Sustainable collaborative planning of green hydrogen refueling networks in urban agglomerations: A three-stage framework integrating geospatially explicit screening, fuzzy MCDM, and stochastic optimization

Urban agglomeration hydrogen refueling station (HRS) networks must coordinate renewable hydrogen supply, spatially heterogeneous demand, and inter-station operation. This study develops a three-stage planning framework that links geospatial screening, fuzzy multi-criteria decision making (MCDM), and stochastic bi-objective capacity optimization for green HRS networks. Candidate areas are identified by GIS (Geographic Information System)-based suitability analysis considering renewable resources, transport demand, infrastructure, environmental constraints, and market-related factors. Candidate stations are prioritized by an integrated CRITIC-DEMATEL-TOPSIS model, and representative wind turbine (WT), photovoltaic (PV), and Hydrogen Fuel Cell Vehicle (HFCV) demand scenarios are generated using kernel density estimation, Latin hypercube sampling (LHS), and Backward Reduction (BR). The optimization model minimizes annual system cost and life-cycle carbon emissions while explicitly accounting for hydrogen production, storage, purchase, and inter-station transfer constraints. The Urumqi-Changji-Shihezi case identifies eight priority HRS sites. The compromise solution yields an annual cost of 1954.01 million CNY and life-cycle emissions of 23,639.56 t CO2, reducing emissions by 28.36% and 40.28% relative to blue and grey hydrogen benchmarks. Inter-station dispatch lowers the average supply-demand gap by 90.62%, while the WT-PV hybrid configuration reduces cost and emissions by 37.88% and 39.33% relative to single-energy alternatives. The framework is most suitable for renewable-resource-rich urban agglomerations and delivers a replicable yet context-specific decision process for regional green hydrogen infrastructure planning.