Applied Thermal Engineering 48 (2012) 465e475

Contents lists available at SciVerse ScienceDirect

Applied Thermal Engineering

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

Thermal management strategies for a 1 kWe stack of a high temperature proton exchange membrane fuel cell
E. Harikishan Reddy a, b, S. Jayanti a, *
a b

Department of Chemical Engineering, IIT Madras, Chennai 600036, India Department of Chemical Engineering, IIT Hyderabad, Yeddumailaram 502205, India1

a r t i c l e i n f o
Article history: Received 1 February 2012 Accepted 18 April 2012 Available online 25 April 2012 Keywords: Thermal management Polymer electrolyte membrane fuel cells High temperature operation Heat removal Temperature mapping Computational uid dynamics

a b s t r a c t
A proper thermal management strategy is needed to maintain uniform temperature distribution and derive optimal performance in high temperature proton exchange membrane fuel cells (HT-PEMFC). In HT-PEMFCs, more than half of the chemical energy is converted into thermal energy during the electrochemical generation of electrical power. We investigate the viability of three heat removal strategies: (a) using cooling plates through which cathode air is passed in excess of stoichiometric requirement for the purpose of heat removal, (b) using forced convection partly in conjunction with cooling plates, and (c) using forced convection alone for heat removal. Calculations, partly done using computational uid dynamics simulations, for a 1 kWe HT-PEMFC stack, which is suitable for scooter type of transport applications, show that a combination of excess stoichiometric factor and forced draft appears to provide the optimal strategy for thermal management of high temperature PEM fuel cells. With proper cooling strategy, the temperature variations within the cell may be reduced to about 20 K over most of the cell and to about 50 K in isolated spots. 2012 Elsevier Ltd. All rights reserved.

1. Introduction Proton exchange membrane fuel cells (PEMFC) are one of the most promising energy technologies for portable applications such as for transportation and for stand-alone or for distributed small power generation at high energy efciencies. Conventional PEMFCs use Naon and other polymer membranes which are restricted in their operation to the rather low temperature range of between 60 and 80  C. Their performance is sensitive to the hydration state of the polymer membrane: too little humidication will lead to reduced proton conductivity and too much humidication will lead to ooding, resulting again in severe loss of performance [1]. In recent years, a number of alternative polymer membranes have been developed which enable operation in the higher temperature range of 120e200  C [2e7]. Higher temperature operation brings in certain advantages: higher tolerance to carbon monoxide (CO) of up to 3e5% by volume enabling a wider choice of fuel; less sensitivity of protonic conductivity to humidication leading to simplication of the water management system; enhanced kinetics; and the possibility of using the exhaust gases for on-board fuel reforming or for other thermal systems [8].
* Corresponding author. Tel.: 91 44 2257 4168; fax: 91 44 2257 4152. E-mail address: sjayanti@iitm.ac.in (S. Jayanti). 1 Currently at IIT Madras. 1359-4311/$ e see front matter 2012 Elsevier Ltd. All rights reserved. doi:10.1016/j.applthermaleng.2012.04.041

An important consideration in the operation of HT-PEMFCs is their thermal management which is needed to prevent the formation of hotspots and to maintain nearly uniform temperature throughout. Despite the advantages of working at high temperatures, the performance of HT-PEMFCs is currently well below that of normal PEMFCs operating at about 80  C as shown in Fig. 1. Thus, during the operation of HT-PEMFCs, say, at a cell voltage of 0.6 V, more than half of the chemical energy of the reactants is converted to thermal energy. The heat energy from the fuel cell is the sum of the irreversible heat, entropic heat (reversible heat) and ohmic heat which account for 55%, 35%, and 10% of the total heat, respectively [10]. This heat is continuously generated as long as the cell is in operation and needs to be removed for the fuel cell to operate at a steady temperature. It is also necessary that, in the process of heat removal (which is primarily by conduction through the media involved), high temperature zones are not created within the cell. Non-uniform temperature distribution leads to variations in the rates of electrochemical reaction. In addition, it may lead to the creation of local hotspots which may lead to damage of the structural components. Although the thermal resistance of PBI membrane, which is studied extensively as a candidate for an HTPEMFC, is high, temperatures higher than 473 K are not advisable because the proton conductivity of PBI depends on the doping level of the phosphoric acid. At high temperature, the membrane may get dried of because of phosphoric acid evaporation. High temperatures

466

E. Harikishan Reddy, S. Jayanti / Applied Thermal Engineering 48 (2012) 465e475

Fig. 1. Typical polarization curves for a low temperature PEM fuel cell using the Naon membrane and operating at a cell temperature of 353 K [9], and a high temperature PEM fuel cell with a PBI membrane operating at 473 K [7].

have a great inuence on the carbon support of the catalysts and degradation of other components [5e7]. Therefore, proper thermal management is necessary in order to maintain good performance of HT-PEMFCs. Generally PEMFCs are cooled by air or water or heat spreaders or antifreeze or other types of coolant that circulates through dedicated cooling plates or bipolar plates [11e13]. A nearly uniform temperature distribution can be obtained by using a large ow rate of coolant which circulates through cooling plate. However, having a separate coolant adds to the complexity of the system, increases the operational cost and the high parasitic power consumption decreases the overall efciency of the fuel cell [11]. Several thermal management models have been proposed for low temperature PEMFCs (LT-PEMFC). Ju et al. [14] developed a numerical model for the evaporation and condensation of water inside LT-PEMFCs and studied the inuence of the cell temperature gradient, liquid saturation along the length of the coolant path. Choi et al. [15] developed a numerical model in order to investigate the effect of parallel and serpentine ow led geometry on the maximum surface temperature and uniformity of temperature distribution in the LT-PEMFC stack. Yu et al. [16] and Baek et al. [17] investigated different multi-pass serpentine ow eld designs in order to achieve uniformity of temperature distribution in the LT-PEMFC stack. Matian et al. [18] developed a simple thermal model to investigate effect of bipolar plate material of construction and external forced convection on LT-PEMFC stack temperature and validated their results with a fuel cell analog composed of an electrically heated plate. Asghari et al. [19] developed a numerical model for a parallel serpentine ow eld design in order to minimum pressure drop and maximum temperature uniformity within the stack and validated with LT-PEMFC stack. Koh et al. [20] investigated the inuence of the current density on the average cell temperature for an air-cooled stack of a Naon membrane-based fuel cell and demonstrated that air cooling by natural convection was not sufciently effective in maintaining a uniform cell temperature. Shon et al. [21] studied the behavior of an air-cooled LT-PEMFC for portable applications. Adzakpa et al. [22] developed a transient air cooling thermal modeling of a LT-PEMFC and investigated the effect of the non-uniformity cell temperature on the efciency of the cell. Cozzolino et al. [23] studied the behavior of a water-cooled LTPEMFC stack suitable for micro-cogeneration systems application. Zhang et al. [24] developed a lumped thermal model and studied

