Control Aspects of the Charging of a Large Population of EVs

EDIC Candidacy Exam
Exam president: Prof. Mario Paolone
Thesis advisor: Prof. Jean-Yves Le Boudec
Co-examiner: Prof. Matthias Grossglauser

Abstract
Electric vehicles (EVs) are already part of today'€™s reality and their number is expected to grow rapidly in the near future. Such growth will directly impact the grid, for instance higher uncoordinated peak load is possible to occur unexpectedly. In literature, this problem is addressed by controlling the charging of EVs using software agents. The controllability of EV charging power makes it an interesting candidate for demand side management applications. Additionally, it is expected that in the future, EVs will also be able to provide energy to the grid by discharging their batteries, in other words to provide vehicle-to- grid services. Thus, the control of EV charging power is of great importance to optimally manage the EV charging/discharging patterns and provide grid ancillary services. However, existing solutions require precise knowledge about the types of all EVs, their arrival and departure times, and hence are limited by the prediction of this information. Furthermore, presence of volatile power sources in the distirbution electrical grids requires the control to be performed in real-time. Hence, adoption of a powerful online optimization framework to perform fast and efficient operations is crucial to control a large population of EVs charging in distribution grids. In this respect, we investigate the use of online convex optimization techniques. Lastly, in order to efficiently control a large number of EVs, we focus on decentralized multi-agent frameworks, which ensure scalability and low computational complexity. We propose to design a controller for EV charging stations that will be responsible for the coordinated charging of EVs under grid constraints and for advertising the real-time power capabilities and operational preferences of its EVs based on local information.

Background papers
A Scalable Three-Step Approach for Demand Side Management of Plug-in Hybrid Vehicles, by Vandael S., et al.
Optimal Scheduling With Vehicle-to-Grid Regulation Service, by Lin J., et al.
Online Convex Programming and Generalized Infinitesimal Gradient Ascen, by Zinkevich M.