Bubble confinement in flow boiling of FC-72in a ‘‘rectangular”microchannel of high aspect ratio
Jacqueline Barber a ,b ,*,David Brutin b ,Khellil Sefiane a ,Lounes Tadrist b
a School of Engineering,University of Edinburgh,The King’s Buildings,Mayfield Road,Edinburgh,EH93JL,UK
b
Aix-Marille Université(UI,UII)–CNRS Laboratoire IUSTI,UMR 6595,5Rue Enrico Fermi,Marille 134
53,France
a r t i c l e i n f o Article history:
Received 20February 2010
Received in revid form 8June 2010Accepted 12June 2010
Keywords:
Confined bubble Flow boiling Experimental
High aspect ratio microchannel Two-pha flow instabilities
a b s t r a c t
Boiling in microchannels remains elusive due to the lack of full understanding of the mechanisms involved.A powerful tool in achieving better comprehension of the mechanisms is detailed imaging and analysis of the two-pha flow at a fundamental level.Boiling is induced in a single microchannel geometry (hydraulic diameter 727l m),using a refrigerant FC-72,to investigate the effect of channel c
on-finement on bubble growth.A transparent,metallic,conductive deposit has been developed on the exte-rior of the rectangular microchannel,allowing simultaneous uniform heating and visualisation to be achieved.The data prented in this paper is for a particular ca with a uniform heat flux applied to the microchannel and inlet liquid mass flowrate held constant.In conjunction with obtaining high-speed images and videos,nsitive pressure nsors are ud to record the pressure drop across the microchan-nel over time.Bubble nucleation and growth,as well as periodic slug flow,are obrved in the microchan-nel test ction.The periodic pressure fluctuations evidenced across the microchannel are caud by the bubble dynamics and instances of vapour blockage during confined bubble growth in the channel.The variation of the aspect ratio and the interface velocities of the growing vapour slug over time,are all obrved and analyd.We follow visually the nucleation and subquent both ‘free’and ‘confined’growth of a vapour bubble during flow boiling of FC-72in a microchannel,from analysis of our results,images and video quences with the corresponding pressure data obtained.
Ó2010Elvier Inc.All rights rerved.
1.Introduction
Microchannels show great aptitude in cooling applications due to their ability to dissipate high heat fluxes.They can be ud as micro-cooling elements for laptop computer chips,electronic com-ponents [1]and aerospace avionics components,and in the design of compact evaporators and heat exchangers [2,3];in the situa-tions their compact size and heat transfer capabilities are unparal-leled.Channels in heat transfer applications have been getting increasingly smaller in dimensions due to their enhanced heat transfer performance.However,this heat transfer performance is accompanied by high pressure drops per unit length.The balance of the two factors is vital when designing a microchannel cooling system for electronics components etc.Flow boiling in microchan-nels is even more attractive than single-pha flow,due to an in-cread heat transfer coefficient,and even greater heat removal capability for a given mass flowrate of coolant,due to the latent heat involved in flow boiling flows being generally greater than
specific heat capacities for single-pha flows.Boiling flows require less pumping power than single-pha liquid flows to achieve a gi-ven heat removal.It has been reported by various rearchers that the heat transfer process and hydrodynamics occurring in micro-channels are distinctly different than that in macroscale flows [2,4–7].This implies that only some of the available macroscale knowledge can be applied to the microscale,and hence new knowledge is required to solve microscale heat tran
sfer.For exam-ple,the condition that triggers the critical heat flux (CHF)in larger channels is postulated to be when the liquid film vanishes (dry-out)at the heated channel wall [8].This is believed to be similar as the cau of CHF in microchannels,where the ont of dry-out caus CHF.It is thought that the controlling heat transfer mecha-nism in microchannels is the evaporation of the thin liquid film around the bubbles inside microchannels [9,10].There are veral general literature reviews on flow boiling heat transfer in micro-channel geometry [11,12].
The development and the progression of a liquid–vapour inter-face through a microchannel have all been well documented,but the mechanisms defining the obrvations are still unclear.Physical phenomena such as bubble confinement and thin film evaporation have been recorded by rearchers,and subquently
0894-1777/$-e front matter Ó2010Elvier Inc.All rights rerved.doi:10.pthermflusci.2010.06.011
*Corresponding author at:Aix-Marille Université(UI,UII)–CNRS Laboratoire IUSTI,UMR 6595,5Rue Enrico Fermi,Marille 13453,France.
