Welcome to Scribd, the world's digital library. Read, publish, and share books and documents. See more
Download
Standard view
Full view
of .
Look up keyword
Like this
3Activity
0 of .
Results for:
No results containing your search query
P. 1
GetPDFServlet Filetype=PDF&Id=JHTRAO000132000003031702000001&Idtype=Cvips

GetPDFServlet Filetype=PDF&Id=JHTRAO000132000003031702000001&Idtype=Cvips

Ratings: (0)|Views: 54|Likes:
Published by patil_vc

More info:

Published by: patil_vc on Feb 05, 2010
Copyright:Attribution Non-commercial

Availability:

Read on Scribd mobile: iPhone, iPad and Android.
download as PDF, TXT or read online from Scribd
See more
See less

11/24/2011

pdf

text

original

 
Boris Schilder
Department of Mechanical Engineering,Technische Universitaet Darmstadt,Petersenstrasse 30,64287 Darmstadt, Germany
Simon Yu Ching Man
School of Mechanical and AerospaceEngineering,Nanyang Technological University,50 Nanyang Avenue,Singapore 639798, Singapore
Nobuhide Kasagi
Department of Mechanical Engineering,University of Tokyo,Tokyo 113-8656, Japan
Steffen Hardt
Center of Smart Interfaces,Technische Universitaet Darmstadt,Darmstadt D-64287, Germany
Peter Stephan
1
Department of Mechanical Engineering,Technische Universitaet Darmstadt,Petersenstrasse 30,64287 Darmstadt, Germanye-mail: p.stephan@ttd.tu-darmstadt.de
Flow Visualization and LocalMeasurement of ForcedConvection Heat Transfer in aMicrotube
The pressure drop and the convective heat transfer characteristics of ethanol and water in a circular tube with a diameter of 600
m with and without phase change have beenstudied experimentally. The test section consists of a glass tube coated with a transparent indium tin oxide heater film. For single-phase liquid flow (including superheated liquid)it was found that the measured Nusselt numbers and friction factors are in good agree-ment with the theoretical values expected from Poiseuille flow. Subsequently, the boilingheat transfer of ethanol was studied. It was found that boiling with bubble growth in bothupstream and downstream directions leaving behind a thin evaporating liquid film on thetube wall is the dominant phase change process. Wavy patterns on the film surfaceindicate shear forces between vapor and liquid phase during slug flow. Temporary dryout  phenomena occur even at a low mean vapor quality due to film rupture as a result of filminstabilities. Local Nusselt numbers are calculated for the two-phase flow at different heat fluxes and Reynolds numbers. Compared with single-phase flow the heat transfer isenhanced by a factor of 3–8.
DOI: 10.1115/1.4000046
1 Introduction
Heat and mass transfer in microchannels have been in the focusof intense research activities in the past decade due to their rel-evance in fields such as electronic equipment cooling and lab-on-a-chip technology. With regard to electronics, the heat flux densityin microelectronic circuits has been constantly increasing, de-manding more efficient cooling technologies. In this context, boil-ing heat transfer in microchannels or microtubes has been identi-fied as a method for removing high heat fluxes.Despite the amount of research work in this area, still numerousunexplained phenomena exist and conflicting results have beenreported by different researchers around the globe. For example,the Nusselt numbers for single-phase flow were found to varyfrom values less than the corresponding value for the Poiseuilleflow
1
to values three times higher than that. The friction factorsfor microchannels and tubes have also shown scattering results,and many researchers attributed the effects to the surface condi-tions of the channels. A good summary of the topic can readily befound in the papers by Sobhan and Garimella
2
, Palm
3
, andmore recently Morini
4
. The scatter of the experimental resultsmay be, in part, due to difficulties associated with the experimen-tal setup and the quantifications of uncertainty levels, the tempera-ture measurements, in particular. Due to the size of the tubesconsidered, direct temperature measurements on the inner walland in the liquid were not possible; they were normally derivedfrom the measurements of the outer wall temperature. Lelea et al.
5
and Celata et al.
6
are among the few who have shownresults that are close to the predictions of the classical theory.Their experiments were conducted in a vacuum chamber to mini-mize heat losses.For experiments involving phase change, one of the prime in-terests has been on the visualization of the bubble formation pro-cess. In many of the experiments previously conducted, nucleateboiling, plug flow, slug flow, and annular flow were identified ascommon flow patterns for microchannels and microtubes
7,8
. Incontrast to macrosize tubes, in microtubes these flow patternstypically alternate with one another even at constant heat andmass fluxes. Besides these, unique microchannel flow patternshave also been described. Hetsroni et al.
9
found a rapid bubblegrowth phenomenon and called it explosive evaporation due to theshort timescales observed. Zhang et al.
10
