International Journal of Heat and Mass Transfer 144 (2019) 118705

Contents lists available at ScienceDirect

International Journal of Heat and Mass Transfer

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

Effect of operating conditions on the performance of a PEM fuel cell

Hossein Askaripour
Department of Chemical and Petroleum Engineering, Sharif University of Technology, Tehran, Iran

a r t i c l e i n f o a b s t r a c t

Article history: In this study, a two-phase flow model based on species transport, energy, and electrochemical equations
Received 31 May 2019 was developed for polymer electrolyte membrane fuel cell (PEMFC). The influences of heat sources and
Received in revised form 9 August 2019 sinks, and water transport inside the fuel cell were also taken into account. In this regard, the effect of
Accepted 6 September 2019
the operating conditions including cell temperature and pressure, anode and cathode inlet humidity,
Available online 12 September 2019
and anode and cathode stoichiometric ratios on the fuel cell performance was investigated. The simula-
tion results show that inlet humidity and stoichiometric ratio of the anode side, cell pressure and tem-
Keywords:
perature, and distribution of the heat sources and sinks are crucial factors affecting the cell
PEM fuel cell
Two-phase flow model
performance and the two-phase flow characteristics. The amount of heat produced (or consumed) as a
Heat sources/sinks result of heat sources (or sinks) considerably decreases at the channel inlet, whereas it decreases slowly
Cell performance in the rest of the flow channel. It was found that, for moderate to high current densities, the fuel cell per-
formance diminishes with an increase of the cell pressure. Increasing inlet humidity and stoichiometric
ratio of the cathode side insignificantly influences the polarization curve; however, the cell performance
enhances with an increase of the cell temperature.
Ó 2019 Elsevier Ltd. All rights reserved.

1. Introduction water distribution in gas diffusion layer and its influence on the
distribution of gas species. Meng and Wang [5] proposed a model
Polymer electrolyte membrane fuel cells (PEMFC) are electro- coupled transport phenomena and electrochemical equations with
chemical devices that convert the chemical energy of oxygen and an interfacial model for liquid water at the interface of gas diffu-
hydrogen into electrical energy [1]. Much attention has been paid sion layer and flow channel. The model was used to examine the
to PEM fuel cells as an alternative power source for automotive and effect of cathode flooding on cell performance. Wang and Wang
portable applications due to their advantages such as fast and easy [6] examined the distribution of liquid water and water flooding
start-up, favorable power to weight ratio, high efficiency, and low in PEM fuel cell using a non-isothermal model. The results showed
noise [2]. that the transport of liquid water via vapor-phase diffusion cannot
To develop the PEM fuel cells technology, significant researches be neglected and vapor-phase diffusion increased water removal
have been conducted to date; however, there are still serious chal- from the gas diffusion layer. A one-dimensional and isothermal
lenges in the design and operation of PEM fuel cells that need to be model taking into account the effects of membrane conduction
overcome to improve the efficiency and performance. Among these and electroosmotic drag was proposed by Springer et al. [7]. The
factors, controlling the membrane water content is an important results showed that the membrane resistance increased with an
issue that causes the phenomenon of cathode flooding or anode increase in current density and a thinner membrane had great
dehydration. Water is transported via three mechanisms in fuel advantages in alleviating this resistance problem. An experimental
cells, i.e., cathode back diffusion, water diffusion by oxidant and work concerning water transport number in NafionÒ 117 was con-
fuel gas streams, and electroosmotic drag of water, hence the net ducted by Fuller and Newman [8]. The results showed that water
water transport as well as water production determine water con- transport number decreased slowly with an increase of the mem-
tent in the membrane [3]. brane dehydration.
Various studies were conducted to describe the mechanisms of To improve the electrical performance of PEM fuel cells, taking
water transport in membrane. Natarajan and Nguyen [4] devel- into account the thermal effects is essential. As stated in the liter-
oped an isothermal three-dimensional model for the cathode flow ature [9–11], charge and mass transfer within the cell and open cir-
channel. The model provided qualitative insights about the liquid cuit voltage strongly depend on temperature. In addition, the heat
transfer is dependent on charge and mass transfer. Therefore, the
temperature effects are not negligible and the heat transfer
E-mail address: h_askaripour@yahoo.com

https://doi.org/10.1016/j.ijheatmasstransfer.2019.118705
0017-9310/Ó 2019 Elsevier Ltd. All rights reserved.
2 H. Askaripour / International Journal of Heat and Mass Transfer 144 (2019) 118705

Nomenclature

a water activity t thickness, m

Acell cell active area, m2 T temperature, K
C concentration, kmol m
3 w channel width, m
Cmix heat capacity of gas mixture, J kg
1 K
1 ws support width, m
dh hydraulic diameter, m y mole fraction
Dw diffusion coefficient of water in the membrane, m2 s
DO2
g diffusion coefficient of oxygen, m2 s
1 Greek symbols
Ecell cell voltage, V a net water drag coefficient
Eoc open circuit voltage, V b2 constant parameter
F Faraday constant, C mol
1 DHv water vaporization heat, kJ mol
1
h channel height, m DS entropy change, J mol
1 K
1
hO2 mass transfer coefficient of oxygen, m s
1 g overpotential, V
I0 exchange current density at a reference pressure, A m
2 k water content of membrane
I(x) local current density, A m
2 m viscosity, Pa s

