Abstract

In this research, the impact of structural parameters of bipolar plates on the proton exchange membrane (PEM) fuel cell performance has been investigated using numerical method, and this model incorporates all the essential fundamental physical and electrochemical processes occurring in the membrane electrolyte, cathode catalyst layer, electrode backing, and flow channel, with some assumptions in each part. In formulation of this model, the cell is assumed to work under steady state conditions. Also, since the thickness of the cell is negligible compared to other dimensions, one-dimensional and isothermal approximations are used. The structural parameters considered in this paper are: the width of channels (Wc), the width of support (Ws), the number of gas channels (ng), the height of channels (hc), and the height of supports (hp). The results show that structural parameters of bipolar plates have a great impact on outlet voltage in high current densities. Also, the number of gas channels, their surface area, the contacting area of bipolar plates, and electrodes have a great effect on the rate of reaction and consequently on outlet voltage. The model predictions have been compared with the existing experimental results available in the literature, and excellent agreement has been demonstrated between the model results and the experimental data for the cell polarization curve.

Introduction

The proton exchange membrane fuel cell (PEMFC) is an electrochemical energy converter that converts chemical energy of fuel directly into DC electricity. A fuel cell using a very thin polymer membrane as an electrolyte has been considered as a promising candidate for future power sources, especially for transportation applications and residential power, due to their high efficiency, high power density, quick startup, quiet operations, and, most importantly, zero emissions, thereby reducing air pollution and greenhouse gas emissions.

The first demonstration of a fuel cell was by lawyer and scientist William Grove in 1839 [1], although it appears that a Swiss scientist Christian F. Shoenbein independently discovered the very same effect at about the same time (or even a year before) [2]. Optimizing of the geometrical parameters is one of the ways to minimize ohmic losses. Some studies have been developed to investigate the effect of using different channel configurations on cell performance. Sun et al. [3] developed a numerical study, which suggested using a trapezoidal channel cross-section rather than a rectangular or square cross-section to improve cell performance. Chiang and Chu [4] conducted numerical simulations of a 3D isothermal model of a PEMFC for various channel configurations (channel length, height, and shoulder width), with particular emphasis on the effect of channel height on cell performance. They carried out simulations with different combinations of channel width and height, while maintaining the same channel cross-sectional area and constant flow rate. Their results indicated that flat channels (i.e., smaller height) gave better performance.

Yi and Nguyen [5] developed an along-the-channel model for evaluating the effects of various designs and operating parameters on the performance of a PEM fuel cell. The results showed that the humidification of the anode gas is required to improve the conductivity of the membrane and the liquid injection and higher humidification temperature can enhance the cell performance. Ge and Yi [6] developed a two-dimensional model to investigate the effects of operation conditions and membrane thickness on the water transport. The results revealed that the cell performance can be enhanced by increasing the cell temperature. Yi and Nguyen [7] employed a two-dimensional isothermal model of a porous electrode to simulate the hydrodynamics of gas flow through the pore volume of the electrode of a PEMFC. It is concluded that, from the predictions of the PEM fuel cells with interdigitated flow field, the higher gas flow rate through the electrode improves the electrode performance. He et al. [8] and Kumar and Reddy [9,10] studied the influences of electrode and flow field design on the performance of a PEM fuel cell with a half-cell model. It was found that higher differential pressure between inlet and outlet channels would enhance the electrode performance. Kim et al. [11] developed a curve-fitting scheme based on experimental data of the first polarization curve of a PEM fuel cell. Ahmed and Sung [12] performed the simulations of PEMFCs with a new design for the channel shoulder geometry, in which the membrane electrode assembly is deflected from shoulder to shoulder. Bernardi and Verbrugge [13,14] used an analytical approach. They developed a mathematical model of a PEMFC from fundamental transport properties. Where the losses incurred by the activation over potential of the anode and cathode reaction, the ohmic losses incurred by the membrane and the ohmic losses due to the electrodes are subtracted from the reversible cell voltage. Marr and Li [15,16] developed a simplified engineering model of a PEMFC based on the catalyst layer of Weisbrod et al. [17] and the membrane model of Bernardi and Verbrugge [13,14]. This is the model that will be used to incorporate the effect of bipolar plate structural parameters on the PEMFC performance. Marr and Li [15] investigated the composition and performance optimization of cathode catalyst platinum and catalyst layer structure in a proton exchange membrane fuel cell by including both electrochemical reaction and mass transport process. They found that electrochemical reactions occur in a thin layer within a few micrometers thick, indicating ineffective catalyst utilization for the present catalyst layer design.

Also Baschuk and Li [18] investigated a polymer electrolyte membrane fuel cell with variable degrees of water flooding of the membrane.

Therefore, the issue of bipolar structural parameters in a PEMFC has not been fully examined in the previous work. In the following sections, the formulation of the model will be presented. The result is then compared to experimental data available in the literature for validation.

