Purdue University

Purdue e-Pubs
Birck and NCN Publications Birck Nanotechnology Center

6-24-2010

Surface Roughness Effects on Flow Boiling in

Microchannels
Benjamin J. Jones
Birck Nanotechnology Center, Purdue University, [email protected]

Suresh V. Garimella
Birck Nanotechnology Center, Purdue University, [email protected]

Follow this and additional works at: http://docs.lib.purdue.edu/nanopub

Part of the Nanoscience and Nanotechnology Commons

Jones, Benjamin J. and Garimella, Suresh V., "Surface Roughness Effects on Flow Boiling in Microchannels" (2010). Birck and NCN
Publications. Paper 1491.
http://dx.doi.org/10.1115/1.4001804

This document has been made available through Purdue e-Pubs, a service of the Purdue University Libraries. Please contact [email protected] for
additional information.
 Surface Roughness Effects on
Flow Boiling in Microchannels
The influence of surface roughness on flow boiling heat transfer and pressure drop in
microchannels is experimentally explored. The microchannel heat sink employed in the
Benjamin J. Jones study consists of ten parallel, 25.4 mm long channels with nominal dimensions of 500
⫻ 500 ␮m2. The channels were produced by saw-cutting. Two of the test piece surfaces
Suresh V. Garimella1 were roughened to varying degrees with electrical discharge machining (EDM). The
e-mail: [email protected] roughness average Ra varied from 1.4 ␮m for the as-fabricated, saw-cut surface to
3.9 ␮m and 6.7 ␮m for the two roughened EDM surfaces. Deionized water was used as
NSF Cooling Technologies Research Center, the working fluid. The experiments indicate that the surface roughness has little influence
School of Mechanical Engineering, on boiling incipience and only a minor impact on saturated boiling heat transfer coeffi-
and Birck Nanotechnology Center, cients at lower heat fluxes. For wall heat fluxes above 1500 kW/ m2, the two EDM
Purdue University, surfaces (3.9 ␮m and 6.7 ␮m) have similar heat transfer coefficients that were 20–35%
585 Purdue Mall, higher than those measured for the saw-cut surface 共1.4 ␮m兲. A modified Bertsch et al.
West Lafayette, IN 47907-2088 [2009, “A Composite Heat Transfer Correlation for Saturated Flow Boiling in Small
Channels,” Int. J. Heat Mass Transfer, 52, pp. 2110–2118] correlation was found to
provide acceptable predictions of the flow boiling heat transfer coefficient over the range
of conditions tested. Analysis of the pressure drop measurements indicates that only the
roughest surface 共6.7 ␮m兲 has an adverse effect on the two-phase pressure drop.
关DOI: 10.1115/1.4001804兴

Keywords: flow boiling, microchannel, heat transfer, surface roughness

1 Introduction coated channels resulted in approximately 1.4 times greater heat

transfer coefficients than uncoated channels for most test condi-
It is well recognized that the limits of heat dissipation that can
tions considered. This level of enhancement was somewhat less
be managed by air-cooled heat sink technology, the primary
than the 1.5–3 times greater heat transfer coefficients compared
means of cooling microprocessors, are at hand. If the current pace
with smooth surfaces that have typically been reported in macros-
of development in the microelectronics industry is to be sustained,
cale tubes with similar coatings 关12,13兴.
new and innovative cooling strategies are needed. Flow boiling in
Microchannels with microfabricated notches and cavities were
microchannel heat sinks is a prime contender for replacement of explored by Zhang et al. 关14兴, Kandlikar et al. 关15兴, Koşar et al.
the air-cooled heat sink due to the very high heat transfer rates 关16,17兴, Kuo and co-workers 关18–20兴, Jones and co-workers
attainable with this technology. Over the last decade, much re- 关21,22兴, and Pate et al. 关23兴. Although the augmented channels
search has, therefore, been devoted toward better understanding resulted in lower wall superheats at incipience than smooth sur-
flow boiling phenomena in microchannels. Recent experimental faces 关14,19兴, the influence of the augmentation on heat transfer
studies 关1–4兴 have shed light on the influence of heat flux, mass rates in fully developed boiling has not been entirely conclusive.
flux, vapor quality, and channel dimensions on the flow boiling From the data of Zhang et al. 关14兴, there appeared to be little
heat transfer coefficients. Improvements in the predictability of benefit from the augmented surface on heat transfer coefficients.
flow boiling in small channels have been supported by recent Kuo and Peles 关19兴 found that the augmented surfaces resulted in
developments in models for both heat transfer 关5,6兴 and pressure higher heat transfer coefficients than a smooth surface at higher
drop 关7–9兴. Despite the advancements in the state of understand- mass fluxes; however, lower heat transfer coefficients were ob-
ing of microchannel flow boiling, several important factors influ- served at lower mass fluxes. From the studies in literature, thus
encing flow boiling in microchannel heat sinks have yet to be far, it appears that the main benefit in using these augmented
fully explored. In particular, and as will be the topic of this paper, surfaces is the suppression of flow instabilities in microchannel
a detailed study of the influence of surface roughness on flow heat sinks 关15,20,22兴.
boiling in microchannels has not been reported in literature to the Although surface roughness produced using standard machin-
authors’ knowledge. ing operations appears to have fallen out of favor as a surface
It has long been known that surface characteristics have a sig- enhancement technique in macroscale boiling 关24兴, it is still rec-
nificant impact on nucleate boiling heat transfer. Numerous inves- ognized as an important factor influencing boiling heat transfer.
tigations have focused on developing methods of enhancing heat Since different methods of manufacturing microchannels impart
transfer rates for in-tube flow boiling. Although most prior studies different levels of roughness on the channel surfaces, an under-
have considered macroscale two-phase flow systems, more re- standing of its effect on flow boiling is needed. An experimental
cently, researchers have begun exploring surface augmentation exploration of the influence of surface roughness in microchannel
techniques in minichannels and microchannels. Ammerman and flow boiling is therefore warranted. In this study, experimental
You 关10,11兴 studied flow boiling in a minichannel and a micro- results on flow boiling heat transfer and pressure drop with deion-
channel with a microporous coated bottom wall. The microporous ized water using three microchannel heat sinks of varying surface
roughness 共Ra = 1.4 ␮m, 3.9 ␮m, and 6.7 ␮m兲 are reported.
1
Corresponding author.
Contributed by the Heat Transfer Division of ASME for publication in the JOUR- 2 Experimental Setup
NAL OF THERMAL SCIENCE AND ENGINEERING APPLICATIONS. Manuscript received May
18, 2009; final manuscript received February 26, 2010; published online June 24, 2.1 Experimental Test Facility. A diagram of the experimen-
2010. Assoc. Editor: Anthony M. Jacobi. tal flow facility is shown in Fig. 1. A differential pressure trans-

