FORCED CONVECTION AND FLOW BOILING OF A DIELECTRIC LIQUID IN A FOAM-

FILLED CHANNEL
Dae Whan Kim1, Avram Bar-Cohen*, Bongtae Han
Department of Mechanical Engineering
University of Maryland at College Park
* Email: [email protected]
1
Currently in Samsung Electronics DM R&D Center

ABSTRACT
Porous metal foams, inserted into the channels of a liquid INTRODUCTION
cold plate, can be used to enhance forced convection and flow Porous foams are an attractive way to enhance convective
boiling heat transfer and may be especially useful for direct, heat transfer in tubes and channels due to the foam’s high ratio
dielectric liquid cooling of electronic and photonic of surface area to volume. Moreover, the dispersion
components. This study explores the thermofluid conductivity which is caused by the recirculation, mixing, and
characteristics of three porous copper foam configurations: “disturbance” of the flow by the porous matrix further
95% porosity and 10 PPI, 95% porosity and 20 PPI, and 92% enhances the heat transfer rate. The consequent relative
porosity and 20PPI, which were soldered to the heated wall of insensitivity of channel heat transfer to the thermal
a 10 mm wide x 37 mm long and 7 mm high channel. The conductivity of the fluid makes porous foams an attractive
results for water are in good agreement with the available candidate for heat exchange with low conductivity fluids such
sparse porous matrix correlations, using the recommended as the dielectric fluids.
dispersion conductivity coefficient of 0.06. Despite this Flow through porous media has been studied in detail
relatively low dispersion coefficient, the porous foam is found beginning with Darcy’s work which was published in 1856
to more strongly enhance the convective heat transfer [1]. With the emergence of the open cell metal foam, it has
coefficients for FC-72 than water. For the two-phase heat been recognized that porous media could be used in creating
transfer rate, the high porosity and large pore size foam, i.e. compact heat exchangers due to their superior characteristics
the 95%, 10 PPI copper foam, was found to provide the best of:
result, achieving a 10 kW/m2-K heat transfer coefficient. • Light weight
• Low mechanical modulus
KEY WORDS: foam, forced convection, flow boiling, • Low cost
dielectric liquid • High contact area/volume with the fluid (much higher
than micro-channel)
NOMENCLATURE • Thermal conductivity that is only weakly dependent on the
Cp specific heat (J/Kg-K) conductivity of the fluid, and
• Lower thermal resistance than plate fin microchannels [1]
f inertial coefficient Therefore, researchers started to study the heat transfer
G mass flux (Kg/m2-s) characteristic of porous media as early as 1974 when Koh et
K permeability (m2) al. [2], investigated the heat transfer characteristic of porous
media. Lage et al. [3] introduced a porous layer in a cold plate
kd dispersion conductivity (W/m-K)
used in a radar system, Boomsma and Poulikakos [4] and
ke effective thermal conductivity (W/m-K) Boomsma et al. [1] studied compact heat exchangers with
metal foam. Tadrist et al. [6] also studied the use of porous
kf fluid thermal conductivity (W/m-K) structures in heat exchangers.
ks solid thermal conductivity (W/m-K) The ability of a porous medium to enhance the effective
thermal conductivity of a fluid flowing through the medium is
h heat transfer coefficient (W/m2-K) a vital consideration in heat exchanger design, and the
Re Reynolds number, Re UVDh / P combined effect of metal foam with a low conductivity fluid
u fluid velocity (m/s) provides far better thermal performance than that of an empty
channel. However, it is difficult to accurately model the
uD Darcian velocity (m/s) thermofluid characteristics of a medium flowing through an
H porosity intricate network of pores and fibers [7]. Therefore, the
J dispersion coefficient characterization of metallic foams, especially with the more
P viscosity (Pa·s) complex geometries, is an ongoing activity [8], [9]. Although
U density (Kg/m3) the metal foam is an excellent material for heat transfer to a
low conductivity fluid, such as dielectric liquids used in
electronic and photonic cooling applications, much of the
available data deals with water and air and it is nearly