E-mail address:barber@polytech.univ-mrs.fr (J.Barber).
attempts have been made to explain the obrvations.It is thought that surface tension,capillary forces and wall effects are dominant in small diameter channels.Various phenomena are ob-rved as the bubble diameter approaches the channel diameter; that is as the bubbles become more confined.The channel’s diam-eter can become so confining that only one bubble exists in the cross-ction,sometimes becoming elongated.This is in stark con-trast toflows en in macrochannels,where numerous bubbles can exist at one time.
Kew and Cornwell obrved threeflow patterns,including iso-lated and confined bubbles,and annular slugflow.Their rearch was carried out using R-113in parallel rectangular minichannels. They prented their confinement number equation(Eq.(2)),bad on transition criterion of confinement for boiling applications[13].
在线英汉互译软件Co¼
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
r=ðgðq
L
Àq VÞÞ
p
d
ð1Þeleft
To fully understand the high heat transfer potential of boiling flows in microchannels,it is vital tofirst understand the mecha-nisms occurring in the small diameter channels.It has been gen-erally accepted that both nucleate boiling and convective boiling mechanisms exist in micro and minichannels;it is however the dominant mechanism that remains inconclusive.
Flow visualisation was made with a high-speed camera by Brutin et al.[14].Theflowing liquid enters the microchannel and nucleation begins.The wall superheat allows the vapour bubble produced to grow rapidly,coalescing and forming a vapour slug. Several investigators[15]have considered this vapour slug as an elongated bubble.This rapid bubble growth is greater than the rate the vapour can exit the channel.There is hence a vapour build-up in the channel,with a thin liquidfilm at the channel walls.The over-pressure produced by the vapour slugs reduces the upstream boilingflowrate.Bubble
s growing before the vapour slug slow down,and quickly the whole channel cross-ction isfilled.The va-pour needs to expand,and the liquid–vapour interface at both sides of the vapour slug is pushed upstream and downstream.This leads to the inflowing liquid being pushed back to the channel he vapour bubbles recoil,creating a reverflow in the microchannel.It has been also noted by Brutin et al.[16]that pres-sure oscillations accompany the(afore mentioned)visual obrva-tions offlow reversal in a microchannel.
Fluctuations in the pressure drop across the channel exist dur-ingflow boiling in microchannels;withflow accelerations to refill the vapour spaces.When the channel is empty,and the pressure drop across the channel has been re-established,the liquid will once again begin toflow into the channel.Bubbles will be formed again quickly,due to uniform heatflux applied at the channel walls,and the phenomena will be repeated.Physical quantities such asflowrate and pressure drop,as well as the visual obrva-tions,will be vital in understanding and predictingflow patterns andflow instabilities.Several references referred to in this paper are concerning boiling experiments in a single microchannel.It is also important to note that in the literature there are many similar experiments conducted on multiple,parallel microchannels.How-ever,since the experimental campaigns conducted in this rearch work are forflow boiling in a single microchannel only,our litera-ture review reflects this.
In the literature there have been veral rearchers that u non-uniform heating in their microchannel with a heater on a ba face(s)of the microchannel,with one face of their channels transparent for visualisation purpos.Several examples of this in the literature follow.Kenning et al.[17]had a single rect-angular minichannel of cross ction2Â1mm,heated on three sides,with the fourth side ud as aflow visualisation window. Brutin and Tadrist[14]ud a single rectangular minichannel,of cross-ction0.5Â8.0mm,with a heater adhered to the back face of the channel with a transparent plexi-glass face for visualisation purpos.The authors typically u two identically dimensioned minichannels;one which is ud to gather heat transfer data,with thermocouples placed along theflow length of the channel,and an-other that is ud purely forflow visualisation with no thermocou-ples prent.This is a common trend among rearchers.
