Droplet-Sidewall Dynamic Interactions in PEMFC Gas
Channels
Preethi Gopalan and Satish G. Kandlikar
J. Electrochem. Soc. 2012, Volume 159, Issue 8, Pages F468-F475.
doi: 10.1149/2.066208jes
Email alerting
service
Receive free email alerts when new articles cite this article - sign up
in the box at the top right corner of the article or click here
To subscribe to Journal of The Electrochemical Society go to:
http://jes.ecsdl.org/subscriptions
© 2012 The Electrochemical Society
F468
Journal of The Electrochemical Society, 159 (8) F468-F475 (2012)
0013-4651/2012/159(8)/F468/8/$28.00 © The Electrochemical Society
Droplet-Sidewall Dynamic Interactions in PEMFC Gas Channels
Preethi Gopalana,∗ and Satish G. Kandlikara,b,∗∗,z
a Microsystems Engineering, Rochester Institute of Technology, Rochester, New York 14623, USA
b Mechanical Engineering, Rochester Institute of Technology, Rochester, New York 14623, USA
Water management in proton exchange membrane fuel cells (PEMFC) has remained one of the most important issues preventing its
commercialization in automotive applications. Water accumulation on the gas diffusion layer (GDL) in a PEMFC acts as a barrier
for the reactant gas transport through the GDL to the catalyst layer. In this paper, a detailed ex situ experimental investigation has
been performed to observe the water droplet dynamics in the gas channel for channel open angles from 30◦ to 90◦ under controlled
gas flow velocities (0.2–2 ms−1 ). The primary goal of these experiments is to elicit the importance and relevance of the channel
open angles as well as the locations of the water droplet emergence on the droplet dynamics. Experimental data were analyzed using
Concus-Finn condition and it was observed that the channel angle along with the air velocities influence the water distribution and
holdup in the channel. Dynamic contact angle was seen as an important parameter in controlling the droplet-wall interaction. It was
revealed from experimentation that location of the droplet emergence has significant impact on water-sidewall interactions and the
resulting pressure drop across a droplet.
© 2012 The Electrochemical Society. [DOI: 10.1149/2.066208jes] All rights reserved.
Manuscript submitted February 29, 2012; revised manuscript received May 10, 2012. Published July 20, 2012. This was Paper 785
presented at the Boston, Massachusetts, Meeting of the Society, October 9–14, 2011.
Fuel cell technology has been one of the major areas of research
that has gained more importance in the past few years due to the
increasing demand of petroleum and its rising cost.1,2 One of the main
issues with Proton Exchange Membrane Fuel Cells (PEMFCs) that
is still preventing its commercialization in the automotive sector is
the efficient removal of the byproducts of reaction, heat and water.3
This is significant for the PEMFC as the pathways for the flow of
reactant gases to the catalyst layer of the fuel cell as well as that of
the liquid water are the same. Accumulation of liquid water on the
GDL surface in a channel hinders the reactants flowing toward the
reaction sites. Inefficient removal of the liquid water leads to flooding
of the gas channel.4 This causes localized fuel starvation which in
turn adversely affects the efficiency of the system.5 Under freezing
conditions, the accumulated liquid water leads to channel blockage
and GDL degradation which also lower the system output further.
At the same time, the proton exchange membrane also needs to be
appropriately hydrated to conduct protons from anode to cathode
side and therefore, lack of water also leads to loss of performance.6
Hence, efficient water management in fuel cells is one of the primary
requirements for proper operation and improved performance of the
system and forms the major motivation for this work. The critical
factor in the water removal from the channels is the droplet stagnation
on the GDL surface and the channel sidewalls. Therefore, a detailed
study of the droplet dynamics is conducted to analyze and explain the
water stagnation phenomenon on the GDL and the surface properties
that determine the criteria for droplet pinning and its removal. This
study is expected to address the fundamental aspects of the dropletsidewall interactions that will help in developing designs for efficient
removal of water from the channels.
In general, droplet interaction with a solid surface is characterized
by the contact angle that the droplet makes with the surface. For
a hydrophilic surface, the droplet would wet the surface and have
a contact angle less than 90◦ . For a hydrophobic surface, the water
droplet would form a hemispherical shape due to non-wetting property
of the surface and would not spread on the material surface. The
contact angle of the droplet on the GDL as well as on the sidewall
determines whether the droplet would stick and spread over the surface
to flood the channel or can be removed easily with the application of
gas flow in the channel.
Droplet pinning on the GDL surface is considered by many researchers as a principal issue in water removal.7–16 Clearly, pinning
of the droplets on the sidewall or on the GDL is not recommended
for efficient PEMFC operation. The static contact angle of the GDL
∗
Electrochemical Society Student Member.
