MOBIDIC

Energy systems are evolving into highly interconnected multi-carrier networks spanning individual processes and large-scale infrastructures. As complexity increases, model-based optimisation becomes essential for designing and operating such systems under dynamic and uncertain conditions. Physics-based models offer interpretability and physical consistency but are often too computationally demanding for real-time or system-level use. Data-driven models, while efficient at capturing complex behaviours, require extensive datasets and often suffer from limited generalisation and physical coherence. Hybrid modelling approaches that blend physical knowledge with machine learning therefore represent a promising path. However, experts in energy systems typically master physics and engineering principles but lack advanced machine-learning skills. Conversely, data-science specialists excel in AI techniques but often lack the domain knowledge required to handle energy systems and their operational constraints. MOBIDIC bridges this gap by training a new generation of researchers who operate at the interface between energy systems engineering and AI. The 15 doctoral candidates advance methodologies for modelling, optimisation, uncertainty handling, and integration, each working on a use case at process, microgrid, or network scale. MOBIDIC ultimately delivers a coherent toolbox of solutions to support the design and control of future energy systems. By being trained at the intersection of physics-based modelling, numerical simulation, data analytics, optimisation, and uncertainty-aware decision-making, competences rarely combined in traditional curricula, and by working with industrial datasets and operational constraints, the MOBIDIC cohort will be able to bridge fundamental multiphysics modelling and deployable digital tools, ensuring immediate employability in academia, industrial R&D, and system operations.