Multi-timescale framework optimizes hybrid electrolyzer green hydrogen operations

A collaborative research team published a refined multi-timescale scheduling strategy for alkaline-proton exchange membrane (ALK/PEM) hybrid electrolyzer systems in Engineering, addressing persistent stability and economic hurdles facing renewable-powered hydrogen production facilities.

Global energy transitions rely on green hydrogen to offset fossil fuel reliance, yet standalone alkaline electrolyzer systems suffer rigid power limits and frequent start-stop losses, while single PEM setups carry prohibitive capital costs. Hybrid electrolyzer systems (HES) pair the low-cost, large-scale capacity of ALK stacks with the fast-response, wide-load flexibility of PEM units, but existing scheduling work lacks fine-grained modeling of individual electrolyzer state transitions and rarely couples hydrogen demand side adjustments with generation-side volatility. The new research fills this gap by integrating joint wind-photovoltaic forecasting, dual-type hydrogen demand response (DR), and two-layer rolling optimization for coordinated system dispatch.

The team built a CNN–BiLSTM–Attention joint wind–solar prediction model to mitigate renewable output uncertainty, filtering input meteorological variables via Pearson correlation coefficient screening to reduce prediction bias across gale, breeze, and cloudy weather conditions. Two forms of hydrogen load demand response are embedded into the system framework: price-based DR adjusts time-varying hydrogen prices inversely to renewable generation output to shift shiftable hydrogen loads, while incentive-based DR enables controlled reduction of interruptible industrial and refueling station hydrogen consumption. Mathematical elasticity matrices quantify load sensitivity to hydrogen price fluctuations, linking supply-side power swings to flexible user demand profiles.

For electrolyzer modeling, distinct binary state variables describe three operating modes for ALK stacks (hydrogen production, standby, shutdown) and three states for PEM stacks (overload, hydrogen production, shutdown), with dedicated cold and hot startup constraints, minimum runtime/downtime limits, and load range boundaries to match real-world safety standards. The Big-M linearization method converts mixed integer nonlinear constraints into a solvable mixed integer linear programming model, balancing computational speed and physical accuracy.

The multi-timescale rolling optimization splits dispatch into a 1-hour day-ahead stage maximizing daily net revenue and a 15-minute intraday rolling stage minimizing deviation from day-ahead schedules. The day-ahead objective accounts for hydrogen sales income, grid power trading gains, equipment operation and maintenance charges, amortized investment costs, and electrolyzer start-stop penalties, while intraday optimization prioritizes steady electrolyzer operation to slow equipment degradation. Case simulations draw on a combined 200 MW wind farm and 110 MW photovoltaic plant in Macheng, Hubei Province, with a configured HES featuring five 37 MW ALK units and three 10 MW PEM units.

Benchmark comparisons against five distinct operational scenarios confirm measurable performance improvements. Deploying HES instead of ALK-only facilities lifts daily net revenue, while integrating hydrogen demand response nearly doubles grid trading income by aligning hydrogen consumption with surplus wind-solar power. Across all tested weather profiles, the multi-timescale rolling framework delivers above 96% day-ahead schedule execution for every electrolyzer unit. Under stable wind-solar conditions, the approach reaches 98% of the theoretical maximum economic return achievable via pure intraday profit maximization, and retains 80% of this economic performance during volatile cloudy weather. The research notes the framework’s extensibility to hydrogen refueling stations, off-grid renewable microgrids and broader integrated energy systems, and identifies stochastic robust optimization for long-duration annual operation as a direction for follow-up work.

The paper “Multi-Timescale Scheduling Optimization of ALK/PEM Hybrid Electrolyzers System Considering Flexible Hydrogen Demand,” is authored by Bowen Wang, Zhaoqing Liang, Kai Yang, Lei Xing, Heng Shao, Zhuorui Wu, Yixin Liu, Li Guo, Ning Yang, Bing Hu, Chengshan Wang, Kui Jiao. Full text of the open access paper: https://doi.org/10.1016/j.eng.2026.02.020. For more information about Engineering, visit the website at https://www.sciencedirect.com/journal/engineering.