The increasing penetration of renewable energy sources makes power system expansion planning with storage strongly dependent on chronological operational conditions. Time-series aggregation (TSA) can reduce the computational burden of full-year planning models, but it distorts the temporal information that determines storage operation and investment decisions. This thesis investigates how SOC-based diagnostic information can improve representative-day-based expansion planning for storage-embedded power systems. The planning model jointly considers transmission expansion, wind investment, and energy storage sizing. Representative days are selected through hierarchical clustering with preserved extreme days, while sequentially linked days are used to reconstruct inter-day chronology. The framework aims to find investment decisions that achieve lower total cost under full chronological evaluation.

Reduced models with different numbers of representative days are evaluated, and their reconstructed SOC trajectories are analyzed. SOC-based diagnostic metrics are developed to characterize the components of the trajectory gap. These metrics are used as diagnostic signals for identifying where the current temporal representation is insufficient for storage-related operation.

The results show that daily-cycling storage is mainly affected by intra-day shape mismatch, while long-duration storage is more sensitive to accumulated drift and inventory-level bias. Natural days with large intra-day shape mismatch are useful feedback candidates because they reveal inadequacy in the current representative-day set for describing storage charging and discharging patterns. Preserving these day-shape-critical days improves storage-relevant temporal representation and can lead to better investment decisions. ... Read more