Model Formulation

The operation of a PEM fuel cell is based on converting the chemical energy of the fuel, such as hydrogen, and oxidant, such as oxygen, into electrical energy, as shown in Fig. 1. The following electrochemical reactions happening on the anode and the cathode are the base of fuel cell function [19].

Fig. 1
Schematic of polymer electrolyte membrane fuel cell [19]
Fig. 1
Schematic of polymer electrolyte membrane fuel cell [19]
Close modal
At the anode,
H22H++2e¯
(1)
At the cathode,
12O2+2H++2e¯H2O
(2)

More precisely, the reactions above happen on a border between the ionically conductive electrolyte and electrically conductive electrode. The electrodes must be porous, allowing the gases to arrive to, as well as product water to leave, the reaction sites, because there are gases involved in fuel cell electrochemical reactions.

In the formulation of this model, the cell is assumed to work under steady state conditions. Also, since the thickness of the cell is negligible compared to its other dimensional, one-dimensional and isothermal approximation is used.

The membrane is assumed to be fully hydrated, according to Marr and Li [15]. In addition, the anode reaction over potential is neglected in the present study, because the model of Marr and Li [15] found the over potential due to the anode reaction to be negligible [18]. Therefore, by calculating the reversible cell voltage and subtracting the losses from it, the overall cell voltage can be found.
E=Er-ηact-ηohmic,e-ηohmic,p-ηohmic,m
(3)

where Er is the reversible cell voltage and ηact is the loss due to resistance to the mass transfer limitations and electrochemical reactions in the cathode catalyst layer.

ηohmic,e, ηohmic,p, and ηohmic,m are the voltage losses due to the ohmic resistance of the electrodes, flow channel plate, and membrane layer. Er is calculated from a modified version of the Nernst equation, which is modified with an extra term to account for changes in temperature from the standard reference temperature [20]. This is given by Eq. (4).
Er=ΔG2F+ΔS2F(T-T0)+RT2F[ln(PH2)+12ln(PO2)]
(4)
ΔG is the change in Gibbs free energy, F is the Faraday constant, ΔS is the change in the entropy, and R is the universal gas constant, while PH2 and PO2 are the partial pressures of hydrogen and oxygen in the atmosphere. Taking into account the quantity of parameters, Eq. (4) is simplified to the form below. Table 1 shows the operating and structural parameters.
Er=1.229-0.85×10-3(T-298.15)+4.31×10-5T[ln(PH2)+12ln(PO2)]
(5)

Electrode.

Ohmic voltage losses are important factors in calculating overall cell voltage; ohmic losses are due to the resistance in the electrode and plate, which are modeled as shown in Fig. 2. This model assumes that current density is constant at the border electrode and the catalyst layer and that the electrons take the shortest path to and from the reaction sites. Total resistance due to the gas flow channels on the bipolar plate results from two equivalent resistances due to the solid portion of flow channels and resistance of the flow channels supports, which can be determined as follows:
Rf=ρR,phpWL
(6)
Rs=ρR,phcwsL
(7)

where hp is the thickness of the solid portion of the plate and hc is the height of the flow channels and supports. W and L are the width and length of the cell, respectively, and ws is the width of the channels supports.

Fig. 2
The flow channel plates and electrode of a polymer electrolyte membrane fuel cell modeled as an equivalent electrical resistance [18]
Fig. 2
The flow channel plates and electrode of a polymer electrolyte membrane fuel cell modeled as an equivalent electrical resistance [18]
Close modal
Also, according to Fig. 2, the relation between ws (width of support) and wc (width of channel) can be determined by the equation below,
ws=(W-ngwc)ng+1
(8)

where wc refers to the width of the flow channels and ws is the width of support, while ng is the number of the flow channels in the cell.