Electrochemical Society Active Member.
z
E-mail: [email protected]
∗∗
is found to be a crucial factor to determine droplet pinning and its removal without considering the effect of sidewall.17 In microchannels,
as in the PEMFC, droplet removal from the gas channel is not only
dependent on the droplet contact angle with the GDL but also depends
on the static contact angle it makes with the sidewall. To ensure removal of the droplets in such scenarios, the airflow in the channel has
to overcome the surface forces to push the droplet out of the GDL. In
the process, the airflow in the channel changes the contact angle of
the droplet as it moves and hence, can be characterized by advancing
and receding contact angles. It was established that the advancing and
receding contact angle values for the droplet are different for different
GDL materials and the airflow rate required to remove a droplet depends on the difference between the advancing and receding contact
angles for a given GDL material.13 For the same material, droplet
removal from the gas channel is not only dependent on the GDL fiber
network but also on the PTFE content on the GDL. The PTFE content
in the GDL increases the hydrophobicity of the GDL which helps in
the water removal much easily by the air flow in the channel. It is
observed that Toray (TGP-H-060) has larger water accumulation in
the channel compared to SGL-25BC which may affect the fuel cell
performance.7,18
A few researchers have studied the effect of sidewall on the water
transport phenomena. Zhu et al.17 proposed that if the sidewall is
hydrophilic then the droplet would spread on the walls and form a
film. Continuous growth of the liquid water film on the GDL would
lead to blockage of the reactant pathways. In the case of hydrophobic
sidewalls it resulted in earlier detachment of the droplets from both the
GDL surface as well as the sidewalls. This result was contradictory
to the one shown by Zhang et al.19 in which the hydrophilic sidewalls
were reported to aid in easier detachment of water droplets compared
to the hydrophobic sidewalls. Akhtar et al.20 researched on the effect of
aspect ratio of the gas channel on the fuel cell efficiency and showed
that the measured pressure drop decreases with increasing channel
cross sectional area. It was also seen that for same cross sectional
area but for different aspect ratio the pressure drop was significantly
different. Hence, aspect ratio is believed to play an important role in the
water transport phenomenon taking place in the fuel cell. To evaluate
the importance of the channel sidewall, Hwang et al.21 conducted a
simulation to verify the droplet dynamics at the channel corner. It was
seen that the effect of the sidewall became more prominent as the
open angle between the sidewall and the base decreased. The effect
of the sidewall on droplet movement in the gas channel was studied
at a more fundamental level by Rath and Kandlikar.15 Filling of the
GDL-sidewall corner was seen as an important consideration. It was
shown that the different sidewall angles have very different effect on
the droplet behavior. It was also verified that Concus-Finn condition
could be used to determine whether the droplet would fill the corner
Journal of The Electrochemical Society, 159 (8) F468-F475 (2012)
F469
Figure 2. Experimental Test Set Up.
Figure 1. (a) Model showing the angles the liquid makes with the wedge
surface (b) Plot representing the Concus Finn condition for wedge surface
where the plot represents the contact angle made by the liquid on the wedge
surfaces. Points that fall in the shaded region R would have solution for the
−
+
Concus-Finn condition equation. For the points falling in the D+
1 , D1 , D2 and
22
no
solution
exists
reproduced
from.
D−
2
between the sidewall and the base and result in slug formation or
would pin on to the sidewall without filling the corner. Concus Finn
condition,22 states that the two surfaces are unrestrained if the given
condition is met.
α + θ < π/2
[1]
where 2α is the open angle of the channel and θ is the static contact
angle that the liquid makes with the surface. Figure 1 schematically
shows this condition. If the contact angle values fall in the shaded
region as shown in Figure 1b then the condition is valid and a solution
exists, i.e. the droplet would fill the corner of the channel and would
continue to flood the GDL surface, whereas if the values fall in the
region of D+ or D- region, then there exists no solution which corresponds to the droplet pinning onto the surface and no filling would
occur in the channel corner.
However, Rath and Kandlikar15 studied the droplet-sidewall interactions without considering any air flow. Therefore, there is a need
for understanding the droplet behavior at the corner of a PEMFC gas
channel to reduce the water accumulation on the GDL surface and
decrease the transport resistance for the gases. This would affect the
fuel cell performance due to lower gas diffusion to the catalyst layer.
The work presented here focusses on the behavior of droplets and their
interaction with the channel sidewalls in the presence of gas flow. This
work corresponds to the actual droplet behavior that prevails in the
PEMFCs and would help in gaining better insight about droplet dynamics, the droplet pinning on the GDL surface, and the effect of gas
flow on the droplet-channel wall interaction. This would aid in reducing the transport resistance in the gas channel and provides guidance
in the gas channel design.
droplet in the system. The gas flow into the channel from the Parker
Balston HPZA 18000 Zero Air Generator was via an air manifold that
was attached to the system at the rear end. The gas flow was controlled
using a rotometer which was attached just before the air manifold. The
entire test section was mounted on top of a test stand on a vibration
isolation table to mitigate any discrepancies due to vibrations. High
speed videos were recorded to observe the droplet dynamics using a
Keyence VW-6000 high speed digital camera. The droplets dynamics
with the sidewall was captured at frame rates of 60–250 fps. For the
different channel angles used in our study, the channel cross section
was varied in order to maintain a constant hydraulic diameter of
3 mm. The channel sizes used were slightly larger compared to the
actual fuel cell dimensions to provide visible access to the droplet and
its behavior inside the channel. Further work on smaller channels will
be performed in the follow-up work.
