An integrated design of hierarchical public transit networks with bike-sharing systems

Abstract

Public transit systems and bike-sharing systems coexist in the urban area. The increasing scale of the modern city results in great demand for long-distance travel in the city, leading passengers to transfer at major transfer stations and making complete trips involving combined modes. The sharing passengers prompt the coexisting transportation networks to interact with each other naturally. Although there have been some studies on the multimodal transportation network design problem, the following research gaps exist. First, the simultaneous design of the feeder bus and shared bike, especially the station location of two feeder services and the bus route, has not been investigated in the previous feeder service design research. Second, very little research distinguishes short-haul and long-haul transit modes from the whole public transit concept. Third, the factors that affect the joint design of metro, feeder buses, and bike-sharing systems have not been identified. The comparisons between the situations that multimodal transportation system operated by public operators or private operators have also not been explored in past studies. Finally, the cooperation strategies of multiple operators in coexisting public transit and bike-sharing networks have not been investigated systemically before.

This study aims to fill the research gaps above. Three studies of the integrated design of hierarchical transit networks coexisting with bike-sharing systems are conducted. The first study introduces a bi-level non-linear mixed-integer programming model to optimize the design of feeder buses and shared bike simultaneously. Under the assumption that both of the two feeder services are operated by the same public operator, the objective of this model is to maximize the improvement of total social welfare. The proposed model is solved by a modified Genetic Algorithm (GA) framework with a fixed-point iteration algorithm. Numerical experiments of small and large realistic networks are conducted to validate the performance of the solution algorithm. The factors that affect the joint design of two types of feeder services are investigated and the comparison between joint design and separate design are presented. The second study extends the first one, including the long-haul transit design of metro station locations and metro lines. A more general multimodal transportation network is considered in this study that covers the whole path of an origin-destination pair. One base model and two derived models are developed to reflect the different operator assumptions. The impact of the addition of shared bikes as a new feeder mode is explored and the comparisons between the design results of public operators and private operators are presented. The last study focuses on the various cooperation strategies of multiple stakeholders in the coexisting networks. Three scenarios of cooperation assumptions are proposed and each scenario has a mathematical formulation, respectively. With the modified GA framework and the modified Non-dominated Sorting Genetic Algorithm II (NSGA-II) framework, the three problems can be solved. Strategies of each scenario are compared through numerical experiments and the ways to select proper cooperation strategies under different assumptions are given.