Table 1

Operational and structural parameters used for the base case condition

ParametersNotationValue
Temperature, °CT60
Gas mixture pressure, atmp5
Oxygen mole fractionXo20.21
CL thickness, μmLC50
GDL porosity, vol. %ɛg40
Volume fraction of membrane in the CL, vol. %Lm, c0.4
Volume fraction of GDL in the CL, vol. %Lg, c0.1
Mass loading of platinum per unit area of the cathode, K gm−2mpt0.01
Mass loading of carbon per unit area of the cathode, K gm−2mc0.003
Platinum density, kg m−3Ppt21,400
Carbon density, kg m−3Pc1800
Reference oxygen concentration, mol m−3Co21.2
Cathodic transfer coefficientdc1
Anodic transfer coefficientda0.5
Bulk protonic conductivity, (Ω m)−1k17
Bulk electronic conductivity, (Ω m)−1ks7.27 × 104
Oxygen diffusion coefficient within the liquid water, m2 s−1D02 –w9.19 × 10−9
ParametersNotationValue
Temperature, °CT60
Gas mixture pressure, atmp5
Oxygen mole fractionXo20.21
CL thickness, μmLC50
GDL porosity, vol. %ɛg40
Volume fraction of membrane in the CL, vol. %Lm, c0.4
Volume fraction of GDL in the CL, vol. %Lg, c0.1
Mass loading of platinum per unit area of the cathode, K gm−2mpt0.01
Mass loading of carbon per unit area of the cathode, K gm−2mc0.003
Platinum density, kg m−3Ppt21,400
Carbon density, kg m−3Pc1800
Reference oxygen concentration, mol m−3Co21.2
Cathodic transfer coefficientdc1
Anodic transfer coefficientda0.5
Bulk protonic conductivity, (Ω m)−1k17
Bulk electronic conductivity, (Ω m)−1ks7.27 × 104
Oxygen diffusion coefficient within the liquid water, m2 s−1D02 –w9.19 × 10−9
Thus, the total plate resistance can be determined by combining the two resistances as the following:
Rp=ρR,pL(hpW+hcW-ngwc)
(9)
In this case, the cell voltage loss due to the resistance of the plate is
ηohmic,p=2RpIδWL
(10)

where Iδ is the cell current density.

The total resistance for one electrode, from series resistance which is shown in Fig. 2 can be calculated from the following equation [18]:
Re=ρR,eeff8ngδeL(wc+ws)
(11)
where δe is the thickness of the electrode. Effective resistivity of the electrode, ρR,eeff, can be derived from the bulk resistivity of the electrode, ρ, and the void fraction of the electrode, ϕ, by the Bruggeman’s correction [20].
ρR,eeff=ρR,e(1-ϕe)32
(12)
The cathode and anode electrodes are assumed to be identical, so the voltage loss due to the resistance of the electrode is
ηohmic,e=2ReIδWL
(13)

Catalyst.

The catalyst layer is modeled with a set of coupled differential equations. In modeling the cathode catalyst layer, it is assumed to be isothermal, one-dimensional, and uniformly distributed. In this model, the other important assumption is that the negated space within the catalyst layer is large enough so that Knudsen diffusion is unimportant [18].

One of the processes which modeled in the catalyst layer is the electrochemical reaction. The rate of this process, with assuming constant proton concentration, is given by the Butler-Volmer equation (14).
dIdz=Avio,ref(CCref)γ[exp(αcFηactRT)-exp(-αaFηactRT)]
(14)

where I is the protonic current density, C is the concentration of oxygen, ηact is the over potential caused mostly by the resistance to the electrochemical reactions and due to the finite rate of mass diffusion, and z is the distance into the catalyst layer measured from the electrode/catalyst layer interface. io,ref denotes the reference current density that is a function of the cell temperature. Also, it is an experimentally derived parameter and is given as a function of temperature for Nafion on platinum in Parthasarathy et al. [21]. Also, γ, reaction order, can be found analytically from the procedure in Newman [22], with a value of 0.5 resulting.

Cref is the reference oxygen concentration, and it is related to io,ref and, in this model, has a value of 12 mol/m3. αc and αa denote the cathodic and anodic transfer coefficient, which, in this model, have the values of 1 and 0.5, respectively. Av is the specific reaction surface and can be derived from the below equation [15],
Av=mptAsδc
(15)

mpt is the catalyst mass loading per unit area of the cathode, δc is the thickness of the catalyst layer, and As is the catalyst surface area per unit mass of the catalyst.

One of the other processes modeled in the catalyst layer is an ohmic loss that was found, using Ohm’s law, as the next equation.
dηactdz=(1Kmeff+1Kseff)I-IδKseff
(16)
This equation describes the ohmic losses through the catalyst layer. The effective conductivities of the membrane and the catalyzed solids are denoted by Kmeff and Kseff, respectively. These can be related to the bulk conductivity and correction of porosity for the catalyst layer as the following:
Kmeff=(lmφc)3/2Km
(17)
Kseff=(lmφc)3/2Ks
(18)
In these equations, Km and Ks are the bulk conductivities of the membrane and solid catalyst, lm is the fraction of membrane in the catalyst layer, and ϕc is the void fraction of the catalyst layer, which can be calculated as Eq. (19) depending upon the type and thickness of the catalyst layer.
φc=1-(1ρPt+1-fPtfPtρc)mPtδc
(19)

ρpt and ρc are the density of the platinum and its carbon support, and fpt represents the amount of platinum catalyst on its carbon support.

