Optimal service and charging scheduling of electric vehicle demand-responsive connectors considering multiple trips

Abstract

Demand-responsive connectors have emerged as a promising solution to the first- and last-mile transit connectivity challenge, particularly for linking metro stations with surrounding communities. This paper investigates the optimal service and charging scheduling problem of electric vehicle demand-responsive connector (EV-DRC) system, in which each vehicle serving multiple consecutive trips to minimize total cost (i.e., sum of the vehicle deployment and daily operational costs). This multi-trip electric vehicle routing problem (MTEVRP) is first formulated as an arc flow model, and then reformulated via Dantzig-Wolfe decomposition into a set-covering problem that can be solved by column generation. Subsequently, we propose an exact branch-and-price algorithm incorporating tailored acceleration strategies, including route classification, column selection, and interior point stabilization. Extensive computational experiments demonstrate the algorithm’s exceptional effectiveness − it achieves global optimality in 29 out of 30 test instances with up to 60 customer nodes, while reducing the runtime by a maximum of 71% compared to a standard branch-and-price. Sensitivity analyses reveal key insights regarding the impacts of vehicle capacity, battery specifications, and customer time windows on operational performance. These findings provide valuable guidance for operators to design economically viable EV-DRC services that meet high standards of customer satisfaction.