Engineers from the Technical University of Denmark have successfully tested an intelligent mass management system for electric vehicle charging. The newly developed architecture allowed the charging station to independently coordinate vehicle connections, avoiding grid overload and reducing electricity costs.
Unlike conventional centralized control, where all decisions are made by a single controller, the new system uses a two-level control scheme. On the upper level, a special program analyzes tariff dynamics, predicts renewable energy generation and ensures that the station’s overall consumption does not exceed a set limit. Its chief goal is to purchase electricity during the cheapest hours and use it to charge vehicles. The lower level is responsible for distributing this power among the connected electric vehicles. The system assigns a priority to each vehicle based on the amount of energy required and the expected departure time. The higher the priority, the faster the car will recharge. If a higher-priority electric vehicle gets connected to the system, the power of the vehicles that are already being recharged will be temporarily reduced to free up capacity.
Testing took place at the Risø Campus near Copenhagen, where six charging stations with 12 connectors, 11 kilowatts each, and a 20-kilowatt solar panel were installed. For two days, the system operated with a limited capacity of 215 kilowatt-hours (kWh) per day, at roughly half of its technical capacity. The electric vehicles of employees would get connected at different times, and the program would recalculate the optimal energy distribution plan every five minutes. This method is called rolling-horizon optimization: the algorithm predicts parameters six hours ahead but only implements the most immediate decisions, constantly refining its calculations when new price and generation data become available.
The results were clear. On the first day, electric vehicles received the entire requested amount of energy: 110 kWh out of 110 kWh. On the second day, despite an increase in the number of connections, they only received 112 kWh out of 143 kWh. That day, one electric vehicle was unable to receive a command due to a short-term failure on its end. As a result, the other vehicles went into sleep mode. This demonstrated the need to automatically exclude faulty devices from general distribution.
For two days, the station would strictly adhere to grid limits and automatically adjust to price dynamics: power increased during low-rate hours and decreased during high-rate hours. A fascinating compromise between the two control levels was also observed: the lower level always maintains a minimum amount of power (2.1 kilowatts per phase) to prevent the vehicle from shutting down, even if the upper level is currently limiting the overall load. While this can cause short-term overshoots, it also prevents interruptions in charging and improves system stability.
The developers also identified other issues arising in real-world conditions, such as short-term communication errors, delays in starting sessions and the incompatibility of some electric vehicles with data exchange protocols. These observations helped refine the algorithms: expanding the range of power reduction for low-priority vehicles, adding protection against switching to sleep mode during short interruptions and improving diagnostics of faulty connections.
In the end, the experiment achieved its goal. The distributed hierarchical system demonstrated stability even with limited resources, proving its ability to perform multiple tasks simultaneously: load smoothing, accounting for price fluctuations and ensuring equitable energy distribution. This technology could potentially be scaled to large parking lots, office centers and residential areas, and its basic principles could be applied to two-way energy exchange systems (vehicle-to-grid), where electric vehicles could not only consume energy but also return their accumulated charge to the grid.
