Scholarly article on topic 'Alloy Selection and Die Design for Stamped Proton Exchange Membrane Fuel Cell (PEMFC) Bipolar Plates'

Alloy Selection and Die Design for Stamped Proton Exchange Membrane Fuel Cell (PEMFC) Bipolar Plates Academic research paper on "Materials engineering"

CC BY-NC-ND
0
0
Share paper
Academic journal
Procedia CIRP
OECD Field of science
Keywords
{"Proton exchange membrane fuel cell" / "bipolar plate stamping"}

Abstract of research paper on Materials engineering, author of scientific article — Travis Lee Smith, Anthony D. Santamaria, Jae Wan Park, Kazuo Yamazaki

Abstract Proton exchange membrane fuel cells have benefits over heat engines and batteries as an energy system for transportation, stationary power, and portable electronics. Before fuel cells can replace current energy systems, their manufacturing cost to performance ratio must be improved. The bipolar plate is a fuel cell component that contributes significantly to fuel cell manufacturing costs and is a key driver of performance. Metallic bipolar plates, with characteristic flow dimensions below the standard 1 mm, were fabricated using direct machining. Hydrogen/air fuel cells constructed of these plates were tested to ensure performance gains, using interdigitated flow operation and automobile-representative channel lengths. Finite Element Modelling (FEM) was used to investigate the manufacture of bipolar plates using low cost sheet metal stamping. Multiple bipolar plate alloys were compared based on their stampability, and die design parameters needed to stamp submillimeter channels were determined. Springback analysis was performed, and the effect of springback on fuel cell stack stresses was investigated.

Academic research paper on topic "Alloy Selection and Die Design for Stamped Proton Exchange Membrane Fuel Cell (PEMFC) Bipolar Plates"

Available online at www.sciencedirect.com

ScienceDirect

Procedia CIRP 14 (2014) 275 - 280

www.elsevier.com/locate/procedia

6th CIRP International Conference on High Performance Cutting, HPC2014

Alloy selection and die design for stamped Proton Exchange Membrane

Fuel Cell (PEMFC) bipolar plates

Travis Lee Smitha , Anthony D. Santamariaa, Jae Wan Parka, Kazuo Yamazakia,b*

aUniversity of California at Davis, 1 Shields Ave., Davis 95616, US bUniversity of California at Berkeley, 101 Sproul Hall, Berkeley 94704, US

* Corresponding author: E-mail address: kyamazaki@ucdavis.edu or kyamazaki@berkeley.edu

Abstract

Proton exchange membrane fuel cells have benefits over heat engines and batteries as an energy system for transportation, stationary power, and portable electronics. Before fuel cells can replace current energy systems, their manufacturing cost to performance ratio must be improved. The bipolar plate is a fuel cell component that contributes significantly to fuel cell manufacturing costs and is a key driver of performance. Metallic bipolar plates, with characteristic flow dimensions below the standard 1 mm, were fabricated using direct machining. Hydrogen/air fuel cells constructed of these plates were tested to ensure performance gains, using interdigitated flow operation and automobile-representative channel lengths. Finite Element Modelling (FEM) was used to investigate the manufacture of bipolar plates using low cost sheet metal stamping. Multiple bipolar plate alloys were compared based on their stampability, and die design parameters needed to stamp submillimeter channels were determined. Springback analysis was performed, and the effect of springback on fuel cell stack stresses was investigated. © 2014 Elsevier B.V. Thisisanopenaccessarticle under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).

Selectionandpeer-review under responsibility of the International Scientific Committee of the 6th CIRP International Conference on High Performance Cutting

Keywords: Proton exchange membrane fuel cell; bipolar plate stamping.

1. Introduction

The continued use of fossil fuels poses two primary concerns: their finite nature and non-renewability combine to create a dwindling supply, and the pollution generated by their combustion can contribute negatively to atmospheric and human health. Renewable and non-polluting hydrogen is an option to one day replace fossil fuel as our primary transportation energy currency. The hydrogen/air PEMFC is an attractive technology to harness hydrogen energy for applications ranging from decentralized electricity generation, to transportation, and portable electronics. PEMFCs can be: more portable and convenient to operate than other fuel cell chemistries, are more quickly "refueled" than batteries, and can reach higher efficiencies and emit dramatically less pollution than internal combustion engines. A main barrier to wider PEMFC adoption is their ratio of performance to manufacturing cost [1].