Experimental details.— Experiments were performed by varying
the air velocity introduced into the channel. As the channel sizes used
were larger than the real world fuel cell dimensions, the air flow rates
used in the system were matched to the air velocities observed in
the real fuel cell. The superficial air velocities of 0.2–2 ms−1 were
simulated for a range of current densities from 0.1 to 1 A/cm2 . This
would correspond to an active area of 18.4 cm2 in a typical fuel cell that
meets the Department of Energy specifications.7 To maintain dynamic
similarity with a real fuel cell, the Re number for the given air flow rate
in the channel was ensured to lie within the range of 39–390. Table
I shows the relation between the air flow velocities corresponding to
current densities for single channel of 183 mm long 0.4 mm deep
and 0.7 mm wide. For these experiments, Mitsubishi Rayon Corp.
(MRC- 105) GDL which was treated with 6% PTFE and coated with
a microporous layer (MPL) in house in GM was used. The thickness
of the GDL is approximately 245 μm. The PTFE gasket of 0.178
mm thickness is used to ensure the appropriate compression under
the applied load. In the GDL, a preferential pore of 180–200 μm was
made for the droplet to appear on the surface at the desired location to
provide easier visualization access and capture the droplet dynamics.
Since the droplet size introduced in the channel before it contacted the
sidewall was much larger than the inlet pore size, the pore diameter
would have negligible effect on the droplet interaction with the gas
Experimental
Experimental set up.— The experimental setup shown in Figure 2
was developed for an ex situ investigation of the effect of air flow on
droplet pinning on the GDL and the sidewall of the channel. The test
section consists of different plates mounted on top of each other. The
test section is made up of polycarbonate. The base plate holds the GDL
sample on the top side and has a water inlet on the bottom as shown in
Figure 3. On top of the GDL, the channel plates are mounted to form
the two sides of the channel. The channel plates have the different
open angles (30◦ , 45◦ , 50◦ , 60◦ , 90◦ ) machined onto them. As the
contact angle of the machined surface would change considerably
than the smooth polycarbonate, the machined surfaces were polished
to make them smooth. The length of the channel was 100 mm. The
top plate is mounted over the channel plates to form the top wall of
the channel. Pressure taps were made on the top plate just before the
droplet inlet to measure the instantaneous pressure drop across the
Figure 3. Ex situ Experimental Test Section Assembly Details.
F470
Journal of The Electrochemical Society, 159 (8) F468-F475 (2012)
Table I. Air flow velocities with corresponding current densities
calculated for a channel active area of 18.4 cm2 .
Current Density
(A/cm2 )
Stoichiometry
Air Velocities
(m/s)
0.1
0.1
0.3
0.3
0.5
0.5
0.7
0.9
1.1
1
2
1
2
1
2
1
1
1
0.24
0.48
0.73
1.46
1.21
2.43
1.70
2.18
2.67
channel walls. A syringe pump from the Harvard Apparatus (Model
11 Plus) was used to introduce water through the polycarbonate base
and the GDL surface. A low water flow rate of 0.005 mL/min was used
for the experiment which corresponds to current density of 0.3 A/cm2
for an active area of 2.9 cm2 . Experiments were also performed with a
water inlet flow rate of 0.05 mL/min which corresponded to the water
generation rate for a current density of 3 A/cm2 for an active area of
2.9 cm2 . There was no difference in the observed droplet behavior
between the two water flow rates used. Therefore, to minimize the
experimental time, water flow rate of 0.05 mL/min was used for all the
experiments. During experimentation, the air flow was set to a desired
value and the droplet was allowed to grow on the GDL surface. As
the droplet would continue to grow, it would come in contact with
the channel sidewall similar to the process that would occur in a
PEMFC channel. This process was captured using high speed video
and was post processed using Keyence Motion Analyzer software
at 3X magnification to manually estimate the instantaneous dynamic
contact angles of the droplet made with the sidewall and the GDL base.
The uncertainty of the contact angle measurement is considered to be
within the range of ± 2◦ . Pressure drop data across the droplet was
also measured using a Honeywell FWD differential pressure sensor
(0–1 psi range) connected to the system via pressure taps. The accuracy
of the pressure sensor was ± 0.25% of full scale reading. All these
data were post processed and are discussed in detail in the results
section.
Dynamic contact angle and contact angle hysteresis
measurement.— Prior to determining the desired angular configuration of the sidewall for efficient water removal from the
PEMFC, it is necessary and important to measure the contact angle
hysteresis. In order to quantify the hysteresis, the static advancing
contact angle, θadv . as well as static receding contact angle, θrec .
were measured for both the sidewall material and the base GDL
material. The contact angle hysteresis was calculated by θhys = θadv .
– θrec . The contact angle was determined using the recorded images
and VCA Optima Surface Analysis System. The corresponding
measurements for both GDL and the polycarbonate sidewall are
given in Table II.