Journal of Thermal Science and Engineering Applications DECEMBER 2009, Vol. 1 / 041007-1
Copyright © 2009 by ASME

Downloaded From: http://thermalscienceapplication.asmedigitalcollection.asme.org/ on 12/13/2013 Terms of Use: http://asme.org/terms

 Fig. 1 Experimental flow loop

ducer is used to measure the pressure drop across the test section
and a pressure transducer at the outlet measures gauge pressure.
The outlet pressure transducer has an uncertainty of ⫾0.31 kPa
while the differential pressure transducer has an experimental un-
certainty of ⫾0.43 kPa. Inlet pressure measurements are calcu-
lated based on the differential pressure transducer and the outlet
pressure transducer readings and, therefore, have an uncertainty of
⫾0.53 kPa.
Subcooled liquid enters the microchannel test section. The sub-
cooled liquid is maintained at the desired temperature using a
liquid-to-liquid heat exchanger in conjunction with a large, tem-
perature controlled, hot water bath circulation unit. The liquid/
vapor exiting the microchannel test section flows into a reservoir.
Fig. 2 Schematic of experimental test section
The vapor is condensed in a glass reflux condenser, which ex-
changes heat from the vapor to cold water from a chilled water
circulation unit. The condensate falls into the reservoir. The con-
denser is open to the atmosphere and due to the low pressure drop sheathed thermocouples with a sheath diameter of 0.8 mm were
in the tubing of the flow loop, atmospheric pressure is maintained used. The thermocouples were referenced to a dry-block ice-point
at the exit of the microchannels. The reservoir is vented with high reference chamber. All thermocouples were calibrated using a dry-
purity nitrogen. The purpose of the vented nitrogen is to displace block thermocouple calibration unit. Since all thermocouples were
the dissolved oxygen in the water, which can contribute to corro- calibrated against the same, stable temperature source and refer-
sion of the copper test surfaces used in the experiments. enced against a uniform ice-point reference temperature, the esti-
A gear pump is used for pumping the fluid. A 7 ␮m sintered mated uncertainty in temperature differences measured by the
filter element removes any particulates from the flow loop. A thermocouples is ⫾0.1° C. However, absolute errors in the cali-
liquid-to-air heat exchanger cools the fluid to near-ambient tem- bration and reference source lead to an estimate of ⫾0.3° C for
peratures prior to entering temperature-sensitive equipment in the absolute temperature measurements. The bottom channel wall
flow loop. A high-accuracy Coriolis flow meter measures the mass temperature is determined by extrapolation of the temperature
flow rate in the test loop. A vacuum degassifying system is used to measurements. Using standard uncertainty analysis, the estimated
remove dissolved gases in the fluid, which can lead to premature uncertainty in the wall temperature ranges from approximately
bubble nucleation and affect the boiling performance 关25,26兴. The ⫾0.4° C near boiling incipience to ⫾1.3° C at the maximum heat
vacuum degassifier consists of a flow passage constructed out of a flux obtained in the experiments. Twelve cartridge heaters, in-
hydrophobic, porous membrane. Vacuum is applied to the outside serted into holes drilled in the bottom of the test piece, supply the
of the flow passage while liquid water flows through it. The mem- desired power input. The test piece is inserted into a polyethere-
brane is gas-permeable allowing the dissolved gases to be ex- therketone 共PEEK兲 共a thermoplastic with good resistance to high
tracted from the liquid as vacuum is applied. The hydrophobic temperatures兲 housing. The cover of the test section is manufac-
nature of the membrane prevents the liquid from passing through tured out of polycarbonate with a glass insert in contact with the
the membrane’s porous structure. A nitrogen sweep gas is used to top of the copper test block. Manifold and plenum regions were
help displace some of the remaining dissolved oxygen in the wa- machined in the PEEK housing and polycarbonate cover to dis-
ter. This method of degassing was found to be very effective in tribute fluid to and from the microchannels. Thermocouple and
removing dissolved oxygen from the liquid, producing dissolved pressure taps in the inlet and outlet plenums of the PEEK housing
oxygen levels of less than 0.1 ppm after approximately 1.5 h of allow measurement of the fluid inlet and outlet temperature and
degassing. pressure drop across the test section. The lower portion of the
copper microchannel test piece is enclosed in an alumina silicate
2.2 Experimental Test Section. A schematic diagram of the ceramic shell for insulation 共not shown in Fig. 2兲.
experimental test section is shown in Fig. 2. The experimental test
piece consists of an oxygen-free copper test block with ten chan- 2.3 Microchannel Test Surfaces. Three microchannel test
nels, each nominally 500 ␮m wide by 500 ␮m high by 25.4 mm surfaces of varying surface roughness were fabricated into test
long, manufactured into the top of the block. Six thermocouple pieces. The channels for one test piece were fabricated using a
taps are drilled into the copper block 共at three different locations saw blade. The channels for the two other test pieces were rough-
along the flow direction and at two different heights兲. T-type ened using electrical discharge machining 共EDM兲. Different EDM

041007-2 / Vol. 1, DECEMBER 2009 Transactions of the ASME

Downloaded From: http://thermalscienceapplication.asmedigitalcollection.asme.org/ on 12/13/2013 Terms of Use: http://asme.org/terms

 Table 1 Test piece characteristics and test conditions. Surface roughness values indicate mean± standard deviation.

Channel dimensions Surface roughness parameters Experimental conditions

Wc Hc Ra Rq Rp Rz Tl,in G Max. xout

Surface preparation 共␮m兲 共␮m兲 共␮m兲 共␮m兲 共␮m兲 共␮m兲 共°C兲 共kg/ m2 s兲 共%兲

Saw-cut 521 496 1.4 1.8 12 18 87.3 199 46.4

86.9 609 24.0
⫾0.1 ⫾0.1 ⫾4 ⫾4
86.8 998 15.4
EDM 458 462 3.9 4.9 18 39 87.2 193 44.7
87.0 604 29.5
⫾0.3 ⫾0.3 ⫾2 ⫾2
87.0 997 19.0
510 480 6.7 8.4 34 58 87.1 203 36.5
86.4 600 25.6
⫾0.4 ⫾0.5 ⫾3 ⫾4
86.6 994 16.4

