EV4EU is proud to announce a new research publication by our consortium partners ΔΕΔΔΗΕ Α.Ε. – HEDNO S.A. and DTU – Technical University of Denmark, appearing in the Sustainable Cities and Society journal. The paper is titled:”Bayesian Optimization of Electric Vehicle Charging Infrastructure Location Planning: An Agent-Based Simulation Approach” and addresses a central challenge for the widespread adoption of electric vehicles: the efficient deployment of charging infrastructure in urban areas. The authors propose a Bayesian Optimisation framework integrated with Multi-Agent Transport Simulation (MATSim) to identify optimal EV charger locations, balancing accessibility and installation costs while accounting for realistic travel and charging behaviours.

MATSim is extended with a probabilistic charging demand model, incorporating plug-in probabilities, initial State of Charge distributions, and charging durations constrained by parking regulations. This high-resolution simulation allows for detailed analysis of individual travel choices, network effects, and interactions between mobility patterns and charging infrastructure.

A case study conducted in Frederiksberg, Denmark, with over 60,000 EV users and nearly 75,000 daily trips, demonstrates the effectiveness of the approach. Results show that public charging infrastructure plays a crucial role in dense urban areas, accounting for 65% of all charging sessions and 68% of delivered energy even at a 10% EV penetration rate.

The study also demonstrates the advantages of Bayesian Optimisation over traditional black-box approaches, achieving 22.4% higher welfare while requiring only one-third of the simulation runs, thereby reducing computational effort. This research provides valuable insights for urban planners and policymakers, offering a robust framework for data-driven decisions on EV charging infrastructure deployment.

Read the full paper here