Electrochimica Acta 54 (2009) 2137–2154

Contents lists available at ScienceDirect

Electrochimica Acta
journal homepage: www.elsevier.com/locate/electacta

Fundamental understanding of liquid water effects on the performance

of a PEMFC with serpentine-parallel channels
Anh Dinh Le, Biao Zhou ∗
Department of Mechanical, Automotive & Materials Engineering, University of Windsor, Windsor, ON, Canada N9B 3P4

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

Article history: A three-dimensional and unsteady proton exchange membrane fuel cell (PEMFC) model with serpentine-
Received 23 September 2008 parallel channels has been incorporated to simulate not only the fluid flow, heat transfer, species transport,
Received in revised form electrochemical reaction, and current density distribution but also the behaviors of liquid water in the
27 September 2008
gas–liquid flow of the channels and porous media. Using this general model, the behaviors of liquid water
Accepted 8 October 2008
were investigated by performing the motion, deformation, coalescence and detachment of water droplets
Available online 25 October 2008
inside the channels and the penetration of liquid through the porous media at different time instants. The
results showed that: tracking the interface of liquid water in a reacting gas–liquid flow in PEMFC can be
Keywords:
PEMFC
fulfilled by using volume-of-fluid (VOF) algorithm combined with solving the conservation equations of
Fuel cell continuity, momentum, energy, species transport and electrochemistry; the presence of liquid water in
Liquid water the channels has a significant impact on the flow fields, e.g., the gas flow became unevenly distributed
Multi-phase due to the blockage of liquid water where the high pressure would be suddenly built up and the reactant
VOF gas transport in the channels and porous media would be hindered by liquid water occupation.
© 2008 Elsevier Ltd. All rights reserved.

1. Introduction In recent years, water management in PEMFCs has been

investigated by a large number of numerical models based on
Due to its high energy density and low operating temperature in the computational fluid dynamics (CFDs) approach. Typical CFD
comparison to other types of fuel cells, the proton exchange mem- models of PEMFCs which simultaneously considered the elec-
brane fuel cell (PEMFC) is one of the most promising fuel cells. trochemical kinetics, current distributions, hydrodynamics, and
With their zero/low emission, quiet operation, and especially high multi-component transport were conducted by Nguyen and co-
efficiency (up to 70%), PEMFCs are also considered to be very suit- workers [1,2], Wang and co-workers [3–6], Um and co-workers [7],
able for automotive and portable electronics applications. In reality, Wang and co-workers [8], Liu [9,10], Dutta and Van Zee [11,12], Dji-
the operation of a PEMFC involves simultaneous and complex pro- lali and co-workers [13,14], among others. These researches have
cesses such as multi-phase, multi-component fluid flow with heat significantly contributed to fuel cell simulation, especially investi-
and mass transfer as well as electrochemical reactions. It is obvious gating and predicting two-phase transport phenomena and water
that one of the most important processes significantly influencing management in PEMFCs. However, by far, to the authors’ knowl-
fuel cell performance is the water transport process, especially the edge, most of the two-phase numerical models have not considered
liquid water transport. The porous media might be dry due to the the interface tracking between liquid water and gas. The liquid
lack of humidified water, resulting in a low ionic conductivity in water in the two-phase models was introduced under the form of
the membrane. In contrary, too much water vapor would produce water saturation through the channels, gas diffusion layers (GDLs)
liquid water in form of droplets. These droplets may hinder the and catalyst layers but not the form of liquid water phase or liquid
mass transport of gases through porous diffusion layers and cata- droplets which remain during the operation of actual PEMFCs. The
lyst layers, hence reducing PEMFC performance. Therefore, water detailed behaviors of liquid water transport inside the PEMFCs were
management, especially liquid water management, to which many rarely discussed until the study [15] made in Zhou’s research group.
engineers and scientists have recently paid particular attention, has The approach to deal with liquid water behaviors in the cathode
been a critical challenge for a high-performance fuel cell design and of a single PEMFC or stack with serpentine or straight parallel fuel
optimization. cell channels with/without GDLs by applying volume-of-fluid (VOF)
technique was firstly developed by Zhou and co-workers [15–19].
The results showed that different designs of gas diffusion layers
∗ Corresponding author. Tel.: +1 519 253 3000x2630; fax: +1 519 973 7007. and flow channels will influence the liquid water flow patterns sig-
E-mail address: [email protected] (B. Zhou). nificantly, thus leading to changes in PEMFC performance. Later,

0013-4686/$ – see front matter © 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.electacta.2008.10.029
2138 A.D. Le, B. Zhou / Electrochimica Acta 54 (2009) 2137–2154

2.1. Governing equations

A three-dimensional and unsteady PEMFC model includes the

following principles of conservation: mass and momentum for
solving fluid flow, energy for solving heat generation and trans-
fer, species for solving species transport, charge for solving current
density distribution. In addition, a volume fraction equation was
used to track liquid water phase in a two-phase mixture. Those
equations are described in the following forms:
∂ε
Continuity equation : + ∇ · (εv) = Sm (1)
∂t

Momentum equation :
∂
(εv) + ∇ (εvv) = −ε∇ p + ∇ [ε∇ v] + Sv (2)
∂t

∂
Volume fraction equation : (εsl l ) + ∇ (εsl l vl ) = Ss (3)
∂t

Energy equation :
⎛ ⎞
∂T 
(cp )eff + (cp )eff (v∇ T ) = ∇ ⎝keff ∇ T − hj Jj + ( · v)⎠ + ST
∂t
j
(4)

Species transport equation :

