CFD-based optimization toolchain for channel geometry in PEM fuel cells with novel BPP material

This work presents a generally applicable, physics-based methodology for optimizing flow channel designs in PEM fuel cells using CFD modeling and multi-objective optimization. A fully resolved 3D CFD model was developed to capture the coupled transport and electrochemical phenomena relevant to Proton Exchange Membrane Fuel Cell (PEMFC) operation. The model is grounded in physical laws and requires only two fitting parameters. These parameters were calibrated using a dedicated fitting algorithm developed within this paper, enabling accurate reproduction of the polarization curve, spatial current density distribution along the channel, and liquid water distribution within the cell across the relevant operating range. The validated model was applied to optimize a novel channel geometry for a low-conductivity polymer-based bipolar plate. The optimization revealed a fundamental trade-off between flow uniformity and electrochemical performance. The selected geometry delivers a current density of j = 1 . 719 A cm 2 , representing an increase of 0 . 172 A cm 2 compared to the reference design, while optimizing media distribution uniformity and reducing the cell pitch by 0.13 mm compared to the reference design. The successful optimization confirms the insights from individual parameter studies and validates the robustness and general applicability of the approach. The combined workflow of CFD simulation, parameter fitting, and Pareto-based optimization offers a transferable framework for future channel design challenges, especially when novel bipolar plate materials introduce complex physical constraints. • 3D CFD model captures transport and electrochemistry in PEM fuel cells. • Only two calibrated parameters are needed to match key experimental data. • Novel channel geometry is optimized for low-conductivity bipolar plates. • Optimization improves current density and reduces cell pitch simultaneously. • The proposed CFD and optimization workflow is transferable and robust.