The high costs associated with fuel cell manufacturing have precluded its production on a large scale. The major emphasis of the present wok is to bring down the overall cost of an independent fuel cell unit. The manufacturing cost can be reduced using commonly available and corrosion resistant materials into the fuel cell assembly. Bipolar plates usually employed in proton exchange membrane fuel cells are fabricated from conducting graphite. Graphite owing to its conductivity, corrosion resistance and easy machinability, is the preferred material in static systems. However, due to its brittle characteristics and failure under bending loads, graphite is inferior in its mechanical properties as compared to metals and their alloys. Dimensional stability is also compromised due to wear and friction. In the present work, an attempt is made to assemble a fuel cell stack which would have durability and sustainability in dynamic conditions, where the setup would be able to withstand periodic shocks, vibrations, and fatigue loads. Instead of employing graphite as the bipolar plate which serves the dual purpose of a current collector and area for flow fields, graphite foil protected aluminum as the current collector and machined plastic slabs on which the flow fields are carved, have been employed. Both the substitutes are easily available owing to mass production and have a small processing cost associated with them. Further, the technique employed for processing of Nafion and hot pressing of the catalyst loaded gas diffusion layer onto the proton exchange membrane have been elaborated in the present paper along with the systematic approach followed by the research group eliminating various current collector candidates for fuel cell applications. The various stages attained towards the final fabrication of the foil protected lightweight current collector, has also been highlighted in the present work.
An optimal coordinated proton exchange membrane fuel cell heat management method based on large-scale multi-agent deep reinforcement learning
