Machine Learning-based uncertainty quantification in Capacity Expansion Models

Strategic energy system planning is crucial for achieving sustainable and resilient energy systems. This process relies on comprehensive analyses that integrate economic, technological, environmental, and social factors to ensure reliable system development. Mathematical optimization forms the foundation of such planning, with Energy System Optimization Models (ESOMs) widely used for designing transition strategies.Within ESOMs, Capacity Expansion Models (CEMs) determine the cost-optimal mix of generation, storage, and transmission infrastructure to meet future energy demands while considering renewable integration, storage deployment, and grid expansion. Accounting for uncertainty is essential for effective energy planning, as future energy systems operate under evolving technological, economic, and policy conditions that cannot be predicted with certainty. However, traditional uncertainty quantification approaches, such as Monte Carlo simulations, become infeasible for large-scale systems due to the computational burden of repeatedly solving the CEM.