Ant colony optimization for complex multi-objective supply chain problems: application to the optimal deployment of green hydrogen in Belgium

Details

Supervisors

Contino, Francesco

Faculty

Ecole polytechnique de Louvain

Degree label

Master [120] : ingénieur civil mécanicien, à finalité spécialisée

Abstract

By 2050, the demand of hydrogen for Europe and UK could reach up to 25% of the total final energy consumption. Most of the industrial demand is due to the production of chemicals, as ammonia or methanol, but also to refineries and steel industry. Today European hydrogen is mostly produced from fossil fuels by Steam Methane Reforming (SMR) while the current green alternatives cover less than 4% of the total hydrogen production. Through the integration of a hybrid metaheuristic algorithm, this work analyses the optimal hydrogen supply chain network design (HSCND) to decarbonize the industrial production of hydrogen in Belgium by 2050. While most literature focuses on optimizing the HSCND from financial perspectives, this study includes the environmental impact as second objective. The Pareto-optimality of cost and CO2 emissions is studied based on the ε-constraint method. The hybrid metaheuristic is based on the integration of the Multi-Variate Ant Colony Optimization (ACO-MV) as master algorithm to handle the variables associated to the nonlinear constraints, while a classical Mixed Integer Linear Programming (MILP) solver is adopted for the solution of the slave problem. A parallel implementation, based on the simplification of the problem with the removal of the non-linearities, is used to test the performance of the hybrid metaheuristic algorithm. The total demand in Belgium is estimated at 19.6 TWh/year, distributed between 7 plants where hydrogen can be produced from SMR and electrolysis or imported from abroad. The results of the case study show that the hybrid implementation offers similar performances compared to the simplified version. Both show a progressive replacement of the operating SMR plants by imported green hydrogen, neglecting electrolysis deployment which is less competitive. The financial impact for installing 1 GW of electrolysis plants in Belgium could cause up to 30% of cost increase compared to the optimal solution. A sensitivity analysis is finally conducted to observe the trend required from the market to make electrolysis in Belgium a profitable option. From worst to best scenario for the price of imported hydrogen, the price of European electricity should fall under 60€ or 80€/MWh respectively, compared to 100€/MWh considered in reference case.