Fig. 1. (a) Schematic diagram of PEMFC. (b) Computation domain. ∂
(εYi ) + ∇ · (εvYi ) = Di,m ∇ 2 (Yi ) + Si (5)
∂t
Djilali and co-workers [20,21] also investigated dynamic behav-
iors of liquid water entering a PEMFC channel through small holes Conservation of charge : ∇ · (s ∇ s ) = Ses (6)
that were assumed as pores in the GDLs using the similar method.
Although the method used in [15–21] can give very useful insights ∇ · (m ∇ m ) = Sem (7)
about water management in the cathode channels or GDL, it can- The physical, electrochemical, heat and mass transport properties
not directly report the liquid water effects on PEMFC performance are presented in Table 1 and the source terms Sm , Sv , Ss , ST , Si , Ses
because the electrochemical reactions were not taken into account. and Sem are summarized in Table 2.
Therefore a general mathematical model for PEMFC describing
all detailed physics involved, e.g., multi-components, multi-phase 2.2. Electrochemical reaction rate and electron transport
with VOF interface-tracking between gas-phase and liquid-water-
phase, heat and mass transport, electrochemical reactions, etc., has The transfer currents are the result of the electrochemical reac-
been proposed and successfully built by Le and Zhou [22]. It showed tions that take place in the catalyst layers, and are driven by the
that this general model can be used to simulate practical PEMFC. respective potential gradients for the electrons and protons [23]. In
In this paper, this general model with all parts of a single PEMFC Eqs. (6) and (7),  s and  m are the electronic conductivity in the
including the membrane, catalyst layers, GDLs, gas flow channels solid phase and ionic conductivity in the membrane phase, respec-
and current collectors is applied to investigate a single PEMFC with tively. s represents the solid phase potential (V) and m represents
serpentine-parallel flow channels. the membrane phase potential (V). The source terms representing
the transfer current are calculated by Butler–Volmer equation:
2. Numerical model   an
ref CH2 ˛a,a F an ˛c,a F an
Ran = Ran exp − exp − (8)
The computation domain consists of current collectors, flow ref
CH RT RT
channels, GDLs, catalyst layers and membrane as shown in Fig. 1. 2
  cat
The gaseous species are considered to be a mixture of hydrogen and
ref CO2 ˛a,c F cat ˛c,c F cat
water vapor (humidified hydrogen) at the anode and a mixture of Rcat = Rcat exp − exp − (9)
ref
CO RT RT
oxygen, nitrogen and water vapor (humidified air) at the cathode. 2
The assumptions used in the model are described as follows:
In order to satisfy the conservation of charge, the total current
of either electrons or protons coming out from the anode catalyst
• Ideal gas law was employed for gaseous species.
layer must be equal to the total current coming into the cathode
• The fluid flow in the fuel cell was laminar.
catalyst layer, which must be equal to the total current caused by
• The porous media including membrane, catalyst layers and GDLs
the proton movement through the membrane [24]:
were considered to be isotropic.
• The fuel cell cooling was controlled by forced convection heat
transfer. (10)
 A.D. Le, B. Zhou / Electrochimica Acta 54 (2009) 2137–2154 2139

Table 1
List of formulae of related parameters used in the model.

Formulae of physical parameter

Parameter Value/formulation

Mixture density  = sl l + (1 − sl )g

Volume fractions relation sl + sg = 1
Mixture viscosity  = sl l + (1 − sl )g
s  v +s  v
Local mass average velocity of mixture v = l s l l +sgg gg g
l l
Effective volumetric heat capacity of porous media (cp )eff = εf cp,f + (1 − ε)s cp,s
Effective thermal conductivity of porous media keff = εks + (1 − ε)kf

Formulae of heat transport parameter

Parameter Value/formulation

Heat transfer rate due to forced convection Q̇convection = hAs (Tsurf − T∞ ) (W )

Heat generation rate by the fuel cell Q̇generated = I(1.25 − Vcell ) (W )

Formulae of species transport parameter

Parameter Value/formulation
n d MH O
Water transport through the membrane by electro-osmotic drag ṁH2 O = F
2
Rcat
Electro-osmotic drag coefficient nd = 2.5
22  
p0 Tp
T
Gaseous diffusivity of specie i Di,m = ε1.5 D0 p T0
i,m
Membrane ionic conductivity m = ε(0.514 − 0.326)e1268(1/303−1/T )
= 0.043 (a < 0)
= 0.043 + 17.18a − 39.85a2 + 36a3 (0 ≤ a ≤ 1)
Water content
= 14 + 1.4(a − 1) (1 ≤ a ≤ 3)
= 16.8 (a > 3)
Water activity a = PPw = XPw P + sl
sat sat
Psat = 101325
Saturated pressure −5
∗(T −273.17)2 +1.4454∗10−7 ∗(T −273.17)3 )
∗10(−2.1794+0.02953∗(T −273.17)−9.1837∗10
Yi 
Species concentrations Ci = Mi

Formulae of phase change parameter

Parameter Value/formulation
 Pwv −Psat

Phase change rate (condensation/evaporation) rw = cr max (1 − sl ) RT
MH2 O , −sl l

Formulae of charge transport parameter

Parameter Value/formulation

Activation overpotential = s − m − Vref

Cell potential Vcell = s,cat − s,an

Average current density

2.3. Water transport gas and liquid is tracked by the volume fraction of liquid water
in the computational cell volume. In the VOF approach, the source
Water is generated in the cathode catalyst layer by electrochem- term of continuity and momentum equations used in porous media
ical reactions. Due to the proton movement from the anode to the include the effects of surface tension and wall adhesion.
cathode through the membrane, water molecules are pulled by
the protons due to an electro-osmotic drag force [25]. Additionally, 2.4.1. Geometric reconstruction scheme
water may diffuse through the membrane due to the concentra- The geometric reconstruction PLIC scheme (piecewise linear
tion differences. The water content also influences the properties interface construction) was employed because of its accuracy and
of membrane such as water diffusivity, ionic conductivity, etc., thus applicability for general unstructured meshes, compared to other
influencing the fuel cell performance. The formulae of membrane methods such as the donor–acceptor, Euler explicit, and implicit
water diffusivity and membrane ionic conductivity are described schemes. A VOF geometric reconstruction scheme is divided into
in related expressions [25,26] listed in Table 1. two parts: a reconstruction step and a propagation step. Details
The net water flux through the membrane results in a water can be found in [23,27].
balance between the electro-osmotic drag of water (from anode to
cathode) and back diffusion (from cathode to anode)
2.4.2. Implementation of surface tension and wall adhesion
MH2 O The addition of surface tension to the VOF method is modeled
os
J w = Jw + Jw
diff
= nd I − D ∇ (Y ) (11)
i,m i
F by a source term in the momentum equation. The pressure drop
across the surface depends upon the surface tension coefficient
2.4. Volume-of-fluid algorithm [23,27]

The VOF technique was implemented in the channels and porous 1 1

p= = + (12)
media (including GDLs and catalyst layers) [23]. Interface between R1 R2
2140 A.D. Le, B. Zhou / Electrochimica Acta 54 (2009) 2137–2154

Table 2
The source terms of governing equations.

Governing equation Volumetric source terms and location of application

Conservation of mass For gas channels, GDL and the membrane: Sm = 0

MH n d MH O
For anode catalyst layer: Sm = − 2F
2
Ran − F
2
Ran

MO MH O nd M H O
For cathode catalyst layer: Sm = − 4F
2
Rcat + 2
2F
Rcat + F
2
Rcat

Volume fraction For gas channels and porous media: Ss = rw

Conservation of For gas channels: Sv = 0

 2 2∇ sl
momentum For GDLs and void of catalyst layers: Sv = g − 
g ε v +  (l +g )
 2∇ s 
For membrane: Sv = g − g ε2 v +   +gl
l
+ p cf nf F ∇ m
2
Conservation of energy For current collectors: ST = I s
For gas flow channel: ST = rw hL
I2
For GDL: ST = eff
+ rw hL
s
I2
For membrane: ST = eff
+ rw hL
m

