Cost-Efficient Allocation of Bikes to Stations in Bike Sharing Systems.

In Bike Sharing Systems, spatio-temporal variation of rentals leads to imbalances in the distribution of bikes causing full or empty stations in the course of a day. Providing a sufficient number of bikes and bike racks is crucial for the viability of these systems. Introducing the notion of service network design, we aim to show the usefulness of tactical planning for bike sharing systems. We design a bike sharing service network considering the suitable aggregation of operational data as well as the anticipation of operational decisions. In particular, we present a mixed-integer programming formulation aiming at cost-efficient allocation of bikes to stations given a predefined service level for different scenarios of bike demand. The scenarios are considered as realizations of typical bike flows between stations in terms of time-dependent origin / destination matrices. Operational relocation decisions are anticipated by a dynamic transportation model. The proposed methodology is exemplified based on two years of operational data from Vienna’s “Citybike Wien”. Computational experiments show how target fill levels vary according to the different scenarios of bike demand. Furthermore, spatio-temporal characteristics of relocation services are derived, which can support operators of bike sharing systems in the planning of relocation services.