Also in the literature there are many rearchers who provide uniform heating to their microchannels,typically via electrical resistance to metal tubing.Kew and Cornwell[2]ud as their minichannels,stainless steel circular tubing of inner diameter range 1.39–3.69mm,heated via direct current.Wambsganss et al.[6]ud a circular minichannel of inner diameter2.92mm,
Nomenclature
a height length scale in aspect ratio,m
b width length scale in aspect ratio,m Co confinement number,–
d h channel hydraulic diameter,m
d channel diameter,m
E power,J sÀ1
g gravitational constant,m sÀ2
h heat transfer coefficient,W mÀ2KÀ1 l channel heated length,m
m liquid massflowrate,kg sÀ1
P pressure bar
Q heatflux density,W mÀ2
r thickness of glass channel,m
t time period,s
T temperature,°C
U velocity,m sÀ1
w channel width,m
Greek
a thermal diffusivity,m2sÀ1
D P pressure drop(D P=P in–P out)bar e channel emissivity,–
l viscosity,kg mÀ1sÀ1
q density,kg mÀ3
r surface tension,N mÀ1
x Stefan–Boltzmann constant,5.67Â10À8W mÀ2KÀ4
Subscripts
avg average value
c convective
elec electrical
i inner dimension
in inlet conditions
o outer dimension
out outlet conditions
sat saturation
本科一批录取结果什么时候出来wall wall condition
L liquid
V vapour
reasons
1376J.Barber et al./Experimental Thermal and Fluid Science34(2010)1375–1388
constructed of stainless steel and heated via direct current.Tran et al.[18]ud two different geometry minichannels;one circular of inner diameter 2.46mm and the other rectangular of hydraulic diameter 2.40mm,both channels were heated by means of electri-cal resistance and direct current.Even though the rearchers manage to provide uniform heating to their mini and microchan-nels,they are not able to visuali the flow inside the channels simultaneously to heating it.A more detailed literature review of flow boiling in microchannels,uniform and non-uniform heating methods and the associated liquid–vapour flow instabilities during flow boiling can be found in Barber et al.[19].
In the literature there has already been rearchers who studied periodic pressure drop and temperature fluctuations during flow boiling in a single microchannel.Huh et al.[20]ud deionid water in a single horizontal microchannel,the periodic fluctuations they obrved matched the transition of two alternating flow pat-terns inside the microchannel.
There are veral rearchers in the literature who have con-ducted boiling experiments with FC-72.Chih Kuang -ducted experiments of pool boiling heat transfer on micro-cavity surfaces using FC-72[21].Yen perimentally investigated boiling heat transfer in tubes of diameter 0.19–
0.51mm with FC-72and HCFC-123[22].Muwanga and Hassan investigated the flow boiling of FC-72in a microtube,paying particular attention to the temperature profiles across their channel during flow boiling [23].An interesting work by Mukherjee and Mudawar in 2002[24],com-mented on the dramatic difference in bubble departure diameter between water and FC-72bad on the Fritz [25]Bond number cor-relation.The differences between water and FC-72exist mainly due to the difference in the surface tensions of the two liquids.The objective of this work is to look in detail at the liquid–va-pour interface behaviour and dynamics of FC-72during flow boil-ing under t experimental conditions,using various data;namely pressure nsors measuring the channel pressure drop and a high-speed camera imaging the flow.The data will be analyd and
discusd in detail,paying particular attention to the ont of bub-ble confinement in the microchannel and its associated pressure fluctuations.
2.Experimental apparatus 2.1.Working fluid
The fluid chon for the experimental campaign is perfluoro-hexane,a Fluorinert™Electronic Liquid and is commercially named FC-72.The main physical properties of this dielectric liquid are reported in Table 1.FC-72is thermally and chemically stable,compatible with nsitive materials,non-flammabl
e,non-toxic,colourless,and has no ozone depletion potential.This combination of properties,together with its particularly low viscosity,makes FC-72appropriate for applications such as heat sinks for electronic components.It is also important to note here that the latent heat of FC-72is significantly higher (88kJ/kg)than its specific heat capac-ity (1.1kJ/kg K).This backs up the theory mentioned earlier,that it is more advantageous to have a boiling flow system than a single-pha flow system,since FC-72can carry larger amounts of ther-mal energy through the latent heat of vaporisation.