86
978-1-4244-1701-8/08/$25.00 ©2008 IEEE.
impossible to obtain useful information related to the behavior and fluid phases, but is relatively insensitive to the thermal
of porous foams with other fluids [10]. Moreover, two-phase conductivity of the fluid.
experimental data, as opposed to single-phase data, is Fluid Mechanics. Transport in porous media involves
relatively rare. complex flow patterns around solid particles or fibers and due
In this paper we experimentally studied the single-phase to the random orientation of the solid phase, exact solutions to
and two-phase heat transfer characteristic of three different the detailed local flow field are nearly impossible. Tenable
metal foams which had different pore size and porosity. Water solutions are found by volume averaging the governing
and a dielectric liquid, FC-72, respectively, were used as the equations. The steady, volume-averaged continuity and
working fluids. Water was used for the comparison of momentum equations for forced convection in a porous
existing correlations of pressure drop and heat transfer for the medium are
validation of the experimental apparatus, and FC-72 was used
for generating heat transfer data for single-phase and two- ’u 0 (3)
phase cooling of electronic and photonic devices. U P f P 2 (4)
u ’u  u  ’ p  Ug  U uu ’ u
Fig. 1 is a schematic diagram of a design which uses H2 K K H
porous foam as a microchannel cold plate to create a high For 2-dimesional flow, Eq. (4) can be reduced to
performance compact heat exchanger at the submount level in
an LED package. In this design the LEDs array is soldered dp P f 2 P d 2u (5)
0   uU u 
to the submount and the porous foam is soldered to the dx K K H dy 2
substrate. As shown in Fig. 1, the foam material “sandwich” u ,
Using the non dimensional terms, u K dP , U
is soldered to the heat source. Wire bonds electrically 
P dx
D
uD
connect the metal core PCB and Silicon submount, which
serves to provide electrical connections and thermal spreading Uu D K , §H ·
1/ 2
and X H ,
Re k Y y¨ ¸ x
from the individual LED’s. P ©K¹ K Re k Pr
Eq. (5) can be expressed as
d 2U (6)
0 1  U  f Re k U 2 
dY 2
Since the velocity boundary layer growth in porous media
takes place over a very short distance, in most practical cases
the hydrodynamic entrance length is negligible [12]. Since
the entering flow can thus be considered fully developed, the
“y” component of equation (6) is identically zero. Then the
average velocity becomes

 1  (1  4 f Re k )1 / 2 (7)
Ua
2 f Re k

Fig. 1 Schematic diagram of porous foam cold plate The momentum equation in the x-direction can be written as
dp P f 2 (8)
 uU u
dx K K
THEORETICAL BACKGROUND
where K is permeability and f is the inertial coefficient, or
Heat Transfer. The effective thermal conductivity of porous the so-called Ergun coefficient.
foam, with heat conducting through the solid and fluid phases Eq. (8) is called the extended Darcy’s equation and is
in parallel, can be written as: widely accepted for steady-state, unidirectional pressure drop
in a homogeneous, uniform and isotropic porous medium,
k e Hk f  (1  H )k s (1)
traversed by a fully saturated incompressible fluid. Both the
Based on the Eq. (1), the empirical correlation of effective permeability, K , and the inertial coefficient, f , are
thermal conductivity of foam material can be expressed by [1]
strongly related to the structure of the medium. In Darcy’s
ke Hk f  A(1  H ) n k s (2) regime (at low velocity), the permeability term is dominant
where, k f is the thermal conductivity of the fluid, k s is but in the non-Darcy regime, viscous shear stresses become
insignificant compared to the form drag.
the thermal conductivity of the solid, H is porosity, n is Several researchers have developed analytical equations
set equal to 0.763 and A is 0.195 for water and 0.181 for air for the prediction of pressure drop in porous foam material.
[10]. In order to develop an approximate pressure drop prediction
As shown in Eq. (2), the effective conductivity of porous for the metal foam structure, Calmidi and Mahajan [11] and
metal foam involves the thermal conductivity of both the solid Du Plessis [13] assume that it has an open cell structure (Fig.
2) composed of dodecahedron-like cells, having 12 ~ 14

87
pentagonal or hexagonal faces, and that the face area of the wT w ª wT º (15)
U 1  J Re k Pre U
pore structure is the same as the pentagonal face. wX wY «¬ wY »¼
With the boundary condition appropriate to a slug flow
approximation, T 0 at X 0 and Y f , T 1 at Y 0 ,
and integrating the non-dimensional energy equation Eq. (15),
yields
­°Y ª 1 / 2
X º ½° (16)
T 1  erf ® « 1  J Re k Pre » ¾
°̄ 2 ¬ Ua ¼ °
¿

