Numerical simulation analysis

The complex interrelationship between liquid water and power generation requires the development of simulation techniques capable of reproducing physical transport phenomena and electrochemical reactions. It mainly includes the continuity equation, Navier-Stokes equation, and the energy equation, potential equation and transport equation of each chemical substance. The commercial computational fluid dynamics (CFD) software Fluent is used to connect these equations as external custom functions, and the corresponding equations such as electrochemical reactions are defined as source terms. The particular focus of this research is liquid water in GDL. Simulations were performed using the Multiphase Mixture (M2) model, which is capable of efficiently analyzing two-phase liquid/gas flow in porous structures. Material parameters are set from ex-situ measurements of the part or values ​​determined during fuel cell testing.

Mechanism analysis

Analytical procedures: Section 4.2 below describes the determination of power generation performance through current-voltage (IV) evaluation tests, the determination of liquid water distribution through visualization experiments, and the use of numerical simulations to determine the interrelationship between power generation and liquid water distribution. Next, Section 4.3 describes numerical simulations to derive the physical properties of GDL that contribute to liquid water distribution. This section also uses liquid water visualization experiments and physical property measurement results to verify the calculation results. This series of analytical work finally clarified the interrelationship between power generation performance, liquid water distribution and GDL physical properties.


Figure 7 Mechanism analysis process.

Analysis of the relationship between power generation performance and liquid water distribution

In order to quantify the impact of liquid water distribution on power generation performance, the conventional method of dealing with liquid water distribution and wet diffusion coefficient (i.e., coupled calculation in numerical simulation) has been changed. Input the visualized liquid water distribution and the diffusion coefficient in the presence of liquid water into the model to calculate the power generation performance in the presence of liquid water (Figure 8). In order to compare the computational performance degradation caused by liquid water with the actual performance degradation, the dimensionless dry-wet performance ratio is used as a comparison metric (Figure 9). In experiments, it was found that performance in wet conditions dropped to approximately 43% of that in dry conditions. The simulation results show that the performance drops by about 61%, indicating that this method can simulate 70% of the performance drop.


Figure 8 Performance prediction under wet conditions.


Figure 9 Dry and wet performance comparison

This method was used to compare the liquid water distribution of two GDLs (A, B) with different power generation performance levels (Figure 10). The factors causing this performance difference were analyzed. As shown in Figure 11, the internal space of the GDL was divided into 12 equal sections. The difference in liquid water volume in each section was used as the factor and the current density was used as the target variable. Through simulation, it was obtained Contribution rate (Figure 12). These results found that the distribution of liquid water under the GDL flow channel (sections 1, 2 and 3) has an important impact on power generation performance, and reducing the distribution of liquid water in these places is very important to improve power density.


Figure 10 Liquid water distribution of A, B-GDL


Figure 11 GDL liquid water cross section


Figure 12 Effect of liquid water cross-section on performance

The 30% difference between the calculated results and the experimental results (Fig. 9) is believed to be due to the inability of liquid water to be expressed through visual experiments and models (Fig. 14). An example is given in Figure 13(1) (details are shown in Figure 14). Since the visualization experiment uses X-ray photography technology, the liquid water in the depth direction is processed into a layered structure. Although the numerical simulations are calculated in three dimensions, liquid water is represented as a spatial average in the computational grid. In contrast, in the phenomenon that actually occurs, liquid water exists in the GDL substrate in the form of droplets hundreds of μm in diameter. The presence of these droplets in the substrate creates local inhomogeneities in the oxygen transport path, and the size of these inhomogeneities creates different concentrations of oxygen transported within the MPL plane. In contrast, in the simulations, oxygen was transported relatively uniformly from the GDL substrate, making differences in oxygen concentration in the planar direction of the MPL unlikely to occur. Therefore, since the performance changes caused by changes in MPL diffusion coefficient are not considered in the model, there is a large difference between the calculated results and the experimental results.


Figure 13 Visualization experiments and phenomena not considered in the model


Figure 14 Differences in oxygen transport paths between experiments and models

As mentioned above, the discrepancy between computational and experimental results is due to the inability of current visualization experiments or models to recognize the microscopic level of liquid water phenomena. Therefore, it is necessary to continue to develop analytical techniques to improve analytical accuracy by more closely reflecting the phenomena (1) to (3) in Figure 13.

Analysis of the relationship between liquid water distribution and GDL physical properties

The analysis is performed by redesigning the revised numerical simulation described in the previous section to achieve coupling to the liquid water distribution. By observing the visualization experiment, the following physical properties can be selected as the main driving factors of liquid water and water vapor in the system: capillary pressure, thermal conductivity (under the base bipolar plate ridge, under the base bipolar plate flow channel, under the MPL ridge and MPL flow channel) and gas diffusion coefficient (MPL in-plane and vertical direction). Through calculations, the sensitivity of the physical properties of GDL to water distribution can be analyzed. Figure 15 shows the results of the analysis of the sensitivity of the water distribution under the flow channel relative to the cathode substrate, which has a significant impact on the performance described in the previous section. These results show that the thermal conductivity of the cathode substrate and cathode MPL has a greater impact on the distribution of liquid water. When the thermal conductivity of the cathode substrate is lower and the thermal conductivity of MPL is higher, the liquid water in the cathode substrate reduce. The reasons are shown in Figure 16: (I) Reducing the thermal conductivity of the cathode substrate increases the substrate temperature, thereby reducing the amount of water condensation; (II) Increasing the thermal conductivity of the cathode MPL increases the heat flow on the cathode side, causing the cathode to The substrate temperature increases, again causing the amount of water condensation to decrease. In addition, lowering the temperature of the membrane facilitates water migration, increasing the amount of water flowing from the cathode to the anode, further reducing the liquid water in the cathode substrate.

In order to verify the validity of the calculation results, the thermal conductivity of GDL (A, B) at two different power generation performance levels (Figure 10) was measured (Figure 17). In GDL B (low liquid water level), the GDL substrate has a lower thermal conductivity than GDL A, and MPL has a higher thermal conductivity than GDL A. The obtained trend is consistent with the calculated results, verifying the validity of the sensitivity calculation.


Figure 15 Liquid water distribution sensitivity analysis results

Conclusion: This paper describes a method to analyze the interrelationship between the power generation performance, liquid water distribution and physical properties of GDL with the aim of determining the drainage mechanism of GDL. The following points were identified using the specific GDL types and conditions described in this article.

1. Liquid water under the cathode channel has the greatest impact on power generation performance. The thermal conductivity of the GDL substrate and MPL has the greatest impact on the distribution of liquid water under the cathode channel.

2. By designing a cathode substrate with low thermal conductivity and a cathode MPL with high thermal conductivity, the amount of liquid water in the cathode substrate can be reduced.

3. This study combines X-ray visualization of liquid water with simulation to demonstrate the importance of GDL thermal design in controlling liquid water behavior in fuel cells. There are plans to continue improving the methods developed in this study to help solve various problems caused by liquid water in polymer electrolyte fuel cells (PEFC) and accelerate the development of low-cost, high-reliability fuel cells.