Sei sulla pagina 1di 18

International Journal of Heat and Mass Transfer 58 (2013) 356373

Contents lists available at SciVerse ScienceDirect

International Journal of Heat and Mass Transfer


journal homepage: www.elsevier.com/locate/ijhmt

Enhanced forced convective cooling of heat sources by metal-foam porous layers


Chih-Cheng Chen a, Po-Chuan Huang a,, Hsiu-Ying Hwang b
a
b

Department of Energy and Refrigerating Air-Conditioning Engineering, National Taipei University of Technology, Taipei 106, Taiwan, ROC
Department of Vehicle Engineering, National Taipei University of Technology, Taipei 106, Taiwan, ROC

a r t i c l e

i n f o

Article history:
Received 7 March 2011
Received in revised form 6 November 2012

Keywords:
Metal foam
DarcyBrinkmanForchheimer ow
Local thermal non-equilibrium

a b s t r a c t
A numerical investigation has been conducted for enhanced heat transfer from multiple discrete heated
sources in a horizontal channel by metal-foam porous layer. Both DarcyBrinkmanForchheimer ow
model and two-equation energy model based on local thermal non-equilibrium are used to characterize
the thermo-ow elds inside the porous regions. Solution of the coupled governing equations for the porous/uid composite system is obtained using a stream functionvorticity analysis.
The results show that an increase in the soliduid interfacial heat exchange results in a decrease in the
temperatures difference between the solid and uid phases for xed Reynolds number, where the porous
media tend to reach local thermal equilibrium (LTE) with the uid and a larger cooling augmentation of
heaters is obtained.
2012 Elsevier Ltd. All rights reserved.

1. Introduction
Considerable attention has been given to thermal control of
electronic equipment and devices in the past decades due to the
requirement of the maintenance of relatively constant electronic
component temperature equal to or below a maximum operation
temperature. Studies have shown that a signicant change in the
design temperature can lead to a perceptible reduction in the reliability of the electronic components [1]. Different high-effective
cooling techniques [2] have been used in the past to obtain heat
transfer enhancement with a minimum of frictional losses including the traditional methods of natural and forced convective cooling. One of the promising techniques is the application of a porous
material. This is due to the high ratio of surface area to volume ratio in the heat transfer process and intense ow mixing, caused by
the tortuous path of the porous matrix, in the thermal dispersion
process.
The transport phenomena through porous media have been of
continuing interest due to its relevance in diverse engineering
applications. Such applications include the thermal insulation, geothermal energy systems, heat exchanger, enhanced oil recovery,
drying processes, and unclear waste disposal. Porous media is also
utilized in applications such as heat pipe technology, industrial furnace, cooling of electronic equipment, and xed-bed nuclear
propulsion.
The problem of forced convection enhancement in a channel
fully or partially packed with a porous material has been studied
extensively in the literature (see Kaviany [3] for a good review
Corresponding author. Tel.: +886 2 27712171x3514; fax: +886 2 27314919.
E-mail address: pchuang@ntut.edu.tw (P.-C. Huang).
0017-9310/$ - see front matter 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ijheatmasstransfer.2012.11.041

on the subject). Koh and Colony [4] analyzed the cooling effectiveness for a porous material in a cooling passage. Kaviany [5] dealt
with convective heat transfer from a steady laminar ow through
a porous channel bounded by two isothermal parallel plates.
Huang and Vafai [6] simulated steady forced convection problem
in an isothermal parallel plate channel with porous block array.
Angirasa [7] numerical reported forced convection in a channel
lled with metallic brous materials. Their results showed that
porous substrate substantially enhance the thermal performance
in a channel. A majority of these studies relate to the aspect of
forced convection over a fully/partial porous channel system with
one-energy-equation model based on the local thermal equilibrium (LTE) assumption. Recently, the forced convection heat transfer in a fully/partially porous channel with discrete heated sources
or blocks was of special interest due to its applications on the
microelectronic cooling. Hadim and Bethancourt [8] investigated
forced convection in fully/partially porous channel containing discrete heat sources on the bottom wall. A signicant increase in
heat transfer rate was observed, as the Darcy number was decrease, especially at the leading edge of each heat source. Angirasa
and Peterson [9] numerically studied forced convection heat transfer augmentation in a channel with a localized heat source using
metal brous material. They concluded that thinner bers and high
porosity media enhance heat transfer because of better distribution ow and mixing. Fu et al. [10] investigated numerically heat
transfer from a spherical-bead porous-block-mounted heated wall
in a channel ow. They reported that for the blocked ratio Hp 0:5
the thermal performances are enhanced by higher porosity and
porous particle diameter. However, the result is opposite for
Hp 1. Huang et al. [11] analyzed forced convective heat transfer
from multiple heated blocks in a channel by porous covers and

C.-C. Chen et al. / International Journal of Heat and Mass Transfer 58 (2013) 356373

357

Nomenclature
asf

K
L
n
Nu
Nusf

specic soliduid interfacial surface area of porous


layer (m1)
surface area (m2)
constant parameter in Eq. (8)
specic heat at constant pressure (J/kg K)
Darcy number, K/R2
ber diameter (m)
mean pore diameter (m)
function used in expression inertia terms
shape function
convective heat transfer coefcient (W/m2 K)
uid-to-solid heat transfer coefcient (W/m2 K)
height (m)
thermal conductivity (W/m K)
dispersion conductivity (W/m K)
effective thermal conductivity of uid (W/m K)
effective thermal conductivity of the solid matrix (W/
m K)
permeability of the porous medium (m2)
length (m)
the normal direction to curves
Nusselt number, hR/kf
h a R2
interfacial Nusselt number, sf ksf

Nusf df

interfacial Nusselt number based on the ber diameter,

P
Pe
PPI
Pr
q00
R
Re
Redf
S
T
uo
u,v
V

pressure (N/m2)
Pelect number, uoR/a
pore density (pores/inch)
Prandtl number, m/a
heat ux (W/m2)
height of channel (m)
!
Reynolds number, uoR/m
j V jd
Reynolds number based on the ber diameter, m f
spacing between heat sources or porous layers (m)
temperature (K)
averaged velocity of the inlet ow (m/s)
velocity components (m/s)
velocity vector (m/s)

Ai
B
Cp
Da
df
dp
F
G
h
hsf
H
k
kd
kfe
kse

W
x, y
hi
jj

width (m)
Cartesian coordinates (m)
volume average
magnitude or absolute value

Greek symbols
a
thermal diffusivity (m2/s)
c
dispersion coefcient
e
porosity of the porous medium
f
vorticity (1/s)
keff
total effective thermal conductivity ratio, keff = keff/kf =
ks + kf
kf
uid-phase effective thermal conductivity ratio, kfe/kf
ks
solid-phase effective thermal conductivity ratio, kse/kf
l
dynamic viscosity (kg/m s)
m
kinematic viscosity (m2/s)
q
density (kg/m3)
w
stream function (m2/s)
u
general property
Superscript

dimensionless quantity

hsf df
kf

found that the recirculation caused by porous-covering block will


signicantly augment the heat transfer rate on both top and right
faces of second and subsequent blocks.
More recently, due to the needs of high-performance highpower electronic devices, there has been an increasing demand to
achieve higher heat transfer removal from the fully/partially porous
channel ow. One of such efforts has been given to exploring the
use of high porosity media (e > 0.85) porous media-metal foams
as heat sink due to the great strength-to-weight ratio, high strength
and rigidity structure, and high effective thermal conductivity. The
man-made porous medium-metal foams have gained attention as
potentially excellent candidates for meeting the high thermal dissipation demands in the thermal applications, such as multifunctional heat exchangers [12,13], high power battery and thermal
shielding, shock-absorbers, corrosion-resistant and hightemperature lters and acoustic applications, compact heat sinks
for electronic cooling [1416] etc. Calmidi and Mahajan [17] conducted an experimental and numerical study of force convection
in a fully metal-foam porous channel with a single heat source,
and the aluminum metal foams (alloy T-6201) is in a variety of pore
densities (540 PPI). Nusselt number data has been obtained as a
function of the Reynolds number based on permeability. Tzeng
[18] numerically investigated the uid ow behavior and the heat

Subscripts
0
reference or ambient
eff
effective
f
uid
fe
effective value for the uid
i
inlet
m
overall mean
non
nonporous
o
outlet
p
porous
s
solid
se
effective value for the solid matrix
sf
specic surface
t
total
w
wall
x
local

transfer mechanism in a bronze-sintered porous channel that contains periodically spaced heated blocks. The results showed that the
average Nusselt number for each block decreased along the direction of the ow until it reached its fully developed value. The Nusselt number increased with relative block height and Reynolds
number. Most of these studies, the heat transfer problems of a fully
metal porous channel were studied with two-energy-equation
model based on the local thermal non-equilibrium (LTNE) assumption. However, due to the mathematic difculties in simultaneously
solving the coupled governing equations for both the metal-foam
porous and uid regions, very little work has been done on forced
convection uid ow and heat transfer in the metal-foam porous/
uid composite system, in which the local thermal non-equilibrium
(LTNE) assumption is used in porous region. Phanikumar and Mahajan [19] numerically and experimentally studied the buoyancyinduced ow in a high porosity aluminum foam heated from below
and indicated the two-equation energy model is a better model
when uid/porous interfaces are involved. The purpose of this
study is to explore the effects of the heat transfer enhancement factor by high-porosity metal-foam porous layers on the cooling of
electronic components in the channel.
This paper presents a numerical investigation of forced convective cooling enhancement of a two-dimensional array of multiple