the effect of operating parameters on the system thermal performance of an LT-PEMFC stack for transportation applications. Yu et al. [25] developed a thermal model in order to estimate parasitic losses with respect to operating temperature of a LT-PEMFC stack for transportation applications. Studies have been reported for HT-PEMFCs too. Sousa et al. [26], Peng and Lee [27] developed a single-phase, non-isothermal numerical model for a single cell of HT-PEMFCs. Andreasen et al. [28] studied heating strategies for an HT-PEMFC stack with air and direct electrical heating methods. Andreasen et al. [29] modeled a 1 kW, air-cooled HTPEM fuel cell in a hybrid electrical vehicle by using external air with a combination storage battery. Scholta et al. [30] developed designs for external liquid cooling for an HT-PEMFC stack in order to avoid liquid coolant within cell active area. Song et al. [31] developed a prototype of a natural circulation-driven water cooling system for an HT-PEMFC stack operating at an exit temperature of 150  C in order to reduce the pumping cost of coolant. There have thus been a number of thermal management studies for LT- and HT-PEMFCs using numerical models based on CFD simulations. While air cooling has been studied, this has been limited to external (forced convective) cooling only. Plate cooling has been studied but with an external coolant loop using a liquid. The objective of the present study is to investigate the possibility of using cathode air, which is required for the electrochemical reaction leading to power generation, to also serve as a coolant of the stack. This serves one other purpose: one of the requirements of the thermal management system for an HT-PEMFC is the need to preheat the air that is entering the cathode side of the cell. While there is a lot of literature on HT-PEMFCs, this need has not been addressed sufciently. Using the cathode air as a stack coolant could simultaneously serve the purpose of having to preheat the air to nearly the cell operating temperature thereby help in minimizing the temperature gradients within the cell. Of interest here are two issues: rstly, how much cathode air is required to maintain cell temperature variations within a reasonable range, and secondly, whether enough heat transfer capability (in terms of surface areas, heat transfer coefcients and temperature gradients) exists within the cell to transfer the heat to the cathode air. In the present study, we address these questions through computational uid dynamics (CFD) simulations of a section of a nominally 1 kWe HT-PEMFC stack focusing on the heat transfer from the cathode catalyst layer (at which most of the heat is generated) through the gas diffusion layer (GDL), the bipolar plates and the cooling plates to the air owing through these ow elds. A CFD simulation will be able to account simultaneously for the uid ow through the channels and the conductive and convective heat transfer occurring through various structural elements of a fuel cell. We also study the possibility of natural and forced convective heat removal from the stack so as to come up with a comprehensive strategy for thermal management of an HT-PEMFC stack. Details of the calculations and the results obtained are discussed below. 2. Model description and problem formulation 2.1. Description of the HT-PEMFC stack The HT-PEMFC under study is based on the Polybenzimidazole (PBI) membrane doped with phosphoric acid which has been studied by a number of researchers [2]. The membrane is 40 mm thick and has fairly high protonic conductivity even without humidication. As in the case of a low temperature PEMFC, platinum catalyst is used on both the anode and the cathode sides, although at a higher loading of 0.2e0.4 mg/cm2. The gas diffusion layers and the bipolar plates are also the same as those for an LTPEMFC. Its typical operating characteristic is given in Fig. 1; the

E. Harikishan Reddy, S. Jayanti / Applied Thermal Engineering 48 (2012) 465e475

467

stack is taken to be made up of cells operating at a constant voltage of 0.6 V and at a current density of 0.42 A/cm2. An HT-PEMFC stack includes serially connected single unit cells to produce the designed output power as shown schematically in Fig. 2. In the present study, we consider possible application in the two-wheeler transport sector. Some commercial bikes with a battery-powered engine with a maximum power of 400 W are already in the market. Given that their maximum speed is limited to about 40 km per hour, we aim at an enhanced maximum engine power of 1 kWe. We therefore design an HT-PEMFC stack capable of providing a power output of up to 1 kWe. Taking active area of each cell to be 100 cm2, 40 such cells are connected in series to make up the stack which generates 1 kWe. For the purpose of the present study, the stack is assumed to be cooled by a graphite plate, one for every four cells, with an embedded ow eld through which air is circulated. The overall dimensions of the stack and the dimensions of the cell and the cooling plates assumed in the present study are given in Table 1. The thermal conductivities [32] and air uid properties are summarized in Table 2 and Table 3. 2.2. Estimation of the thermal load The hydrogen gas ionizes at the anode catalyst layer and releases hydrogen ions and electrons. The electrons ow from the anode to the cathode by an external circuit while the hydrogen ions ow through the membrane. At the cathode catalyst layer, oxygen reacts with the electrons and hydrogen ions and produces water, electrical energy and thermal energy. Most of the heat produced in the cell is generated in the cathode catalyst layer due to asymmetry of the entropy of change and the overpotential on the cathode electrode [33,34]. Ohmic heating attributable to proton and electron conduction through the membrane and the GDL etc, respectively, also contributes to the heat generation, although only to a small extent. Since the anode overpotential is small, its contribution can be neglected [35]. There are sources of heat removal from the cell. Firstly, due to the high operating temperature, all

water produced at the cathode will be vapourized; the contribution of the latent heat of vapourization must be accounted for in estimating the heat load. Similarly, the heating up of hydrogen in the anode ow eld from its feed temperature to the cell operating temperature also requires heat. The same is true also for the heating up of the cathode air. Finally, heat may be lost by natural (or forced) convection and radiation to the surroundings. In the present study, we neglect the contribution of the minor heat sources or sinks such as those on the anode side and those associated with ohmic heating and take specic account of the other factors in estimating the thermal load on the heat removal system. This is done as follows. The 1 kWe HT-PEMFC stack is designed based on a cell voltage of 0.6 V and a uniform current density 0.42 A/cm2 [7]. The stack operational temperature is taken to be 200  C. The current produced by the stack, Ist, is given by

Ist i Acell

(1)

where i is the current density and Acell is the total active area of a single cell (all cells are assumed to be operating identically), which is taken in the present case to be 100 mm 100 mm or 0.01 m2. The stack voltage, Vst, is given by

Vst Vcell Ncell

(2)

where Vcell is the cell operating voltage (0.6 V) and Ncell is the number of cells connected in series, which is determined from the requirement that the power of the stack, which is given by,

Pst Ist Vst

(3)

should be 1 kWe. This gives the number of cells to be 40. The heat generation rate for each cell in the stack at any working temperature can be determined using thermodynamic relations. The total amount of heat released, Qcell, from the each cell is the sum of the reversible heat generation, Qrev, and the irreversible heat generation, Qirrev [34]. Thus,

Fig. 2. Schematic diagram of a portion of the fuel cell stack showing the arrangement of one cooling plate for every four cells. The gure on the right shows the portion modeled in the CFD simulations as Case A.