In order to analyze the droplet dynamics, the droplet contact angles
with the sidewall and the base GDL need to be studied. Figure 4 shows
the different contact angles the droplet makes with the sidewall and
the base. Droplet pinning on the surface is dependent on the lower
contact line (LCL) and inner contact line (ICL) angles.23 LCL is the
Figure 4. Graphical representation of the different contact lines made by the
droplet with base and sidewall of the channel reproduced from.15
angle made by the droplet on the sidewall near the channel corner
and ICL is the angle made by the droplet on the base near to the
channel corner. These angles determine the droplet filling the corner
of the channel where sidewall and the base meet. Once the droplet
fills the corner, it is very difficult to remove the droplet because this
filled corner acts as a wetted site for other droplets, which grows
on the GDL, and then stick on to the wetted site and form a layer
of liquid water on the GDL surface. This in turn leads to a reduced
diffusion of the gases through the GDL to the catalyst layer and
hence, the efficiency of the system is compromised. It becomes very
important to prevent the droplet filling the corners of the channel. To
avoid this filling situation, the sidewall angle needs to be appropriately
evaluated.
As mentioned earlier, the Concus-Finn condition was used to predict the rise height of the column of the liquid in the wedge shaped
space. The same relation could be used to predict the liquid behavior in the corner of a channel. Using the same graph as shown in
Figure 1b, the droplet filling the corner could be predicted if the contact angle point was inside the shaded region or not. To make the plots
simple to understand in this particular application, instead of drawing
the whole box to indicate the filling region as shown by Concus-Finn
in their work, only the upper line of the box was used. This would
imply that for any contact angle points below the limit line, the droplet
would move toward the corner of the channel and would eventually
fill the corner of the channel. If the point is above the Concus-Finn
limit (CFL) line, then the droplet would pin to the sidewall and would
not move toward filling the corner of the channel. In this manuscript,
to simplify the notation, the contact angles θICL and θLCL will be represented by θB and θW corresponding to base and wall respectively.
Therefore, the limit line for the Concus-Finn condition using the static
contact angles is predicted using the Equation 2
2α = (θ B + θW ) − π
[2]
where θ B is the static contact angle for the base material (GDL) and
θW is the static contact angle for the sidewall (polycarbonate).
Results
Table II. Static Contact Angle Measurement.
Material
Advancing Contact
Angle (θadv )
Receding Contact
Angle (θrec )
Hysteresis
(θhys )
Baseline GDL
Polycarbonate
148
85
138
61
10
24
The high-speed videos at frame rates of 60–250 fps were captured
for the droplet dynamic behavior in the gas channel with different
angular configurations. These videos were examined to characterize
the effect of air flow in the channel on the behavior of the droplet
and determine the effect of location of the droplet emergence on the
droplet dynamics. Frame by frame analysis was performed on the
videos to estimate the instantaneous dynamic contact angle (IDCA)
Journal of The Electrochemical Society, 159 (8) F468-F475 (2012)
F471
Figure 5. Sequence image of the droplet dynamics in 30◦ open angle channel made of polycarbonate sidewall and GDL base with 0.4 ms−1 air velocity introduced
into the channel from the air manifold. The red circle shows that the droplet does not fill the corner of the channel. The top wall is difficult to be visualized in the
image as the length is only about 500 μm.
and the movement of the droplet along the air flow. Pressure drop
data were also recorded for all experiments and processed to evaluate
the effect of the droplet dynamics in the system. One of the major
observations was that the IDCA made by the droplet on the sidewall
and the GDL were outside the range of static advancing and receding
contact angles. The air flow causes oscillations in the droplet interface
introducing significant variations in the associated dynamic contact
angles.
Effect of sidewall angle.— To evaluate the effect of sidewall angle
on the droplet dynamics, experiments were performed with five sidewall angles (30◦ , 45◦ , 50◦ ,60◦ , 90◦ ) for flow rates 0.4 and 1.6 ms−1 .
The image sequence of the video for the droplet behavior for 30◦ is
shown in Figure 5. The video was captured for whole sequence of
droplet dynamics i.e. from droplet growth till the droplet is removed
from the channel. It was observed that the droplet in a channel would
undergo a series of transitions before being removed from the channel.
For the 30◦ channel, the droplet would emerge on the GDL surface
and continue to grow until it touches the top wall before spreading
on the channel sidewall. In this case if the force from the air is much
higher than the surface forces that are holding the droplet on the GDL,
the droplet would start to move on the GDL surface and eventually
get pushed out of the channel. The whole sequence of droplet behavior was common for all channel angles but the only difference
that was observed was the corner filling and non-filling for different
open angles. For 30◦ channel, the droplet did not fill the corners of
the channel. Similar trend was observed for the 45◦ channel, whereas
for 60◦ channel, the droplet filled the corner of the channel. Figure 6
shows the sequence image of the video for a single droplet behavior
in 60◦ channel. Similar results were observed for the 90◦ open angle channel. All these video images are recorded for a longer time
span to visualize successive droplet emergence and departure however only one droplet emergence and departure has been shown in the
sequence of images. To evaluate the experimental observations, the
IDCA values the droplet made with the sidewall was measured from
the videos and plotted against the IDCA made on the GDL. IDCA
values were obtained by measuring the contact angle θW and θB the
droplet made with the wall and the base respectively for each video
frame till it filled the channel corner. This was extended for the entire
range of open angles tested. The IDCA (θW and θB ) data obtained
from the experiments was plotted with the theoretical CFL. In the
plots, even if one of the IDCA point falls below the CFL line then
the droplet would not pin on the sidewall. Instead, it would slowly
move toward the channel corner and eventually fill the corner. The
IDCA plot for 30◦ and 45◦ open angle channel is shown in Figure 7a,
and for the 60◦ and 90◦ open angle channels, IDCA plot is shown in
Figure 7b. In these plots the solid line shows the theoretical CFL for
a particular channel angle. The dotted line for θB and θW shows
the hysteresis observed in the static advancing and receding contact
angle valves for base GDL and sidewall respectively. The filled boxes
in the plot show the filling contact angle i.e. the contact angle values
that would attract the droplet toward the corner of the channel and
the open boxes show the non-filling IDCA values which would not
lead to corner filling. Figure 7 clearly shows that the IDCA points for
Figure 6. Sequence image of the droplet dynamics in 60◦ open angle channel made of polycarbonate sidewall and GDL base with 0.4 ms−1 air velocity introduced
into the channel from the air manifold. The red circle shows that the corner of the channel is completely filled.