parameters were used to vary the surface roughness between the 2.4 Microchannel Characterization. The channel widths
two test pieces. These latter channels were first cut with a saw were measured using an optical microscope with a digital mi-
blade to undersized dimensions before undergoing the EDM crometer stage while the channel depths were measured with a
roughening process. Enough material was removed with the EDM Zygo NewView 7300 optical profiling system based on white
machining process to ensure that the final roughness had no re- light, vertical scanning interferometry. The optical profiling sys-
sidual features from the saw-cutting operation. It should be noted tem was also used to measure the surface roughness of the bottom
that the EDM process roughens all three heated sides of the cop-
channel surfaces. The channel dimensions were measured at ap-
per microchannel surface.
Ram-type EDM, as opposed to wire EDM, was chosen due to proximately 30 locations distributed across the microchannel heat
its ability to produce channels with a form that more closely sink while the surface profilometry was performed at 16 different
matches the shape of the channels produced with saw-cutting. A locations. The average channel dimensions and surface roughness
graphite electrode was fabricated for the EDM cutting process. measurements for each test piece are shown in Table 1. The esti-
The different EDM parameters used in roughening the two test mated uncertainty in the channel dimension measurements is
pieces necessitated a different set of electrodes for each test piece 5 ␮m. The average channel dimensions are within ⫾10% of the
in order to produce channels of similar widths since the amount of nominal 500 ␮m by 500 ␮m for each test piece. Several surface
overburn 共the gap between the electrode surface and the resulting roughness parameters are provided in Table 1, according to ASME
machined surface兲 changed with the EDM parameter adjustment. B46.1-1995. For the remainder of this paper, Ra will be used to
During the EDM roughening process, the corners of the electrodes
denote the roughness of the different surfaces. The surface topog-
tend to wear preferentially resulting in a channel with rounded
corners. This required periodic replacement of the electrodes with raphy of the bottom channel walls, as measured by the optical
a new set in order to reduce the amount of corner-rounding. How- profiling system is shown in Fig. 3. The saw-cut operation pro-
ever, the fixed overburn distance over the periphery of the elec- duces a channel with a series of grooves 共see Fig. 3共a兲兲. The EDM
trode leads to some corner-rounding being inevitable with EDM. operation produces an irregular pattern of cavities on the surface.
A black layer was formed on the electrical discharge machined The Ra = 6.7 ␮m EDM surface 共as seen in Fig. 3共c兲兲 appears to
copper surfaces. The exact nature of the black layer was not pre- have larger cavities than the Ra = 3.9 ␮m surface 共shown in Fig.
cisely known but is believed to be a carbon residue, perhaps de- 3共b兲兲.
posited from the graphite electrode or EDM bath fluid products. Representative profiles of the bottom channel wall are shown in
Since the surface wettability was likely to be affected by such a Fig. 4. These profiles were obtained from the optical profilometer
layer, it was desirable to remove the layer and expose the bare measurements by averaging the surface height along the flow di-
copper surface prior to testing. Most of the layer could be re- rection of the channel. It is evident that the saw-cut operation
moved by cleaning the test piece with deionized water in an ul-
produces a channel with a rather flat bottom wall 共see Fig. 4共a兲兲.
trasonic bath for approximately 2 h. The remainder of the layer
was removed using a weak copper etchant solution to etch the The channel profiles were nearly square as intended. For the EDM
underlying copper and lift off the black residue. The copper test channels, the profile is slightly more rounded, as is particularly
pieces were placed in an ultrasonic bath containing 0.3% hydro- evident in the Ra = 6.7 ␮m surface 共see Figs. 4共b兲 and 4共c兲兲. As
gen peroxide and 1% sulfuric acid solution. Approximately 20 discussed in the previous subsection, although attempts were
min of ultrasonic agitation with the acid solution were required to made to produce a channel cross section that was as square as
completely remove the remaining residue. possible, some rounding of the bottom wall did occur.

Fig. 3 Topographies of bottom channel surface

Journal of Thermal Science and Engineering Applications DECEMBER 2009, Vol. 1 / 041007-3

Downloaded From: http://thermalscienceapplication.asmedigitalcollection.asme.org/ on 12/13/2013 Terms of Use: http://asme.org/terms

 Lsp + Lsub Tsat,0 − Tsat,in
= 共6兲
L Tsat,out − Tsat,in
where Tsat,in and Tsat,out are the saturation temperatures evaluated
based on the inlet and outlet pressure readings, respectively. Solv-
ing Eqs. 共5兲 and 共6兲 yields the lengths of the subcooled and satu-
rated regions.
The saturated boiling heat transfer coefficient can then be cal-
culated as

q
hsat = 共7兲
T̄w − T̄sat

where T̄w is the mean wall temperature and T̄sat is the mean satu-
rated fluid temperature in the saturated boiling region. The esti-
mated uncertainty in the heat transfer coefficients ranges from
⫾25% at heat fluxes near boiling incipience to ⫾13% at higher
heat fluxes.
Fig. 4 Profiles of bottom channel surface
3.2 Pressure Drop. The pressure taps are located upstream
and downstream from the inlet and outlet of the microchannels,
respectively, 共see Fig. 2兲. Therefore, determination of the pressure
3 Data Reduction drop across the microchannels ⌬Pc requires correction of the mea-
sured pressure drop ⌬P using estimates of the pressure losses in
3.1 Heat Transfer. The rate of heat transfer to the fluid Q the manifolds and pressure losses/recoveries due to flow area
was calculated using the readings from the thermocouples embed- changes. However, calculations revealed that the significant pres-
ded in the copper test piece. The effective wall heat flux can be sure corrections are only due to the contraction from the inlet
defined as manifold to the microchannels and the expansion from the chan-
Q nels to the outlet manifold. Therefore, the channel pressure drop
q= 共1兲 can be calculated as
Aw
⌬Pc = ⌬P − ⌬Pcon + ⌬Pexp 共8兲
where
The fluid enters the microchannels as subcooled, single-phase
Aw = Nc共Wc + 2␩Hc兲 共2兲 liquid. Therefore, the contraction losses at the inlet of the micro-
channel can be estimated from Ref. 28.
and ␩ is the fin efficiency. Due to the very large fin thickness
共about 2 mm兲 relative to the channel dimensions, the fin efficiency
is close to 1 共estimated value is 0.99 for a heat transfer coefficient
of 50,000 W / m2 K兲. The estimated uncertainty in the wall heat
冋 冉 冊
⌬Pcon = 1 −
N cA c
Aman
2
+ Kcon 册冉 冊 G2
2␳l
共9兲

flux measurements ranges from ⫾22% of the measured wall heat where
flux near boiling incipience to ⫾12% at higher heat fluxes.
Since water enters the microchannel at subcooled temperatures, Kcon = 0.0088␣2 − 0.1785␣ + 1.6027 共10兲
the channel length can be partitioned into three regions. The pressure recovery due to flow expansion at the exit of the
microchannel depends on whether there is single-phase or two-
L = Lsp + Lsub + Lsat 共3兲 phase flow at this location. For a thermodynamic exit quality of
xout ⬍ 0, single-phase flow is assumed while two-phase is assumed
where Lsp, Lsub, and Lsat are the lengths of the single-phase, the for xout ⬎ 0. For single-phase flow, the pressure recovery can be
subcooled boiling, and the saturated boiling regions, respectively. expressed as 关28兴
The point of zero thermodynamic quality defines the boundary
between the subcooled and saturated boiling regions, assuming 1
thermodynamic equilibrium between the liquid and vapor phases. ⌬Pexp = KexpG2 共11兲
The local thermodynamic quality can be defined as 2␳l
where