468

E. Harikishan Reddy, S. Jayanti / Applied Thermal Engineering 48 (2012) 465e475 Table 3 Air properties. At 303 K Density, r Specic Heat, Cp Thermal conductivity, K Thermal Expansion coefcient, b Kinematic viscosity, y Prandtl number, Pr 1.255 1006 0.0242 3.2 103 1.46045 0.712 388 K 0.922 1011 0.032 2.615 103 2.413105 0.702 Kg/m3 J/kg K W/m K 1/K m2/s

Table 1 The cell and the stack dimensions and operating conditions of an HT-PEMFC operating at a cell voltage of 0.6 V and a current density of 0.42 Acm2. Bipolar plate channels Channel depth Channel width Channel and rib width Channel Length Cooling plate channels Channel depth Channel width Channel Length Single Cell layers Anode /cathode GDL thickness Anode catalyst layer thickness Membrane layer thickness Cathode catalyst layer thickness Monopolar plate thickness Bipolar plate thickness Cooling plate thickness Stack dimensions Number of cells Active area of the cell Height of the stack Breadth of the stack Width of the stack Stack operating conditions Inlet air temperature of the coolant air Atmospheric air temperature Stack operating temperature 1 103 1 103 2 103 0.1 1 103 1 103 0.1 2 104 1.3 105 4 105 1.3 105 1.5 103 3 103 2 103 40 102 0.1 0.1 0.152 303 303 473

m m m m m m m m m m m m m m e m2 m m m K K K

vap 3 2 DHT 3:6985x104 Tcell 0:4834Tcell 152:4258Tcell

68260:5789

(9)

vap where DHT is in Joule and Tcell in K. From Eqs. (1)e(8), the total heat released from each cell in the stack can be written as

 Qcell

DHg;T
nF

 Vcell Icell

(10)

and the total amount of heat released from the stack can be written as

Qst Qcell Ncell

(11)

The above calculation shows that for a stack of 40 cells operating at temperature of 200  C, a voltage of 0.6 V and a current density of 0.42 A/cm2, the total amount of heat generated is 913 W. 2.3. Estimation of heat removal by natural convection and radiation It is expected that some amount of the heat generated in the stack will be removed by natural convection and radiation. We estimate these by assuming that the four side edges, two of which are horizontal (the top and the bottom) and the other two (side edges) vertical, are free of encumbrances and are therefore free to remove heat by natural convection and radiation. Since the cell cross-section is rectangular, all the four faces have an area of 0.15 m 0.1 m or 0.015 m2. The amount of heat removed by natural convection can be calculated using appropriate correlations expressed in terms of Grashof and Rayleigh numbers [37] for natural convective heat transfer from vertical and horizontal plates. Assuming that all the surfaces are at a temperature of 200  C and that ambient air temperature is 30  C, the heat removal rate by natural circulation comes to about 45 W, which amounts to only w5% of the total thermal load. The radiative heat transfer rate from the stack can also be similarly estimated. The emissivity of the graphite is reported to be between 0.3 and 0.7 [38]. Taking a mean value of 0.5 and taking the surface temperature of the plates to be 200  C, the total radiative heat transfer rate amounts to about 60 W. The actual value may be less than this because not all the exposed surface of the plate is made of graphite. 2.4. Problem formulation It can be seen from the above calculations that natural convection and radiative heat transfer together are expected to contribute to the removal of only w12% of the total heat generated from a 1 kWe HT-PEMFC stack at the nominal operating temperature of 200  C. For steady operation, the rest of the heat, amounting to nearly 800 W, needs to be removed by other means. In the present study, we consider two possibilities to remove the excess heat: (A) using air which needs to be fed to the cathode to cool the stack by passing it through cooling plates prior to it being fed to the cathode, and

Qcell Qrev Qirrev

(4)

The reversible heat release can be written as a function of the cell operating temperature (Tcell), the cell current (Icell) and the entropy change of the overall reaction (DST):

Qrev Tcell DST

Icell nF

(5)

where F is Faradays constant and n is the number of electrons transferred. The entropy change for the reaction can be written in terms of the cell operating temperature as [36]

DST 9967:35lnTcell 12414:83

where DST in J K evaluated as
1

(6)

and Tcell in K. The irreversible heat release can be

 Qirrev

DGT
nF

 Vcell Icell

(7)

where DGT is the Gibbs free energy change for the reaction. In the evaluation of DGT, one can account for the fact that it is water vapor that is produced in HT-PEMFCs rather than liquid water. The enthalpy of water formation in gaseous phase, DHg,T, can be evaluated as
vap DHg;T DHlf DHT

(8)

where the enthalpy of water formation in liquid phase, DHlf , is vap , is evaluated 285830 J and the heat of vaporization of water, DHT from [36] the following equation
Table 2 Thermal properties of fuel cell layers [18]. Anode/cathode GDL Membrane Anode/cathode catalyst Bipolar/monopolar plate Cooling plate 1.70 0.95 0.30 20 20 W/m W/m W/m W/m W/m K K K K K