I average current density, A m
2 mx velocity in x-direction, m s
1
Ilim limiting current density, A m
2 n stoichiometric parameter
kc homogeneous rate constant for condensation and evap- qm,dry dry membrane density, kg m
3
oration of water qw density of liquid water, kg m
3
km membrane conductivity / GDL porosity
kmix thermal conductivity of gas mixture, W m
1 K
1
Kw permeability of liquid water, m2 Subscripts/superscripts
L flow channel length, m
a anode
Mm,dry equivalent weight of a dry membrane, kg kmol
1 ad adsorption
Mw molecular weight of water, kg kmol
1 c cathode
nd electroosmotic drag coefficient cat catalyst
nlv water flow rate subjected to phase change, kmol s
1
cond condensation
N molar flow rate, kmol s
1 eff effective
Nch number of serpentine channels evp evaporation
P pressure, Pa
g gas phase
Pc capillary pressure, Pa i the ith species
Q energy, J kmol
1 l liquid
R universal gas constant, J mole
1 K
1 m membrane
RH relative humidity
rxn reaction
s liquid water saturation sat saturation


s average water saturation std standard
sd liquid water saturation at flow channel-GDL interface v vapor
S stoichiometric ratio
w water
Sh Sherwood number

phenomenon plays an important role on the performance and elec- however, most of them considered the electrochemical reactions
trical behavior of PEMFCs. A study about the thermal and water as an isothermal process and did not take into account the thermal
management in a PEM fuel cell was conducted by Nguyen and effects. In addition, the emerging field of heat transfer in PEM fuel
White [12] and the effects of water diffusion and latent heat were cell have not received much attention to date, and the numerical
considered. The results showed that, at high current densities, a and analytical models studied the heat and mass transfer were
large fraction of the voltage loss in fuel cell is due to ohmic overpo- based on local thermal equilibrium [15,16]. Assuming local ther-
tential and the back diffusion of water is insufficient to keep the mal equilibrium, fluid and solid have the same temperature.
membrane hydrated. Hwang and Chen [13] developed a coupled Because of large difference between the conductivities of fluid
model of heat and mass transfer to predict the transport phenom- and solid phases, e.g., air and carbon fibers, the temperature of
ena inside the porous electrode of a fuel cell. Zong et al. [14] pro- fluid is quite different from that of solid.
posed a non-isobaric and non-isothermal model to study the In this study, a two-phase model was proposed based on spe-
effect of water phase change, pressure variation along the flow cies transport, energy, and electrochemical equations. The influ-
channel, and water transport through the membrane. It was found ences of various heat sources/sinks, i.e., water sorption and
that humidification of cathode and anode sides affected the cell desorption at the interface of gas distribution layer (GDL) and
performance and proper injection of liquid water at anode side membrane, water phase-change in GDL, anode and cathode
could improve the cell performance. Yan et al. [10] examined water half-reactions, cathode activation loss, and convective heat trans-
and thermal management in the membrane of a fuel cell and fer between the gas and solid phases, were taken into account.
solved the transport equations using different boundary condi- Initially, a comparison between the significance of different heat
tions. The results showed that the anode temperature had remark- mechanisms was carried out, and then the effects of anode and
able effect on water content in the membrane and high current cathode inlet humidity, anode and cathode stoichiometric ratios,
density could cause non-uniform distribution of the temperature and operating pressure and temperature on the fuel cell perfor-
in membrane. mance were explored. We believe that such a parametric study
Investigators have been proposed different mathematical mod- can be useful and beneficial to improve the performance of PEM
els for the simulation of PEM fuel cell in the past two decades; fuel cells.
 H. Askaripour / International Journal of Heat and Mass Transfer 144 (2019) 118705 3

2. Mathematical model where Ni is the molar flow rate of species i along the flow channel, w
is the channel width, n is the stoichiometric parameter, a is the net
Fig. 1 shows a schematic diagram of a PEM fuel cell. A single cell water drag coefficient, I(x) is the local current density, and the sub-
is composed of cathode flow channel, cathode GDL, cathode cata- scripts a and c denote the anode and cathode sides, respectively.
lyst layer, electrolyte membrane, anode catalyst layer, anode The net water drag coefficient means that a mole H2O is trans-
GDL, and anode flow channel. The flow channels supply the reac- ported from anode to cathode side together with 1.0 mol H+ and
tants of H2 and O2 and the gas distribution layers cause even distri- can be evaluated from [12]:
bution of reactant species. Hydrogen molecules are introduced into F C w;c
C w;a
the anode side and separated into proton (H+) and electron (e
) a ¼ nd
Dw ð5Þ
IðxÞ tm
using the catalyst layer. The protons are transferred via the ion-
conducting membrane while the electrons flow through an exter- where nd is the electroosmotic drag coefficient, Dw is the diffusion
nal circuit. Oxygen molecules enter the cathode side and diffuse coefficient of water in the membrane, tm is the membrane thickness,
through GDL towards the catalyst layer where combine with pro- and Cw,a and Cw,c are water concentration at the anode- and
tons and electrons to produce water and heat. cathode-membrane interfaces, respectively. The diffusion coeffi-
cient of water can be calculated from [7]:
2.1. Species transport equations   
1 1
Dw ¼ nd D0 exp 2416
ð6Þ
303 T
The molar flow rate of hydrogen and oxygen at the inlet of
anode and cathode flow channels can be determined based on The electroosmotic drag coefficient, which is related to water
the stoichiometric ratios and average current density [17]. transport mechanism through the membrane with protons, can