F472
Journal of The Electrochemical Society, 159 (8) F468-F475 (2012)
the filling and non-filling criteria according to the Concus-Finn condition determined using static contact angle values, the results showed
different behavior for different air flow rates.24 These results are discussed in more detail in the next section.
Effect of air flow.— To evaluate the effect of air flow on the droplet
behavior, experiments were conducted in 90◦ open angle channel by
varying the air flow in the channel from 0.2–2 ms−1 . It was observed
that the droplet filled the corner of the channel for all air flow rates.
Similar experiments were conducted for other open angles as well.
It was observed that for all air velocities, the 60◦ channel showed
corner filling whereas for the 30◦ and 45◦ open angle channels the
corners were not filled. However, there were some distinct differences
observed in the droplet dynamics for lower air flow rates compared to
the higher flow rates. These differences are listed below:
1.
2.
3.
Figure 7. Instantaneous dynamic contact angles made by the droplet with the
sidewall and the base GDL plotted against each other along with the theoretical
Concus-Finn limit line shown by solid line. The solid boxes in the plot shows
the angles that lead to droplet fill the corner of the channel and the open box
shows the angle showing pinning of the droplet. For channel angle 30◦ and
45◦ the corner of the channel is not filled and for channel angles 60◦ and 90◦
the corner of the channel is filled.
the 30◦ and 45◦ open angle channel, obtained from the experimental results, fell above the CFL line which indicates that the droplet
would not fill the corner, as was observed in the videos. For the 60◦
and 90◦ open angle channel it was observed that the contact angle
points fell below the CFL line showing that the corner should be
filled. These results were similar for all air flow rates. These results
were consistent with the findings shown by Rath and Kandlikar.15,24
However for the 50◦ channel, which is near the transition point for
The droplet blocks the channel completely in the case of lower
flow rates before the air pushes the droplet out of the channel,
whereas for higher air flow rates the droplet is pushed out of the
channel well before it blocks the channel completely.
At lower air flow rates, the droplet would first translate in the
direction of the sidewall which eventually leads to filling or nonfilling condition according to the channel angle and then would
move out of the channel, whereas at higher flow rates it was observed that the droplet would first translate in the airflow direction
and then move on to the sidewall later.
Time required for the droplet to be removed from the channel
from the time it makes its appearance on the top of the GDL was
much higher for lower flow rates (15–20 sec) whereas, using high
air flow rates the droplet removal time was almost cut down by
half (5–10 sec).
Similar behaviors were observed for all the channel angles. However, there was another significant effect that was observed only on
the 50◦ open angle channel. It was observed that the droplet filling
and non-filling in the corner of the channel was dependent on the
air velocity. For lower flow rates of 0.2–0.4 ms−1 , the droplet filled
the channel corner as shown in Figure 8 whereas for the higher flow
rates i.e. 0.5 ms−1 onwards the droplet did not fill the channel corner
as shown in Figure 9. To evaluate the actual cause for this behavior,
IDCA obtained from the experiments before filling of the droplet in
the channel corners were plotted along with the theoretical CFL in
Figure 10. It was examined that for lower air velocities the IDCA
made with the sidewall and the base GDL measured from the video
before the droplet filled the channel corners were found to lie below
the CFL line implying that the droplet should fill the corner of the
channel. While for higher air flow velocities, the contact angles were
above the CFL line. These results were similar to what was observed
from the experimental videos. From these observations, it could be
inferred that the air velocities in the channel manipulates the droplet
and oscillates it in the channel before it contacts the sidewall. These
oscillations drive the corner filling and non- filling in the channel
for different air velocities at the transition angle. According to the
static advancing contact angle values for the GDL and polycarbonate
sidewall, the transition angle from non-filling to filling region is 53◦ .
Figure 8. Sequence image of the droplet interaction with the polycarbonate sidewall and
GDL base in 50◦ open angle channel with
0.4 ms−1 air velocity introduced into the channel.
Red circle in the image shows the corner filling of
the channel by the droplet.
Journal of The Electrochemical Society, 159 (8) F468-F475 (2012)
F473
Figure 9. Sequence image of the droplet interaction with the polycarbonate sidewall and GDL base in 50◦ open angle channel with 1.6 ms−1 air velocity introduced
into the channel. Red circle in the image shows that the corner of the channel is not filled by the droplet.
