Removal of Flooding in a PEM Fuel Cell at Cathode by Flexural Wave
Article information
Abstract
Energy is an essential driving force for modern society. In particular, electricity has become the standard source of power for almost every aspect of life. Electric power runs lights, televisions, cell phones, laptops, etc. However, it has become apparent that the current methods of producing this most valuable commodity combustion of fossil fuels are of limited supply and has become detrimental for the Earth’s environment. It is also self-evident, given the fact that these resources are non-renewable, that these sources of energy will eventually run out. One of the most promising alternatives to the burning of fossil fuel in the production of electric power is the proton exchange membrane (PEM) fuel cell. The PEM fuel cell is environmentally friendly and achieves much higher efficiencies than a combustion engine. Water management is an important issue of PEM fuel cell operation. Water is the product of the electrochemical reactions inside fuel cell. If liquid water accumulation becomes excessive in a fuel cell, water columns will clog the gas flow channel. This condition is referred to as flooding. A number of researchers have examined the water removal methods in order to improve the performance. In this paper, a new water removal method that investigates the use of vibro-acoustic methods is presented. Piezo-actuators are devices to generate the flexural wave and are attached at end of a cathode bipolar plate. The “flexural wave” is used to impart energy to resting droplets and thus cause movement of the droplets in the direction of the traveling wave.
1. Introduction
Most energy comes from fossil fuels. However, reserves of fossil fuels will soon reach their limits. Furthermore, the problems of environmental pollution and climate change due to fossil fuel usage, have created the need to develop new energy sources. Thus, study on developing various renewable energy sources is being conducted. The South Korean government is also actively supporting the development of various renewable energy resources. Energy alternatives – including wind energy, solar cells, and fuel cells – are under development. In fuel cell research, researchers are addressing flow field patterns to supply fuel and oxygen more effectively [1]. Since the internal structure in a fuel cell cannot be directly observed, measurement is difficult [2].
Therefore, research combining computational analysis with experimentation is active. Looking at some of the leading research, Li and Sabir [3] introduced the separation plate design and latest flow path design made by various research institutions. Additionally, problems were identified, and several suggestions were made regarding the optimization of each channel pattern. Perng and Wu [4] compared fuel cell performance through computational analysis of flow field characteristics in the catalyst layer surface and inside channel, using different forms in the GDL of PEM fuel cells. In addition, they varied the number of forms, to compare the flow characteristics and fuel cell performance [5]. Karthikeyan et al. [6] developed a serpentine flow channel with uniform and zigzag positioned porous carbon inserts on cathode flow plate. Experimental results show that the flow channel with zigzag positioned porous inserts has improved the power density and current density when compared to the conventional serpentine flow channel. Han et al. [7] proposed a novel cathode channel fabricated with a wave shape bottom. Results show that concentration loss induced by unstable mass transfer was delayed and the fuel cell’s performance was improved slightly.
Jiao et al. [8] studied effect of cross flow on flow pattern and performance of a PEMFC with a novel parallel flow channel design numerically and experimentally. Results show that this novel parallel flow channel design is effective in improving the PEMFC performance. Xu et al. [9] investigated the performance of thin stamped bipolar plates with a tapered channel shape. Results show that the aspect ratio and the side wall angle have great influences on the flow distribution in the channel. For the Z-type flow field, lower Reynolds number leads to a uniform flow distribution. Arun et al. [10] investigated the performance of a PEMFC with different structures including serpentine parallel, serpentine zig-zag, straight parallel and straight zig-zag flow channels. Results illustrate current and power density of the straight zigzag flow channel is quite higher than other flow channels because of better consumption of reactant, better water removal in the flow channels and uniform and optimal water distribution in the electrolyte.
In this paper, a new water removal method that investigates the use of vibro-acoutic methods is presented. Piezo-actuators are devices to generate the flexural wave is attached at end of a cathode bipolar plate. Flexural wave is used to impart energy to resting droplets and thus cause movement of the droplets in the direction of the traveling wave.
2. Theoretical Background
The water management methods proposed in this work are intended to effect water removal by generating a force, Fdrop, that is equal to the force of adhesion between the water and the supporting surface, Fadh. In equation form, the basis for water removal may the expressed as:
The force of adhesion exerted on a droplet resting on a surface if given by [11,12]
where σ is the surface tension of the liquid droplet and is the width of the droplet base, ω = 2Rsin(ϕ0).
