A COMPARATIVE STUDY OF CANONICAL NSGA-II AND NSGA-IIHHO FOR POWER DISTRIBUTION CONDUCTOR ALLOCATION

The potential of hybridizing Non-dominated Sorting Genetic Algorithm II (NSGA-II) with Harris Hawks Optimization (HHO) algorithm remains largely unexplored, especially for distribution feeder planning problems; thereby leaving a research gap to be filled. This article therefore investigates the aspects of performance improvement when the NSGA-II is hybridized with HHO algorithm, for multi-objective distribution system planning. Both algorithms (NSGA-II and NSGAII-HHO) were deployed on the IEEE 33-bus feeder with no DG under the same load-flow and cost models, with equal computational budgets. Each was executed independently for 20 runs, and results were collated across three representative loading scenarios, with three conductor types. Performance was evaluated using convergence/divergence metrics, Pareto quality metrics, and statistical significance testing via Mann-Whitney U test. Results indicate that NSGA-II-HHO achieves a 43.59% reduction in IGD+ relative to NSGA-II (p = 0.021, r = 0.43), indicating significantly improved convergence to the true Pareto front. However, runtime was 9.46% higher for the hybrid; with a moderate effect size (r = −0.34) and a meaningful median difference, this increase was however not statistically significant (p = 0.072). All other metrics recorded statistically insignificant improvement i.e., no proven loss in diversity. Smoother convergence and reduced run-to-run variability, particularly around the critical cost/loss knee region, were also observed. The modest rise in runtime and mean conductor capital cost are arguably acceptable trade-offs for the hybrid’s better convergence and reliability.