358

C.-C. Chen et al. / International Journal of Heat and Mass Transfer 58 (2013) 356373

erally arranged in rows in such a way that the spacing between


components in each row (in the third dimension inside paper) is
small compared to the spacing between the rows. Also, the length
of each row is at least one order of magnitude higher than the width
of each component. Therefore, the two-dimensional model for the
ow and temperature elds provides a good approximation. The
uid enters the channel at uniform temperature T0 with a parabolic
RR
velocity prole ui = uo6(y/R)[1  (y/R)], where u0 1R 0 ui dy are the
averaged velocity of the inlet ow. The ow is assumed to be steady,
incompressible, and two-dimensional. The buoyancy and radiation
effects are neglected. Besides, since the metal foam matrix is isotropic [19], the thermophysical properties of the uid and the porous
matrix are assumed to be constant. The uid-saturated porous
medium is considered in local thermal non-equilibrium (LTNE)
with the uid. Possible channelling near the wall is neglected in
the present study because brous media are considered for which
the porosity and permeability are relatively constant even close
to the wall [20]. The local volume averaging conservation equations have been derived by Vafai and Tien [21], this technique is
used in developing the governing equations in the region of porous
matrix. The ow is modeled by the DarcyBrinkmanForchheimer
equations [22] in the porous matrix to incorporate the viscous
and inertial effects, and by the NavierStokes equations in the uid
domain. The mathematical model for energy transport is based on
the two-temperature-equation model which assumes local thermal
non-equilibrium (LTNE) between the uid and the solid phases
[17,22,23]. The temperature of the solid and the uid are solved
separately.
With the above assumptions, the governing conservation equations for the present problem can then be separately written for
the porous and uid regions. Treating the uid-saturated porous

heated sources located upon lower plate of an insulated channel


utilizing the high-porosity metal foams porous layer. In this work,
the thermal interactions between the solid and uid phases inside
the porous regions and how the heater cooling is affected by these
interactions will be analyzed in detail. Furthermore, the inuences
of various parameters governing the hydrodynamic and thermal
characteristics of the problem are also examined to establish the
fundamental effects and provide practical results. Another objective is to attain a better understanding of the situations under
which the local thermal equilibrium (LTE) or local thermal nonequilibrium (LTNE) assumption would be justiable. It is shown
that specic choices in certain governing parameters, such as the
permeability, pore density, ber diameter and the effective thermal conductivity of the metal-foam porous layer, can have profound effects on the cooling of the heat sources.
2. Mathematical formulation
The schematic of the physical model and coordinate system under investigation is shown in Fig. 1(a). It includes ow through a
horizontal parallel-plate channel mounted with 4 split heat sources
on the bottom plate. The porous layer mounted on each heat source
is used as heat sink. Both upper and lower channel walls are insulated. Each heat source dissipates an equal and uniform heat ux
q00 over its length. The conguration and assumptions are assumed
to simulate the cooling problem of electronic components. Such a
conguration can be observed in electronic cooling applications,
where the IC boards usually have a repeating pattern, so the top
plate and the bottom plat without heat source are assumed to be
insulated. An inline discrete heat sources on bottom plat represents
an array of IC components. For a PC board, IC components are genHeat Source
Ti

Porous Medium
Lt

Hp*=1

ui

y
x

Parabolic Velocity
Profile

Li

Wp

Sp

Lo

(a)

10PPI

20PPI

40PPI

(b)

(c)
Fig. 1. (a) Schematic diagram of the problem and the corresponding coordinate systems, (b) Samples of different pore density aluminum foam with a graduated millimeter
scale [29], and (c) a typical non-uniform grid system for the whole computational domain.

359

C.-C. Chen et al. / International Journal of Heat and Mass Transfer 58 (2013) 356373

medium as a continuum, the local-volume averages of the conservation equations for mass, momentum, and energy in the porous
region are

r  hV i 0

!
!
1 !
1
1
h V  r V i  rP meff r2 h V i 

qf


eF ! !
p jh V ij h V i
K
K

mf

Solid-phase energy equation:

0 r  kse rhT s i  hsf asf hT s i  hT f i

The conservation equations for mass, momentum, and energy in the


pure uid region are

!
r V 0
!
!
!
1
V  r V  rP mf r2 V

qf

In the above formulations, h i represent the aforementioned volume-averaged quantities. Notably, the relevant empirical coefcients in the present numerical model, such as kfe (the effective
stagnant conductivities of uid), kse (the effective stagnant conductivities of solid matrix), kd (the dispersion conductivity), asf (the specic soliduid interfacial surface area), hsf(uid-to-solid heat
transfer coefcient), K (permeability) and F (inertial parameter),
generally do not have universal values. This is because these empirical coefcients mainly depend on the geometry of metal foam, the
thermal properties of the solid and the uid, and the ow rate.
Moreover, all these empirical coefcients are difcultly measured
and the suitable formulas for these coefcients can be found in
the open literatures to build a numerical model herein. The thermal
properties of metal foam, kfe and kse, kd, asf and hsf, for energy equation are taken from Calmidi and Mahajan [24], Koch and Brady [25],
Calmidi and Mahajan [17], and Zhukauskas [26]. The properties of
porous matrix of metal foam, K and F, for momentum equation
are taken from Calmidi [27]. These empirical correlation formulas
are shown as follows:

B 0:181 for air 24

the dispersion coefficient for air 17


asf

3pdf
0:59dp

1  e1e=0:04 

8
9
10

8
0:37
; 1 6 Redf 6 40
0:75Re0:4
>
df Pr
>
>
<
hsf df
0:5
0:37
0:51Redf Pr ; 40 6 Redf 6 1000
Nusf df
>
kf
>
>
:
0:37
0:26Re0:6
; 1000 6 Redf 6 2  105
df Pr
where Redf

14
The ber diameter, df, is measured by using a microscope. The pore
diameter dp is estimated by counting the number of pores in a given
length of material. Both df and dp are average values. Besides, the
pore density is the number of pores per unit length of the material
(PPl, pores per inch, see Fig. 1(b) [29]), which usually is a nominal
value supplied by the manufacturer.
The associated boundary conditions necessary to complete the
formulation of the present problem are:
(1) At the channel (x = 0, 0 6 y 6 R), the ow with a parabolic
velocity prole is given.

u uo 6y=R1  y=R;
6

!
V  rT f af r2 T f

keff kfe kse ekf B1  e0:763 ks ;


p !
kd cqC p K j V j; where c 0:06

r
df
1  e 1
1:18
; where G 1  e1e=0:04 shape function
3p G
dp

Fluid-phase energy equation:


!
qCpf h V i  rhT f i r  kfe kd rhT f i hsf asf hT s i  hT f i

As it was mentioned earlier, Calmidi [27,28] developed a model for


df/dp as a function of porosity after modication to reect the difference between the open cell represent and the three-dimensional
dodecahedron structure.

v 0;

Tf T0

15

(2) At the exit (x = Lt, 0 6 y 6 R), the fully developed conditions


are satised.

@v
0;
@x

@u
0;
@x

@T f
0
@x

16

(3) Along the lower channel walls (0 < x < Lt, y = 0) [18,23], the
no-slip conditions are taken.

u 0;

v 0

17a

without heat sources :

@T f
0;
@y

with heat sources : q00 kfe

@T s
0
@y

@T f
@T s
 kse
;
@y
@y

17b

Tf Ts

17c

(4) Along the upper plate (0 < x < Lt, y = R), the no-slip conditions at the perfect-insulated plates are taken.

u 0;

v 0;

@T f
0;
@y

@T s
0
@y

18

(5) Along the uid/porous interface (the continuities of the


velocity, pressure, stress, temperature, and heat ux are satised.) [18,23]

huip jgx;y0 uf jgx;y0

19a

hv ip jgx;y0 v f jgx;y0

19b

hPip jgx;y0 Pf jgx;y0






@huip @hv ip 
@uf @ v f 

leff

f
@n
@t gx;y0
@n
@t gx;y0

19c

hT f ip jgx;y0 T f jgx;y0



@hT f ip 
@hT s ip 
@T f 


kfe

k
se
f
@n gx;y0
@n gx;y0
@n gx;y0

19d
19e
19f

where g(x, y) = 0 are the curves dening the uid/porous interfaces,


and the derivative with respect to n and t represents the normal and
tangential gradients to these curves at any point on the interfaces.

!
j V jdf

m
11

 1:63
df
F 0:002121  e
dp
 1:11
K
0:224 df

0:000731

e

2
dp
dp
0:132

Introducing the stream function w and vorticity f as

12
13

@w
;
@y
@hv i @hui


@x
@y

u hui

v hv i 

@w
;
@x

@ v @u

@x @y
20

360

C.-C. Chen et al. / International Journal of Heat and Mass Transfer 58 (2013) 356373

and the following dimensionless variables:

x
y
Lt
Wp
Hp
; y ; Lt ; W p
; Hp
;
R
R
R
R
R
Li
Lo
Li ; Lo
R
R
u
v ! p
u ; v  ; j V  j u2 v 2
u0
u0
P
T  T0
w
Rf


; w
; f
P 2 ; T 00
u0 R
u0
q R=kf
qu0
df
dp


asf asf R; df ; dp
R
R

keff can be either measured experimentally or estimated from the


porosity and the thermal conductivities of the solid matrix and
uid. The dimensionless empirical correlation formulas are shown
below:

x

21a

kfe
e;
kf
 
kse
ks
ks
0:1811  e0:763
kf
kf

keff
21b
21c
21d

Then dimensionless stream functionvorticity formulation of the


above two sets of conservation equations are as follows:

keff
ks kf ;
kf

kf

30

aSf asf  R

3pdf

0:59dp 2
asf

Nusf Nusf df

1  e1e=0:04 

31
32

df

i.e.
1. Dimensionless governing equations in the uid domain:

@ 2 w @ 2 w

f
@x2 @y2

22

"
#
@f
@f
1 @ 2 f @ 2 f

u  v  
Ref @x2 @y2
@x
@y
" 2 
#
2 
@T f
@T f
1 @ Tf @ Tf
u  v  

Pef @x2 @y2


@x
@y

23

up

24

25

" 2 
#
2 
@fp
@fp
1 @ fp @ fp
e


vp 

f
Reeff @x2 @y2
@x
@y
Reeff Da p
F e2 ! 
 p  V p fp
Da
! 
! 3
2
 
 
@  V p 
F e2 4  @  V p 

p


v p   up  5
@x
@y
Da

"
#

p !  @ 2 T f @ 2 T f
@T f
@T f
kf



u
v

c Daj V j

@x
@y
Pef
@x2 @y2

Nusf  

T s  T f
Pef
"
#


@ 2 T s @ 2 T s



T

T
0 ks

Nu
sf
s
f
@x2 @y2

! 
where Redf Re  j V  jdf


Reeff

u0 R

meff

Da

hsf asf R2
Nusf
;
kf

K
R2

kf

kfe
;
kf

ks

kse
kf

26

asf R2
df

34

dp


 2
df

Da 0:00073 dp 1  e0:224 
dp
r

df
1  e
1
 1:18
3p 1  e1e=0:04
dp

!1:11
35
36

(1) At the channel inlet x Li ; 0 6 y 6 1,

v  0;

u 6y  y2 ;


