Renewable Energy 30 (2005) 1557–1572

www.elsevier.com/locate/renene

The effects of porosity distribution variation

on PEM fuel cell performance
R. Roshandela,*, B. Farhanieha, E. Saievar-Iranizadb
a
School of Mechanical Engineering, Sharif University of Technology, Tehran, Iran
b
Department of Physics, Faculty of Science, University of Tarbiat Modarres, Tehran, Iran
Received 2 May 2004; accepted 22 November 2004
Available online 17 March 2005

Abstract
Gas diffusion layers (GDL) are one of the important parts of the PEM fuel cell as they serve
to transport the reactant gases to the catalyst layer. Porosity of this layer has a large effect on
the PEM fuel cell performance. The spatial variation in porosity arises due to two effects: (1)
compression of the electrode on the solid landing areas and (2) water produced at the cathode
side of gas diffusion layers. Both of these factors change the porosity of gas diffusion layers
and affect the fuel cell performance. To implement this performance analysis, a mathematical
model which considers oxygen and hydrogen mass fraction in gas diffusion layer and the
electrical current density in the catalyst layer, and the fuel cell potentials are investigated. The
porosity variation in the GDL is calculated by considering the applied pressure and the amount
of the water generated in the cell. The validity of the model is approved by comparing the
computed results with experimental data. The obtained results show that the decrease in the
average porosity causes the reduction in oxygen consumption, so that a lower electrical current
density is generated. It is also shown that when the electrical current density is low, the
porosity variation in gas diffusion layer has no significant influence on the level of polarization
whereas at higher current density the influence is very significant. The porosity variation causes
non-uniformity in the mass transport which in turn reduces the current density and a lower fuel
cell performance is obtained.
q 2005 Elsevier Ltd. All rights reserved.

* Corresponding author.

0960-1481/$ - see front matter q 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.renene.2004.11.017
1558 R. Roshandel et al. / Renewable Energy 30 (2005) 1557–1572

Nomenclature
A electrode surface area (cm2)
C concentration (mol/cm3)
C channel concentration (mol/cm3)
CHC fixed charge concentration (mol/cm3)
F faraday constant (C)
I electrical current density (A/cm2)
I0 exchange electrical current density (A/cm2)
Ilim limiting current density (A/cm2)
k membrane ionic conductivity (1/U cm)
k4 membrane electrokinetic permeability (cm2)
kp membrane hydraulic permeability (cm2)
lm membrane thickness (cm)
MK water balance in anode (mol/s)
MC water balance in cathode (mol/s)
M1,Max water amount corresponding to complete flooding (mol/s)
P total pressure (atm)
pH 2 hydrogen pressure (atm)
pO 2 oxygen pressure (atm)
psat
w saturation vapor pressure (atm)
R gas constant (J/mol K)
rH catalyst transfer rate
T temperature (K)
U reactant velocity in GDL (cm/s)
v water velocity in membrane (cm/s)
V operational voltage (V)
VOCV open circuit voltage (V)
wel water produced in cathode (mol/cm2 s)
wtr water transport in membrane (mol/cm2 s)
whu water entering the cell due to reactant humidification (mol/cm2 s)
wout water outgoing from the cell (mol/cm2 s)
x mole fraction
hact activation overpotential (V)
hohm ohmic overpotential (V)
hdif diffusion overpotential (V)
hconv convection overpotential (V)
d water density in membrane pores (kg/cm3)
4 electrical potential (V)
m viscosity (kg/m s)
r density
tR membrane resistance ratio
z stoichiometric ratio
x membrane water transport ratio
 R. Roshandel et al. / Renewable Energy 30 (2005) 1557–1572 1559

3m
w membrane water porosity
30 initial porosity
3com porosity corresponding to compression pressure
3hum porosity corresponding to humidification condition