冉 冊冉 冊
i − il,sat
x= 共4兲 N cA c N cA c
h fg Kexp = 2.66 1− 共12兲
Aman Aman
where i is the local enthalpy and il,sat is the liquid enthalpy evalu-
ated at saturated conditions corresponding to the local pressure. For two-phase flow, the pressure recovery due to expansion is
The length of the single-phase and subcooled boiling regimes given by 关29兴
can be determined from an energy balance.
1
⌬Pexp = KexpG2共1 − xout兲2␾l2 共13兲
ṁc p共Tsat,0 − Tl,in兲 2␳l
L − Lsat = Lsp + Lsub = 共5兲
q共Aw/L兲 where Kexp for two-phase flow will be taken as 关28兴
where Tsat,0 is the saturation temperature at x = 0. Assuming that
the local saturation temperature follows a linear relationship be-
tween the inlet and outlet of the microchannel as suggested by
Kexp = 2 冉 冊冉
N cA c
Aman
1−
N cA c
Aman
冊 共14兲

Collier 关27兴, the length of the subcooled regions can also be ex- The two-phase friction multiplier ␾l is calculated using the
pressed as Mishima and Hibiki 关30兴 correlation.

041007-4 / Vol. 1, DECEMBER 2009 Transactions of the ASME

Downloaded From: http://thermalscienceapplication.asmedigitalcollection.asme.org/ on 12/13/2013 Terms of Use: http://asme.org/terms

 Fig. 5 Flow boiling curves for different surface conditions at different mass fluxes

C 1 curve for the range of conditions tested. According to Lee and

␾l2 = 1 + + 共15兲 Mudawar 关32兴, the influence of mass flux on boiling heat transfer
X X2
is dependent on the experimental conditions; with some research-
where ers, similar to the current study, finding a weak dependence
C = 21关1 − exp共− 319Dh兲兴 共16兲 关2,4,33兴 while others have reported a strong dependence 关34,35兴.
Critical heat flux 共CHF兲 was experimentally achieved in the
where Dh is in meters. The Martinelli parameter depends on G = 200 kg/ m2 s and 600 kg/ m2 s cases. For the G
whether the liquid and vapor phases are in the laminar or turbulent = 1000 kg/ m2 s cases, the maximum heat flux was limited by the
regimes. For the current set of experiments, most of the data fall power supply used in the experiments rather than CHF. The criti-
in the viscous-liquid, viscous-vapor regime. However, the vapor cal heat flux measurements are shown in Fig. 6. For the lower
Reynolds number is sufficient in some instances for turbulent- mass flux case 共200 kg/ m2 s兲, the Ra = 3.9 ␮m and 6.7 ␮m sur-
vapor conditions. The Martinelli parameter is calculated as 关31兴 faces had an 18% and 22% lower CHF value than the saw-cut
2
Xvv = 冉 冊冉 冊冉 冊
1−x
x
␳v
␳l
␮l
␮v
;
Rel ⬍ 2000
Rev ⬍ 2000
共17兲
surface, respectively. At a mass flux of 600 kg/ m2 s, the 3.9 ␮m
surface had a 9% higher CHF than the saw-cut one while the
6.7 ␮m surface had a nearly identical CHF value as the saw-cut

冉 冊冉 冊冉 冊
surface. It should be noted that the precision with which the CHF
1−x ␳v ␮l Rel ⬍ 2000
Xv2t = 348 Rev−0.8 ; 共18兲 values are determined is limited to the heat flux increment used in
x ␳l ␮v Rev ⬎ 2000 the experiments near CHF 共150– 250 kW/ m2兲. Nonetheless, it
does not appear that the surface roughness has a large influence on
4 Results and Discussion the critical heat flux under the conditions tested.
The influence of surface roughness on the saturated boiling heat
4.1 Heat Transfer Results. The experimental conditions are transfer coefficients can be seen in Figs. 7共a兲–7共c兲 for mass fluxes
reported in Table 1. The inlet temperature of the test section was of 200 kg/ m2 s, 600 kg/ m2 s, and 1000 kg/ m2 s, respectively.
maintained at 87° C for all tests. Data were collected at three All mass fluxes showed a similar trend. For heat fluxes below
different mass fluxes for each experimental test piece: nominally approximately 700 kW/ m2, there does not appear to be a signifi-
200 kg/ m2 s, 600 kg/ m2 s, and 1000 kg/ m2 s. Heat fluxes up to cant effect of surface roughness on the saturated boiling heat
approximately 3000 kW/ m2 were achieved, although the heat transfer coefficient. Although for G = 1000 kg/ m2 s, the 6.7 ␮m
flux was limited to lower values by critical heat flux in some of surface does have notably higher heat transfer coefficients than
the tests. The calculated equilibrium thermodynamic exit quality
xout ranged from approximately 45% at the lowest mass flux
共200 kg/ m2 s兲 to approximately 15% at the highest
共1000 kg/ m2 s兲. The saw-cut and 3.9 ␮m EDM tests were con-
ducted at a mass flux of 1000 kg/ m2 s numerous times to ensure
repeatability of the results. For both test sections, the measured
wall temperature was repeatable to within approximately 2 ° C.
The flow boiling curves for the Ra = 1.4 ␮m, 3.9 ␮m, and
6.7 ␮m surfaces are shown in Figs. 5共a兲–5共c兲, respectively. For
low superheats, single-phase fluid flows through the entire micro-
channel. Boiling commences at wall superheats ranging from 2 K
to 3 K for a mass flux of 200 kg/ m2 s and from 6 K to 8 K for a
mass flux of 1000 kg/ m2 s. The requirement of higher wall su-
perheats at the higher mass fluxes may be due to a larger tempera-
ture gradient being present in the near wall region, producing a
thinner superheated liquid layer that suppresses nucleate boiling.
The surface roughness does not appear to have a substantial effect
on the wall superheat at incipience. Once boiling commences, a
higher slope in the boiling curve compared with the single-phase
region is apparent due to the higher heat transfer coefficients with
boiling heat transfer. In the fully developed boiling region, mass Fig. 6 Critical heat flux versus surface roughness at two dif-
flux does not appear to have a major influence on the boiling ferent mass fluxes