Depending on the case examined, the supporting surface may be a membrane or a bipolar plate. Note that, in order to initiate droplet movement, Fdrop must actually be greater than Fadh. However, once initiated, movement at a constant velocity require only that Eq. (1) be satisfied and that equality will be taken as the benchmark for realizing water droplet removal.
A droplet resting on a surface is shown in Fig. 1. The assumption is that the droplet can be represented as the intersection of a sphere with a flat surface. The amount of water depends, on the contact angle ϕ0 and the radius of the assumed sphere (R), both of which are shown in Fig. 1. The contact angle is a property of the surface on which the droplet rests. The bipolar plate is generally coated with a hydrophobic material (such as Teflon) to make water removal easier. The contact angle for a surface made of Teflon is around 105° [13]. The percentage of the spherical volume that makes up the droplet increases with the droplet angle. For example, the extreme case of ϕ0 = 180° would imply a perfect sphere resting on top of a surface. The effective volume Veff and surface area Seff of the spherical droplet cap may be calculated using the relations [14].
and
Experiment to remove water drop on surface of plate using flexural wave was conducted by Scortesse et al. [15] and Biwersi et al. [16].
The flexural waves were used to impart energy to resting droplets and thus cause movement of the droplets in the direction of the traveling wave, as illustrated in Fig. 2. The main forces acting on a droplet due to flexural wave propagation are due to the radiation pressure [17]
the capillary pressure
and the hydrostatic pressure
where k = 2π/λ is the wave number, λ the corresponding wave length, ρ the liquid density and ω is the excitation frequency in rad/s. The parameters B and A are non-linear effect constants with the ration B/A=5.2 for liquids, u the vibration amplitude in meters and x is the direction of propagation of the flexural wave. The equation of capillary pressure is derived according to Laplace Law, explained in Reference [18]. In this equation, σ is the surface density of the liquid droplet and Rx and Ry are the radii of curvature on the horizontal surface of the plate in the respective directions. For a spherical droplet, Rx = Ry = R.
Defining the pressure amplitude required to move a water droplet as:
The force Fx exerted on the droplet by radiation pressure can then be calculated by integrating the radiation pressure over the angles ψ and θ, shown in Fig. 1, as:
Equating this force to the adhesive force in Eq. (2) and solving for the pressure amplitude Por,
The resulting Por can then be used with Eq. (8) to solve for the minimum vibration amplitude required to generate flexural waves that will cause droplet movement according to:
Once the vibration amplitude u has been obtained, the corresponding vibration energy required for droplet movement using this method can also be calculated. This is given by [13]
where U is velocity amplitude of vibration, ρω the water density = 1,000 kg/m3 and Veff is the effective volume of the droplet given by Eq. (3).
3. Experimental
3.1 Experimental apparatus
To measure the performance of the PEM fuel cell, we configured the performance test equipment for a unit cell as shown in Fig. 3 and Fig. 4 shows a schematic of the fuel cell experimental apparatus.
For the experiment with the fuel cell, we adjusted the flow rates of hydrogen and air by having 99.9 percent of them pass through the mass flow controller (MFC). If the hydrogen and air that pass through the MFC at the controlled flow rate are too dry, the performance of the PEM fuel cell may deteriorate. If the relative humidity of the reactant gas is low, the polymer membrane would not be hydrated sufficiently, reducing the ion conductivity and performance of the membrane. If the relative humidity of the reactant gas is high, the excessive amount of water would cover the surface of electrode catalyst to decrease the performance of the fuel cell. For this reason, in this study, the gas that passed through the MFC was allowed to go through the humidifier to maintain its relative humidity. The reactant gas, which was allowed to pass through the MFC and the humidifier so that its flow rate and relative humidity could be controlled, should be maintained at a proper temperature before it enters the unit cell. Therefore, in this study, we installed a line heater up to the inlet of the unit cell to maintain the temperature of the reactant gas constant. Since the reactant gas, whose temperature is kept constant by using line heater, may change in temperature due to heat transfer in the unit cell, we installed a heater at end plate of the unit cell to keep the bipolar plate of the unit cell at a constant temperature. Furthermore, we installed a pressure gauge at the inlet of the cathode to measure pressure at the inlet of the unit cell and installed a differential pressure gauge between the inlet and outlet of the cathode of the unit cell to calculate pressure at the outlet of the cathode of the unit cell. The data from the pressure transducer, differential pressure transducer and thermocouple were recorded and transmitted to a computer in the network connecting process by a channel recorder (Fluke). Components of the unit cell performance test system were the MFC (manufactured by Bronkhorst High-Tech), which controls the flow rate of the reactant gas, the humidifier, which supplies humidifies the reactant gas, and the 300 W DC-Loader (manufactured by E.L.P Tek), which controls the current and voltage generated in the fuel cell. The maximum power of the electronic loader was 300 W and the maximum current 50 A. The experiment can be performed at a constant current, a constant voltage and a constant resistance. The BCN terminal on the back can be used to adjust the current of the loader to an accuracy of 60.1 percent. The response and output characteristics of the unit cell can be simulated by using an external signal. In order to measure pressure of the reactant gas at the inlet and outlet of the cathode of the unit cell, we used the differential pressure gauge Model-EJA110A and the pressure gauge Model-EJA530A (both manufactured by YOKOGAWA), which can provide measurements to an accuracy of 60.1 percent. For measurement of humidity at the inlet and outlet of the cathode of the unit cell, we used a hygrometer of the HMT 337 series, which can provide measurements to an accuracy of 1.5 percent. (manufactured by VAISALA). The specifications of the experimental device are similar to those shown in Table 1.