1. Introduction

Fuel cells convert the chemical energy of hydrogen in the presence of oxygen into
electricity. Their high efficiency and low emissions have made them a very suitable power
system for next generation of vehicular application as well as small stationary power
plants. Fuel cells are usually classified by the nature of electrolyte they use. Polymer
electrolyte membrane (PEM) fuel cell systems are considered to be promising and
environmentally friendly power source. This is due to the attributes of high energy density
at low operating temperature, good performance in intermittent operation, zero emission,
quick start up capacity, and minimal problems from components corrosion and electrolyte
leakage [1–5].
The properties of the diffusion layer effects the optimum performance of the catalyst
and electrode [6]. These layers are porous to allow for distribution of the gases to
unexposed areas of the flow channel and for complete utilization of the electrode area. The
electrical conductivity of these layers could affect the transport of electrons to the current
collector from the electrode.
The development of a theoretical model of the PEM fuel cell as well as corresponding
analyses, are crucial to gain a good understanding of the effect of the operating conditions
on the cell potential. Most of the studies in this area has assumed for simplicity constant
porosity of the GDL [7–16]. This, however, cannot reflect the importance of GDL porosity
on fuel cell performance. The hydrophobicity of the diffusion layers may interact with the
amount of water available for hydration at the membrane. The thickness and porosity of
the gas diffusion layers can also change due to the applied compression. These layers can
be porous carbon cloths or paper coated with polytetrafluoroethylene (PTFE). Although it
has been recognized that the performance of the fuel cell can be significantly influenced
due to the porosity variation [17–21], systematical studies are still rarely reported in the
open literature.
Gurau et al. [22] considered the non-uniformity of the porosity of the gas diffusion
layers by developing a one-dimensional half-cell model in which the concept of the
effective porosity was employed to account for the fact that the pores may be partially
filled with liquid water. Chu et al. [23] investigated the effects of non-uniform porosity on
fuel cell performance in terms of physical parameters such as oxygen consumption,
current density, power density, etc. The non-uniformity of the porosity was accounted for
by four different continuous functions of position, each of which had a different averaged
value of porosity and a different type of distribution across the diffusion layers. Lee et al.
[24] experimentally studied the effects of porosity change on the performance of the fuel
cell due to the compression applied onto the GDL.
1560 R. Roshandel et al. / Renewable Energy 30 (2005) 1557–1572

In the present study, fuel cell performance is investigated by considering the effects of
porosity variation distribution in gas diffusion layers. The effect of compression of the
electrodes on the solid landing areas and the water generated at cathode side of GDL are
both considered in expressing the variation of porosity distribution.

2. Mathematical model

The operation principle of a PEM fuel cell is presented in Fig. 1. Humidified air enters
the cathode channel and hydrogen diffuse through the anode diffusion layer towards the
catalyst, where each hydrogen molecule splits up into two hydrogen protons and two
electrons according to:

2H2 / 4HC C 4eK (1)

The protons migrate through the membrane and the electrons travel through the
conductive diffusion layer and an external circuit where they produced electric work.
On the cathode side the oxygen diffuses through the diffusion layer, splits up at the
catalyst layer surface and reacts with the protons and electrons to form water:

O2 C 4eK C 4HC / 2H2 O (2)

From above, it can be seen that the overall reaction in a PEM fuel cell can be written as:

2H2 C O2 / 2H2 O (3)

Fig. 1. Operating scheme of a PEM fuel cell.

 R. Roshandel et al. / Renewable Energy 30 (2005) 1557–1572 1561

2.1. Gas diffusion layer modeling

Gas diffusion layer does much more than diffuse the gas. It forms an electrical
connection between the catalyst layer and the bipolar plate, or other current collectors. In
addition, it carries the product water away from the electrolyte surface, and also forms a
protective layer over the very thin layer of catalyst [4]. They are usually made of a porous
carbon paper or carbon cloth, typically 100–300 mm thick.
The gas diffusion layer also assists in water management during the operation of the
fuel cell. Too little or too much water can cause the cell to cease operation. The correct
backing material allows the right amount of water vapor to reach the membrane and keep it
humidified. This layer also allows the liquid water produced at the cathode to leave the
cell, so it does not flood [3]. Therefore, it is important to investigate and analyze the effect
of reactant distribution in this layer on the fuel cell performance. A cross-sectional view
through a GDL is depicted in Fig. 2.
The upper boundary (I, II) represents the interface between the GDL and either the solid
graphite plate or an open Gas-filled flow channel. The lower boundary (IV) corresponds to
the catalyst layer separating the GDL and the polymer electrolyte membrane.
Single-phase flow in porous media is typically modeled using Darcy’s law [26]