1 1
For catalyst layer: ST = Ran,cat + I 2 eff
+ eff
+ rw hL
s m
MH
Hydrogen transport For anode catalyst layer: SH2 = − 2F
2
Ran
MO
Oxygen transport For cathode catalyst layer: SO2 = − 4F
2
Rcat

nd M H O
Water vapor transport For anode catalyst layer: SH2 O = − F
2
Ran − rw

MH O nd M H O
For cathode catalyst layer: SH2 O = 2
2F
Rcat + F
2
Rcat − rw

Ses = −Ran
Conservation of charge For anode catalyst layer:
Sem = Ran
Ses = Rcat
For cathode catalyst layer:
Sem = −Rcat
Ses = 0
For other parts:
Sem = 0

where denotes the surface tension coefficient,  is the surface ating temperature. Contrarily, liquid water is evaporated when
curvature; R1 and R2 are the two radii, in orthogonal directions, the water vapor pressure is smaller than its saturated value.
to measure the surface curvature. The surface tension force can be The phase change process (condensation/evaporation) is also an
written in terms of the pressure jump across the interface, which is important factor in determining the presence of liquid water in
expressed as a volumetric force F added to the momentum equation
∇ sl
F =  (13) Table 3
(l + g )/2 Geometrical properties and operation conditions.

where the term /(1 + 2 )/2 is used to improve the continuum Physical properties or parameters Value
surface force (CSF) method’s capability of modeling surface ten- Total channel length 0.115 m
sion in the case for fine meshes at high density ratio interfaces. Channel width 0.001 m
This method is called “density-scaling of the CFS”. It helps to main- Channel height 0.001 m
Membrane thickness 50 × 10−6 m
tain a constant interface thickness when the interface between a
GDL thickness 300 × 10−6 m
dense phase (namely, liquid water) and a light phase (namely, air) Catalyst layer thickness 10 × 10−6 m
is tracked by using VOF algorithm [28]. The source terms for differ- Anode and cathode inlet temperature 300 K
ent regions of the fuel cell are given in Table 2. The surface curvature Anode inlet excess coefficient, ωa 3
 can be defined in terms of the divergence of the normal unit vector Cathode inlet excess coefficient, ωc 3
Open-circuit voltage, Voc 1.15 V
of the interface n̂
Faraday constant, F 96,487,000 C kmol−1
8314 J kmol−1 K−1
 = ∇ · n̂ = ∇ · (n̂ss cos ss + t̂ss sin ss ) (14) Gas constant, R
Anode volumetric reference exchange current 7 × 1010 A kmol−1
ref ref an
where n̂ is the unit vector normal to the interface between two density/reference hydrogen concentration, Ran /(CH )
2
phases near the solid surfaces, n̂ss is the unit vector normal to the Cathode volumetric reference exchange current 7 × 105 A kmol−1
ref ref cat
solid surfaces, t̂ss is the unit vector tangential to the solid surfaces, density/reference oxygen concentration, Rcat /(CO )
2

and  ss is the static contact angle at the solid surfaces. For the elec- Anode transfer coefficient for anode reaction, ˛a,a 0.5
Anode transfer coefficient for cathode reaction, ˛a,c 0.5
trode surfaces with different wetabilities, different static contact
Cathode transfer coefficient for anode reaction, ˛c,a 0.5
angles could be assigned, and different contact angles could result Cathode transfer coefficient for cathode reaction, ˛c,c 0.5
in different surface tensions (F), thus influencing the water trans- Anode concentration dependence, an 0.5
port. The values of surface tension and contact angle used in this Cathode concentration dependence, cat 1.0
Factor accounts for energy release, ˇ 0.5
model are listed in Table 3.
Membrane porosity, εmem 0.5
Diffusion layer porosity, εgdl 0.5
2.5. Phase change Catalyst layer porosity, εcatalyst 0.5
Permeability of porous media,  1.76 × 10−11 m2
Liquid water is considered to appear in the fuel cell when the Contact angle,  w 90◦
Surface tension, 0.065 N m−1
water vapor pressure reaches its saturated value at the cell oper-
 A.D. Le, B. Zhou / Electrochimica Acta 54 (2009) 2137–2154 2141

Table 4 that of hydrogen/oxygen theoretically consumed. Then, inlet veloc-

Boundary conditions.
ity expression is given by the following equations:
Locations of application Boundary conditions
Iave Asurf
u = Uan,in
UH2 ,in = ωH2 MH2 (15)
2FH2 Ain
YH2 = YH2 ,in
Inlet of the anode flow channel
YH2 O = YH2 O,an,in Iave Asurf
Tan,in = TH2 ,in UO2 ,in = ωO2 MO2 (16)
4FO2 Ain
u = Ucat,in
YO2 = YO2 ,in Outlet of flow channels: the fully developed condition was
Inlet of the cathode flow channel
YH2 O = YH2 O,cat,in
assumed to be applied for the velocity field and species concen-
Tcat,in = Tair/O2 ,in
∂uout,an trations.
=0
∂x The terminal collectors: the anode and cathode collectors were
∂YH2 ∂YH2 O,an
Outlet of the anode flow channel = 0; =0 connected to the external electric circuit. It was necessary to
∂x ∂x
∂Tan,out determine the boundary conditions for phase potentials on these
=0
∂x interfaces.
∂uout,cat
=0 External boundaries: external boundaries are defined as all out-
∂x
∂YO2 ∂YH2 O,cat side surfaces of PEMFC except the terminals (the outside surfaces
Outlet of the cathode flow channel = 0; =0
∂x ∂x of current collectors where the current entry to or departure from).
∂Tcat,out
=0 Zero-current-flux condition was applied for the external bound-
∂x
s = 0 aries due to no currents coming or leaving.
The anode terminal ∂m Cooling assumptions: for the sake of simplicity, the cooling chan-
=0
∂y
s = Vcell
nel was not included in the present model. However, it is necessary
The cathode terminal ∂m to control the cell temperature. In order to ensure that the forced
=0
∂y convection dissipates the heat generated from the electrochemical
∂s ∂ ∂ ∂m
External boundaries ∂x
= 0; ∂zs = 0; ∂xm = 0; ∂z
=0 reaction (described in Table 1); the following condition must be
satisfied [29]:
multi-phase model. In this study, the variation of water mass flow
Q̇convection = Q̇generated (17)
rate due to condensation/evaporation was considered as described
in Table 1. Then convective coefficient h was chosen as
I(1.25 − Vcell )
2.6. Heat transfer h= (18)
As (Tsurf − T∞ )

A general conservation equation of energy is proposed to solve It should be noticed that it is necessary to have an estimated
the heat generation and heat transfer processes for any domain value of surface temperature, Tsurf , of the fuel cell to calculate the
in the fuel cell. Eq. (4) describes effective properties to account convective heat transfer coefficient in Eq. (18). In this study, it was
for the porous media parts (GDLs, catalyst layers and the mem- assumed that the temperature difference between the fuel cell sur-
brane). It also includes the solid zone and fluid zone in the porous face and the ambient is about 50 ◦ C when estimated the convective
media. The first three terms on the right-hand side of Eq. (4) rep- heat transfer coefficient.
resent energy transfer due to conduction taking place in solid and
fluid zones, species diffusion, and viscous dissipation, respectively 2.8. Numerical procedures
[23]. For the collectors what are pure solid materials, the species
diffusion, viscous dissipation and convection term are negligible. The computation domain consists of current collectors, flow
For the channel and porous media, those three terms were con- channels, GDLs, catalyst layers and membrane. Fig. 1 shows that the
sidered since the fluid flow is presented in such regions. Note that
the porosity specified in the energy equation determines different
zones, depending on its values: 1 for the pure fluid zone (the chan-
nels), 0 for pure solid zone (the collectors and ribs), and between 0
and 1 for porous zones. The source term in the energy conservation
includes heat from chemical reactions (only in the catalyst layers),
Ohmic heating due to Ohmic resistance of solid zones, and phase
change (condensation/evaporation processes). Hence, the sources
terms of energy equation for different layers are different that listed
in detail in Table 2.

