Energy Management of Electric Vehicle Supply Equipment in Multi-Unit Residential Buildings

Abstract

As EV adoption continues to grow, the demand for charging infrastructure is increasing, particularly in multi-unit residential buildings (MURBs), which account for a substantial share of urban housing. However, most existing MURBs were not designed to support the additional electrical demand introduced by EV charging, resulting in limitations in capacity. While electrical upgrades could address these challenges, they are often prohibitively expensive. A more affordable and scalable alternative is the deployment of Energy Management Systems (EMS) dedicated to Electric Vehicle Supply Equipment (EVSEs), which can intelligently allocate power and enhance utilization of existing electrical infrastructure without major physical modifications.

To address these challenges across different electrical topologies, a suite of EMS methods is developed for MURBs. For a single-source feeder supplying EVSEs, we develop a Mixed Integer Linear Programming (MILP) based EMS that explicitly models the two-phase connections of EVSEs in a three-phase system to ensure balanced operation between connected phases. A Nonlinear Auto-Regressive with Exogenous Input (NARX) forecasting model is employed to predict non-EV load consumption, enabling real-time EMS operation. To manage the uncertainties introduced by non-EV load demand and stochastic EV availability across multiple sub-feeders, a Distributionally Robust Optimization (DRO) framework with a Wasserstein ambiguity set is developed to determine the optimal charging profiles for the EVSEs under such uncertain conditions. Furthermore, to better capture the hierarchical and multi-tier electrical system of MURBs, a game-theoretic EMS based on the Stackelberg model is developed, where the main panel acts as the leader by setting capacity limits, and each sub-feeder, as a follower, maximizes its own consumption. The proposed EMSs are designed to prioritize non-EV loads and allocate the remaining feeder capacity to active EVSEs while adhering to feeder constraints.

Beyond energy allocation, to mitigate excessive curtailment and ensure fairness among EVSEs, a dynamic Round-Robin (d-RR) scheduling mechanism is proposed that can be lightly integrated with the developed EMS methods. By dynamically changing the time quantum, the d-RR approach prevents starvation and equitably distributes charging opportunities across EVs. This scheduler enhances energy utilization and leads to overall customer satisfaction.

Simulation results confirm that the proposed EMSs effectively curtail EVSE loads during high consumption of non-EV loads, which facilitates the integration of EVSEs into existing MURB infrastructures. Moreover, the results of the d-RR scheduler demonstrate that EVSE penetration can increase and provide fair access to the limited charging capacity among EVSEs, without requiring major electrical system reinforcements.