Journal of Thermal Science and Engineering Applications DECEMBER 2009, Vol. 1 / 041007-5

Downloaded From: http://thermalscienceapplication.asmedigitalcollection.asme.org/ on 12/13/2013 Terms of Use: http://asme.org/terms

 Fig. 7 Influence of surface roughness on saturated heat transfer coefficients at different mass fluxes

the other surfaces yet still within the estimated range of experi- the study兲 while the Bertsch et al. correlation 共Eq. 共19兲兲 predicts
mental uncertainty. At higher heat fluxes, the effect of surface heat transfer coefficients at a particular quality. Based on the defi-
roughness on the heat transfer coefficient is more marked. For nition of the saturated heat transfer coefficient in this study 共Eq.
heat fluxes above 1500 kW/ m2, the EDM surfaces have approxi- 共7兲兲, an average flow boiling heat transfer coefficient can be cal-
mately 20–35% higher heat transfer coefficients than the saw-cut culated from a local heat transfer coefficient using
surface at a given heat flux. However, there is no significant dif-
1

冕
ference in the measured heat transfer coefficients between the two hFB,avg = xout 共21兲
EDM surfaces 共3.9 ␮m and 6.7 ␮m surfaces兲 at these higher heat 1 1
fluxes. dx
xout hFB共x兲
As discussed earlier 共and as indicated in Table 1兲, the channel 0

dimensions of the three test pieces vary by ⫾10% of the nominal Figure 8共a兲 presents the comparison between the Bertsch et al.
500⫻ 500 ␮m2 channel dimensions. The experimental results of correlation predictions 共see Eq. 共19兲兲 and the experimental results.
Harirchian and Garimella 关2,4兴, using microchannels with dimen- The predictions for water with the channel exit at atmospheric
sions ranging from 100⫻ 400 ␮m2 共Wc ⫻ Hc兲 to 5850 pressure deviate from the experimental results. The reason for this
⫻ 400 ␮m2, indicate that saturated flow boiling heat transfer co- deviation is the unsuitability of the Cooper correlation at low re-
efficients are not significantly influenced by small changes in duced pressures. As described in Jones et al. 关41兴, Cooper ac-
channel dimensions. It is, therefore, believed that the differences counted for surface roughness by reformulating the Nishikawa
共1−P 兲/5
in heat transfer coefficients measured in the current tests are a and co-workers 关42,43兴 relationship of h ⬀ R p,oldr into h
result of the different surface roughness rather than due to small ⬀ Pr−0.2 log10 Rp,old. The reformulation provides a close approxima-
differences in channel dimensions. tion to the original formulation of Nishikawa et al. for reduced
4.2 Heat Transfer Predictions. Bertsch et al. 关36兴 compiled pressures above 0.08 or for a surface roughness near R p,old
an extensive review of heat transfer correlations for saturated flow = 1 ␮m but is unsuitable for lower reduced pressures when the
boiling in small channels. They found that the Cooper 关37,38兴 roughness deviates from R p,old = 1 ␮m.
pool boiling correlation outperformed most dedicated flow boiling For low reduced pressures, either the Gorenflo 关44兴 nucleate
correlations when evaluated against a large database of experi- boiling correlation or the Cooper 关37,38兴 correlation reformulated
mental results assembled from a wide range of studies in litera- to include the surface roughness factor originally proposed by
ture. The Cooper correlation resulted in mean absolute error Nishikawa et al. 关42,43兴 should prove more suitable 共see Table 2兲
共MAE兲 of 36% when compared with this large database. Bertsch for use in Eq. 共19兲. It should be noted that such a modified Cooper
et al. 关5兴 later developed a composite model based on the formu- correlation provides nearly identical predictions to the original
lation of Chen 关39兴, offering improved predictions for flow boiling formulation by Cooper for reduced pressures above 0.08. As indi-
in small channels 共MAE of 28% compared with the same data- cated in Figs. 8共b兲 and 8共c兲, the Bertsch et al. correlation utilizing
base兲. the Gorenflo or modified Cooper correlation offers much im-
proved predictions with MAEs of 22.5% and 21.8%, respectively,
hFB = hNB共1 − x兲 + hconv,tp关1 − 80共x2 − x6兲exp共− 0.6Co兲兴 共19兲 and predicting 70% and 72% of the data within a deviation of
where ⫾30%, respectively. Considering the expected errors of current

hconv,tp = hconv,l共1 − x兲 + hconv,vx 共20兲

The nucleate boiling heat transfer coefficient hNB is evaluated Table 2 Correlations for calculating nucleate boiling heat
transfer coefficient hNB
from the Cooper 关37,38兴 correlation 共see Table 2兲 and the liquid
and vapor convective heat transfer coefficients are evaluated from Author共s兲 Correlation
the Hausen 关40兴 correlation. Since the Cooper correlation already
includes a factor accounting for the effect of surface roughness, it Cooper 关37,38兴 h = 55共Pr0.12−0.2 log10 Rp,old兲共−log10 Pr兲−0.55M −0.5q0.67
is instructive to ascertain how well the Bertsch et al. flow boiling h = 55Pr0.12R共1−P r兲/5
共−log10 Pr兲−0.55M −0.5q0.67

冋 冊册
Modified Cooper
correlation 共Eq. 共19兲兲 predicts saturated heat transfer coefficients
冉
p,old
h 0.68
over the wide range of surface roughness considered in the current = 1.73Pr0.27 + 6.1 + P2
h0 1 − Pr r
study.
冉 冊 冉冊
0.15
It should be noted that the experimentally determined saturated Ra 0.133 q 0.9−0.3Pr
heat transfer coefficients presented in this study are averaged over ⫻
Ra0 q0
a thermodynamic quality ranging from 0 to the channel exit qual- Gorenflo 关44兴 q0 = 20,000 W / m2, h0 = 5600 W / m2 K, Ra0 = 0.4 ␮m
ity xout 共Table 1 presents the maximum exit qualities obtained in

041007-6 / Vol. 1, DECEMBER 2009 Transactions of the ASME

Downloaded From: http://thermalscienceapplication.asmedigitalcollection.asme.org/ on 12/13/2013 Terms of Use: http://asme.org/terms

