Search

Abstract: Water management in PEM fuel cells has received extensive attention due to its key role in fuel cell performance. The unavoidable water, from humidified gas streams and electrochemical reaction, leads to gas–liquid two-phase flow in the flow channels of fuel cells. The presence of two-phase flow increases the complexity in water management in PEM fuel cells, which remains a challenging hurdle in the commercialization of this technology. Unique water emergence from the gas diffusion layer, which is different from conventional gas–liquid two-phase flow where water is introduced from the inlet together with the gas, leads to different gas–liquid flow behaviors, including pressure drop, flow pattern, and liquid holdup along flow field channels. These parameters are critical in flow field design and fuel cell operation and therefore two-phase flow has received increasing attention in recent years. This review emphasizes gas–liquid two-phase flow in minichannels or microchannels related to PEM fuel cell applications. In situ and ex situ experimental setups have been utilized to visualize and quantify two-phase flow phenomena in terms of flow regime maps, flow maldistribution, and pressure drop measurements. Work should continue to make the results more relevant for operating PEM fuel cells. Numerical simulations have progressed greatly, but conditions relevant to the length scales and time scales experienced by an operating fuel cell have not been realized. Several mitigation strategies exist to deal with two-phase flow, but often at the expense of overall cell performance due to parasitic power losses. Thus, experimentation and simulation must continue to progress in order to develop a full understanding of two-phase flow phenomena so that meaningful mitigation strategies can be implemented. [Copyright &y& Elsevier]

Abstract: Two-phase flow in horizontal parallel channels has been experimentally investigated under fuel cell related operating conditions. Pronounced hysteresis is observed in the pressure drop versus flow characteristic curve when starting from either flooded or dry conditions. When gas is introduced into channels initially filled with water (flooded initial condition), both gas and liquid tend to flow predominantly in one channel at low gas or liquid flow velocities. As the gas flow velocity increases, even distribution of gas and liquid flow in both channels is observed, accompanied with a sudden decrease in the pressure drop. On the other hand, even gas and liquid flow distribution between both channels is found at comparatively lower gas flow velocities when starting with dry-gas flow conditions with liquid introduced into channels filled with gas (stratified flow regime). The flow regimes of this system are visualized in plots of the pressure drop against gas and liquid flow velocities. However, this phenomenon tends to vanish at high gas and liquid flow velocities, suggesting that high gas and liquid flow velocities are required to ensure even flow distribution in parallel channels. The hysteresis points appear at the same level of the pressure drop, reflecting intrinsic characteristics of the parallel channels used in this study. These results have important implications for PEM fuel cell operational strategies. In order to avoid reactant mal-distribution in parallel flow channels in the flow field in the two-phase flow regime, fuel cells should be operated at sufficiently high gas flow velocities. [Copyright &y& Elsevier]

The transient 3-D visualization of the desaturation process of flooded gas diffusion layers (GDLs) is presented for the first time. Desaturation rates and pathways are reported for two commercial GDL samples, one with a PTFE treatment and the other without. On-the-fly Synchrotron X-ray CT is performed while the GDL sample is subjected to a humidified or dry air flow stream to visualize the transient desaturation. The two humidification conditions assist in separating the convective and evaporative components of the desaturation process, showing a slight contribution from the convective effect, while the majority of the desaturation is due to evaporative removal. The convective removal is found to be insufficient to fully desaturate either GDL, with water remaining trapped underneath the channel rib with the more hydrophobic GDL and within the pore space in the more hydrophilic GDL. Stop-and-shoot Synchrotron X-ray computed tomography (CT) is then used in conjunction with program-assisted segmentation to determine initial saturation water volumes. These are then combined with the desaturation times found from the on-the-fly experiments to determine desaturation rates for both GDLs. The desaturation rate using dry air flow for the more hydrophobic GDL is found to be nearly four times faster than that of the more hydrophilic GDL. Results demonstrate that an evaporative contribution is necessary for either GDL sample to reach full desaturation. [ABSTRACT FROM AUTHOR]

The visualization of the thawing and desaturation process on an initially saturated, frozen gas diffusion layer (GDL) with a serpentine gas flow channel was performed based on synchrotron X-ray computed tomography images. High speed CT scanning during the experiments allowed the dynamic desaturation process to be quantified under the cold-start with air purging condition. The saturation profiles and the desaturation rates were studied over the entire GDL domain, through-plane, and in selected regions of interest for localized behavior. Sigracet 35AA and 35BA GDLs were selected for the experiments to study the effects of GDL hydrophobicity. Along with the real-time saturation profiles, the average desaturation rates for the entire GDL domain over the whole purging process were 0.000186 μL cm−2 s−1, 0.000470 μL cm−2 s−1, 0.000516 μL cm−2 s−1 and 0.000901 μL cm−2 s−1 with the superficial gas velocity of the purging air at 2.88 m/s, 4.26 m/s, 5.98 m/s and 9.02 m/s, respectively. In addition, the dynamic saturation contours and 3-D GDL geometry models were constructed to show the liquid water movement through a GDL. Although the GDL desaturation curves for each experiment share similar trends, the results show that different conditions including air flow rate, GDL geometric location, initial water saturation, and GDL boundary condition could cause heterogeneous desaturation behavior on both overall and localized GDL regions. These data provide valuable information for future modeling studies that involve the thawing process in the GDL, and could be used to optimize the cell design and develop cold-start protocols. • Dynamic GDL thawing and desaturation profile is quantitatively analyzed via synchrotron X-ray CT. • Higher purging rate speeds up the thawing and desaturation process; GDL hydrophobicity has no significant impact on these. • Overall GDL saturation may not describe the blockage level accurately due to heterogeneity. [ABSTRACT FROM AUTHOR]

Abstract The 4-D imaging and quantification of the desaturation of an initially flooded gas diffusion layer (GDL) with a serpentine gas flow channel by synchrotron radiography is presented. An imaging area with a diameter of 10 mm allows for the study of how the natural anisotropy of a GDL affects the desaturation profile. The GDL is progressively spatially segmented from the overall domain down to sections of the individual channels and ribs. Temporal saturation profiles and desaturation rates are presented for each step of this progressive segmentation. Although the desaturation of the overall domain initially appears to be a constant and steady process with a desaturation rate of 0.0030 μL cm−2 s−1, segmented results display heterogeneous behavior with over two-fold differences between areas of the flow field. Segmented saturation surface plots are presented that spatially show how this heterogeneity occurs within the entire domain. Results indicate that initial conditions, anisotropic material properties, and flow field geometry all play a key role in determining local desaturation behavior. The larger domain considered here provides valuable data for future modeling studies that combine continuum studies of flow in the gas flow channels with GDL pore network models that consider two-phase flow behavior. Highlights • Spatial and temporal GDL saturation is analyzed from Synchrotron X-ray CT data. • As the GDL is segmented spatially, variations in the desaturation rate are noted. • Overall desaturation rate of 0.0030 ± 0.0013 μL cm−2 s−1 is spatially investigated. • Spatially and temporally resolved desaturation rates are key to model validation. [ABSTRACT FROM AUTHOR]

Proper water management in proton exchange membrane fuel cells (PEMFCs) remains a challenge, hindering the commercialization of this technology. Liquid water emerges in the form of droplets from micrometer-scaled pathways of porous gas diffusion layers (GDLs) into millimetre-scaled gas flow channels. Afterward, droplets grow to certain critical sizes before being removed by the convective gas flow. The opaque nature of fuel cell components, including GDL materials, poses significant challenges to understand droplet dynamics and develop more efficient water management techniques. In this work, an advanced synchrotron X-ray radiographic imaging technique available at the Canadian Light Source (CLS) was employed to visualize time dependent droplet behaviour in a serpentine PEMFC flow channel for the first time. High spatial and temporal resolution coupled with high energy photons of an X-ray beam provided high resolution images of growing droplets. From the captured images, the whole growth cycle of liquid droplets was obtained under various gas flow rates. A cyclic pattern of droplet dynamics including emergence, growth, and detachment was visualized from high resolution images. From the X-ray images, the time-evolution of dynamic contact angles, height, and chord of a growing droplet was quantitatively analyzed. At higher gas velocities, the critical detachment diameter of droplets became smaller and contact angle hysteresis was found be lower, while larger contact angle hysteresis and bigger droplets were observed at lower gas velocities. The critical droplet size at which the detachment of droplets occurred was compared to the predicted results from an analytical approach and good agreement was found. [ABSTRACT FROM AUTHOR]

Abstract: Two-phase flow pressure drop hysteresis was studied in a non-operational PEM fuel cell to understand the effect of stoichiometry, GDL characteristics, operating range, and initial conditions (dry vs. flooded) for flow conditions typical of an operating fuel cell. This hysteresis is noted when the air and water flow rates are increased and then decreased along the same path, exhibiting different pressure drops. When starting from dry conditions, the descending pressure drop tended to be higher than the ascending pressure drop at lower simulated current densities. The hysteresis effect was noted for stoichiometries of 1–4 and was eliminated at a stoichiometry of 5. It was found that the hysteresis was greater when water breakthrough occurred at higher simulated current densities, which is a function of GDL properties. The operating range had to reach a critical simulated current density (800mAcm−2 in this case) between the ascending and descending approach to create a pressure drop hysteresis zone. The descending step size does not change the size of the hysteresis effect, but a larger step size leads to lower fluctuations in the pressure drop signal. An initially flooded condition also showed hysteresis, but the ascending approach tended to have a higher pressure drop than the descending approach. [Copyright &y& Elsevier]