Experimentally, 50◦ open angle channel shows the transition point
from non-filling to filling.
To further characterize the effect of air flow on the droplet in
different channel angular configurations, the length of the sidewall
from the corner of the channel which was left non-wetted by the
droplet (HLCL ) and the length of the sidewall that was wetted by the
droplet (HUCL ) for a given instant of time were measured and plotted.
This was done to measure the height clearance that is left when the
droplet gets pinned on to the surface which would clearly show that
the droplet did not fill the corners of the channel and also this would
be helpful in getting an insight on the time it requires for the droplet to
fill the corner of the channel in a particular angular configuration. The
length measurement was done by manually going through each frame
in the video and measuring the wetted and non-wetted length values
for a particular time instant. The length of the non-wetted sidewall
was plotted against the time. It was seen from the plot that the droplet
pinning on the wall was prominent for the 30◦ and 45◦ angle channels
as shown in Figure 11. Here the non-wetted length remained constant
after certain time. This indicates the droplet pinning phenomenon on
the surface which would not allow the droplet to move toward corner
of the channel to fill it. Once the droplet is pinned to the sidewall
from the lower contact line, the upper contact line starts to move more
rapidly to fill the channel and increase the pressure upstream for it to
be pushed out of the channel. For 60◦ and 90◦ channels, the non-wetted
length of the wall near lower contact line HLCL became zero after 35
ms showing that the corner is filled. This sequence was similar for all
air flow rates studied. Droplet filling the channel corner took place
rapidly in these channels. When the same experiment was performed
for the 50◦ channel it was observed that for the lower air flow rates
the non-wetted length of wall approached to zero at 27 ms showing
that the droplet filled the corner of the channel whereas for the higher
air flow rates the non-wetted length of the wall never reached zero.
Figure 10. Instantaneous dynamic contact angles made by the droplet in a
50◦ open angle channel made up of polycarbonate sidewall and the base GDL
plotted against each other along with the theoretical Concus-Finn limit line
shown by solid line at different air velocities.
This corresponds to non- filling criteria where the droplet is pinned to
the sidewall and does not move forward on the wall surface to fill the
corner. This also verifies the different behavior seen in the gas channel
as an effect of airflow in the channel.
Effect of droplet inlet location.— In the PEMFC, the droplet introduction on the top of the GDL surface is not bounded to the central
location in the channel. It could appear anywhere on the GDL surface
such as near the channel wall surface or at the channel center or under
the land area. All the above experimental and the IDCA plot results
were recorded when the droplet was introduced at the center of the
channel. To verify the same droplet behavior when it appeared near
to the sidewall, preferential pore for the droplet inlet was introduced
at 1 mm away from the sidewall and base interface. It was observed
that for 45◦ channel the droplet filled the corner of the channel for
all air flow velocities. According to the CFL condition, the droplet
should not fill the corner of the channel made by polycarbonate and
the GDL for 45◦ channel when the advancing contact angle values
for the sidewall and base materials are used to determine the corner
filling criteria. Using IDCA values from the experiments against the
CFL condition revealed that the droplet should indeed fill the corners
of the channel as shown in Figure 12 and this behavior was confirmed
in the experiments. This implies that the dynamic contact angle of the
droplet with the sidewall was altered dynamically by the airflow in the
system. This accordingly changes the criteria of the droplet filling and
non-filling within the system. Hence the CFL condition based on advancing contact angles for predicting the droplet filling or non-filling
in the channel is not valid. Using IDCA values in the CFL condition
will give the actual filling or non-filling criteria for the system.
Experiments were also performed to visualize the droplet behavior
when the droplet was introduced under the land region. In this case, the
droplet was introduced 0.5 mm away from the channel sidewall and
base intersection into the land region. It was observed that irrespective
of the channel angle, the droplet would be introduced into the channel
by filling of the corner. When the droplet is introduced under the land,
it spreads along the land area and when it comes into the channel area
it wets the channel sidewall as well as the GDL surface. To avoid
such scenario different design for gas channel needs to be developed
to avoid corner filling of the channel. Moreover, in this case, the
Concus-Finn condition cannot be used to evaluate the results since
the Concus-Finn condition is valid only when the droplet emerges
near the opening of the channel corner. It cannot be used when the
droplet approaches the channel corner from the other end.
Pressure drop across the droplet.— The corner filling of the gas
flow channel lead to an undesired water holdup within the channel
which may eventually degrade the performance of the PEMFC. Pressure drop across the droplet is one of the reliable ways to measure
the gas transport resistance due to the water hold up in the channel.
When the water builds up in the channel, pressure drop increases accordingly. Lower pressure drop corresponds to less amount of water
hold up in the channel and thereby lower gas transport resistance. In
order to study the pressure drop across the droplet for various channels, the pressure tap was made behind the droplet inlet which would
read the pressure change across the droplet as the droplet would grow.