be given by [7]:
Sa Acell I
N_ H2 ¼
cell
ð1Þ 2:5km
2F nd ¼ ð7Þ
22


Sc Acell I where km is the water content of membrane and can be calculated
N_ O2 ¼
cell
ð2Þ as a function of water activity at the anode side, aa [7]:
4F
(
where Sa and Sc are the anode and cathode stoichiometric ratios, 0:043 þ 17:81aa
39:85a2a þ 36:0a3a 0 < aa 6 1

km ¼ ð8Þ
respectively, Acell is the cell active area, I is the average current den- 14:0 þ 1:4ðaa
1Þ aa > 1
sity, and F is the Faraday constant.
Water concentration at the anode- and cathode-membrane
H2 and O2 are transported from flow channels to the reaction
interfaces can be evaluated by [12]:
sites where react with each other and produce water. Under
8q
steady-state conditions, species conservation equation along the < Mm;dry ð0:043 þ 17:81ak
39:85a2k þ 36:0a3k Þ 0 < ak 6 1
m;dry
flow channel can be expressed by [18]: C w;k ¼
: qm;dry ½14 þ 1:4ðak
1Þ ak > 1
M
dNi wIðxÞ m;dry
¼ ni ð3Þ ð9Þ
dx 4F
In which: In which [12]
(
Anode channel : nH2 ¼
2; nH2 O;a ¼
4a yH2 O;k Pk
ak ¼ ð10Þ
Cathode channel : nO2 ¼
1; nH2 O;c ¼ 2 þ 4a; nN 2 ¼ 0 Psat
w;k

ð4Þ where qm,dry is the dry membrane density, Mm,dry is the equivalent
weight of a dry membrane, a is the water activity, P is the pressure,
Psat
w is the water saturation pressure, y is the molar fraction, and the
subscript k denotes either the cathode or anode.

2.2. Mass transport in GDL

The difference between the vapor pressure and saturation pres-

sure of water determines the amount of liquid water that forms
inside the fuel cell. Condensation of water vapor starts in the
GDL pores when water vapor pressure exceeds the saturation pres-
sure; however, the liquid water will be evaporated, if there is any
liquid water present in the cell, when the saturation pressure
becomes more than water vapor pressure. Hence, the state of
GDL flooding and two-phase flow inside the gas flow channels
depend on net changes in the water vapor condensation and liquid
water evaporation in the electrolyte membrane and can be evalu-
ated from [19]:
 
dNlw;k kc wh  v 
¼ Pw
Psat
w;k ð11Þ
dx RT

dNvw;a dNlw;a aIðxÞ

Fig. 1. Schematic diagram of a PEM fuel cell.
¼

w ð12Þ
dx dx F
4 H. Askaripour / International Journal of Heat and Mass Transfer 144 (2019) 118705

dNvw;c dNlw;c ð1 þ 2aÞIðxÞ Ecell ¼ Eoc
gact
gohm
gconc ð22Þ

¼
þw ð13Þ
dx dx 2F where gact, gohm, and gconc are the activation, ohmic, and concen-
where N lw is the molar flow rate of liquid water, Nvw is the molar flow tration overpotentials, respectively, and Eoc is the open circuit volt-
rate of water vapor, kc is the homogeneous rate constant for con- age. The open circuit voltage is a function of temperature and
densation and evaporation of water, h is the channel height, T is partial pressure of oxygen and hydrogen [21].
 
the temperature, and Pvw is the water vapor pressure. Eoc ¼ 1:229
8:456  10
4 ðT
298:15Þ þ 4:31  10
5 Tln P H2 P0:5
O2
The condensation of water vapor starts at the interface of cata-
lyst layer and GDL, and then liquid water disperses in GDL pores ð23Þ
and flows towards the gas flow channel. Liquid water is driven Neglecting the anode overpotential, the cathode activation
by the capillary pressure inside GDL, so Darcy’s law can be overpotential can be calculated from [12]:
employed to describe the velocity and molar flow rate of liquid " #
water [19]. RT IðxÞ
gact ðxÞ ¼ ln ð24Þ
Kw 0:5F I0 Pcat
O2 ðxÞ
ml ¼
rP l ð14Þ
lw where I0 and Pcat
O2 are the exchange current density at a reference

qw K w pressure and partial pressure of oxygen at the catalyst layer, respec-

Nlw ¼
rP l ð15Þ tively. Assuming ideal gas mixture, P cat
M w lw O2 ðxÞ can be calculated from:

cat
where ml, mw, qw, Mw, Pl are the velocity, viscosity, density, molecu- Pcat
O2 ðxÞ ¼ C O2 ðxÞRT ð25Þ
lar weight, and pressure of the liquid water, respectively, and Kw is
In which [22]:
the permeability of liquid water in GDL pores. The capillary pres- !
sure is defined as the difference between pressure of the gas phase IðxÞ 1 tGDL
and liquid water. C cat bulk
O2 ðxÞ ¼ C O2 ðxÞ
þ ð26Þ
4F hO2 Deff
O2
g
Pc ¼ Pg
Pl ð16Þ
where hO2 and Deff
O2 are the mass transfer coefficient and effective dif-
Hence, Eq. (15) can be rewritten as
fusion coefficient of oxygen, respectively, and C bulk
O2 is the oxygen
q Kw
Nlw ¼
w rðPg
P c Þ ð17Þ concentration in flow channel that can be calculated from
M w lw
PO2 ðxÞ
Assuming negligible pressure gradient of the gas phase in GDL, C bulk
O2 ðxÞ ¼ ð27Þ
RT
Eq. (17) is simplified to
  To determine oxygen transfer inside the GDL pores, Bruggeman
qw K w q K dP ds equation relating Deff
Nlw ¼
rP c ¼
w w
c ð18Þ O2 to GDL porosity, /, was applied [7].
M w lw M w lw ds dy
3
where s is the liquid water saturation, defined as volume fraction of Deff
O2
g ¼ DO2
g ½ð1
sÞ/
2 ð28Þ
the GDL pores that is occupied by liquid water. The permeability
and capillary pressure of liquid water are functions of liquid water where [23]
saturation. In this study, dPc =ds was assumed to be constant and the 1
y O2