2

T f 0

37a

3

w 3y  2y ;

f 0

37b


(2) At the exit x Lt  Li ; 0 6 y 6 10

27
28

29a
29b

@u
0;
@x

@T f
0;
@x

@v 
0;
@x

@ 2 w
0;
@x2

@f
0
@x

38

(3) Along the upper channel wall

v  0;

w 0;

f 

@ 2 w
@y2

39a

Without heat source

@T f
0;
@y

@T s
0
@y

39b

With heat source

kf

hsf df
Nusf df
;
kf

where Nusf Nusf df

!1:63

The associated dimensionless initial conditions and boundary conditions necessary to complete the formulation of the present problem are:

The non-dimensional parameters in above-mentioned equations are as follows:

mf
kf
u0 R
Ref
; Prf ; Pef
Ref Prf ; af
mf
af
af
qC p

df

F 0:002121  e0:132

u 0;

u0 R

33

2. Dimensionless governing equations in the porous medium


domain:

@ 2 wp @ 2 wp
2 fp
@x2
@y

8
a
0:37
>
; 1 6 Redf 6 40
0:75 dsf Re0:4
>
df Pr
>
f
>
>
>
<

a
0:37
Nusf 0:51 dsf Re0:5
; 40 6 Redf 6 1000
df Pr
f
>
>
>
>

>
>
: 0:26 asf Re0:6 Pr0:37 ; 1000 6 Red 6 2  105
df
f
d

@T f
@T 
ks s 1;

@y
@y

T f T s

39c


(4) Along the lower channel wall Li < x <> Lt  Li ; y 1

29c

meff is almost equal to mf from the experiment of Lundgren [30]. kf is


uid-phase effective thermal conductivity ratio, and ks solid-phase
effective thermal conductivity ratio depending on the organization
of porous media. The effective stagnant thermal conductivity ratios

u 0;

v  0;

w 1;

f 

@T f
0;
@y

@T s
0
@y

@ 2 w
@y2

(5) Along the uid/porous interface

40a
40b

361

C.-C. Chen et al. / International Journal of Heat and Mass Transfer 58 (2013) 356373

41a
41b
41c
41d
41e

41f

gx;y0

To assess the effects of the metal foam porous media on the heat
transfer rate at the heat sources, the local Nusselt number along
the surface of heat sources is evaluated as [18,19]

hR
Nux s Nux f
kf


  
kf @T f 
ks @T s 
 


T w @y y 0 T w @y y 0

Nux

42

T 0
where T w Tqw00 R=k
is the dimensionless surface temperature at the
f
heat source.
The overall mean Nusselt number for a heat source is calculated
as follow

Num

1
W

Z
0

Nux dx Num s Num f

43

where W is the overall exposed length of heat source.


3. Numerical method
A control-volume integration method developed by Patankar
[31] is employed to solve the governing equations along with the
boundary conditions. Due to stability considerations, the
convective terms are discretized with upwind differencing and
the diffusion terms are discretized with the second-order difference
scheme. The second-order central difference approximations are
used for most of the spatial derivatives for the inner grid points.
The interface between the porous medium and uid space requires
special consideration due to the sharp change of thermophysical
properties, such as the permeability, porosity, and the thermal conductivity, across the interface. The harmonic mean formulation
suggested by Patankar [31] was used to handle these discontinuous
characteristics in the porous/uid interface. This ensures the continuity of the convective and diffusive uxes across the interface
without an excessively ne grid. The computational domain is chosen to be larger than the physical domain to eliminate the channel
exit effects and to satisfy continuity at the exit. A non-uniform grid
system is employed for the present calculations over discrete cells
surrounding the grid points, as shown in Fig. 1(c). This grid system
is designed to capture the steep gradients near the porous/uid
interfaces as well as boundaries, and to provide sufcient grid density at the interface regions. The resulting nite difference equations are solved by GaussSeidel iteration scheme with the
under-relaxation factor. The non-uniform grid system is generated
by using the Robertss general stretching transformations [32].
Here, a Fortran computer program was developed to carry out the
numerical resolution of the coupled governing stream function
vorticitytemperature equations for the present porous/uid
composite system and to generate the meshes. In this work,

convergence was considered to have been achieved when the


absolute value of relative error on each grid point between two
successive iterations was found to be less than 104. The iterative
procedure was then terminated. In addition, the grid independent
test is performed by examining the impact of grid size on Nusselt
number of heat source. The aluminum-foam heat sinks (qs = 2690
kg/m3 and ks = 218 W/mK for aluminum-alloy T-6201 [17]) are
employed, and air is used as the working uid (qf = 1.16 kg/m3
and kf = 0.0263 W/mK). The major physical and geometric

parameters are Re 2000; Pr 0:7; e 0:95; df 0:02; W p 1;

and Hp 1. The proper combinations of Dx and Dy are employed
to assure stability for the grid independent test. Five sets of grid
systems, 154  51, 183  61, 212  71, 251  81 and 299  101,
are investigated. Fig. 2(a) shows the convergence trend of local
Nusselt number distribution from rough meshes to ne meshes,
and there is only less than 3% relative error in the overall mean
Nusselt number for the 251  81 grid system compared with nest
grid system in Fig. 2(b). All computations for the present work are
based on the 251  81 grid meshes for the conservative reason.
The mathematical model and the numerical scheme are to be
validated by comparing the current numerical results with three
relevant limiting cases available in the literatures. The relevant
studies for our case correspond to the problems: (1) laminar forced
convection in a channel with a non-metal porous-block-mounted
heat source in the upper channel wall (i.e., W p 1:0; Hp 0:5,
and T f T s ), typifying the local thermal equilibrium (LTE) case
for Re = 250, Pr = 0.72, e = 0.9, F = 0.55, Da = 1  102, and keff = 1;
(2) a steady laminar forced convection in a channel with two heat
Grid points
154 X 51
183 X 61
212 X 71
251 X 81
299 X 101

800

600

Nux 400
200

x*

(a)
(100%)

30

Absolute value of relative error of Num









up 
uf 
; v p 
v f 
gx;y0
gx;y0
gx;y0
gx;y0








wf 
; fp 
ff 
wp 
gx;y0
gx;y0
gx;y0
gx;y0
  
  
@v f 
@v p 


leff
lf
@n gx;y0
@n gx;y0
 

 

@uf @ v f 
@u
@v  

leff p p 
lf

@n
@t gx;y0
@n @t gx;y0

D E 

T f 
T f 
p gx;y0
gx;y0
D E 

  


@ Tf 
@ T s p 
@T f 
p

ks

kf


@n 
@n 
@n gx;y0

gx;y0

4
3

25

No.4 Heat Source


No.3 Heat Source

No.2 Heat Source


4
3

20
1

15
1

10

4
3
2

No.1 Heat Source

5
4
3
2
1

154X51

183X61

212X71

251X81

2
3
4
1

299X101

Grid points

(b)
Fig. 2. The diagram of grid independent test on (a) the local Nusselt number
distributions, and on (b) relative error of mean Nusselt number.

362

C.-C. Chen et al. / International Journal of Heat and Mass Transfer 58 (2013) 356373

solid blocks at uniform temperature (i.e., Da ! 0; T f


T s ; ks =kf ! 1, and Hp 0:25), representing the solid-block heater
with non-porous heat sink case for Sp 1; Li 5:0;
Lo 29:0; Pr 0:7 at Re = 500 and 700 for 0:5 6 W p 6 3e; (3) a
steady forced convection in a full high porosity
metal-foam
p
channel and air as working uid for Rek u0 K =m 15  90 and
Pr = 0.7 (Aluminum metal alloy T-6201 foam sample # 3 [17] is
selected to compare with the present results, i.e., W p 1:81;
Hp 1; R 0:063 m for physical dimensions; and e = 0.9486,
F = 0.097, df = 0.0004 m, dp = 0.00313 m, K = 1.2  107 m2, kse =
4.1, kfe = 0.0248). The results for the rst case are within less than
1 percent agreement with the numerical results reported by Sung
et al. [33] for both streamlines and isotherms, as shown in the
Fig. 3(a) and (b). For the second case, the results agree to better
than 2.1% with data provided by Kim and Kang [34] for streamlines

-----Sung et al. [33]


Authorsresults

and space-averaged Nusselt number (Num)s of each solid block for


the steady forced convection over two heat blocks in a channel, as
shown in Fig. 3(c) and (d). The results for the third case are shown a
better agreement with the experimental and numerical simulation
results of Calmidi and Mahajan [17] for an average Nusselt number
of the heated surface (NuL = hf L/ke, where L = 0.114 m is the length
of the heated section of metal foam.) under different Reynolds
number based on permeability, as displayed in Fig. 3(e).
4. Results and discussion
The aluminum-foam heat sink, made of aluminum-alloy T-6201
[17], is employed in the present study. The density and thermal
conductivity are almost identical for the above aluminum-alloy
heat sinks, i.e. qs = 2690 kg/m3 and ks = 218 W/mK, in spite of the

Kim and Kang [34]


Authors results

0-1

0
-1

(a) Streamlines

(c) Streamlines
13
12

R e = 70 0
- - - - R e = 50 0

Kim and Kang [34]

R e = 70 0

A u tho rs' res ults

R e = 50 0

First block

11

(Num )s

10
1

Seco nd blo ck

8
0.01

0
-1

6
5

0. 5

1. 5

2. 5

w /H

(b) Isotherms

(d) overall mean Nusselt number


H*p=1
Y

15

NuL=hfL/ke

++

10

++

Presents results
Experimental results by Calmidi and Mahajan [17]
Simulation results by Calmidi and Mahajan [17 ]

20

40

60

80

100

120

Re k

(e)
Fig. 3. The results compared with other literatures: (a) and (b) streamlines (Dw = 0.2, 0 6 Dw 6 1) and isotherms (DT = 0.03 beginning with T = 0.01) compared with Sung
et al. [33], (c) and (d) streamlines (Dw = 0.2, 0 6 Dw 6 1) and overall mean Nusselt number compared with Kim and Kang [34], and (e) overall mean Nusselt numbers of the
heated surface compared with Calmidi and Mahajan [17].

