Optimisation of a grid-connected electrolyser: time-ahead dispatch and energy portfolio planning

This study investigates short-term power dispatch optimisation and long-term energy capacity planning for a grid-connected water electrolysis system (WES). It explicitly accounts for key technical characteristics often overlooked in existing studies, including dynamic specific energy consumption (SEC; incorporating both stacks and balance of plant (BoP)), minimum load, ramping, and usage-based stack degradation. To enable computationally tractable optimisation without sacrificing these key technical aspects that introduce nonlinear dynamics, a two-step approach is adopted. First, nonlinear governing relationships (e.g., between input power and SEC) are derived using a detailed cell-level stack model. Second, these relationships are approximated using piecewise linearisation (PWL) with minimal error. The results show that time-ahead dispatch optimisation significantly reduces total electricity cost relative to constant operation; however, this benefit diminishes at higher required average WES utilisation due to reduced temporal flexibility (i.e., less freedom to shift production), which in turn constrains the effective use of operational flexibility (e.g., ramping). Optimal energy portfolio design is highly sensitive to yearly average WES utilisation and power purchase agreement (PPA) prices. Higher utilisation rates (URs) require increased grid-sourced firming of the power supply, whereas lower utilisation enables greater reliance on variable renewable energy (VRE), albeit with larger contracted capacities, yet remaining economically favourable. Sensitivity analysis further shows that wind PPA plays a dominant role in portfolio design, driven by its higher base cost relative to solar PPA and more consistent generation profile.