2015 Volume 11 Pages 1378-1385
This study aims to consider strategies for public bicycle systems with different demand patterns. We propose operating cost considering a penalty function representing the expected number of shortages of bikes or lockers. Example networks consisting of 10 stations are created to illustrate the proposed model. The model used in this study is Mixed Integer Nonlinear Programming (MINLP) and determines what route the driver chooses in order to minimize repositioning costs. It is conducted with MS Excel 2013 and GAMS solver (SCIP). Demands of each station were generated stochastically at every iteration and normally distributed. This study analyzes five scenarios and shows that solving time is high when most of stations have similar demand. In addition, the more rental and return demands are clearly separated, the lower the standard deviation is. This result implies that an additional algorithm which helps to reduce the fluctuation is needed when demands are similar.