~ Z K K VP
U (4)
m
In this equation, the averaged velocity is related to pressure gradient.
The equation of continuum for the reactant gas is:
~ÞZ0
V$ðrU (5)
where U~ is a mass-averaged velocity. The conservation law for the reactant concentration
takes the form
~ C DVCÞ Z 0
V$ðC U (6)
The diffusive flux is given by Fick’s law, which states that the flux of one component
relative to the molar average velocity is proportional to the gradient in mole fraction via
J~ Z KDVC (7)
where D is the coefficient of diffusion and it is related to the fluid properties and the
porosity of gas diffusion layer. In isotropic porous media, the dependence of the diffusion

Fig. 2. Cross-section of the gas diffusion layer and gas flow channels.
1562 R. Roshandel et al. / Renewable Energy 30 (2005) 1557–1572

coefficient on porosity is often approximated using the empirical formula [27]:

Deff Z D31:5 (8)
This relates the effective diffusivity in the porous medium to that of a free gas. In an
ideal gas mixture, the pressure depends linearly on the mixture concentration (and hence
also the density) via the ideal gas law:
rRT
P Z CRT Z (9)
M
Thus, the Darcy equation for gas reactant becomes:

~ Z K KRT VC
U (10)
3m
The boundary consists of four distinct components, labeled I–IV in Fig. 2.
The boundary conditions on each of the four sections are obtained as follows [24]:

(I) The permeable boundary, where the mixture concentration immediately inside the
GDL is taken to be identical to that in the channel
C Z C (11)
That is, the pressure is assumed to be uniform throughout the depth of the channel.
Further it is assumed that diffusive flux of the component across the channel/GDL
interface is proportional to the difference in the concentration on either side
J Z r0 ðC K CÞ (12)
(II) The impermeable boundary between the graphite palate and the GDL where no-flow
condition on the vertical component of fluxes is imposed
vC
J Z 0; Z0 (13)
vy
(III) The open side boundaries, where a periodic solution in x-direction is assumed.
(IV) The boundary conditions at the permeable boundary between the catalyst and GDL
are:
J Z rH C (14)
vC r
ZK H C (15)
vy D

The mass transfer coefficient rH models the reactions and electrochemistry taking place
in the catalyst region.

2.2. Porosity variation due to compression

Most of the models have assumed for simplicity that the porosity of the GDL is
constant. This, however, may not reflect the importance of GDL porosity on fuel cell
 R. Roshandel et al. / Renewable Energy 30 (2005) 1557–1572 1563

Fig. 3. Effect of compression of gas diffusion layer porosity.

performance. Any change in the composition or the morphology of the GDL can lead to a
substantial influence on fuel cell performance owing to a porosity change [25].
In order to seal the fuel cell against any gas leakages, gas diffusion layer and bipolar
plate are bolted together under significant pressure. Compressing the GDL reduces the
porosity over the landing area, which in turn decreases the overall flux of reactant to the
catalyst layer, particularly above the landing area (Fig. 3).
The effect of compression on the porosity variation is considered by allowing the
porosity to vary as a function of x.
Using a composition of function of the form sin2nx which is appropriately scaled and
translated, a function for the variation of porosity across gas diffusion layer is expressed as
!
comp 0
X 2n
3 ðxÞ Z 3 An sin x (16)
n

where 30 is initial porosity and 3comp is the porosity of the gas diffusion layer after
compression process.

2.3. Porosity variation due to water generation

It is clear from the description of the proton exchange membrane that there must be
sufficient water content in the polymer electrolyte, otherwise the conductivity
decreases. However, there must not be so much water that the electrodes which
bounded to electrolyte, flood, blocking the pores in the gas diffusion layer [4]. Thus,
the presence of water generated in the GDL during fuel cell operation can change the
effective porosity [25].
To consider the non-uniformity of the porosity of the GDL caused by presence of water,
all flow rates related to the different phenomena should be determined. Then the effective
porosity can be specified by considering the presence of additional water amount in the
fuel cell. Fig. 4 outlines the water flows inside the cell.
1564 R. Roshandel et al. / Renewable Energy 30 (2005) 1557–1572

Fig. 4. Schematic of water flows in a PEM fuel cell.

2.3.1. Water input

Water entering the cell due to the fuel and oxidant humidification:

hum I xhumK
w
wK Z z (17)
2F 1 K xhumK
w

!
hum I xhumC
w x0N
wC Z zC humC
1C 02 (18)
4F 1 K xw xO2
hum hum
The mole fraction of water corresponding to humidification condition (xK , xC ) can
be determined by considering the humidification temperature.

