|
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30-June 4, 2010
Visualization of Water Distribution in Gas Diffusion Layer of Fuel cell using 2-D and 3-D
X-ray radiography
Jongrok Kiml, Junho Je2, Massoud Kaviany3, Sang Young Son4
and Moo Hwan Kims
SDepartment of Mechanical Engineering, POSTECH, Pohang, Korea
2 Department of Mechanical Engineering, POSTECH, Pohang, Korea
SDepartment of Mechanical Engineering, University of Michigan, Ann Arbor
4 Department of Mechanical Engineering, University of Cincinnati, Cincinnati
5 Department of Mechanical Engineering, POSTECH, Pohang, Korea
jongrok a postech.ac.kr, sofujuno a postech.ac.kr, kaviany aumich.edu, sangyoung.son(Auc.edu, and
mhkim~postech.ac.kr
Keywords: Visualization, X-ray radiography, Gas Diffusion Layer, Water distribution, Fuel cell
Abstract
Low temperature fuel cell, e.g. PEMFC, is operated under boiling temperature of water. Therefore, liquid water exits in fuel
cell and effects deeply to the performance of fuel cell. So, understanding of water behavior in fuel cell is one of important
issues for high efficiency fuel cell. Especially, visualization techniques of water distribution in fuel cell are very important for
understanding and validation of models to predict phenomena in fuel cell. The objective of this study is the visualization of
water distribution in gas diffusion layer (GDL) as liquid water is injected through a face of GDL. For this aim, 2-D and 3-D
X-ray radiography were employed with the X-ray microscopy (7B2) in Pohang Accelerator Laboratory and the experimental
facilities to inject water in GDL were developed. 3-D X-ray radiography (tomography) shows more accurate result than 2-D
X-ray radiography. Although X-ray 2-D radiography has error of quantification of water, this is very helpful to analyze the
behavior of water because this can measure the water distribution on unsteady state. Water distributions were convexity shape.
Injected water penetrated deeply to GDL as pressure of injected water was increased and water penetrated shape was like as an
inverted tree-like transport structure.
Introduction
A fuel cell is an electrochemical device that converts the
bond energy of reactants directly to electricity (and heat).
Polymer electrolyte membrane fuel cells (PEMFCs) employ
a solid-state proton-conducting polymer membrane as the
electrolyte. PEMFCs have an enormous power range from a
few watts to several hundred kilowatts, and a large number
of applications. Especially, PEMFCs are adoptable on
vehicle because they can be compact design and has low
operating temperature. However, there are many problems
for commercializing of PEMFC vehicles. One of these
problems is liquid water distribution in fuel cell.
Water can be either brought into the fuel cell by external
humidification or generated by electrochemical reaction.
This water is very important for PEMFC operating. Without
sufficient water in the polymer electrolyte and catalyst layer,
proton transport and reactivity are poor, causing
performance to suffer. Excess liquid water in the catalyst
and GDL, though, also impedes reactant transport. Unable
to continue operating with water clogging "breathing
channel" channels, the fuel cell drowns (M. Eikerling 2006)
Especially, visualization techniques of water distribution in
fuel cell are very important for understanding and validation
of models to predict phenomena in fuel cell. The objective
of this study is the visualization of water distribution in gas
diffusion layer (GDL) as liquid water is injected through a
face of GDL. For this aim, 2-D and 3-D X-ray radiography
were employed with the X-ray microscopy (7B2) in Pohang
Accelerator Laboratory and the experimental facilities to
inject water in GDL were developed.
Nomenclature
Thickness in beam direction (m)
Ratio of beam attenuation
Greek letters
Radiation beam intensity (n cm2 s ')
6 thickness in beam direction (m)
Sattenuation coefficient [nun ]
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30-June 4, 2010
explicitly given for the bravery of notation and consistency
with the actual computational procedures. With the radiation
beam flux at reference time,
[ J
this allows to define the relative radiation beam
transmission,
T=f =t exp p,~, AS (5)
Hence, the division of two images obtained under different
conditions allows extracting the difference between both
images and the relative radiation beam transmission is
directly related to the thickness difference of the attenuating
materials.