Fig. 2. Photograph of open cell Assuming that the velocity at the wall is zero, i.e. V 0,
Y 0

and excluding mass flow through the plate surface, the local
They compared their analytical equation to the existing Nusselt number can be calculated from differentiation of Eq.
experimental results and developed equations for the (16) and can be expressed by
permeability and inertial coefficient, as: § wT · (17)
Nu x ¨ ¸
H 2d 2 (9) © wY ¹Y
K 0
36 F ( F  1)
Consequently, the length averaged Nusselt number can be
calculated from the integration of Eq. (17) and expressed by
f 1.8 /(180H 5 )1 / 2 (10) 1/ 2
1 L ª1 º (18)
L ³0
Nu Nu x dx 2 « Re k Pre Da L (1  J Re k Pre )»
¬ S ¼
Heat transfer. The volume averaged homogeneous energy
equation for porous media is given by where Da L is the inverse Darcy number, Da L L / K , the
~
UC p u ˜ ’T ’ ˜ k e ’T  UC p ’ ˜ u~T (11) effective Prandtl number is defined as Pre PCp f / k e , the
Reynolds Number is as before, Re k Uu a K / P .
where k e is the effective conductivity [11].
In this study, we measured pressure drop and heat transfer
In Eq. (11) the convective term is divided into average capability and compared the results to these best-available
and spatially varying components. The last term is due to correlations.
thermal dispersion resulting from flow disturbance, mixing,
and recirculation of local fluid streams as the fluid passes
EXPERIMENTAL STUDY
through the porous media [14]. This can be viewed as a Experimental apparatus. The goal of the single phase
diffusive process related to the overall temperature gradient experiments was to measure the hydraulic and thermal
and the enhanced “dispersion” conductivity, representing the performance of the porous foam cooler and compare the
combined thermal transport effects of conduction and results with existing theory and correlations to establish the
microconvection. Therefore the spatially varying term in Eq. accuracy and reproducibility of the experimental data obtained
(11) can be expressed as with this apparatus, validate the use of a computational fluid
~
UC p ’ ˜ u~T k d ’T (12) dynamics software product, and set the stage for later two-
with the dispersion conducivity k d given by phase experiments. As described in [16] , the test “clamshell”
channel is made of a copper base plate and a Plexiglas top
kd UC p J K u (13)
plate, which has 0.2 W/m-K of thermal conductivity. It was
where J is dispersion coefficient and K is the above placed within a Styrofoam box with a wall thickness of 20mm.
The coolant was pumped by a commercial pump (200
defined permeability. GB, Micropump) which has a maximum capability of 3200
For a two-dimensional flow in uniform porosity media, the ml/min at 5000 rpm. The flow rate was controlled by a
energy equation is express by regulating valve (Pro-spense, Cole-Parmer), which can control
wT w ª§ k d · wT º (14) the flow rate with and accuracy of ± 0.5%. The flow rate was
u D «¨1  ¸¸ »
wx wy ¬¨© ke ¹ wy ¼ measured by two different flow meters: up to 200 ml/min by a
liquid flow sensor (FPR1502, Omega), which provides ± 3%
Or, substituting the previously defined non dimensional full scale (FS) accuracy and 0.2% FS repeatability and for
velocity and Reynolds Number, i.e. with the following non- flow rates greater 200 ml/min by a rotameter (FL-3840ST,
dimensional term, Omega), which offers ± 2% FS accuracy and 0.25% FS
U
u , Uu D K , H , Y y§¨ H ·¸
1/ 2
repeatability.
Re k X x
uD P K Re D Pr ©K¹ Two pressure ports (1.6 mm diameter) were bored into the
top plate to measure the pressure drop across the channel.
and T T  Ti
The static pressure drop of the fluid was measured by two
Tw T i
different differential pressure transducers manufactured by
Eq. (14) can be expressed as Valydine; Model P55D-32 for the low pressure ranging from 0

