Data-driven ferry network design with candidate service arcs: the case of Zhuhai Islands in China

We investigate a ferry network design problem with candidate service arcs (FNDP-SA), which is a new variant motivated by areal-world case from Zhuhai Islands. In this research, the design of a ferry schedule must conform to strict constraints including technical feasibility, safety issues, environmental impacts and moreover, avoid conflicting with ferries from neighboring cities (i.e. HongKong, Shenzhen and Macao). Therefore, a set of cautiously approved service arcs are firstly given, based on which the ferry service network is optimized. This study is among the first to formally describe the FNDP-SA and model it as an integer program. A hybrid variable neighborhood descent (VND)-based algorithm is developed. Two sets of instances are generated based on the case of Zhuhai Islands, where the first set is based on historical ticket sales data while the second set is derived by incorporating increased demands in the future according to a questionnaire survey. Numerical studies have shown that 59% cost reduction on the first set can be achieved by the proposed VND algorithm when compared with manual results. Research outcomes of this study have been adopted and implemented to facilitate the sustainable development of the ferry service of Zhuhai Islands.