A framework integrating multi-physic and data-driven models and optimization approaches to maximize electrical efficiency and power of PEMFC

The performance of open-cathode air-cooled proton exchange membrane fuel cells (PEMFCs) is critically dependent on fan characteristics, which can significantly impact electrical efficiency and output. This paper proposes a PEMFC mathematical model, a surrogate model driven by a database containing 120,000 data sets, and an optimization algorithm to form a framework for addressing low efficiencies in existing air-cooled stacks caused by fan operation. The purpose is to maximize the electrical efficiency and output power of an open-cathode air-cooled PEMFC stack at various currents, and to identify the optimal power combinations for fan operation within the stack. The results indicate that fan operation has multiple interactive effects on the stack. The developed data-driven surrogate model demonstrates excellent predictive performance, validated by three robust evaluation metrics. By adjusting the operating power of the fans, the maximum increments in electrical efficiency of the stack at 6 A, 27 A and 48 A are 49.79 %, 10.00 % and 5.91 %, respectively. For the stack rated at 1 kW, the maximum output power reached up to 236.52 W, 839.43 W and 1091.04 W at 6 A, 27 A, and 48 A respectively, with the maximum increments in output power of 4.14 W, 64.80 W and 59.52 W.