A Markov Simulation Approach to Balancing Bike-Sharing Systems

This research presents a simulation-based approach to establishing stations for an urban bike-sharing system. Movements of bicycles from one station to another are simulated via a Markov process. During the simulation, some stations will lose bicycles, while other stations will gain bicycles. This simulation process is tantamount to the transient phase of a Markov process. Sparsely-used stations are closed, resulting in increased system utilization. The simulation is applied to several test problems. Experimentation shows 86.53% average system utilization, with a standard deviation of 6.81%. It was also discovered that more bikes in the system requires more simulation time, along with more sensitivity to minimum utilization thresholds.