88
to 3.2 psi and Model P55D-40 for higher pressure drops of up
to 12 psi. Both models offer an accuracy of ± 0.25% FS. The Foam specimen. In the present study of the thermal
output of the pressure transducer is a DC signal (± 5V in FS) performance of an open cell, metal foam filled channel, the 3
and was documented by a data logger (34970A, Agilent). different copper foams shown in Fig. 4 were tested. These 3
The temperature of the coolant was controlled by a chiller samples were selected to provide a contrast in pore size, using
(CH/P 2067, Forma Scientific). The chiller could remove up samples with 10 and 20 pores-per-inch (PPI) nickel-plated
to 600W from the coolant, and thus regulate the inlet copper, respectively, at a porosity of 95% and a contrast in
temperature to within 0.3qC in the range of 50qC. porosity, by adding a copper foam sample with 92% porosity
For wall temperature measurement, 10 holes (1mm in and 20 PPI. The channel was made to accommodate a 10 mm
diameter) were drilled in the plane of the heater plate and wide by 37 mm long foam layer. The foam thickness and,
placed just 1mm below the wetted surface, from both sides (5 consequently, channel height of 7 mm, was selected to assure
holes in each side), E-type thermocouples (diameter of 75 µm) that at least two full pores would be present in the foam layer.
were placed in the holes. Four additional thermocouples were The foam specimens were carefully soldered to the copper
inserted through the thermocouple ports in the top plate (Fig. base block. As a first step, the copper base block was heated
4 (b)) to measure the temperature of the coolant at both the by placement on a hot plate and then a layer of solder
inlet and the outlet. The inlet temperature of the coolant was distributed across its surface. When the solder melted, the
maintained at 25 ± 0.2qC. foam specimen was placed on the melted solder and manual
The heat flux on the surface of the channel wall was pressure was used to press the foam to the copper surface.
provided by a ceramic leaded chip resistors (LR Lastly, a solder wick was used to remove excessive solder and
13737T0100J; American Technical Ceramics). Each such to prevent the solder from blocking the pores of the foam
resistor is 1 cm x 1 cm and it can generate a maximum of 250 specimen. Visual inspection of the soldered foam specimen
W (250 W/cm2 of heat flux) and operate up to 100°C. Three revealed no unsoldered areas and little solder penetration into
resistors were soldered to the bottom surface of the copper the open pores of the foam. Due to the high surface tension
base plate, assuring minimal contact resistance and high and viscosity of water, only FC-72 is used as the working
conversion efficiency and provided the constant wall heat flux fluid in the less porous, smaller pore (92%, 20 PPI) foam
condition. The overview of the experimental set-up and detail sample. For the two-phase experiment, only FC-72 was used
of channel is shown in Fig. 3. as working fluid.

Test “clamshell”
Heater channel
Plexiglass
Power Foam
supply
Copper block

Thermocouple

Flow meter

Pump
Data Acquisition Pressure
Device Transducer

Chiller
Fig. 4. Photograph of open cell metal foam specimen
(a)
EXPRIMETAL RESULT
Single phase.
a. Pressure drop
Fig. 5 shows the fluid effect on the pressure drop of porous
media, revealing the pressure drop of water in 95% porosity
foam which was calculate by Eq. (8) to be 3 ~ 4 times greater
than FC-72 with 10 PPI. Fig. 5 also shows the pressure drop
of an empty channel for the flow of water and FC-72, which is
calculated by Eq. (19) ~ (21) [17].
UV 2 ª
'P «
2 ¬
^1  V 2  K ` 4 ˜ f L  ^1  V 2  K `º
c c
Dh
e e »
(19)
¼

(b)
Fig. 3. (a) Schematic of experimental set-up and (b) detail of
channel

89
 ª 0.674 3.44 º differences in the manufacturing processes used by different
« 24  
 1/ 2 » vendors, the material properties of the purchased foam could
4˜ x x (20)
f « 3.44  » / Re vary from one manufacturer to the next, masking the
« x 1 / 2
1  0.000029 ˜ x  »
2

« » anticipated porosity effect. The 92% porosity foam used in

¬ ¼ these experiments was manufactured with organic polymer
x / Dh foam which may not have been fully removed from the porous
x (21)
medium during the burn-out process. Such an organic
Re
residue was observed in the foam which was used for our
Due to the large gap of the channel, the pressure drop experiment.
for the unfilled channel is nearly negligible. The water The dispersion coefficient of 0.06 which was suggested
pressure drop for 95% porosity and 10PPI foam is 50 times by Calmidi and Mahajan [11] is used for the prediction of heat
greater than the open channel and with FC-72 it is 10 times transfer capability for water and agrees with the experimental
greater than the open channel. As the pore size becomes result to within 10%, thus confirming the suggested dispersion
smaller (higher PPI), the deviation of pressure drop between coefficient for water and the non-equilibrium effects looks
water and FC-72 becomes greater. In the foam configuration negligible for water. However due to limited pores in the
of 95% porosity and 20 PPI, the pressure drop of water is 4 ~ channel and foam specimen it would be needed further
4.5 times greater than FC-72. As expected, the higher pressure experimental investigation. Assuming that only small
drop is for water at the same volumetric flow rate as FC-72 departures from local thermal equilibrium occur, as has been
due to the high viscosity of water. The experimental found for water, for FC-72, the predicted correlation values
uncertainty of 15% is indicated in the experimental results. with different dispersion coefficients are compared with
Overall experimental results of high PPI foam are closer to the experimental results. The comparison is shown in Fig. 6 (b).
prediction. The larger discrepancy for the low PPI foam may The black and magenta lines shows the result of different
well be due to the low pressure drop which is nearly at the low dispersion coefficient of 95% 10 PPI foam and the dots show
limit of the pressure transducer. the experimental values and 10% error bar is drawn in the Fig.
6 (a) and (b). The change of dispersion coefficient from 0.06
Prediction Exp.
Water (H95% 10PPI)
to 0.03 decreases the predicted heat transfer coefficients 10%.
Water (H95% 20PPI) However, it increases improves the agreement between data
FC-72 (H95% 10PPI)
1000 FC-72 (H95% 20PPI) and prediction from a discrepancy of 17% to 11%. However,
FC-72 (H92% 20PPI) the previous assumption of local thermal equilibrium will
800
need to be verified in future experimental studies.
It is interesting to observe that while in the micro single
Pressure drop (Pa)

