Least-regret hydrogen infrastructure design under demand uncertainty

This study presents a routing-and-sizing framework for hydrogen pipeline networks that minimises max regret across uncertain demand. An obstacle-aware genetic algorithm generates corridors over a weighted-GIS surface; a multi-period hydraulic sizing step selects commercial diameters subject to pressure, velocity, and wall-thickness constraints. Decisions are taken in a rolling-horizon so topology and capacity adapt as information arrives. Applied to the UK Humber cluster from 2030 to 2050, built length reaches 165–200 km, with a Spine-First routing strategy averaging 185 km. Least-regret oversizing adds £40 m compared to a myopic approach but cuts 2040 worst-case incremental outlay from £260 m to < £80 m. By 2050, Spine-First achieves £2.07 m/km, LCOT 54 £/kt and regret 12 £/kt, rivalling a Perfect-Foresight strategy (44 £/kt; 3 £/kt). The results show how a short, centrally aligned trunk combined with anticipatory sizing reduces stranded-asset risk and budget shocks, providing a transferable least-regret template for hydrogen pipelines under deep uncertainty.