2.7. Boundary conditions

The boundary conditions are required to close the system of

equations including conservation equations of continuity, momen-
tum, volume fraction, energy, species, and charge with the
unknowns: u, v, w p, sl T, Yi and s , m . The summary of boundary
conditions is listed in Table 4. Below a brief description is provided.
Inlet of flow channels: inlet velocities can be preliminarily calcu-
lated by the inlet flow rates based on the average current density
(Iavg ) and excess coefficient ω. The excess coefficient is defined as
the ratio of the molar flow rate of total hydrogen/oxygen supplied to Fig. 2. A series of water droplets was initially suspended into the cathode channel.
2142 A.D. Le, B. Zhou / Electrochimica Acta 54 (2009) 2137–2154

Fig. 3. The motion, deformation, coalescence and detachment of droplets versus time in the cathode channel from t = 1.00025 s to t = 1.00275 s.
 A.D. Le, B. Zhou / Electrochimica Acta 54 (2009) 2137–2154 2143

Fig. 4. Flow velocity vectors and water droplets distribution in the middle-plane of the cathode channel (Y = 0.0015 m).

channels are serpentine-parallel, total path length for a channel of Fig. 2. The reactant flow was supplied to the cathode inlet and hence,
115 mm with 1-mm width × 1-mm height in the cross-section flow impacting the liquid drops suspending in the channel. As time pro-
area. The set of coupled governing equations and relative equations gressed, the development of droplet deformation (elongation) in
were solved in Fluent® 6.2 [23] by writing own user-defined- the flow direction is due to a balance of droplet surface tension
functions (UDFs). The UDFs written in C language are used to define and shear stress from surrounding gas flow, as can be observed
the physical parameters, source terms and boundary conditions in in Fig. 4 at different time instants. In other words, the enlarge-
the equations. There were 247,000 grid cells in the computation ment of droplet surface area increased the surface tension such that
domain. The solution procedure for pressure–velocity coupling was the force balance on the droplets can be achieved. From t = 1.0 s to
based on PISO algorithm [30]. t = 1.001 s, the deformation of the droplets distributed in the two
branches of inlet channel is almost symmetric due to the symme-
3. Results and discussion try of channel shape and flow direction and magnitude. This could
be easily seen from the flow velocity vectors and water droplets dis-
To observe the presence of liquid water in the porous media tribution in the middle-plane of the cathode channel (Y = 0.0015 m)
and channels as well, a long enough physical time period in the with t = 1.00025 s and 1.00075 s (Fig. 4). As the time progressed after
unsteady cases is required—usually needs to have at least several- milliseconds, the deformed droplets that enter the turning area
minute operation in experiments, and therefore, consuming a large from the straight channels hit the turning-wall surface. As a result,
amount of CPU (computer processing unit) calculation time. In the deformed droplets were further elongated and the strong “turn-
order to save computing time, an unsteady multi-phase model ing” gas flow breaks these droplets into even smaller ones sticking
was simulated from the time t = 0 s to t = 1.0 s. At specific times on the turning-wall surface. Such water droplets would coalesce,
t = 1.0 s and t = 1.002 s, a series of liquid water droplets was ini- then spread and deform into “water bands” due to the shear stress
tially suspended on the cathode channel to further investigate from “turning” gas flow and pressure drop, as shown in Fig. 3 at
two-phase flow behavior, especially liquid water behavior across t = 1.002 s. Note that, at this moment, a series of water droplets were
porous media, together with electrochemical reaction, heat and further added into the channel. As the time progressed, as shown
mass transfer phenomena. Fig. 2 shows the arrangement of the ini- in Fig. 3 at t = 1.0025 s, the water bands were broken again into
tial liquid water distribution. There were 8 equidistant spherical smaller droplets distributed over the straight channel wall since
droplets (r = 0.4 mm) freely suspended along the cathode channel these bands left the turning area. Subsequently, the formed small
at the two different times. The calculation time step is set to be water drops are easily removed and distributed everywhere on the
10−5 s for VOF tracking algorithm. The cell voltage was constantly straight channel walls. However, the deformation of the droplets
set to be 0.5 V. The operating condition and fuel cell parameters are distributed in the two branches of inlet channel was not symmet-
listed in Table 4. ric anymore after the moment t = 1.0075 s as shown in Fig. 4. This
is due to the fact that the water droplets at the turning area not
3.1. The motion, deformation, coalescence, and detachment of only affected the flow distribution in the downstream, but also
liquid water inside the cathode channel influenced the flow distribution in its upstream. As a result, the
interaction between the flow and droplets is different for the two
Fig. 3 shows the motion and deformation of droplets versus time branches of the cathode channel.
on the cathode channel of the PEMFC. At t = 1.0 s, the droplet placed At t = 1.010 s as shown in Fig. 5, the small droplets and slugs
at the channel inlet was in its initial spherical shape as shown in were scattered in the channel. In addition, the coalescence of water
2144 A.D. Le, B. Zhou / Electrochimica Acta 54 (2009) 2137–2154

Fig. 5. The motion, deformation, coalescence and detachment of droplets versus time in the cathode channel from t = 1.010 s to t = 1.050 s.
 A.D. Le, B. Zhou / Electrochimica Acta 54 (2009) 2137–2154 2145

Fig. 6. The motion, deformation, coalescence and detachment of droplets versus time in the cathode channel from t = 1.056 s to t = 1.070 s.