E. Harikishan Reddy, S. Jayanti / Applied Thermal Engineering 48 (2012) 465e475

469

(B) forced circulation of external air over the stack. It may be noted that, in a moving vehicle, this may be achieved by directing the air to over the stack; thus a blower may not be required for this purpose. Case A has two additional advantages; rstly, in the process of cooling the stack, the cathode air gets pre-heated to the cell temperature and therefore avoids cold spots in the entrance region. Secondly, by feeding the air through the cooling plates, the interior region of the cell gets cooled directly which gives rise to a more uniform temperature distribution over the cell. Case B has the advantage that the additional complexity of cooling plates and ow passage through its ow elds can be avoided. However, external cooling alone may create large temperature differences between the exterior and the core region of the cell. The variation of the temperature due to internal conduction through the composite plates is to be determined. Similarly, in Case A, it is not known if sufcient heat transfer area exists within the cooling plates to effectively transfer heat to the air. In order to determine these, two computational models have been set up to calculate the temperature variation within the cells. For the sake of simplicity, the fuel cell stack is assumed to consist of 40 cells connected in series with each cell having bipolar plates with 50 parallel channels of a length of 100 mm and a cross-section of 1 mm 1 mm. In such a stack, the possibility of having cooling plates is investigated. Given the good thermal conductivity of graphite, we assume that the cooling plates are also made of graphite and are thick enough to have cooling channels of 1 mm 1 mm crosssection and a length of 100 mm running parallel to the ow elds (see Figs. 2 and 3). One cooling plate is provided for every four cells; thus, the cathode air required for the four cells rst passes through the channels in the cooling plate and then gets redistributed among the four adjacent channels. The ne details of the ow distribution through the parallel channels in the ow eld are not considered and the ow rate is assumed to be uniformly distributed through the 50 channels on each plate. With this assumption, CFD simulations of ow and heat transfer have been carried out for two simplied geometries representing the two cases:

 Case A is meant to simulate the heat generation in the cathode catalyst layer; its removal by conduction through the membrane on one side and the cathode side GDL on the other side; its further transfer by conduction to the other layers and removal by convection by the air owing through the cathode ow elds and the cooling plate. To this end, a ow domain (Fig. 3) of a length of 100 mm, width of 9.558 mm and thickness of 2 mm is considered. The domain represents one-half of a four-cell unit with one cooling plate consisting of one-half of the cooling channel and the two cathode air channels on one side along with the composite solid media in-between. The length of the domain here corresponds to the length of a parallel ow eld on the cell or the cooling plate. The width corresponds to half the width of four cells at the middle of which a graphite cooling plate is inserted. The thickness of 2 mm covers the channel width and the rib width in a typical PEM fuel cell ow eld. The segment considered in Fig. 3 for the ow simulation is such that (i) 50 such segments in the thickness direction will equal the cell breadth of 100 mm and (ii) 20 segments in the width direction will equal the 40 cells (with 10 cooling plates) of the stack. In the 100 9.558 2 mm3 segment, each cathode catalyst layer is modeled separately (because a volumetric heat source term has to be specied within it) while the membrane, the anode side GDL and the anode catalyst layer have been clubbed together as a single layer with a specied thermal conductivity. The model is such that the air meant for the cathode side ow elds of two cells is rst made to go through one-half of the cooling plate (where it picks up heat) and is then redistributed to the two channels. The thermophysical properties and thicknesses of the various materials used in the study are given in Tables 1e3  Case B is meant to simulate the heat generation within the catalyst layers and its subsequent removal by forced convection from the sides. This requires consideration of the entire width, i.e., all the 50 channels, of each bipolar plate. This makes the computations too cumbersome. Therefore the heat loss to the cathode air is not explicitly modeled; the strength of the heat generation term in the MEA is reduced to account for the heat loss (which is equal to the mass ow rate* specic heat * temperature difference between outlet and inlet of the cathode air). Thus, a domain (Fig. 4) of length of 100 mm, width of

Fig. 3. Schematic diagram of the computational domain for Case A showing the halfchannel in the cooling plate, and cathode ow channels in two cells.

Fig. 4. Schematic diagram of the computational domain for Case B with four fuel cell units placed sandwiched between two half-cooling plates.

470

E. Harikishan Reddy, S. Jayanti / Applied Thermal Engineering 48 (2012) 465e475

100 mm and thickness of 15.864 mm is modeled. It may be noted that the thermal conductivities of the GDL, the membrane and the catalyst layer are very nearly the same while that of the graphite plate is an order of magnitude higher. Therefore, the calculation domain is divided into two parts: that belonging to the MEA (membrane electrode assembly) with a thermal conductivity of 1.7 W/m K and that belonging to the graphite with a thermal conductivity of 20 W/m K. A distributed and equivalent volumetric heat generation term is specied in the MEA while only conduction takes place in the graphite region. Convective heat transfer from the sides is modeled by specifying a convective heat transfer boundary condition. Since the heat loss to the cathode air is not explicitly modeled, it is not necessary to solve for the velocity eld and the problem reduces to one of conduction through a composite domain with convective thermal boundary conditions. The computational domain and the boundary conditions applied for the two cases are shown schematically in Figs. 3 and 4 for Case A and Case B, respectively. The CFD simulations have been carried out using the commercial CFD software ANSYS-FLUENT, version 6.3.26. The discretization is formally second-order accurate. Grid independence of the results has been veried by comparing predictions of temperature contours in Case A on the catalyst surfaces for 1.03 105 cells, 7.7 105 cells and 11.7 105 cells. Marginal differences (of the order of 0.1 K in the predicted maximum temperature) were found between the predictions with the last two grids. The results obtained with 11.7 105 cells are used in the discussion below. 3. Results and discussion 3.1. Heat removal primarily by cathode air In order to study heat removal by cathode air alone, the ow domain shown in Case A is used with external cooling restricted to

that by natural convection, which is represented by specifying a convective heat transfer coefcient of 9 W m2K1, and an ambient air temperature of 303 K. Air, at a ow rate equal to a specied stoichiometric factor and at an inlet temperature of 303 K, enters the cooling plate. As it ows through, it picks up heat at a rate determined by the temperature distribution in the cooling plate. At the exit, it splits into parts and ows through the cathode side ow channels of two cells. (The geometry modeled here has one-half of the cooling plate only; in the real case, the ow from the cooling plates splits into four parts and enters the four adjacent channels.) Depending on the air temperature and the surrounding bipolar plate temperature, it may pick up heat or it may even lose heat to the bipolar plate. A zero-gauge pressure boundary condition is employed at the exit from the cathode ow elds so that the ow distribution in the two adjacent cells may be different. Each of the two adjacent HTPEMFCs has a cathode catalyst layer. A volumetric heat generation term, representing the net heat to removed, amounting to 913 W for the entire stack, is specied in each of the catalyst layers. Typical results from these calculations are shown in Fig. 5 where the temperature contours are compared for different air ow rates, expressed in terms of stoichiometric factor. The temperature contours in a z-x plane passing through the mid-ow eld height are shown here. This enables one to see the air temperature in the cathode side ow elds and in the cooling plate as well as the temperature in the surrounding membranes, GDLs and bipolar plates. It can be seen that for a stoichiometric factor of 3, the temperature in the cell is of the order of 700e800 K, which is well above the limit for the safe operation of the PBI membrane. As the stoichiometric factor is increased, the temperature in the cells decreases, and at a stoichiometric factor of 10, the cell temperature varies in the more acceptable range of 400e500 K. The results from these calculations are summarized in Table 4 which shows the air ow rate, the amount of heat removal by air and by free convection, and the maximum temperature within the cell. It can be seen that when the air ow rate is twice that required to supply the oxygen necessary to maintain a current density of

Fig. 5. Temperature contours of cell layers for stoichiometric factor of cathode air of (a) 3, (b) 5, (c) 7, (d) 9 and (e) 10.

