Placement optimisation for station‐free bicycle‐sharing under 1D distribution assumption

Abstract

Bicycles become popular again in the transportation system because they could serve as useful tools for convenient, economical, and environmentally friendly short-trips. In the traditional public bicycle system, customers need to reach a fixed public bicycle station before they can rent a public bicycle. However, the current station-free bicycle-sharing systems can allow customers to borrow or return shared bicycles, which would be distributed almost anywhere. Such new systems bring new optimisation problems for bicycle management. The authors study a placement optimisation problem that highlights the distribution characteristics of shared bicycles, aiming to minimise the total walking distance of customers. The proposed model is a 0–1 mix-integer non-linear programme. To solve the model, they propose a bi-level solving framework. The upper-level model optimises the locations of supply stations. The lower-level model optimises the number of bikes assigned to each demand site. A test based on the campus of Tsinghua University is employed to validate the proposed model. The optimal location suggested by their model is significantly different from the location suggested by models that ignore the distributed feature. Their model performs better in terms of reducing the total walking distance.