363

C.-C. Chen et al. / International Journal of Heat and Mass Transfer 58 (2013) 356373

the ow and temperature elds near the porous-layer-mounted


strip heat sources clearly, only this region and its vicinity were
presented. However, at all time, the much larger domain was used
for numerical calculations and interpretation of the results.

difference in their material composition. Air is used as the working


uid (Pr = 0.7, qf = 1.16 kg/m3, and kf = 0.0263 W/mK). In addition,
the xed input parameters utilized in the simulation are
R 20 mm; W p 1; and Hp 1. In this study, the effects of
governing physical parameters, such as the Darcy number
(4.58  105 6 Da 6 1.15  103), Reynolds number (250 6 Re 6

2000), porosity (0.85 6 e 6 0.95), ber diameter 0:025 6 d f 6
0:01, and pore density (5 6 PPI 6 40) on the ow and heat transfer
characteristics are explored. Note that for illustrating the results of

4.1. Effect of Darcy number Da


The Darcy number, Da = K/R2, is the dimensionless quantity
related to the permeability of porous medium. The permeability

Table 1
Properties for different porosities of aluminum-alloy T-6201 [17].

e
0.85
0.875
0.9
0.925
0.95

df (mm)

dp (mm)

0.4
0.4
0.4
0.4
0.4

K (m2)

R (mm)

2.62
2.81
3.02
3.22
3.32

Da
8

20
20
20
20
20

4

6.20  10
8.03  108
1.05  107
1.37  107
1.65  107

1.55  10
2.01  104
2.63  104
3.41  104
4.12  104

asf (m1)

keff (W/m K)

keff

0.058
0.067
0.078
0.089
0.099

1536
1307
1089
886
701

9.30
8.10
6.83
5.49
4.04

353.66
307.86
259.83
208.82
153.53

Noted that: the value of dp is calculated by Eq. (14), K by Eq. (13), F by Eq. (12), asf by Eq. (10) and keff by Eq. (8).

0.5

Da=1.15x10-3
-4
Da=1.03x10
-5
Da=4.58x10
Da=

y*
0.25

0.5

0.5

0.5

0.25

0.25

0.25

at x*=0.5 plane
0
0

0.5

u*

at x*=2.5 plane
0
0

0.5

at x*=4.5 plane
0
0

u*

0.5

at x*=6.5 plane
0
0

u*

0.5

u*

(a)
0 .2

D a=

0 .1
0
0 .2

D a=1 .1 5 x 1 0 -3

0 .1
0

* 0 .2

D a=1 .0 3 x 1 0 -4

0 .1
0
0 .2

D a=4 .5 8 x 1 0 -5

0 .1
0

(b)
1
0.5

Da=1.15x10-3

0.0003
0.0003

0
1

0.0003

0
1

-4

Da=1.03x10

0.0003

0.5

0.5

0
1

0.0003
0.0003

-5

0.0006

0.0006

Da=1.15x10-3

0
1
0.5

10

12

0.0003
0.0006

0.0003
0.0006

0.0006

0.0006

Da=1.03x10

-4

Da=4.58x10

0.5
0
-2

0.0006

0.0003
0.0006

0.5

0
-2

0.0003
0.0006

0.0003
0.0003
0.0003
0.0006
0.0006
0.0006

Da=4.58x10-5

10

12

x*

(c)

(d)

Fig. 4. Effects of the Darcy number on (a) the velocity proles u  y, (b) velocity distribution, (c) expanded uid-phase isotherms (DT = 0.0004 for 0 6 T 6 0.004), (d) solidphase isotherms (DT = 0.0004 for 0 6 T 6 0.004), (e) mean temperature proles T m  y , (f) local temperature distribution along the heat source surfaces, (g) mean Nusselt
number Num, (Num)s and (Num)f, and (h) heat transfer enhancement factor Num/(Num)non at Re = 2000, e = 0.95, Pr = 0.7.

364

C.-C. Chen et al. / International Journal of Heat and Mass Transfer 58 (2013) 356373

Da = 1.15 10 3

Da = 1.03 10 4

Da = 4.58 10 5

0.75
*

0.75
*

0.75
*

0.5

0.5

0.5

0.25

0.25

0.25

0
0

0.001

0.002

0.003
*

0
0

0.004

0.001

0.002

0.003
*

0.004

0
0

0.001

0.002

T
(e)

0.003
*

0.004

T
No.1 heat source
No.2 heat source
No.3 heat source
No.4 heat source

800

0.15

without metal foam

Tf for No.1 heat source


T*f for No.2 heat source
T*f for No.3 heat source
T*f for No.4 heat source
T*s for No.1 heat source
T*s for No.2 heat source
T*s for No.3 heat source
T*s for No.4 heat source

700

: Num=(Num)s+(Num)f
: (Num)s

: (Num)f

(Tw)non0.1
0.05

Num

600

0.01

(Num)s

500

with metal foam

0.003

0.002
*
w

400

-3

Da=1.15x10
-4
Da=1.03x10
-5
Da=4.58x10

0.001

x*

30

(Num)f

20

10

300
40

0 -5
10

10-4

Da

10-3

10-2

(g)

(f)
60

50

Num

40

4
3

(Num)non
30

No.1 heat source


No.2 heat source
No.3 heat source
No.4 heat source

1
2
3
4

2
1

4
3
2
1

4
3
2

4
3
2

4
3
2

20

10

0 -5
10

-4

-3

10

10

-2

10

Da

(h)
Fig. 4. (continued)

can be calculated by the empirical formula Eq. (13), as shown in


Table 1. The effect of variations in the Darcy number is depicted
in Fig. 4 for Re = 2000, df = 0.4 mm, e = 0.95, Pr = 0.7 with Da ? 1
(i.e., K ? 1 for nonporous case), Da = 1.15  103, 1.03  104
and 4.58  105 (i.e., K = 1.65  107 m2 for R = 12 mm, 40 mm,
and 60 mm, respectively). The velocity proles u  y near lower
plate at four monitoring planes x = 0.5, 2.5, 4.5, and 6.5, respectively, (i.e. in the middle of each heat sources) are displayed in
Fig. 4(a) and (b). In the rst nonporous layer (1 qx 5 0), the

prole is identical to that in the nonporous channel. As the uid


progresses through the rst porous layer (0 qx 5 1), hydrodynamic boundary layer thickness decreases, and the velocity distribution develops from a parabolic prole to a at prole. The
magnitude of the axial centerline velocity decreases, and axial
velocity near heater surface increases. The reason for this is the
bulk damping caused by the porous matrix that the viscous effects
are conned to the region near the heater wall. This leads to a at
prole of velocity in the core ow region. When the ow passes

C.-C. Chen et al. / International Journal of Heat and Mass Transfer 58 (2013) 356373

through the second nonporous layer (1 qx 5 2), the hydrodynamic boundary layer thickness lightly increases and at portion
of the velocity prole contracts little toward the centerline of channel because the viscous effects do not penetrate deeply into the
core region. The same process is repeatedly appeared in the following porous/uid layers.
As the Darcy number decreases from 1 to 4.58  105, the
boundary layer thickness decreases, the at portion of the velocity
prole extends gradually toward the heated surfaces, The axial
centerline velocity becomes smaller and velocity near heater surface becomes larger. This is because the smaller values of Da relate
to larger bulk frictional resistance to the core ow inside the porous layer, which in turn causes more uid to ow near the heater
surface regions to satisfy mass continuity, as shown in Fig. 4(b).
Fig. 4(c) and (d) shows the temperature elds in the expanded uid
phase (including the whole uid eld outside and inside
metal-foam porous layers) and solid-matrix phases, respectively,
corresponding to the above ow eld. Here, an intermittent development of the thermal boundary layer is observed in the expanded
uid-phase region due to the discrete heating condition. The isotherms concentrate more near each heater surface and expand to
its downstream. The thermal boundary-layer thickness over each
heat source increases with increasing distance along the channel;
while in the adiabatic portions, the thermal boundary-layer thickness keeps the same, as shown in Fig. 4(c). The penetrating depth
of heat into the solid phase of porous layer increases for downstream porous layer array. The reason is that as the uid passes
over the sources, its temperature increases and the heat exchange
between the solid and uid phases decreases, as shown in Fig. 4(d).
As Da decreases, both the thermal boundary-layer thickness in the
uid phase and heat-penetrating depth in the solid phase decrease
due to the larger solid-to-uid interfacial heat exchange, which
transport more convective energy away from heater through the
solid phase to the uid phase (The detailed illustration is given
in the following paragraphs).
In order to validate the qualitative results of the assumption of
local thermal equilibrium (LTE), the volume-averaged temperature
proles of solid phase are compared quantitatively with those of
the uid phase, both of which are obtained from the two-equation
model. When the temperature distributions of solid phase are
nearly identical to those of the uid phase, the assumption of local
thermal equilibrium becomes more valid. In other words, as the
difference between T s and T f approaches zero, the local thermal
non-equilibrium (LTNE) becomes the local thermal equilibrium
(LTE). For observing the phenomenon of local thermal equilibrium,
Fig. 4(e) displays the dimensionless mean temperature proles
T m  y of solid and uid phases over each uid-saturated metalfoam porous layer mounted on the heaters for different Darcy
number, where the dimensionless mean temperature T m over each
heat source is dened by

R x2
T m y

x1

T  x ; y dx
R x2 
x dx

44

where x1 is 0, 2, 4, and 6, and x2 is 1, 3, 5 and 7 for No. 1, No. 2, No. 3,
and No. 4 heat sources, respectively. It can be seen that as the Darcy
number decreases, the difference temperatures between the solid
and uid phase decrease. The reason is that a decrease in the Darcy
number translates into a decrease in the dimensionless pore diam

eter dp or ber diameter df (from Eqs. (35) and (36)) for xed poros

ity e. As the ber diameter df or pore diameter dp decreases, the
specic surface area asf of the porous media increases (see Eqs.
(31) and (36)), thus increasing the solid-to-uid heat transfer interaction by offering a larger interfacial surface area. The heat exchange between the solid and uid phases becomes more
efcient. Therefore, the local thermal equilibrium (LTE) assumption

