Multilayered topology optimization model of two-phase nonisothermal PEM fuel cell

Purpose This study aims to advance the design of proton exchange membrane (PEM) fuel cells by introducing a novel topology optimization approach. It seeks to overcome the computational challenges of full 3D simulations through a cost-effective pseudo-3D multilayered model. Additionally, it explores the simultaneous optimization of both the anode and cathode flow fields to enhance performance. The objectives are to maximize power generation, ensure uniform current density distribution and facilitate prototype manufacturing. Design/methodology/approach The proposed methodology employs a pseudo-3D multilayered model that captures key physical phenomena in fuel cells, including electrochemistry, species transport, multiphasic fluid flow, energy conservation and membrane water transport. Governing equations are discretized using the finite element method, and sensitivities are computed via the adjoint method with automatic differentiation. The optimization uses a density-based material model and a gradient-based interior point algorithm to iteratively refine topologies. Multiple objective functions are considered, balancing power generation and current density homogenization. Finally, the feasibility of the optimized designs is validated through the manufacturing of 3D-printed prototypes. Findings The study demonstrates the efficacy of the pseudo-3D model in reducing computational costs while maintaining high fidelity in simulating fuel cell behavior. By integrating simultaneous anode-cathode optimization, the approach provides novel insights into performance tradeoffs and optimized design configurations. Results reveal distinct optimized topologies can be achieved by varying the objective function weights. The fabrication of a 3D-printed prototype underscores the practical applicability of the proposed designs. Originality/value This work introduces an innovative approach to fuel cell topology optimization by combining a pseudo-3D multilayered model with simultaneous anode-cathode design, providing a computational viable optimization framework. Furthermore, the integration of multi-objective optimization with manufacturability considerations provides a unique contribution to the field, highlighting the applicability.