Benchmarking control strategies for multi-stack electrolyzer systems under renewable energy variability

This work presents a benchmarking framework for evaluating control strategies in multi-stack proton-exchange membrane electrolyzer systems operating under renewable energy variability. As the deployment of electrolyzers grows, especially in distributed and renewable-integrated settings, ensuring high efficiency, low degradation, and robust operation becomes increasingly important. A novel semi-empirical degradation model is developed to assess the long-term performance of different control strategies, accounting for membrane, catalyst, and bipolar plates degradation. The paper compares state-of-the-art rule-based and model-predictive control based distribution algorithms, under realistic renewable energy input profiles. Rule-based strategies such as Optimal Steady Efficiency and Sequential then Equal demonstrate strong hydrogen production performance while balancing degradation effects. Model-predictive control based strategies can surpass most rule-based methods, but only when using sufficiently detailed production-curve modeling. Their performances remain slightly below Optimal Steady Efficiency but achieve fewer module start-ups. Parametric studies demonstrate that rule-based approaches can be improved through parameterization, while model-predictive control strategies benefit from concave relaxation and reduced time horizons. Sensitivity analyses reveal that while prediction errors and representation inaccuracies affect performance, strategies remain robust The findings highlight the trade-offs between computational complexity, efficiency, and degradation, providing insights for the design of flexible and durable multi-stack electrolyzer systems.