 Fig. 8 Comparison of experimental measurements of saturated heat transfer coefficients with predictions

state-of-the-art flow boiling correlations, the modified Bertsch et pieces used in the development of the various previous pressure
al. correlations appear to offer acceptable predictions even over drop correlations. It was found that the Lee and Garimella 关9兴
the range of surface roughness considered in this study. However, correlation provided the most accurate predictions of the mea-
the reasons for the similar heat transfer characteristics of the Ra sured pressure drop results for the current set of experiments. A
= 3.9 and 6.7 ␮m surfaces are not adequately explained by the comparison of the experimental pressure drop and the predictions
current models, which illustrates the general weakness in using Ra from Lee and Garimella 关9兴 is shown in Fig. 10. The predictions
to characterize boiling surfaces. Improved methods of accounting match the experimental data with a mean absolute error of 17.6%
for the surface roughness effect on flow boiling heat transfer are with 83% of the experimental data falling within ⫾30% of the
certainly desired. predictions. Therefore, the Lee and Garimella 关9兴 correlation is
used to predict the influence of channel dimensions on pressure
4.3 Pressure Drop Results. The pressure drop results are drop.
shown in Fig. 9. At low heat fluxes in the single-phase regime, the On average, the Lee and Garimella 关9兴 correlation predicts that
pressure drop is independent of heat flux. Once boiling com- the 3.9 ␮m roughness channels should have approximately a 40%
mences, the pressure drop increases with an increase in heat flux. larger pressure drop than the 1.4 ␮m roughness channels due to
Similar pressure drop values were measured for the two EDM the smaller channel dimensions 共see Table 1兲 while the 6.7 ␮m
surfaces 共Ra = 3.9 and 6.7 ␮m兲. In the saturated boiling regime, surface is predicted to have approximately 20% larger pressure
the pressure drop is on average 40% and 45% higher for the drop than the saw-cut channels. As previously noted, this com-
3.9 ␮m and 6.7 ␮m surfaces, respectively, compared with the pares to the experimental measurements of 40% and 45% larger
1.4 ␮m surface. However, unlike the heat transfer coefficients, pressure drops for 3.9 ␮m and 6.7 ␮m surfaces, respectively,
the differences in pressure drop are due to differences in the chan- compared with the saw-cut surface. Based on this analysis, it ap-
nel dimensions in addition to possible surface roughness effects. It pears that the larger pressure drop of the 3.9 ␮m surface is due
is, therefore, necessary to gauge the influence of channel dimen- purely to smaller channel dimensions than the 1.4 ␮m surface.
sions on the pressure drop before the surface roughness effect can
However, the increased roughness of the 6.7 ␮m surface does
be ascertained.
appear to have a noticeable adverse impact on the pressure drop.
The influence of channel dimension can be estimated from two-
phase pressure drop correlations available in literature. However,
there is no general agreement on the most applicable pressure 5 Conclusions
drop correlation to microchannels. Several small-channel correla- Experiments were conducted on three microchannel heat sinks
tions 共such as those by Lee and Lee 关45兴, Mishima and Hibiki with different surface roughness 共Ra = 1.4 ␮m, 3.9 ␮m, and
关30兴, Mudawar and co-worker 关7,8兴, and Lee and Garimella 关9兴兲 6.7 ␮m兲. For the current test surfaces, roughness did not appear
were compared for validation against the experimental results for to have a significant impact on the boiling incipience wall tem-
the Ra = 1.4 ␮m since this surface most closely matches the test perature. Surface roughness was also found to have only a minor

Fig. 9 Influence of surface roughness on channel pressure drop at different mass fluxes

Journal of Thermal Science and Engineering Applications DECEMBER 2009, Vol. 1 / 041007-7

Downloaded From: http://thermalscienceapplication.asmedigitalcollection.asme.org/ on 12/13/2013 Terms of Use: http://asme.org/terms

