Static Repositioning Operation in Bicycle-Sharing Systems

There has been a rapid growth of bicycle sharing systems around the world. As a bicycle sharing system is a relatively newer transportation system, it faces many issues related to the planning and operation of the system. One of the most important operational issues is the issue of the system imbalance. This issue is very critical in determining the success of the system. This is due to the fact that users may not be able to either pick up or drop off bicycles at their desired station if the system is imbalanced (i.e. some stations are empty while some others are full). Moreover, the repositioning operation constitutes a major portion of the total operating costs of a bicycle sharing system. Developing methods by which the agencies can rebalance the system in a low-cost and efficient manner would lead to huge cost savings for the agencies. This thesis focuses on the static repositioning operation in a bicycle sharing system. It proposes a method of modeling the demand for withdrawing or returning bicycles at a station as a Markov process. The estimated penalty values for the stations are calculated for the different starting inventory levels. A Non-Linear Program is formulated to calculate the optimal inventory levels at the stations which minimize the total estimated penalty incurred by the entire system and also conserve the total number of bicycles in the system before and after the repositioning operation. Then, given the starting inventory levels at the stations and the desired inventory levels, a model is proposed for carrying out the repositioning activity in a specific time deadline which minimizes the cost of the entire operation. The model is formulated as a Mixed Integer Linear Program which can be solved to optimality for small networks. To solve the problem for real world networks, a heuristic algorithm based on clustering and the Simulated Annealing algorithm is proposed. The algorithm estimates the optimal solution for the real world networks in a reasonable computational time. The results show that by using the proposed approach instead of the traditional methods of repositioning, the level of service of the system can be improved and considerable cost savings can be obtained by the agencies