The New Jersey Institute of Technology's
Electronic Theses & Dissertations Project

Title:	Optimizing feeder bus network based on access mode shifts
Author:	Huang, Zhaodong
View Online:	njit-etd2012-092 (xv, 127 pages ~ 3.0 MB pdf)
Department:	Department of Civil and Environmental Engineering
Degree:	Doctor of Philosophy
Program:	Transportation
Document Type:	Dissertation
Advisory Committee:	Liu, RongFang (Committee chair) Bladikas, Athanassios K. (Committee member) Chien, I-Jy Steven (Committee member) Daniel, Janice Rhoda (Committee member) Yang, Jian (Committee member)
Date:	2012-08
Keywords:	Feeder bus network Elastic demand Network optimization Tabu search Grid street network Service frequency
Availability:	Unrestricted
Abstract:	The methodology introduced in this dissertation is to optimally find a feeder bus network in a suburban area for an existing rail system that connects the suburban area with the Central Business District (CBD). The objective is to minimize the total cost, including user and supplier costs. Three major access modes (walk, feeder bus, and auto) for the rail station are considered and the cost for all modes makes up the user cost. The supplier cost comes from the operating cost of the feeder bus network. The decision variables include the structure of the feeder bus network, service frequencies, and bus stop locations. The developed methodology consists of four components, including a Preparation Procedure (PP), Initial Solution Generation Procedure (ISGP), Network Features Determination Procedure (NFDP) and Solution Search Procedure (SSP). PP is used to perform a preliminary processing on the input data set. An initial solution that will be used in SSP is found in ISGP. The NFDP is a module to determine the network related features such as service frequency, mode split, stop selections and locations. A logit-based Multinomial Logit-Proportional Model (MNL-PM) model is proposed to estimate the mode shares of walk, bus and auto. A metaheuristic Tabu Search (TS) method is developed to find the optimal solution for the methodology. In the computational experiments, an Exhaustive Search (ES) method is designed and tested to validate the effectiveness of the proposed methodology. The results of networks of different sizes are presented and sensitivity analyses are performed to investigate the impacts of various model parameters (e.g., fleet size, parking fee, bus fare, etc.).