2.3.2. Water production

Water produced in the cell (on the cathode side) from the electrochemical reaction:
I
wel Z (19)
2F

2.3.3. Water transport

Water transport in the membrane depends on the electro-osmotic transport associated
with the flow of the HC protons, and the convective transport due to a pressure gradient in
 R. Roshandel et al. / Renewable Energy 30 (2005) 1557–1572 1565

the membrane
 
k4 kp
v Z 3m
w Z C FVf K Vp (20)
m f f m
The membrane water transport ratio can be calculated from:
d
x Z 3m
wF v (21)
I
The water transport in the membrane pores can be determined using the following
expression:
I
wtr Z x (22)
F

2.3.4. Water outlet

Water outlet from the fuel cell is calculated using

out I xsatK
w
wK Z ðz K 1Þ (23)
2F 1 K xsatK
w

" ! #
0
out I x N xsatC
w
wC Z zC 1 C 0 2 K 1 (24)
4F xO 2 1 K xsatC
w

The water balance is expressed by the following relationships:

hum out
MK Z wK C wtr K wK (25)

hum out
MC Z wC C wel K wtr K wC (26)
Finally, the gas porosity in the cathode diffusion layer is determined using the
expression below
 
jMCj
3hum 0
Z3 1K ; if MCO 0 (27)
M1;Max
where Mmax is the amount of water corresponding to a complete flooding.

2.4. The effect of porosity on fuel cell performance

The performance of a PEM fuel cell can be illustrated by a voltage vs. current density
plot, or polarization curve. The polarization curve can be divided into four regions
characterized by activation overpotentials, ohmic overpotentials, concentration over-
potentials and convection or internal current over potentials. So, the operational fuel cell
voltage can be written as:
V Z VOCV K hact K hohm K hdif K hconv (28)
1566 R. Roshandel et al. / Renewable Energy 30 (2005) 1557–1572

2.4.1. Activation overpotential

In the activation overpotential region, the dominate source of losses is due to resistance
to electrochemical reaction. These losses also are referred to as activation losses, occurring
when slow electrochemical reactions are driven from equilibrium in order to produce
electric current [26]. The activation overpotential can be calculated from:
 
RT I
hact Z ln (29)
F I0

2.4.2. Ohmic overpotential

The ohmic overpotential losses from electronic and ionic resistance in the cell is the
most significant source of performance degradation. Ohmic overpotential is determined by
the following equation:
hohm Z IRcell (30)

2.4.3. Concentration (diffusion) overpotential

In the concentration overpotential region, losses due to mass transport limitation are
dominant. Concentration overpotential occurs when the chemical reaction is limited by the
rate of which reactants can be supplied. This lack of reactant shows the electrochemical
reaction, resulting in a lower cell potential. This amount can be calculated by:
 
Ilim
hdif Z uTI ln (31)
Ilim K I
where Ilim is the limiting current density that is affected by additional water amount and
effective porosity of the gas diffusion layer in a PEM fuel cell
 T 0:823
2FDO2KN2 ð3dg Þ1:5 273
Ilim Z K lnð1 K xO2 Þ (32)
Vm ldC

2.4.4. Convection overpotential (internal current)

The electrolyte should only transport ions through the cell, but a certain amount of fuel
diffusion and electron flow is always possible. The fuel looses and its effect is usually not
very important. The convection overpotential can be determined using the following
equation:
lm
hconv Z FCHC v (33)
k

2.5. Grid size effects and numerical accuracy

Accuracy tests were performed on the numerical procedure, particularly to study the
effect of the grid fineness on the solution. The important parameters used in this analysis
are shown in Table 1. Comparison of the mole fraction distribution under various grid
sizes were made and presented in Fig. 5. The results show that by increasing the fineness of
 R. Roshandel et al. / Renewable Energy 30 (2005) 1557–1572 1567