linear function of K w ¼ K w s was considered for the permeability,
DO2
g ¼ yN yH ð29Þ
 DO
2
þ DO 2 O;c
2
N 2 2
H2 O
where K w is the permeability at 100% saturation [20].
Integrating Eq. (18), distribution of liquid water saturation and the binary diffusion coefficients can be expressed as follows
across the GDL is obtained [19]. [24]:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi  2=3  
2Nlw ðy
t GDL Þ T 1
s ¼ s2d þ ð19Þ DO2
N2 ¼ Dstd
O2
N2 ð30Þ
b 273 Pc

and  2=3  
T 1
   DO2
H2 O ¼ Dstd
O2
H2 O ð31Þ
qw K w dPc 273 Pc
b¼

ð20Þ
M w lw ds Assuming a fully-developed laminar flow, O2 mass transfer
where sd is the liquid water saturation at the interface of flow chan- coefficient can be expressed as follows [25]:
nel and GDL. The average saturation along the flow channel can be hO2 dh
calculated from [19]: Sh ¼ ¼ 2:7 ð32Þ
DO2
g
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
s2d b
2Nlw t GDL 2 s2d b
2N lw t GDL l where Sh is the Sherwood number and dh is the channel hydraulic
Z t GDL
sd b þ 2 Nw tGDL þ s3d b

1  
b b diameter.
s¼ sðy Þdy ¼
t GDL 0 3t GDL Nlw The ohmic overpotential associated with the proton transport
through the membrane is given by [12]
ð21Þ
tm
gohm ðxÞ ¼ IðxÞ ð33Þ
2.3. Electrochemistry km ðxÞ

The cell voltage, Ecell, can be calculated by subtracting the over- where km is the membrane conductivity and can be evaluated from
potentials from the open circuit voltage. [7]:
 H. Askaripour / International Journal of Heat and Mass Transfer 144 (2019) 118705 5

  
1 1 where S is the species entropy, and entropy change for the half-
km ðxÞ ¼ ð0:5139km
0:326Þexp 1268
ð34Þ
303 T reactions of hydrogen oxidation and oxygen reduction are
DSa = 84.7 J mole
1 K
1 and DSc =
247.9 J mole
1 K
1, respectively
The concentration overpotential arising from variations in the
[30]. The anode and cathode reaction heat is proportional to the
concentration of reactants or products between the electrode sur-
entropy change of each half-reaction and the current density, and
face and flow channel can be calculated from [26]:
can be evaluated by [14]
 b2
IðxÞ
gconc ðxÞ ¼ IðxÞ b1 ð35Þ
Q arxn ¼
DSa
T a ðxÞIðxÞ ð43Þ
Ilim
2F
where Ilim is the limiting current density, b2 is a constant parameter,
and b1 is a parameter depending on the O2 partial pressure and tem- DSc
Q crxn ¼ T c ðxÞIðxÞ ð44Þ
perature [26]. 4F
The current density is not distributed evenly along the flow Because of activation losses, some of the electrical energy turns
channel, so the average current density can be calculated by inte- into heat. Knowing the overpotentials, the heat generated from
grating the local current density along the whole length of the flow cathode activation overpotential is given by [14]
channel:
Z L h i Q cact ¼ gact ðxÞIðxÞ ð45Þ

1

I¼ 1
s ðxÞ IðxÞdx ð36Þ The thermal energy balance for the anode and cathode gas
L 0
streams can be expressed as
2
2.4. Energy balance dT a d Ta
qa;mix C a;mix ma;x ¼ ka;mix þ Q asorp=desorp þ Q acond=ev p þ Q arxn
dx dx2
Since water is in liquid phase in the membrane and vapor phase þ ha ðT elec;a
T a ðxÞÞ ð46Þ
in the GDL pores, water in fuel cell is subjected to sorption/desorp-
tion at the membrane-GDLs interfaces and the heat produced or 2
consumed as a result of phase change should be considered. The dT c d Tc
qc;mix C c;mix mc;x ¼ kc;mix 2 þ Q csorp=desorp þ Q ccond=ev p þ Q crxn
sorption enthalpy, DHsorp, can be expressed as follows [27]: dx dx
þ Q cact þ hc ðT elec;c
T c ðxÞÞ ð47Þ
DHsorp ¼ DHfH2 OðadÞ
DHfH2 OðgasÞ ð37Þ
where mx is the velocity in x-direction, ha and hc are the heat transfer
Various values have been reported for the sorption enthalpy in
coefficients for anode and cathode flow channels, T surf
a and T surf
c are
the literature [28]; however, most of them are close to the latent
the temperatures of anode and cathode electrodes, and Cmix and
heat of water (Lv = 41.6 kJ mol
1 at 353 K [29]). The amount of heat
kmix are the heat capacity and thermal conductivity for gas mixture,
that is produced/consumed by the sorption/desorption process is a
respectively.
function of sorption enthalpy and molar flow rate of water sub-
jected to the phase change. According to the mass transport in
the membrane electrolyte assembly, molar flow rate of water at 3. Solution procedure
the interfaces of membrane and GDLs can be calculated. Therefore,
sorption heat at the anode and cathode sides of PEM fuel cell can To solve the governing equations of PEM fuel cell, the method of
be evaluated by [27] keeping the cell voltage constant was adopted in this study. The
cathode and anode flow channels were divided into a number of
Q asorp ¼ DHsorp NaH2 O and Q csorp ¼ DHsorp NcH2 O ð38Þ
equally-sized segments. An initial value was assumed for average


