New Drain-Vortex Optimization tops benchmarks on 34 of 58 CEC 2017 cases

Drain-Vortex Optimization (DVO), introduced by Mohsen Omidi and Brian Vaughan, models candidate solutions as particles moving in a multi-drain vortex field. The update rule decomposes motion into radial attraction toward assigned drain centers and tangential rotation governed by a regularized free-vortex law. A three-phase mechanism switches between far-field exploration, spiral inward movement, and localized core exploitation based on normalized distance. The method also includes adaptive spiral exploitation, population-level basin assignment, and optional stochastic basin switching to maintain diversity.

DVO was rigorously evaluated against PSO, GWO, WOA, SCA, AOA, EO, and SVOA using a calibration-validation protocol with CEC 2022 for tuning and CEC 2017, classical functions, and five engineering problems for out-of-sample testing. On CEC 2017, DVO achieved the best mean log10 error on 34 out of 58 cases and the best Friedman average rank (1.67), significantly outperforming every baseline under Holm-corrected Wilcoxon tests. On CEC 2022, it obtained the best Friedman rank (2.13) and was significantly better than five of seven baselines. The algorithm also comes in a vectorized GPU implementation for parallel runs.