Fig. 7. Water mass flux increased on the mid-plane of the membrane (Y = 0.002335 m) at t = 1.010 s: (a) due to electro-osmotic drags; (b) due to diffusion in the X-direction;
(c) due to diffusion in the Y-direction; and (d) due to diffusion in the Z-direction.
2146 A.D. Le, B. Zhou / Electrochimica Acta 54 (2009) 2137–2154

Fig. 8. The distributions on the cathode channel and porous media at t = 1.010 s in the case without the presence of liquid water: (a) pressure; (b) temperature; (c) oxygen
mass fraction; and (d) water vapor mass fraction.

droplets was also observed. First, water droplets were detached tine channel and details about this will be published in a separate
by advection from the gas flow. Consequently, these droplets may paper.
encounter other water droplets downstream. As a result, some
bigger droplets were formed. This growth of the droplets would 3.2. Water transport in the membrane
increase the volume of liquid water in some local cross-sections
of the channel, resulting in a possible blockage. Again, under the Water transport through the membrane due to electro-osmotic
high-pressure gradient across the blockage gap, the shear stress drag force and diffusion process was performed via water mass flux
of the flow on the droplets would force water droplets to expand J as defined in Eq. (11) and shown in Fig. 7. Fig. 7a shows water mass
and deform into “water bands”. For a small droplet in the channel, flux due to electro-osmotic drag which pulls water molecules from
the cross-section-area of the channel blocked by the small water the anode to the cathode through the membrane while Fig. 7b–d
droplet is smaller, thus the pressure difference across the blocking shows the water mass flux due to the diffusion in the X-, Y-, and Z-
cross-section would be smaller. One more thing is mentioned that a directions on the middle-plane of the membrane (Y = 0.002335 m).
few water droplets are kept in the corner of the channel throughout Note that positive value of the value of water flux due to electro-
the process due to the fact that the shear stress of the flow in the osmotic drag means that the flux vectors from the anode side,
corners is insignificant comparing with the wall adhesion effects through the membrane, to the cathode side. However, the value of
around the corner. water flux due to diffusion may be positive or negative, depend-
More than 40% of liquid water amount was moved out of the ing on the water concentration gradients of different regions in
channel by the gas flow at t = 1.060 s. At this moment, it could the porous media. Conventionally, the positive sign of water dif-
be observed that there were small droplets remaining in cor- fusion flux means the water diffused in the positive directions in
ners of the channel without moving to other places as shown in the 3D-coordinate and the negative sign means water diffused in
Fig. 6. A portion of liquid water retained in the porous media. negative directions. As can be seen in Fig. 7b and c, water diffu-
This is the fact that the gas flow velocity in these regions is insuf- sion in X- and Z-directions (or X–Z planes) is insignificant and takes
ficient to have a strong shear stress to force the small droplets place around the channels and the ribs, compared with water dif-
that is of high surface tension. The liquid water in the channel is fusion in Y-direction between cathode side and anode side. Fig. 6c
gradually removed with a trivial amount, as time progressed to also shows that a large portion of water flux is positive, accounting
t = 1.070 s, which means that the operation conditions somehow for the fact that a large water amount diffused through the mem-
reach a “stable” state. This phenomenon was also observed from brane from cathode to anode. The highest values of water mass
experiments for a single transparent PEMFC with single serpen- flux concentrate in the regions underneath the ribs where water
 A.D. Le, B. Zhou / Electrochimica Acta 54 (2009) 2137–2154 2147

Fig. 9. The distributions on the cathode GDL/catalyst layer interface (Y = 0.0023 m) at t = 1.010 s in the case without the presence of liquid water: (a) pressure; (b) temperature;
(c) oxygen mass fraction; and (d) water vapor mass fraction.

was built-up during fuel cell operation. Moreover, along the chan- tributed in the fuel cell at a specified time were presented via
nel from the inlet to the outlet, the direction of water diffusion volume fraction of water. By tracking the presence of volume frac-
is opposite. One can see if the water diffuses from anode to cath- tion of water, it could be determined where is the liquid water and
ode at the inlet region then it conversely diffused from cathode the effects of presence of liquid water on transport parameters and
to anode towards the downstream along the channel. Comparing characteristics of the fuel cell. Fig. 10a shows volume fraction of
Fig. 7a and c, the water flux due to electro-osmotic force from cath- liquid water in the cathode in three dimensions and Fig. 10b and c
ode to anode is higher that due to diffusion from cathode to anode. shows volume fraction of liquid water in the cathode on the middle
It is obvious right since the electro-osmotic force is proportional to planes of cathode channel and cathode GDL/catalyst layer interface,
the current density of the cell. The higher current density, the more respectively (The blue lines in Fig. 10c indicates the position of the
electro-osmotic water flux dominates. It is very significant since channels and the ribs on top of the porous media). It should be
water transport management through the membrane is important noticed that a small amount of liquid water moved through porous
to balance water amount in the porous media when the fuel cell media at the region under corners of the channel. This is reason-
operates at high current density values. able due to the fact that the corners and turning regions caused a
high resistance to the gas flow in the channel and the motion of
3.3. Effects of liquid water on velocity physical and transport liquid droplets as well. Therefore, a portion of water liquid tends to
properties move through porous media instead of moving forward along the
channel to the exit.
3.3.1. Physical and transport properties of the fuel cell in the case
without the presence of liquid water 3.3.3. Pressure field
Figs. 8 and 9 show distributions of pressure, temperature and Fig. 11 shows the pressure distribution in the cathode channel
species mass fractions in the cathode channel and porous media at and porous media at 1.01 s. It could be seen that the magnitude
1.01 s without the presence of liquid water. Since no liquid water of pressure is high at the inlet and decreases along the channel to
was taken into account, the system is treated as single, homoge- the outlet. However, around the location of the droplets, the pres-
neous phase with multi-components. Pressure, velocity and species sure drop reaches high values due to the resistance caused by liquid
mass fraction distributions mostly depend on the transport and water in gas flow. The pressure peaks were obtained at the location
properties of single-phase flow (namely, gas mixture phase). There- in which liquid water took place with highest values of volume
fore, these distributions have no unexpected changes that are due to fraction. Obviously, the presence of water liquid caused a dramatic
the presence of liquid phase. The comparison between single-phase change on pressure distribution in the cathode. Compared with the
and two-phase VOF models is given in later sections when physical case without taking the liquid water into account (Fig. 8), the peak
and transport properties in the case that liquid water presents. of pressure distribution in the case including water liquid is much
higher. And also the pressure change is much smoother in the case
3.3.2. Volume fraction of liquid water without liquid water. This also explains the inaccuracy and unre-
Fig. 10 displays the distribution of volume fraction of liquid ality of single-phase PEMFC models since that cannot simulate the
water in the cathode channel at 1.01 s. The liquid droplets dis- liquid water presence while it is a fact for real PEMFC operation.
2148 A.D. Le, B. Zhou / Electrochimica Acta 54 (2009) 2137–2154