Km in Eq. (17) is the bulk conductivity of membrane, and its value depends on the membrane type and the temperature of the cell. For a Nafion membrane, the electric conductivity is given by [14]
Km=F2CH+DH+RT
(20)
Also, the mass diffusion process was modeled using conservation of species and Fick’s law of diffusion in the catalyst layer as the following equation:
dCdz=I-Iδ4FDO2eff
(21)
This equation describes the mass transport through the cathode catalyst layer. In this equation, I is the protonic current density and DO2eff denotes the effective diffusion coefficient of oxygen in the catalyst layer, which is determined as the following equation [23]:
DO2,bulkeff=φe3/2DO2,bulk
(22)

In this model, the mass transfer of oxygen from the flow channel to the reaction site of the cathode catalyst layer will determine the shape of the polarization curve in the concentration over the potential region [18].

The boundary conditions that apply to Eqs. (23)–(25) require that the cathode electrode/catalyst layer interface and the protonic current density must be equal to zero, because the electrode is ionically insulated. Also, at the other end of the catalyst layer, the protonic current density must be equal to the cell current density. Finally, the concentration profile of the oxygen at the catalyst layer must be the same as the one calculated for the electrode at the point. These boundary conditions are summarized below [18].
I(Z=0)=0
(23)
I(Z=δc)=Iδ
(24)
C(z=0)=Ccatalyst
(25)

Membrane.

The membrane in this model is assumed to be fully hydrated and one-dimensional. As a result, the equation for the ohmic losses incurred by the membrane is [20]
ηohmic,m=δm(IδKm+FKPCH+ΔPa-cKmμH2Oδm)(1+F2KECH+2KmμH2O)
(26)

In this equation, δm is the thickness of the membrane, Km is the electrical conductivity of the membrane, which is a function of cell temperature, Kp is the hydraulic permeability, KE is the electrokinetic permeability, ΔPa-c is the pressure differential across the membrane, CH+ is the fixed charge concentration, and μH2O denotes the viscosity of liquid water. The values for the Km, Kp, CH+, and KE for Nafion can be found in Bernardi and Verbrugge [13,14].

Bipolar Plate.

In a typical frame and PEM cell, the bipolar plate is a plate of graphic with a serpentine flow channel cut into it. These channels are used to deliver reactants to the electrode and to collect current from the electrode. Two different phenomena are modeled in the plate:

  1. (1)

    Mass transport in the flow channel

  2. (2)

    Electric resistance in the plate

Channel Flow.

Many models [13,14,24] use a plug flow assumption to model mass diffusion to the electrode surface. Making this assumption effectively ignored the convective effect on mass transport in the channel. In this model, mass transport is approximated using dimensionless parameters. A Sherwood number approximation [25] is used to estimate the amount of reactant that migrates from the bulk of the stream to the surface of the electrode. This equation is valid for a binary system. Unmodified, a Sherwood number approximation would not be valid for a humidified air stream, because this stream constitutes a ternary system. To overcome this complication, the air stream is reduced to a binary system of oxygen/bulk mixture of H2O and N2 by using a bulk diffusion correction. It is assumed that there is a constant concentration of O2 at the electrode surface. Additionally, inlet and secondary flow patterns at the corners of the channel were ignored. These assumptions will give more conservative predictions, because adding in correlations for inlet and flow around the corners would increase the mass transfer rate. The mass transport has the form
N=shDibulkAcdhΔCln
(27)
where N is the molar flow rate and ΔCln is the logarithmic average of concentration change between bulk gas concentration and electrode surface. The exposed area of the electrode on the channel is symbolized with Ac, the diffusion coefficient, Dibulk, is the bulk-corrected diffusion, and dh is the hydraulic diameter of the channel. However, the Sherwood number, Sh, is a dimensionless parameter that is defined as
Sh=hmdhDij
(28)
hm is the mass transfer coefficient (mol/m2.s) and Dij is the diffusion coefficient (m2/s) of the species i in j. For the flow channel, the Sherwood number was derived by the analogy between heat and mass transfer. The Nusselt number correlation [26] for a three-sided adiabatic square duct with one constant heat flux wall was converted into a Sherwood number by using the following conversion formula [25]:
Sh=(ScPr)1/3Nu
(29)
Pr is the Prandtl number, and the Schmidt number is defined as
Sc=μρDij
(30)
The concentrations of reactants are calculated using logarithmic averages. A logarithmic average is defined as the equation below [20],
Cln=ΔC1-ΔC2ln(ΔC1ΔC2)
(31)

In this equation, ΔC1 = CsCinlet represents the inlet concentration difference and ΔC2 = CsCoutlet represents the outlet concentration difference.

