Proceedings
the ASME 2011
9th International
on
Proceedings of the 9th International Conference
on of
Nanochannels,
Microchannels,
andConference
Minichannels
Nanochannels, Microchannels, and Minichannels
ICNMM2011
ICNMM2011
June 19-22, 2011, Edmonton,
Canada
June 19-22, 2011, Edmonton, Alberta, CANADA
ICNMM2011-58252
ICNMM2011-58252
EFFECT OF CHANNEL GEOMETRY ON TWO-PHASE FLOW STRUCTURE IN FUEL
CELL GAS CHANNELS
Cody D. Rath and Satish G. Kandlikar*
Department of Mechanical Engineering
Rochester Institute of Technology
Rochester, NY USA
*[email protected]
ABSTRACT
Water management issues continue to be a major
concern for the performance of polymer electrolyte membrane
(PEM) fuel cells.
Maintaining the optimal amount of
hydration can ensure that the cell is operating properly and
with high efficiency. There are several components that can
affect water management, however one area that has received
increased attention is the interface between the gas diffusion
layer (GDL) and the gas reactant channels where excess water
has a tendency to build up and block reactant gasses. One key
parameter that can affect this build up is the geometry of the
microchannels. The work presented here proposes an optimal
trapezoidal geometry which will aid in the removal of excess
water in the gas channels. The Concus-Finn condition is
applied to the channel surfaces and GDL to ensure the water
will be drawn away from GDL surface and wicked to the top
corner of the channel. An ex situ setup is designed to establish
the validity of the Concus-Finn application. Once validated,
this condition is then used to design optimal channel
geometries for water removal in a PEM fuel cell gas channel.
NOMENCLATURE
PEM – Polymer Electrolyte Membrane
GDL – Gas Diffusion Layer
GC – Gas Channel
1.0 INTRODUCTION
Over the past few years, polymer electrolyte membrane
(PEM) fuel cells have become an increasingly popular as a
possible replacement for the internal combustion engine (ICE).
These fuel cells are an attractive alternative due to the fact that
they operate without the use of fossil fuels, are more efficient
than typical ICE’s, and have no harmful emissions, with heat
and water as the only byproduct. However, the management of
this product is still a fundamental issue that needs to be
addressed. In order for the cell to operate efficiently, there
must be a certain level of hydration present in the membrane.
On the other hand, if there is too much water being generated,
a phenomenon known as flooding occurs in which the excess
water blocks the reaction sites causing poor performance.
There have been several methods to investigate the water
management issues in PEM fuel cells. Recently, focus has
been on the interface between the gas diffusion layer (GDL),
which is a porous layer that allow for an even distribution of
reactant gases at the catalyst layer, and the reactant gas
channels (GC) [1-6]. Tuber et. al [6] was the first group to use
a transparent fuel cell to study the effects of the GDL materials
on water management. It was found that GDLs with a
hydrophobic Teflon™ (PTFE) coating tend to distribute water
more randomly within the channels. The results from this
work highlighted the need for a understanding ways of
controlling over water within the fuel cell components.
In 2007, Ous et. al [2] visualized water accumulation in
the flow channels using a transparent fuel cell to study the
effects of air flow, external loading, and droplet location. It
was found that droplet size is inversely proportional to air
velocity. It was also found that droplets growing in the center
of the channel developed much more rapidly than those
growing in contact with the side wall indicating the
importance of the side wall effects. Simulations were also
1
Copyright © 2011 by ASME
Downloaded 29 Jun 2012 to 129.21.225.197. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
completed which indicate capillary force to be the dominant
driving force behind droplet growth. This agrees with the
more previous work of Wang et al. [7] in 2001 in which a
model was developed to predict single and two phase flow
within a channel. This model also determined that water
accumulation is controlled by capillary action and molecular
diffusion.
Kandlikar et. al [3, 4], Lu et. al [8], and Owejan et. al [5,
9] have done extensive work involving visually accessible fuel
cells. Transparent components were used to visually study the
effects of GDL structure, channel geometry and maldistribution
in a 50 cm2 fuel cell both ex situ and in situ [3, 4, 8]. Neutron
Radiography was also used to identify water accumulation
within an operating fuel cell for various channel geometries
and material sets [5, 9]. It was found that flow patterns, i.e.
slug, film and mist, emerged based on the operating conditions
of the fuel cells and that more water accumulates in the
channels at lower air flow rates causing a decrease in
operational efficiency.
