A Framework for Solving Sequential Charging Facility Location Estimation Problem in Urban Setting

Abstract

The rapid growth of electric vehicles (EVs) necessitates efficient charging infrastructure planning, considering existing facilities. In urban contexts, EV charging times depend on activity durations rather than charging time itself. Considering both effects, this study proposes a sequential, two-step urban EV charger allocation framework. Step 1 uses a modified K-means algorithm to identify candidate locations, incorporating activity locations, participations, and durations. Step 2 employs metamodel-based optimization to allocate charger types and plug counts under setup, operational budget, and power constraints to the candidate locations. Applied to a 10% MATSim Montreal scenario with 74,542 EV users with only 1,392 public chargers, the framework reduced average peak-hour queues by 21% from the benchmark while respecting 60% increases in setup, operational, and power budgets. Results highlight a preference for deploying more slow chargers over fewer fast chargers in this high-demand scenario. Demand elasticity was observed, suggesting the need for improved behavioral modeling.