Int. J. Heat Mass Transfer. Vol. 39, No. 12, pp.

2599-2608, 1996
~ ) Pergamon Copyright © 1996 Elsevier Science Ltd
Printed in Great Britain. All fights reserved
0017-9310/96 $15.00+0.00

0017-9310(95)00327-4

Convective heat transfer and flow friction for

water flow in microchannel structures
X. F. PENG
Thermal Engineering Department, Tsinghua University, Beijing 100084, China

and

G. P. PETERSON?
Mechanical Engineering Department, Texas A&M University, College Station, TX 77843, U.S.A.

(Received 15 May 1995)

Abstract--The single-phase forced convective heat transfer and flow characteristics of water in mic-
rochannel structures/plates with small rectangular channels having hydraulic diameters of 0.133q).367 mm
and distinct geometric configurations were investigated experimentally. The results indicated that the
geometric configuration had a significant effect on the single-phase convective heat transfer and flow
characteristics. The laminar heat transfer was found to be dependent upon the aspect ratio and the ratio
of the hydraulic diameter to the center-to-center distance of the microchannels. The turbulent heat transfer
was found to be a further function of a new dimensionless variable, Z, such that Z = 0.5 will be the
optimum configuration for turbulent heat transfer regardless of the groove aspect ratio. The friction factor
or flow resistance reached a minimum value as Z approaches 0.5. The turbulent flow resistance was usually
smaller than that predicted by classical relationships, and the Reynolds number for flow transition to fully
developed turbulent flow became much smaller than the ordinary channel flow. Empirical correlations
were suggestedfor calculatingboth the heat transferand pressuredrop. Copyright© 1996ElsevierScienceLtd.

INTRODUCTION [1, 2], Wu and Little [3, 4] measured the flow friction
and heat transfer characteristics of gases flowing
The importance of convective heat transfer with and through microchannels and observed that the con-
without phase change in microchannels and/or micro- vective heat transfer characteristics departed from the
channel structures has increased dramatically due to typical experimental results for conventionally sized
practical applications involving the thermal control channels. In addition, the frictional pressure drop for
of electronic devices. In the earliest investigation of laminar flow was found to be higher than the classical
microscale flow and heat transfer, Tuckermann and prediction. These measurements indicated a transition
Pease [1] studied the fluid flow and heat transfer from laminar to turbulent flow at Reynolds numbers
characteristics in microchannels, and demonstrated of 400-900 for a number of different test configur-
that electronic chips could be effectively cooled by ations. Pfahler et al. [5, 6] later found that the fric-
means of the forced convective flow of water through tional flow constant, C = f ' R e , was generally lower
microchannels fabricated either directly in the silicon than what was normally obtained from theoretical
wafer or in the circuit board on which the chips were predictions, and that this value increased with increas-
mounted. The dissipated heat flux was of the order of ing Reynolds number. Other investigations by Choi
1.3 x 107 W m -2 while the surface temperatures were et al. [7], Weisberg et al. [8] and Bowers and Mudawar
maintained at a level of less than 130°C. These results [9] all provided additional information and con-
confirmed the potential of this technology. Since this siderable evidence that the behavior of fluid flow and
initial study, other investigations have supported the heat transfer in microchannels or microtubes without
earliest findings and have served to illustrate the phase change is substantially different from that which
unusually high levels of heat removal that can be typically occurs in larger channels and/or tubes.
accomplished using microchannel structures [2]. Sev- In an attempt to clarify some of the questions sur-
eral of these investigations have suggested that the rounding this issue, Peng and Wang [10, 11] and Peng
heat transfer coefficient for laminar flow through mic- et al. [12] investigated microchannels and micro-
rochannels might be higher than for turbulent flow channel structures both with and without phase
through larger more conventionally sized channels. change. In these investigations, the heat transfer and
Shortly after the initial work of Tuckermann et al. flow mode conversions for single-phase forced con-
vection in microchannels, and the transitions induced
f Author to whom correspondence should be addressed. by or associated with variations in the liquid ther-

