ww.sciencedirect.com

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 9 ( 2 0 1 4 ) 8 8 6 1e8 8 6 7

Available online at w

ScienceDirect

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

Simulation and study of proposed modificationsover straight-parallel flow field design

Gerardo Martın Imbrioscia a,b,*, Hector Jose Fasoli c

aDepartamento de Investigacion y Desarrollo de Energıas Renovables (CITEDEF-EST), San Juan Bautista de La Salle

4397, Villa Martelli B1603ALO, Provincia de Buenos Aires, Argentinab Laboratorio de Simulacion y Diseno, Escuela Superior Tecnica del Ejercito General Manuel Nicolas Savio, Cabildo 15,

C1426AAA Ciudad Autonoma de Buenos Aires, ArgentinacEscuela Superior Tecnica del Ejercito General Manuel Nicolas Savio, CITEDEF-EST, Cabildo 15, C1426AAA Ciudad

Autonoma de Buenos Aires, Argentina

a r t i c l e i n f o

Article history:

Received 6 November 2013

Accepted 17 November 2013

Available online 23 January 2014

Keywords:

PEMFC

Simulation

Bipolar plate

* Corresponding author. Departamento de Inv4397, Villa Martelli B1603ALO, Provincia de B

E-mail address: gmimbrioscia@gmail.com0360-3199/$ e see front matter Copyright ªhttp://dx.doi.org/10.1016/j.ijhydene.2013.11.0

a b s t r a c t

Diverse CAD (Computer aided-Design) 3D bipolar plates model are presented. By using the

OpenFOAM software, an open source CFD (Computational Fluid Dynamic), hydrogen flow

simulations were carried out, obtaining velocities and pressure maps for each model.

Main objective resides on predict the flow behavior in response to the modifications

proposed on the bipolar plate geometry, such as width, depth and shape of the distributing

channels (collectors) as over the main channels. Channelers fins are also besought with the

purpose of direct the flow towards different zones, in order to homogenize the flow

distribution.

Copyright ª 2014, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights

reserved.

1. Introduction

Bipolar plates, constitutive elements of PEM fuel cells, have

different slotted designs on their faces in order to allow the

flow of the reactant gases. These channels have distinctive

patterns, being the “straight-parallel” design the case of study

of this paper. The main functions of bipolar plates are:

distributing the reactant gases inside the cell avoiding their

mixture, collecting the electric current outside the cell, man-

aging the water formed by the electrochemical reaction and

preventing the cell from flooding and transfer the heat pro-

duced inside the cell to the environment.

estigacion y Desarrollo deuenos Aires, Argentina. T(G.M. Imbrioscia).

2014, Hydrogen Energy P79

The gas flow field design has a fundamental role on the

gases pressure variation along the channels. This pressure

variations affect directly the amount of gases driven through

the Gas Diffusion Layer (GDL) to the catalytic reacting layer, as

is stated in Barreras [1], thus achieving a better cell

performance.

In this work several flow field designs are presented and

studied by Computational Fluid Dynamic technique (CFD). All

the designs shown in this research were devised with the aim

of solving problems detected by the author in a previous work

[2] and making it possible the manufacture in our facilities.

Different angle inlets, channel collectors configurations and

Energıas Renovables (CITEDEF-EST), San Juan Bautista de La Salleel.: þ54 11 4709 8100x1472.

ublications, LLC. Published by Elsevier Ltd. All rights reserved.

i n t e rn a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 9 ( 2 0 1 4 ) 8 8 6 1e8 8 6 78862

channel ratios (width to depth) are evaluated using Open-

FOAM, an open source finite volume code with remarkable

results.

2. Mathematical model

In order to numerically study the proposed flow field designs,

NaviereStokes (NS) equations were solved, considering

laminar flow in all cases. Steady-state was considered. GDL

was not taken into account here; water formation and heat

transfer have been neglected.

In the models, a 3D steady version of the incompressible

NaviereStokes equation is used as described in equations (1)

and (2), where r is the density, y the kinematic viscosity and

ui the i ¼ 1,2,3 component of the velocity field:

Momentum:

ujvui

vuj¼ �1

r

vpvxi

þ mv2ui

vxjvxj; (1)

Continuity:

vuj

vxj¼ 0: (2)

By using the SIMPLE (Semi-Implicit Method for Pressure

Linked Equations) algorithm [3], the result is reached when

the specified convergence criterion takes the value of 10�6.

The resulting system of linear equations is solved using a

Geometric Agglomerated Algebraic Multigrid (GAMG), jointly

with the relaxation factor of pressure (0.3) and velocity (0.7).

Even though flow was considered steady, the numerical

scheme needs a velocity and pressure initial conditions to

start the calculation, which can be seen in Table 1. The flow

selected was hydrogen at NTP (Normal Temperature and

Pressure) conditions.