E. Harikishan Reddy, S. Jayanti / Applied Thermal Engineering 48 (2012) 465e475 Table 4 Variations in the amount of heat removed with stoichiometric ratio of air in Case A. Stoichiometric factor for cathode air 2 3 4 5 6 7 8 9 10 11 Velocity m/s 1.963 2.945 3.926 4.908 5.889 6.871 7.852 8.834 9.815 10.797 Amount energy carried in though inlet air W 9.53 5.90 9.88 13.30 16.60 19.70 22.80 25.90 28.90 31.90 Amount heat removed out by coolant W 732.89 779.70 812.48 835.24 852.40 865.70 877.00 886.70 895.10 902.50 Heat removed by free convection W 189.74 139.30 110.50 91.160 77.30 67.10 58.90 52.30 46.90 42.50

471

Maximum temperature in the stack K 983 814 714 646 595 577 526 501 481 464

0.42 A/cm2, i.e., for a stoichiometric factor of two, the air ow rate is so low that it heats up considerably; the cell temperature is also very high even though a signicant amount of heat is removed by free convection. It is only when the stoichiometric factor is increased to 10, i.e., only when ten times more oxygen than what is electrochemically required is supplied, do we see the maximum cell temperature within acceptable limits. The above results have been shown for a current density of 0.42 A/cm2, which is fairly small compared to that for an LT-PEMFC. Studies have indicated [7] that it can be as much as 0.93 A/cm2 at a cell voltage of 0.5 V. The higher current density implies higher heat generation rate and therefore a requirement for a higher heat removal rate. In order to see the effect of this higher current density on the temperature distribution, calculations have been repeated for the same computational domain as in Fig. 3 with a higher volumetric heat generation rate corresponding to an average current density of 0.93 A/cm2. The results, in terms of the variation of the maximum material (GDLs, catalyst layers, bipolar plates, membranes; note that the temperature variation is not very large in the different media) temperature in the cell with the stoichiometric factor, are shown in Fig. 6. It can be seen that while there is a signicant increase in the maximum surface cell temperature at low stoichiometric factors, the effect is not so large at high stoichiometric factors. It may be noted that the air ow rate in the two cases (low and high current density) is different for the same stoichiometric factor because the oxygen requirement is directly proportional to the current density. Thus, it appears that a stoichiometric factor of about ten is required to maintain the temperature of the materials to within 473 K over a range of current densities.

3.2. Taking account of current density variation The above results show that there is a temperature variation across the cell. It is well known that, at a constant voltage, the current density varies with temperature. For HT-PEMFCs, this relation among the three variables is captured in the following empirical formula [39,40]:

Vcell

  V0 RTcell i i0 R i Rcon icell ohmic cell ln cell 4ac F i0 l1

(12a)

where ac is the cathode transfer coefcient, io is the exchange current density, Rohmic and Rconc are the ohmic and concentration losses (expressed as equivalent resistances) and l is the stoichiometric factor. Korsgaard et al. [39,40] expressed these as the linear functions of temperature:

ac a0 Tcell b0
Rohmic a1 Tcell b1 Rcon a2 Tcell b2 io a3 eb3 Tcell

(12b) (12c) (12d) (12e)

They obtained the constants using their own experimental data from a PBI membrane-based HT-PEMFC; these are given in Table 5. This formula enables one to take account of current density variations arising out of temperature variations if the voltage is known. In the present study, since the problem is formulated as a constant voltage problem, Eq. (12) is expressed in terms of a current density as the dependent variable and temperature as the independent variable in the following way for a constant voltage of 0.6 V:
3 2 icell 1:377107 Tcell 0:0001767Tcell

0:078555Tcell 11:77

(13)

Using Eq. (13), it is now possible to calculate the local current density which will then dene the equivalent heat source term
Table 5 Values used for varying current density with temperature [39,40]. Charge transfer constant, a0 Charge transfer constant, b0 Ohmic loss constant, a1 Ohmic loss constant, b1 Diffusion limitation constant, a2 Diffusion limitation constant, b2 Limiting current constant, a3 Limiting current constant, b3 Open circuit voltage, V0 2.761 103 0.9453 1.667 104 0.2289 1.667 104 0.4306 33.3 103 0.04368 0.95 K1 e U K1

U U K1 U

Fig. 6. Variation of the maximum surface temperature with the stoichiometric factor of air at current density of 0.42 A/cm2 (solid line) and 0.93 A/cm2 (dashed line).

