Optimal strategic deployment of green hydrogen production and supply infrastructures in Belgium based on MINLP model, solved with multi-variate metaheuristic algorithm

Ambitious targets have been set to bring the world to net zero carbon emissions in the coming decades. The replacement of today’s energy sources with renewable, stable and cheap sources is key, however it represents a challenge from a societal, economical and technological point of view. Among the solutions to be adopted, the introduction of a new energy carrier like hydrogen can fasten the defossilization of sectors heavily dependent on fossil sources, for which few mature technological alternatives exist. This master thesis aims at developing an approach to optimize the future hydrogen production-distribution system of a country, while taking into account major sources of uncertainty. Indeed, integrating unknown future hydrogen demand and energy costs changes the way the design of the system balances imports and local production. That is what we will show by applying our developments to Belgium as case study. We develop an algorithm combining mixed-integer non-linear programming with a multi-stage development of scenarios to integrate sources of uncertainty. More specifically, we develop a two-level algorithm where an Ant-Colony metaheuristic embeds a classic solver, which allows to clearly separate linear from non-linear objectives and constraints. Our results are promising, as seen on the application to the case study. Our approach finds stable network configurations at different points in time (2030, 2040, 2050), at minimum capex and opex levels. The computational time remains reasonable for different problem sizes and constraint types. Our approach is generic enough to introduce new dimensions to the objective function, new constrains, more time-steps and new technology of production and transport.