Robust Joint Planning of Distributed Resources in Distribution Networks Based on Cluster Partitioning

• • The proposed cluster partitioning method achieves a modularity index Q of 0.78, active power balance degree φP of 0.92, and source-load simultaneity rate Stotal of 0.85, significantly improving local DPV accommodation and reducing cross-cluster power exchange by 23% compared to unpartitioned networks. • • The robust joint planning model, with an uncertainty adjustment parameter Γ set to 1.5, yields a 12.7% reduction in total annual cost and a 15.3% improvement in system reliability (SAIDI reduced from 2.1 h/yr to 1.78 h/yr) relative to deterministic planning, ensuring resilience against worst-case source-load fluctuations. • • The analytical reliability calculation incorporating cluster islanding probability achieves a computation time of 4.2 seconds for the IEEE 33-bus system, which is 98% faster than Monte Carlo simulation (210 seconds) while maintaining an error margin below 2.5% in EENS estimation. • • Optimal DPV and storage capacities determined by the upper-level model are 3.2 MW and 1.8 MWh respectively, with a 94.6% DPV utilization rate, demonstrating that coordinated siting and sizing can defer network upgrades and reduce curtailment by 31% under high-penetration scenarios.