postulated differentboiling mechanisms depending on the channel size. For channelslarger than 100
m in diameter, they expect nucleate boiling tobe the major heat transfer process. In channels smaller than50
m, they identified explosive boiling without bubble nucle-ation as being dominant. Hardt et al.
11
observed explosive boil-ing processes with subsequent film evaporation in channels with
h
=50
m and assumed bubble nucleation to trigger this process.In the present investigation, an indium tin oxide
ITO
coatedmicroglass tube has been used for heat and mass transfer studies.The arrangement can effectively generate a uniform heat fluxalong the outer surface of the tube without providing optical ob-struction to the test section. By employing a high speed camera,the phase change at high heat fluxes could be visualized clearly.
2 Experimental Arrangement and Procedure
Figure 1 shows a schematic view of the experimental setupused in the present investigation. Distilled and degassed water anddegassed ethanol are used as the working fluids for the convectiveheat transfer experiments. The working fluid is delivered to thetest section via a HPLC Pump
GL Sciences, PU 714
. The flow
1
Corresponding author.Contributed by the Heat Transfer Division of ASME for publication in the J
OUR-NAL OF
H
EAT
T
RANSFER
. Manuscript received October 20, 2008; final manuscript re-ceived July 31, 2009; published online January 4, 2010. Assoc. Editor: YogeshJaluria.
Journal of Heat Transfer
MARCH 2010, Vol. 132
/ 031702-1Copyright © 2010 by ASME
Downloaded 22 Jan 2010 to 121.242.76.214. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 
rate is accurate to within +
/
−1
%
. Experiments have been carriedout in an inlet Reynolds number range of 25–202.The fluid temperatures before and after the glass tube
in
and
out
are measured by 0.5 mm diameter K-type thermocouples inthe inlet and outlet reservoir. The volume of the reservoirs is3.5 cm
3
each. K-type thermocouples with a diameter of 25
mare used for the measurement of the temperatures along the tube
0
6
. These thermocouples are glued by silicone adhesivewith a thermal conductivity of 1.53 W/m K to the outer wall of thetube. All thermocouples are calibrated for the experimental tem-perature range and have an accuracy of +
/
−0.1 K.Pressure measurements are performed by a Sokken Pz-77 pres-sure transducer with an accuracy of +
/
−98.1 Pa. The flow visu-alization is conducted via a high speed camera
VISION RESEARCHPHANTOM V5.0
operated at a frame rate of up to 8100 per secondwith an interrogation area of 1024
1024 pixels. The glass tubetest section is a 600
m internal diameter tube
outer diameter1000
m
with a rather smooth inner surface; only shallow cavi-ties with a diameter of 2–5
m could be identified. The outersurface of the glass tube is coated with a 2
m ITO/Ag film insuch a way that Joule heating can be applied uniformly along thecoated surface, which remains transparent. The uniformity of theelectric resistance of the ITO film is within +
/
−5
%
. Silver pastewith a specific electric resistance of 5
10
−5
cm is used toconnect the copper wires to the ITO film to supply electric cur-rent.It became evident that the ITO film is sensitive to oxidation.Figure 2 shows the electric resistance over time for the unpro-tected ITO film in the presence of air. Without heating the electricresistance monotonously increases; but if the film is exposed toexternal heating or Joule heating is applied, it decreases. There-fore, we protected the ITO film from the environmental oxygen bya 0.42
m Parylene-F coating to avoid oxidation. A constant heatflux is created along the outer surface of the tube by applying aconstant voltage between the two wires connected to the test sec-tion. The heating is stable to within +
/
−2
%
over a period of 6 h.The total length of the glass tube is 200 mm, of which a section of 110 mm is heated. The entry section in front of the heated sectionis 85 mm long to ensure hydrodynamically developed flow.The experiments were conducted as follows. The working fluidis fed to the test section at room temperature
=23°C
. The flowrate of the working fluid and the heating voltage were adjusted tothe desired values. To ensure quasisteady-state conditions, themeasurement data were taken after a waiting time of 20 min.Temperature and pressure data were recorded every 464 ms. Therecorded data of 40 measurement cycles were averaged for calcu-lating the Nusselt number and friction factor.
3 Data Reduction and Uncertainties
3.1 Estimation of Heat Losses.
In the present study the testsection is not surrounded by a vacuum chamber to allow for flowvisualization by a high speed camera. As a consequence specialefforts need to be made to account for heat losses. Preliminaryexperiments with an empty test section were conducted to esti-mate heat losses due to free convection, conduction, and radiation.The heat transfer coefficient of the heat losses is defined as
h
loss
=
q˙
loss
wall,out
amb
1
Due to the fact that no working fluid is employed in this case,the Joule heat is transferred to the ambience
q˙
loss
=
q˙
heat
=
VI 
 