Table 1
Modeling parameters

Parameter values
H 0.005
L 1
LC 0.25
LS 0.25
K 10K8
30 0.74
R 8.31!107
T 346.15
CHC 1.2!10K3
k 0.151
kp 1.8!10K14
k4 7.18!10K16
lm 0.023
dw 0.0543
3m
w 0.3
t 0.77
Anode Cathode

m 2.1!10 K4
1.2!10K4
D 0.089 0.123
r0 10 10
rH 0.8 0.2
T 358 353
p 3 5
z 1.2 3

the grid no significant changes appear in the mole fraction distribution. The final
computations were performed with 40!40 grid points to maintain relatively moderate
computing times in the final calculations. It is reasonable to assume that the accuracy in
the presented mole fraction distribution is less than 1%.
To further assess the accuracy of the mathematical model, the comparison of the
computed polarization curve for air and oxygen with the experimental data of Maggio et al [4]

Fig. 5. Comparison of the mole fraction distribution under various grid sizes.
1568 R. Roshandel et al. / Renewable Energy 30 (2005) 1557–1572

Fig. 6. Comparison of polarization curve with experimental and numerical data.

is presented in Fig. 6. As seen from this figure, the results of the present computations agree
well with experimental data.

3. Results and discussion

To investigate the effect of porosity on gas reactant distribution in anode and cathode
gas diffusion layer the porosity distribution in the first step is considered to be constant. In
Fig. 7, the reactant mole fraction is presented. The mole fraction plot shows that gas

Fig. 7. Reactant mole fraction in gas diffusion layer with constant porosity distribution.
 R. Roshandel et al. / Renewable Energy 30 (2005) 1557–1572 1569

Fig. 8. Reactant mole fraction in gas diffusion layer with consideration of compression effect on porosity of GDL.

concentration decreases from channel to membrane as the reactant is consumed. However,

the reactant concentration in the GDL/membrane interface as well as the electrical current
density are both uniform. The average of electrical current density is 0.61 A/cm2.
Fig. 8 presents the reactant distribution in GDL when porosity is considered as a
function of compression pressure. In this case, due to the compression the porosity
decreases specially under the landing area. This causes decrease in overall reactant flux to
the catalyst layer particularly under the landing areas and thus creating a more non-
uniformity in the gas diffusion layer. Therefore, the variation of produced electrical
current increases across the electrodes. It means that some of the catalyst particles,
especially under the landing area remain unused and the average electrical current density
reduces to 0.47 A/cm2 (w23% less!)
Another source of variation in the GDL’s porosity is presence of water in the layers. If
too much water is present, flooding may occur resulting in the pores of the gas diffuser
filled by liquid water, which blocks the transport of reactant to the reaction site in the
cathode side. It is shown in Fig. 9 that the water decreases the effective porosity of GDL

Fig. 9. Reactant mole fraction in gas diffusion layer with consideration of compression effect and presence of
water droplet on porosity of GDL.
1570 R. Roshandel et al. / Renewable Energy 30 (2005) 1557–1572

Fig. 10. Electrical current density distribution in cathode and anode side of a PEM fuel cell.

and therefore, casing the reduction in the oxygen concentration in GDL/Membrane

interfaces. In fact the gas porosity in the diffusion layer decreases from the initial 74% to
55% considering the blockage of the pores by water.
Fig. 10 shows the electrical current density in electrodes corresponding to reactant
concentrations. It can be observed that compression pressure affects the electrical current
distribution along the electrodes and makes peaks in the electrode current densities.

Activition Activition
63 % 58 %
Ohmic
Ohmic 24 %
25 %

1% 11 % 1% 17 %
Convection Diffusional Convection Diffusional

Ohmic
Activition
20 %
49%

1% 30 %
Convection Diffusional

Fig. 11. Percentage of overpotential contribution corresponding to (a) constant porosity along GDL, (b) spatial
variation of porosity due to compression pressure and (c) porosity as a function of water amount in GDL at
0.85 A/cm2.
 R. Roshandel et al. / Renewable Energy 30 (2005) 1557–1572 1571

As already mentioned, porosity variation changes the limiting current density and so
the diffusion overpotential increases. However, the potential drop related to the diffusion
overpotential appears to be too sharp compared with other overpoentials.
Fig. 11 presents the contributions of the relative share of the calculated overpotential
corresponding to different porosity distributions. It can be observed that diffusion
overpotential increases from 11% to 30% when effective porosity deceases due to the
compression pressure and presence of liquid water in gas diffusion layers.