2599
2600 X.F. PENG and G. P. PETERSON

NOMENCLATURE

Ct- empirical coefficient for flow friction W~ center-to-center distance of

Dh hydraulic diameter microchannel
f friction factor Wt width of rnicrochannel structure/plate
f* dimensionless variable defined by Z dimensionless variable defined by
equation (13) equation (9)
H height of microchannel tz dynamic viscosity
h heat transfer coefficient v viscosity
k conductivity p density
L length of microchannel structure/plate r shear stress.
Nu Nusselt number
Pr Prandtl number
p pressure drop Subscripts
Re Reynolds number ex at exit
q" heat flux f liquid
T temperature H heat transfer
T temperature difference in at inlet
Tm log mean temperature difference 1 laminar flow
U velocity t turbulent flow
W width of microchannel w wall.

mophysical properties due to the increases in the liquid viously [15, 16]. However, for the current inves-
temperature through the heated microchannels, were tigation, twelve different microstructure configur-
studied. Wang and Peng [13] also experimentally stud- ations, each with several rectangular microchannels
ied the forced flow convection of liquid in micro- were designed and evaluated. The geometric and
channels and found that the fully developed turbulent characteristic parameters are summarized in Table 1.
convection was initiated at Reynolds numbers in the The test module shown in Fig. 1 consisted of a stainless
range of 1000-1500, and that the conversion from the steel plate substrate into which the microchannels
laminar to transition region occurred in the range of were machined, and a cover which served as both an
300-800. The heat transfer behavior in the laminar insulator and a sealant. On the microchannel plate or
and transition regions was found to be quite unusual test section, two sumps were machined and connected
and complicated. Peng and Peterson [14] later con- by microchannels. A pressure tap was located at both
firmed these experimental observations using meth- the inlet and outlet sumps in order to measure the inlet
anol flowing through similar microchannel structures and outlet pressures and the pressure drop through the
and analyzed experimentally the effects of the ther- microchannels. Thermocouples were installed in the
mofluid and the geometric variables on the heat trans- sumps at the entrance and exit to measure the inlet
fer. Most recently, Peng et al. [15, 16] measured both and outlet liquid temperature. In addition, six thermo-
the flow friction and the heat transfer for single phase couples were mounted on the back of each micro-
convection in an array of parallel microchannets. channel plate to measure the axial wall temperature
These measurements validated previous observations distribution and ensure that the flow rate was equally
and conclusions and were used to develop a set of distributed in the channels. These thermocouples were
heat transfer and friction factor correlations for both distributed longitudinally at three different locations.
laminar and turbulent flow. During the experiment, the test module was wrapped
The present work was undertaken to investigate in an insulating shroud to minimize the heat loss to
the liquid flow and heat transfer characteristics in the surroundings through convection and radiation.
microchannel structures in an attempt to exper- Water was selected as the working fluid. To account
imentally determine the effects of the geometric con- for fluid property variations, the experiments were
figuration on the flow and heat transfer and to further conducted over an inlet liquid temperature range of
develop heat transfer and flow friction correlations, 20-45°C. The liquid velocity was relatively high, rang-
which would be readily applicable to engineering ing from 0.20 to 12 m s-~. Accordingly, the Reynolds
design and practical applications. number spanned a range of pure laminar to highly
turbulent, 50-4000. The flow rate of water entering
the test module was measured using a rotameter,
EXPERIMENTAL DESCRIPTION
which had been calibrated by the weight method. The
The experimental test apparatus employed in the pressure drop of the liquid flowing through the micro-
current investigation was similar to that described pre- channel was measured to determine the flow friction.
 Convective heat transfer and flow friction 2601

