Data Driven Optimization of Charging Station Placement for EV Free Floating Car Sharing

Abstract

Free Floating Car Sharing (FFCS) is a transport paradigm where customers are free to rent and drop cars of a fleet within city limits. In this work we consider the design of a FFCS system based on Electric Vehicles (EVs), We face the problem of finding the minimum number of charging stations and their placement, given the battery constraints of electric cars, the cost of installing the charging network, and the time-varying car usage patterns of customers. Differently from other studies, we base our solution on actual rentals collected from traditional combustion FFCS systems currently in use in two cities. We use about 450 000 actual rentals to characterize the system utilization. We propose a user-behavior model and system policies for the charging events. Then we evaluate via accurate trace driven simulations the performance with different charging station placement policies. We first present greedy solutions, and then perform a local optimization with a meta-heuristic that 1) guarantee system operativeness, i.e., car batteries never get depleted, and 2) minimize users' discomfort, i.e., users are only seldom forced to drop cars in a far-away charging station. Results show that it is possible to guarantee service continuity by installing charging stations in just 6 % of city areas, while 15% of equipped zones guarantee limited impact on users' discomfort.