P-D pair based EV-sharing vehicle relocation problem with route familiarity

Abstract

This paper focuses on the electric vehicle (EV) relocation problem with dispatcher-specific times (E-VRePDST) in one-way station-based EV-sharing systems. An innovative modeling method based on the p-d pairs is developed, and the dispatcher-specific travel time is adopted to model the dispatchers’ route familiarity. Thus, preconditions and time windows for every dispatcher to traverse every pair of pickup and delivery node (p-d pair) are solved out. Then, the p-d pair based model is constructed. The proposed modeling method efficiently reduces the complexity of the associated constraints for E-VRePDST, and the numerical experiments indicate that the p-d pair based model is more efficient in terms of the solution quality and run time than the existing models based on separately treated pickup and delivery nodes. Next, a large neighborhood search (LNS) algorithm is designed and proved to be efficient compared with the results by using CPLEX. Finally, a case study of the EV-sharing system is conducted, and it shows that each dispatcher is distributed to the route with relatively more requests located in the familiar stations.