3. Bipolar plates

Taking into account the modifications suggested by Dong

Hyup [4] related to collector dimensions in order to improve

the uniformity of the velocity fields and pressure drop, a wider

collector than that used in Ref. [2] is common through all

models shown below.

Bipolar plates presented consist of an active area of

40 � 40 mm, with 20 channels; where both inlet and outlet

ducts are square shaped with an area of 2 mm2. Models BP1,

BP3, BP4 and BP5 have inlet and outlet ducts oriented in the

same way as the channels, criterion which was changed after

the early results, as will be discussed in Section 4.

Designs were meshed using hexagonal elements, with

non-uniform mesh size due to a bell type biasing with a

compression factor of 1.5 in order to refine the mesh close to

the walls intersection.

Table 1 e Initial conditions for simulation.

Inlet velocity [m/s] 2

Pressure at exhaust [Pa] 101.325

Detailed information is attached to each model:

3.1. BP1

The relation between collectors and channel width is 2 to 1,

being the channel size of 1 � 1 mm.

3.2. BP3

The upper collector was designed with a negative slope line

from the inlet duct to the farthest channel, beginning with

2 mm and finishing with 1 mm width. Regarding the down-

stream collector, its shape is the exactly the inverted opposite

to the upstream one.

3.3. BP4

In this case, the upper collector was designed with a negative

slope curve from the inlet duct to the farthest channel,

beginningwith 2mmand finishingwith 1mmwidth, but with

a more pronounced decrease. About the downstream collec-

tor, its shape is exactly the inverted opposite to the upstream

one.

3.4. BP5

This model, the upper collector, was designed with an arc

curve with its maximum located over the central channels. As

regards the downstream collector, its shape is exactly the

opposite to the upstream one.

3.5. BP6

This design has several changes with respect to the models

presented before, in the way that the inlets as well as the

outlet ducts were collinear with the collectors and both col-

lectors width were increased to 4 mm.

A special arrangement was carried out over the channels,

increasing width over the central channel and decreasing it

towards the sides. This configurationwas kept formodels BP6,

BP7, BP9 and BP10.

3.6. BP7

A special intake design was applied at the channels entrance.

This modification was implemented to add resistance in the

first two channels where flow path was noticed to be more

evident [2]. Channelers were placed in the rest of the en-

trances, except in the last two channels, with the objective of

capturing and directing the largest amount of gas flow to the

central area of the cell.

3.7. BP9

This cell structure is similar to BP6. The difference lies on the

collectors depths, which were increased 0.5 mm.

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 9 ( 2 0 1 4 ) 8 8 6 1e8 8 6 7 8863

3.8. BP10

The last proposed geometry is based on BP6, where the depth

of the entire cell was modified. This modification implies

gradually reducing both channels and collectors depth from

the lateral areas, thus leaving a greater volume in the middle

of the cell.

4. Results

4.1. BP1

There is a noticeable tendency of the flow to circulate through

the externals channels as we can see in Fig. 1. This behavior

Fig. 1 e Numerical simulation results of BP1: (a) average

velocity values across the plate, (b) velocity field, (c)

pressure field.

matches with the work of Barreras [5], in spite of imple-

mentation of a wider collector to act as a compensation area,

to regularize the flow pattern.

Although a nonsymmetrical flow distribution was ex-

pected, special attention was placed on the inlet/outlet ducts

orientation. The fact of their collinearity with the external

channels increases the flow amount going across them,which

is reflected in a lower amount of gas available for the rest of

the cell.

Referring to the pressure field, gradual pressure gradient is

shown, without the existence of any overpressure spot.

4.2. BP3

At first sight there is no difference between the flow pattern

of this geometry and its predecessor, a remarkable non-

uniformity was obtained. A detailed comparison between

Figs. 1a and 2a shows that BP3 central channels maximum

Fig. 2 e Numerical simulation results of BP3: (a) average

velocity values across the plate, (b) velocity field, (c)

pressure field.

i n t e rn a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 9 ( 2 0 1 4 ) 8 8 6 1e8 8 6 78864

velocity is higher and smoother than that reached in the

previous case, although the flow uniformity problem was no

solved yet.

Looking at the pressure map, we can clearly see that the

difference between the zone close to the inlet and the outlet is

greater than the results from Fig 1(c).

4.3. BP4

In this model we can become aware of the importance of the

collector design. Only a variation of the selected type of curve

significantly impacts on the velocity profiles, proving to have

opposite operating conditions contrasted to other models

studied here. Nevertheless, flows still behavior in different

manners across the cell.

Fig. 3 e Numerical simulation results of BP4: (a) average

velocity values across the plate, (b) velocity field, (c)

pressure field.

Fig. 3(c) shows a poor performance of the design in the path

for reaching a stable pressure.

4.4. BP5

This configuration, contrary to what was expected, presented

a deficient flowmanagement, with almost an inexistent mass

flow in the middle of the cell; disregarding the advantages

obtained with high pressure rates, as we can see at Fig. 4.