600
gap channel, the heat transfer rate of FC-72 was 25% of water
for the same geometry [16], with the porous foam the FC-72
400
heat transfer coefficients rise to approximately 35% of water,
at same configuration. However, even though the metal foam
200 Open channel, FC-72 could increase the heat transfer rate above that of an empty
Open channel, water channel, the achievable heat transfer coefficient is still lower
0 than that of a single micro-gap which was 24 kW/m2-K for
0 2 4 6 8 10 12
water and 5 kW/ m2-K for FC-72 [16].
Inlet flow rate (ml/s)

Fig. 5 Experimental result of pressure drop of metal foam 14000

Fluid: water
12000 Porosity:95%
b. Heat transfer coefficient.
Fig. 6 shows the effect of the pore count (PPI) and 10000
porosity on the metal foam heat transfer coefficient for each of
the two working fluids. While the change from 10 PPI to 20
h (W/m -K)

8000
PPI is theoretically expected to result in a 5% increase in the
2

water heat transfer coefficients, as seen in Fig 6 (a), the 6000

present experiment is not sufficiently precise to confirm this

4000 10PPI 20PPI
dependence. Fig. 6 (b) shows the experimental results for heat
Prediction
transfer to FC-72 and reveals that if the porosity is held 2000 Exp.1
constant, changing the pore size (PPI) by a factor of two does Exp.2
Exp.3
affect the dispersion conductivity but does not significantly 0
4 6 8 10 12 14
affect the heat transfer rate. Based on the working relations for
the Nu number (18), a reduction in porosity could be expected Inlet flow rate (ml/s)
to increase the effective thermal conductivity and lead to an (a)
increase by approximately 30% for the 3% change in porosity,
though no such shift is seen in the data. However, due to the

90
 Two-phase result. Fig. 8 shows the two-phase experimental
5000 results for 95% porosity and 10 PPI metal foam, which was
solder to the copper base block, at various mass fluxes. The
4000 experiment was done twice under the same conditions and the
average deviation between the experiments was found to be
less than 7%. The increase of mass flux from 20 kg/m2-s to 48
h (W/m -K)

3000
kg/m2-s enhanced the heat transfer rate by as much as 15%.
2

2000 95% 92% Further increasing the mass flux, up to 72 kg/m2-s, did not
10PPI 20PPI 20PPI enhance the cooling capability.
Prediction
Fig. 9 shows the two-phase experimental results for heat
1000
Exp.1 transfer to metal foam of 95% porosity and 20 PPI.
Exp.2
Exp.3 Increasing the mass flux from 20 kg/m2-s to 48 kg/m2-s
0
5 6 7 8 9 10 11
enhanced the heat transfer rate by as much as 10%. The
Inlet flow rate (ml/s)
repeatability which could be defined as the deviation of two
different experiments is less than 5%. The heat transfer
coefficient increased as the heat flux increased, due to the
(b) flow condition transitioning from subcooled to saturated flow
Fig. 6 Heat transfer rate comparison of different foam boiling. Further increasing of the mass flux from 48 kg/m2-s
configuration with same coolant (a) water (b) FC-72 as
to 72 kg/m2-s did not increase the cooling capability.
coolant

Fig. 7 shows the comparison of heat transfer rate vs. pressure

14000
drop of the foam filled channel and open channel ((the ratio of Mass flux (kg/m -s)
2

Porosity: 95%
foam heat transfer coefficient to open channel heat transfer , 20
12000 10PPI
, 48
coefficient) to (the ratio of foam pressure drop to open , 72
channel pressure drop)). It is to be noted that across the 10000
parametric range studied, the thermal enhancement provided
htp (W/m -K)

