Planning of local-scale hydrogen-based renewable energy systems: a review of uncertainty, optimization, and decision methods

Hydrogen can improve energy system reliability and cost-effectiveness by acting as both a storage medium and an energy carrier. This review focuses on local-scale hydrogen-based renewable energy system (H2-RES) planning, where uncertainties, component-sizing complexity, and conflicting objectives must be addressed together. It provides a focused methodological synthesis of uncertainty modeling, optimization, and multi-criteria decision-making (MCDM) as interconnected elements of the planning process. These aspects form an interconnected chain: uncertainty modeling defines realistic inputs such as renewable resources and demand; optimization uses these inputs to size components and generate feasible alternatives; and MCDM compares optimized solutions across multiple criteria for final design selection. A key contribution of this review is the integrated examination of uncertainty modeling, optimization, and MCDM as interconnected stages of local-scale H2-RES planning. Three methodological trends emerge from the reviewed literature. First, probabilistic methods, particularly Monte Carlo simulation, are widely applied, while static distributions, limited hydrogen-demand data, and neglected dependencies highlight the need for generative data extension and correlation-aware uncertainty modeling. Second, artificial intelligence (AI)-based and hybrid metaheuristics receive increasing attention for optimization, but black-box behavior, limited generalizability, and inconsistent benchmarking require more transparent and standardized evaluation. Third, MCDM methods support comparison among design alternatives, but sensitivity to criterion weights and ranking methods, together with narrow techno-economic criteria, highlights the need for robustness testing and broader sustainability assessment. Overall, reliable local-scale planning requires uncertainty modeling, optimization, and final decision-making to be treated as an integrated process rather than as isolated methodological stages.