The peak pressure drop value across the droplet for different air flow
F474
Journal of The Electrochemical Society, 159 (8) F468-F475 (2012)
Figure 12. Plot of instantaneous dynamic contact angle made by the droplet
with the polycarbonate sidewall and the GDL base for 45◦ open angle channel
when the droplet emergence is near the channel corner. Here the contact angle
points fall below the Concus-Finn limit line and shows corner filling of the
channel.
velocities in a channel is plotted in Figure 13. The uncertainty in the
peak pressure drop value was about 0.25%. The solid line in the plot,
which is plotted by visual observation of the data, shows the pressure
drop limit for attaining the droplet filling and non-filling criteria. For
peak pressure drop values falling below the line would give non-filing
of the channel whereas if the peak values fall above the limit line the
droplet would fill the corner of the channel. Figure 13 clearly shows
that in 90◦ channel the pressure drop across the droplet increases as the
air flow rate increases. However, at lower air flow rates the pressure
drop is high. This is observed when the droplet grows and completely
blocks the channel, the pressure in the channel starts to build up. At a
given instant when the pressure in the system increases tremendously,
the droplet is removed from the channel. The time taken by the lower
air flow to remove the droplet from the channel is comparatively large
which leads to high pressure built up in the channel. However at higher
flow rates, the droplet experiences much higher force due to the air
flow which pushes the droplet out of the channel sooner as compared
to the lower velocities. This trend was similar for all channel angles
showing corner filling of the channel. Also at higher air flow rates
Figure 11. Plot of lower contact line and upper contact line distance (height)
from the corner of the channel for different channel angle as a function of
air flow velocities (a) 45◦ channel where the droplet is pinned to the channel
sidewall and does not fill the corner of the channel for any air flow rate
(b) 60◦ channel where the droplet does not pin to the channel sidewall and fill
the corner of the channel for all air flow rates (c) 50◦ channel where for lower
air flow rates the droplet fills the corner of the channel shown by square box
and for higher flow rates the droplet remains pinned to the sidewall and does
not fill the channel corner shown by filled circle.
Figure 13. Plot of peak differential pressure drop values across the droplet for
different channel angles as a function of air flow velocity. The solid line shows
the P limit for filling and non-filling of the corner of the channel through
experimental data.
Journal of The Electrochemical Society, 159 (8) F468-F475 (2012)
the droplet behaves as a film in the channel which helps in keeping
the pressure drop much low. In Figure 13 it is also observed that the
pressure drop value for the 60◦ channel is higher than that of the
90◦ channel. The channel area for the 60◦ channel is larger compared
to the 90◦ channel. Hence when the droplet fills the channel corner,
more amount of liquid water is present in the same cross section as
compared to the 90◦ channel and the amount of pressure required to
remove the droplet becomes large.
For the channels 45◦ and 30◦ , the droplet does not fill the corner
of the channel for any air flow rates. For these channels, the observed
pressure drop value is significantly lower compared to the 60◦ or
90◦ channels where the corner of the channel is filled. Similarly for
50◦ channel, at higher flow rates where the droplet does not fill the
corners the pressure drop values are much lower compared to the
lower air flow rates where the droplet fills the corner and hence larger
pressure drop values in the channel. In 50◦ channel, at higher flow rates
(1.6–2 ms−1 ) the droplet was removed from the channel before it
touched the channel wall. This behavior was reflected in the pressure
drop plot by a drop in the peak pressure drop value. It was also
observed that the pressure drop values varies directly proportional
to the channel angle for the non-filling channel angles i.e., the 30◦
channel has the lowest pressure drop for a given flow rate and the 50◦
channel has the highest pressure drop among the three. Main reason
for such lower pressure drop values for the smaller channel angle
is mainly due to the gap near the channel corner. For example, in
30◦ channel, the droplet squeezes in the channel due to the channel
configuration before it is removed in such a way that there is larger
gap near the sidewall corner for the air to squeeze through and escape
to the atmosphere and hence the lower pressure drop in the channel
is observed. However, in the 50◦ channel, the space for the air to
squeeze through the channel is less compared to the 30◦ channel
which increases the velocity of the airflow around the droplet. This
leads to building up of pressure inside the channel before the droplet
is pushed out of the channel.
It was observed through the experiments that the droplet dynamics
in the corner of the gas channel is dependent mainly on the channel
angle, material used in the PEMFC and on the gas flow velocities in
the channel. This work provides an insight on the importance of the
design of the gas channel configuration on the droplet dynamics. The
experiments performed in this manuscript were in a larger dimension
channels. Keeping the air velocities similar to the real world fuel cell,
the dynamic similarities were maintained. To validate this behavior for
a real fuel cell scenario, similar experiments are underway on small
sized channels. Also, different channel materials would have different effects on the droplet dynamics. Polycarbonate was used in these
experiments as it allows visual access through it and also has similar
static contact angle values as graphite and stainless steel which are
used in real fuel cell applications. Further work is in progress to evaluate the material effects on the droplet behavior. Also, compression of
the GDL would lead to larger intrusion in the channel which would affect the droplet dynamics. In this work, compression effects were not
studied. However, to avoid compression effect, minimal compression
was applied on the system.
