DOI QR코드

DOI QR Code

딥러닝 기반 PEMWE 중첩 기포 분리 및 전류 밀도에 따른 유동 패턴 분석

Deep learning-based analysis of flow patterns and separation of overlapping bubbles in PEMWE under varying current densities

  • Seungmin Han (Department of Biomedical Engineering, Soonchunhyang University) ;
  • Sohyeon Kim (Department of Biomedical Engineering, Graduate School, Soonchunhyang University) ;
  • Hyeonjin Lee (Department of Biomedical Engineering, Graduate School, Soonchunhyang University) ;
  • Hanwook Park (Department of Biomedical Engineering, Soonchunhyang University)
  • 투고 : 2026.04.16
  • 심사 : 2026.05.08
  • 발행 : 2026.08.31

초록

Proton exchange membrane water electrolysis (PEMWE) faces efficiency degradation as increased current density leads to bubble accumulation, which reduces the active reaction area. To investigate this phenomenon, an analysis of bubble dynamics is required. This study proposes a method that sequentially utilizes nnU-Net for segmentation and You Only Look Once (YOLO) for key point extraction. The analysis is performed by identifying concave points as geometric landmarks to distinguish individual bubbles. This framework enables the characterization of flow patterns during PEMWE operation. Experimental results showed that the average number of bubbles in the upstream channel (2nd channel) slightly increased before declining, while it consistently decreased in the downstream channel (10th channel). The void fraction increased with current density regardless of the channel location. This study provides fundamental data for analyzing two-phase flow characteristics in water electrolysis systems.

키워드

과제정보

이 논문은 2025학년도 순천향대학교 교수 연구년제에 의하여 연구하였음. 2026년도 정부(교육부)의 재원으로 한국연구재단의 지원을 받아 수행된 기초연구사업임(RS-2021-NR066087).

참고문헌

  1. Li, K. and Lin, B., 2015, "Impacts of urbanization and industrialization on energy consumption/CO2 emissions: Does the level of development matter?," Renewable and Sustainable Energy Reviews, Vol. 52, pp.1107~1122. https://doi.org/10.1016/j.rser.2015.07.185
  2. International Energy Agency, 2025, Global Hydrogen Review 2025, IEA Publications, Paris.
  3. Akpasi, S. O., Smarte Anekwe, I. M., Tetteh, E. K., Amune, U. O., Mustapha, S. I. and Kiambi, S. L., 2025, "Hydrogen as a clean energy carrier: advancements, challenges, and its role in a sustainable energy future," Clean Energy, Vol. 9(1), pp.52~88. https://doi.org/10.1093/ce/zkae112
  4. Carmo, M., Fritz, D. L., Mergel, J. and Stolten, D., 2013, "A comprehensive review on PEM water electrolysis," International Journal of Hydrogen Energy, Vol. 38(12), pp.4901~4934. https://doi.org/10.1016/j.ijhydene.2013.01.151
  5. Li, Y., Kang, Z., Mo, J., Yang, G., Yu, S., Talley, D. A., Han, B. and Zhang, F.-Y., 2018, "In-situ investigation of bubble dynamics and two-phase flow in proton exchange membrane electrolyzer cells," International Journal of Hydrogen Energy, Vol. 43(24), pp.11223~11233. https://doi.org/10.1016/j.ijhydene.2018.05.006
  6. Sangtam, B. T. and Park, H., 2023, "Review on Bubble Dynamics in Proton Exchange Membrane Water Electrolysis: Towards Optimal Green Hydrogen Yield," Micromachines, Vol. 14(12), p.2234. https://doi.org/10.3390/mi14122234
  7. Colliard-Granero, A., Gompou, K. A., Rodenbücher, C., Malek, K., Eikerling, M. H. and Eslamibidgoli, M. J., 2024, "Deep learning-enhanced characterization of bubble dynamics in proton exchange membrane water electrolyzers," Physical Chemistry Chemical Physics, Vol. 26(20), pp.14529~14537. https://doi.org/10.1039/D3CP05869G
  8. Barredo Arrieta, A., Díaz-Rodríguez, N., Del Ser, J., Bennetot, A., Tabik, S., Barbado, A., Garcia, S., Gil-Lopez, S., Molina, D., Benjamins, R., Chatila, R. and Herrera, F., 2020, "Explainable Artificial Intelligence (XAI): Concepts, taxonomies, opportunities and challenges toward responsible AI," Information Fusion, Vol. 58, pp.82~115. https://doi.org/10.1016/j.inffus.2019.12.012
  9. Isensee, F., Jaeger, P. F., Kohl, S. A. A., Petersen, J. and Maier-Hein, K. H., 2021, "nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation," Nature Methods, Vol. 18(2), pp.203~211. https://doi.org/10.1038/s41592-020-01008-z
  10. Grabowski, B., Ziaja, M., Kawulok, M., Long'ep'e, N., Saux, B. L. and Nalepa, J., 2022, "Self-Configuring nnU-Nets Detect Clouds in Satellite Images," ArXiv, Vol. abs/2210.13659.
  11. Ronneberger, O., Fischer, P. and Brox, T., 2015, "U-Net: Convolutional Networks for Biomedical Image Segmentation," Medical Image Computing and Computer-Assisted Intervention - MICCAI 2015, pp.234~241.
  12. Jocher, G., Chaurasia, A. and Qiu, J., 2023, "Ultralytics YOLO," https://github.com/ultralytics/ultralytics.
  13. Lee, C. U., Sung, J. and Lee, D. H., 2025, "Application of AI Instance Segmentation Models for Microbubble Diameter Measurements and Its Performance Evaluation," Journal of the Korean Society of Visualization, Vol. 23(4), pp.30~41.
  14. Tian, Y., Liao, W., Xu, Z. and Wang, W., 2024, "A Binocular Vision Fish Body Length Measurement Method Based on Improved YOLOv8-pose," Proc. 5th International Symposium on Computer Engineering and Intelligent Communications (ISCEIC), Vol. 1, pp.622~625.
  15. Chen, Z. and Wang, T., 2025, "Lightweight Excavator Bucket Pose Estimation Algorithm via Improved YOLOv8," Proc. 6th International Conference on Machine Learning and Computer Application (ICMLCA), Vol. 1, pp.71~75.
  16. Schepperle, M., Junaid, S. and Woias, P., 2024, "Computer-Vision-and Deep-Learning-Based Determination of Flow Regimes, Void Fraction, and Resistance Sensor Data in Microchannel Flow Boiling," Sensors (Basel), Vol. 24(11), p.3505. https://doi.org/10.3390/s24113363