Figure 3 shows how the channel is constructed in the cell. The area exposed to the electrode is defined as the length of the channel multiplied by the channel width. The length of the channel is derived by setting the length of the channel in the direction of W equal to the total width of the end supports and the width of the channel. The length is measured from the center of the channel. For the distance in the W direction, we have
lw=W-(2ws-wc)
(32)
where W is the width of the cell, lw is the length of the channel in the W direction, and ws and wc are the widths of the support and channel, respectively. The channel length in the L direction is more complicated, because the inlets and outlets are a different length than the inner channels. The length of the inlet or outlet is
louter=L-(ws-12wc)
(33)
L is the length of the cell. The length of an inner channel is
linner=L-(2ws+wc)
(34)
and the total distance in the L direction is
lL=linlet+loutlet+(ng-2)linner
(35)
lL=ngL-(ng-1)(2ws+wc)
(36)
Summing up the total for both width and length gives
lt=ng(L-(2ws+wc))+W
(37)

For the second group of terms, it is assumed that ws has the same distance with respect to both length and width. Also, ohmic resistance, which is calculated in Sec. 2.1, is assumed constant within the temperature range 300–370 K, and diffusion is assumed to spread O2 evenly throughout the electrode surface such that there is a constant O2 concentration at the electrode catalyst layer interface.

Fig. 3
Top view of channel showing how dimensions are defined for calculating channel surface area
Fig. 3
Top view of channel showing how dimensions are defined for calculating channel surface area
Close modal

Boundary Conditions.

In order to obtain a well-posed problem, the derived equations must be supplemented with a number of boundary conditions. At the gas diffusion layer (GDL)/catalyst layer (CL) interface, Z = 0, oxygen enters the CL pores from the GDL, and, since the pores are assumed to be flooded with liquid water, the oxygen passes through the GDL/CL interface at Z = 0 by first dissolving in water. The oxygen concentration at the interface is therefore determined using Henry’s law.
CO2|z=0=PO2HO2
(38)
where the oxygen partial pressure is
PO2=χO2P
(39)
χO2 is the oxygen mole fraction, and P is the gas mixture pressure at the GDL/CL interface. The Henry constant Ho2 (in atm m3 mol−1) is typically assumed to be a function of temperature [13] according to
HO2=1.33exp(-666T)
(40)
In this equation, T is measured in degrees Kelvin. Because the GDL is very thin, the pressure conditions at the GDL/CL interface are set to the cathode channel values, which, in this model, have been taken to be Xo2 = 0.21 and P = 5 atm. The boundary condition for the protonic current density comes from the assumption that all protons are consumed before they reach the GDL/CL boundary, so that
i|z=0=0
(41)
Also, at the CL/membrane interface, z = lc, the protonic current density at the membrane approaches its ultimate value (i.e., the cell current density, Iδ), so that
i|z=1c=Iδ
(42)
Also, oxygen cannot diffuse into the membrane, so that
-dCO2dz|z=lc=0
(43)

The procedure for solving the formulated model to yield a cell voltage for given structural and operational parameters is as follows: first, the reversible voltage of the cell is calculated for the reference temperature and pressure (T = 60 °C, P = 5 atm) using Eq. (5). Then, the losses are calculated in all parts of the cell for calculating overall cell voltage in Eq. (3). Also, for calculating the catalyst influence on overall cell voltage, it is needed to solve ordinary differential equations consisting of Eqs. (14), (16), and (21) along with boundary conditions, which are stated for the catalyst layer in Eqs. (38), (41), and (43) that represent a two-point boundary value problem. The built-in shooting method algorithm within the boundary value problems function of the Matlab software package has been used for the present computation [27].

Validation

The polarization curve is obtained from the present calculations in base case condition (T = 353.15 K, P = 5 atm), except with mPt = 0.35 mg/cm2 (f = 10 Pt/C Mass %) and ROhmic = 0.225 (Ω/cm2), and is compared with the experimental Ticianelli et al. [28]. As shown in Fig. 4, the results predicted by the numerical are in good agreement with the experimental data.

Fig. 4
Comparisons of numerical results with experimental data [28]
Fig. 4
Comparisons of numerical results with experimental data [28]
Close modal

Results

The study on the structural parameters of the bipolar plates, such as Wc, Ws, ϕ, ng, hp, and hc, which represent the channel width, support width, void fraction, height of support, and channel, respectively, has been done by using the Marr [15] model (Fig. 2). The amounts of other parameters are listed in Table 1.

Considering any structural changes in channels makes change in the amounts of pressure losses values, so it was assumed that its losses will be relieved by external tools and the pressure will be constant(5 bar). By comparing the results obtained from this research with other models and other sizes of the PEMs, it is revealed that these results are extendable for other fuel cells with any dimensions and structural and operational parameters.

Width of Channels (Wc).

According to obtained results, when the other parameters are constant and Wc is changed, the outlet voltage will be changed according to Fig. 5. As shown in this diagram, it is obvious that, when the number of channels or grooves in the plate is constant, the voltage increases with the decrease of Wc.

Fig. 5
Voltage versus Wc diagram in Iδ = 10 000 and ng = 12
Fig. 5
Voltage versus Wc diagram in Iδ = 10 000 and ng = 12
Close modal

Of course, this change is very insignificant and, according to Fig. 6, the changes in Wc are more sensible in high current densities. It can be a good manner to increase total voltage in high current densities.

