• Citer

Development of Optimisation tools for the design under uncertainty of multi-energy hubs

par Garcia Arenas, Julien
Président du jury Parente, Alessandro
Promoteur Hendrick, Patrick
Co-Promoteur Henneaux, Pierre
Publication Non publié, 2026-07-02

Thèse de doctorat

• Accès en ligne
• Détails
• Contenu
• Statistiques

Résumé :

The European Union has committed to achieving climate neutrality by 2050, implying a profound transformation of energy systems through large-scale deployment of renewable energy sources, widespread electrification of end-use sectors, and increasing reliance on decentralised resources. While these developmentsare essential for decarbonisation, they also introduce significant challenges related to renewablevariability, system flexibility, infrastructure planning and investment risk. In this context, multi-energy hubs, which integrate multiple energy vectors such as electricity, heat, hydrogen and other synthetic molecules, emerge as a promising approach for improving the efficiency, resilience and flexibility of future energy systems.This thesis develops optimisation-based methodologies for the techno-economic design and operation of decentralised multi-energy systems (DMES) under uncertainty. Building upon the energy hub concept, a unified modelling framework is proposed to represent energy conversion, storage and exchange technologies within a common optimisation environment. The framework combines deterministic optimisation, Monte Carlo simulation and stochastic programming, thereby enabling uncertainty to be considered at different levels of modelling complexity. A central challenge addressed throughout this work is the trade-off between model accuracy and computational tractability. Increasing temporal resolution, technological detail and uncertainty representation rapidly leads to large-scale optimisation problems affected by the curse of dimensionality. The proposed methodologies therefore seek a pragmatic balance between realism and solvability through the use of linearised techno-economic formulations, representative temporal resolutions and uncertainty reduction techniques. This approach allows uncertainty to be explicitly incorporated while preserving acceptable computational requirements.The modelling framework is implemented through an integrated workflow combining GAMS for optimisation, Matlab for input generation and scenario construction, and Python for automation, post-processing and visualisation. Particular attention is devoted to aligning modelling assumptions with the intended decision context, with a focus on behind-the-meter and community-scale applications. The framework is applied to four complementary case studies covering multiple spatial scales and uncertainty representations. At the regional scale, a Brussels-Capital Region case study demonstrates the value of sector coupling for reducing system costs and facilitating renewable integration. At the district scale, the H2GridLab living-lab case study investigates the role of hydrogen technologies, heat recovery and environmental constraints within local multi-energy systems. At the residential scale, a UK household case study employs Monte Carlo simulation to capture climatic variability and occupant behaviour, highlighting the limitations of deterministic designs based on average profiles. Finally, a hydrogen refuelling station case study introduces a two-stage stochastic programming formulation to investigate robust system design under uncertainty in photovoltaic generation and dynamic electricity prices. Across all case studies, flexibility emerges as the central enabler of highly renewable DMES. The results demonstrate that sector coupling, thermal flexibility, battery storage, hydrogen storage and demand-side flexibility significantly improve techno-economic performance. The analyses further reveal that uncertainty cannot be adequately represented through average conditions alone, and that robust investment decisions require explicit consideration of variability. In particular, the stochastic case study highlights the limitations of representative average-year approaches and demonstrates the value of flexibility-oriented investment strategies capable of performing satisfactorily across multiple future operating conditions. Overall, this thesis contributes both methodological developments and practical insights for the design of DMES. By combining deterministic, Monte Carlo and stochastic optimisation approaches within a common framework, it provides decision-support tools for researchers, policymakers, system planners and investors seeking to accelerate the deployment of flexible, resilient and low-carbon energy systems.