Model Complexity and Optimization Trade-offs in the Design and Scheduling of Hybrid Hydrogen-battery Systems

The production of hydrogen from renewable sources could play a significant role in supporting the transition toward a decarbonized energy system. This study has involved investigating optimization strategies − mixedinteger linear programming (MILP), a hybrid particle swarm optimization (PSO)-MILP framework, and PSO combined with a rule-based energy management strategy (EMS) − applied to a power-to-hydrogen system for industrial applications. The analysis evaluates the levelized cost of hydrogen production (LCOH), carbon emissions, and the impact of key factors, such as battery degradation, electrolyzer efficiency, real-time pricing, and hydrogen load management. The obtained results indicated that the MILP-based models achieved moderate LCOH values (10.1–10.7 €/kg) but incurred higher CO2 emissions (20.2–24.6 kt/y). Instead, the PSO model, combined with the rule-based EMS, lowered emissions to 14.3 kt/y (a 27–45% reduction), albeit with a higher LCOH (11.6 €/kg). The hybrid PSO-MILP models struck a balance, achieving LCOH values of between 9.2 and 9.7 €/kg, with CO2 emissions of 19.7–20.3 kt/y, as they benefited from the integration of piecewise affine linearization for modeling electrolyzer efficiency and battery degradation. In terms of computational efforts, the MILP-based models required more than 48 h to converge, while the PSO-MILP models completed within 27–35 h, and the PSO model with rule-based EMS achieved results in 1.5 h. These findings offer guidance that can be used to select the most suitable optimization method on the basis of the desired performance targets, resource constraints, and computational complexity, thereby contributing to the design of more sustainable energy systems.