Currently, the massive integration of electric vehicles (EVs) into the global mobility market has generated a growing and urgent need for the adequate planning, integration, and sizing of Electric Vehicle Charging Stations (EVCSs) within Distribution Systems (DSs). The primary objective of this integration is to maximize the available charging capacity while preserving the secure and reliable operation of the electrical network. However, the large-scale deployment of EVCS introduces a substantial increase in power demand, which can significantly affect grid performance. In particular, higher charging penetration levels intensify operational stresses on the distribution infrastructure, leading to potential violations of voltage regulation limits, thermal constraints of distribution lines, among others.
From a modeling perspective, the integration and sizing of EVCS constitute a highly complex optimization problem. This complexity arises from the nonlinear nature of power flow equations governing distribution systems, as well as from the coexistence of discrete decision variables, associated with the location of EVCS within electrical system, and continuous decision variables, related to their charging capacities. As a result, the problem is naturally formulated as a Mixed-Integer Nonlinear Programming (MINLP) problem, which is known to be computationally challenging, especially for large-scale networks and realistic operating scenarios.
To effectively address this challenge, it is essential to develop an accurate mathematical model that captures both the electrical behavior of the distribution system and the operational characteristics of EVCS. Moreover, the solution of such a model requires the use of high-efficiency optimization techniques capable of handling nonlinearity, mixed decision variables, and large solution spaces. In this context, this work proposes a comprehensive optimization framework for the integration and sizing of EVCS in distribution systems, with the objective of maximizing the total charging capacity while strictly satisfying all operational constraints of the DS under EVCS-intensive conditions.
As solution strategies, parallel implementations of Particle Swarm Optimization (PSO), a Population-based Continuous Genetic Algorithm (PGA), and Monte Carlo methods are employed. These methodologies are evaluated using a benchmark 33-bus distribution system, considering realistic variability in both photovoltaic (PV) generation and load demand. Each algorithm is executed 100 independent times to assess solution quality, robustness, repeatability, and computational efficiency, enabling a fair comparison and the identification of the most suitable approach for solving the proposed optimization problem.
