Parallel Metaheuristic-Based Optimization for Electric Vehicle Charging Station Integration and Sizing in Distribution Systems

The large-scale integration of electric vehicles (EVs) has made the siting and sizing of electric vehicle-charging stations (EVCSs) a critical challenge in distribution systems, as inadequate deployment may compromise secure network operation due to voltage and thermal limit violations. This problem is formulated as a mixed-integer nonlinear programming (MINLP) model, where discrete variables define EVCS locations and charging capacities expressed in terms of the number of EVs served. To address this problem, this paper proposes a unified parallel AC-feasible optimization framework that maximizes EV hosting capacity while explicitly enforcing all operational constraints of the distribution system. Particle Swarm Optimization (PSO), a Population-based Continuous Genetic Algorithm (PGA), and Monte Carlo (MC) optimization are evaluated under a common decision-variable encoding, objective function, AC power-flow evaluator, and constraint-handling strategy, enabling a fair comparison among methodologies under identical operating conditions. The proposed framework is assessed on a modified 33-bus distribution system considering a representative weekly operating scenario and 100 independent runs. Results show that PSO achieves the highest hosting capacity, integrating up to 1246 EVs and an average of 1213.3 EVs, compared with 1225 and 1196.1 EVs for PGA and 1077 and 1049.4 EVs for MC, respectively. All methodologies exhibit standard deviations below 3%, confirming robust and repeatable performance, while requiring less than 1428 s on average to identify feasible planning solutions. In addition, the parallel implementation reduces computational times by 42.22%. These results demonstrate the effectiveness of the proposed framework for identifying high-capacity EVCS planning solutions while preserving secure network operation.