A e Volts

472

E. Harikishan Reddy, S. Jayanti / Applied Thermal Engineering 48 (2012) 465e475

through Eq. (10) for a local temperature on the catalyst surface calculated through the CFD calculations of the mass, momentum and energy balance equations. We have introduced Eq. (13) into the CFD calculations in the form of a user dened function within FLUENT and calculated the temperature eld again. Since the current density decreases with temperature as per Eq. (12), the heat source decreases in areas of low temperature. Since the design criterion is that the temperature on the catalyst surface should not exceed 473 K, this results in a lesser current density and thus lesser power output as compared to the case of assumed constant current density which then led to a decrease in the maximum temperature. This gave rise to the possibility of decreasing the cathode air ow rate. Eventually, through an iterative calculation, it was found that a reduced cathode stoichiometric factor of 9.6 gave a maximum catalyst temperature of 463 K. The spatial variation of the current density, the heat source and the temperature on the surface of the catalyst layers and in the bipolar plates predicted under this condition, namely, stoichiometric factor of 9.6 and constant voltage of 0.6 V, are shown in Fig. 7. It may be noted that in order to bring out the contours clearly, a scale factor of 10 is used in the y-direction in Fig. 7. It can be seen that from Fig. 7a that there is nearly 20% variation in the current density over the mean value which itself is reduced by about 17% compared to the nominal value. As a result, the heat source also varies considerably (Fig. 7b). The predicted temperature on the catalyst layers (Fig. 7c) varies between 423 K and 463 K with a relatively cold spot forming on the catalyst layer closest to the cooling plate and a hot spot forming on the other catalyst layer. The predicted temperature in the two bipolar plates is shown in Fig. 7d shows a nearly monotonic decrease in the temperature in the region under the rib; however, in the under the channel region, the ow effects on the temperature distribution are clearly seen. The temperature proles at various z-positions, i.e., in the air ow direction in the cooling plate, at mid-channel height, are shown in Fig. 8 for an air ow rate corresponding to a stoichiometric factor of 9.6. Here, the position z 0 corresponds to the entry of the air into the cooling channel. It can be seen the temperature variation in the z-direction, i.e., along the ow direction, is rather small except in the ow elds where ow effects can be seen. However, in the xdirection, i.e., across the GDLs, the catalyst layers and the bipolar plates, there is rather considerable temperature difference, amounting to about 45 K between the maximum and the minimum values. As expected, the temperature gradients within the cooling channel are rather high; these can be attributed to the fact that the cathode air enters the cooling plate at a temperature of 303 K and gets heated, rather fairly uniformly, to about 435 K at the exit. These calculations show that cooling of the stack by cathode air alone is possible but at a high stoichiometric factor of 9.6. Under these conditions, the temperature variation over the cell and across the different layers in the stack is conned to about 25 K around the mean value. As shown in Fig. 7, this leads to a variation of about 20% in the current density about the mean value. The power generated is also reduced by about 15% from its nominal value. 3.3. Heat removal assisted by forced convection Since there is no liquid water formation in HT-PEMFCs, there is no pronounced mass transport-limited drop in the polarization curves at high current densities [7]. Experimental data show that the stoichiometric factor does not have any effect on the polarization curve for current densities of up to 1 A/cm2. In view of this, it is not necessary to use high stoichiometric factors and increase thereby the parasitic cost of pumping air. Thus, if stack cooling can be assisted by having a forced convection (natural circulation alone is not enough as shown in Section 2.3), the stack can be operated at

Fig. 7. The spatial variation of the (a) current density (b) the volumetric heat source and (c) the temperature in the catalyst layers and (d) the temperature variation in the bipolar plates.

lower stoichiometric factors. This can be readily induced in a transport application by directing the ambient air to ow over the cell when the vehicle is moving. In order to study this possibility, the computational domain shown in Case B (Fig. 4) is used. As noted earlier, forced circulation may introduce large temperature variations within the cell and the objective of these simulations is to estimate how much external cooling is required. Noting that the temperature gradients within the cell are very small if cooling is achieved by passing air through the cooling plate and the cathode ow elds (Fig. 2), the effect of cathode air cooling is represented in Case B simulations by reducing the heat generation rate appropriately. This enables the computational domain to be divided into two regions, one of high thermal conductivity 20 W/m2 K corresponding to the region occupied by the bipolar plates and the

E. Harikishan Reddy, S. Jayanti / Applied Thermal Engineering 48 (2012) 465e475

473

Fig. 8. Temperature proles along the stack thickness of two cells and one cooling plate for a stoichiometric factor of 9.6.

cooling plate and the other having low thermal conductivity (1.5 W/m2 K) consisting of the membrane, the catalyst layers and the GDLs. The total heat generation rate (which is now reduced by an amount equal to sensible heating of the cathode air) is distributed uniformly over the MEA. The temperature distribution over this composite material plate is calculated subject to a specied convective heat transfer coefcient and with an ambient temperature of 303 K. Typical results obtained from these calculations are shown in Fig. 9 for two convective heat transfer coefcients, namely, 37 and 50 W/m2 K at a stoichiometric factor of 3. It can be seen immediately that the temperature in the core region is higher, by about to 20e30 K, than the sides through which heat is being removed. The results from such calculations, performed for a range of stoichiometric factors, are summarized in Table 6 which shows the maximum and the minimum surface temperatures as a function of the stoichiometric factor and the convective heat transfer coefcient. Also shown in the table is the estimated air velocity required to achieve the specied heat transfer coefcient. This is obtained by modeling the stack as a rectangular block and using the following correlation for heat transfer in cross-ow of air over it [41]:

Nu 0:14 Re0:66

(14)

Here Nu is the Nusselt number dened as hD/k where h is the convective heat transfer coefcient, D is the hydraulic mean diameter and k is the thermal conductivity and Re is the Reynolds number dened as UD/n where U is the free stream velocity and n is the kinematic viscosity. It can be seen from the table that for a given stoichiometric factor, as the convective heat transfer coefcient increases, the maximum temperature (Tcore) decreases. Also, the difference between the core temperature (Tcore) and the surface

Fig. 9. Temperature contours for external ow and coolant ow in channel with a convective heat transfer coefcient of (a) 37 W/m2 K and (b) 50 W/m2 K.

474

E. Harikishan Reddy, S. Jayanti / Applied Thermal Engineering 48 (2012) 465e475

Table 6 Minimum and maximum temperature of the stack in Case B. Stoichiometric factor for cathode air h 25 W/m2 K (U 3.1 m/s) Tsur (K) 0 1 2 3 4 5 6 826 795 732 670 607 545 482 Tcore (K) 878 837 769 701 634 566 497 h 37 W/m2 K (U 5.6 m/s) Tsur (K) 671 630 588 547 505 464 422 Tcore (K) 719 672 625 578 532 485 438 h 50 W/m2 K (U 8.9 m/s) Tsur (K) 571 541 510 480 450 420 389 Tcore (K) 618 583 547 511 476 441 405 h 75 W/m2 K (U 16.4 m/s) Tsur (K) 475 456 436 417 398 378 359 Tcore (K) 522 498 473 448 423 398 374 h 100 W/m2 K (U 25.4 m/s) Tsur (K) 428 414 400 386 371 358 343 Tcore (K) 474 455 435 416 397 378 358