Table 1. Geometric parameters of the test sections

W Wc Wt L H Dh
Plate [mm] [mm] [mm] [mm] [mm] [mm] Dh/Wc HI W Z Cf,I Cf,t

1 0.4 4.5 18 45 0.2 0.267 0.0593 0.5 0.5 28 600 40 400

2 0.4 2.8 18 45 0.3 0.343 0.1225 0.75 0.75 44 800 34 200
3 0.4 2.0 18 45 0.3 0.343 0.1715 0.75 0.75 44 800 34 200
4 0.3 4.6 18 45 0.2 0.24 0.0533 0.667 0.667 42 600 18 200
5 0.3 2.8 18 45 0.3 0.30 0.107 1.00 1.00 109 000 38 600
6 0.3 2.0 18 45 0.3 0.30 0.15 1.00 1.00 109 000 38 600
7 0.2 4.5 18 45 0.2 0.20 0.0444 1.00 1.00 32 400 20 100
8 0.2 2.8 18 45 0.3 0.24 0.0857 1.50 0.667 42 600 18 200
9 0.2 2.0 18 45 0.3 0.24 0.12 1.50 0.667 42 600 18 200
10 0.1 4.5 18 45 0.2 0.133 0.0296 2.0 0.50 5200 1820
11 0.1 2.8 18 45 0.3 0.15 0.0536 3.0 0.333 24 200 6920
12 0.1 2.0 18 45 0.3 0.15 0.075 3.0 0.333 24 200 6920

The microchannel stainless steel plate was elec- state conditions. F o r each different set of conditions,
trically heated by directly connecting the test sections the experimental data, including flow rate, liquid tem-
to a d.c. transformer that provided low voltage and peratures, inlet and outlet pressures (or pressure drop
high electric current. In this way, heat generated in between the entrance and exit), the test section wall
the thin section of the plate was transferred directly surface temperatures, and the input voltage and
to the liquid from the two sides and the bottom of the current, were measured and recorded. The measure-
microchanne!, as illustrated in Fig. 1(c). Because the ment uncertainties were analyzed and are given in
center section of the test plate was uniform in thick- Table 2.
ness and width, i.e. cross-sectional area, the heat flux The parameters used in the data reduction and
was considered to be uniform along both the longi- analysis are summarized below :
tudinal length and around the wetted periphery,
the hydraulic diameter, Dh
except for the top surface, which was insulated. The
input voltage and current were used to measure and 4(cross-section area) 2 WH
control the heat flux. These parameters were varied D h -- = (1)
wetted perimeter W+ H
from 0.15 to 0.92 V and 50 to 400 A, respectively.
The experiments were carried out under steady- the Reynolds number, Re

SUMP / SUMP

r'/W////ll DCCABLE

i \
6THERMOCOUPLES STAINLESS
S1EELPLATE W
(2ATEACHLOCATION)
{a] L C ) N G ~ D I N A L V I E W

ESSURp~PORT /MICROCI-IANNEL / PRESSURE

PORT
(c)MICROCHANNEL
CRC6S-SECTION
SUM / / SUMP

(b)TOPVIEW
\\
STAINLESS
S'IEELPLATE
Fig. 1. Experimental section.
2602 X.F. PENG and G. P. PETERSON

Table 2. Experimental uncertainties where Tm is the log-mean temperature difference,

determined from
Variables Uncertainty [%]
aT, n -ATex
Flow rate 1.5 ATm - ln(AT~,/AT~x) . (7)
Velocity 3.5
Voltage 0.5
Current 0.5 For all cases, the relative size of the heat losses was
Heat flux 8 small.
Length scale 12 The final results are presented as a function of the
Pressure drop 1.5 Nusselt number, Reynolds number and friction factor.
Liquid temperature 1.5 Using conventional techniques, the uncertainties for
Wall temperature I
Heat transfer coefficient 10 all of the experimental values measured were esti-
Flow friction factor 10 mated and are given in Table 2.
Nusselt number 16
Reynolds number 8
HEAT TRANSFER PERFORMANCE