365

would be more justiable at low Darcy number. In other words, the


difference between T s and T f becomes less pronounced. Fig. 4(f)
shows the variations of local temperature distribution along each
heat source surface for various values of Da. As seen in Fig. 4(f),
the surface temperature of heaters increases downstream the heat
source array for xed Da. This is because the temperature of the
uid increases as the uid passed over the heaters. Consequently,
the heat transfer from the heater surface to the bulk of the ow decreases as the uid moves downstream. As a result, the local temperature distribution increases progressively for downstream
heater. In addition, the surface temperature of heaters decreases
with decreasing Darcy number due to the larger soliduid interfacial heat exchange occurring on metal foam heat sink for smaller
Darcy number. Comparison of the local temperature distribution
along each heater surface with and without metal-foam porous
layer in Fig. 4(f) shows that each heater surface temperature has a
much signicant improvement by mounting with metal-foam porous heat sink.
To understand the relative magnitudes of the various thermal
transport phenomena, it is useful to examine the solid-and uidphase heat transfer rates, (Num)s and (Num)f, at the heater surfaces.
In Fig. 4(g), the componential variations of the overall mean Nusselt number Num over each heater are plotted. Clearly, heat transfer
rate in the uid-phase component (Num)f accounts for a negligibly
small portion of the total heat rate Num. These can be explained as
follows. The predominant mode of heat transport from the heater
s
surface to the uid is by conduction, qs ks @T
W p (for unit length
@y
in the vertical direction of gure), through the solid phase and then
interfacial heat transfer, qsf = hsfasf(Ts  Tf) HpWp = qs, from the solid
to the uid phase. Since the solid-phase thermal conductivity
(ks = 218 W/mK) is much larger than the uid-phase thermal
conductivity (kf = 0.0263 W/mK) inside the metal foam porous
layers, the heat transferred directly from the heater to the uid,
qf = hf(Tsurf  Tf)Wp, is low. In addition, as Daincreases, the heat
transfer rate (Num)s contributed by porous solid matrix decreases
slightly due to the decreasing of interfacial heat transfer from the
solid to the uid phase, as explained earlier, while the heat transfer
rate (Num)f contributed by uid increases due to the existence of
larger temperature difference between the heater surface and uid.
Summarily, the total heat transfer rate Num = (Num)f + (Num)s decreases slightly with increasing Darcy number, as shown in
Fig. 4(g).
Furthermore, in order to obtain an overall measure of heat
transport characteristics in the present study, the inuence of metal-foam heat sink on the heat transfer enhancement factor Num/
(Num)non, which gives the overall mean Nusselt number Num over
a heat source normalized by the corresponding nonporous-layer
value (Num)non, is calculated. Fig. 4(h) exhibits the effect of Da on
Num/(Num)non. It is clear that the overall mean Nusselt number
Num for the case with metal-foam heat sink is much larger than
that without metal-foam heat sink due to the larger surface velocity for convective heat transport and much larger thermal conductivities ks of solid matrix for conductive heat transport. In addition,
for each heater as Da increases, the gain in Num/(Num)non decreases
because the total heat transfer rate Num decreases with increasing
Darcy number, as discussed earlier. The last heat source has the
largest heat transfer enhancement factors because it has the smallest value of (Num)non.
4.2. Effect of Reynolds number Re
To investigate the effect of Reynolds number on the ow and
temperature elds, computations were carried out at Re = 250,

500, 1000, 1500, and 2000, respectively, with e 0:95; df 0:02;
Da 4:12  104 , and Pr = 0.7. Comparison of the velocity proles
u  y variations in Fig. 5(a) shows that as Re increases, the porous

366

C.-C. Chen et al. / International Journal of Heat and Mass Transfer 58 (2013) 356373
0.1

0.1

Re=250
Re=500
Re=1000
Re=2000

y*
0.05

0.05

at x*=0.5 plane
0
0

0.5

0.1

0.1

0.05

0.05

at x*=2.5 plane
0
0

u*

0.5

at x*=4.5 plane
0
0

u*

0.5

at x*=6.5 plane
0
0

u*

0.5

u*

(a)
Re=250

Re=500

Re=1000

Re=2000

0.75

0.75

0.75

0.75

y*

y*

y*

y*

0.5

0.5

0.5

0.5

0.25

0.25

0.25

0.25

0
0

0.001

0.002

0.003
*

0
0

0.004

0.001

0.002

0
0

0.004

0.3

(T )

0.1

0.002

(b)
600
500

0.003
*

0.004

0
0

0.001

Num
(Num)s

400

300

0.004

0.002

0.003

0.004

T*

No.1 heat source


No.2 heat source
No.3 heat source
No.4 heat source

0.001

without metal foam

*
0.2
w non

0.01

0.003
*

Tf for No.1 heat source


*
Tf for No.2 heat source
T*f for No.3 heat source
T*f for No.4 heat source
T*s for No.1 heat source
T*s for No.2 heat source
T*s for No.3 heat source
T*s for No.4 heat source

: Num=(Num)s+(Num)f
: (Num)s

with metal foam


0.003
*

Tw

0.002

1500

2000

500

1000

2500

(d)

60

No.1 heat source


No.2 heat source
No.3 heat source
No.4 heat source

1
2
3
4

70

Re

80

(Num)non

(c)

Num

(Num)non 10
0

: (Num)f
: (Num)non

(Num)f 20

Re=250
Re=500
Re=1000
Re=2000

0.001

200
30

50

40

4
3
2

4
3

30

4
3
2

4
3
2
1

20
10
0

500

1000

1500

2000

2500

Re

(e)
Fig. 5. Effects of the Reynolds number on (a) the velocity proles u  y, (b) mean temperature proles T m  y , (c) local temperature distribution along the heat source

surfaces, (d) the mean Nusselt number Num, (Num)s and (Num)f, and (e) the heat transfer enhancement factor Num/(Num)non at e 0:95; df 0:02; Da 4:12  104 , and
Pr = 0.7.

matrix has a greater effect on velocity proles, which becomes atter. The hydrodynamic boundary layer thickness decreases, and the
at portion of the velocity prole extends gradually toward the
heater surface in the porous-layer regions. The isotherms in the expanded uid phase and solid phase (not shown) display that at
higher Reynolds number, both the thermal boundary-layer thickness in the uid phase and heat-penetrating depth in the solid
phase decrease because the heat transport away from the heater
surface through the solid phase to the uid phase increases by

the larger interfacial heat transfer coefcient hsf for xed asf. The
difference between the mean temperatures of solid and uid
phases in the porous layers increases with an increase in Reynolds
number, as also shown in Figs. 5(b). This increase should be attributed to the velocity of the ow since the specic surface area asf of
the porous layer is the same in all of these cases. As the velocity of
the ow increases, the time for the solid-to-uid heat interaction
decreases. This will cause a decrease in the efciency of heat exchange between the solid and uid phases and hence the deviation

C.-C. Chen et al. / International Journal of Heat and Mass Transfer 58 (2013) 356373

from the local thermal equilibrium (LTE) will increase. The local
temperature distribution along heater surfaces, shown in
Fig. 5(c), becomes lower for higher Reynolds number. The reason
for this trend is that the larger the Reynolds number, the more
the low temperature core uid passes over heater surfaces. This,
in turn, augments signicantly the transfer of the convective energy away from the heaters. For the same Reynolds number, as expected, the surface temperature of heater increases downstream
the heat source array.
The effect of Re on Num over each heat source are exhibited in
Fig. 5(d). It can be seen that as Re increases, the heat transfer rate
(Num)f contributed by uid increases due to a increasing in convective heat transport caused by larger heater surface velocity, and the
heat transfer rate (Num)s contributed by porous solid matrix increases slightly due to a increasing in interfacial heat transfer from
the solid to the uid phase (see Eq. (11) or Eq. (33)). Again, since
the heat transferred directly to the uid from the heated wall is
negligible compared to the heat transferred interfacially from the
solid to the uid, the total heat transfer rate Num increases slightly
with increasing Re. Fig. 5(e) exhibits how Num/(Num)non changes
with Re along the heater array. It is evident that as Re increases
the gain in Num/(Num)non decreases for each heater owing to the
amount of increasement in (Num)non is larger than that in Num with
increasing Re.
4.3. Effect of porosity e
The effect of an increase or decrease in the porosity e is shown

in Fig. 6 for Re = 2000 and df 0:02 with e = 0.85, 0.9, and 0.95. As
seen from the Fig. 6(a) that the porosity has a much less pronounced inuence on the velocity proles inside the porous region.
This is because the porosity effect is not sensitive to the porous
permeability K or Darcy number Da (see Table 1). Comparison of
the corresponding isotherms in the expanded uid and solid
phases portraits that both the thermal boundary-layer thickness
in the uid phase and heat-penetrating depth in the solid phase increase as e increases (not shown). That is due to less solid matrix
existing in the metal foam at higher porosity, which results in a
diminish activity in the conductive thermal energy from heater
to the solid matrix and nally to the uid by interfacial heat convection. Fig. 6(b) gives the mean temperature proles of the solid
and uid phases in the porous layers. It can be seen from
Fig. 6(b) that the mean temperature difference between solid and
uid phases becomes larger at higher porosity. The reason is that
an increase in the porosity e relate to an increase in the pore diameter dp (from Eq. (14)) and a decrease in the specic surface area asf

(from Eq. (10)) for xed ber diameter df , as displayed in Table 1.
This will cause a decrease in the uid-to-solid heat transfer interaction and hence the local thermal non-equilibrium (LTNE)
assumption becomes more pronounced. Fig. 6(c) demonstrates
the local temperature distribution along heater surfaces. It is found
that the local temperature distribution along each heater surface
becomes higher at higher porosity. The reason for this trend is
the principal heat transport through the solid phase by conduction
from the heater surface to the uid becomes less signicant at
higher porosity. Fig. 6(d) displays the variation of overall mean
Nusselt number with e. Again, compared to the solid-phase heat
transfer rate component (Num)s, the uid-phase heat transfer rate
component (Num)f only is a negligible small portion of the total
heat rate Num for each heat source. Further, this heat transfer rate
component (Num)f increases as porosity increases due to the larger
temperature gradient near the heater surface for high porosity
cases. While the heat transfer rate component (Num)s decreases
with increasing porosity due to the smaller percentage solid matrix
components for conduction at high porosity cases. As expected, the
total mean Nusselt number of each heater decreases as porosity in-