out
l
heat
2
where
and
are the voltage and the current measured at thecopper wires connected to the tube,
out
is the outer diameter of the tube, and
l
heat
is the heated length. The experimentally esti-mated
h
loss
has been compared with calculations following theNusselt number correlations for natural convection at the wall of ahorizontal cylinder
12
. Depending on the tube temperature, theexperimentally derived
h
loss
is up to 40% higher than the theoret-ical value for convection only. Additional heat losses due to ra-diation and conduction
axial along the tube and via the thermo-couple and heater wires
are the likely reasons for thisdiscrepancy. Nevertheless, it was found that the heat losses areless than 10% of the total heat input for all experiments withworking fluid.A second order polynomial is used to provide a continuousrelationship for
q
loss
depending on
wall,out
amb
q˙
loss
=
wall,out
amb
2
c
1
+
wall,out
amb
c
2
+
c
3
3
c
1
,
c
2
, and
c
3
are the coefficients that fit the experimental data bestin a least-squares sense.
3.2 Calculation of Local Heat Transfer Performance.
Thelocal heat transfer performance for single-phase and two-phaseflow between the inner wall of the tube and the working fluid isexpressed by the dimensionless Nusselt number. The local Nusseltnumber is calculated for all thermocouple positions at the heatedlength of the tubeNu
loc
=
h
loc
in
liquid
4
The thermal conductivity of the pure liquid
liquid
and the innerdiameter of the tube have been used for calculating Nu
loc
both forsingle-phase and two-phase flows.The local heat transfer coefficient is given as
Fig. 1 Schematic view of the experimental setupFig. 2 Change of electric resistance of an unprotected ITO film
031702-2
Vol. 132, MARCH 2010
Transactions of the ASME
Downloaded 22 Jan 2010 to 121.242.76.214. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 
h
loc
=
q˙
loc
wall,in
bulk 
5
When pumping a working fluid through the tube, a part of theheat is transferred to the working fluid and another part to theambience. The heat transferred to the working fluid is calculatedas follows:
q˙
loc
=
q˙
heat
q˙
loss
6
q˙
loc
=
VI 
 