Experimental Facility
The main technical characteristic of 7B2 X-rav microscopy
beam line is the absence of the monochromator and hard
X-ray optical components except windows: that gives high
flux (The range of photon energy is 2.8~5.5 keV.) and rather
large beam size (8.7 x0.7 cm full width at half maximum)
suitable for large field of view imaging. In what follows, the
7B2 beam line will be described. Synchrotron X-ray is
generated due to rotation of high energy electron in storage
ring and it move from front end to hutch through spool with
10-10Torr. The source (bending magnet port of the PLS
storage ring) has 60 pm (vertical) and 160 pm (horizontal)
size, and is equipped with a radiation shutter. Right after the
front end and the corresponding 0.25mm thick polished
beryllium window, a water-cooled slit system makes it
possible to define the effective source geometry. The system
consists of two perpendicular slits with width adjustable
from zero to 5cm. The total length of the beam line, 34.8m,
was selected to provide a relatively large beam on the object.
And, the transverse coherence length increases with the
source-object distance and therefore improves the
performances or the beam line. The 34.8m length is the
maximum allowed by the PLS building and is perfectly
suitable for the X-ray radiography experiments. The next
beam line component is the second slit system, again with
two perpendicular slits. Furthermore, there is the radiation
hutch with the third manual slit set (the other two sets are
steppmng-motor-driven). X-ray beam which passed spool has
the distribution of energy and flux along lateral plane so we
can select the coherence and the flux of X-ray beam by
controlling the size of slit in hutch. If slit is opened largely
then we can take high flux X-ray beam but the coherence of
beam is bad and if slit is minified then beam has good
coherence and low flux. In between these second and third
slit systems, there is the rear-end 0.25mm thick, polished,
beryllium window. This is an essential feature for this type
of beam line, to avoid scattering background and other
possible spurious features magnified by the large distance
with respect to the detector. The exit of the beam pipe is
covered by a Kapton window (polvimide). (The space
between last beryllium window and Kapton foil (polvimide)
being evacuated to low vacuum (10 5mbar) ). To block and
control the intensity of X-ray beam, we used mechanical
shutter after slit. Mechanical shutter consists of Pt-Ir blade
Subsripts
ref Reference value
Theory of Radiography
All radiographic methods, whether making use of X-rays-
gamma-rays or neutrons, are based on the same general
principle; that radiation is attenuated on passing through
matter. The object under examination is placed in the
incident radiation beam. After passing through, the beam
that remains enters a detector that registers the fraction of
the initial radiation intensity that has been transmitted by
each point in the object. Any homogeneity in the object or
an internal defect (such as e.g. void, crack, porosity or
inclusion) will show up as a change in radiation intensity
reaching the detector. Thus detection of defects in
radiography image is based on the observation of
differences in radiation intensity after passing through the
object under examination. This occurs according to the basic
law of radiation attenuation:
O = Goe (1)
as is graphically presented in Figure 1.
Figure 1: Attenuation of radiation by matter
The investigation of highly structured objects such as fuel
cells implies the need to separate the contributions of
different materials or layers to the formed radiation beam
flux behind the object. This is achieved by referencing: the
scaling of the recorded radiation beam flux by a convenient
reference.
The simplest approach is to separate by temporal
dependence. Under the constraints of a negligible secondary
radiation beam flux, the time-dependent radiation beam flux
behind the measurement object (at the detection plane) is
described bv:
O(x, y, t) = Go(x, y, t) exp ,6, (x, yt) (2)
Herein, the index i sums over all materials and layers with 3;
being the thickness in beam direction. The radiation beam
flux provided by the source is denoted by 4b. If the
temporal dependence is expressed as the difference from a
given reference state at time t,4 according to 6(t) = 6 .g +A 6,
above equation becomes
O = G ep pI (6, +fi S)I (3)
1 I
Herein, the dependence on the spatial variables is not
a 1~1
11111
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30-June 4, 2010
blocks those had a 1mm*1mm*10mm channel and 3 port -
fluid inlet, outlet, and pressure port. A block had a gap
where GDL was laid down. GDL compressed rate is
controlled by depth of this gap (Figure 2).
capable of blocking X-ray energy up to 30 keV and housing.
