A Dynamic Inventory Routing Approach for the Public Bike Sharing Rebalancing Problem: A Case Study of Seoul

Seoul’s public bicycle system faces operational inefficiencies due to an imbalance in availability, resulting to bicycle shortages or surpluses at specific times and locations. The current redistribution method is reactive and based on human operators’ experience, and it tends to fail to proactively address real-time demand. This study proposes a data-driven, proactive redistribution system. We first develop a prediction model for the number of bikes at the stations, then propose “acceptable inventory levels” for bike stations based on time-dependent demand patterns, and finally design a three-phase heuristic algorithm, called “Clustering-Routing-Relocation,” to minimize deviations from the predefined inventory levels and total travel time. We also present a proof-of-concept mobile app that implements the proposed algorithm to enable staff to perform rebalancing duties. Experimental validation using real operational data showed that the proposed process may increase the inventory-level satisfaction rate by more than 10% compared with the existing method. This study demonstrates that a proactive, data-driven approach can significantly improve operational efficiency and user satisfaction.