Tabu search for solving multiple-vehicle bike sharing system routing problem with real port distribution

The bicycle sharing system (BSS) is one of the eco-friendly transportation systems introduced in major cities world-wide. However, BSS faces one crucial challenge: an imbalance of bicycles among the ports, because users are free to rent and return bicycles freely. It is necessary to restore the bicycles among the ports using vehicles to overcome this issue. We have already proposed search strategies for finding the tours of the vehicles in a reasonable time frame. In our previously proposed method, we used the tabu search which harnesses executing two local searches: the CROSS-exchange and the Or-opt. In the numerical experiment, we used instances with uniformly distributed ports to evaluate the performance of our previous method; however, real BSS ports are not uniformly distributed. Thus, in this paper, we use instances with real port distributions from four BSSs (Boston, Washington D.C., New York, Chicago) to investigate whether our method is effective for non-randomly distributed ports of larger-size instances (200 ports). From the numerical experiments, our proposed method is also effective for instances based on real BSS port locations.