Metz et. al [1], investigated the effect of the channel
geometry on the water removal characteristics of the reactant
gas channels. A novel trapezoidal channel design was used to
remove water The work by Metz et al. primarily used the
simulation of droplet growth near a channel wall to determine
the force drawing the droplet away from the GDL surface. The
results show that the surface properties and channel geometries
have a profound effect on the water accumulation. This is in
agreement with Zhang et. al [10] which found
that
hydrophilic channels are preferable for water removal.
Venkatraman et. al [11] established the need for the
incorporation of contact angle and contact angle hysteresis in
fuel cell applications. This has become an increasingly
popular area of study. A balance of forces must occur in order
for the contact angle to remain stable and several factors can
contribute to the shape of a droplet, i.e. – surface tension,
gravity, and density [12]. Fang et. al [13] conducted 3-D
simulations of contact angle hysteresis and highlighted the
importance of understanding the effect of surface tension on
the contact line. It was also found that there is an inherent
relationship between the hysteresis and the fluid Capillary
number.
The work presented here will incorporate the importance
of contact angle into the fundamental understanding of PEM
fuel cell water accumulation. Experiments are conducted to
investigate the dynamics of a droplet growing near the side
wall of a PEM gas channel. High-speed visualization is used
to analyze the transient hysteresis that occurs as a droplet is
introduced to two perpendicular surfaces with varying static
contact angles. The results obtained with this work will be
extended in future work into a fundamental model that will be
used to describe the forces that drive slug growth and water
accumulation within PEM fuel cells.
2.0 EXPERIMENTAL
The objective of the experimental setup described below is
to visualize a droplet emerging during the dynamic transition
from growing on the GDL surface to simultaneous contact at
the GDL-droplet-channel side wall interfaces. Once the
droplet behavior is characterized, the results can be used to
validate a new channel design that will help improve water
management based on the channel geometry.
2.1 Surface Characterization
Before the corner effects could be investigated, the contact
angles were established to characterize the side wall surface as
well as the GDL base material, namely SGL-25BC. Typically,
PEM fuel cell gas channels are made out of graphite, however,
for this setup, a polycarbonate acrylic, namely Lexan™ was
used to represent the graphite for improved imaging and
machinability. The contact angles for both surfaces are very
comparably as shown in Table 1. These angles were measured
using a VCA Optima Surface Analysis System from AST
Products, Inc. The advancing contact angle, θa, and receding
contact angle, θr were measured for all surfaces to determine
the contact angle hysteresis which is given as ∆θ = θa – θr. An
example image of the contact angle measurement can be seen
in Figure 1, which shows the advancing contact angle for (a)
Lexan, (b) graphite and (c) GDL. The lines represent the
measured contact angle. The values listed in Table 1are an
average of five samples measured with the advancing angles
within 2% of the average and the receding angles within 10%
of the average.
(a)
(b)
(c)
2
Copyright © 2011 by ASME
Downloaded 29 Jun 2012 to 129.21.225.197. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
Fig. 1: Image of advancing contact angle for (a) Lexan, (b)
gra
Surface
Material
θa (°)
θr (°)
∆θ (°)
phit
Base
SGL -25BC
147
138
9
e
and
Side Wall
Lexan™
85
61
24
(c)
Side Wall
Graphite
89
34
55
GD
L
Table 1: Static advancing and receding contact angles
2.2 Test Setup
A test setup was designed to simulate droplets that would
be produced by the electrochemical reaction in a PEM fuel cell.
Figure 2 shows a general overview of the system which is
comprised of a constant supply syringe pump, high speed
digital camera capable of up to 24,000 fps, a computer for
control and analysis, and a test section. All components are
mounted on a vibration isolation table.