3.3.4. Liquid water effects on concentrations

Figs. 12 and 13 show the mass fraction distribution of species
in the cathode and on the cathode GDL/catalyst layer interface
(Y = 0.0023 m). The oxygen fraction decreased in two ways: along
the cathode channel from the inlet to the outlet and from the chan-
nel to the catalyst layer due to the oxygen consumption on the
catalyst layer. Mass transport of oxygen from the channel to the
GDL was mainly driven by its concentration gradient (diffusion)
and this effect was also partially influenced by the velocity com-
ponent (convection) inside the porous media. The flow in cathode
channel enhanced the gas transport in the region under the chan-
nel, rather than in the region under the rib (the land area). As
shown in Figs. 12a and 13a, high oxygen region under the cath-
ode channel was observed. On the other hand, the land area had
a low amount of oxygen that was presented by oxygen mass frac-
tion. Moreover, the oxygen mass fraction is too low in the corners
of porous media where oxygen transport is limited by both con-
vection and diffusion. It could be imagined that the corners are
considered as “dead-end” zones in which the flow field has no way
to go and species concentration are insufficiently supplied. Regard-
ing the two-phase flow, the presence of liquid water significantly
influences the oxygen mass fraction in the channel as well. The oxy-
gen mass fraction is quite low at places occupied by water droplets.
The higher volume fraction of liquid water occupies, the lower the
oxygen mass fraction could be observed. It could be seen in Fig. 13a
that a small amount of oxygen was replaced by a corresponding
amount of liquid water (the regions were circled), resulting in a
reduction of oxygen concentration.
On the other hand, the hydrogen mass fraction gradually
increased from the inlet to the outlet as shown in Figs. 12b and 13b.
The difference of the trend of mass fraction change along the
channel for oxygen and hydrogen could be explained by the dif-
ferent compositions of the gas mixtures in cathode and anode and
their changes along the channel. Note that the anode gas mixture
includes only two species: hydrogen and water vapor meanwhile
the cathode gas mixture includes three species: oxygen, water
vapor and nitrogen. It could be noticed that in the anode catalyst
layer, an amount of water vapor is consumed by penetrating to the
cathode side due to the electro-osmotic drag, resulting in a decrease
of water mass fraction along the channel. As a result, hydrogen mass
fraction must increase along the channel because the sum of mass

Fig. 10. The distribution of volume fraction of liquid water at t = 1.010 s: (a) on the
cathode channel and porous media; (b) on the middle-plane of cathode channel
(Y = 0.0015 m); and (c) on the cathode GDL/catalyst layer interface (Y = 0.0023 m).

Fig. 11. Pressure distribution on the cathode channel and porous media at t = 1.010 s.
 A.D. Le, B. Zhou / Electrochimica Acta 54 (2009) 2137–2154 2149

Fig. 12. Species mass fraction distribution on the cathode channel and porous media at t = 1.010 s: (a) oxygen in the cathode; (b) hydrogen in the anode; (c) water vapor in
the cathode; and (d) water vapor in the anode.

fractions of species in the anode is one. In addition, there is no density vectors (defined as vectors whose magnitude is the electric
appearance of liquid water in the anode and so the influence of current per cross-sectional area) can be performed via the electron
liquid water on hydrogen mass transport cannot be seen. current density vectors with reverse direction. Electrons are first
The water vapor mass fraction increased along the cathode generated in the anode catalyst layer by the hydrogen oxidization
channel by two sources: by-product water was generated from the reaction, and then the electrons moved through an external circuit
electrochemical reaction at the cathode and the transport of water to the cathode terminal. When reaching the cathode catalyst layer,
from the anode to the cathode due to the electro-osmotic drag. the electrons recombine with protons and oxygen to produce water.
Figs. 12c and 13c show that there was a high water remaining in In fact, the current generated in the fuel cell is caused by the move-
the land area at the cathode as a result of water build-up due to the ment of electrons from anode to cathode (namely, in this model,
lower velocity. In addition, it could be seen that water vapor con- the electron movement is in Y-direction), then it can be determined
centration is highest in the edges and corners of the porous media. using Ohm’s law as follows:
As mentioned above, convection and diffusion are limited in these
regions due to “dead-end” effect. Therefore, water vapor was built Ie,y = −s ∂s (19)
∂y
up and not easy to be removed from there. This phenomenon is
the same for both cathode and anode porous media as shown in In the anode porous media, electron currents tend to directly
Figs. 12d and 13d. In addition, similar to the distribution of oxygen move along the Y-direction to the anode terminal where is con-
mass fraction in the cathode with presence of liquid water, a por- nected to the external circuit. However, in the anode porous region
tion of liquid water also affects on the water vapor concentration under the channel where no electrically conductive material is
by increasing the mass fraction in the region that marked by circles available to directly transfer the electrons to the collector, the elec-
as shown in Fig. 13c. tron currents must detour to the collector in different directions.
Similarly, in the cathode side, electron currents tend to directly
move to the cathode porous media along the Y-direction from the
3.3.5. Current density distribution cathode terminal. Because of the channel (it is physically a void
The current density vectors in Y-direction on the X–Z plane are space and electrons cannot transport through it), the electron cur-
described in Fig. 14. In a PEMFC, current is determined by the elec- rents must detour to the region in the cathode catalyst layer in
tron movement through solid conductive materials (including the which the chemical reaction takes place. This also accounts for the
collectors and solid part of porous media). Because the electron car- reason why the current density vectors in Y-direction under the
ries negative charge in solid phase, so the electron current is in the channel is usually smaller than those under the ribs as shown in
direction opposite that of the electrical current. Therefore, current Fig. 15a.
2150 A.D. Le, B. Zhou / Electrochimica Acta 54 (2009) 2137–2154

Fig. 13. Species mass fraction distribution at t = 1.010 s: (a) oxygen on the cathode GDL/catalyst layer interface (Y = 0.0023 m); (b) hydrogen on the cathode GDL/catalyst
layer interface (Y = 0.00237 m); (c) water vapor on the cathode GDL/catalyst layer interface (Y = 0.0023 m); and (d) water vapor on the cathode GDL/catalyst layer interface
(Y = 0.00237 m).

Effect of liquid water on current density distribution was inves- 3.3.6. Temperature distribution
tigated in this model as displayed in Fig. 15a (the case with liquid Fig. 16 shows the temperature distribution in the cathode chan-
water) and Fig. 15b (the case without liquid water) with the com- nel and porous media. The heat generation in the fuel cell was
pared regions that marked by circles. Even with a small amount, due to the electrochemical reaction and Ohmic heating. Note that
liquid water may hinder the current movement, resulting in a low the heat generated by electrochemical reactions has taken place in
current density in the region where the water liquid is placed. the catalyst layers and the heat caused by Ohmic heating occurs
In addition, the current density distribution in terms of the inte- in the zones in which the current goes through depending on the
grated cathode catalyst-layer current density presented in Fig. 15c zone materials. Therefore, the highest temperature concentrates
was defined as follows: in the catalyst layers in which the reactions take place and the
current is generated. The sum of heat sources would transfer to
I = Rcat,an tlayer (20) other regions by heat conduction (in solid materials) and/or heat
convection (in the channels). By controlling the convective coeffi-
Since the current density is proportional to the reaction rate as
cient h and ambient temperature at 300 K to cool down external
defined [26]:
walls of the fuel cell, the average cell temperature was approxi-
I = nF (21) mate 318 K. It could be observed that the presence of liquid water
also affects the distribution of cell temperature by absorbing heat
The reaction rate is intuitively defined as how fast a reaction from the surroundings and hence it locally decreases tempera-
takes place and it depends on oxygen consumption rate in the cath- ture in the liquid water region, as shown in Fig. 16a. In contrary,
ode catalyst layer as well. Fig. 15c shows that the current density there is no liquid water presented in the anode (Fig. 16b) or in the
distribution under the cathode channel is usually higher due to cathode (Fig. 8b), resulting in a less-change distribution of temper-
higher oxygen concentration. On the other hand, it can be indicated ature over the cell excluding the region near the inlet where the
that the current density vectors distribution presented by the pro- local temperature is directly affected by the inlet gas flow temper-
ton movement through the membrane (Fig. 14b) is the same with ature.
the integrated cathode catalyst-layer current density distribution.
Note that the proton current is defined as
3.3.7. Phase change
Ip,y = −m ∂m (22)
∂y As discussed in Section 2.5, the phase change process was con-
sidered in terms of the variation of water mass flow rate due
In a similar way, the proton current density vectors in the mem- to condensation/evaporation process with the presence of liquid
brane region under the channel are higher than those in the water. If the partial pressure at a certain cell is smaller than satu-
membrane under the ribs due to larger reaction rate. rated pressure at the same location then an amount of water liquid
 A.D. Le, B. Zhou / Electrochimica Acta 54 (2009) 2137–2154 2151