367

creases, as shown in Fig. 6(e). In addition, the results indicate also


that Num generated by the lower porosity open-cell aluminum
foam (e = 0.85) is about two times larger than that by the higher
porosity open-cell aluminum foam (e = 0.95). Moreover, for each
heater, the gain in Num/(Num)non decreases with increasing e because Num decreases with increasing e, as shown in Fig. 6(e).
4.4. Effect of pore density (PPI)
Fig. 7 illustrate the ow and thermal elds inside metal foam
samples with different pore densities (PPI = 5, 20, and 40) from
Bhattacharya et al. [28] at Re = 2000, e = 0.95. The properties of
these metal foam samples are shown in Table 2. The velocity proles, depicted in the Fig. 7(a), show that inside the porous layers
the magnitude of the axial centerline velocity slightly decreases
and axial velocity near heater surface increases with increasing
pore density. It is owing to the fact that an increase in the pore
density translates into a decrease in the Darcy number or permeability, which leads to more uid owing near the heater surface
regions as illustrated earlier. The thermal penetration depth into
both the uid and solid phases decrease as pore density increases
(not shown) due to the higher heat transport away from the heater
surfaces through the solid phase to the uid phase caused by larger
uid-to-solid interfacial convective heat transfer for xed e and keff.
The difference between the solid-and uid-phase mean temperatures in the metal foam porous layer decreases as pore density increases, as shown in Fig. 7(b). Since an increase in pore density
(PPI) is characterized by a decrease in ow passage width (dp, pore
diameter), this will lead to an increase in the specic interfacial
surface area asf, which in turns enhance the uid-to-solid heat
transfer interaction. Because of the efcient solid-to-uid interfacial heat transfer, the uid temperature is close to the solid temperature and hence the local thermal non-equilibrium (LTNE)
assumption becomes very erroneous for higher pore-density ow.
The local temperature distribution along each heater surface
shown in Fig. 7(c) becomes lower at higher pore density because
of the larger heat transport from the heater surface to the uid
caused by the larger solid-to-uid interfacial heat transfer. The effect of PPI on Num is show in Fig. 8f. Again, the heat transferred directly to the uid phase from the heater surface is negligible
compared to the heat transferred interfacially from the solid phase
to the uid. As seen from Fig. 7(d) that the value of (Num)s increases
with increasing pore density due to an increasing in interfacial heat
transfer from the solid to the uid phase. While the value of (Num)f
decreases as pore density increases due to a decreasing in the uid
temperature gradient at the heater surfaces. As expected, the total
mean Nusselt number Num of each heater increases as pore density
increases, as shown in Fig. 7(d). In addition, the increases in Num
generated by the higher pore density open-cell aluminum foam
(40 PPI) is about 12% larger than that by the lower pore density
open-cell aluminum foam (5 PPI) for the rst heat source and
about 5% for other heat sources. Fig. 7(e) gives the results for heat
transfer enhancement factor vs. pore densities. It can be seen that
as pore density increases the gain in Num/(Num)non increases
slightly for each heater.


4.5. Effect of ber diameter df

Based on Eq. (14) for the xed porosity e, the smaller the ber
diameter, the smaller the pore diameter. In other words, the metal
foam with higher pore density has the smaller ber diameter, as
shown in Table 2. Therefore, a decrease in the ber diameter has
the same qualitative effect as an increase in the pore density on
the ow and temperature elds. Fig. 8 displays the effect of ber
diameter on the ow and thermal elds for e = 0.95, Re = 2000 with

df 0:01; 0:015; 0:02 and 0.025 (i.e., df = 0.2, 0.3, 0.4, and 0.5 mm

368

C.-C. Chen et al. / International Journal of Heat and Mass Transfer 58 (2013) 356373

0.1

=0.85
=0.9
=0.95

y*
0.05

0.1

0.1

0.1

0.05

0.05

0.05

at x*=0.5 plane
0
0

0.5

at x*=2.5 plane
0
0

u*

0.5

at x*=4.5 plane
0
0

u*

0.5

at x*=6.5 plane
0
0

u*

0.5

u*

(a)
=0.85

=0.9

=0.95

0.75
*

0.75
*

0.75
*

0.5

0.5

0.5

0.25

0.25

0.25

0
0

0.001

0.002

0.003
*

0
0

0.004

Tf for No.1 heat source


*
Tf for No.2 heat source
T*f for No.3 heat source
T*f for No.4 heat source
T*s for No.1 heat source
T*s for No.2 heat source
T*s for No.3 heat source
T*s for No.4 heat source

0.001

0.002

0.003
*

0
0

0.004

0.001

0.002

0.003
*

0.004

(b)
0.15

1200

without metal foam

*
0.1
w non

(T )

Num
600

0.01

: (Num)s

: (Num)f

300

30

=0.85
=0.9
=0.95

: Num=(Num)s+(Num)f

0
40

*
w

0.001

(Num)s

with metal foam

0.003

0.002

900

0.05

No.1 heat source


No.2 heat source
No.3 heat source
No.4 heat source

(Num)f

10

0.85

0.9

0.95

0
0.8

20

(d)

(c)
80

1
2
3
4

70
3

60

Num

No.1 heat source


No.2 heat source
No.3 heat source
No.4 heat source

3
4

50

(Num)non

40

4
3
2
1

30

4
3
2
1

20

4
3
2
1

10
0
0.8

0.85

0.9

0.95

(e)
Fig. 6. Effects of the porosity on (a) the velocity proles u  y, (b) mean temperature proles T m  y , (c) local temperature distribution along the heat source surfaces, (d)

the mean Nusselt number Num, (Num)s and (Num)f, and (e) the heat transfer enhancement factor Num/(Num)non at Re = 2000, df 0:02, Pr = 0.7.

based on R = 20 mm), respectively. Comparison of the velocity proles in Fig. 8(a) shows that as ber diameter decreases, the magnitude of the axial centerline velocity slightly decreases, and axial

velocity near heater surface increases in the porous layers. The


larger the ber diameter, the thicker the thermal penetrating
depths for both uid and solid phases (not shown). The mean

369

C.-C. Chen et al. / International Journal of Heat and Mass Transfer 58 (2013) 356373

0.1

5PPI
20PPI
40PPI

y*
0.05

0.1

0.1

0.1

0.05

0.05

0.05

at x*=0.5 plane
0
0

0.5

at x*=2.5 plane
0
0

u*

0.5

at x*=4.5 plane
0
0

u*

0.5

at x*=6.5 plane
0
0

u*

0.5

u*

(a)
5 PPI

20 PPI

40 PPI

0.75
*

0.75
*

0.75
*

0.5

0.5

0.5

0.25

0.25

0.25

0
0

0.001

0.002

0.003
*

0
0

0.004

Tf for No.1 heat source


*
Tf for No.2 heat source
T*f for No.3 heat source
T*f for No.4 heat source
T*s for No.1 heat source
T*s for No.2 heat source
T*s for No.3 heat source
T*s for No.4 heat source

0.001

0.002

0.003
*

0
0

0.004

0.001

0.002

0.003
*

0.004

(b)
No.1 heat source
No.2 heat source
No.3 heat source
No.4 heat source

600

0.15

without metal foam

(Tw)non0.1
500

0.05

Num

0.01

0.002

T*w

400

10

20

300
40

5PPI
10PPI
20PPI
40PPI

0.001

: (Num)s

: (Num)f

(Num)s

with metal foam

0.003

: Num=(Num)s+(Num)f

30

(Num)f

20

10

x*

(e)

PPI

30

40

50

(f)
50

1
2
3
4

40

Num

30

(Num)non

4
3
2

3
2

3
2

10

20

No.1 heat source


No.2 heat source
No.3 heat source
No.4 heat source
4
3
2
1

20

10

30

40

50

PPI

(g)
Fig. 7. Effects of the pore density on (a) the velocity proles u  y, (b) mean temperature proles T m  y , (c) local temperature distribution along the heat source surfaces,
(d) the mean Nusselt number Num, (Num)s and (Num)f, and (e) the heat transfer enhancement factor Num/(Num)non at Re = 2000, e = 0.95, Pr = 0.7.

temperatures difference between the solid and uid phases, shown



in Fig. 8(b), increases with increasing df , and hence the local ther-

mal non-equilibrium (LTNE) assumption becomes more pronounced for larger ber diameter. The local temperature

370

C.-C. Chen et al. / International Journal of Heat and Mass Transfer 58 (2013) 356373

Table 2
Properties for different pore densities of aluminum-alloy T-6201 [28].
PPI

df (mm)

dp (mm)

R (mm)

K (m2)

Da

asf (m1)

keff (W/m.DK)

Rk

5
10
20
40

0.95
0.95
0.95
0.95

0.47
0.37
0.32
0.24

3.90
3.10
2.70
1.98

20
20
20
20

2.17E07
1.49E07
1.185E07
5.62E08

5.43E04
3.73E04
2.96E04
1.41E04

0.099
0.099
0.100
0.0976

597
744
848
1183

4.04
4.04
4.04
4.04

153.53
153.53
153.53
153.53

0.1

d*f =0.01
*
df =0.015
d*f =0.02

y*
0.05

0.1

0.1

0.1

0.05

0.05

0.05

at x*=0.5 plane
0
0

0.5

at x*=2.5 plane
0
0

u*

0.5

at x*=4.5 plane
0
0

u*

0.5

at x*=6.5 plane
0
0

u*

0.5

u*

(a)
*

d f =0.01

*
d f =0.02

d f =0.015

*
d f =0.025

0.75

0.75

0.75

0.75

y*

y*

y*

y*

0.5

0.5

0.5

0.5

0.25

0.25

0.25

0.25

0
0

0.001

0.002

0.003
*

0
0

0.004

0.001

0.002

0.003
*

0
0

0.004

0.001

0.002

0.003
*

(b)
(Tw )non0.1

0.003

0.004

: Num=(Num)s+(Num)f
: (Num)s

: (Num)f

500

Num

0.01

(Num)s

with metal foam

0.003

400

300
40

d*f =0.01
*
df =0.015
*
df =0.02
*
df =0.025

0.001

0.002

0.05

0.002

0.001

T*

No.1 heat source


No.2 heat source
No.3 heat source
No.4 heat source

600

without metal foam

T*w

0
0

0.004

0.15
*

Tf for No.1 heat source


*
Tf for No.2 heat source
T*f for No.3 heat source
T*f for No.4 heat source
T*s for No.1 heat source
T*s for No.2 heat source
T*s for No.3 heat source
T*s for No.4 heat source