The bipolar plate is a PEMFC component that is a primary driver both of performance and manufacturing cost. The bipolar plate is typically the main structural element of a fuel cell stack (where many bipolar plates are compressed together in series), and serves as the transport pathway for the reactant gases and product water. In a fuel cell stack, the bipolar plates typically account for 60% of the mass and 30% of the manufacturing cost [2]. Improvements to the design and manufacturing processes of bipolar plates can therefore both increase fuel cell performance and decrease manufacturing cost, encouraging wider commercialization of this attractive technology.

Fuel cell performance, especially at high power operation, is limited by mass transport. Maximization of fuel cell performance is possible by optimizing bipolar plate geometry, especially by the reduction of fluid flow path lengths.

The manufacturing process typically associated with high quality, submillimeter feature generation is Computer Numeric Controlled (CNC) machining. Although flexible and

2212-8271 © 2014 Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).

Selection and peer-review under responsibility of the International Scientific Committee of the 6th CIRP International Conference

on High Performance Cutting

doi:10.1016/j.procir.2014.03.078

capable, this process is not an economically viable option for mass production of bipolar plates. Stamping is a low cost, high volume manufacturing process used for sheet metal fabrication. By applying state of the art CNC micromachining techniques to the fabrication of long life stamping dies, the economical mass production of high performance bipolar plates may be realized.

This study aims to determine optimal bipolar plate geometries and the stamping dies required to manufacture these bipolar plates. Prototype bipolar plates were machined by CNC milling and tested for performance. Stamping of these designs was then studied using the FEM software Abaqus. Five bipolar plate alloys were evaluated based on stampability. Stamping die parameters were then investigated for features down to 0.25 mm. Finally, an evaluation of springback, and its effect on fuel cell stack stresses was performed. The optimized die designs from this study will inform future work, where the stamping dies will be manufactured using CNC micromachining of long life die materials such as tungsten carbide.

2. Bipolar plate fundamentals

2.1. Bipolar plate design

Figure 1 depicts the most basic repeating unit of a PEMFC stack. An anode bipolar plate (the H2 side) and a cathode bipolar plate (the O2 side) sandwiching a Membrane Electrode Assembly (MEA). The bipolar plates consist of open gas flow channels (1), and lands (2) which contact the MEA. The MEA consists of the proton exchange membrane (4), surrounded by two electrodes (5) and two porous Gas Diffusion Layers (GDL) (3).

Fig. 1. Basic fuel cell unit (not to scale).

The anode bipolar plate channels evenly distribute hydrogen to the catalyst-containing electrode layer in the MEA, where it is broken down to its constituent electrons and protons. The electrons are conducted out of the fuel cell, to

the external load, through the bipolar plate. The protons travel through the proton exchange membrane, to the opposite electrode, where they combine with oxygen ions and electrons to form water. To maintain low electrical resistance between the bipolar plate and the GDL, these two layers must be compressed together. The GDL, typically a porous carbon paper or cloth, electrically connects the electrode and bipolar plate, while allowing fluid flow under the otherwise blocked lands.

To obtain the required voltage, current, or power, these repeating units of bipolar plates and MEAs are stacked together in series. A molded graphite bipolar plate and matching MEA are shown in Fig. 2 along with a fuel cell stack of similar components.

To replenish fuel consumed at the electrode, gas must flow from the channels, through the GDL. One method of increasing reactant flux to the electrode is reducing bipolar plate feature sizes. By decreasing channel and land widths, the characteristic diffusion lengths are shortened, increasing mass transport, and leading to higher performance. In machined graphite bipolar plates, a 79% increase in maximum current output was observed by decreasing channel and land widths from 1 mm to 0.25 mm [3].

In interdigitated bipolar plates, the pressure in adjacent channels is separately controlled, generating pressure driven convective flow under the lands. Testing [4] and modeling [5] of machined, interdigitated bipolar plates, with 1 mm wide channels, showed higher maximum power than for diffusion dominated parallel patterns, such as those used in [3].

Fig. 2. Graphite bipolar plate and MEA.

