UC Berkeley

This dissertation shares a unique perspective to plug-in electric vehicle (PEV) charging infrastructure planning and operations, in which various optimization, control, and data analytic techniques are explored. The critical infrastructure supports to the rapid growth of plug-in electric vehicle adoption is lagging. To efficiently and cost-effectively build and operate a network of charging infrastructure, understandings toward the characteristics both at a single station level as well as at the complex network level are crucial to a societal success. This dissertation shares perspectives and advances knowledge to fill this gap.

First we emphasize the poorly understood and often neglected issue, the \textit{overstay} problem at a single charging station level. The overstay problem is described when a PEV continues to occupy a charger even after its charging session has been completed. It significantly hinders the utilization of the charging infrastructure, leading to wasteful resources, disappointing customer satisfaction, and discounted revenue return. This motivates a strategy for increasing utilization by interchanging fully charged PEVs with those waiting for service. An interchange mechanism is defined and the planning and operation models are developed to account for the phenomena. Numerical experiments are conducted to illustrate the performance and demonstrate decreased planned chargers yet increased economic benefits.

Secondly, we seek to further improve the charging station efficiency through optimal operation strategies. This work stands out from the others by acknowledging and incorporating human users decision process. Human factor is a crucial element in the decision loop and cannot be forced. We achieve the control strategies by nudging users' choices with time flexibility and monetary incentives. The formal process will be defined in the corresponding chapter. The overall control framework is evaluated with three metrics, (i) net profits, (ii) overstay duration, and (iii) number of sessions served. Furthermore, this work served as proof-of-concept and has enabled multiple real life hardware testbeds, including the parking lots on the UC Berkeley and UC San Diego campuses. Pricing experiments have been conducted and the data will be shared to advance community understanding of PEV drivers' sensitivity to charging flexibility and prices.

Thirdly, the scope is enlarged to consider a network of charging infrastructure. The context is electric trucking logistics with cargo movement. We propose an innovative modeling perspective to consider the non-cooperative nature between charging service provider and the fleet operator(s). Often, it is assumed a powerful system planner can control and management all assets. We, on the contrary, highlight the necessity to consider different entities within the context of transportation electrification.

The final phase proposes a computationally efficient and scalable framework to size the ride hailing fleet, manage it at large-scale, design and match the charging infrastructure. Contrary to current market trends, the results of this work reveal that neither large-battery-size AEVs nor high-power charging infrastructure is necessary to achieve efficient service. This effectively alleviates financial and operational burdens on fleet operators and power systems. Furthermore, strategic fleet management results in low mileage, reducing emissions detrimental to human health. Finally, the reduced travel time and emissions resulting from efficient fleet management create an economic value that exceeds the total capital investment and operational costs of fleet services. The associative policy implications are also revealed in the chapter.

In summary, this dissertation shares the unique perspectives and tackles the often poorly understood problems from practice. The chapters are organized in sequence from planning phase to operation phase, and from single atomic scale to large network level. Yet, each chapter is self-contained. Hence, readers may jump to any chapter depending on their interests and needs.