Green hydrogen production system design optimisation with strategic reliability decisions

Green hydrogen (GH) is central to global decarbonisation strategies, yet its large-scale deployment remains constrained by high costs, limited operational experience, and the lack of integrated planning frameworks that jointly account for reliability, maintenance, and renewable resource variability across diverse grid connectivity and demand conditions. This study develops a novel reliability-based mixed-integer nonlinear programming (MINLP) framework that uses stochastic optimisation to simultaneously optimise technology selection, subsystem sizing, maintenance planning, and availability investment decisions for GH production systems, with green ammonia serving as the downstream energy carrier for long-distance transport. A case study in Eastern Canada identifies a wind-dominated configuration with supplemental solar and battery storage as the optimal solution, achieving a levelised cost of hydrogen (LCOH) of $4.44 kg −1 . Comparative geographic analyses reveal that wind-dominant regions achieve costs 18% to 41% lower than solar-dominant regions across best- to worst-case operating conditions, where lower solar capacity factors and stronger seasonal generation patterns necessitate substantially larger storage investments with limited production benefits. The framework also identifies two primary sources of renewable energy curtailment: the mismatch between installed renewable capacity and electrolyser absorption capability, and electrolyser downtime. Improving electrolyser availability reduces total curtailment by up to 48%, yet imposing stricter curtailment limits increases system costs without improving hydrogen production, with low-demand systems experiencing the greatest economic penalty due to higher incremental maintenance requirements. The findings demonstrate that maintenance strategies, curtailment management, grid interaction, and technology configuration must be jointly optimised under region-specific resource and demand conditions to achieve viable GH production. This study provides a practical framework to support these decisions during the design phase.