The US Department of Energy (DOE) has set a 2015 target of $3/kw for the manufacturing cost of bipolar plates, with a 2 Hz production rate [6]. These targets exclude direct machining as a viable manufacturing option. Stamping of thin metal sheets has been shown to be an effective manufacturing process for bipolar plates with channel sizes down to approximately 0.7 mm, using metal foils down to

0.05 mm in thickness [7]. Austenitic stainless steel is often the material of choice for bipolar plates, owing to its high strength and stampability, but less expensive alloys have been proven effective as well [8]. Indeed, since even stainless steels require a coating that simultaneously has high electrical conductivity and low corrosion potential (the fuel cell environment is acidic, warm, and wet), many options exist for bipolar plate alloys [9].

Long life stamping dies are required to maintain low bipolar plate manufacturing costs, with wear mitigation becoming increasingly urgent as the bipolar feature size decreases. H13 tool steel stamping dies experience significant die surface roughness evolution while stamping uncoated 316L stainless steel bipolar plates [10].

Advances in diamond based tooling and machine tool design have made possible the precision micromachining of extremely hard materials, such as tungsten carbide, with nanometer level surface roughness [11]. With the high hardness and low roughness of tungsten carbide stamping dies, the die replacement frequency, and therefore the manufacturing cost of bipolar plates may be dramatically improved.

FEM has proven to be an effective tool in analyzing the bipolar plate stamping process, eliminating the need for costly iterative die design experiments [12].

2.2. Springback and fuel cell stack stresses

In sheet metal stamping, the sheet is plastically deformed by rigid dies to permanently change its shape. After the die pressures have been removed, stamped parts tend to elastically relax back toward their original configuration. Springback can therefore give rise to finished parts that differ greatly from the expected geometry, possibly requiring die design changes [13].

Springback is strongly controlled by die geometry, such as the die radii and die clearances, as well as the stiffness and strain hardening behavior of the material being stamped [14].

Determining the springback of the stamped bipolar plates is crucial, as this can affect the compression stresses of the GDL in the assembled fuel cell stack. Fuel cell stacks are held in compression to ensure good electrical contact between GDL and bipolar plates, and to prevent gas leakage. Excessive GDL compressive stresses, caused by springback, can affect fuel cell performance by decreasing the GDL porosity, choking gas flow [15].

3. Fundamental study on the die design for thin foil bipolar plates

3.1. Testing of machined bipolar plate prototypes

Two sets of bipolar plates were machined from 6061-T6 aluminum (with 0.25 mm and 1 mm features respectively) using carbide tooling and a Sodick HS430L CNC milling machine. The manifold region of these two bipolar plates (showing separate high and low pressure channels for interdigitated operation) are shown in Fig. 3.

Fig. 3. 1 mm and 0.25 mm featured bipolar plates.

After machining, these bipolar plates were coated with a Ni/Rh coating to improve corrosion and electrical conductivity, then assembled into PEMFCs for performance testing.

The testing was conducted on an Arbin Instruments FCTS using 99.95% pure hydrogen and medical grade air. The MEA was comprised of SGL10 BC based GDL, bonded to a 0.4 mg/cm2 platinum loading catalyst layer. The membrane was Nafion 112. Fig. 4 shows that the 0.25 mm bipolar plate produces more power than the 1 mm.

2 0.8 .o

£ 0.6 (0

1mm _J"L A

-0.25mm ^¿¡¡¡¡¡H^L.

0 0.5 1 1.5 2 2.5 Current Density, amps/cm2

Fig. 4. 1 mm and 0.25 mm power curves.

The prototype testing indicates that submillimeter bipolar plate features are viable, from a performance standpoint. To investigate if submillimeter bipolar plates are manufacturable within the DOE guidelines, FEM was used to study the stamping process for 1 mm, 0.5 mm, and 0.25 mm bipolar plate features.

3.2. Bipolar plate material model

Five common bipolar plate alloys were selected for stampability evaluation: 316L stainless steel [16], Crofer 22APU (a high chromium ferritic stainless steel) [17], 5086-0 aluminum [18], 1100-0 aluminum [19], and Commercially Pure (CP) titanium [18]. The engineering stress strain curves for the five materials can be seen in Fig. 5. These curves represent annealed material at room temperature. This data was used to model the elastic-plastic behavior of each alloy in Abaqus.

Using this data, the strain hardening exponent (n) for each material was calculated, using the technique described in [20]. These values are listed in Table 1, along with each alloy's in-plane anisotropy, or Lankford coefficient (R).