The test section is comprised of three main components: a
base piece machined out of Lexan, which is used to hold the
GDL sample as well as to facilitate the water supply, a side
wall, which is mounted on the top side of the GDL near a
preferential pore and, finally a rotating end cap that the side
wall is clamped to and allows for various open angles between
the base and side wall surfaces. Figure 2 also shows a
graphical representation of the test section design. It can be
seen that the droplet is pushed through the GDL through the
pore at a constant flow rate until it emerges on the top side. As
the droplet emerges, it eventually contacts the side wall where
the movement of the droplet is then dictated by the corner
geometry.
Fig. 2: Schematic of test setup
2.3 Data Analysis
As the droplet grew from the GDL, high speed video was
captured at 50x magnification with frame rates ranging from
250 to 1000 fps. After contact was made and the droplet had
reached equilibrium the test was completed and the videos
were saved for further analysis. The GDL and sidewall
surfaces were dried off and the test was repeated several times
to establish repeatability and validate the results comparatively.
After a few tests were completed at the 90° angle, the side wall
was rotated so that the open angle between the GDL and side
wall was decreased. The whole process was then repeated for
various angles ranging from 15° to the previously mentioned
90°. A database was collected for various side wall materials
at varying open angles for further video analysis.
Figure 3 shows a simplified schematic of a droplet to
illustrate the key data points measured, namely the overall
droplet height (diameter) just before contact, the height of the
side wall upper and lower contact lines as well as the distance
between the corner intersection and the inner and outer contact
lines on the base. The contact angles are also measured for all
four contact lines and are given as θUCL and θLCL respectively
for the upper and lower side wall contact lines and θICL and
θOCL respectively for the inner and outer contact lines on the
base. From these measurements, the contact line velocities can
be calculated and the contact angle information is used to
determine the Concus-Finn condition as discussed in the
following section.
Fig. 3: Graphical representation of droplet interfaces
3.0 RESULTS AND DISCUSSION
3.1 Droplet Dynamics
3
Copyright © 2011 by ASME
Downloaded 29 Jun 2012 to 129.21.225.197. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
Using the test setup and procedure explained in Section 2,
the movement of the contact lines, as well as the change in
contact angles is measured on a frame by frame basis to help
establish the contact line velocities and position relative to the
droplet/side wall. An example of this can be seen in Figure 4
which shows images taken from a droplet growing on the GDL
with a perpendicular untreated Lexan side wall. In these
images, the wall is highlighted by the dark line in the center,
with the original droplet on the left, and a reflection in the
wall surface appearing on the right. It can be seen that at time
t = 1 msec, the droplet contacts the side wall and begins to
spread rapidly on the surface. The lower contact line (LCL)
quickly reaches the corner, at t = 3 msec, allowing for it to be
completely filled. At the same time, the upper contact line
(UCL) climbs the wall until the equilibrium is reached.
Fig. 5: Plot of contact line heights at 2α = 90°
The velocity of the upper and lower contact lines, shown
in Figure 6 as VUCL and VLCL respectively, is calculated from
these height measurements based on a 1 msec time step. The
two velocities are initially high as the droplet begins to
transitions from equilibrium. Both velocities then decrease as
the equilibrium height is reached. It can also be seen that an
oscillating trend in the movement of the contact line as well,
especially for the UCL (solid symbols). This indicates that
there is a stick-shift type of movement experienced by the
contact line as it moves along the surface which can be
explained by the change in the contact angle as the contact line
gets pinned on the surface. This mechanism is explained in
more detail in the following section.
Fig. 4: Sequence of images showing a droplet growing on a GDL
base with Lexan side wall at 2α = 90°
The height of the upper and lower contact lines, HUCL and
HLCL respectively, were measured from the frames and are
plotted in Figure 5. For the UCL (solid symbols) it can be seen
the height of the contact line increases until the equilibrium
height is reached at t = 4 msec. The LCL (open symbols)
however, decreases exponentially until the contact line reaches
the corner, where HLCL = 0, at t = 3 msec.
Fig. 6: Plot of contact line velocities at 2α = 90°
3.2 Concus – Finn Condition
The Concus-Finn condition can be a useful tool to help
describe the liquid behavior at the intersection of two surfaces
with an open angle of 2α. Figure 7 illustrates an example of
4
Copyright © 2011 by ASME
Downloaded 29 Jun 2012 to 129.21.225.197. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
the plot similar to the one developed by Concus-Finn [14].