4.5. BP6

With the modification stated in the previous section, we ob-

tained a really smooth velocity profile. The reduction of the

external width led to a sensitive decrease of the first and last

Fig. 4 e Numerical simulation results of BP5: (a) average

velocity values across the plate, (b) velocity field, (c)

pressure field.

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 9 ( 2 0 1 4 ) 8 8 6 1e8 8 6 7 8865

two channels flow speed, without sacrificing active area since

it was compensated by the increase in the middle channel

width in equal number of millimeters.

The fact that the collectors had been enlarged beyondwhat

was proposed at BP1, had a positive effect harmonizing

channels speed, yet the overall speeds dropped.

The pressure contour exhibits a quite homogeneous dis-

tribution at Fig. 5(c).

4.6. BP7

Obstruction placed in the channel closer to the intake gases

seemed to have accomplished its function, avoiding the

large entry of gas and redirecting it to adjacent channels. An

incorrect angle of fins position made the main gas flow

current drive to the first half of channels, leading to a lack of

Fig. 5 e Numerical simulation results of BP6: (a) average

velocity values across the plate, (b) velocity field, (c)

pressure field.

reactant gas in the other half, bringing to light the need for

redesigning this flow field geometry.

Again, due to the blockage, a zone of high pressure was

noted near the inlet duct as seen in Fig. 6(c). However, this

situation did not spread along the cell, but was contained and

lessened by the geometry, leaving a pattern of acceptable

pressures.

4.7. BP9

Increasing the collectors volume had a positive impact on the

flow pattern obtained. Aswe can see in Fig. 7(a), themaximum

velocity scalar value was now found in the middle channel,

but as opposed to the results obtained in BP4, the remainder of

the velocities were very close to this.

Fig. 6 e Numerical simulation results of BP7: (a) average

velocity values across the plate, (b) velocity field, (c)

pressure field.

Fig. 7 e Numerical simulation results of BP9: (a) average

velocity values across the plate, (b) velocity field, (c)

pressure field.

Fig. 8 e Numerical simulation results of BP10: (a) cross area

velocity field, (b) pressure field.

i n t e rn a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 9 ( 2 0 1 4 ) 8 8 6 1e8 8 6 78866

Fortunately, the pressure plot showed a regular gradient

across the cell marking this flow field design as the best option

in-between the models presented.

4.8. BP10

This last model showed promising results because its design

drastically reduced the flow in the first and last channel, while

maintaining all velocities at the same level. A particularity of

this bipolar plate is the low average speed compared with the

other models under study, clearly seen at Fig. 8(a), which

could cause several problems at the time of removing the

water produced into the cell.

A high pressure spot was expected at the inlet duct owing

to the small passage area at the beginning of the collector,

which resulted in an overall cell pressure increase.

Nonetheless, the gradient is similar to the one observed in the

last 3 models mentioned before.

5. Conclusions

Several bipolar plates have been numerically studied. CFD

simulations based on OpenFOAM SIMPLE code provided a

reliable response, being a useful tool to validate designs

modifications.

From models BP1, BP3, BP4 and BP5 we can emphasize the

importance of the inlet and outlets orientation, which could

pre-direct flow leading to a malfunctioning of the cell. It was

also demonstrated how a small change in the design of the

collector could drastically change the flow behavior across the

bipolar plate.

The functionality of the BP6 as a basic design criterion was

achieved, giving rise to numerous models. For flow manage-

ment, different techniques were proved: from BP7, where the

uses of obstructions revealed the importance of the correct

position of fins, to BP9 and BP10 where the modification of the

volumes of collectors and channels drove to uniform flow

distribution.

From the present studymodels BP6, BP9 and BP10 showed a

good performance both in the field of velocities as in the

pressure gradient, making them viable options for the next

part of the research, where the GDL will be included.

r e f e r e n c e s

[1] Lozano A, Valino L, Barreras F, Mustafa R. Fluid dynamicsperformance of different bipolar plates: part II. Flow throughthe diffusion layer. J Power Sources 2008;179(2):711e22.

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 9 ( 2 0 1 4 ) 8 8 6 1e8 8 6 7 8867

[2] Imbrioscia G, Lavorante M, Franco J, Heidenreich E.Simulacion de flujo gaseoso en placa bipolar de celda PEMFC.Mecanica Computacional 2011;vol. XXX. p. 2513e22.

[3] Patankar SV. Numerical heat transfer and fluid flow. NewYork, USA: McGraw-Hill; 1980.

[4] Kim Kwang Nam, Jeon Dong Hyup, Nam Jin Hyun, Kim ByungMoon. Numerical study of straight-parallel PEM fuel cells at

automotive operation. Int J Hydrogen Energy2012;37:9212e27.

[5] Barreras F, Lozano A, Valino L, Marın C, Pascau A. Flowdistribution in a bipolar plate of a proton exchange membranefuel cell: experiments and numerical simulation studies. JPower Sources 2005;144:54e66.