-316l \

-22APU \

f^D Xi »

0.00 0.20 0.40 0.60 Strain, %

Fig. 5. Stress strain behavior of evaluated alloys. Table 1. Lankford coefficients and strain hardening exponents.

clearance, die radius, and die/sheet coefficient of friction on the stampability of each alloy. The values of the tested parameters are listed in Table 2 and described in Fig. 7.

Table 2. Die design parameters.

-CP Ti * Parameter Low Medium High

-5086 Die Clearance, mm 0.187 0.276 0.368

1100 Die Radius, mm 0.05 0.075 0.1

Coefficient of Friction 0.1 0.3 0.5

Material

316L 22APU CP Ti 5086 1100

1.03 1.07 2.85 0.67 0.7

0.55 0.29 0.31 0.41 0.27

3.3. Strain based damage model

The typical mechanism of failure in sheet metal stamping is localized necking through the sheet thickness. Rather than the difficult to produce forming limit diagrams typically used to predict stampability, localized necking can be predicted using widely available uniaxial tensile data [21].

Equation 1 [22] shows the equivalent strain at the onset of localized necking (eiw), based on the n and R values for each material from Table 1.

£ln +R

1 + 2 R

A stamping simulation is only considered successful (damage free) if none of the bipolar plate mesh elements exceeded this critical strain. The maximum stamped channel depth prior to the onset of localized necking was used as a measure of stampability.

3.4. Alloy selection screening and die design

The FEM tests were conducted on 0.05 mm thick foils, using a die that generated seven, 1 mm wide channel and lands (Fig. 6).

Fig. 6. Detail of stamping die.

A three factor, three level Box-Behnken experimental design was implemented to determine the effects of die

Clearance

Radius

Fig. 7. Die parameters.

A two dimensional planar Abaqus dynamic explicit simulation was used to test each combination of alloy and die design. The bipolar plate meshes consisted of 5,000 CPE4R elements, using enhanced hourglass controls. A mass scaling factor of 100 was used to speed the computation times.

3.5. Channel size effects

The initial die design simulations were all for 1 mm channels and lands. To stamp submillimeter features, it was necessary to determine how the findings from the 1 mm testing scaled to smaller features. The stamping response of a single 0.25 mm is different than a 1.0 mm, also, a larger number of 0.25 mm channels will fit on a bipolar plate of a given width than would 1.0 mm channels. These two effects were separately investigated.

More accurate simulations were conducted to study the channel size effects. These models used a finer mesh (20,000 elements) and no mass scaling. All size effect simulations used a die radius of 0.1 mm and a coefficient of friction of 0.1, thereby focusing on the dominating die clearance. The bipolar plate material used in the channel size study was 316L.

To determine the isolated effect of channel width on stampability, seven-channel bipolar plates with widths of 1 mm, 0.5 mm, and 0.25 mm were evaluated. Since die clearance was the controlling factor in the 1 mm screening tests, the channel size simulations used a die clearance scaled to the 1 mm value. Therefore, the 1 mm, 0.5 mm, and 0.25 mm bipolar plates had die clearances of 0.368 mm, 0.184 mm, and 0.092 mm respectively.

To determine the isolated effect of the total number of channels, 1 mm features were stamped in bipolar plates having 1, 3, 5, and 7 channels.

To investigate if the 0.25 mm prototype bipolar plate was manufacturable, a die design capable of stamping a bipolar plate with 0.25 mm by 0.25 mm channels and 0.25 mm lands was determined. Starting from the scaled 0.091 mm value, die clearance was increased until a 0.25 mm deep, damage free channel could be stamped.

3.6. Springback and fuel cell stack compression

To determine the effect of springback, each alloy was formed into a seven-channel, 1 mm bipolar plate with consistent channel depth of 0.21 mm (the max channel depth for the least stampable alloy, CP Ti).

The "sprung" meshes were then imported into a separate dynamic explicit model where each bipolar plate was compressed between rigid platens (Fig. 8), until they were returned to a flat configuration.

Fig. 8. Bipolar plate compression model.

4. Results and discussion

4.1. Alloy selection screening and die design

The marginal means plot for the alloy selection and die design simulation is seen in Fig. 9. Each value on the x-axis is read left to right: low, medium, and high (see Table 2).