Water condensation occurs when the partial pressure of water current density, I ini , and the inlet molar flow rates of hydrogen
vapor exceeds the saturation pressure. To determine condensa- and oxygen were calculated based on Eqs. (1) and (2), respectively.
tion/evaporation sites, partial pressure of vapor is compared with A local current density was assumed for the first segment and vari-
the saturation pressure, which is a function of temperature distri- ations in the species flow rate were calculated from Eq. (3), then
bution inside the gas flow channel. After identifying the condensa- the segment voltage was obtained from Eq. (22) after open circuit
tion and evaporation sites, the relevant heat is calculated from [14] voltage, and activation, ohmic, and concentration overpotentials
were calculated. If the calculated voltage for the first segment dif-
Q cond=ev p ¼ nlv DHv ð39Þ fered from predefined cell voltage, the assumed local current den-
sity should be corrected using Newton-Raphson method;
otherwise, calculations were repeated for the next segment. Calcu-
DHv ¼ 45070
41:9½T
273 þ 3:44  10
3 ½T
2732
lations proceeded to the last segment, and then the average current
þ 2:54  10
6 ½T
2733
8:98  10
10 ½T
2734 ð40Þ density was calculated for the whole channel length. Finally, a
comparison between the calculated average current density and
where nlv is the water flow rate subjected to phase change and DHv
the assumed initial value was implemented, and if the convergence
is the water vaporization heat. criterion did not meet, Newton-Raphson method was employed to
A part of chemical energy cannot be transformed into electrical
modify the initial value assumed for the average current density;
work in an electrochemical system. In this study, the concept of otherwise, the calculations ended. Flow chart of the solution proce-
entropy was employed to calculate the energy resulting from
dure is shown in Fig. 2.
anode and cathode reactions. The entropy change for the half-
reactions of anode and cathode are as follows [14]:
4. Results and discussions

DSa ¼ 4SHþ þ 4Se

2SH2 ð41Þ

In simulation of PEM fuel cell, the governing equations of spe-
cies transport, energy, and electrochemistry for both the anode
DSc ¼ 2SH2 O
SO2
4Se

4SHþ ð42Þ
and cathode gas streams were solved together. The information
6 H. Askaripour / International Journal of Heat and Mass Transfer 144 (2019) 118705

Table 1
Parameters and properties used in present simulation.

Parameters Values
Active cell dimensions 99  99 mm2
Channel width (w) 1 mm
Channel height (h) 1 mm
No. of serpentine channels (Nch) 5
Support width (ws) 1 mm
GDL thickness (tGDL) 300 mm
GDL porosity (/) 60%
1  10
10 m2

Permeability at 100% saturation (K w )
dPc/ds
28.42 Pa m
1
Homogeneous rate constant (kc) 1
Membrane thickness (tm) 0.1275 mm
D0 5.5  10
11 m2 s
1
Equivalent weight of a dry membrane (Mm,dry) 1100 kg kmol-1
Dry membrane density (qm,dry) 2000 kg m
3
Oxygen exchange current density (I0) 0.01 A cm
2
Cell operating pressure (Pcell) 1.5 atm
Inlet humidity of anode and cathode (RHin) 100%
Anode stoichiometric ratio (Sa) 1.2
Cathode stoichiometric ratio (Sc) 1.8
Cell operating temperature (Tcell) 75 °C
Limiting current density (Ilim) [31] 2.2 A cm
2
b2 [30] 2.0

Fig. 2. Flow chart of solution procedure.

regarding the gas flow channels, GDLs, and membrane as well as

the cell operating parameters are summarized in Table 1.

4.1. Validation of numerical simulations Fig. 3. Comparison of polarization curve between the simulation results and
experimental data.
In this work, a PEM fuel cell with serpentine flow channels and
active cell area of 100 cm2 was simulated and the results were
compared with the experimental data reported in Ref. [32]. Li channel at Vcell = 0.2, 0.4, 0.6, and 0.8 V, respectively. Fig. 4(b),
et al. [32] conducted an experimental work on a single cell with (d), (f), and (h) illustrates variations of the different heat mecha-
channel width of 1 mm and land width of 1 mm. The inlet streams nisms along the anode flow channel at Vcell = 0.2, 0.4, 0.6, and
for both anode and cathode sides were considered to be fully 0.8 V, respectively. It can be observed that the heat resulting from
humidified, the stoichiometric ratios of air and hydrogen were cathode activation overpotential and anode half-reaction has pos-
set as 3 and 1.5, and the cell operating pressure and temperature itive values, whereas the heat resulting from cathode half-reaction
were set as 1 atm and 80 °C, respectively. Fig. 3 shows the polariza- has negative values. This is due to positive and negative entropy
tion curve for both the numerical simulation and experimental changes of the anode and cathode reactions, respectively. In addi-
data. It can be seen that simulation results are in good agreement tion, amount of the heat of cathode half-reaction is higher than
with the experimental data; therefore, validity of the present mod- that of anode half-reaction owing to the higher entropy change.
eling is verified. It can be also seen that amount of the heat produced (or con-
sumed) as a result of cathode activation, and anode and cathode
4.2. Comparison of heat sources and sinks reactions considerably decreases at the channel inlet, whereas it
decreases slowly in the rest of the flow channel. The reason is that,
In this section, sorption, half-reaction, and activation heat activation overpotential and heat loss, and half-reactions heat are
sources/sinks for both anode and cathode sides were compared functions of the local current density. In addition to fully accessible
at different cell voltages. Fig. 4(a), (c), (e), and (g) illustrates varia- GDL pores, the reactants concentration and pressure, and mem-
tions of the different heat mechanisms along the cathode flow brane conductivity are the highest at cell inlet, so it is expected
 H. Askaripour / International Journal of Heat and Mass Transfer 144 (2019) 118705 7