Peng et al. [15] demonstrated that the critical Reyn-

olds number for laminar transition in microchannel
UD h
Re - (2) structures occurs at approximately 200-700 and for
fully developed turbulent flow at Reynolds numbers
where U represents the mean channel velocity, the of 400-1500. In addition, the transition Reynolds
flow friction factor, [; numbers have been shown to decrease as the hydraulic
diameter decreased. The experiments conducted in the
8rw 2Dh Ap current investigation demonstrated that these con-
f- - (3)
pU 2 pU2L clusions are true only for liquid flow in microchannel
structures with identical microchannel dimensions.
and the Nusselt number, Nu,
The current data indicates that the geometric par-
hDh ameters such as hydraulic diameter, ratio of the
Nu - (4) microchannel height and width, H / W , and ratio of
k
hydraulic diameter and microchannel center-to-center
In the calculation of the friction factor,.L and the distance, Dh/Wc, all have a significant influence on
Reynolds number, Re, the thermophysical properties both the flow regime and the heat transfer.
of water were assumed to be independent of liquid
pressure. Because of variations in the temperature of Laminar Jtow
the liquid, the friction factor and the Reynolds num- The laminar convective heat transfer data for water
ber varied along the length of the plate. For simplicity, flowing through the twelve different microchannel
the inlet properties and conditions of the liquid were structures was shown to be correlated by the following
used to calculate these values. Since there existed a relationship
sudden flow contraction and expansion at the entrance
and exit of the microchannels, the actual measured
pressure drop included these contraction and expan-
sion losses. The pressure drop, Ap, in equation (3),
however, represents the pressure drop only along the In Fig. 2, this correlation is compared with the exper-
microchannels, hence, the calculated pressure drop imental results. Although the Reynolds number range
caused by the contraction and expansion were sub- for laminar convective heat transfer is not identical,
tracted from the measured values. These were 5 7% due to the difference in the transition Reynolds num-
of the total. ber for different microchannel structures or plates,
In the determination of the total heat generation this comparison indicates reasonably good agreement
or heat flux, the heat losses induced by conduction, between the experimental data obtained from the
convection, and radiation were modified using an twelve microchannel plates and the correlation given
energy balance. The heat flux of the microchannel by equation (8). The range of deviation for this cor-
structure was defined as relation is approximately _4-30%.
Figures 3 and 4 depict the average experimental
Q
q" = --- (5) results for the individual test plates as a function of
A O h / W c and H / W , respectively, and illustrate the effects

where Q is total heat input, and A denotes the micro- of the geometric parameters on the laminar convective
structure plate area (i.e. A = WtL). The heat transfer heat transfer. Careful thought was given to the selec-
coefficient was evaluated as, tion of these dimensionless parameters in order to
explore and generalize the effect of the various channel
q" dimensions. Figure 3 indicates that the laminar heat
h(x) - (6)
AT,, transfer will be augmented by enlarging the hydraulic
 Convective heat transfer and flow friction 2603

0
0
0 V 0 000-~ ' ~
[k

0+ O ~ v •
t 00 I1~ ~ • Plate t
& ~ + • 4- 0 Plate 2
(5 A A + • Plate3
Z~ A [] Plate 4
• Plate 5
,b • Plate 6
c~ A Plate 7
V Plate 8
Plate 9
Plate 10
4- Plate 11
Z @ Plate 12
Eq.(8)

10o i . I . , ,I I i I , I , i , l , , I , , . l , , . l . l , l . l . t

8 9102 2 3 4 5 6 7 8 9103
Re
Fig. 2. Experimental results of laminar heat transfer.

diameter or decreasing the center-to-center distance. more heavily dependent on this ratio than the laminar
Figure 4 indicates that increasing the width or decreas- case. The significance of H and W is nearly the same,
ing the height of the microchannel enhances the lami- however, the effect of these two parameters on the
nar heat transfer. turbulent heat transfer cannot be simply described by
the dimensionless parameter, H/IV. For this reason, a
Turbulentflow new dimensionless variable, Z, defined as
For turbulent convection, the ratio Dh/Wc, was
found to be a very important parameter, particularly min(H, IV)
Z = (9)
when compared to the case of laminar convective heat max(H, I40
transfer. Figure 5 illustrates the change of the tur-
bulent heat transfer for each test plate with the dimen- has been introduced, which helps to define the effect
sionless value Dh/Wc. This relationship is similar to of the aspect ratio. The significance of this relationship
that shown in Fig. 3, however, a more detailed analysis was determined empirically.
indicates that the turbulent convection is somewhat The measured results of the turbulent heat transfer

0.035

0.030

O_
d.
0.025
(1)
n.."

