Distribution network reconfiguration—the strategic adjustment of feeder topology through switching actions—offers utilities a low-cost operational lever to reduce losses and enhance service reliability. Traditional approaches often focus narrowly on minimizing active power losses, but this objective may not optimize the overall balance between efficiency and continuity of supply.
A new study addresses this limitation by combining Pareto-based multi-objective optimization with detailed AC power-flow analysis in OpenDSS, a widely used distribution modeling platform. The framework employs NSGA-II evolutionary search to explore non-dominated solutions along the efficiency-reliability frontier, enabling utilities to select configurations that reflect their operational priorities rather than defaulting to a single minimum-loss topology.
The methodology integrates three key elements: an evolutionary search strategy adapted for radial network topology, rigorous AC electrical evaluation including voltage compliance and convergence checks, and explicit feasibility constraints ensuring radiality, connectivity, and grid stability. Rather than inventing new optimization algorithms, the researchers applied proven evolutionary methods with problem-specific adaptations, including a graph-aware branch-exchange operator suited to radial network structure.
Testing on the IEEE 33-node benchmark system revealed that minimum-loss configuration reduces losses by 26%, but a compromise solution achieves nearly identical loss reduction (25.88%) while improving SAIDI—a standard reliability metric—by 21.52%. Critically, evaluation on a real 13.2 kV Colombian distribution network demonstrated even larger gains: 38.88% loss reduction coupled with 23.11% SAIDI improvement and substantial voltage profile enhancement.
These results validate the fundamental premise that single-objective loss minimization may sacrifice reliability benefits. By explicitly mapping the Pareto frontier, utilities can make informed trade-off decisions aligned with their operational strategies, asset conditions, and service quality targets. The reproducible framework also supports planning studies on non-standard networks beyond classic benchmarks, addressing a gap in practical distribution optimization tools.
