Scholarlink

<p>Optimizing solid oxide fuel cell (SOFC) performance requires balancing competing electro-mechanical demands. This paper presents an integrated framework leveraging artificial neural networks (ANNs) and the multi-objective grey wolf optimizer (MOGWO) to improve flow channel geometry under coupled electro-mechanical constraints. Channel-shaping parameters, including width, height, and inclination angle, were optimized along with the interdependent interconnect rib dimensions. A Sobol sequence-based design of experiments ensures robust surrogate training, enabling rapid simultaneous maximization of power density and minimization of the interconnect von Mises stress and electrolyte failure probability. Optimal trade-off solutions are identified using the TOPSIS method and validated via FEM simulations. The results reveal two dominant design regimes: wide, shallow channels favor power output but increase mechanical risk, whereas taller, narrower channels enhance structural reliability. The proposed framework significantly reduces computational cost compared to direct FEM approaches, offering a practical tool for comparative thermo-mechanical screening of candidate SOFC designs.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Electro-mechanical trade-offs in solid oxide fuel cell flow channels: multi-objective optimization with machine learning-assisted grey wolf optimizer

  • 期刊论文

  • 发表日期：

    2026年7月2日

查看全文
  • Aimen A. Bouaiss,
  • Mohamed Souri Mimoune,
  • Djafar Chabane,
  • Achraf Senoussi,
  • Oussama Bouaiss,
  • Ali Mohammadi,
  • Nadhir Lebaal

摘要

Optimizing solid oxide fuel cell (SOFC) performance requires balancing competing electro-mechanical demands. This paper presents an integrated framework leveraging artificial neural networks (ANNs) and the multi-objective grey wolf optimizer (MOGWO) to improve flow channel geometry under coupled electro-mechanical constraints. Channel-shaping parameters, including width, height, and inclination angle, were optimized along with the interdependent interconnect rib dimensions. A Sobol sequence-based design of experiments ensures robust surrogate training, enabling rapid simultaneous maximization of power density and minimization of the interconnect von Mises stress and electrolyte failure probability. Optimal trade-off solutions are identified using the TOPSIS method and validated via FEM simulations. The results reveal two dominant design regimes: wide, shallow channels favor power output but increase mechanical risk, whereas taller, narrower channels enhance structural reliability. The proposed framework significantly reduces computational cost compared to direct FEM approaches, offering a practical tool for comparative thermo-mechanical screening of candidate SOFC designs.