3.2 Experiment conditions
To measure fuel cell performance, the apparatus was stabilized for 30 minutes in steady state. Humidity, temperature and pressure were maintained constant at the inlet and outlet. The temperature and pressure at the inlet and outlet, relative humidity, unit cell temperature, and open circuit voltage were measured every 1 second. Relative humidity in the unit cell was controlled by controlling the humidifier temperature. The relative humidity of the humidified air was measured with a hygrometer installed at the cathode inlet.
To operate the unit fuel cell with an operational temperature of 50°C, the unit cell temperature was set to 54°C, so the inlet and outlet temperature were maintained constant for 30 minutes, before the actual measurement. If the calculated humidity and the relative humidity measured by hygrometer differed, the humidifier temperature was adjusted to achieve the desired relative humidity. During the experiment, the relative humidity was maintained at 35, 50, 75, and 90 percent, respectively. The error range was maintained within ±0.2 percent.
In this study, pure hydrogen gas and compressed air were used, with stoichiometry of 1.5 pure hydrogen, and 2.0 compressed air, respectively. The variance of fuel cell performance was observed for both cases of fixed pure hydrogen stoichiometry and varied air stoichiometry, and the secondary fixed air stoichiometry and varied pure hydrogen stoichiometry.
Fig. 5 shows a view of the channel shape used in the unit cell performance experiment. The basic channel shape was a serpentine 5 channel, of total area 50 mm × 50 mm, and separation plate area of 80×80 mm. To make the diffusion of reaction gases faster, the rib width was 0.5 mm, so it had as small a value as possible, and the channel width was 0.9 mm.
In this study, the performance measuring experiment is conducted to suggest flexural wave capacity, in order to effectively remove the flooding phenomenon in fuel cells. Performance is compared by attaching the piezo-actuator that can generate flexural wave to the separation plate and organizing unit cell of PEMFC. Variations of conditions as follows were applied to compare unit cell performances.
Performance variation according to variation of frequency in the piezo-actuator.
Performance variation according to variation of relative humidity.
Details of experiment conditions are shown in Table 2.
3.3 Piezo-Actuator
In this study, the piezo-actuator is attached to separation plate to remove the flooding effect, using flexural wave. Power and frequency is supplied to piezo-actuator by the signal generator, and the piezo-actuator converts electric energy to mechanical energy through an inverse or indirect piezo-electric effect. Converted mechanical energy acts on a separation plate and the flexural wave is generated. The result is, subsequently removing the flooding effect by driving water drops in the direction of the wave propagation.
Fig. 6 shows piezo-actuator, which consists of 2 PZT ceramics and Beryllium copperplate in between. Fig. 7, is the fuel cell used in this study, along with the piezo-actuator attached to it. The piezo-actuator is attached so the flexural wave can be generated in direction of reaction gas propagation, in the left side of cathode separation plate.
4. Results and Discussion
4.1 Variation of performance according to variation of piezo-actuator frequency
The piezo-actuator consists of PZT ceramic and Beryllium copper plate. PZT ceramic comprises the piezo-actuator and converts electrical energy to mechanical energy – which is referred to as the “piezoelectric adverse effect.” The signal generator provides power and frequency, and the piezo-actuator generates the flexural wave from mechanical energy. But a limit exists in the frequency range. The Beryllium copper plate located between 2 PZT ceramics is unable to follow ceramic vibration in high frequencies. The piezo-actuator’s limit frequency is about 50 Hz. Therefore, the experiment is conducted within the limit frequency range. The frequencies of 10, 30 and 50 Hz are applied in these conditions where the fuel cell’s reacting gas temperature is 50°C, and relative humidity is 90 percent. The final performance is compared.
