A PILOT/VND/GRASP Hybrid for the Static Balancing of Public Bicycle Sharing Systems

Due to varying user demands in bicycle sharing systems, operators need to actively shift bikes between stations by a fleet of vehicles. We address the problem of finding efficient vehicle tours by an extended version of an iterated greedy construction heuristic following the concept of the PILOT method and GRASP and applying a variable neighborhood descend (VND) as local improvement. Computational results on benchmark instances derived from the real-world scenario in Vienna with up to 700 stations indicate that our PILOT/GRASP hybrid especially scales significantly better to very large instances than a previously proposed variable neighborhood search (VNS) approach. Applying only one iteration, the PILOT construction heuristic followed by the VND provides good solutions very quickly, which can be potentially useful for urgent requests.