 M ⫽ molecular weight
ṁ ⫽ mass flow rate
Nc ⫽ number of channels
P ⫽ pressure
q ⫽ wall heat flux
R a, R q, R z, R p ⫽ roughness parameters according to ASME
B46.1-1995
R p,old ⫽ “Glättungstiefe” according to DIN 4762/
1:1960, R p,old ⬇ 0.4Ra 关44兴
Rel ⫽ Reynolds number if liquid flows alone, Rel
= G共1 − x兲Dh / ␮l
Rev ⫽ Reynolds number if vapor flows alone,
Rev = GxDh / ␮v
T ⫽ temperature
Wc ⫽ channel width
X ⫽ Martinelli parameter
x ⫽ thermodynamic equilibrium quality
Fig. 10 Comparison of experimental pressure drop results to Greek
those of Lee and Garimella †9‡
␣ ⫽ aspect ratio
␩ ⫽ fin efficiency
␮ ⫽ dynamic viscosity
influence on the critical heat flux. For wall heat fluxes below
␳ ⫽ density
700 kW/ m2, only small differences in the saturated boiling heat
␴ ⫽ surface tension
transfer coefficient due to surface roughness were measured.
␾ ⫽ two-phase friction multiplier
However, at higher heat fluxes, the EDM surfaces 共3.9 ␮m and
6.7 ␮m兲 have notably higher heat transfer coefficients than the Subscripts
saw-cut surface 共1.4 ␮m兲 at a given heat flux. The two EDM 0 ⫽ value at a thermodynamic quality of x = 0
surfaces have similar values for the heat transfer coefficients at avg ⫽ average
higher heat fluxes. For a fixed heat flux above 1500 kW/ m2, the c ⫽ channel
EDM surfaces have between 20% and 35% higher heat transfer con ⫽ contraction
coefficients than the saw-cut surface. CHF ⫽ critical heat flux
The Bertsch et al. 关5兴 correlation from literature, modified for conv ⫽ convective
the low reduced pressures of the current study using the Gorenflo exp ⫽ expansion
or modified Cooper correlation to predict the nucleate boiling heat FB ⫽ flow boiling
transfer coefficient, provided acceptable flow boiling heat transfer in ⫽ inlet
predictions over the entire range of test conditions considered in l ⫽ liquid
the present study. man ⫽ manifold
From analysis of the pressure drop measurements, it is not be- NB ⫽ nucleate boiling
lieved that the 3.9 ␮m surface causes a measurable increase in out ⫽ outlet
the pressure drop. However, the 6.7 ␮m surface does appear to r ⫽ reduced property
have an adverse effect on the pressure drop compared with the sat ⫽ saturated
1.4 ␮m and 3.9 ␮m surfaces. The results of this experimental sp ⫽ single-phase
study demonstrate that under certain conditions, surface roughness sub ⫽ subcooled
can have a notable influence on heat transfer and pressure drop in tp ⫽ two-phase
microchannel flow boiling and therefore is an important parameter v ⫽ vapor
for consideration when designing microchannel heat sinks. vt ⫽ viscous liquid, turbulent vapor
vv ⫽ viscous liquid, viscous vapor
Acknowledgment w ⫽ wall
Dr. Dong Liu, Dr. Tailian Chen, and Dr. Poh-Seng Lee are
thanked for their helpful and insightful discussions. Financial sup- References
port from the Indiana 21st Century Research and Technology 关1兴 Bertsch, S. S., Groll, E. A., and Garimella, S. V., 2008, “Refrigerant Flow
Fund and the Cooling Technologies Research Center, an NSF IU- Boiling Heat Transfer in Parallel Microchannels as a Function of Local Vapor
CRC at Purdue University, is gratefully acknowledged. Quality,” Int. J. Heat Mass Transfer, 51, pp. 4775–4787.
关2兴 Harirchian, T., and Garimella, S. V., 2008, “Microchannel Size Effects on
Local Flow Boiling Heat Transfer to a Dielectric Fluid,” Int. J. Heat Mass
Nomenclature Transfer, 51, pp. 3724–3735.
关3兴 Bertsch, S. S., Groll, E. A., and Garimella, S. V., 2009, “Effects of Heat Flux,
A ⫽ area
confinement number, Co= 冑␴ / 共g共␳l − ␳v兲D2h兲
Mass Flux, Vapor Quality, and Saturation Temperature on Flow Boiling Heat
Co ⫽ Transfer in Microchannels,” Int. J. Multiphase Flow, 35, pp. 142–154.
cp ⫽ specific heat 关4兴 Harirchian, T., and Garimella, S. V., 2009, “Effects of Channel Dimension,
Heat Flux, and Mass Flux on Flow Boiling Regimes in Microchannels,” Int. J.
Dh ⫽ hydraulic diameter Multiphase Flow, 35, pp. 349–362.
G ⫽ mass flux 关5兴 Bertsch, S. S., Groll, E. A., and Garimella, S. V., 2009, “A Composite Heat
g ⫽ acceleration of gravity Transfer Correlation for Saturated Flow Boiling in Small Channels,” Int. J.
Heat Mass Transfer, 52, pp. 2110–2118.
Hc ⫽ channel height 关6兴 Kandlikar, S. G., 1998, “Heat Transfer Characteristics in Partial Boiling, Fully
h ⫽ heat transfer coefficient Developed Boiling, and Significant Void Flow Regions of Subcooled Flow
h fg ⫽ heat of vaporization Boiling,” ASME J. Heat Transfer, 120, pp. 395–401.
i ⫽ enthalpy 关7兴 Qu, W., and Mudawar, I., 2003, “Measurement and Prediction of Pressure
Drop in Two-Phase Micro-Channel Heat Sinks,” Int. J. Heat Mass Transfer,
K ⫽ pressure loss coefficient 46, pp. 2737–2753.
L ⫽ length 关8兴 Lee, J., and Mudawar, I., 2005, “Two-Phase Flow in High-Heat-Flux Micro-

041007-8 / Vol. 1, DECEMBER 2009 Transactions of the ASME

Downloaded From: http://thermalscienceapplication.asmedigitalcollection.asme.org/ on 12/13/2013 Terms of Use: http://asme.org/terms

