The electrification of cities, and especially the increase in electric vehicles (EVs), is driving the need to expand the public network of charging points in a sustainable manner. Unlike traditional ‘Greenfield’ planning approaches that assume an empty scenario, this work proposes a non-greenfield planning strategy that explicitly integrates pre-existing charging infrastructure, thereby reflecting real-world urban conditions. To address this, a two-phase hybrid optimization framework is presented, combining a Genetic Algorithm (GA) with a weighted K-Means clustering technique, performing its calculations while taking into account and preserving the existing stations during the optimization phases. In the GA phase, different candidate solutions are generated and evaluated through a customized fitness function, designed to maximize population demand coverage while penalizing excessive or redundant station installations; the individual’s encoding and the design of genetic operators have been modified in such a way that already-installed stations remain fixed throughout the GA optimization. In the second phase, the weighted K-Means algorithm refines the position of the new stations, considering the existing ones as fixed centroids and optimizing the placement of the new points, based on population density. Two case studies show that while the pre-existing network induces a slight increase in the total number of stations compared to a theoretical optimal greenfield solution, the proposed method significantly reduces the number of new installations required to achieve comparable coverage. The proposed method provides urban planners and decision-makers with realistic, sustainable and cost-effective growth plans that are fully compatible with current infrastructure and aligned with long-term urban sustainability goals.
Monteagudo et al. (Thu,) studied this question.