The deployment of proton exchange membrane (PEM) water electrolysis plants is accelerating, creating a need for accurate lifetime prediction to support reliable and cost-effective operation. Yet, degradation models suitable for plant-level lifetime estimation remain scarce. To address this gap, this study introduces a layer-unspecific physical model to estimate voltage degradation and the remaining useful life of PEM water electrolysis stacks from current-voltage data. The model extracts quasi-steady polarization curves from operation without diagnostic downtime. As a central novelty, it fits a simplified electrochemical equation assuming a fixed Tafel slope, which enables stable tracking of the temporal evolution of resistance and exchange current density by avoiding parameter collinearity. Uncertainty is quantified through Monte Carlo propagation of regression parameters, enabling lifetime forecasts with confidence intervals. Validation is conducted with a 26 200 h synthetic dataset and a 2200 h experimental six-cell stack dataset. The synthetic data enable validation over industry-relevant lifetimes without multi-year testing, while the experimental data confirm practical applicability. Across both datasets, the model reproduced voltages with millivolt-scale error and provided prognostic horizons suitable for maintenance planning. Overall, the approach offers a practical and interpretable model for PEM water electrolysis efficiency prediction and accelerates model development while reducing reliance on extended lifetime experiments.
Building similarity graph...
Analyzing shared references across papers
Loading...
Felix Dittmar
Thomas Lickert
Uwe Küter
Journal of Power Sources
Building similarity graph...
Analyzing shared references across papers
Loading...
Dittmar et al. (Sat,) studied this question.
www.synapsesocial.com/papers/69a75a0dc6e9836116a1f8bf — DOI: https://doi.org/10.1016/j.jpowsour.2025.239122