Time aggregated method based on clustering applied to energy scenario optimization

The transition towards decarbonized energy systems significantly increases the complexity of power grids, making long-term simulations computationally intensive. Specifically, the COMESE framework utilizes a Differential Evolution (DE) optimization algorithm that evaluates parallel populations of scenarios, leading to a very high computational cost when simulating a full 8760-hour year. The objective of this work is the structural integration of a Time-Series Clustering module within the COMESE model to drastically reduce simulation times. Utilizing a Partitioning Around Medoids (K-Medoids) algorithm, the calendar year is synthesized into a limited set of representative periods (clusters). This novel architecture allows the computational engine to simulate exclusively these reduced timeframes while preserving the original dispatch logic. Finally, the outputs are reaggregated to accurately project the annual behavior of the system, unlocking unprecedented computational efficiency without compromising the validity of the results.