TY - JOUR
T1 - Cell-by-cell temperature observer for commercial PEMFC stacks
T2 - a thermal-hydraulic approach
AU - Fang, Zhendu
AU - Zhu, Yujin
AU - Wei, Zhongbao
AU - He, Hongwen
AU - Liu, Xinghua
N1 - Publisher Copyright:
© 2025 Elsevier Ltd.
PY - 2026/2/15
Y1 - 2026/2/15
N2 - Precise knowledge of the internal temperature field is essential for safeguarding proton exchange membrane fuel cell (PEMFC) stacks, yet existing observers either rely on intrusive hardware, provide only lumped averages, or are validated exclusively on laboratory-scale hardware. This study develops a coupled thermal–hydraulic model and embeds it in an extended Kalman-filter (EKF) observer that reconstructs the cell-by-cell temperature of a 40 kW, 180-cell commercial liquid-cooled stack. The observer exploits only routine operating signals—stack current, voltage, and coolant pressures and temperatures—to infer otherwise unmeasurable states. Validation during pre-heating, start-up with deliberate coolant interruption, and dynamic load operation yields a root-mean-square error of 0.95 °C and confines the maximum relative error to < 2.5 % across all cells. Benchmarking against data-driven hybrid models demonstrates a 16.9 % improvement in RMSE, highlighting the advantage of physics-based estimation in data-sparse conditions. Furthermore, a systematic robustness analysis confirms the observer’s stability against parameter uncertainties and sensor noise, ensuring reliable performance under practical operating conditions. To the authors’ knowledge, this is the first observer that (i) delivers full-stack, cell-resolved temperatures using only standard PEMFC measurements and (ii) is experimentally verified on a commercial-scale stack rather than on small laboratory stacks. The method provides a practical foundation for hot-spot detection, durability enhancement, and model-predictive thermal control in fuel-cell electric vehicles.
AB - Precise knowledge of the internal temperature field is essential for safeguarding proton exchange membrane fuel cell (PEMFC) stacks, yet existing observers either rely on intrusive hardware, provide only lumped averages, or are validated exclusively on laboratory-scale hardware. This study develops a coupled thermal–hydraulic model and embeds it in an extended Kalman-filter (EKF) observer that reconstructs the cell-by-cell temperature of a 40 kW, 180-cell commercial liquid-cooled stack. The observer exploits only routine operating signals—stack current, voltage, and coolant pressures and temperatures—to infer otherwise unmeasurable states. Validation during pre-heating, start-up with deliberate coolant interruption, and dynamic load operation yields a root-mean-square error of 0.95 °C and confines the maximum relative error to < 2.5 % across all cells. Benchmarking against data-driven hybrid models demonstrates a 16.9 % improvement in RMSE, highlighting the advantage of physics-based estimation in data-sparse conditions. Furthermore, a systematic robustness analysis confirms the observer’s stability against parameter uncertainties and sensor noise, ensuring reliable performance under practical operating conditions. To the authors’ knowledge, this is the first observer that (i) delivers full-stack, cell-resolved temperatures using only standard PEMFC measurements and (ii) is experimentally verified on a commercial-scale stack rather than on small laboratory stacks. The method provides a practical foundation for hot-spot detection, durability enhancement, and model-predictive thermal control in fuel-cell electric vehicles.
KW - Cell-resolved temperature
KW - extended Kalman filter
KW - Non-invasive monitoring
KW - PEMFC stack
KW - State observer
KW - Thermal management
UR - https://www.scopus.com/pages/publications/105025146737
U2 - 10.1016/j.enconman.2025.120965
DO - 10.1016/j.enconman.2025.120965
M3 - Article
AN - SCOPUS:105025146737
SN - 0196-8904
VL - 350
JO - Energy Conversion and Management
JF - Energy Conversion and Management
M1 - 120965
ER -