Fig. 4. Variations of heat sources and sinks along the cathode and anode flow channels: (a, b) Ecell = 0.2 V; (c, d) Ecell = 0.4 V; (e, f) Ecell = 0.6 V; and (g, h) Ecell = 0.8 V.

that the current density to be maximum at the inlet of flow chan- reduction in the catalyst accessibility due to GDL flooding. More-
nel and decreases along the flow channel. The decrease of the cur- over, comparison of these heat sources and sinks indicates that
rent density can be attributed to lower reactants concentration and with an increase of the cell voltage, amount of the activation and
8 H. Askaripour / International Journal of Heat and Mass Transfer 144 (2019) 118705

half-reactions heat decrease. The polarization curve shows that cell fuel cell is shown in Fig. 6. As seen from this figure, fuel cell perfor-
voltage is inversely proportional to the average current density, so mance increases monotonically with an increase in cell pressure up
average current density decreases with an increase in cell voltage. to the current density of 0.4 A/cm2. The reason is that, higher cell
Based on Eqs. (1) and (2), inlet molar flow rates of fuel and oxidant pressures increase the partial pressure of hydrogen and oxygen
diminish with a decrease of the average current density. As a in the anode and cathode flow channels, which in turn increase
result, local current density drops resulting in the anode and cath- the open circuit voltage and decrease the cathode activation over-
ode reactions as well as the activation heat loss decrease. potential. In addition, higher cell pressures increase the water
As shown in this figure, sorption and desorption phenomena activity and water content in membrane, which in turn decrease
occur at the anode GDL- and cathode GDL-membrane interfaces the ohmic overpotential. As a result, cell voltage as well as fuel cell
and this causes the relevant heat to be positive and negative, performance are enhanced at low current densities. It can be also
respectively. The results indicate that condensation of water vapor seen that, for current densities larger than 0.4 A/cm2, cell perfor-
takes place at the anode side, whereas evaporation of liquid water mance decreases with an increase of the cell pressure. The higher
takes place at the cathode side. It can be seen that the anode sorp- current densities dominate the effects of oxygen partial pressure
tion heat and cathode desorption heat show a decreasing trend and water content of membrane, which cause the cathode activa-
along the gas flow channel. As explained in Section 2.4, sorption tion and ohmic losses to increase. In addition, higher current den-
and desorption heat are functions of the water flow rate subjected sities increase the concentration overpotential. Increasing the
to phase change, which in turn they are functions of the local cur- activation, ohmic, and concentration losses dominate the positive
rent density based on Eqs. (12) and (13). Therefore, sorption and effects of reactants partial pressures on open circuit voltage and
desorption heat diminish along the gas flow channel due to the result in the cell voltage and performance decrease at moderate
reduction in local current density. Besides, as explained in the pre- to high current densities.
ceding paragraph, inlet flow rates of the reactants and local current
density drop as the cell voltage increases. Hence, higher cell volt-
ages lead to decrease of the cathode desorption and anode sorption
heat.
Overall comparison of these figures shows that amount of the
heat produced (or consumed) via different heat sources (or sinks)
decreases intensively at channel inlet due to sudden drop in the
local current density. Moreover, the heat resulting from cathode
activation, anode and cathode half-reactions, anode sorption, and
cathode desorption have significant effects on the performance of
PEM fuel cell; however, at high cell voltages, where the current
density is small, the effect of these heat sources and sinks can be
neglected.

4.3. Cathode and anode stoichiometric ratios

The effect of cathode and anode stoichiometric ratios on the fuel

cell performance is shown in Fig. 5(a) and (b), respectively. The
cathode stoichiometric ratio was considered to be 1.8, 2.7, and
3.6, whereas the anode stoichiometric ratio was considered to be
1.2, 2.2, and 3.2 and variations in the polarization curve were
investigated. As illustrated, change in the cell performance is neg-
ligible with an increase of the cathode stoichiometric ratio.
Increasing anode stoichiometric ratio does not alter cell perfor-
mance at low current densities; however, at moderate to high cur-
rent densities, the cell performance is enhanced. The higher
stoichiometric ratios increase inlet molar flow rates of the reac-
tants and as a result, more hydrogen fuel and oxidizing agent are
available in the flow channels. As known to all, enthalpy and
entropy changes are the driving forces for chemical reaction and
the processes in which both the tendency toward minimum
enthalpy and maximum entropy favor the products. As mentioned
in Section 2.4, entropy change of the anode half-reaction is positive
while that of the cathode half-reaction is negative. Therefore, it is
expected that increasing the amount of reactant in anode
side positively affects fuel cell performance, whereas variation
in the amount of reactant in cathode side slightly affects the
cell performance. In addition, more influence of the anode
stoichiometric ratio on cell performance at moderate to high cur-
rent densities are due to the higher intensity of electrochemical
reactions.

