Optimal Management for Megawatt Level Electric Vehicle Charging Stations With a Grid Interface Based on Modular Multilevel Converter

Abstract

This paper proposes a strategy to manage an electric vehicle charging station (EVCSs) with a grid-side interface based on a Modular Multilevel Converter (MMC). The MMC topology is studied due to its potential for reducing the footprint and the use of active material in the internal distribution system by allowing for transformer-less connection to the medium voltage distribution grid. However, heterogeneous charging demands and arrival-departure profiles of the electric vehicles (EVs) could lead to significant loading unbalances among the MMC arms and among the modules of a single arm. Nevertheless, the current in the grid interface must be kept balanced and sinusoidal. Furthermore, the voltages of the modules of an arm must be balanced. This work combines a load management (LM) algorithm with a power flow management (PFM) algorithm to achieve the required characteristics of grid current and module voltages under the heterogeneity of the charging demand in MMC-based EVCSs. The PFM algorithm controls the circulating currents to compensate the phase-to-phase, arm-to-arm and intra-arm unbalances of the given loading. To minimize the additional losses resulting from active balancing by the PFM, the LM optimizes the charging schedules and allocations of incoming EVs into charging units in order to minimize phase-to-phase and arm-to-arm unbalances in the system. The performance of the proposed optimization-based LM is compared with a rule-based benchmark LM by simulating the daily operation of an example shopping mall parking with MMC-based grid interface. In scenarios with pronounced unbalance limitations, the optimization-based LM increases the supplied energy significantly. Real-time (RT) simulations demonstrate a balanced and sinusoidal grid current profile and balanced module voltages in MMC arms over the daily scenarios. These results indicate that the proposed strategy combining LM and PFM is applicable for real-world deployments.