30

(Num)f

20

10

0.01

0.015

0.02

0.025

0.005

(c)

df

0.03

(d)
50

No.1 heat source


No.2 heat source
No.3 heat source
No.4 heat source

1
2
3
4

40
4

Num

(Num)non30

2
1

4
3
2
1

4
3
2

4
3
2

0.4

0.5

20

10

0.1

0.2

0.3

0.6

df (mm)

(e)
Fig. 8. Effects of the ber diameter on (a) the velocity proles u  y, (b) mean temperature proles T m  y , (c) local temperature distribution along the heat source surfaces,
(d) the mean Nusselt number Num, (Num)s and (Num)f, and (e) the heat transfer enhancement factor at Re = 2000, e = 0.95, Pr = 0.7.

distribution along heater surfaces becomes lower for smaller ber



diameter, as illustrated in Fig. 8(c). The effect of df on Num is show

in Fig. 8(d). As seen that the value of (Num)s decreases with increas
ing df due to a decreasing in interfacial heat transfer from the solid

371

C.-C. Chen et al. / International Journal of Heat and Mass Transfer 58 (2013) 356373


to the uid phase. While the value of (Num)f slightly increases as df


increases due to an increasing in the uid temperature gradient at

the heater surfaces for larger df . As expected, the total mean Nus
selt number Num of each heater decreases as df increases, as shown
in Fig. 8(d). In addition, when ber diameter is increased from
0.025 to 0.001, the increases in Num is about 16% for the rst heat
source and about 8% for other heat source. Again, as ber diameter
increases the gain in Num/(Num)non decreases for each heater, as
displayed in Fig. 8(e).

and effective thermal conductivity ratio keff for different metal



foam materials are shown in Table 3. Since Re; e; df , and Da are
xed, the variation of the effective conductivity ratio has no effect
on the ow eld, and therefore the velocity proles u  y are the
same, as in Fig. 5(a) at Re = 2000 for all effective conductivity ratios. The expanded uid-phase isotherms and solid-phase isotherms are shown in Fig. 9(a) and (b). As seen from the Fig. 9(a)
that decreasing effective thermal conductivity ratio from 276.01
to 49.94 results in an increases in both thermal boundary-layer
thickness in the uid phase and heat-penetrating depth in the solid
phase with high temperature gradient distribution. The reason is
that the higher thermal resistance weakens the heat transfer from
the solid-matrix surfaces by allowing less heat ow conducted to
the uid phase at lower conducting metal foam material. While keff
is decreased further to 6.81 (for RVC material), an opposite tendency is found. Both thermal boundary-layer thicknesses in the
uid phase and heat-penetrating depth in the solid phase decrease
with higher temperature gradient distribution. This can be explained by noting that the heat transfer direct from the heater to
the uid qf is larger than the interfacial heat transfer qsf from the

4.6. Effects of the effective thermal conductivity ratio keff


The effect of the effective thermal conductivity ratio keff = keff/kf
is shown in Fig. 9 for different commercial foam materials with

Re 2000; e 0:95; df 0:02; Da 4:12  104 , and Pr = 0.7.
These foam materials investigated here include Cu (copper
ks = 393 W/mK, and keff = 276.01), Al (aluminum ks = 170 W/mK
and keff = 119.93), Ni (nickel ks = 70 W/mK and keff = 49.94) and
RVC (Reticulated Vitreous Carbon, ks = 8.37 W/mK and keff = 6.81)
[20,35,36]. The values of the effective thermal conductivity keff

Cu

0.0005

Cu

0.0005

Al

0.001

0
1

0.0005

0
1

Al

0.001

0.5

0.0005
0.0005

0.5

0.5

0
1

0.0005
0.001

Ni

0
1

Ni

0.0005

0.5

0.5

0
1

0
1

RVC

0.0005

RVC
0.0005

0.5

0.0005

0.5

0.0005

0.5

0.0005

0
-2

10

0
-2

12

(a)

(b)
Al

0.75

0.75

0.75

y*

y*

y*

0.5

0.5

0.5

0.5

0.25

0.25

0.25

0.25

0
0

0.008

0.002

0.004

0.006
*

10

without metal foam


RVC
Ni
with metal foam
Al
Cu

300

T*w
10-2

0
40

-3

10

(d)

(Num)f

: (Num)s

: (Num)f

No.1 Heat source


No.2 Heat source
No.3 Heat source
No.4 Heat source

1
2
3
4

50

0.02

4
3
2

40

1
Cu

30

4
3
2
1
Al

20

100

150

200

250

4
3
2
1
Ni

10

0
0

0.015

(Num)non

0.01

T*

50

Num

20
10

0.005

60

: Num=(Num)s+(Num)f

0
0

0.008

30

0.006
*

600

(Num)s

-1

0.004

(c)
No.1 heat source
No.2 heat source
No.3 heat source
No.4 heat source

900

0.002

Num

10

0
0

0.008

0.006
*

Tf for No.1 heat source


*
Tf for No.2 heat source
T*f for No.3 heat source
T*f for No.4 heat source
T*s for No.1 heat source
T*s for No.2 heat source
T*s for No.3 heat source
T*s for No.4 heat source

0.75

0.004

12

RVC

y*

0.002

10

Ni

0
0

Cu
1

x*

300

4
3
1
2
RVC

50

100

150

keff/kf

keff/kf

(e)

(f)

200

250

300

Fig. 9. Effects of the effective conductivity ratio for different solid matrixes on the variations of (a) expanded uid-phase isotherms (DT = 0.0005), (b) solid-phase isotherms
(DT = 0.0005), (c) mean temperature proles T m  y , (d) local temperature distribution, (e) mean Nusselt number Num, (Num)s and (Num)f, and (f) heat transfer enhancement

factor Num/(Num)non at Re = 2000, e 0:95; df 0:02; Da 4:12  104 , and Pr = 0.7.

372

C.-C. Chen et al. / International Journal of Heat and Mass Transfer 58 (2013) 356373

Table 3
Properties for different metal foam materials.

Material
Cu
Al
Ni
RVC

df (mm)

0.95
0.95
0.95
0.95

dp (mm)

0.4
0.4
0.4
0.4

Da
4

3.32
3.32
3.32
3.32

4.12  10
4.12  104
4.12  104
4.12  104

0
-10

ks (W/m.DK)

keff (W/m.DK)

Rk

0.099
0.099
0.099
0.099

393
170
70
8.37

7.26
3.15
1.31
0.18

276.01
119.93
49.94
6.81

Da=
(without metal foam)

Re=2000(without metal foam)


Re=1000(without metal foam)
Re=250(without metal foam)

Da=1.15x10

-20

Re=2000(with metal foam)

=1
(without metal foam)

-3

-5

-20

-40

-40

-60

10

-100

12

-15

-25

12

=0.95
=0.9
=0.85
0

without metal foam

df =0.025
*

P-P0

d*f =0.015

12

df =0.02

-20

20PPI

-30

10

without metal foam

-10

5PPI
10PPI

-20

(c)

-10

x*

(b)

u20

10

-10

x*

(a)

P-P0

2
0

-20

Da=4.58x10-5

Re=250(with metal foam)


0

P-P0

Da=1.03x10-4

-80

-50
-60

2
0

-30

u20

P-P0

Re=1000(with metal foam)

P-P0

u20

-30
-40

-40

40PPI
*

-60

df =0.01

-50

-50

10

12

-60

(d)

10

12

(e)


Fig. 10. The variations of pressure drop factor along the upper plate for various values of (a) Re, (b) Da, (c) e, (d) PPI, and (e) df .

solid phase to the uid phase. Therefore, the depth of heat penetrating into the solid and uid phases inside the metal foam porous
layer is dominated by qf. As expected, the mean temperature difference between solid phase and uid phase becomes larger at lower
keff, as shown in Fig. 9(c), and hence the local thermal nonequilibrium (LTNE) assumption becomes more viable. As seen in
Fig. 9(d), the local temperature distribution along the heat source
surfaces decreases with increasing effective conductivity ratio keff
due to the larger conduction heat loss from the heater surface
through the solid phase at higher keff. Fig. 9(e) gives the results
for the mean Nusselt numbers. It is observed that (Num)s is increased signicantly with increasing keff as above-mentioned, but
(Num)f decreases due to the smaller temperature gradient near
each heater surface. The total mean Nusselt number Num of each
heater increases as keff increases, as shown in Fig. 9(e), since (Num)s
is much larger than (Num)f. Again, the gains in Num/(Num)non increase signicantly with an increase in Rk, as shown in Fig. 9(f).
4.7. Pressure drop calculation
When using a metal foam material for augmenting heat transfer,
an important factor to consider is the penalty arising from increased pressure drop (based on the DarcyBrinkmanForchheimer
equations Eq. (2)). In the stream functionvorticity formulation, the
pressure eld is eliminated in obtaining the solution. However, the

pressure eld can be recovered from the converged stream function


and vorticity elds. This is done by integrating the pressure gradient along the upper channel wall. The pressure gradient is derived
from the momentum equation using the no-slip boundary conditions on the solid wall. The total pressure drop DP along the upper
channel wall is then obtained from (Huang et al. [11])


DP

Z
0

Lt





Z Lt 


 @P 
1 @f
 
 


dx
dx

 