4.4. Cell operating pressure

In this section, inlet pressures of 1.5, 2.0, 2.5, and 3.5 atm were Fig. 5. Effect of (a) cathode stoichiometric ratio and (b) anode stoichiometric ratio
considered and the effect of cell pressure on polarization curve of on polarization curve.
 H. Askaripour / International Journal of Heat and Mass Transfer 144 (2019) 118705 9

Fig. 6. Effect of cell operating pressure on polarization curve.

4.5. Cathode and anode inlet humidity

To examine the effects of cathode and anode humidification,

inlet relative humidity of 0.55, 0.70, 0.85, and 1.0 were considered
and the fuel cell performance was studied. Fig. 7(a) and (b) shows
the effects of cathode and anode inlet humidity on polarization
curve, respectively. As seen from this figure, increasing cathode
humidity slightly affects the performance of fuel cell. The partial
pressure of water vapor in cathode flow channel increases with
an increase in cathode inlet humidity, which in turn decreases
the oxygen partial pressure. Variations in the oxygen pressure will
influence the open circuit voltage and activation overpotential.
Since open circuit voltage and activation loss are in relation to nat-
ural logarithm of the oxygen pressure, they are insignificantly
influenced by the variations in oxygen pressure which result in
the cell voltage remains unchanged. It can be also seen that
increasing the anode inlet humidity leads to the improvement in
fuel cell performance. The decrease in hydrogen partial pressure,
which resulted from the increase in anode humidity, causes the
open circuit voltage to change slightly owing to the logarithmic
relationship. In addition, higher anode humidity increases the Fig. 7. Effect of (a) cathode inlet humidity and (b) anode inlet humidity on
water content in membrane which results in the membrane con- polarization curve.

ductivity enhances. Due to improvement in proton transport

through the membrane, ohmic loss decreases and as a result, the
cell voltage and fuel cell performance increases.

4.6. Cell operating temperature

In this section, inlet gas temperatures of 75, 80, 85, and 90 °C

were considered and the effect of cell temperature on polarization
curve is depicted in Fig. 8. It can be observed that fuel cell perfor-
mance slightly changes with an increase of the inlet temperature at
low current densities; however, at moderate to high current densi-
ties, fuel cell performance significantly goes up with increasing
inlet temperature. The higher temperatures for inlet gas reinforce
amount of the water vapor produced in the fuel cell, which in turn
increase water activity and water content of membrane and as a
result, membrane conductivity increases and ohmic loss decreases.
On the other hand, open circuit voltage goes up with an increase in
gas temperature. Despite the higher temperatures increase the
activation overpotential, increasing open circuit voltage and
decreasing ohmic loss dominate the activation loss; consequently,
the fuel cell performance enhances. Fig. 8. Effect of cell operating temperature on polarization curve.
10 H. Askaripour / International Journal of Heat and Mass Transfer 144 (2019) 118705