temperature (Tsurf) decreases. Both these effects are as expected; what is perhaps surprising is that in spite of the good thermal conductivity of graphite, the temperature difference can be as much as w30 K. The table also shows that as the stoichiometric factor increases, the core temperature as well as the temperature difference between the core and the surface decreases. However, it may be noted that the simulation in Case B does not consider the temperature variations caused by the cooling air entering at 303 K which is shown to induce up to 50 K variation in the cell temperature. The implications of these results on the operation of the stack are as follows. With external convection-assisted case, the designer will have the option of running the cell at a high stoichiometric factor or with a high convective heat transfer coefcient to maintain the cell temperature within reasonable limits. In a transport application, the estimated free stream air velocity required to provide the heat transfer coefcient listed in Table 6, may be related to the speed of the moving vehicle. The present study shows that when the vehicle is stationary, it may be necessary to use a high stoichiometric factor to keep the engine cool while at high speeds, say, of the order of 60 km/h (i.e., 16.7 m/s) which corresponds to an h value of 75 W/m2 K in Table 6, a stoichiometric factor of two is sufcient to maintain cell temperatures to within 200  C. 3.4. Heat removal by forced convection alone It can be seen from the results of Sections 3.1 and 3.2 that fairly large temperature differences, of the order of 50 K, are created within the cell in this case due to the feeding of the air into the cooling plate at 303 K. Results from Section 3.3 show that forced convection at an air velocity in the range of 16e25 m/s introduces a temperature variation of only w20 K in the cell. Thus, one way of reducing temperature variations within the cell would be not to use cathode air for cooling at all; forced convection alone could be used to effect the cooling. This case is simulated by performing the calculations for the computational domain shown in Fig. 4 with a heat generation term corresponding to a stoichiometric factor of zero, i.e., without considering the cooling coming from the preheating of air from 303 K to 473 K. The results of calculations for different convective heat transfer boundary conditions are shown in the rst row of Table 6. It can be seen here that the core temperature is at an acceptable value of 474 K only at a convective heat transfer coefcient of 100 W/m2 K corresponding to an air velocity of 25 m/s. However, at this condition, the minimum temperature in the cell is 428 K, i.e., again a temperature difference to 46 K. Increasing the heat transfer coefcient further to 150 W/m2 K reduces this temperature slightly to 45 K but also reduces the core temperature to 427 K which is well below the target cell operating temperature of 473 K. Thus, it appears that temperature differences of the order of 50 K over the entire cell are inevitable with any of the three strategies considered here. However, over most of the region, the variation may be only of the order of 20 K.

4. Conclusion The present CFD simulations of the ow and the temperature eld within the stack of a 1 kWe HT-PEMFC have given a temperature mapping over a cell under various operational strategies. The following conclusions can be drawn from this study:  HT-PEMFCs require external cooling in order to maintain reasonably low cell temperatures.  It is possible to use the cathode air supply to also serve as a coolant to keep the cell temperatures low. However, a large stoichiometric factor, of the order of 10, is required to maintain cell temperatures of w200  C without forced convective cooling. For transport applications, where the stack is mounted on a moving vehicle, the required external cooling can be achieved by directing the ambient air to ow over the stack. In such a case, the stack can be operated at a reduced stoichiometric factor.  Too high an air draft may reduce the cell temperatures significantly to well below the desired operating temperature of the cell; a combination of cathode air cooling and induced draft may be the right choice in a transport application.  There can be about 15% deviation in the local current density from its mean value due to the variation in the temperature across the catalyst layer. This has the effect of reducing the maximum power that can be drawn from a given fuel cell. Achieving more uniformity in the temperature, through proper thermal management, is therefore necessary to extract the maximum power from an HT-PEMFC.  With proper cooling strategy, the temperature variations within the cell may be reduced to about 20 K over most of the cell and to about 50 K in isolated spots. References
[1] X. Li, Principles of Fuel Cells, Taylor & Francis, New York, 2006. [2] S. Bose, T. Kuila, T.X.H. Nguyen, N.H. Kim, K.T. Lau, J.H. Lee, Polymer membranes for high temperature proton exchange membrane fuel cell: recent advances and challenges, Progress in Polymer Science 36 (2011) 813e836. [3] H.J. Kim, S.Y. Cho, S.J. An, Y.C. Eu, J.Y. Kim, H.K. Yoon, H.J. Kweon, K.H. Yew, Synthesis of poly(2,5-benzimidazole) for use as a fuel-cell membrane, Macromolecular Rapid Communications 25 (2004) 894e897. [4] M. Geormezi, V. Deimede, N. Gourdoupi, N. Triantafyllopoulos, S. Neophytides, J.K. Kallitsis, Novel pyridine-based poly (ether sulfones) and their study in high temperature PEM Fuel Cells, Macromolecules 41 (2008) 9051e9056. [5] J. Lobato, P. Caizares, M.A. Rodrigo, D. beda, F.J. Pinar, A novel titanium PBIbased composite membrane for high temperature PEMFCs, Journal of Membrane Science 369 (2011) 105e111. [6] J. Lobato, P. Caizares, M.A. Rodrigo, D. beda, F.J. Pinar, Enhancement of the fuel cell performance of a high temperature proton exchange membrane fuel cell running with titanium composite polybenzimidazole-based membranes, Journal of Power Sources 196 (2011) 8265e8271. [7] J. Zhang, Y. Tang, C. Song, J. Zhang, Polybenzimidazole membrane based PEM fuel cell in the temperature range of 120e200  C, Journal of Power Sources 72 (2007) 163e171. [8] J. Zhang, Z. Xie, J. Zhang, Y. Tang, C. Song, T. Navessin, Z. Shi, D. Song, H. Wang, D.P. Wilkinson, Z.S. Liu, S. Holdcroft, High temperature PEM fuel cells, Journal of Power Sources 160 (2006) 872e891.