0.020

I
0.015

Z l
0.01 0

I i I i I i t i I i L I L I
0.005
0.01 0.03 0.05 0.07 0.09 0.11 0.13 0.t5 0.17 0.19
D,/Wc
Fig. 3. Effect of Dh/Weon laminar heat transfer.
2604 X . F . PENG and G. P. PETERSON

0.22

0.20

0.18
El_
0.16 • •
d
©
0:::: 0.14

0.12

0.10
0
,_,._,

0.08
Z
0.06

0.04
I t I i i I I i I

0.1 0.6 1.1 1 6 2.1 2.6 ,3.1

H/W
Fig. 4. Effect of H/W on laminar heat transfer.

for different test plates are depicted as a function of Figure 7 compares the correlation given in equation
Z, in Fig. 6. This figure indicates that there exists an (10) with the turbulent heat transfer data obtained for
optimum value of Z which maximizes the turbulent the twelve test plates. As shown, the experimental data
convective heat transfer, and that this value, approxi- for all twelve test plates fall along this line with a
mately Z = 0.5, is constant, regardless of the ratio of maximum deviation of _+25%.
H/W or W/H. While providing a means for predicting the heat
For fully developed turbulent flow in microchannel transfer characteristics in microchannel structures,
structures, the forced convective heat transfer cor- these measurements indicate that the transition heat
relation can be expressed as transfer for a liquid flowing through microchannel
plates is highly complicated. As noted above, the geo-
metric configuration is of significant importance for
Nu = O.O72(D-~)' 's [1- 2.421(Z-O.5)2]Re°S Pr '/3. the single phase forced-convective heat transfer, and
has a somewhat different effect for laminar and tur-
(lO) bulent convection.

0.1,3 i i i i i i i i

0.12

0_
0.11
co
,5
(1) 0.10

,m,, 0.09
LQ
O
0.08

N
0.07
0.06

c4 0.05
L
O.04
0.03 • 11

Z 0.02
I
0.01 [ i i i I i I ~ I l I i I i

0.01 0.03 0.05 0.07 0.09 0.11 0.13 0.15 0.17 0.19
D,,/Wo
Fig. 5. Effect of Dh/W~on turbulent heat transfer.
 Convective heat transfer and flow friction 2605

0.08 I i i i

0.07

(3-
~o

r,f
'~ 0.06

~" 0.05

0.04
z

I
0.03
I I , I i I , I i I i I , I
0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Z
Fig. 6. Eff~t of Z on turbulent heat trans~r.

FLOW RESISTANCE for fully developed turbulent flow, where Cr.l and Cf,,
are empirical coefficients for laminar and turbulent
As a result of the experimental flow measurements
flow, respectively. These values are also dependent
of Peng e t al. [15] empirical correlations for the cal-
upon the microchannel configuration and geometric
culation of the friction factor for single-phase liquid
parameters.
flow in microchannels have been proposed. These
The experimental investigation of microchannel
relations are given as :
plates or structures indicates that the analyses of Peng
q,, et al. [15] are applicable to the current case, and that
f= (ll) the flow resistance has the same form given by equa-
R e 1.98
tions (11) and (12) for laminar and turbulent flow,
for laminar flow and respectively. Comparison of experimental data also
indicates that the hydraulic diameter, Dh, and the
f= Cr,t (12) dimensionless variable, Z, are critical parameters in
R e 1.72

' , ' , ' ' , , * , . , . , ' , ' , ' , ' , ' 1 ' ' • , ' ' ' ' " ' ' ' '

e..
t3_

c~ ~/f~V DD
I
N

I A
A" A+/--'-~
" ~ r- ' - " ~D ~P_!t
P!o~e 3
i~.~ • Plate5
~. v4, • • _P!ate_6
7-. / '• ~ Plate 8
j v ~ Plate 1 0
d 101 4- Plate 1 1
v
9 6` Plate 1 2
6, Eq.(1 O)
8
, i , I i i i I , , • I • , , I . , . I , , , I I i . i . , , i
7
5 6 7 8 9103 2 3 4

Re
Fig. 7. Experimental results of turbulent heat transfer.
2606 X. F. P E N G and G. P. PETERSON

,, .,.,.i.,. I , i , , , , . , , E . , , r . , , , . , ,...,,,

"A. • EXPERIMENTAL DATA

THEORETICAL RESULT
...... Eq. (1'1) C,L=32400
• - Eq. (12), Cft:201 O0

10 0 !

