EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Design and screening of complex serpentine flow fields in redox flow batteries with expanded topological space

Xue Yang, Jinpeng Hu, Mingzhe Jiang, Hong Yang, Chengqian Xu, Honggang Wang, Dongxu Ji

Clean Energy · 09.07.2026 · 10 (4) · 117-135

DOI: 10.1093/ce/zkag030 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Clean Energy, online erschienen am 14.08.2026. Verfasst von Xue Yang und 6 weiteren Autor:innen. Die Arbeit nennt 50 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Natural Science Foundation of China, Shenzhen Science and Technology Innovation Commission, Shenzhen Science and Technology Program.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Redox flow batteries are promising for grid-scale energy storage, but the design of high-performance flow fields remains constrained by a narrow set of predefined serpentine prototypes and the high cost of case-by-case numerical evaluation. Here, we develop a physics-labeled machine-learning framework for the rapid design and screening of complex serpentine flow fields in an expanded topological space. Five representative topology-generation algorithms were used to construct a heterogeneous serpentine design space, from which 4000 representative candidates were selected and evaluated by three-dimensional numerical simulations. The resulting dataset provides two hydrodynamic labels: a flow-uniformity metric, y1, and a pressure-drop metric, y2. Random forest, conventional convolutional neural network, and residual-network-based convolutional neural network models were then systematically compared in terms of predictive accuracy, generalization, stability, robustness, interpretability, and inference efficiency. The residual-network-based model delivered the best overall performance, achieving test R2 values of 0.8031 for y1 and 0.8583 for y2, with mean absolute percentage errors of 1.53% and 0.48%, respectively. Large-scale screening identified several complex serpentine designs that outperformed the conventional serpentine flow field, increasing y1 from 0.668 to 0.765 and reducing y2 from 245.93 Pa to 220.64 Pa. Additional coupled electrochemical analysis further showed that the screened candidates reduced the concentration overpotential from 120.21 mV to 97.44–98.73 mV, confirming that the hydrodynamic improvements translate into reduced concentration-polarization losses. This work demonstrates that topology-space expansion combined with physics-labeled surrogate screening can accelerate flow-field discovery and identify designs with improved electrolyte distribution, lower hydraulic resistance, and enhanced electrochemical transport behavior.

Förderung

  • National Natural Science Foundation of China
  • Shenzhen Science and Technology Innovation Commission
  • Shenzhen Science and Technology Program

Zitieren

APA

Yang, X., Hu, J., Jiang, M., Yang, H., Xu, C., Wang, H., & Ji, D. (2026). Design and screening of complex serpentine flow fields in redox flow batteries with expanded topological space. Clean Energy, 10(4), 117-135. https://doi.org/10.1093/ce/zkag030

Vancouver

Yang X, Hu J, Jiang M, Yang H, Xu C, Wang H, et al. Design and screening of complex serpentine flow fields in redox flow batteries with expanded topological space. Clean Energy. 2026;10(4):117-135. doi:10.1093/ce/zkag030

BibTeX
@article{Yang2026,
  author = {Yang, Xue and Hu, Jinpeng and Jiang, Mingzhe and Yang, Hong and Xu, Chengqian and Wang, Honggang and Ji, Dongxu},
  title = {Design and screening of complex serpentine flow fields in redox flow batteries with expanded topological space},
  journal = {Clean Energy},
  year = {2026},
  volume = {10},
  number = {4},
  pages = {117--135},
  doi = {10.1093/ce/zkag030}
}
Kostenlos starten →