22,105114.
80.Lin, Y. and Farouk, B. (2008). Heat transfer in a rectangular chamber with differentially
heatedhorizontalwalls:effectsofavibratingsidewall.Int.J.HeatMassTransf.51,31793189.
81. Lemmon,E.W.,Peskin,A.P.,McLinden,M.O.,andFriend,D.G.(2000).NISTStandard
ReferenceDatabase12:ThermodynamicandTransportPropertiesofPureFluids,version5.
National Institute of Standards and Technology, Standard Reference Data program,
Gaithersburg,MD.
82. Lin,Y.,Farouk,B.,andOran,E.S.(2006).Flowsinducedbythermoacousticwavesinan
enclosure:effectsofgravity.J.Thermophys.HeatTransf.20,376383.
83.Lin, Y., Farouk, B., and Oran, E. S. (2008). Interactions of thermally induced acoustic waves
withbuoyancyinducedflowsinrectangularenclosures.Int.J.HeatMassTransf.51,16651674.
84. Bruel & Kjr (2006). Acoustic Transducers: Pressure-Field Microphones. http://www.
bkhome.com
136 B.FAROUKETAL.
85. Kline,S.J.andMcClintock,F.A.(1953).Describinguncertaintiesinsingle-sampleexperi-
ments.Mech.Eng.75,38.
86. Wagner, H. and Bayazitoglu, Y. (2002). Thermal disturbances in near critical fluids.
J.Thermophys.HeatTransf.16,273279.
87. Wagner, H., Hos, P., and Bayazitoglu, Y. (2001). Variable property piston effect.
J.Thermophys.HeatTransf.15,497503.
88. Patankar, S. V. (1980). Numerical Heat Transfer and Fluid Flow. Hemisphere,
Washington,DC.
89. Lei,Z.andFarouk,B.(2007).Generationandpropagationofthermallyinducedacoustic
waves in supercritical carbon dioxide. In ASME International Mechanical Engineering
CongressandExposition.pp.18,Seattle,Washington,DC.
90. Gillis,K.A.,Shinder,I.I.,andMoldover,M.R.(2004).Thermoacousticboundarylayers
neartheliquid-vaporcriticalpoint.Phys.Rev.E70,119.
91. Zappoli, B., Amiroudine,S., Carles, P., and Ouazzani, J. (1996). Thermoacoustic and
Buoyancy-driven transport in a square side-heated cavity filled with a near-critical fluid.
92. Lemmon,E.W.,Peskin,A.P.,McLinden,M.O.,andFriend,D.G.(2000).NISTStandard
Reference Database 12: Thermodynamic and Transport Properties of Pure Fluids,
Version5.0.NationalInstituteofStandardsandTechnology.
93. Brown,M.A.andChurchill,S.W.(1995).Experimentalmeasurementsofpressurewaves
generatedbyimpulsiveheatingofasurface.AIChEJ.41,205213.
94. Lin, Y. and Farouk, B. (2008). Experimental and numerical studies of thermally induced
acousticwavesinaclosedenclosure.AIAAJ.Thermophys.HeatTransf.22,105114.

Farouk 2010 Advances in Heat Transfer

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Farouk 2010 Advances in Heat Transfer

Uploaded by

Copyright:

Available Formats

ADVANCESINHEATTRANSFERVOL.

0 1000 2000 3000 4000 5000 6000 7000 8000 900010,000

2E-05 2.5E-05 3E-05

0 5E-06 1E-05 1.5E-05 2E-05 2.5E-05 3E-05

0 5E-06 1E-05 1.5E-05

(a) 0.2 (b) 0.2

FIG. 81. Velocity vectors at t=1.27 s (t

(a) 0.1 (b)

0 0.0005 0.001 0.0015 0.002

You might also like