AI-enhanced techno-economic feasibility of off-grid renewable hydrogen refueling stations

This study investigates a wind–solar–storage-driven renewable hydrogen refuelling station for Northern Oman (Barka, Sohar, Al Hadd), targeting reliable service of an AC auxiliary demand of 334 kWh/day with 33.06 kW peak under local resources (mean solar irradiance 5.54–5.68 kWh/m2/day; mean wind speed 7.4–10.4 m/s). The study’s hybrid optimisation workflow (metaheuristics/learning-assisted search and multi-objective Bayesian optimisation) uses HOMER Pro as the main energy-balance and lifecycle-cost engine to screen four architectures (PV-WT-FC-B, WT-FC-B, PV-WT-B, and WT-B) across hydrogen-price sensitivity (0.1-0.3 USD/kg). Optimised designs are consistently wind-led (10 × 330 kW turbines) and electrolyser-limited (100 kW) with 1000 kg hydrogen storage; PV appears only as a marginal contributor (0–61.8 kW) alongside 305–574 kWh batteries and, where installed, a 250 kW fuel cell. Annual electricity throughput ranges from 4.78 × 10 6 to 6.97 × 10 6 kWh, and hydrogen yield ranges from 3,814 to 4,464 kg/yr across sites and configurations. Battery-dominated layouts (WT-B and PV-WT-B), which deliver LCOE = 0.0166-0.0170 USD/kWh and LCOH = 0.456-0.592 USD/kg, achieve the most competitive economics. Beyond “zero on-site emissions”, a cradle-to-grave screening indicates zero operational air pollutants (CO 2 /NO x /SO 2 /PM/VOC = 0 kg/yr) but non-zero embodied climate burdens with GWP100 = 0.30–0.60 kgCO 2 e/kg H 2 and an implied 20-year indirect GWP of 22.9–53.6 tCO 2 e, reinforcing the need for coupled techno-economic and life-cycle reporting for deployment-grade sustainability claims.