Fig. 6
Voltage and power difference versus current density when Wc changes from 0.003 (m) to 0.00001 (m)
Fig. 6
Voltage and power difference versus current density when Wc changes from 0.003 (m) to 0.00001 (m)
Close modal

Figure 5 shows the change of voltage in variable sizes of channel width in ng = 12 and Iδ = 10 000 (A/m2). Figure 6 shows the voltage and power difference when Wc changes from 0.003 (m) to 0.00001 (m) and other parameters are constant.

More surface area is necessary in high current densities; however, when the width of channels decreases, the support area will increase.

The increment of the support area will change the void fraction of the electrode as compression decreases and consequently change the ability of reactant to diffusion to the catalyst surface directly underneath the support changes. So, with decreasing Wc, the ohmic resistance will decrease and also changes of void fraction of the electrode will have influence on cell voltage. The electrode void fraction influence on cell voltage will be discussed in Sec. 4.2.

Void Fraction.

Electrode void fraction is one of the important structural parameters that has influence on cell voltage. According to Marr [20], when void fraction in the electrode increases, mass transfer increases as well and, similarly, electrode resistance increases. The corresponding increase in resistivity from the increase of the electrode void fraction is swamped by increasing performance from better gas diffusion. As shown in Fig. 7, the increasing of the void fraction will lead to enhancement of cell voltage.

Fig. 7
Polarization curve with different amount of void fraction [27]
Fig. 7
Polarization curve with different amount of void fraction [27]
Close modal

Width of Support (Ws).

According to Eq. (8), it can be seen that, for a constant number of channels in the bipolar plate with defined dimensions, Ws is increased as Wc decreases. So the outlet voltage is increased by increase of Ws, and this increase in voltage is due to the decrease of ohmic resistance in addition to changes in void fraction, like in Sec. 4.2. Also, this increase of outlet voltage, due to increase of Ws, is more obvious in high current densities.

Figure 8 shows the change of voltage in variable sizes of Ws when ng = 12 and Iδ = 10 000 (A/m2). Figure 9 shows the voltage and power difference when Ws changes from 0.003 (m) to 0.00001(m) and other parameters are constant.

Fig. 8
Voltage versus Ws diagram for reference current density = 10,000 (A/m2) and ng = 12
Fig. 8
Voltage versus Ws diagram for reference current density = 10,000 (A/m2) and ng = 12
Close modal
Fig. 9
Voltage and power difference versus current density when Ws changes from 0.0038 (m) to 0.0011 (m)
Fig. 9
Voltage and power difference versus current density when Ws changes from 0.0038 (m) to 0.0011 (m)
Close modal

Number of Channels (ng).

It was seen that the outlet voltage is increased because of decrease in Wc. Also, it is clear that, in high current densities, the effects of structural parameters are more sensible and more area of reaction is needed in these high current densities. Therefore, by increase of ng, the surface of contact can be increased, but on the other side, the width of supports will decrease by the increase of the number of channels. This causes changes in the void fraction and, accordingly, the ohmic resistance of the plate. Therefore, the number of channels must have an optimized amount. It has been clearly shown in Fig. 10. According to the voltage-ng diagram, it is obvious that, when ng increases in constant amounts of Wc and Iδ, the voltage increases and has a maximum amount in a definite amount for each size of Wc.

Fig. 10
Voltage versus ng, current density = 10,000 (A/m2)
Fig. 10
Voltage versus ng, current density = 10,000 (A/m2)
Close modal

Height of Channels (hc).

Through obtained data by numerical solution method and according to Fig. 11, it is clear that, as hc (the height of channels shown in Fig. 2) is decreasing, the voltage is increased. This is due to the decrease of ohmic resistance. In this diagram, the number of channels used for the transfer of gas, the width of channel, and mass flow rate of channel are constant and hc is variable.

Fig. 11
Voltage versus hc diagram, reference current density = 10,000(A/m2), wc = 0.002 (m), ng = 12
Fig. 11
Voltage versus hc diagram, reference current density = 10,000(A/m2), wc = 0.002 (m), ng = 12
Close modal

Also, in Fig. 12, the polarization curve for two different hc demonstrates the increase of voltage in higher current densities because of decreasing height of channels. That change in hc has more impact on outlet voltage in higher current densities.

Fig. 12
Polarization curve for two different sizes of channel height, wc = 0.002 and ng = 12
Fig. 12
Polarization curve for two different sizes of channel height, wc = 0.002 and ng = 12
Close modal

Height of Support (hp).

By study effect of the height of supports, which is shown in Fig. 2, it is evident that, by reducing hp, the outlet voltage increases due to reduction of ohmic resistance. Figure 13 shows the influence of hp on the outlet voltage.