Fig. 8 shows performance curve as frequency varies. There is a visible 16 to 20 percent improvement in performance, when the flexural wave is generated. There is a particularly large difference in performance in high-current density because the water droplets generated during reaction was constantly removed by the flexural wave.
The general fuel cell shows a severe drop of voltage at high-current density. However, a fuel cell with a piezo-actuator attached maintains steady cell voltage at high density.
Performance according to variation in frequency is most effective at 50 Hz improved by about 20 percent.
4.2 Evaluation of performance to variation of relative humidity
Controlling the relative humidity of reacting gas is important in a PEM fuel cell. Reacting gas must be sufficiently dry to evaporate generated water drops, but must also be sufficiently humid to maintain high percentage of water in electrolyte surface.
The relative humidity of reacting gases can be controlled by adjusting the temperature of the humidifier wet-bulb at fuel cell inlet. Reacting gas through MFC can be passed through humidifier, to adjust its relative humidity.
Unit cell performances with a relative humidity of 35, 50, 75, and 90 percent are measured to compare PEM fuel cell performances, according to relative humidity. And the piezo-actuator was operated at a frequency of 50 Hz, selected through frequency varying experiments designed to verify effects of the flexural wave.
The reacting gas inlet temperature was maintained at 50°C. The reacting gas inlet temperature was assumed as dry-bulb temperature, and humidifier temperature was adjusted through calculation of wet-bulb temperature at relevant relative humidity. A humidity sensor from the Vaisala Corporation is used to maintain more accurate relative humidity.
Fig. 9 shows the performance curve, according to variation of relative humidity. Improvement in fuel cell performance, in accordance with an increase of relative humidity, can be observed. It indicates humidity as important factor in fuel cell performance variation. This is especially true with an operating piezo-actuator. An improvement in efficiency of 18 to 25 percent, and steady cell voltage, was observed - regardless of relative humidity in the high-current density regions of CD 1~CD 1.6.
Fig. 9(a) shows the performance curve when the relative humidity is 35 percent. Differences in performance between general fuel cell and fuel cell with a piezo-actuator, is the highest in high-current density by 25 percent. This is because cell voltage is the lowest in the high-current density region. The change is relatively large. Fig. 9(b) shows a slightly smaller efficiency difference of about 18 percent, when relative humidity is 90 percent. This is because fuel cell maintains relatively high cell voltage in CD 1.6.
Because the flooding phenomenon is removed by flexural wave in a fuel cell performance experiment according to variation of relative humidity, stable cell voltage is maintained in high-current density. Efficiency is also improved by about 18 to 25 percent. Thus the following experiment compares cell voltage variation in high-current density, when flooding occurs. The experiment is conducted in different current densities because performance varies, according to relative humidity. The current density for each relative humidity is as follows: with a relative humidity 35 percent, the humidity with the lowest performance CD is 1.5 (37 A); for relative humidity 50 percent is CD 1.6 (38 A); for relative humidity 75 percent is CD 1.6 (39 A); and 90 percent humidity with the highest performance showed relatively high and stable cell voltage in CD 1.6. Therefore the flooding experiment was conducted in the higher CD of 1.7 (43 A).
Fig. 10 shows the result of voltage as a function of time of different RH. Times where fuel cell performance drops dramatically are compared in cases where the flooding phenomenon occurs, and also in cases where flooding is removed by flexural wave. Overall in every relative humidity, cell voltage output was high – as much as 2 times – and stable cell voltage was maintained when the flexural wave was applied. In general, the fuel cell performance decreased dramatically in a range of 200 to 700 sec and became inoperative. However, the fuel cell with a piezo-actuator maintained its performance over 1,000 sec. In other words, the fuel cell was able to operate more than 6 times, in spite of the flooding phenomenon due to flexural waves.
4.3 Performance evaluation with respect to variation in reacting gas temperature
PEM fuel cell operating temperature is an important factor related to MEA lifespan, or the life span of fuel cell. An increase in operating temperature decreases fuel cell loss and results in performance improvement - but reduces MEA lifespan. In this study, the performance experiment and the flooding experiment are both conducted to verify effects of the flexural wave, according to variation in fuel cell reacting gas temperature.
