Dynamic Electric Vehicle Fleets Management Problem for Multi-Service Platforms with Integrated Ride-Hailing, on-Time Delivery, and Vehicle-to-Grid Services

Abstract

The rapid expansion of electric vehicle (EV) adoption, coupled with the growing demand for mobility services, calls for solutions to manage EV fleet operations efficiently. More importantly, mobility service vehicles often remain idle for extended periods due to fluctuations in mobility demand. In this context, this paper examines a multi-service platform that dynamically coordinates ride-hailing, on-time delivery, and vehicle-to-grid (V2G) based energy services. The proposed multi-service coordination offers significant benefits, including improved resource utilization, reduced operational costs, and increased profits by leveraging synergies across multiple services. On arriving at the multiservice platform, users submit (heterogeneous) service requests in terms of origin, destination, service time window, the number of riders and/or the weight of goods, etc. To meet users' heterogeneous service requests in real-time, we propose a dynamic multi-service electric vehicle fleet management (MEFM) problem to optimize the platform's operational strategies, including the allocation, routing, and scheduling of EV fleets, aiming to maximize the platform's profits in each time period. We formulate the MEFM problem as an arc-based mixed-integer quadratically constrained programming (MIQCP) model. Furthermore, we develop a customized branch-and-price-and-cut (B&P&C) algorithm to solve the proposed dynamic MEFM problem effectively. The proposed B&P&C algorithm integrates Dantzig-Wolfe decomposition where the master problem is further strengthened with subset row cuts and a novel labelling sub-algorithm that can capture multi-service coordination, fleet capacity, and battery-level constraints with recharging flexibility. Numerical experiments conducted in the context of Shenzhen, China validate the efficiency and effectiveness of our proposed algorithm, achieving computation speeds 309.87 times faster than the state-of-the-art commercial solver (Gurobi) on average (with speed-ups ranging from 18.34 to 715.50 times), and consistently obtaining optimal solutions for large-scale instances where the commercial solver (Gurobi) fails. The numerical results further highlight the benefits of integrating on-time delivery and V2G based energy services into the platform, e.g., while operational costs show a slight increase, the corresponding surge in profits indicates the potential for significant economic returns.

Keywords: Dynamic electric vehicle fleets management, Multi-service ride-hailing platforms, Vehicle-to-grid