A Three-Layered System Level Modeling Approach to Electric Transportation.

摘要

Transportation electrification offers solutions to an array of current societal issues, ranging from unstable oil prices to environmental concerns. The promised cost-savings prompt a growing interest to push Electric Vehicles (EVs) into the market. On the other hand, the all-electric-range of current EVs is relatively short compared to gas powered competitors, while the need for longer travel ranges requires development of a network of public fast charging facilities. Hence the envisioned change demands synchronized deployment of new vehicles and infrastructure on a massive scale. However, the required upgrades are very costly and straining the grid beyond its limits could easily lead to cascading failures and outages.;In this thesis we propose a three layered system level modeling approach to electric transportation. First layer (Design) includes the stochastic modeling of a single charging station architecture. Proposed charging station architecture ensures grid reliability at all times while sacrificing to reject small amount of EVs defined as the Quality-of-Service (QoS). First half of this layer explores the system dynamics and solves the optimal energy storage sizing problem respect to QoS targets. Second part of the Design layer examines how the charging station performance is affected both by the energy storage technology used, and the employed charging strategy.;Second Layer (Control) considers control and coordination of customer chargings in a network of fast charging stations. Acknowledging the fact that the non-uniform spatial distribution of EVs creates uneven power demand at each charging facility, the goals of the proposed control mechanism are threefold: (i) avoid straining power grid resources, (ii) increase the percentage of served customers with the same amount of grid resources and consequently maximize the revenue of charging facilities operator, and (iii) provide charging service to customers with a certain level of QoS. We further divide Control layer into two; control for (1) cooperative EV fleets and (2) selfish drivers.;Third Layer (Communications) binds the customers to the charging infrastructure. We explore the communication requirements of EV charging requirements. In order to quantify the communications system performance, we propose a Markov-Modulated Poisson Process based model. Further, we show that as the EV population increases, the network operators will need better communications technology to handle additional demand.;We further extend our modeling approach to resource provisioning in large scale public charging stations. Acknowledging the fact that the current generation capacity could be a bottleneck during the busy hours, we present a capacity planning framework by exploiting the statistical behavior of customers. We modeled the customer demand at each charging slot with an On-Off process. Then, we introduced the concept of "effective power" that is strictly less than the peak power demand during On periods. This notion significantly reduced the required power resources when compared to the capacity planning approach based on peak demand.