In order to meet the users' demand, bike sharing systems must be regularly rebalanced. The problem of balancing bike sharing systems (BBSS) is concerned with designing optimal tours and operating instructions for relocating bikes among stations to maximally comply with the expected future bike demands. In this paper, we tackle the BBSS by means of Constraint Programming: first, we introduce two novel constraint models for the BBSS including a smart branching strategy that focusses on the most promising routes. Second, in order to speed-up the search process, we incorporate both models in a Large Neighborhood Search (LNS) approach that is adapted to the respective CP model. Third, we perform a computational evaluation on instances based on real-world data, where we see that the LNS approach outperforms the Branch & Bound approach and is competitive with other existing approaches.