The novel aspect of this rearch is the simultaneous data acqui-sition and flow visualisation.This has been achieved due to a uni-form,transparent,metallic deposit of Tantalum on the exterior walls of the rectangular microchannels investigated.This tantalum deposit is both conductive and transparent at the thickness (nm)sputtered,hence enabling simultaneous uniform heating and visual-isation of the microchannel flow.The cross-ction and sizing of the particular rectangular microchannel ud can be en schemati-cally in Fig.1,where for a hydraulic diameter microchannel 727l m,the dimensions are:d i =400l m,w i =4000l m and r =400l m,with a cross-ctional aspect ratio of 10,and approxi-mate heated channel length,l ,of 80mm.The majority of the geom-etries in real world applications of micro and minichannels are not circular.This is why our results are interesting and po
ssibly more relevant to real world applications of flow boiling in microchannels.An experimental apparatus has been designed to induce boiling,to measure parameters across the channel such as the pressure drop,and to visuali and record the phenomena occurring inside the microchannel test ction.The final flow loop consists of an injection system providing a constant liquid mass flowrate into the system,an interchangeable microchannel test ction,a condensation system,a flow visualisation system using a high-speed camera,and a data acquisition system,e Fig.2.This is all houd inside a temperature regulated box of volume 1m 3,at a regulated temperature of 34°C (at atmospheric pressure)which is below the saturation
Table 1
confusionPhysical property data of FC-72.Boiling point (1atm)56°C
Liquid density (25°C)1680kg m À3Vapour density (56°C)13.24kg m À3Liquid viscosity (25°C)0.64mPa s Vapour viscosity (56°C)–
Surface tension (25°C)
0.0105N m À1Liquid specific heat capacity (25°C) 1.1kJ kg À1K À1Latent heat of vaporisation (25°C)88kJ kg À1
Liquid thermal conductivity (25°C)0.057W m À1K À1Liquid thermal diffusivity (25°C)
3.08Â10À8m 2s À1
l
alert是什么意思
d i
r
w i
J.Barber et al./Experimental Thermal and Fluid Science 34(2010)1375–13881377
temperature (56°C)of FC-72.The pressure nsors ud (Honeywell 24PC differential ries)are accurate to ±0.25%span.The microchan-nel test ction,in which the microchannel is placed vertically,can be en in more detail in Fig.3.Further descriptions and diagrams of the experimental apparatus can be found in Barber et al.[19].
2.2.Experimental procedure
The FC-72liquid is considered clean;due to the high purity it was purchad at,however it was still degasd before entry to the flow loop to remove any non-condensables.The degassing is
Heat Sink
Pressure nsors
Electrodes
Thermocouples
ias(b)
Liquid flow direction through microchannel
1378J.Barber et al./Experimental Thermal and Fluid Science 34(2010)1375–1388
achieved by boiling the FC-72liquid in the rervoir vigorously using an imbedded320W cartridge heater for one and a half hours, which is overfive times the required time to boil the volume of FC-72liquid prent in the rervoir.It should be noted here that the degassing system is a parate unit
from the experimentalflow loop,and hence does not appear in Fig.2.The inlet liquid mass flowrate is held constant by the syringe pump during an experi-mental run.It is also possible to apply a range of heatflux to the test ction via electrical resistance and the power regulator that is connected to the microchannel.Steady state is reached after about30min in each test run.The barrel size of the syringe ud limits the operational time of the liquidflow to just under an hour, hence implying that once steady state is reached(after approxi-mately30min),there is still27min remaining to record steady state data.Bubble nucleation and growth,slugflow and other phe-nomena can be obrved in the test ction with the u of a high-speed camera and macro lens.Obrvations of the nucleation of a vapour bubble,its growth,and subquent blockage of the micro-channel cross-ction are recorded in the test ction.Pressure and temperature readings at the inlet and outlet of the microchannel in conjunction withflow visualisation images obtained provide simultaneous data.
2.3.Pressure drop data
The pressure nsors are located upstream and downstream of the microchannel,and so the measured pressure drop is the sum-mation of the pressure drops across the inlet and outlet manifolds, the microchannel,and the pressure drop resulting from the inlet and outlet contraction and expansion.The acquisition frequency for all the pressure data is133Hz.
2.4.Heat transfer data
A thorough heat transfer reduction has been performed.Initially the IR thermography data was analyd during single-phaflow conditions in the microchannel.In this way we could be confident on the uniformity of the resistive metallic deposit on the exterior of the microchannels,and we obrved the constant temperature across and along the microchannel wall beforeflow boiling took place.
Heat loss,both convective and radiative were calculated,and it was found that the maximum heat loss was<5%for all heat and massflux cas investigated.Thus implying that over95%of the power provided at the microchannel wall was transmitted directly to theflowing liquid inside the microchannel.
