Service Network Design of Bike Sharing Systems with Resource Constraints

Station-based bike sharing systems provide an inexpensive and flexible supplement to public transportation systems. However, due to spatial and temporal demand variation, stations tend to run full or empty over the course of a day. In order to establish a high service level, that is, a high percentage of users being able to perform their desired trips, it is therefore necessary to redistribute bikes among stations to ensure suitable time-of-day fill levels. As available resources are scarce, the tactical planning level aims to determine efficient master tours periodically executed by redistribution vehicles. We present a service network design formulation for the bike sharing redistribution problem taking into account trip-based user demand and explicitly considering service times for bike pick-up and delivery. We solve the problem using a two-stage MILP-based heuristic and present computational results for small real-world instances. In addition, we evaluate the performance of the master tours for multiple demand scenarios.