Dynamic optimization of bicycle-sharing systems

Bicycle-sharing systems stations are often imbalanced. Stations are said to be in excess when there is no empty dock for users willing to deposit their bicycle. Conversely, stations aresaidtobeinshortagewhenthereisnoavailablebicycle. Rebalancingcanbeachievedby night when the system is closed and when the users are not permitted to move the bicycles. Insuchcase, rebalancingissaidtobestatic. Whenitisperformedduringthedaywhilethe users are moving the bicycles, rebalancing is said to be dynamic. The present work deals with the dynamic rebalancing of bicycle-sharing systems. Truck-based and pricing-based regulation heuristics are designed and implemented. Their performance is assessed and compared using real instances of up to 30 stations from Washington bicycle-sharing system. The contribution of this study resides in the development of easy-to-implement yet effective and high-impact heuristics.