The two contact points that are of interest for the fuel cell gas
channel application are the two contact angles that are interior
to the corner, given as θICL and θLCL. These values are plotted
on the graph with coordinates given as (x,y) equal to (θICL,
θLCL). The location of this point can be used to graphically
predict whether or not the corner will fill. A point located
within the rectangle below the limiting line indicates the
corner will be filled when the droplet contacts the side wall. If
the point falls in the D1- region, the corner will not fill. For
the geometry and orientation being tested, sections D1+, D2+,
and D2- are absorbed into the R region to form to main regions,
filling and non-filling. Using the measured values listed
previously in Table 1, it can be seen in Figure 7 that the Lexan
surface falls below the line for a perpendicular wall. This
indicates that the corner should be filled when a droplet is in
contact with the GDL base and the side wall as shown in the
images in Figure 4.
Fig. 7: Plot of Concus-Finn Condition for a GDL base and
Lexan side wall at 2α = 90°
As mentioned in Section 1, due to the manufacturing of
gas channels, as well as GDL intrusion into the GDL, the
interface between the side wall and GDL might not make a
perfect 90° angle. Therefore, the effect of varying side wall
angles on the contact line height and velocity is also
investigated. To illustrate this, Figure 8 shows plot of the
theoretical Concus-Finn condition for an open angle of 2α =
60° and 2α = 45° respectively. It can be seen that the point
falls directly in between the two open angles and that for 2α =
60° (–), the point is below the line indicating that the corner
should fill. For the 2α = 45° (---), the point falls above the line
indicating non-filling of the corner.
Fig. 8: Plot of Concus-Finn Condition for a GDL base and
Lexan side wall at 2α = 60° and 2α = 45°
The corresponding images for these two values of open
angle can be seen in Figures 9 and 10. Similarly to the images
shown for 2α = 90°, the wall is highlighted by the dark line in
the center with the main droplet on the left and a reflection of
that droplet on the right. From the sequence of images it can
be seen that for and open angle of 2α = 60° the droplet behaves
similarly to that of 2α = 90°. The droplet spreads as soon as it
makes contact with the side wall and after the first 5 msec the
entire corner has filled. However, for 2α = 45°, the droplet
remains pinned near the point of contact on the side wall at t =
1 msec. Because of the transition phase, the contact line doe
move slightly however, once an equilibrium contact angle is
established, it can be clearly seen that the corner is not being
filled by the droplet after the same amount of time. This is due
to the fact that the surface tension is acting as the driving force
and is preventing the droplet from spreading to the corner as
determined by the CF condition.
5
Copyright © 2011 by ASME
Downloaded 29 Jun 2012 to 129.21.225.197. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
untreated Lexan side wall at 2α = 50°. Figure 11 shows a
sequence of images taken from one of these testes. Given a
certain droplet size and starting height, it was observed that
once the droplet contacts the side wall, it is completely drawn
to that surface and detaches from the GDL base. These images
show that at t = 6 msec the droplet has completely detached
from the GDL base and remains pinned on the Lexan surface.
The still images from t = 8 msec and t = 10 msec also show
that another droplet is beginning to grow out of the pore where
the original droplet emerged.
Fig. 9: Sequence of images of a droplet contacting GDL
and untreated Lexan at 2α = 60°
Fig. 11: Sequence of images showing a droplet in contact
with a GDL base and an untreated Lexan side wall
where 2α = 50°
Fig. 10: Sequence of images of a droplet contacting
GDL and untreated Lexan at 2α = 45°
Through this experimental setup, the Concus-Finn
Condition has been shown to be a useful for designing the
geometry of a PEM FUEL CELL gas channel for improved
water management. By evaluating the limits shown by the plot
in Figures 7 and 8, the limiting value of the open angle, 2α
Limit, can be expressed as a function of the wall and base contact
angles, θW and θB respectively and is given as:
2α Limit = (θW + θ B ) − π
Subsequent droplets continued to grow out of the pore and
were drawn into the initial droplet that remained on the side
wall. This process continued until the droplet was large
enough to contact both the side wall and the GDL again.