8000
by the foam is between 20% and 40% for FC-72 but only a
2

few percent for water. As the flow rate increases, the heat 6000
transfer to pressure drop ratio appears to drop slightly and
then increase before reaching a peak or plateau, with the low 4000
porosity foam displaying this trend more distinctly than the
high porosity foam. It is noteworthy that due to a higher 2000

pressure drop than the open channel, even though the porous 0
foam enhances the heat transfer capability, due to the low 50 100 150 200 250
value of the relative improvement, less than 40% for FC-72 Heat flux (kW/m )
2

and less than 10% for water, it appears that the higher porosity
and lower PPI foam constitutes the better choice.
Fig. 8 Mass flux effect on the two-phase heat transfer
0.6 coefficient of 95% porosity and 10PPI foam
Water FC-72
95% 10PPI
95% 20PPI
92% 20PPI 14000 2
Mass flux (kg/m -s)
, 20 Porosity: 95%
0.4 12000 , 48 20PPI
, 72
h/ho/'P/'Po

10000
htp (W/m -K)

8000
2

0.2
6000

4000

0.0 2000
5 6 7 8 9 10 11 12 13
0
Inlet flow rate (ml/s) 50 100 150 200 250 300
2
Heat flux (kW/m )
Fig. 7 Heat transfer rate vs. pressure drop comparison of
different foam configurations Fig. 9 Mass flux effect on the two-phase heat transfer rate of
95% porosity and 20 PPI foam

91
 Fig. 10 shows the two-phase experimental results for the 12000
95% 10PPI
metal foam of 92% porosity and 20 PPI. The two-phase heat 95% 20PPI G=20 kg/m -s
2

transfer coefficient variation of this foam configuration is as 10000 92% 20PPI

the same form as for 95% porosity, 20 PPI foam. Increasing
the mass flux from 20 kg/m2-s to 48 kg/m2-s shifted the
8000
highest heat transfer coefficient to the high heat flux and

htp (W/m -K)

weakened the dependence of the heat transfer coefficient on

2
6000
the heat flux. The enhancement was not distinct compare to
95% porosity foams. A further increase of mass flux up to 72
kg/m2-s did not enhance the heat transfer rate. 4000

14000 2 2000
Mass flux (kg/m -s)
Porosity: 92%
, 20
12000 20PPI
, 48 0
, 72 50 100 150 200
10000 2
Heat flux (kW/m )
(a)
htp (W/m -K)

8000
2

6000 12000
95% 10PPI 2
G=48 kg/m -s
95% 20PPI
4000 10000 92% 20PPI

2000
8000

0 htp (W/m -K)

2

50 100 150 200 250 6000

2
Heat flux (kW/m )
4000
Fig. 10 Mass flux effect on the two-phase heat transfer rate of
92% porosity and 20PPI foam
2000
Fig. 11 shows the effect of porosity and pore size (PPI) on
the wall heat transfer rate at three distinct flow rates. As 0
50 100 150 200 250
shown in Fig. 11 (a), at mass flux of 20 kg/m2-s, it is hard to
2
see the difference between foams of different pore size (PPI) Heat flux (kW/m )
and porosity. It is noticed that from the subcooled to the early (b)
boiling region, low pore count (large pore size) is slightly
better than the high PPI (smaller pore size). However, as the 12000
95% 10PPI
heat flux increases, the high pore count (smaller pore size) 95% 20PPI G=72 kg/m -s
2

shows higher heat transfer coefficients. The difference of heat 10000 92% 20PPI
transfer coefficient between 10 PPI and 20 PPI at 95%
porosity is less than 10% which is within the experimental
8000
error. The foam of 92% porosity and 20 PPI shows a lower
htp (W/m -K)

heat transfer rate, 20% lower than that of 95% porosity foam.
2

6000
Fig. 11 (b) shows the experimental results for of a mass flux
of 48 kg/m2-s. The thermal performance of the 95% porosity,
4000
10 PPI foam is 10% higher than 95% porosity, 20 PPI foam.
The lower porosity foam, 92% porosity and 20 PPI foam, is
20% lower than 95% porosity foam. A similar result is seen in 2000