In fact, the phase change from the gas phase to the liquid phase
can be taken place in the cathode catalyst layer where water is
generated and water concentration in this place is higher than the
others places. However, it is possible to observe vapor-to-liquid
process if the calculation time is long enough.

Fig. 14. (a) Distribution of the local current density vectors on Y–Z cross-sectional
plane at X = 0.01 m. (b) Distribution of the proton current vectors in the membrane on
Y–Z cross-sectional plane at X = 0.01 m. (c) Distribution of the proton current vectors
in the anode catalyst layer on Y–Z cross-sectional plane at X = 0.01 m.

would transfer to the same amount of water vapor, and vice versa.
Fig. 17 shows that the phase change process taken place in the fuel
cell depends on the temperature and pressure distributions in dif-
ferent zones. The negative value means the water transferred from
liquid phase to gas phase and positive value means the water in gas
phase is changed to liquid phase. It can be seen that phase change
from liquid phase to gas phase is dominant in the fuel cell since it
operates under the temperature of 318 K and gauge pressure vary-
ing from 0 to 700 Pa. The phase change rate is proportional to the
volume fraction of liquid water. It means that the more water in liq-
uid state, the phase change from liquid phase to gas phase is taken
Fig. 15. Current density distribution at t = 1.010 s: (a) on the cathode GDL/catalyst
place more easily. The phase change also appears in the porous
layer interface (Y = 0.0023 m) with the presence of liquid water, (b) on the cathode
media where the liquid water occupies, even with a small amount, GDL/catalyst layer interface (Y = 0.0023 m) without the presence of liquid water (c)
as shown in Fig. 17c. on middle-plane of the membrane (Y = 0.002335 m).
2152 A.D. Le, B. Zhou / Electrochimica Acta 54 (2009) 2137–2154

3.4. Liquid water amount and average current density variation

with respect to time

Fig. 18a and b shows the liquid water fraction in the channel
and average current density with respect to time. For the conve-
nience, the factor  was defined as the ratio of an amount of liquid
water in the channel at a certain time to the amount of liquid water
added at initial time. The time period was analyzed from t = 1.004 s
to t = 1.07 s. As mentioned above, the initial liquid water was added
as droplets suspending in the cathode channel. As the time pro-
gressed, liquid water would be gradually removed from the channel
to ambient air under the impact of gas flow. However, at certain
time, the amount of liquid water in the channel is smaller than the
initial amount due to phase change process occurring in the fuel
cell and some liquid water beginning to come out from the channel
exit. As a result, it could be seen in Fig. 18a that  starts to drop from
t = 1.004 s to t = 1.04 s. From t = 1.04 s to t = 1.05 s, the water amount
in the channel suddenly increased after smoothly decreasing. This
sudden increase accounts for the fact that an amount of liquid water
accumulated in the porous media comes back to the channel. How-

Fig. 17. Phase change rate of water in the cathode at t = 1.010 s: (a) on the cath-
ode channel and porous media; (b) on the middle-plane of cathode channel
(Y = 0.0015 m); and (c) on the cathode GDL/catalyst layer interface (Y = 0.0023 m).

Fig. 16. Temperature distribution at t = 1.010 s: (a) on the cathode channel and
porous media and (b) on the anode channel and porous media.
 A.D. Le, B. Zhou / Electrochimica Acta 54 (2009) 2137–2154 2153

parallel flow channels. This model includes the complete MEA and
bipolar plates. It is transient in nature and takes fluid dynamics, heat
and mass transfers, electrochemistry and species transport into
account. In addition, by implementing VOF interface tracking algo-
rithm, the flow behaviors of liquid water, their effects on physical
and transport parameters and performance were also investigated.
The main conclusions are summarized as

• Numerical analysis coupling a three-dimensional, unsteady, elec-

trochemical fuel cell model with two-phase flow and VOF
technique for interface-tracking could be a powerful optimization
tool for practical engineering designs of PEMFC.
• The motion, deformation, coalescence and detachment of water
droplets in the PEMFC were illustrated at different time steps
in the numerical simulation. The presence of liquid water in
the channel significantly influences the flow and transport
parameters such as velocity, pressure, temperature and species
concentrations. Due to the blockage of droplets, the flow would
become unevenly distributed, high pressure-drop regions would
occur around the location of liquid droplets, the liquid water
would cause a decrease of local cell temperature, and the oxygen
mass fraction would be quite low at the place where the water
droplets occupy.
• The structure of the channel would significantly influence the
distribution of liquid water inside a PEMFC. As simulated in
a serpentine-parallel channel, the liquid water could be easily
removed from the channel, compared with a serpentine chan-
nel [22]. Consequently, there is a small portion of liquid water
that tends to move through the gas diffusion layer in the porous
media. However, there were small droplets remaining in the cor-
ners of the channel due to the fact that the gas flow velocity in
such regions is insufficient to force the small droplets to move.
• The presence of liquid water significantly influences the local cur-
Fig. 18. (a) Liquid water fraction in the channel with respect to time. (b) The cell rent density in two aspects: (1) increasing the current density by
average current density with respect to time. directly increasing the ionic conductivity of the catalyst layers
and the membrane; (2) decreasing the current density by hinder-
ing the mass transport of reactant gas in the channel and porous
ever, this increase did not last long. Liquid water in the channel then
media as well. This model would be very useful for investigat-
go down very fast with nearly 30% during the period from t = 1.05 s
ing the trade-off between these two aspects by conducting some
to t = 1.059 s.
optimization studies on the performance of PEMFCs.
The change of average current density also is considered respect
to the change of liquid water. From t = 1.004 s, the average current
density reached a constant value at 5975 A/m2 . Note that the cell Acknowledgments
potential is set to be constant at 0.5 V. With the presence of liquid
water, the average current density tended to decrease as shown in The authors are grateful for the support of this work by the
Fig. 18b. It could be realized that the presence of liquid water in Auto21TM Networks of Centers of Excellence (Grant D07-DFC and
the fuel cell might be the main reason for the variation of current D303-DFC), the Natural Sciences and Engineering Research Council
density of the fuel cell. As mentioned in the previous section, water of Canada (NSERC), the Canada Foundation for Innovation (CFI), the
in the form of liquid may hinder the gas transport in the channel Ontario Innovation Trust (OIT), and the University of Windsor.
and porous media, resulting in a concentration loss and therefore
causing a reduction of the current density. However, the cell current Appendix A. Nomenclature
density may increase due to a high conductivity of the catalyst lay-
ers and the membrane with the presence of liquid water. Hence, it
is necessary to have a considerable effort to investigate the effects a water activity (dimensionless)
of liquid water on fuel cell performance. Due to the limitation of Asurf reactive surface area (m2 )
this paper, this would be investigated a future study. As heat transfer surface area (m2 )
cf concentration of sulfonic acid ions (HSO3 − ) (kmol m−3 )
4. Conclusion cp specific heat capacity (J kg−1 K−1 )
Ci species concentration i (kmol m−3 )
In this study, a three-dimensional, unsteady, multi-phase, Di diffusion coefficient of species i in gas mixture (m2 s−1 )
multi-component PEMFC model with VOF interface tracking was F Faraday constant (9.6487 × 107 C kmol−1 )
implemented using commercial CFD package—FLUENT by devel- h convective heat transfer (W m−2 K−1 )
oping own user-defined-functions to customize the governing hH2 O the enthalpy of formation of water vapor (N m kg−1 )
equations, relative equations, and boundary conditions. Numerical I current density (A m−2 )
simulations were conducted for a single PEMFC with serpentine- Iave average current density (A m−2 )
2154 A.D. Le, B. Zhou / Electrochimica Acta 54 (2009) 2137–2154

