Optimal deployment of charging infrastructures for urban networks.

Electric Vehicles (EVs) have emerged as a promising solution for fossil fuel diminution, greenhouse gas emissions, and noise pollution in urban areas. Transport electrification poses a necessitous problem for charging infrastructure deployment. In order to promote the gradual increment in market penetration of EVs, more public charging facilities should be located to ease EV users’ range anxiety. However, inappropriate deployment of charging infrastructure may cost a significant amount of resources and potentially pose more congestion on the network as it alters the traffic flow pattern. Despite extensive studies on the EV charging facility, modelling and investigating the impact of travellers’ route choice behaviours on the decision-making process is still in a state of flux. This thesis aims to develop optimization frameworks for deploying EV charging infrastructures, i.e. fast-charging stations (FCS) and wireless-charging lanes (WCL), considering their mutual interaction with route choice behaviours and various realistic traffic considerations. Firstly, the thesis contributions put forward the literature on charging facility location problems by proposing bi-level optimization programs in which the upper-level problem deploys FCS to minimize the system cost or maximize the network performance. Meanwhile, the lower-level problem captures the route choice behaviour of different vehicle classes by solving a traffic assignment problem. The frameworks have been extended to incorporate different traffic settings via different traffic assignments, i.e. User Equilibrium (UE) and Stochastic User Equilibrium (SUE). Besides, a mixed-integer linear framework for WCL deployment has been developed, considering traffic dynamics and congestion under multiple vehicle classes. Specifically, a multi-class Dynamic System Optimum (DSO) model with EVs has been developed to compute an approximate representation of the dynamic traffic flows. Furthermore, various scenarios of driving range limitation and travelling demand have been investigated. Finally, a multi-objective framework for locating WCL has been proposed to support system planners in making decisions based on the trade-off between capital cost and network performance while tracking the energy consumption over the net- work. From the computational perspective, meta-heuristic algorithms combining the Cross- Entropy Method and traffic assignment algorithms (e.g., Frank-Wolfe algorithm and Method of Successive Average) have been proposed to solve complex bi-level optimization problems. Under system optimal route choice, the problem can be formulated as a mixed-integer linear program by incorporating the multi-class DSO traffic assignment into the WCL location problem. The solution then can be obtained by optimized solvers with large-scale capabilities exist. In addition, a constrained optimization approach has been utilized to solve the multi-objective framework. Numerical tests have been carried out to examine the efficacy of the proposed frameworks and provide managerial insights into charging infrastructure planning. The proposed frameworks have demonstrated the ability to capture potential congestion due to the re-routing behaviour of road users and the increasing EV penetration in the network. The thesis concludes with an outlook for potential research directions