A Multi-Stage Optimization Framework for Battery Swapping in Urban E-Bike Sharing Systems

Abstract

Shared electric bicycles (e-bikes) are increasingly central to urban transportation, providing a faster, more accessible alternative to conventional bicycles for short trips. As their use grows, maintaining sufficient battery levels across large fleets becomes critical. Among battery management strategies, battery swapping, which replaces depleted batteries with fully charged ones, offers a scalable solution, especially for dockless systems that lack fixed charging stations. However, it presents logistical challenges related to vehicle routing, battery delivery, and capacity constraints. This thesis introduces an optimization framework that combines dynamic clustering with a dual-objective vehicle routing model. E-bikes requiring service are grouped into van-feasible clusters based on battery level, spatial proximity, and fleet availability. Each cluster is then optimized to minimize travel distance while maximizing economic return. The framework is applied to real-world data from San Francisco’s Bay Wheels system, accessed via the General Bikeshare Feed Specification (GBFS). Results show that the proposed method reduces travel distance and enhances the efficiency of battery-swapping operations.