Multi-objective optimization of a hydrogen-fueled PEMFC with multi wavy channels via machine learning and CFD simulation

Abstract

Proton exchange membrane fuel cells (PEMFCs) are crucial for clean energy conversion and hydrogen energy systems. However, practical PEMFC operations face conflicting requirements between gas supply efficiency and water removal under high-current conditions, leading to performance degradation and flooding risks. This study proposes a multi-wavy channels flow field (MWCFF) design with a stretch factor to achieve progressive flow oscillation for synergistic optimization of mass transfer and water management. This study integrates a 3D Multiphysics PEMFC model with a neural network-NSGA-II (Non-dominated Sorting Genetic Algorithm II) optimization framework. Using this integrated approach, five critical parameters (temperature, cathode stoichiometry, stretch factor, channel width, and channel depth) are systematically optimized. The optimized configuration shows 13.2–22.5 % power density enhancement over conventional designs while reducing flooding risks. This work provides a novel flow field optimization paradigm through geometric-function integration, offering theoretical guidance for two-phase flow regulation and technical solutions for next-generation fuel cell stack engineering.