The power provided to thefluid(E p)was calculated bad on the power applied to the microchannel deposit(E elec)and corrected for loss to the environment(E loss),i.e.E p=E elecÀE loss.The loss include both convective and radiative heat transfer.The con-vective loss(E c)were calculated using Eq.(3),where h c is the convective heat transfer coefficient.The value obtained for E c is
0.059W.
E c¼h c AðT wallÀT inÞð2Þ
The radiative loss(E rad)were calculated using Eq.(4),where e is the emissivity of the microchannel deposit(e=0.76),and x is the Stefan–Boltzmann constant(5.67Â10À8W/m2K4).The value obtained for E rad is0.082W.
E rad¼e x AðT4wallÀT4inÞð3Þ
The power provided to thefluid inside the microchannel(E p) could then be calculated bad on Eq.(1):
E P¼E elecÀE cÀE radð4Þ
The power actually provided to thefluid via the deposit at the microchannel wall is E p%3.00W.This translates to a heatflux density of Q=4.26kW/m2.
Percentage loss from the microchannel could then be calcu-lated bad on:E p/E elec.The loss were of the magnitude of5% or less,for a worst ca scenario.Hence implying that the majorit
大学四级成绩查询官网y of the heating provided at the microchannel exterior wall(over 95%)was transmitted directly to theflowing liquid inside the microchannel,with only small loss to the surroundings.
英国教育
3.Results
Prented in the following ction are results illustrating peri-odicflow boiling,in terms of both pressure data and high-speed flow imaging.The conditions of the ca prented are:a single microchannel of hydraulic diameter727l m,rectangular cross-ction0.4Â4.0mm,heated channel length80mm,uniform heat flux(Q)applied to the microchannel is4.26kW/m2and inlet liquid massflowrate(m in)is held constant at1.33Â10À5kg/s(injection speed of inlet liquid=4.95Â10À3m/s).Several interesting fea-tures are noted during the high-speedflow imaging,namely;bub-ble nucleation and growth,slugflow and vapour blockage in the microchannel leading to pressure build-up in the channel.Due to space and time limitations only bubble confinement and its effect on the microchannel pressure drop is analyd here.
3.1.Periodic pressurefluctuations and bubble dynamics
Pressurefluctuations during two-phaflow boiling are re-corded in the microchannel test ction,e Fig.4.The pressure drop(D P)is simply the difference between the inlet and outlet pressure across th
ballroome microchannel,and is measured in mbar.The pressure signalfluctuations are correlated to the passage of vapour bubbles and slugs inside the microchannel,e Figs.5and6for the correlations.The experimental conditions for allfigures in this paper are for the same heat and massflux ca,noted previ-ously,where theflow has already pasd the ont of boiling con-dition for the hydraulic diameter channel given,bad on previous data obtained.In Fig.4the pressurefluctuations shown are tho during periodicflow boiling.
In the top plot of Fig.4,data over a time period of700s is pre-nted.Large temporalfluctuations can be en in the pressure data,for both the inlet and outlet pressure of the microchannel. It can be en in the pressure data of Fig.4,that there exists both large positive amplitudefluctuations(approximately+106mbar), and also large negative amplitudefluctuations(approximately À66mbar)with similar time periods of approximately40–50s. Thanks to the high nsitivity of our pressure measurements,we could pick outfluctuations at much smaller timescales,as en in the bottom two plots of Fig.4.The middle plot of Fig.4is over a time period of57s,which corresponds to the period oscillations highlighted in the top plot of Fig.4.This time period shown in the middle plot of Fig.4,draws attention to the constant pressure readings of both the inlet and outlet pressures between pressure fluctuations.This time period between pressurefluctu
ations might be related to veral effects,for example a nucleation waiting time period,or the microchannel wall heating time period and/or the li-quid heating time period.The effects are currently under inves-tigation.It is also interesting to note that inside the periodic oscillations,exists smaller frequency oscillations as prented in the bottom plot of Fig.4,which shows thefluctuations of the pres-sure drop across the microchannel over a time period of10s.Sim-ilar obrvations by Barber et al.[19]offlow instabilities during flow boiling of n-pentane were recorded,again with pressure
J.Barber et al./Experimental Thermal and Fluid Science34(2010)1375–13881379
