Optimization of hydrogen supply operations for decarbonizing energy-intensive industries: A multi-timescale rolling horizon approach

The adoption of green hydrogen, produced via water electrolysis using renewable energy sources, is a promising decarbonization strategy for energy-intensive industries. However, the feasibility of this transition depends on the economic viability of hydrogen supply and the ability to ensure a stable hydrogen supply under fluctuating and uncertain renewable energy availability. This study develops a Mixed-Integer Nonlinear Programming (MINLP) model to optimize hydrogen supply operations. On-site hydrogen production is supported by grid electricity purchases and complemented by external green hydrogen truck deliveries to ensure the continuous fulfillment of hydrogen demand in industrial furnaces. The model captures key electrolyzer operational dynamics, including variable loads, state transitions, and stack efficiency degradation. The formulation is embedded in a multi-timescale framework that accounts for different decision frequencies and implementation lead times of electrolyzer operations and external hydrogen delivery. The problem is solved using a rolling horizon approach to reduce reliance on long-term forecasts and enable reactive scheduling of hydrogen supply operations under renewable energy uncertainty. Results indicate that seasonal variations in renewable energy availability and grid electricity prices can cause operating cost differences of up to 60%. In contrast, renewable energy forecast inaccuracies result in cost variations limited to 2.47% under the rolling horizon approach, which achieves operating cost reductions of up to 4.15% compared to static optimization, demonstrating the robustness and relevance of the proposed framework. The use of real historical forecast datasets or the integration of forecasting algorithms represents an interesting direction for future research.