Optimal electric vehicle charging station planning in distribution networks using a Pareto-based parallel multi-verse optimizer

This study develops a Pareto-based framework for optimal electric vehicle charging station (EVCS) planning in distribution networks under time-varying conditions. EVCS locations and integer EV allocations are jointly optimized to minimize weekly technical energy losses and maximize EV hosting capacity under nonlinear AC constraints. A Parallel Multi-Verse Optimizer (PMVO) is combined with successive-approximation power flow and a feasibility-first archive, with each candidate evaluated over a 168-h weekly horizon using Colombian demand and multicluster EV charging profiles. The framework is tested on modified 33-bus and 136-bus systems and benchmarked against Monte Carlo (MC), a Population-based Genetic Algorithm (PGA), Hybrid Adaptive Simulated Annealing (HASA), and Parallel Particle Swarm Optimization (PPSO) over 100 independent runs per method. In the 33-bus feeder, the PMVO Pareto approximation spans 210 EVs at 11,601.55 kWh of weekly losses to 1,311 EVs at 13,667.86 kWh, while PMVO attains the highest average compromise-point hosting capacity among the compared methods, with 921.3 EVs. In the 136-bus feeder, PMVO spans 725 EVs at 30,612.05 kWh to 3,930 EVs at 33,583.88 kWh, reaching both the lowest-loss and highest-hosting extremes among the compared fronts. A dedicated 33-bus convergence diagnostic reaches 95% of the terminal best-so-far hypervolume at iteration 353, with 94.12% of the population electrically feasible at termination. Overall, PMVO provides broad access to competitive electrically feasible loss–hosting alternatives, although the results do not support uniform superiority across all methods and Pareto regions.