Figure 12 shows an image taken after 5 seconds for this test. It
can be seen that when the droplet reconnects with the GDL
base the CF condition still holds and the droplet still does not
get drawn into the corner.
(1)
The value for limiting open angle can be used to design an
optimal corner geometry for a PEM fuel cell gas channel. For
untreated Lexan and a GDL base, the limiting open angle is
found to be 52°. Therefore, a corner angle of 50° is proposed
to be the most efficient corner open angle for improved water
management. This is tested using the test setup up discussed
previously with a droplet growing on a GDL base with an
6
Copyright © 2011 by ASME
Downloaded 29 Jun 2012 to 129.21.225.197. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
like this is so the water can be completely removed from the
GDL surface allowing for the maximum transport of reactant
gases to the catalyst layer.
4.0 CONCLUSION
Fig. 12: Image of a droplet in contact with a GDL base and
an untreated Lexan side wall where: 2α = 50° at t = 5 sec
This phenomenon also occurred for certain tests carried
out with an open angle of 2α = 45°. This prompted a closer
look into the forces acting on the droplet. Not only are the
surface tension forces determining whether or not the corner
will fill, they are also imposing a net force on the droplet,
potentially causing it to detach from the GDL surface even
without a shearing gas flow (as typically seen in PEM fuel cell
gas channels). Through this, the new channel is designed such
that both open angles in contact with the GDL are 50°. Figure
13 shows a conceptual schematic of this proposed gas channel.
50°
0.77 mm
1 mm
Fig. 13: Schematic of proposed channel design
The intent of this new design is such that when the droplet
contacts the side wall, it will not be drawn into the corner
formed by the GDL and channel wall. Instead, the droplet will
detach from the GDL and become wicked in the top corner.
Due to the geometry of this corner, the water will completely
fill the corner and the water can then be removed from the
channel through an annular film in the upper right and left
corners. The benefit of having a water removal mechanism
The work presented here has brought attention to some of
the fundamental issues that are still affecting water
management in PEM fuel cells. An experimental setup has
been designed to investigate the effect of corner geometry and
surface properties on water buildup in polymer electrolyte
membrane (PEM) fuel cells. The approach simulated the
water droplet growth over the porous gas diffusion layer (GDL)
and focused on the interaction between the emerging droplet
and corner interfaces, namely the GDL base and channel side
wall. This experimental setup allows for the changing of the
side wall angle to various degrees as a way of testing the
geometry of the corner interface.
It has been shown that the contact angle is an important
parameter that must be incorporated in order to accurately
study and understand the transient behavior of water-channel
interactions. High speed video has shown that the transition
from a resting droplet on a flat surface to a droplet pinned in a
corner is a complex process that requires a more in depth
investigation. Once initial contact was made between the
droplet and the wall, the droplet was drawn into a corner and a
transition took place from the hydrophobic shape resting on
the GDL surface to a hydrophilic equilibrium on the channel
wall. The height of the contact line was found to rise until a
stable height was reached. The actual transition from
hydrophobic to hydrophilic was shown to be an oscillatory
process inducing a contact angle hysteresis on both the GDL
surface as well as the perpendicular wall.
Through this work, a new channel design was proposed to
improve water management in PEM fuel cells. The ConcusFinn Condition was shown to successfully predict the behavior
of the droplet. This was then used to design a channel in
which the droplet will detach from the base and be wicked into
the top corner away from the GDL surface. This process will
prevent the water from blocking the pores on the GDL which
causes poor performance and low efficiency.
5.0 ACKNOWLEDGEMENTS
This work was completed in the Thermal Analysis,
Microfluidics, and Fuel Cell Laboratory in the Department of
Mechanical Engineering at the Rochester Institute of
Technology and was supported by the US Department of
Energy under contract No. DE-FG36-07G017018.
