Please use this identifier to cite or link to this item: http://hdl.handle.net/10397/83771
Title: Advanced models for transit network design and operation under uncertainties
Authors: Zhang, Yuqing
Degree: M.Phil.
Issue Date: 2011
Abstract: Transit network uncertainties commonly exist in transit systems due to day-to-day and within-day demand variation, congestion, adverse weather, and road incidents. Typical phenomena include frequency instability at downstream stops such as vehicle bunching, unexpectedly lengthy passenger waiting times, and over-crowded services followed by empty runs. Under such circumstances, the interests of both transit passengers and service suppliers are affected. Transit passengers may be subjected to travel time unreliability, and service suppliers may be subjected to profit fluctuation and poorly specified levels of service. Hence, uncertainty is an inevitable aspect of transit planning, especially affecting passenger flow prediction and transit network design. In this study, transit network uncertainties have been examined from the perspectives of both transit passengers and suppliers. Two new dynamic transit network assignment models have been developed to reflect passenger reaction to network uncertainties. The first is a single-class reliability-based transit assignment model, developed to reflect risk-averse passenger travel decisions as to departure time and route choices. In order to account for travel time reliability, passenger effective travel time, which includes average travel time plus a safety margin to cope with uncertainty, is adopted as dis-utility function. This model is formulated as a fixed point problem which can be solved by a heuristic solution algorithm. A numerical example shows the existence of service deviations under transit network equilibrium conditions, such as vehicle bunching and overtaking. The second is a new multi-class reliability-based transit assignment model. A safety margin is differentiated for different passenger classes, as passengers have different risk-taking attitudes towards random generalized travel costs (including both travel times and monetary costs). Network congestion is also reflected in this model by introducing an overload parameter on vehicle design capacity constraint on random passenger boarding demand. A network example connecting the Kowloon area to Hong Kong International Airport illustrates the ability of this model to demonstrate that different passenger risk-taking attitudes greatly impact passenger route and departure time choices, and subsequently both monetary and time costs.
With taking account of the effects of network uncertainties, another two transit line scheduling models have been developed to serve a transit supplier's different planning purposes. Of concern to the transit authority, social welfare aspects, such as passenger travel time efficiency and reliability, are improved by a proposed line scheduling model. The interaction between service supply and passenger behavior response is reflected in the bi-level formulation of the scheduling problem. The upper level of the model optimizes the integrated transit service attributes, while the lower level predicts passenger travel decisions under transit network uncertainties. The bi-level problem is solved by applying the genetic algorithm (GA). The numerical results show that transit service reliability under network uncertainties can be improved by the adjustment of line schedules, without the need for extra vehicle resources. Of relevant to the private transit operator, a new transit line scheduling model has been proposed to reflect the competition between operators in the deregulated transit market under conditions of network uncertainty. The operator's profit is considerably affected by service irregularity, as well as the passenger's response to the irregular service. The operator's risk preference determines how the variability of random profit is measured. Thus, the objective of the operator in the transit line scheduling model is to maximize the α-confident profit, defined as the stochastic profit within a confidence threshold. The passenger's response to the change of line schedules is formulated as a reliability-based user stochastic equilibrium (RSUE) constraint. The α-confident profit maximization model is formulated as a variational inequality (VI) problem and solved by an adapted diagonalization algorithm. This model shows that the ignorance of network uncertainties and operator risk preferences can result in over-optimism on profit when developing the transit line schedules.
Subjects: Urban transportation -- Planning.
Local transit.
Hong Kong Polytechnic University -- Dissertations
Pages: xxiv, 128 leaves : ill. ; 30 cm.
Appears in Collections:Thesis

Show full item record

Page views

47
Last Week
0
Last month
Citations as of Mar 24, 2024

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.