4. Conclusion

The gas diffusion layer is an essential component of the proton exchange membrane
fuel cell, and limitations due to mass transport in the GDL are known to be significant. The
properties of the diffusion layer will impact the optimum performance of the catalyst and
the electrodes. These layers are porous to allow for distribution of the gases to unexposed
areas of the flow channel and this distribution allows for complete utilization of the
electrode area. Many of simulation, have assume for simplicity that the porosity of GDL is
constant, but in practice, the compression pressure corresponding to assembly process and
presence of water in these layers may change the porosity distribution and affect gas
diffusion coefficient in GDLs. Any change in porosity or diffusion coefficient can lead to a
substantial influence on fuel cell performance owing to a diffusion overpotential change.
The results in the present work demonstrate that the compression pressure can lead to
32% decrease in porosity and so reduce the reactant transport in GDL upto 10%. In
addition, water management simulation shows that the water droplet can block the pores in
GDL especially in high current density. In fact The average porosity of gas diffusion layer
decreases 23% at 0.85 A/cm2.

References

[1] Larminie J, Dicks A. Fuel cell system explained. New York: Wiley; 2000.
[2] Bernardi DM, Verbrugge MW, Mathematical model of the solid polymer electrolyte fuel cell. General
Motors Report, GMR-7360; 15 May 1991.
[3] Hirschenhofer JH, Stauffer DB, Engleman R. Fuel cells—a hand book (revision 3), US Department of
Energy Undercontract No. DE-AC01-88FE61684; 1994.
[4] Maggio G, Recupero V, Pino L. Modeling polymer electrolyte fuel cells. J Power Sources 2001;101:
275–86.
[5] Stockie JM, Promislow K, Wetton BR. Int J Numer Meth Fluids 2001;1–35.
[6] Scherer GG. Solid State Ionics 1997;94:249–57.
[7] Bernardi DM. J Electrochem Soc 1990;137:3344–50.
[8] Verbrugge MW, Hill RF. J Electrochem Soc 1990;137:886–93.
[9] Verbrugge MW, Hill RF. J Electrochem Soc 1990;137:1131–8.
[10] Bernardi DM, Verbrugge MW. AlChE J 1990;37:1151–63.
[11] Verbrugge MW, Bernardi DM. J Electrochem Soc 1992;139:2477–90.
[12] Springer TE, Zawodziniski TA, Gottesfeld S. J Electrochem Soc 1992;138:2477–90.
[13] Singh D, Lu DM, Djilali N. Int J Eng Sci 1999;37:431–52.
1572 R. Roshandel et al. / Renewable Energy 30 (2005) 1557–1572

[14] Rowe A, Li X. Mathematical modeling of proton exchange membrane fuel cells. J Power Sources 2001;102:
82–96.
[15] Gurau V, Liu H, Kakac S. ALCHE J 1998;44:2410–22.
[16] Yi JS, Nguyen TV. J Electrochem Soc 1999;146:38–45.
[17] Um S, Wang CY, Chen KS. Computational fluid dynamic modeling of proton exchange membrane fuel
cells. J Electrochem Soc 2000;147(12):4485–93.
[18] Jordan LR, Shukla AK, Behrsing T, Avery NR, Muddle BC, Forsyth M. J Power Sources 2000;86:250–4.
[19] Jordan LR, Shukla AK, Behrsing T, Avery NR, Muddle BC, Forsyth M. J Power Sources 2000;30:641–6.
[20] Giorgi L, Antlini E, Pozio A, Passalacaqua E. Electrochim Acta 1998;43:3675–80.
[21] Wohr M, Bolwin K, Schnurnberger W, Fischer M, Neubrand W, Eigenberger G. Int J Hydrogen Energy
1998;23:213–8.
[22] Gurau V, Barbir F, Liu H. J Electrochem Soc 2000;147:4485–93.
[23] Chu HS, Yeh C, Chen F. Effects of porosity change of gas diffuser on performance of proton exchange
membrane fuel cell. J Power Sources 2003;123:1–9.
[24] Lee WK, Ho CH, Van Zee JW, Murthy M. The effect of compression and gas diffusion layers on the
performance of a PEM fuel cell. J Power Sources 1999;84:45–51.
[25] Thomas S, Zalbowitz M. Fuel cell green power, Los Alamos National Laboratory, New Mexico.
[26] Nield DA, Bejan A. Convection in porous media. 2nd ed. Berlin: Springer; 1998.
[27] Akanni KA, Evans JW, Abramson IS. Effective transport coefficients in heterogeneous media. Chem Eng
Sci 1987;42(8):1945–54.