Fig. 13
Voltage versus hp diagram. Current density = 10,000 (A/m2), Wc = 0.002 (m), ng = 12.
Fig. 13
Voltage versus hp diagram. Current density = 10,000 (A/m2), Wc = 0.002 (m), ng = 12.
Close modal

Other structural and operational parameters remain constant in the investigation of hp parameter.

Figure 14 shows the voltage and power difference when hp changes from 0.009 (m) to 0.00001 (m) and other parameters are constant.

Fig. 14
Voltage and power difference versus current density when hp changes from 0.009 (m) to 0.00001 (m)
Fig. 14
Voltage and power difference versus current density when hp changes from 0.009 (m) to 0.00001 (m)
Close modal

Furthermore, it can be seen that the effect of the channel height is more sensible in high current densities, and according to Fig. 15, the effect of channel height is more sensible than the height of support on cell voltage.

Fig. 15
Voltage versus hp and hc, current density = 10,000 (A/m2) and ng = 12
Fig. 15
Voltage versus hp and hc, current density = 10,000 (A/m2) and ng = 12
Close modal

Conclusion

By investigation of the isothermal, steady, and one-dimensional PEM fuel cell, it is shown that the present model is in excellent agreement with experimental data. Also, it was obtained that structural parameters of bipolar plates and the method of their designing and construction has impact on outlet voltage in high current densities. Also, the result indicates decreasing width of the gas channels, increasing outlet voltage. Similarly, it is evident that the height of channel, the height of support, and the number of gas channels has influence on outlet voltage. Also, the result shows that the contacting area of bipolar plates and the electrode and surface area of gas channels have great effect on the rate of reaction and, consequently, on cell voltage. Changes in the number of gas channels show that there are an optimum number of them for each size of channel width. Also, the increase in number of bipolar plate channels is not a benefit to reach maximum power density and depends on other parameters like contact area or width of channels and width of supports.