£ 0.5

15 с с со

О 0 25

5086 APU22 1100 CP Ti

Clearance

Radius

Friction

Fig. 9.Alloy selection marginal means.

Fig. 9 shows that 316L is the most stampable alloy. Over the range of tested parameters, the die clearance is the dominant parameter affecting channel depth, followed by die radius. The coefficient of friction had a slight and

inconsistent effect on the maximum damage free channel depth.

4.2. Channel size effects

Table 3 shows the maximum damage free channel depth for the three tested feature widths. A normalized value is also listed (channel depth/width).

Table 3. Effect of channel width on stampability.

Channel width, mm Depth, mm Depth/width

1.0 0.516 0.516

0.5 0.325 0.65

0.25 0.208 0.832

Table 3 shows that the normalized stampability goes up for smaller feature sizes, when die clearance is linearly scaled. This result indicates that the prototype 0.25 mm bipolar plates could be stampable, with sufficient die clearance.

Table 4 shows the maximum damage free channel depth for 1 mm bipolar plates of 1, 3, 5, and 7 channels.

Table 4. Effect of the total number of channels on stampability.

# of channels

Depth, mm

0.663 0.588 0.514 0.592

A consistent decrease in stampability is noted from 1 to 5 channels, but the 7 channel case does not follow the trend (it is however, less stampable than the 1 channel case). This result indicated that the scaled 0.091 mm die clearance for 0.25 mm channels would likely need to be increased to stamp a bipolar plate of the prototype design.

The die clearance required to achieve a damage free channel depth of 0.25 mm was 0.175 mm (almost double the 0.091 mm linearly scaled value). Fig. 10 shows a section of the 0.25 mm feature bipolar plate.

Fig. 10. Detail of 0.25mm bipolar plate.

The large required die clearance, along with the large relative die radii, generated a bipolar plate resembling a rectangular aspect ratio sine wave, which is quite different than the square wave shape of the prototype plates. Triangular channels have been shown to improve liquid water transport, and the rounded lands may improve gas flow to the

electrodes [23], but these more pointed lands might also lead to increased GDL compressive stress.

4.3. Springback and fuel cell stack compression

The peak displacement (the maximum distance moved) for each alloy is reported in Table 5, along with the peak platen contact pressure. These values were measured at the specified platen separation, which yielded the largest contact area between the platen and bipolar plate.

Table 5. Springback comparison.

Material Platen separation, ^m Peak displacement, ^m Peak contact pressure, MPa

316L 228 27 11.5

5086 246 26 10.1

22APU 265 11 13.8

1100 244 25 2.1

CP Ti 241 40 5.4

The worst case peak displacement is still less than the foil thickness, so springback should not pose a difficulty in fuel cell stack assembly. Although springback displacement is easier to determine than contact pressure, the difference in alloy ranking between displacement and contact pressure indicates that displacement is insufficient to determine the full effect of springback on fuel cell stack stresses. The difference in platen separation indicates that each alloy, even with the same die design, creates bipolar plates of variable overall thickness.

5. Conclusions

PEMFC power output was shown to be higher for bipolar plates with 0.25 mm features than for 1 mm.

Due to its high stampability, 316L stainless steel is the most suitable bipolar plate alloy of those tested. Over the range of tested die design parameters, die clearance had the largest effect on stampability, die radius had a moderate effect, and the coefficient of friction had an inconsistent effect.

The required die clearance does not scale linearly with feature width. Generally, decreased stamapability with increasing channel number is expected.

316L requires moderate contact pressure to return its sprung configuration back to flatness. For low stiffness GDL, designing stamping dies to mitigate springback may be required with this alloy, though easier to flatten alloys can be selected if this is the driving engineering concern.

Using a sufficiently large die clearance, a 24-channel bipolar plate with 0.25 mm features can be formed at room temperature, from 0.05 mm thick 316L stainless steel.

Acknowledgements

The authors would like to thank Sodick and NS Tools for supporting the bipolar plate machining research and Professor

Masakazu Soshi for supplying additional Abaqus computation resources. Travis Smith is supported by the NSF, grant DGE-0948021.

References

[1] Wang Y, Chen KS, Mishler J, Cho SC, Adroher C. A review of polymer electrolyte membrane fuel cells: Technology, applications, and needs on fundamental research. Appl. Energy 2011;88:981-1007.