Fig. 11 (c), where there is nearly no distinction between the

thermal performance of the 92% and 95% foams at low heat 0
50 100 150 200 250 300
flux, but up to a 23% lower thermal transport capability for
2
the 92% porosity and 20 PPI material at the highest heat Heat flux (kW/m )
fluxes tested. Interestingly, Fig. 11 (c) reveals 95% porosity (c)
and 10 PPI foam, shows highest thermal transport capability at Fig. 11. Porosity and pore size effect on the two-phase heat
same test condition. transfer rate at same mass flux (a) Experimental result of 20
kg/m2-s of mass flux, (b) Experimental result of 48 kg/m2-s of
mass flux and (c) Experimental result of 72 kg/m2-s of mass
flux

92
 As shown in Fig. 11, the increase of mass flux from 20 Experimental Thermal and Fluid Sciences, V. 28, pp. 193-199,
kg/m2-s to 48 kg/m2-s enhanced the heat transfer coefficient. 2004
However further increasing of mass flux to 72 decreased the [7] Leblond, G., Gosselin, L., “Effect of Non-local
heat transfer coefficient. This is due to the high mass flux Equilibrium on Minimal Thermal Resistance Porous Layered
suppress the flow boiling inside of the channel. Systems”, I. J. of Heat and Fluid Flow
[8] Giani, L., Groppi, G., Tronconi, E., “Heat Transfer
CONCLUSIONS AND DISCUSSION Characterization of Metalic Foams”, Industrial and
In the analysis of the characteristics of the porous foam Engineering Chemistry Research, V. 44, 1005, pp. 9078 -
channel in single phase heat transfer, a heat transfer 9085.
coefficient of 10 kW/m2-K was achieved with 95% porosity [9] Crittenden, P. E., Cole, K. D., “Design of Experiments for
and 20 PPI with water and a heat transfer coefficient of 2.85 Thermal Characterization of Metallic Foam”, J. of
kW/m2-K was achieved at the same configuration of foam Thermophysics and heat Transfer, V. 19, pp. 367 – 374, 2005
with FC-72 for the single phase. As expected, for single-phase [10] Calmidi, V. V., “Transport Phenomena in High Porosity
flow, the porous foam provides higher heat transfer rates than Fibrous Metal Foams”, Ph. D. thesis, U of Colorado, 1998
the open channel of the same dimensions, with the FC-72 heat [11] Calmidi, V. V., Mahajan, R. L., “Forced Convection in
transfer capability reaching 35% of the values attained with High porosity Metal Foams”, J of Heat Transfer, V.122, pp.
water. Based on the present experimental results, it would 557-565, 2000
appear that the dispersion coefficient for FC-72 may lie [12] Poulikakos, D., Renken, K., “Forced Convection in a
between 0.03 and 0.06. However this would need to be Channel Filled with Porous Medium, Including the Effects of
verified in future experimental studies due to limited pores in Flow Inertia, Variable Porosity and Brinkman Friction”, Trans.
channel and did not considering local thermal non-equilibrium of the ASME, V. 109, pp. 880-888, 1987
condition. Even though the porous foam channel provides [13] Prieur Du Plessis, “Pressure Drop Prediction for Flow
high heat transfer rate, the pressure drop was also greater than through High Porosity Metallic Foams”, Vhem. Eng. Sci.
the open channel. V.49, pp. 3545-3553, 1994
The heat transfer coefficient of horizontal two-phase flow [14] Hunt, M. L., Tien, C.L., “Effects of Thermal Dispersion
in the porous media reached up to 10 kW/m2-K at 95% on Forced Convection in Fibrous Media”, Int. J. of Heat Mass
porosity and 10PPI. For horizontal two-phase flow within the Transfer, V.31, n.2, pp. 301-309, 1988
parametric range studied – the foam porosity can have a [15] Bhattacharya, A., Calmidi, V. V., Mahajan, R. L.,
modest effect, up to 23%, on the heat transfer coefficient at “Thermophysical Properties of High Porosity Metal Foams”,
high heat fluxes. Alternatively, if the porosity is the same, the IJ of Heat and Mass Transfer, V.45, pp. 1017-1031, 2002
large pore size has better capability on heat transfer. It might [16] Kim, D. W., “Convection and Flow Boiling in Microgaps
be argued that the reason the high porosity foams do better at and Porous Foam Coolers”, Ph. D. thesis, U of Maryland,
high fluxes is that there is more vapor that needs to flow 2007
through the foam (i.e. higher quality) and that it is the porosity [17] Shah, R. K. and London, A.L., “Laminar Flow Forced
and large pore size – that determines how easy it is for the Convection in Ducts”, Supplement 1 to Advances in Heat
vapor to escape from the wall area. Transfer, Academics, New York,1978
[18] D. Poulikakos, K. Renken, “Forced Convection in a
REFERENCES Channel Filled with Porous Medium, Including the Effects of
[1] Boomsma, K., Poulikakos, D., Zwick, F., “Metal foams as Flow Inertia, Variable Porosity and Brinkman Friction”, Trans.
compact high performance heat exchangers”, Mechanics of Of the ASME, V. 109, pp. 880-888, 1987
Materials V.35, pp. 1161–1176, 2003 [19] Hunt, M. L., Tien, C.L., “Effects of Thermal Dispersion
[2] Koh, J. C. Y., Colony, R., “Analysis of Cooling on Forced Convection in Fibrous Media”, Int. J. of Heat Mass
Effectiveness for Porous Material in a Coolant Passage”, J. of Transfer, V.31, n.2, pp. 301-309, 1988
Heat Transfer-Transaction s of ASME 96, pp. 324-330, 1974 [20] Boomsma, K., Poulikakos, D., Zwick, F., “Metal Foams
[3] Lage, J. L., Weinert, A. K., Price, D. C. and Weber, R. M., as Compact High Performance Heat Exchangers”, Mechanics
“Numerical study of a Low Permeability Microporous Heat of Materials, V. 35, n. 12, pp. 1161-1176, 2003
Sink for Cooling Phased-array Radar Systems”, I. J. of Heat [21] Bejan, A., “Designed Porous Media: Maximum Heat
and Mass Transfer V. 39, pp. 3633-3647, 1996 Transfer Density at decreasing Length Scales”, Int. J. of Heat
[4] Boomsma, K., Poulikakos, D., “On the Effective Thermal and Mass Transfer, V. 47, n.14-16, pp.3073-3083, 2004
Conductivity of a Three-dimensionally Structured Fluid- [22] Amiri, A., Vafai, K., Kuzay, T. M., “Effects of Boundary
saturated Metal Foam”, I. J. of Heat and Mass Transfer, V. 44, Conditions on non-Darcian Heat Transfer through Porous
pp. 827-836, 2001 Media and Experimental Comparisons”, Numerical Heat
[5] Jeng, T.-M., Hwang, G. J., Hung, Y. H., “Thermal Transfer, V. 27, pp.651-654, 1995
Performance of Metallic Porous Channels for Electronics [23] Prieur Du Plessis, “Pressure Drop Prediction for Flow
Cooling Application”, Advances in Electronic Packaging V. 2, through High Porosity Metallic Foams”, Vhem. Eng. Sci.
pp. 943-949,2001 V.49, pp. 3545-3553, 1994
[6] Tadrist, L., Miscevic, M., Rahli, O., Topin, F., “About the [24] Alazmi, B., Vafai, K., “Analysis of Variable Porosity,
Use of Fibrous Materials in Compact Heat Exchangers”, Thermal Dispersion, and Local Thermal Nonequilibrium on