Conclusions
In this work, ex situ experiments were conducted to study the water
droplet dynamics in a gas channel of 3 mm hydraulic diameter for
different channel sidewall angles under controlled air flow velocities
of 0.2–2 ms−1 . The pressure drop across the droplet was recorded
simultaneously. It was revealed through the high speed videos that the
droplet dynamics at the channel corner are significantly affected by
the changes in the channel angle, airflow velocities and the droplet
inlet location. All these data confirmed the validity of Concus-Finn
condition if the instantaneous dynamic contact angle values were used
instead of static contact angles. The channel sidewall angle plays a
F475
major role in determining the droplet removal from the GDL and
associated flow pattern and pressure drop characteristics.
r Effect of Channel Angle – For GDL base and polycarbonate
sidewall, an open angle of 50◦ is the transition point between a droplet
filling (for higher angles) and non-filling (for lower angles) of the
corner. The surface energies of the channel wall and the base material
have significant impact on the corner filling characteristics of the
channel at a given channel open angle.
r Effect of Air Flow Velocities – Air flow introduces oscillations
in the droplet before it touches the sidewall which significantly affects
the instantaneous dynamic contact angle the droplet makes with the
sidewall. This instantaneous dynamic contact angle determines the
corner filling or non-filling criteria.
r Effect of Droplet Location – Depending on the droplet emergence location the droplet filling and non-filling the channel corner
for a particular channel configuration changes. Using instantaneous
dynamic contact angle the corner filling criteria could be predicted
accurately. Also, Concus-Finn condition cannot be used to evaluate
the droplet behavior when the droplet emerges under the land region.
Acknowledgment
This work was conducted 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-EE0000470.
References
1. M. S. Dresselhaus and I. L. Thomas, Nature, 414, 332 (2001).
2. P. Hoffmann, Tomorrow’s Energy: Hydrogen, Fuel Cells and the Prospects for a
Cleaner Planet, The MIT Press, Cambridge, MA (2002).
3. J. T. Gostick, M. W. Fowler, M. A. Ioannidis, M. D. Pritzker, Y. M. Volfkovich, and
A. Sakars, Journal of Power Sources, 156, 375 (2006).
4. F. Y. Zhang, X. G. Yang, and C. Y. Wang, Journal of The Electrochemical Society,
153, A225 (2006).
5. J. J. Kowal, A. Turhan, K. Heller, J. Brenizer, and M. M. Mench, Journal of Electrochemical Society, 153, A1971 (2006).
6. B. Markicevic, A. Bazylak, and N. Djilali, Journal of Power Sources, 171, 706
(2007).
7. J. P. Owejan, T. A. Trabold, D. L. Jacobson, M. Arif, and S. G. Kandlikar, International Journal of Hydrogen Energy, 32, 4489 (2007).
8. J. P. Owejan, J. J. Gagliardo, J. M. Sergi, S. G. Kandlikar, and T. A. Trabold, International Journal of Hydrogen Energy, 34, 3436 (2009).
9. T. Metz, N. Paust, C. Müller, R. Zengerle, and P. Koltay, Sensors and Actuators A:
Physical, 143, 49 (2008).
10. Z. Lu, S. G. Kandlikar, C. Rath, M. Grimm, W. Domigan, A. D. White,
M. Hardbarger, J. P. Owejan, and T. A. Trabold, International Journal of Hydrogen Energy, 34, 3445 (2009).
11. Z. Lu, M. M. Daino, C. Rath, and S. G. Kandlikar, International Journal of Hydrogen
Energy, 35, 4222 (2010).
12. A. Bazylak, D. Sinton, and N. Djilali, Journal of Power Sources, 176, 240
(2008).
13. A. Theodorakakos, T. Ous, M. Gavaises, J. M. Nouri, N. Nikolopoulos, and
H. Yanagihara, Journal of Colloid and Interface Science, 300, 673 (2006).
14. T. Ous and C. Arcoumanis, International Journal of Hydrogen Energy, 34, 3476
(2009).
15. C. D. Rath and S. G. Kandlikar, Colloids and Surfaces A: Physicochemical and
Engineering Aspects, 384, 653 (2011).
16. V. Gurau, H. Liu, and S. Kakaç, AIChE Journal, 44, 2410 (1998).
17. X. Zhu, P. C. Sui, and N. Djilali, Journal of Power Sources, 181, 101 (2008).
18. D. Spernjak, S. Advani, and A. K. Prasad, Experimental Investigation of Liquid Water
Formation and Transport in a Transparent Single-Serpentine PEM Fuel Cell, in ASME
Conference Proceedings, p. 453, ASME (2006).
19. F. Y. Zhang, X. G. Yang, and C. Y. Wang, J. of Electrochem. Soc., 153(2), A225
(2006).
20. N. Akhtar, A. Qureshi, J. Scholta, C. Hartnig, M. Messerschmidt, and W. Lehnert,
Inter. J. of Hydro. Energy, 34(7), 3104 (2009).
21. C.-C. Hwang, J.-R. Ho, and S.-C. Lee, Phy. Rev. E, 60(5), 5693 (1999).
22. P. Concus and R. Finn, Microgravity Sci. & Tech., 7, 152 (1994).
23. Y. V. Kalinin, V. Berejnov, and R. E. Thorne, Langmuir, 25, 5391 (2009).
24. P. Gopalan and S. G. Kandlikar, ECS Transactions, 41, 479 (2011).