Optimizing the Relocation Operations of Free-Floating Electric Vehicle Sharing Systems

Abstract

Free-floating electric vehicle sharing (FFEVS) systems require nightly relocation and recharging operations to better meet the next day's spatial demand with sufficient battery-levels. Such operations involve not only a crew of drivers to move the shared electric vehicles (EVs), but also a fleet of shuttles to transport those drivers. We consider a decision-making problem for routing shuttles and drivers to recharge and relocate EVs in FFEVS systems. The free-floating nature of FFEVS systems makes their operations costly and cumbersome. Comprehensive studies for relocating EVs and routing shuttles are limited in the literature, and an optimal mix of shuttles and drivers is unknown. We fill this gap by providing a modeling framework for joint decision making and efficient computational tools. We formulate mixed integer programs to model the relocation and recharging operations. Two approaches are devised: sequential and synchronized approaches. In the sequential approach, the movement of EVs is first decided, then the routing of shuttles and drivers is determined. In the synchronized approach, all decisions are made simultaneously. To solve large-scale problems, we devise an efficient computational method, called an exchange-based neighborhood-search method. Our computational method can solve real large-scale instances of car2go in Amsterdam within 10 minutes on a generic computer. Our synchronized approach saves the total shuttle route up to 15% compared to the sequential approach. Wait times of drivers and EVs are also reduced. Our extensive numerical experiments show that when the service area is large, increasing the number of shuttles is more cost efficient than increasing the number of drivers. We also find that when the service area is small, the charging infrastructure is scarce, or the recharging requirements are low, increasing the number of drivers can be more beneficial.