E. Harikishan Reddy, S. Jayanti / Applied Thermal Engineering 48 (2012) 465e475 [9] DuPont fuel cells, Membrane Electrode Assemblies MEA3 and MEA5, http:// www2.dupont.com/FuelCells/en_US/assets/downloads/dfc501.pdf, 2011-11-01. [10] S.G. Goebel, Evaporative cooled fuel cell, US Patent 6960404, assigned to General Motors Corporation, 2005. [11] A. Faghri, Z. Guo, Challenges and opportunities of thermal management issues related to fuel cell technology and modeling, International Journal of Heat and Mass Transfer 48 (2005) 3891e3920. [12] S.G. Kandlikar, Z. Lu, Thermal management issues in a PEMFC stack - A brief review of current status, Applied Thermal Engineering 26 (2009) 1276e1280. [13] G. Zhang, S.G. Kandlikar, A critical review of cooling techniques in proton exchange membrane fuel cell stacks, International Journal of Hydrogen Energy 37 (2012) 2412e2429. [14] H.C. Ju, C.Y. Wang, Effects of coolant channels on large-scale polymer electrolyte fuel cells (PEFCs), International Journal of Automotive Technology 9 (2008) 225e232. [15] J. Choi, Y.H. Kim, Y. Lee, K.J. Lee, Y. Kim, Numerical analysis on the performance of cooling plates in a PEFC, Journal of Mechanical Science and Technology 22 (2008) 1417e1425. [16] S.H. Yu, S. Sohn, J.H. Nam, C.J. Kim, A numerical study to examine the performance of multi-pass serpentine ow elds for cooling plates in polymer electrolyte membrane fuel cells, Journal of Power Sources 194 (2009) 697e703. [17] S.M. Baek, S.H. Yu, J.H. Nam, C.J. Kim, A numerical study on uniform cooling of large-scale PEMFCs with different coolant ow eld designs, Applied Thermal Engineering 31 (2011) 1427e1434. [18] M. Matian, A. Marquis, N. Brandon, An experimentally validated heat transfer model for thermal management design in polymer electrolyte fuel cells Proceedings of IMechE Part A, Journal of Power and Energy 224 (2010) 1069e1082. [19] S. Asghari, H. Akhgar, B.F. Imani, Design of thermal management subsystem for a 5 kW polymer electrolyte membrane fuel cell system, Journal of Power Sources 196 (2011) 3141e3148. [20] J.H. Koh, A.T. Hsu, H.U. Akay, M.F. Liou, Analysis of overall heat balance in self-heated proton-exchange-membrane fuel cells for temperature predictions, Journal of Power Sources 144 (2005) 122e1228. [21] Y.J. Sohn, G.G. Park, T.H. Yang, Y.G. Yoon, W.Y. Lee, S.D. Yim, C.S. Kim, Operating characteristics of an air-cooling PEMFC for portable applications, Journal of Power Sources 145 (2006) 604e609. [22] K.P. Adzakpa, J. Ramousse, Y. Dub, H. Akremi, K. Agbossou, M. Dostie, A. Poulin, M. Fournier, Transient air cooling thermal modeling of a PEM fuel cell, Journal of Power Sources 179 (2008) 164e176. [23] R. Cozzolino, S.P. Cicconardi, E. Galloni, M. Minutillo, A. Perna, Theoretical and experimental investigations on thermal management of a PEMFC stack, International Journal of Hydrogen Energy 36 (2011) 8030e8037. [24] Y. Zhang, M. Ouyang, Q. Lu, J. Luo, X. Li, A model predicting performance of proton exchange membrane fuel cell stack thermal systems, Applied Thermal Engineering 24 (2004) 501e513. [25] S. Yu, D. Jung, Thermal management strategy for a proton exchange membrane fuel cell system with a large active cell area, Renewable Energy 33 (2008) 2540e2548. [26] T. Sousa, M. Mamlouk, K. Scott, A dynamic non-isothermal model of a laboratory intermediate temperature fuel cell using PBI doped phosphoric acid membranes, International Journal of Hydrogen Energy 35 (2010) 12065e12080. [27] J. Peng, S.J. Lee, Numerical simulation of proton exchange membrane fuel cells at high operating temperature, Journal of Power Sources 162 (2006) 1182e1191. [28] S.J. Andreasen, S.K. Kr, Modelling and evaluation of heating strategies for high temperature polymer electrolyte membrane fuel cell stacks, International Journal of Hydrogen Energy 33 (2008) 4664e4995. [29] S.J. Andreasen, L. Ashworth, I.N.M. Reman, P.L. Rasmussen, M.P. Nielsen, Modeling and implementation of a 1 kW, air cooled HTPEM fuel cell in a hybrid electrical vehicle, ECS Transactions 12 (2008) 639e650. [30] J. Scholta, M. Messerschmidt, L. Jrissen, Ch. Hartnig, Externally cooled high temperature polymer electrolyte membrane fuel cell stack, Journal of Power Sources 190 (2009) 83e85.

475

[31] T.W. Song, K.H. Choi, J.R. Kim, J.S. Yi, Pumpless thermal management of water-cooled high-temperature proton exchange membrane fuel cells, Journal of Power Sources 196 (2011) 4671e4679. [32] T. Sousa, M. Mamlouk, M.,K. Scott, An isothermal model of a laboratory intermediate temperature fuel cell using PBI doped phosphoric acid membranes, Chemical Engineering Science 65 (2010) 2513e2530. [33] H. Ju, C.Y. Wang, S. Cleghorn, U. Beuscher, Nonisothermal modeling of polymer electrolyte fuel cells I. Experimental validation, Journal of the Electrochemical Society 152 (2005) A1645eA1653. [34] H. Ju, H. Meng, C.Y. Wang, A single-phase, non-isothermal model for PEM fuel cells, International Journal of Heat and Mass Transfer 48 (2005) 1303e1315. [35] A. Shamardina, A. Chertovich, A.A. Kulikovsky, A.R. Khokhlov, A simple model of a high temperature PEM fuel cell, International Journal of Hydrogen Energy 35 (2010) 9954e9962. [36] K. Scott, M. Mamlouk, A cell voltage equation for an intermediate temperature proton exchange membrane fuel cell, International Journal of Hydrogen Energy 34 (2009) 9195e9202. [37] J.P. Holman, Heat Transfer, sixth ed. McGraw-Hill, Singapore, 1986. [38] B. Cunningham. The development of compression moldable polymer composite bipolar plates for fuel cells, Ph.D. Thesis, Virginia Polytechnic Institute and State University. 2007. [39] A.R. Korsgaard, R. Refshauge, M.P. Nielsen, M. Bang, S.K. Kr, Experimental characterization and modeling of commercial polybenzimidazole-based MEA performance, Journal of Power Sources 162 (2006) 239e245. [40] A.R. Korsgaard, M.P. Nielsen, M. Bang, S.K. Kr, Modeling of Co Inuence in PBI Electrolyte PEM Fuel Cells, the 4th International Conference on Fuel Cell Science, Engineering and Technology, Irvine, CA, June 19e21, 2006. [41] T. Igarashi, Heat transfer from a square to an air stream prism, International Journal of Heat and Mass Transfer 28 (1985) 175e181.

Nomenclature
vap DHT : enthalpy of vaporization (J) DHg;T : enthalpy of water formation in gas phase (J) DHlf : enthalpy of water formation in liquid phase (J) DGT: Gibbs free energy (J) DHT: enthalpy of water formation (J) DST: entropy of reaction (J K1)

A: area (m2) F: Faraday constant (C mol1) i: current density (A m2) I: current (A) io: exchange current density (A m2) n: number of electrons in electrochemical reaction Ncell: number of cells Nu: Nusselt number P: electrical power (W) Q: thermal energy produced by reaction (W) Qirrev: irreversible heat of reaction (W) Qrev: reversible heat of reaction (W) Rconc: concentration losses Re: Reynolds number Rohmic: ohmic losses T: temperature (K) V: voltage (V) Subscripts cell: single cell st: stack Surf: surface of the layer Core: core of the layer Greek letters

ac: cathode transfer coefcient l: stoichiometric factor