A Two-Phase Framework with Loop Reduction and Meta-Heuristic Evaluation for Distribution Network Reconfiguration

This paper presents a novel two-phase framework for solving the Distribution Network Reconfiguration (DNR) problem by combining loop-based search-space reduction with Meta-Heuristic (MH) algorithms. In Phase #1, fundamental loops are identified using Dijkstra's algorithm, and duplicated switches are eliminated through a winner–loser rule, which drastically reduces the search space while preserving feasible topologies. In Phase #2, multiple MH algorithms are simultaneously applied to the reduced search space, allowing an efficient evaluation of their suitability for DNR. The proposed approach was validated on three benchmark systems: IEEE 33-bus, IEEE 69-bus, and IEEE 118-bus distribution networks. In all cases, the method consistently identified optimal network configurations with active power loss reductions of up to 36.4%, 32.8%, and 28.9% for the IEEE 33-bus, IEEE 69-bus, and IEEE 118-bus systems, respectively, while improving the minimum bus voltage by 2–6% compared to the original configurations. The results also reveal algorithm-specific behaviors, highlighting that some MH algorithms (e.g., SADE, JADE, HS, GWO) perform robustly across all network sizes, while others are more problem-dependent. Importantly, the loop-reduction step leads to a substantial reduction of the search space, thereby facilitating faster convergence and enhancing scalability. These findings confirm that the proposed framework provides a practical, flexible, and effective solution to the DNR problem in both small-scale and large-scale distribution networks.