For a protection of the sample, unnecessary shining of
X-ray needs to be shut off. Shutter switches open and close
by electrical signal from drive unit only for a taking an
image.
The optimized X-ray beam passed through slit and
attenuator is illuminated to sample. The position of X-ray
beam is fixed so that we can take image of any position
what we want by controlling sample. So we use steerable
sample stages for three-dimensional movements. This motor
control unit includes rotation stage, tilting stage and
translational stage which have high resolution of 0.00090,
0.0020 and 250nm respectively. This is very important
feature for the coherent X-ray imaging since the detection of
the edge-enhancing fringes depends of the detector
resolution and on the object-detector distance. Small size
samples (and samples which do not require specific sample
environment) are mounted on the nano-stage which is then
directly mounted on the rotation stage. Nano-stage moves
allowing small linear motion. (This is used for alignment
prior to tomography scan, where, during rotation, sample
has to remain side the field of view for all angular range
00 to 1800). X-ray has information of the sample after
passing it. This X-ray with information is converted to
visible light and magnified by optical lens. The role of
scintillation crystal scintillatorr) is to shift X-ray to visible
light. Its material is both faces polished CdWO4 single
crystal (30x30x0.3 mm', Nihon Kessho Koogaku CO., LTD.)
and it generates visible light within the range of 470~540nm
wavelength. The information of visible light is reflected
perpendicularly due to silicon wafer and is magnified by
optical lens of 3 ~ 20 magnification. In these optical lenses,
using revolver, technologies to image samples increase
levels of resolution, as hierarchical imaging. The high
penetration properties of X-rays and this novel approach
allow nondestructive three-dimensional visualization and
quantification of large samples, while retaining a precise
anatomical context for regions of interests scanned at very
high resolution. After magnification, image is monitored
and is saved using software. The detection system includes
scintillator, silicon wafer mirror and camera. (Soeun Chang
2006)
Figure 2: GDL housing for 2-D radiography.
TPGH-090 was used for experiments. Test sections were
consisted with GDL housing, water supply system, and
pressure sensors. There were two type GDL housings for
2-D radiography and tomography.
2-D radiography housing was made of Teflon. This had two
~I
.
Figure 3: GDL housing for tomography. a) is two blocks, b)
is cross section of housing
Tomography housing was made of polyimide. This had also
two blocks. A block had two ports, those are liquid inlet and
pressure port, and 1.4mm diameter hole that supply water to
GDL. Another block had a 1.4mm hole that was gas outlet.
(Figure 3).
Results and Discussion
Figure 4 is an original image of X-ray 2-D radiography. This
shows the liquid channel (upper square) and gas channel
(bottom square), and GDL (middle line). And this also
shows un-deformed GDL on channel. In this picture, liquid
channel is darker than gas channel because liquid channel
was filled with water.
Figure 4: Original 2-D radiography image
Distribution of water thickness was evaluated with
quantification process. Figure 5 is processed result from
Figure 4. Water in liquid channel and top region of GDL is
visible.
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30-June 4, 2010
reconstructed with OCTOPUS 8.3 that is reconstruction
program. Slices were created after reconstruction as Figure
8. Figure 8 and Figure 9 are reconstructed slices for A-A'
and B-B' cross section on Figure 7. In Figure 9, brightest
lines are carbon fibers of GDL, mid-gray areas are water,
and dark areas are air. These slices were rendered with
AMIRA (rendering program for reconstructed slices) and
the results, which are side cross section, are on Figure 10.
Water is not in Figure 8 because water didn't permeate to
cross section A-A'. From these slices, averaged saturation
on each slice was evaluated.
(wa'Cter area)
averaged saturation = (6)
1- (fiber area)
Averaged saturation for a case was plotted on Figure 11
High
Figure 5: Processed 2-D radiography image. This includes
the water thickness distribution.
However, it is not enough to analyze the water distribution
in GDL. So, water thickness values were read and plotted on
graph. Distribution of water thickness in GDL is plotted on
Figure 4.22 for several liquid pressures 0.46KPa, 0.85KPa,
1.27KPa, 1.58KPa. In this case, flooding was happened
when liquid pressure was 1.58KPa. Left of graph is top of
GDL on Figure 5 and right of graph is bottom of GDL on
Figure 5. As shown in Figure 6, water distribution had
convexity shape and water permeated deeply as liquid
pressure was increased.