References

1.
Grove
,
W. R.
,
1839
, “
On Voltaic Series and the Combination of Gases by Platinum
,”
London Edinburgh Philos. Mag. J. Sci., Series 3
,
14
, pp.
127
130
.
2.
Bossel
,
U.
,
2000
, “
The Birth of the Fuel Cell 1835–1845
,”
European Fuel Cell Forum
,
Oberrohrdorf, Switzerland
.
3.
Sun
,
L.
,
Oosthuizen
,
P. H.
, and
McAuley
,
K. B.
, “
A Numerical Study of Channel-to-Channel Flow Cross-Over Through the Gas Diffusion Layer in a PEM-Fuel-Cell Flow System Using a Serpentine Channel With a Trapezoidal Cross-Sectional Shape
,”
Int. J. Therm. Sci.
(in press)
.
4.
Chiang
,
M. S.
, and
Chu
,
H. S.
, “
Numerical Investigation of Transport Component Design Effect on a Proton Exchange Membrane Fuel Cell
,”
J. Power Sources
(in press)
.
5.
Yi
,
J. S.
, and
Nguyen
,
T. V.
,
1998
, “
An Along-the-Channel Model for Proton Exchange Membrane Fuel Cells
,”
J. Electrochem.
,
145
(
4
), pp.
1149
1159
.10.1149/1.1838431
6.
Ge
,
S. H.
, and
Yi
,
B. L.
,
2003
, “
A Mathematical Model for PEMFC in Different Flow Modes
,”
J. Power Sources
,
124
, pp.
1
11
.10.1016/S0378-7753(03)00584-6
7.
Yi
,
J. S.
, and
Nguyen
,
T. V.
, “
Multicomponent Transport in Porous Electrodes of Proton Exchange Membrane Fuel Cells Using the Interdigitated Gas Distributors
,”
J. Electrochem. Sci.
,
146
, pp.
38
45
.10.1149/1.1391561
8.
He
,
W.
,
Yi
,
J. S.
, and
Nguyen
,
T. V.
,
2000
, “
Two-Phase Flow Model of the Cathode of PEM Fuel Cells Using Interdigitated Flow Fields
,”
AIChE J.
,
46
(
10
), pp.
2053
2064
.10.1002/aic.690461016
9.
Kumar
,
A.
, and
Reddy
,
R. G.
,
2006
, “
Effect of Gas Flow-Field Design in the Bipolar/End Plates on the Steady and Transient State Performance of Polymer Electrolyte Membrane Fuel Cells
,”
J. Power Sources
,
155
, pp.
264
271
.10.1016/j.jpowsour.2005.05.006
10.
Kumar
,
A.
, and
Reddy
,
R. G.
,
2003
, “
Effect of Channel Dimensions and Shape in the Flowfield Distributor on the Performance of Polymer Electrolyte Membrane Fuel Cells
,”
J. Power Sources
,
113
, pp.
11
18
.10.1016/S0378-7753(02)00475-5
11.
Kim
,
J.
,
Lee
,
S.
, and
Srinivasan
,
S.
,
1995
, “
Modeling of Proton Exchange Membrane Fuel Cell Performance With an Empirical Equation
,”
J. Electrochem. Soc.
,
142
(
8
), pp.
2670
2674
.10.1149/1.2050072
12.
Ahmed
,
D. H.
, and
Sung
,
H. J.
,
2008
, “
Design of a Deflected Membrane Electrode Assembly for PEMFCs
,”
Int. J. Heat Mass Transfer
,
51
, pp.
5443
5453
.10.1016/j.ijheatmasstransfer.2007.08.037
13.
Bernardi
,
D. M.
, and
Verbrugge
,
M. W.
,
1991
, “
A Mathematical Model of a Gas Diffusion Electrode Bonded to a Polymer Electrolyte
,”
AIChE J.
,
37
(
9
), pp.
1151
1163
.10.1002/aic.690370805
14.
Bernardi
,
D. M.
, and
Verbrugge
,
M. W.
,
1992
, “
A Mathematical Model of a Solid Polymer Electrolyte Fuel Cell
,”
J. Electrochem. Soc.
,
139
(
9
), pp.
2477
2491
.10.1149/1.2221251
15.
Marr
,
C.
, and
Li
,
X.
,
1999
, “
Composition and Performance Modeling of Catalyst Layer in a Proton Exchange Membrane Fuel Cell
,”
J. Power Sources
,
77
, pp.
17
27
.10.1016/S0378-7753(98)00161-X
16.
Marr
,
C. L.
, and
Li
,
X.
,
1998
, “
An Engineering Model of Proton Exchange Membrane Fuel Cell Performance
,”
ARI
50
, pp.
190
200
.10.1007/s007770050014
17.
Weisbrod
,
K. R.
,
Grot
,
S. A.
, and
Vanderborgh
,
N. E.
,
1995
, “
Through-the-Electrode Model of a Proton Exchange Membrane Fuel Cell
,”
Electrochem. Soc. Proc.
23
, pp.
153
167
.
18.
Baschuk
,
J. J.
, and
Li
,
X.
,
2000
, “
Modeling of Polymer Electrolytemembrane Fuel Cells With Variable Degrees of Water Flooding
,”
J. Power Sources
,
86
, pp.
181
195
.10.1016/S0378-7753(99)00426-7
19.
Larminie
,
J.
, and
Dicks
,
A.
,
2003
,
Fuel Cell Systems Explained
, 2nd ed.,
Wiley
,
New York
.
20.
Marr
,
C. R.
,
1996
, “
Performance Modelling of a Proton Exchange Membrane Fuel Cell
,”
MSc thesis
,
University of Victoria
,
Victoria, Canada
.
21.
Parthasarathy
,
A.
,
Srinivasan
,
S.
, and
Appleby
,
A. J.
,
1992
, “
Temperature Dependence of the Electrode Kinetics of Oxygen Reduction at the Platinum Nafion Interface—A Microelectrode Investigation
,”
J. Electrochem. Soc.
,
139
(
9
), pp.
2530
2537
.10.1149/1.2221258
22.
Newman
,
J. S.
,
1991
,
Electrochemical Systems
, 2nd ed.,
Prentice-Hall
,
Englewood Cliffs, NJ
.
23.
Amphlett
,
J. C.
,
Baumert
,
R. M.
,
Mann
,
R. E.
,
Peppley
,
B. A.
, and
Roberge
,
P. R.
,
1995
, “
Performance Modeling of the Ballard Mark IV Solid Polymer Electrolyte Fuel Cell II: Empirical Model Development
,”
J. Electrochem. Soc.
142
(
1
), pp.
9
15
.10.1149/1.2043959
24.
Springer
,
T. E.
,
Zawodzinski
,
T. A.
, and
Gottesfeld
,
S.
,
1991
, “
Polymer Electrolyte Fuel Cell Model
,”
J. Electrochem. Soc.
138
(
8
), pp.
2334
2342
.10.1149/1.2085971
25.
Incropera
,
F. P.
, and
De Witt
,
D. P.
,
1990
,
Fundamentals of Heat and Mass Transfer
,
John Wiley
,
New York
.
26.
Kakac
,
S.
, and
Shah
,
R. K.
,
1987
,
Handbook of Single Phase Heat Transfer
,
John Wiley
,
New York
.
27.
Gerald
,
C. F.
, and
Wheatley
,
P. O.
,
1999
,
Applied Numerical Analysis
, 6th ed.,
Addison-Wesley
,
Reading, MA
.
28.
Ticianelli
,
E. A.
,
Derouin
,
C. R.
,
Redondo
,
A.
,
Sirinivasan
,
S.
,
1988
. “
Methods of Advance Technology of Proton Exchange Membrane Fuel Cells
,”
J. Electrochem. Soc.
,
135
(
9
), pp.
2209
2214
.10.1149/1.2096240