6
5
4
%

PLATE 7
D h - - 0 . 2 0 r~w" "b
Z=I
10 1 i,i,l,l,lll i

8 9102 2 5 4 5 6 7 8

Re
Fig. 8. Friction factor of plate 7 [15]•

the determination of the empirical coefficients in eq ua-

tions (11) and (12). Figures 8 and 9 illustrate some of \.fi,,heo/R~, "
experimental results, and the values for C,u and C,:,
for the plates tested are listed in Table I. This value, f * , is depicted as a function of Z in Fig.
Comparing Figs. 8 and 9, it is clear that by selecting 10. Clearly.f* has a minimum value, which might
the appropriate geometric configuration, the friction be used as an optimum parameter in the design of
factor for the laminar conditions can be reduced and microchannel plates or structures and other similar
made smaller than the value predicted by the classical devices. For turbulent flow, the measured flow resist-
relation• A new parameter, f * , can be introduced to ance in microchannels is lower than the theoretical
relate the experimental and theoretical friction factor results predicted by conventional relationships, and
for the laminar conversion Reynolds number, Re~,, the Reynolds number becomes smaller than the Reyn-

10 o '','"~'"1 • • , , , , , t , , . , , , , , . , , , ,

PLATE 10
6 Dh=0135 m m
5 H/W = 0.5
4
3

10-1

6
5
4
5 • EXPERIMENTAL DATA
THEORETICAL RESULT
..... Eq. (11 ), C . = 5 2 0 0
2 -- EQ (12), C~',t=1820

itll ,.I.,.I i i i , i , i I , i I L , I ¸ , .'.,

10-2
8 9102 2 5 4 5 6 7

IRe
Fig. 9. Friction factor of plate 10 [15]•
 Convective heat transfer and flow friction 2607

4 i i i I i

,L

0 I i I i I i I i I l I i I J I i I

0.2 0.,3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Z
Fig. 10. Variation off* with Z [15].

olds number at which the flow transitions to fully plate with the same size and number of channels, the
developed turbulent flow occur. However, it is not convection heat transfer from the two sides of the
clear whether Wc has any influence on the flow. How channel in the center section of the plate decreases
and how much these geometric factors affect the flow as the center-to-center distance is decreased. These
resistance or friction factor requires additional inves- conclusions are of fundamental importance in the
tigation. practical applications and design of these types of
structures.
The flow resistance of the liquid flow in the micro-
CONCLUSIONS
structures was also investigated experimentally and
The forced convective heat transfer and flow analytically, and correlations were proposed for the
characteristics of water flowing through microchannel calculation of the flow resistance. The experiments
plates with extremely small rectangular channels hav- demonstrated the importance of the geometric par-
ing hydraulic diameters of 0.133~.367 mm and ameters, including the hydraulic diameter, H! W or Z,
different geometric configurations were investigated and Dh/We, on the friction factor. The laminar friction
experimentally. Measurements indicated that the geo- factor or flow resistance reaches a minimum value as
metric configuration of the microchannel plate and Z approaches 0.5. The turbulent flow resistance is
individual microchannels has a critical effect on the usually smaller than the value predicted by the classi-
single phase convective heat transfer, and that the cal relationships and the Reynolds number becomes
effect on the laminar and turbulent convection is quite smaller than the Reynolds number at which the flow
different. While the thermal conductivity of the transitions to fully developed turbulent flow, as the
material from which the plates were fabricated could flow resistance decreases, i.e., the smaller the resist-
be a factor, the microchannels are so small that the ance, the larger the flow transition Reynolds number.
hydraulic radius is comparable to the sublayer thick-
ness and therefore, the resistance in the sublayer for Acknowledyements--This material is based on part upon
these cases becomes much more important than for work supported by the Texas Advanced Technology Pro-
gram (Grant No. 71440) and the Center for Energy and
larger more conventional channels. As a result, for Mineral Resources.
channels as small as those evaluated here, the shape
of the channels plays a negligible role for both the
REFERENCES
laminar and turbulent flow conditions. The laminar
heat transfer, however, does depend on the parameters 1. D. B. Tuckermann and R. F. Pease, Optimized con-
vective cooling using micromachined structure, J. Elec-
D h / W c and H / W , while the turbulent convection is trochem. Soc. 129(3), C98 (1982).
related to Dh/Wc and Z rather than H/IV. It was found 2. D. B. Tuckermann and R. F. Pease, High-performance
that the dimensionless ratio, Z = 0.5, is the optimum heat sinking for VLSI, IEEE Electronic Device Lett.
configuration for turbulent heat transfer regardless of EDL-2, 126-129 (1991).
3. P. Y. Wu and W. A. Little, Measurement of friction
H/W. In addition, empirical correlations were sug- factor for the flow of gases in very fine channels used for
gested for predicting the heat transfer for both the microminiature Joule-Thompson refrigerators, Cryo-
laminar and turbulent case. Physically, for a given #enics 23(5), 273-277 (1983).
2608 X . F . PENG and G. P. PETERSON

