Dynamic bike redistribution strategies considering on-site demand patterns for bike sharing systems

More and more cities have deployed their own bike sharing systems in recent years, which enable seamless door-to-door public transportation and foster the trend of sustainable urban development. One of the critical issues for system operation is the imbalance between demand and supply due to the spatiotemporal dynamics of travel patterns, underlain by the urban activity system over the service areas. This results in the situation that users find no available on-site bikes/docks when they want to rent/return bikes at some popular stations during certain time periods, and in the long run it can lower user satisfaction toward a system. Hence, the bike redistribution problem is developed for bike sharing systems, aiming at avoiding/reducing the unmet demand of users to rent or return bikes. In practice, bike redistribution is conducted throughout daily operation hours and particularly needed for the busiest ones, generally using trucks to transport bikes from the stations with surplus bikes to the stations lacking them. This research proposes a model for dynamic bike redistribution which accounts for the demand of both renting and returning bikes across stations. It seeks to enable system operators to promptly respond to the current on-site demand-supply conditions or even to prevent potential imbalance. Given estimated user dissatisfaction on station, the proposed model determines the numbers of bikes to be picked up from and reloaded to each station. Additionally, the model also leverages the routing of the trucks used for redistribution to attain the minimum operational cost.