in
l
heat
q˙
loss
7
q˙
loss
is calculated depending on the measured temperatures usingEq.
3
. Due to the constant electric resistance of the ITO filmalong the tube,
q˙
heat
is constant over the total heated area. The heatloss
q˙
loss
is a function of the temperature difference
wall,out
amb
, which changes along the heated tube because of the heattransferred to the working fluid. Therefore,
q˙
loss
and
q˙
loc
are func-tions of the axial distance
x
.The temperature at the inner wall of the tube at each thermo-couple position is computed using the one-dimensional heat con-duction equation in cylindrical coordinates
wall,in
=
wall,out
q˙
loc
in
2
tube
ln
out
in
8
The axial conduction number introduced by Maranzana et al.
13
has been calculated to find out whether the assumption of purelyradial heat transfer is reasonableMa=
Q˙
cond,axial
Q˙
conv,axial
=
tube
l
heat
 A
cond
 M ˙c
 p
9
According to Maranzana axial conduction can be neglected if theconduction number Ma is lower than 10
−2
. Depending on the flowrate, we found values ranging between 4.0
10
−6
and 4.8
10
−5
.The average fluid temperature, referred to as bulk temperature,is calculated for each thermocouple position by taking the energybalance
bulk 
=
in
+1
 x
0
 x
th
q˙
loc
dx
 
in
 x
th
 M ˙c
 p
10
where
x
th
is the axial distance of the thermocouple measured fromthe starting point of the heated length. For the evaluation of theintegral, the heated length of the tube has been discretized by 110cylindrical elements with an increment of 1 mm. For experimentswhere boiling has been observed, the maximum bulk temperatureis specified as the saturation temperature of the liquid. The satu-ration pressure inside the tube is calculated by linear interpolationbetween the pressure in the inlet reservoir and the ambientpressure.The heat transferred from each cylindrical element to the work-ing fluid is then calculated in the same way as the local heat flux
Eq.
7

. The heat losses from the elements are estimated byusing Eq.
3
.The mean outside wall temperature of the elements is interpo-lated by a third order polynomial. The coefficients of the polyno-mial are found by fitting the polynomial to the temperatures mea-sured by the microthermocouples at the outside wall of the tube.The local vapor quality for each measurement point is calcu-lated by the energy balance
 
=1
 x
0
 x
th
q˙
loc
dx
 
in
 x
th
 M ˙
h
lv
c
 p
sat
in
h
l
v
11
The temperature dependent fluid properties, such as viscosity,density, thermal conductivity, and specific heat, are calculated bypolynomials provided in Ref.
12
. Due to a lack of physical prop-erties for superheated liquid ethanol, data for higher pressure atthe same temperature have been used. The database has been ex-trapolated to obtain liquid properties at temperatures higher than100°C.The equivalent wall heat flux due to heating of the liquid byviscous dissipation has been calculated to be less than 40 W
/
m
2
,which is in all cases less than 1% of the Joule heating and, thus,has been neglected.The standard deviation of the Nusselt number due to the mea-surement uncertainties is calculated following the Gaussian errorpropagation formula to be less than
Nu
/
Nu=
20
%
for all testruns.
4 Results and Discussion
4.1 Single-Phase Flow.
The pressure drop for water and eth-anol was measured at room temperature
liquid
=23°C
. The de-rived friction factor is well in agreement with the theory for lami-nar flow
friction factor=64
/
Re
for both liquids
Fig. 3
. Thesecondary pressure losses due to hydrodynamically developingflow and the tube inlet/exit losses have been estimated and werefound to be marginal
see the Appendix
.Figure 4 shows the typical temperature distribution for a single-phase flow heat transfer test run. The wall temperatures and thebulk temperatures are almost on parallel lines, indicating that theheat flux generated is uniformly distributed along the surface, andheat losses are marginal.Figure 5 shows the local Nusselt numbers for two Reynoldsnumbers and six different heat fluxes. They converge toward theclassical value for constant heat flux and Poiseuille flow
Nu=4.36
at about 80% of the heated tube length. Higher Nusseltnumbers at the entrance to the heated section are a consequence of 
Fig. 3 Friction factor compared with theory for laminar flowFig. 4 Outer wall, inner wall, and bulk temperature
Journal of Heat Transfer
MARCH 2010, Vol. 132
/ 031702-3
Downloaded 22 Jan 2010 to 121.242.76.214. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

You're Reading a Free Preview

Download
scribd
/*********** DO NOT ALTER ANYTHING BELOW THIS LINE ! ************/ var s_code=s.t();if(s_code)document.write(s_code)//-->