Figure 7: A image for tomography
Figure 7 is a projection image for tomography. 300 images
were recorded for each 0.6 degree. And these images were
15r00
0
B 0
-- TGPOPH9 I50am 0 46KPa
TGPH90_150hum 0_85KPa
--- TGPYH90_150um L_27KPa
-- TGPH90 lIOum I $8KPa
I-
exgure rr: H reconstruclea suce Ior A-A cross secuon1 on
Figure 8
100 150 200 \rZ, ,5 300
Depth of DL (pm)
Figure 6: [2-D radiography result] Distribution of water
thickness in GDL for various liquid pressures 0.46kPa,
0.85kPa, 1.27kPa, 1.58kPa (flooding pressure)
Figure 9: A reconstructed slice for B-B` cross section on
Figure 8
Figure 10: Side cross section of tomography results for
various liquid pressure Dry, 0.66kPa, 1.93kPa, 2.97kPa,
4.13kPa(flooding pressure)
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30-June 4, 2010
Conclusions
The objective of this study was the visualization of water
distribution in gas diffusion layer (GDL) as liquid water was
injected through a face of GDL. For this aim, 2-D and 3-D
X-ray radiography were employed with the X-ray
microscopy (7B2) in Pohang Accelerator Laboratory and the
experimental facilities to inject water in GDL were
developed. 3-D X-ray radiography (tomography) shows
more accurate result than 2-D X-ray radiography. Although
X-ray 2-D radiography has error of quantification of water,
this is very helpful to analyze the behavior of water because
this can measure the water distribution on unsteady state.
Water distributions were convexity shape. Injected water
penetrated deeply to GDL as pressure of injected water was
increased and water penetrated shape was like as an inverted
tree-like transport structure.
Acknowledgements
This study was supported by the National Research
Foundation of Korea (NRF) through a grant provided by the
Korean Ministry of Science & Technology (MOST) in
M60602000005 -06EO200-00410.
References
M. Eikerling, A.A. Komnyshev, A.P. Kucemnak, "Water in
polymer electrolyte fuel cell: friend or foe?", Physics Today,
38-44 (2006)
Soeun Chang, Study on Tissular Morphology of Distal
Airway Using Synchrotron X-ray Microtomography, Master
Thesis 2006, Department of Materials Science &
Engineering, Pohang University of Science & Technology
Jin Hyun Nam and Massoud Kaviany, Effective diffusivity
and water-saturation distribution in single- and two-layer
PEMFC diffusion medium, Intemnational Joumnal of Heat
and Mass Transfer, Vol. 46, 4595-4611( 2003)
o. '--a '
a son io 2* 2o
--TOPHO90O 0posesit
TOPHO90P 00.66KPs
- *TOPHO90 011.93Da
- -TOPHO90_014.13K.
*fGRL mrDL)
Figure 11: [Tomography result] Distribution of water
saturation in GDL for various liquid pressures 0.66kPa,
1.93kPa, 2.97kPa, 4.13kPa(flooding pressure)
As you can see, water distributions from both methods
showed convexity shape although detail distributions were
different. For tomography result, water penetrated shape
was like as an inverted tree-like transport structure. This
shape agrees with Jin Hyun Nam and Massoud Kaviany
(2003).
2-D X-ray radiography showed fluctuation of water
distribution. This comes from x-ray reflection. X-ray is
reflected on interface of phase. In these experiments,
interface was very complex because GDL has porous
structure. Additionally, interface geometries for each liquid
pressure case were different from reference (dried) GDL
although each x-ray image should refer x-ray image of dried.
However, tomography had low fluctuation because
tomography doesn't refer dried image although x-ray
diffraction also exists. Therefore, tomography shows more
accurate result than 2-D X-ray radiography. However, 2-D
X-ray radiography is very helpful to analyze the behavior of
water although it has high error. 2-D X-ray radiography
requires few seconds for a result while tomography requires
about 10minutes. Therefore, 2-D X-ray radiography is
possible to visualize the transient behavior of water. (In this
study, transient results are not represented)
|
RSS