6.0 REFERENCES
7
Copyright © 2011 by ASME
Downloaded 29 Jun 2012 to 129.21.225.197. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
[1] Metz, T., Paust, N., Mfiller, C., Zengerle, R., and Koltay,
P., 2007, "Passive Water Removal in Fuel Cells by Capillary
Droplet Actuation," Proc. 20th IEEE International Conference
on Micro Electro Mechanical Systems (MEMS 2007), Kobe,
JAPAN, pp. 49-57.
[2] Ous, T., and Arcoumanis, C., 2007, "Visualisation of Water
Droplets During the Operation of Pem Fuel Cells," Journal of
Power Sources, 173(1), pp. 137-148.
[3] Kandlikar, S. G., Lu, Z., Domigan, W. E., White, A. D.,
and Benedict, M. W., 2009, "Measurement of Flow
Maldistribution in Parallel Channels and Its Application to ExSitu and in-Situ Experiments in Pemfc Water Management
Studies," International Journal of Heat and Mass Transfer,
52(7-8), pp. 1741-1752.
[4] Kandlikar, S. G., Lu, Z., Lin, T. Y., Cooke, D., and Daino,
M., 2009, "Uneven Gas Diffusion Layer Intrusion in Gas
Channel Arrays of Proton Exchange Membrane Fuel Cell and
Its Effects on Flow Distribution," Journal of Power Sources,
194(1), pp. 328-337.
[5] Owejan, J. P., Gagliardo, J. J., Sergi, J. M., Kandlikar, S.
G., and Trabold, T. A., 2009, "Water Management Studies in
Pem Fuel Cells, Part I: Fuel Cell Design and in Situ Water
Distributions," International Journal of Hydrogen Energy,
34(8), pp. 3436-3444.
[6] Tuber, K., Pocza, D., and Hebling, C., 2003, "Visualization
of Water Buildup in the Cathode of a Transparent Pem Fuel
Cell," Journal of Power Sources, 124(2), pp. 403-414.
[7] Wang, Z. H., Wang, C. Y., and Chen, K. S., 2001, "TwoPhase Flow and Transport in the Air Cathode of Proton
Exchange Membrane Fuel Cells," Journal of Power Sources,
94(1), pp. 40-50.
[8] Lu, Z., Kandlikar, S. G., Rath, C., Grimm, M., Domigan,
W., White, A. D., Hardbarger, M., Owejan, J. P., and Trabold,
T. A., 2009, "Water Management Studies in Pem Fuel Cells,
Part Ii: Ex Situ Investigation of Flow Maldistribution, Pressure
Drop and Two-Phase Flow Pattern in Gas Channels,"
International Journal of Hydrogen Energy, 34(8), pp. 34453456.
[9] Owejan, J. P., Trabold, T. A., Jacobson, D. L., Arif, M., and
Kandlikar, S. G., 2007, "Effects of Flow Field and Diffusion
Layer Properties on Water Accumulation in a Pem Fuel Cell,"
International Journal of Hydrogen Energy, 32(17), pp. 44894502.
[10] Zhang, F. Y., Yang, X. G., and Wang, C. Y., 2006, "Liquid
Water Removal from a Polymer Electrolyte Fuel Cell," Journal
of the Electrochemical Society, 153(2), pp. A225-A232.
[11] Venkatraman, M., 2009, "Estimates of Pressure Gradients
in Pemfc Gas Channels Due to Blockage by Static Liquid
Drops," Elsevier, International Journal of Hydrogen Energy.
[12] Diaz, M. E., Fuentes, J., Cerro, R. L., and Savage, M. D.,
2010, "Hysteresis During Contact Angles Measurement,"
Journal of Colloid and Interface Science, 343(2), pp. 574-583.
[13] Fang, C., Hidrovo, C., Wang, F. M., Eaton, J., and
Goodson, K., 2008, "3-D Numerical Simulation of Contact
Angle Hysteresis for Microscale Two Phase Flow,"
International Journal of Multiphase Flow, 34(7), pp. 690-705.
[14] Concus, P., and Finn, R., 1994, "Capillary Surfaces in a
Wedge - Differing Contact Angles," Microgravity Science and
Technology, 7(2), pp. 152-155.
8
Copyright © 2011 by ASME
Downloaded 29 Jun 2012 to 129.21.225.197. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm