Computational Modeling of Tubular Flow Fields for PEM Fuel Cells

Abstract

The bipolar plates of polymer electrolyte membrane fuel cells (PEMFCs) are of focus for improvement since they are responsible for most of the fuel cell stack weight and volume. Traditionally these flow fields are machined or stamped out of a graphitic or metallic material. Recent research has shown that additively manufactured metal plates show comparable performance to those made with conventional method. Due to the layer-by-layer nature, there is a great deal of design flexibility for additively manufactured parts. This allows for far more freedom in flow field designs, such as a tubular flow field with internal channels.In recent years there have been developments of tubular PEMFCs. Numerical simulation reported in literature have shown that tubular PEMFCs provide higher power density [1], more uniform reactant gas velocity distribution [2], and a smaller pressure gradient [3].The goal of the tubular model in our study was to replicate in size a standard AAA battery , so the total dimensions of the fuel cell were limited at a height of 40.5 mm and a diameter of 10.5 mm, with a cell active area of 987 mm2. The tubular channel length was approximately 400 mm. For reference, a conventional (planar) single serpentine fuel cell with a comparable 982 mm2 active area was used, replicating a scaled down commercially available design from Fuel Cell store. This planar design had square channels with a depth of 1.016 mm and a channel width of 0.7874 mm. The length of the channel was approximately 550 mm. For both designs, the land widths were 0.7874 mm to maintain a 1:1 channel-land width ratio. The gas diffusion layer (GDL), microporous layer (MPL), catalyst layer, and membrane had respective thicknesses of 360 μm, 50 μm, 10 μm, and 50 μm. The channel geometry for the tubular fuel cell was changed from rectangular to a semicircle, as the additive manufacturing process is not adept at building overhangs with supports. To maintain an accurate comparison, the cross-sectional area of the planar and tubular channels was kept constant at 0.79 mm2. This flow field would be incorporated into the outer anode flow field of the final tubular design, so hydrogen at 80°C was used as the fluid. A volumetric flow of 0.24 L/min was used as the inlet boundary condition.Initially, basic fluid flow simulations were performed on several tubular serpentine flow field designs (Figure 1). The double corkscrew Tubular v4 design had the greatest pressure drop, which was due to the fact the cross-sectional area was divided into two smaller sections. Tubular v2 and v3 had similar results being the most alike geometrically. Tubular v1, which split off into two symmetrical flow channels, exhibited the lowest pressure drop as the half channels cut down on the distance the hydrogen would have to travel inlet to outlet.With the chosen design Tubular v1, this geometry was used for the PEMFC module. Various changes were made to the tubular model to simulate the new design. Our proprietary membrane electrode assembly (MEA) design had no GDL present in the cathode layer, with oxygen being pumped directly through the cathode catalyst layer. This design assured that the supplied O2 concentration is much higher since its transport is by convection instead of diffusion. The ANSYS PEMFC model will not run if there is no defined GDL between the gas channel and MPL, so this was circumvented by directly changing the properties of the catalyst layer. To model this, the mass fraction of the oxygen in the cathode catalyst layer was increased from 0.2 to 0.3. It is also expected for the flowing of oxygen to flush away any water formation, so the water formation was limited. Due to the water mass balance taking place in the model, limiting liquid water formation increased the dissolved water inside of the membrane. The electrochemical surface area was increased to account for the higher surface area of the Pt catalyst in our proprietary MEA design.These modifications showed an increase in performance for the tubular cell when compared with the standard planar design as seen in Figure 2. Limiting the liquid water formation and selecting a geometry with a more uniform pressure distribution resulted in a cell with higher power density. The removal of the GDL and increase of air flow allowed for a higher oxygen concentration at the cathode catalyst interface. With oxygen no longer being a limiting reactant, the cell was able to perform at a higher power output. Figure 1