Structure design for the gas distribution zone of PEMFC bipolar plates: A data-driven multi-objective optimization framework integrating the Taguchi method and grey relational analysis

The gas distribution zone (GDZ) is a critical flow field component for enhancing the overall performance of proton exchange membrane fuel cells (PEMFCs). However, conventional monolithic designs often cause problems such as uneven gas distribution, low mass transfer efficiency, and inadequate water-thermal management. To address these challenges, this study proposes a partitioned GDZ design concept for precise gas distribution control and a two-stage optimization framework based on limited data for the synergistic improvement of performance metrics. In the first stage, the GDZ was divided into primary and secondary distribution zones, and then seven configurations were evaluated using a 3D two-phase PEMFC model. A hybrid configuration was identified as the optimal case (dots in primary and ribs in secondary distribution zones). Building on this configuration, four key geometric parameters of the GDZ were further optimized in the second stage. An orthogonal array was first employed to efficiently design 16 runs of experiments, with four targets defined: net output power density ( W net ), oxygen concentration non-uniformity ( U O 2 ), flow velocity non-uniformity ( U flow ), and membrane water content ( λ ¯ ). The Taguchi method and analysis of variance (ANOVA) were then applied to assess parameter significance and quantify their contribution rates. Finally, grey relational analysis (GRA) delivered the optimal solution, which improved all four targets versus the first-stage baseline: a 4.328% increase in W net , a 0.612% increase in λ ¯ , and reductions of 6.160% in U O 2 and 10.913% in U flow . Compared to the first-stage optimum, three targets were further enhanced, with the slight W net loss (only 1.838%) outweighed by gains in U O 2 , U flow , and λ ¯ . This study validates the novel partitioned GDZ design concept and establishes a comprehensive, transferable optimization framework for the systematic design of various PEMFC components.