The practical integration of smart charging on EV fleet charging

Abstract

The goal of this thesis is to analyze the limitations encountered in real-time smart charging during DPMand to integrate them into a smart charging algorithm that schedules EVs. A base case scenario willbe set up to reflect the present dynamics of DPM in a three-phase grid. This scenario will serve as a benchmark for evaluating the smart charging algorithm. The smart charging algorithm itself will employ mixed-integer linear programming (MILP) within a Receding Horizon Optimization (RHO) framework aimed at minimizing charging costs while maintaining a high State of Charge (SOC) for the EVs. To integrate the practical limitations, the grid constraints of a three-phase system will be analyzed and incorporated into the optimization.
Furthermore, a laboratory setup will be constructed to validate the charging behaviors that need to be incorporated into smart charging algorithms and to explore new EV charging behaviors in an experiment. The charging phenomena identified as significant will be integrated into the smart charging algorithm, with a lookup table, to assess their influence on the overall system performance.
The results of this investigation provide valuable insights regarding the charging behaviors and the impact of incorporating these behaviors into a smart charging algorithm. A degradation of the charging efficiency was noted at low charging currents, with the significance of the degradation varying between EV types. Additionally, a voltage discrepancy of the pulse width modulation (PWM) signal of the EVSE was observed, causing an offset between the setpoint of the charging current and the real charging current transmitted by the EVSE.
When these results were incorporated into a smart charging algorithm with a lookup table, clear improvements were seen in the total charged capacity compared to a smart charging algorithm that did not include the lookup table. At a grid capacity of 125 A, the smart charging algorithm with the lookup table provided an increase in charging capacity of 5.23% and an increase of 1.40% for a grid capacity of 175 A. Furthermore, it was observed that the smart charging algorithm with the lookup table could increase the total charged capacity for the low grid capacity of 125 A by 3.97% compared to the charging algorithm that performed immediate charging and DPM.
Both smart charging algorithms, with and without lookup table, show the same reduction in cost compared to the base case algorithm that performs DPM and immediate charging. A decrease in the average charging cost to 4.88% at a grid capacity of 125 A was noted. When the grid capacity is increased, the charging cost could be further decreased to 12.23%, as the increased capacity allows for greater flexibility in scheduling EVs during periods of lowest prices, optimizing the utilization of available power.