J mass flux (kg m−2 s−1 ) e electrochemical reaction

keff effective thermal conductivity (W m−1 K−1 ) eff effective
Mi molecular weight of species i in gas mixture (kg kmol−1 ) g gas phase
nd electro-osmotic drag coefficient (dimensionless) H2 hydrogen
nf charge number of the sulfonic acid ion H2 O water
P pressure (Pa) i species i
Pi partial pressure of species i (Pa) in inlet
R universal gas constant (8314 J kmol−1 K−1 ) l liquid phase
Q̇ heat rate (W) m membrane phase
Rcat , Ran volumetric current density (A m−3 ) out outlet
S source term N2 nitrogen
t time (s) O2 oxygen
T temperature (K) ref reference
u, v, w velocities in X, Y, and Z directions, respectively (m s−1 ) s solid phase
Voc open-circuit potential (V) ss solid surface
Vcell cell potential (V) sat saturated
Vref reference potential (V) surf surface
volume (m3 ) w water vapor
Xi mole fraction of species i (dimensionless)
Yi mass fraction of species i (dimensionless) References

Greek symbols [1] G. Lin, T.V. Nguyen, J. Electrochem. Soc. 153 (2006) A372.
[2] Q. Ye, T.V. Nguyen, J. Electrochem. Soc. 154 (2007) B1242.
˛ transfer coefficient [3] Y. Wang, C.Y. Wang, J. Electrochem. Soc. 153 (2006) A1193.
ˇ the factor accounts for energy release [4] H. Ju, G. Luo, C.Y. Wang, J. Electrochem. Soc. 154 (2007) B218.
concentration dependence [5] U. Pasaogullari, P. Mukherjee, C.Y. Wang, K.S. Chen, J. Electrochem. Soc. 154
(2007) B823.
p, T exponent factors [6] Y. Wang, S. Basu, C.Y. Wang, J. Power Sources 179 (2008) 603.
ε porosity [7] S. Um, C.Y. Wang, K.S. Chen, J. Electrochem. Soc. 147 (2000) 4485.
 the factor accounts for presence of liquid water [8] Z.H. Wang, C.Y. Wang, K.S. Chen, J. Power Sources 94 (2001) 40.
[9] L. You, H. Liu, Int. J. Heat Mass Transf. 44 (2002) 2277.
overpotential (V) [10] J. Ge, H. Liu, J. Power Sources 142 (2005) 56.
w contact angle (◦ ) [11] S. Dutta, S. Shimpalee, J.W. Van Zee, J. Appl. Electrochem. 30 (2000) 135.
 surface curvature (m−1 ) [12] S. Shimpalee, S. Greenway, D. Spuckler, J.W. Van Zee, J. Power Source 135 (2004)
79.
water content [13] T. Berning, N. Djilali, J. Electrochem. Soc. 150 (2003) A1589.
 dynamic viscosity (kg m−2 s−1 ) [14] D.H. Schwarz, N. Djilali, J. Electrochem. Soc. 154 (2007) B1167.
 density of gas mixture (kg m−3 ) [15] P. Quan, B. Zhou, A. Sobiesiak, Z. Liu, J. Power Sources 152 (2006) 131.
i density of species i (kg m−3 ) [16] K. Jiao, B. Zhou, P. Quan, J. Power Sources 154 (2006) 124.
[17] K. Jiao, B. Zhou, P. Quan, J. Power Sources 157 (2006) 226.
 phase conductivity (−1 m−1 ) [18] K. Jiao, B. Zhou, J. Power Sources 169 (2007) 296.
 viscous stress tensor [19] K. Jiao, B. Zhou, J. Power Sources 175 (2008) 106.
g gaseous permeability (m2 ) [20] X. Zhu, P.C. Sui, N. Djilali, J. Power Sources 172 (2007) 287.
[21] A. Bazylak, D. Sinton, N. Djilali, J. Power Sources 176 (2008) 240.
l liquid permeability (m2 ) [22] A.D. Le, B. Zhou, J. Power Sources 182 (2008) 197.
p hydraulic permeability (m2 ) [23] Fluent® 6.2 Documentation, Fluent Inc., 2005.
 electrokinetic permeability (m2 ) [24] F. Barbir, PEM Fuel Cells, Elservier Academic Press, 2005.
[25] T. Springer, T. Zawodzinski, S. Gosttesfeld, J. Electrochem. Soc. 138 (1991)
 reaction rate (kmol m2 s−1 ) 2334.
 phase potential (V) [26] R. O’Hayre, S.W. Cha, W. Colella, F.B. Prinz, Fuel Cell Fundamentals, John Wiley
ϕ relative water content & Sons, New York, 2006.
[27] C.S. Peskin, J. Comput. Phys. 25 (1997) 220.
surface tension coefficient (N/m) [28] D.B. Kothe, W.J. Rider, S.J. Mosso, J.S. Brock, Volume tracking of interfaces having
ω excess coefficient surface tension in two and three dimensions, AIAA 96-0859, 1996.
[29] J. Larminie, A. Dicks, Fuel Cell Systems Explained, 2nd ed., John Wiley & Sons,
England, 2003.
Subscripts and superscripts [30] J.H. Ferziger, M. Peric, Computational Methods for Fluid Dynamics, 3rd ed.,
an anode Springer, 2002.
cat cathode