4. P. Y. Wu and W. A. Little, Measurement of the heat microchannel structures, J. Enhanced Heat Tran4Ter (in
transfer characteristics of gas flow in fine channels heat press).
exchangers used for microminiature refrigerators, C~ivo- I 1. X. F. Peng and B. X. Wang, Experimental investigation
genics 24(8), 415-420 (1984). on flow boiling of liquid through microchannels, Chin.
5. J. Pfahler. J. Harley, H. H. Bau and J. Zemel, Liquid J. Engn 9 Thermophys. 14(3), 281 286 (1993).
transport in micron and submicron channels, J. Sensors 12. X. F. Peng and B. X. Wang, G. P. Peterson and H. B.
Actors 21,431-434 (1990). Ma, Experimental investigation of heat transfer in flat
6. J. Pfahler, J. Harley, H. H. Bau and J. Zemel, Gas plates with rectangular microchannels, Int. J. Heat Mass
and liquid flow in small channels. In Micromechanical Tran~'fer38(1), 127 137 (1995).
Sensors, Actuators and Systems (Edited by D. Cho et 13. B. W. Wang and X. F. Peng, Experimental investigation
al.), ASME DSC-Vol. 32, pp. 49 60 (1991). on forced-flow convection of liquid flow through
7. S. B. Choi, R. F. Barron and R. O. Warrington, Liquid microchannels, Int. J. Heat Mass Transfer 37(Suppl. 1).
flow and heat transfer in microtubes, In Micromechanical 73 82 (1994).
Sensors, Actuators and ,~vstems (Edited by D. Cho et 14. X. F. Peng and G. P. Peterson, The effect of thermofluid
al.), ASME DSC-Vol. 32, pp. 123 134 (1991). and geometric parameters on convection of liquid
8. A. Weisberg, H. H. Bau and J. Zemel, Analysis of through rectangular microchannels, Int. J. Heat Mass
microchannels for integrated cooling, Int. J. Heat Mass Tran~'fer 38(4), 755 758 (1995).
Transfer 35(10), 2465 2474 (1992). 15. X. F. Peng and G. P. Peterson, Frictional flow charac-
9. M. B. Bowers and I. Mudawar, High flux boiling in low teristics of water flowing through rectangular micro-
flow rate, low pressure drop mini-channel and micro- channels, J. Exp. Heat Tran.~Jer 7(4), 249 264 (1995).
channel heat sinks, Int. J. Heat Mass Tran,~[er 37(2), 16. X. F. Peng, G. P. Peterson and B. X. Wang, Heat transfer
321 332 (1994). characteristics of water flowing through microchannels,
10. X. F. Peng and B. X. Wang, Cooling characteristics with J. Exp. Heat Tran,~Jer 7(4), 265 283 (1995).