[2] Li X. Review of bipolar plates in PEM fuel cells Flow-field designs. Int. J. Hydrogen Energy 2005;30:359-71.

[3] Goebel SG, Impact of land width and channel span on fuel cell performance. J. Power Sources 2011;196:7550-54.

[4] Santamaria AD, Bachman J, Park JW. Design strategy for a polymer electrolyte membrane fuel cell flow-field capable of switching between parallel and interdigitated configurations. Int. J. Hydrogen Energy 2013;38:5807-12.

[5] Santamaria AD, Cooper NJ, Becton MK, Park JW. Effect of channel length on interdigitated flow-field PEMFC performance: A computational and experimental study. Int. J. Hydrogen Energy 2013;38:16253-63.

[6] Sinha J, Marcinkoski J, Katie O. V . A . 3 Cost Analyses of Fuel Cell Stacks / Systems 2010; DOE report.

[7] Lim JW, Lee DG. Development of composite-metal hybrid bipolar plates for PEM fuel cells. Int. J. Hydrogen Energy 2012;37:12504-12.

[8] Brady MP, Elhamid MA, Dadheech G, Bradley J, Toops TJ, Meyer HM, Tortorelli PF. Manufacturing and performance assessment of stamped, laser welded, and nitrided FeCrV stainless steel bipolar plates for proton exchange membrane fuel cells. Int. J. Hydrogen Energy 2013;38:4734-9.

[9] de Oliveira MCL, Ett G, Antunes RA. Materials selection for bipolar plates for polymer electrolyte membrane fuel cells using the Ashby approach. J. Power Sources 2012;206:3-13.

[10] Peng L, Liu D, Hu P, Lai X, Ni J. Fabrication of Metallic Bipolar Plates for Proton Exchange Membrane Fuel Cell by Flexible Forming Process-Numerical Simulations and Experiments. J. Fuel Cell Sci. Technology 2010;7:0310991-9.

[11] Peker MF, Cora ÖN, Ko$ M. Investigations on the variation of corrosion and contact resistance characteristics of metallic bipolar plates manufactured under long-run conditions. Int. J. Hydrogen Energy 2011;36:15427-36.

[12] Nakamoto K, Katahira K, Ohmori H, Yamazaki K, Aoyama T. A study on the quality of micro-machined surfaces on tungsten carbide generated by PCD micro end-milling. CIRP Ann. - Manuf. Technology 2012;61:567-70.

[13] Ling YE, Lee HP, Cheok BT. Finite element analysis of springback in L-bending of sheet metal. J. Mater. Process. Technology 2005;168:296-02.

[14] Panthi SK, Ramakrishnan N, Ahmed M, Singh SS, Goel MD. Finite Element Analysis of sheet metal bending process to predict the springback. Mater. Design 2010;31:657-62.

[15] Ge J, Higier A, Liu H. Effect of gas diffusion layer compression on PEM fuel cell performance. J. Power Sources 2006;159:922-7.

[16] Blandford RK, Morton DK, Snow SD, Rahl TE. Tensile stress-strain results for 304L and 316L stainless steel plate at temperature. Mater. Fabrication 2007;6:617-28.

[ 17] Thyssen Krupp VDM, Crofer 22 APU. Material Datat Sheet No. 4046, 2010.

[18] US Department of Defense. MIL-HDBK5J 2003.

[19] Takata N, Okitsu Y, Tsuji N. Dynamic deformation behavior of ultrafine grained aluminum produced by ARB and subsequent annealing. J. Mater. Science 2008;43:7385-90.

[20] ASTM International. Standard test method for tensile strain-hardening exponents ( n -values ) of metallic sheet materials. ASTM 2013;E636-07:1-8.

[21] Sing KP, Rao WM. Prediction of sheet-metal formability using tensile-test results. J. Mater. Process. Technology 1993;37:37-51

[22] Hill R. Theoretical plasticity of textured aggregates. Math. Proc. Cambridge Philos. Society 1979;85:179-91.

[23] Manso AP, Marzo FF, Barranco J, Garinkano X, Mujika MM. Influence of geometric parameters of the flow fields on the performance of a PEM fuel cell a review. Int. J. of Hydrogen Energy 2012;37:15256-87