 Channel Heat Sink for Refrigeration Cooling Applications: Part I—Pressure Flow Boiling of a Dielectric Coolant in a Microchannel Heat Sink,” ASME J.
Drop Characteristics,” Int. J. Heat Mass Transfer, 48, pp. 928–940. Electron. Packag., 128, pp. 398–404.
关9兴 Lee, P.-S., and Garimella, S. V., 2008, “Saturated Flow Boiling Heat Transfer 关27兴 Collier, J. G., 1981, Convective Boiling and Condensation, 2nd ed., McGraw-
and Pressure Drop in Silicon Microchannel Arrays,” Int. J. Heat Mass Trans- Hill, New York.
fer, 51, pp. 789–806. 关28兴 Blevins, R. D., 1992, Applied Fluid Dynamics Handbook, Van Nostrand Re-
关10兴 Ammerman, C. N., and You, S. M., 1998, “Enhancing Small-Channel Convec- inhold Company, New York.
tive Boiling Performance Using a Microporous Surface Coating,” 1998 Inter- 关29兴 Chisholm, D., and Sutherland, L. A., 1969, “Predictions of Pressure Gradients
national Mechanical Engineering Congress and Exposition, Anaheim, CA, Vol. in Pipeline Systems During Two-Phase Flow,” Symposium on Fluid Mechan-
3, pp. 467–476. ics and Measurements in Two-Phase Flow Systems, University of Leeds.
关11兴 Ammerman, C. N., and You, S. M., 2001, “Enhancing Small-Channel Convec- 关30兴 Mishima, K., and Hibiki, T., 1996, “Some Characteristics of Air-Water Two-
tive Boiling Performance Using a Microporous Surface Coating,” ASME J. Phase Flow in Small Diameter Vertical Tubes,” Int. J. Multiphase Flow, 22,
Heat Transfer, 123, pp. 976–983. pp. 703–712.
关12兴 Kovalev, S. A., and Shklover, E. G., 1988, “Heat Transfer in the Boiling of 关31兴 Lockhart, R. W., and Martinelli, R. C., 1949, “Proposed Correlation of Data
Water on a Porous Surface in an Annular Channel,” High Temp., 26, pp. for Isothermal Two-Phase, Two-Component Flow in Pipes,” Chem. Eng. Prog.,
712–717. 45, pp. 39–48.
关13兴 Savkin, N. N., Kuzma-Kichta, Yu. A., and Komendantov, A. S., 1988, “Inves- 关32兴 Lee, J., and Mudawar, I., 2005, “Two-Phase Flow in High-Heat-Flux Micro-
tigating Enhancement of Heat Transfer With Water Boiling Under Conditions Channel Heat Sink for Refrigeration Cooling Applications: Part II—Heat
of Forced Flow in a Tube With a Porous Coating,” Therm. Eng., 35, pp. Transfer Characteristics,” Int. J. Heat Mass Transfer, 48, pp. 941–955.
295–297. 关33兴 Chen, T., and Garimella, S. V., 2006, “Measurements and High-Speed Visual-
关14兴 Zhang, L., Wang, E. N., Goodson, K. E., and Kenny, T. W., 2005, “Phase izations of Flow Boiling of a Dielectric Fluid in a Silicon Microchannel Heat
Change Phenomena in Silicon Microchannels,” Int. J. Heat Mass Transfer, 48, Sink,” Int. J. Multiphase Flow, 32, pp. 957–971.
pp. 1572–1582. 关34兴 Yan, Y.-Y., and Lin, T.-F., 1998, “Evaporation Heat Transfer and Pressure Drop
关15兴 Kandlikar, S. G., Kuan, W. K., Willistein, D. A., and Borrelli, J., 2006, “Sta- of Refrigerant R-134a in a Small Pipe,” Int. J. Heat Mass Transfer, 41, pp.
bilization of Flow Boiling in Microchannels Using Pressure Drop Elements 4183–4194.
and Fabricated Nucleation Sites,” ASME J. Heat Transfer, 128, pp. 389–396. 关35兴 Lee, H. J., and Lee, S. Y., 2001, “Heat Transfer Correlation for Boiling Flows
关16兴 Koşar, A., Kuo, C.-J., and Peles, Y., 2005, “Boiling Heat Transfer in Rectan- in Small Rectangular Horizontal Channels With Low Aspect Ratios,” Int. J.
gular Microchannels With Reentrant Cavities,” Int. J. Heat Mass Transfer, 48, Multiphase Flow, 27, pp. 2043–2062.
pp. 4867–4886. 关36兴 Bertsch, S. S., Groll, E. A., and Garimella, S. V., 2008, “Review and Com-
关17兴 Koşar, A., Kuo, C.-J., and Peles, Y., 2005, “Reduced Pressure Boiling Heat parative Analysis of Studies on Saturated Flow Boiling in Small Channels,”
Transfer in Rectangular Microchannels With Interconnected Reentrant Cavi- Nanoscale Microscale Thermophys. Eng., 12, pp. 187–227.
ties,” ASME J. Heat Transfer, 127, pp. 1106–1114. 关37兴 Cooper, M. G., 1984, “Saturation Nucleate Pool Boiling—A Simple Correla-
关18兴 Kuo, C.-J., Koşar, A., Peles, Y., Virost, S., Mishra, C., and Jensen, M. K., tion,” First UK National Conference on Heat Transfer, University of Leeds, pp.
2006, “Bubble Dynamics During Boiling in Enhanced Surface Microchan- 785–793.
nels,” J. Microelectromech. Syst., 15, pp. 1514–1527. 关38兴 Cooper, M. G., 1984, “Heat Flow Rates in Saturated Nucleate Pool
关19兴 Kuo, C.-J., and Peles, Y., 2007, “Local Measurement of Flow Boiling in Struc- Boiling—A Wide-Ranging Examination Using Reduced Properties,” Adv.
tured Surface Microchannels,” Int. J. Heat Mass Transfer, 50, pp. 4513–4526. Heat Transfer, 16, pp. 157–239.
关20兴 Kuo, C.-J., and Peles, Y., 2008, “Flow Boiling Instabilities in Microchannels 关39兴 Chen, J. C., 1966, “Correlation for Boiling Heat Transfer to Saturated Fluids in
and Means for Mitigation by Reentrant Cavities,” ASME J. Heat Transfer, Convective Flow,” I&EC Process Des. Dev., 5, pp. 322–329.
130, p. 072402. 关40兴 Hausen, H., 1943, “Darstellung des Wärmeüberganges in Rohren durch ver-
关21兴 Jones, R., Pate, D., and Bhavnani, S., 2006, “Phase Change Thermal Transport allgemeinerte Potenzbeziehungen,” Z. VDI-Beiheft Verfahrenstechnik, 4, pp.
in Etched Silicon Microchannel Heat Sinks,” Proceedings of the 13th Interna- 91–102.
tional Heat Transfer Conference, Sydney, Australia. 关41兴 Jones, B. J., McHale, J. P., and Garimella, S. V., 2009, “The Influence of
关22兴 Jones, R. J., Pate, D. T., and Bhavnani, S. H., 2007, “Control of Instabilities in Surface Roughness on Nucleate Pool Boiling Heat Transfer,” ASME J. Heat
Two-Phase Microchannel Flow Using Artificial Nucleation Sites,” Technical Transfer, 131, p. 121009.
Conference and Exhibition on Integration and Packaging of MEMS, NEMS, 关42兴 Nishikawa, K., Fujita, Y., Ohta, H., and Hidaka, S., 1982, “Effect of the Sur-
and Electronic Systems, Vancouver, Canada. face Roughness on the Nucleate Boiling Heat Transfer Over the Wide Range
关23兴 Pate, D. T., Jones, R. J., and Bhavnani, S. H., 2006, “Cavity-Induced Two- of Pressure,” Proceedings of the Seventh International Heat Transfer Confer-
Phase Heat Transfer in Silicon Microchannels,” Proceedings of the Tenth Inter- ence, Müchen, Germany, Vol. 4, pp. 61–66.
society Conference on Thermal and Thermomechanical Phenomena in Elec- 关43兴 Nishikawa, K., Fujita, Y., Ohta, H., and Hidaka, S., 1982, “Effects of System
tronics Systems, San Diego, CA, pp. 71–78. Pressure and Surface Roughness on Nucleate Boiling Heat Transfer,” Mem.
关24兴 Webb, R. L., 1981, “The Evolution of Enhanced Surface Geometries for Fac. Eng., Kyushu Univ., 42共2兲, pp. 95–111.
Nucleate Boiling,” Heat Transfer Engineering, 2, pp. 46–69. 关44兴 Gorenflo, D., 1993, Pool Boiling, VDI Heat Atlas, VDI Verlag, Düsseldorf,
关25兴 Steinke, M. E., and Kandlikar, S. G., 2004, “Control and Effect of Dissolved Germany.
Air in Water During Flow Boiling in Microchannels,” Int. J. Heat Mass Trans- 关45兴 Lee, H. J., and Lee, S. Y., 2001, “Pressure Drop Correlations for Two-Phase
fer, 47, pp. 1925–1935. Flow Within Horizontal Rectangular Channels With Small Heights,” Int. J.
关26兴 Chen, T., and Garimella, S. V., 2006, “Effects of Dissolved Air on Subcooled Multiphase Flow, 27, pp. 783–796.

Journal of Thermal Science and Engineering Applications DECEMBER 2009, Vol. 1 / 041007-9

Downloaded From: http://thermalscienceapplication.asmedigitalcollection.asme.org/ on 12/13/2013 Terms of Use: http://asme.org/terms