Optimal Deployment of Public Charing Infrastructures in Transportation and Power Network

Abstract

Battery Electric Vehicles (BEVs) can reduce the emission of air pollution and greenhouse gas. One difficulty in promoting the usage of PEVs is the availability of charging infrastructures. Currently, the PEVs can be recharged at home and in public charging stations. A recharging method still under development and experiment is called dynamic wireless charging lanes. This method could charge the electric vehicles while driving on the road, so it is more time-efficient than charging stations.This paper represents the first attempt to develop an optimal deployment design of charging lanes and fast charging stations, considering the traffic flow and electricity generation network simultaneously. Since the charging lane is still under experiment, we first designed a stated preference (SP) survey to seek the most attributes to influence BEV drivers' charging choice. Next, given a transportation network and the corresponding power grid, we build a bi-level optimization frame so the user cost of BEV drivers could be minimized. We first assume that one BEV can only use one type of charging method. Based on this assumption, we build a traffic assignment model given the demand and location of charging facilities and combined it with the prediction model based on our stated-preference survey results. Then we design a two-stage solution algorithm to solve the prediction model and the traffic assignment model simultaneously. However, because the two-stage algorithm is too slow to converge, we release the assumption that all the BEVs can only use one charging method. Then we reformulate the lower-level optimization problem. Given the path between each driver’s origin and destination, the objective function is to minimize user costs. The decision variables include charging time at charging stations and charging lanes and the state-of-charge (SOC) at each node. The upper level is to minimize the transportation cost and electricity generation cost. The decision variables include the locations of charging lanes and charging stations and the electricity generated in each bus. We use Karush–Kuhn–Tucker conditions (KKT) to turn the bi-level optimization problem into a single-level optimization, and finally using binary variables to turn it into a mixed-integer optimization problem. A numerical study and a sensitivity analysis are conducted with an integrated transportation-power network.