93
Free Surface Flows Through Porous Media”, Journal of Heat
Transfer, V.126, pp. 389-399, 2004
[25] Fouriea, J. G., DuPlessis, J. P., “Pressure Drop Modeling
in Cellular Metallic Foams”, Chemical Eng. Science V. 57, pp.
2781 – 2789, 2002
[26] Larkins. R., White, R., “Two-Phase Co-Current Flow in
Packed Beds”, AIChE. J. V.7, pp. 231-239, 1961
[27] M. Saada, “Fluid Mechanics of Co-Current Two-Phase
Flow in Packed Beds: Pressure Drop and Liquid Hold-up
Studies” Chem. Ind. Genie Chem. V.105, pp. 1415-1421, 1972
[28] M. Jamialahmadi, H. Muller-Steinhagen, M.R.
Izadpanah, “Pressure Drop, Gas Hold-up and Heat Transfer
during Single and Two-phase Flow through Porous Media”,
Int. J of Heat and Fluid Flow, V. 26, pp. 156–172, 2005
[29] K.M NG, “A Model for Flow Regime Transitions in
Cocurrent Down-flow Trickel-bed Reactors”, AlChE J. V. 32,
n.1, pp. 115-122, 1986
[30] Saez, A. E., and Carbonell, R. G., “Hydrodynamics
Parameters for Gas-Liquid Co-current Flows in Packed Beds”,
AIChE J., v. 31, pp. 52-62, 1985
[31] Schulenburg, T., Muller, U., “An Improved Model for
Two-Phase Flow through Beds of Coarse Particles”, Int. J.
Multiphase Flow, v. 13, pp. 87-97, 1987

94