 @x y 1 
Re @y y 1 
0 

45

where pressure P is nondimensionalized with respect to qu2o . The



effects of Re, Da, e, PPI, and df on the pressure distribution along
the upper plate wall for a channel with and without the porous layers is presented in Fig. 10. In this partially full porous channel, the
pressure drop in the non-porous layers is negligible compared with
that of the porous layers, and the pressure varies linearly with axial
distance in the porous layers. In Fig. 10(c), the pressure drop is less

affected by a change in e (at Re 2000; df 0:02; Da 4:12  104
and keff = 153.53). However, pressure drop increases substantially as
PPI (at Re = 2000, e = 0.95, and keff = 153.53) increases, and or as Da

(at Re 2000; e 0:95; df 0:02, and keff = 153.53), Re (at
e 0:95; df 0:02; Da 4:12  104 , and keff = 153.53) or df (at
Re = 2000, e = 0.95, Da = 4.12  104, and keff = 153.53) decreases,
as seen in Fig. 10(a), (b), (d) and (e). A decrease in Da provides a
larger bulk frictional resistance to ow, which results in a profound

increase in the pressure drop. A decrease in df or increase in PPI also

C.-C. Chen et al. / International Journal of Heat and Mass Transfer 58 (2013) 356373

leads to a larger pressure loss due to a smaller passage and larger


specic surface, inducing larger resistance to ow as stated previously. In addition, an increase in Re causes a decrease in pressure
drops because the viscous effects do not penetrate deeply into the
core ow in the porous layers, as explained earlier. As expected,
the total pressure drop through the channel for partially or full porous case is always larger than for non-metal-foam case. Therefore,
the required pumping power to maintain a ow increases with
increasing Reynolds number and pore density, but with decreasing
Darcy number, ber diameter, and porosity.
5. Conclusions
This paper performs a numerical study of forced convection in a
partial-full metal-foam porous channel with discrete heat sources
on the bottom wall. Due to the complexity of foam microstructure,
governing equations with the empirical correlations and local thermal non-equilibrium (LTNE) assumption have been employed to
analyze the enhanced heat transfer in high porosity metal foam.
The major conclusions can be drawn as follows:
1. The less the soliduid interfacial heat transfer rate is, the
stronger the phenomena of local thermal non-equilibrium
(LTNE) exhibits, and the smaller the convective cooling of heater is. On the contrary, the more the soliduid interfacial heat
transfer rate is, the more pronounced the phenomena of local
thermal equilibrium (LTE) displays, and the better the convective cooling of heater is
2. For aluminum-alloy metal foam heat sink, the phenomenon of
local thermal equilibrium (LTE) becomes more obvious at lower
Darcy number, porosity, and ber diameter, or at higher pore
density and effective thermal conduction ratio.
3. Comparison of local temperature distribution along the heat
source surfaces with and without metal foam heat sink clearly
shows that signicant cooling enhancement of the heaters can
be achieved through the mounting of metal foam porous layer.
4. The mean Nusselt number Num generated by the lower porosity
open-cell aluminum foam (e = 0.85) are about two times larger
than the higher porosity open-cell aluminum foam (e = 0.95).
The increases in Num generated by the higher pore density
open-cell aluminum foam (40 PPI) from the lower pore density
open-cell aluminum foam (5 PPI) is about 12% for the rst heat
source and about 5% for other heat source. When decreasing
ber diameter from 0.5 mm to 0.2 mm, the increases in Num is
about 16% for the rst heat source and about 8% for other heat
source.
5. The enhancement in heat transfer always accompanies with
increase in pressure drop as a penalty. This pressure drop is
higher at the higher pore density, lower permeable, smaller
porosity, or ber diameter of metal foams.
Acknowledgements
The second author grateful acknowledges the support of the National Science Council of the Republic of China under contacts NSC
99-2221-E-027-076 and 99-ET-E-027-004-ET.
References
[1] A.D. Kraus, A. Barcohen, Thermal Analysis and Control of Electronic Equipment,
McGraw Hill, 1983.
[2] S.J. Kim, S.W. Lee, Air Cooling Technology for Electronic Equipment, CRC press,
New York, 1996.

373

[3] M. Kaviany, Principles of Heat Transfer in Porous Media, Springer-Verlag, New


York, 1991.
[4] J.C.Y. Koh, R. Colony, Analysis of cooling effectiveness for porous material in a
coolant passage, ASME J. Heat Transfer 96 (1974) 324330.
[5] M. Kaviany, Laminar ow through a porous channel bounded by isothermal
parallel plate, Int. J. Heat Mass Transfer 28 (1985) 851858.
[6] P.C. Huang, K. Vafai, Analysis of forced convection enhancement in a parallel
plate using porous blocks, AIAA J. Thermophys. Heat Transfer 8 (1994) 563
573.
[7] D. Angirasa, Forced convective heat transfer in metallic brous materials,
ASME J. Heat Transfer 124 (2002) 739745.
[8] H.A. Hadim, A. Bethancourt, Numerical study of forced convection in a partially
porous channel with discrete heat sources, ASME J. Heat Transfer 8 (1995)
465472.
[9] D. Angirasa, G.P. Peterson, Forced convection heat transfer augmentation in a
channel with a localized heat source using metal brous material, ASME J.
Electron. Packag. 121 (1999) 17.
[10] W.S. Fu, H.C. Huang, W.Y. Liou, Thermal enhancement in laminar channel ow
with a porous block, Int. J. Heat Mass Transfer 39 (1996) 21652175.
[11] P.C. Huang, C.F. Yang, J.J. Hwang, M.T. Chiu, Enhancement of forced-convection
cooling of multiple heated blocks in a channel using porous covers, Int. J. Heat
Mass Transfer 48 (2005) 647664.
[12] K. Boomsma, D. Poulikakos, F. Zwick, Metal foams as compact high
performance heat exchangers, Mech. Mater. 35 (2003) 11611176.
[13] S.Y. Kim, J.W. Paek, B.H. Kang, Flow and heat transfer correlations for porous
n in a plate-n heat exchanger, ASME J. Heat Transfer 45 (2000) 572578.
[14] S.Y. Kim, J.W. Peak, B.H. Kang, Thermal performance of aluminum-foam heat
sinks by forced air cooling, IEEE Trans. Compon. Packag. Technol. 26 (2003)
262267.
[15] K.C. Leong, L.W. Jin, An experimental study of heat transfer in oscillating ow
through a channel lled with an aluminum foam, Int. J. Heat Mass Transfer 48
(2005) 243253.
[16] K.C. Leong, L.W. Jin, Characteristics of oscillating ow through a channel lled
with open cell metal foam, Int. J. Heat Fluid Flow 27 (2006) 144153.
[17] V.V. Calmidi, R.L. Mahajan, Forced convection in high porosity metal foams,
ASME J. Heat Transfer 122 (2000) 557565.
[18] S.C. Tzeng, Convective heat transfer in a rectangular channel lled with
sintered bronze beads and periodically spaced heated blocks, ASME J. Heat
Transfer 128 (2006) 453464.
[19] M.S. Phanikumar, R.L. Mahajan, Non-Darcy natural convection in high Porosity
metal foams, Int. J. Heat Mass Transfer 45 (2002) 37813793.
[20] M.L. Hunt, C.L. Tien, Effects of thermal dispersion on forced convection in
brous media, Int. J Heat Mass Transfer 31 (1988) 301308.
[21] K. Vafai, C.L. Tien, Boundary and inertia effects on ow and heat transfer in
porous media, Int. J. Heat Mass Transfer 24 (1980) 195203.
[22] A. Amiri, K. Vafai, Analysis of dispersion effects and non-thermal equilibrium,
non-Darcian, variable porosity incompressible ow through porous media, Int.
J. Heat Mass Transfer 37 (1994) 939954.
[23] B. Alazmi, K. Vafai, Constant wall heat ux boundary conditions in porous
media under local thermal non-equilibrium conditions, Int. J. Heat Mass
Transfer 45 (2002) 30713087.
[24] V.V. Calmidi, R.L. Mahajan, The effective thermal conductivity of high porosity
metal foams, ASME J. Heat Transfer 121 (1999) 466471.
[25] D.L. Koch, J.F. Brady, The effective diffusivity of brous media, AIChE J 32
(1986) 575591.
[26] A. Zhukauskas, Heat Transfer from Tubes in Cross Flow, in: J.P. Hartnett, T.F.
Irvine Jr. (Eds.), Advances in Heat Transfer, vol. 8, Academic Press, New York,
1972.
[27] V.V. Calmidi, Transport phenomena in high porosity metal foams, PhD. Thesis,
University of Colorado, Boulder, CO, 1998.
[28] A. Bhattacharya, V.V. Calmidi, R.L. Mahajan, Thermophysical properties of high
porosity metal foams, Int. J. Heat Mass Transfer 45 (2002) 10171031.
[29] W.E. Azzi, A systematic study on the mechanical and thermal properties of
open cell metal foams for aerospace applications, Master Thesis, North
Carolina State University, 2004.
[30] T.S. Lundgren, Slow ow through stationary random beds and suspensions of
spheres, ASME J. Fluid Mech. 51 (1972) 273299.
[31] S.V. Patankar, Numerical Heat Transfer and Fluid Flow, McGraw-Hill, New
York, 1980.
[32] G.O. Roberts, Computational Meshes for Boundary Layer Problems, in:
Proceedings of Second International Conference on Numerical Methods Fluid
Dynamics, Lecture Notes in Physics, vol. 8, Springer-Verlag, New York, 1971,
pp. 171177.
[33] H.J. Sung, S.Y. Kim, J.M. Hyun, Forced convection from an isolated heat source
in a channel with porous medium, Int. J. Heat Fluid Flow 16 (1995) 527535.
[34] S.Y. Kim, B.H. Kang, Forced convection heat transfer from two heated blocks in
pulsating channel ow, Int. J. Heat Mass Flow 41 (1998) 625634.
[35] K.C. Leong, L.W. Jin, Heat transfer of oscillating and steady ows in a channel
lled with porous media, Int. Commun. Heat transfer 31 (2004) 6372.
[36] H.L. Fu, K.C. Leong, X.Y. Huang, C.Y. Liu, An experimental study of heat transfer
of a porous channel subjected to oscillating ow, ASME J. Heat Transfer 123
(2001) 162170.

Potrebbero piacerti anche