Inlet and outlet temperature were maintained 30°C - similar to atmospheric temperature - for the purpose of comparing performance curves, according to operating temperature. Relative humidity was maintained at 90 percent in a temperature range controllable by humidifier. A figure of 1.5 was selected for stoichiometry, due to its high performance. Piezo-actuator operating frequency was assigned 50 Hz, which was effective for performance improvement.
Fig. 11 shows the performance curve, according to variation in reacting gas temperature. Efficiency improved by approximately 17 percent in the high-current density region of CD 1 ~ CD 1.6, and stable cell voltage output was observed by the flexural wave effect. Comparing each performance with respect to reacting gas temperature, efficiency decreased about 15 percent – compared to that of 50°C. This result shows performance improves with higher reacting gas temperature, due to more active an electrochemical reaction.
Fig. 12 shows results of the flooding experiment. The flooding experiment was conducted in CD 1.5(43 A), and in the case of the general fuel cell performance, there was a drastic decrease after about 50 sec and became inoperable. However, after the piezoactuator was attached, cell voltage output was observed over 1,000 sec.
Due to the resulting lower temperature, a decrease in fuel cell performance was observed – much similar to the case of variation in stoichiometry. Fuel cell performance became higher, as a reaction to the gas temperature. In other words, fuel cell operating temperature increased.
4.4 Flexural wave reliability experiment
So far, the piezo-actuator frequency varying experiment, and the performance flooding experiments – both regarding important factors in fuel cell such as: relative humidity, stoichiometry and reacting gas temperature – were conducted.
Through previous experiments, the most effective condition for fuel cell performance were deduced: frequency of 50 Hz, relative humidity 90 percent, stoichiometry of 1.5, and a reacting gas temperature 50°C. Thus in this experiment, a fuel cell is operated for long time in effective fuel cell operating conditions, to contemplate the effect of the flexural wave.
Fig. 13 shows result of the flexural wave reliability experiment. The differences of cell voltage between the general fuel cell and the fuel cell with a piezoactuator are approximately 5 percent in CD 1.1. However, after 15 hours of operation, the cell voltage difference was approximately 18 percent. Increase in cell voltage difference is due to the removal of droplets, generated through an electrochemical reaction by a flexural wave. Removal of droplets lead to high efficiency at high-current density, and therefore cell voltage output was stable. Comparing cell power instead cell voltage comparison as shown in Fig. 14 can verify flexural wave effect more firmly. Initial cell power showed no significant difference, around 12 to 13 Watts. After 15 hours, the difference in performances of two fuel cells was about 15 percent. Due to flexural wave, fuel cell with a piezo-actuator attached showed about 13 percent decrease in performance of 12.7 to 11 Watt, while general fuel cell showed about 28 percent decrease of 12.4 to 9 Watt.
The flexural wave reliability experiment verified that the flexural wave produced movement of droplets, possible and delayed droplet accumulation inside channel, thus producing stable cell voltage and performance improvement.
5. Conclusions
In this study, vibro-acoustic method is applied to fuel cells, in order to remove the flooding phenomenon. The piezo-actuator is attached to the fuel cell separation plate, along with power and frequency being supplied by the signal generator - thus the flexural wave is generated by piezo-adverse effect.
Conclusion of this study is as follows:
Observing results of the PEM fuel cell performance experiment, it can be seen that fuel cell performance improves as relative humidity increases. However at high-current density, cell voltage is unsteady due to flooding phenomenon and performance decreases rapidly.
It is verified through the frequency variation experiment that performance is improved the most at 20 percent at 50 Hz, which is the limit frequency of piezo-actuator.
The piezo-actuator is operated at optimal frequency of 50 Hz, and relative humidity is varied while the performance measuring experiment is conducted. As a result, the flooding phenomenon was removed by the flexural wave and rapid decline of fuel cell performance was prevented.
In conclusion, fuel cells with a piezo-actuator showed higher output – density at high-current density – and maintained steady cell voltage.
Acknowledgement
This study was supported by 2016 Research Grant from Kangwon National University (No. 620160067) and this work (Grants No. C0533189) was supported by Business for Cooperative R&D between Industry, Academy, and Research Institute funded Korea Small and Medium Business Administration in 2017.
Glossary
List of Symbols
F
Force, N
P
Pressure, kPa
R
Radius, N
ρ
Liquid density
Ψ
Contact angle, °
σ
Surface tension
ω
Width of the droplet
λ
Corresponding wave length
Subscripts
gen
generating force
adh
adhesion
r
radiation
c
capillary
in
inlet
out
outlet
h
hydrostatic