5. Conclusions [6] Y. Wang, C.Y. Wang, A non-isothermal, two-phase model for polymer
electrolyte fuel cells, J. Electrochem. Soc. 153 (2006) A1193–A1200.
[7] T.E. Springer, T.A. Zawodizinski, S. Gottesfeld, Polymer electrolyte fuel cell
Over the past few years, many numerical studies have been pro- model, J. Electrochem. Soc. 138 (1991) 2334–2342.
posed to simulate the PEM fuel cell; however, heat transfer phe- [8] T.F. Fuller, J. Newman, Experimental determination of the transport number of
water in Nafion 117 membrane, J. Electrochem. Soc. 139 (1992) 1332–1337.
nomena were usually disregarded and fuel cell was isothermally
[9] D. Hyun, J. Kim, Study of external humidification method in proton exchange
simulated. In the present study, a model for a single PEM fuel cell membrane fuel cell, J. Power Sources 126 (2004) 98–103.
was developed and different heat sources and sinks including sorp- [10] W.M. Yan, F. Chen, H.Y. Wu, C.Y. Soong, H.S. Chu, Analysis of thermal and water
management with temperature-dependent diffusion effects in membrane of
tion and desorption, anode and cathode half-reactions, water
proton exchange membrane fuel cells, J. Power Sources 129 (2004) 127–137.
phase-change, and activation loss were taken into account. The [11] H. Wu, P. Berg, X. Li, Non-isothermal transient modeling of water transport in
model was able to predict the profile of key parameters along PEM fuel cells, J. Power Sources 165 (2007) 232–243.
the anode and cathode flow channels such as temperature, humid- [12] T.V. Nguyen, R.E. White, A water and heat management model for proton
exchange membrane fuel cells, J. Electrochem. Soc. 140 (1993) 2178–2186.
ity, pressure, cell current density, and so on. Based on this study, [13] J.J. Hwang, P.Y. Chen, Heat/mass transfer in porous electrodes of fuel cells, Int.
the following conclusions can be drawn: J. Heat Mass Transf. 49 (2006) 2315–2327.
[14] Y. Zong, B. Zhou, A. Sobiesiak, Water and thermal management in a single PEM
fuel cell with non-uniform stack temperature, J. Power Sources 161 (2006)
(1) The amount of heat produced (or consumed) through the 143–159.
activation loss, anode and cathode reactions, anode sorption, [15] T. Ackmann, L.G.J. de Haart, W. Lehnert, D. Stolten, Modeling of mass and heat
and cathode desorption monotonically decrease along the transport in planar substrate type SOFCs, J. Electrochem. Soc. 150 (2003)
A783–A789.
flow channels. [16] J. Yuan, M. Rokni, B. Sunden, Three-dimensional computational analysis of gas
(2) The cell performance is slightly changed with the variations and heat transport phenomena in ducts relevant for anode-supported solid
in stoichiometric ratio and inlet humidity of cathode side. oxide fuel cells, Int. J. Heat Mass Transf. 46 (2003) 809–821.
[17] J. Park, X. Li, Effect of flow and temperature distribution on the performance of
(3) Increasing the anode inlet humidity significantly enhances
a PEM fuel cell stack, J. Power Sources 162 (2006) 444–459.
the fuel cell performance. [18] Z. Liu, Z. Mao, C. Wang, A two dimensional partial flooding model for PEMFC, J.
(4) With an increase of the cell voltage, the heat resulting from Power Sources 158 (2006) 1229–1239.
[19] G. Karimi, A. Jamekhorshid, Z. Azimifar, X. Li, Along-channel flooding
activation overpotential, half-reactions, and sorption and
prediction of polymer electrolyte membrane fuel cells, Int. J. Energy Res. 35
desorption diminishes. (2011) 883–896.
(5) The higher cell pressures increase cell performance at low [20] G. Lin, W. He, T.V. Nguyen, Modeling liquid water effects in the gas diffusion
current densities, whereas worsen the cell performance at and catalyst layers of the cathode of a PEM fuel cell, J. Electrochem. Soc. 151
(2004) A1999–A2006.
moderate to high current densities. [21] W. Sun, B.A. Peppley, K. Karan, An improved two-dimensional agglomerate
(6) Increasing the anode stoichiometric ratio improves the cell cathode model to study the influence of catalyst layer structural parameters,
performance at moderate to high current densities. Electrochim. Acta 50 (2005) 3359–3374.
[22] X. Li, Principles of Fuel Cells, Taylor & Francis, New York, U.S.A, 2006.
(7) Increasing the cell temperature results in the fuel cell perfor- [23] H. Li, Y. Tang, Z. Wang, Z. Shi, S. Wu, D. Song, J. Zhang, K. Fatih, J. Zhang, H.
mance significantly goes up at moderate to high current Wang, Z. Liu, R. Abouatallah, A. Mazzab, A review of water flooding issues in
densities. the proton exchange membrane fuel cell, J. Power Sources 178 (2008) 103–
117.
[24] R.E. Treybal, Mass Transfer Operations, third ed., McGraw Hill, New York, 1990.
[25] S. Kakac, R.S. Shah, W. Aung, Handbook of Single-Phase Convective Heat
Transfer, Wiley, New York, 1987.
Declaration of Competing Interest [26] X. Zhang, J. Guo, J. Chen, The parametric optimum analysis of a proton
exchange membrane (PEM fuel cell) and its load matching, Energy 35 (2010)
The authors declared that there is no conflict of interest. 5294–5299.
[27] J. Ramousse, O. Lottin, S. Didierjean, D. Maillet, Heat sources in proton
exchange membrane (PEM) fuel cells, J. Power Sources 192 (2009) 435–441.
[28] T. Watari, H. Wang, K. Kuwahara, K. Tanaka, H. Kita, K. Okamoto, Water vapor
References
sorption and diffusion properties of sulfonated polyimide membranes, J.
Membr. Sci. 219 (2003) 137–147.
[1] A.J. Del Real, A. Arce, C. Bordons, Development and experimental validation of a [29] P. Atkins, J. de Paula, Physical Chemistry, seventh ed., Oxford University Press,
PEM fuel cell dynamic model, J. Power Sources 173 (2007) 310–324. 2002.
[2] C.K. Dyer, Fuel cells for portable applications, J. Power Sources 106 (2002) 31– [30] B.E. Conway, D.P. Wilkinson, Non-isothermal cell potentials and evaluation of
34. entropies of ions and of activation for single electrode processes in non-
[3] D.S. Falcão, V.B. Oliveira, C.M. Rangel, C. Pinho, A.M.F.R. Pinto, Water transport aqueous media, Electrochim. Acta 38 (1993) 997–1013.
through a PEM fuel cell: a one-dimensional model with heat transfer effects, [31] X. Zhang, J. Guo, J. Chen, The parametric optimum analysis of a proton
Chem. Eng. Sci. 64 (2009) 2216–2225. exchange membrane (PEM) fuel cell and its load matching, Energy 35 (2010)
[4] D. Natarajan, T.V. Nguyen, Three-dimensional effects of liquid water flooding 5294–5299.
in the cathode of a PEM fuel cell, J. Power Sources 115 (2003) 66–80. [32] X. Li, I. Sabir, J. Park, A flow channel design procedure for PEM fuel cells with
[5] H. Meng, C.Y. Wang, Model of two-phase flow and flooding dynamics in effective water removal, J. Power Sources 163 (2007) 933–942.
polymer electrolyte fuel cells, J. Electrochem. Soc. 152 (2005) A1733–A1741.