A rolling horizon approach to optimizing green hydrogen production with grid-connected proton-exchange membrane electrolyzers

Green hydrogen production using grid-connected Proton Exchange Membrane (PEM) electrolyzers can play an important role in decarbonization. An approach to advance adoption through improved economically viability is short-term scheduling to align hydrogen production with fluctuating electricity prices. Existing methods typically rely on fixed forecasts or assume perfect foresight, which may not be realistic in practice. This study proposes a novel adaptive and scalable scheduling framework that integrates electricity price forecasting directly into the operation of a multi-electrolyzer system. A mixed-integer linear programming (MILP) model is developed, incorporating on/off/standby decisions, compression, hydrogen storage, and an objective that penalizes unmet demand while maximizing overall profit. A baseline is first established under perfect foresight, showing that allowing the schedule to adapt to time-varying electricity prices increases profit by 25% compared to a fixed operating schedule. This is followed by the implementation of a rolling horizon optimization (RHO) framework that uses Seasonal Auto-Regressive Integrated Moving Average (SARIMA) generated forecasts, updated at each step using the most recent data. Case study results show that on average RHO outperforms single-forecast scheduling, with the timing of price peaks having a greater influence on performance than average forecast error.