Bicycle Allocation Methods Within Public Bicycle-Sharing Systems

This paper is a study of the paper of J. Shu et al.[2], which discusses the development of operational research models that simplify decision making in the management and design of public bicycle-sharing systems.. To begin with, the results of a simple example from the paper of J. Shu et al.[2] will be reproduced. This is a three-station example, for which the time-average level of bicycles that are traveling in between stations is calculated both by implementing a theoretical analysis, as well as running a simulation. The results of the two different techniques will be compared, and turn out to be quite similar. To obtain better insight in the results, the time-average amount of bicycles at each station and the percentage of customers that is lost is calculated. The simple example is extended by taking into account arbitrary travel times, as a result of which the theoretical model and the simulation need to be adapted. In the second part of the paper, two larger and more complex problems are discussed. In this way, we can test if the simulation model also works for problems other than the simple example. The results of the simulation and the theoretical model for the 4-node example with steady travel rates and travel time excluded, show on average a 33.8% difference in the average amount of bicycles moving in between stations for all combination of stations, and the 6-node example shows a 10.3% difference. But we cannot jump to the conclusion that the theoretical model is not accurate, moreover we will see that the ratios of the bicycles leaving from station i traveling to station j divided by total bicycles number of leaving from station i are the same for the simulation and the deterministic model for all stations.