Joint design of multimodal transit networks and shared autonomous mobility fleets☆
Introduction
Designing public transit networks to efficiently serve heterogeneous travelers subject to various constraints (e.g. budgetary, political, equity) is a challenging problem for transit agencies. Residents of low-density areas and employees working outside of central business districts tend to experience low-quality transit service (e.g. long walk distances to or from a transit stop, long wait times due to infrequent transit service, circuitous routes, and multi-transfer trips) due to the difficulty of serving these areas in a cost-effective manner.
Autonomous vehicles (AVs) and their inclusion in mobility service fleets may offer transit agencies a potential solution to this problem. Agencies could replace inefficient transit routes/patterns operating in certain regions, and during certain times of the day, and reallocate those resources to operate (or subsidize) shared-use AV mobility services (SAMSs). Given the considerable operational and capital cost advantages of SAMSs over fixed-route transit services and driver-operated mobility services (e.g. flex-transit, ridesourcing, and taxi service), it is conceivable that reallocating resources from less cost-efficient transit patterns may produce better service for travelers and/or reduce overall transit agency costs. This study uses an agent-based modeling approach that captures the system-level impacts of redesigning transit networks and operating (or subsidizing) SAMSs taking into account traveler behavior and the interactions between traveler agents in the system. In their cost-based analysis of future transport systems, Bösch et al. (2018) indicate that there is an opportunity for SAMSs to replace bus service in low-density areas and allow transit agencies to focus their resources on mass transit in dense urban areas.
To help achieve the Pareto-improving outcome of better service for travelers and lower costs for agencies, this study presents a modeling framework to optimize the joint design of transit networks and SAMS fleets. Specifically, the objective is to formulate and solve the joint transit network redesign and SAMS fleet size determination problem (JTNR-SFSDP) subject to user-equilibrium constraints at the mode and route choice levels.
This study presents conceptual, theoretical, and methodological contributions. First, as far as the authors are aware, there are no other studies that define, model, or solve a joint transit network redesign and mobility service fleet size determination problem. The study models the problem using a bi-level mathematical programming formulation. The problem definition and modeling framework represent a timely contribution to the existing literature, especially with the emergence of AVs, rapid growth of mobility services, and their interaction with transit services. The second and third contributions relate to the formulation of the upper- and lower-level problems in the bi-level JTNR-SFSDP, respectively. This study modifies a transit network frequency setting problem (TNSFP) formulation for the upper-level problem via allowing transit frequencies to be set to near-zero and incorporating SAMS fleet size as a decision variable. For the lower-level problem, this study employs a dynamic combined mode choice—traveler assignment problem (DCMC-TAP) formulation. This appears to be the first study to incorporate time-dependent mode (and route) choice in the lower-level problem of a transit network design problem. Hence, the modeling framework captures the modal split response to the frequency setting of transit patterns and operation/subsidization of SAMS fleets. The fourth contribution is the use of a detailed agent-based simulation tool with three components to address the lower-level problem. The three component models include a multinomial logit mode choice model, a transit traveler assignment-simulation model, and a SAMS fleet simulation model. Although data-intensive, the agent-based model for the lower-level provides information about individual travelers, the transit network, and the SAMS fleet. This information is valuable for model verification, model validation, and understanding the complex interactions between design decisions and travel behavior.
Relative to the small but growing literature modeling the intersection and integration of SAMSs with public transit, the modeling framework in this paper imposes fewer restrictions. For example, this study makes no a priori assumptions about efficient joint designs of transit networks and SAMSs; whereas, existing studies define scenarios in which SAMSs (i) replace specific transit routes (Pinto et al., 2018, Winter et al., 2018), (ii) replace transit systems (Basu et al., 2018), (iii) are implemented instead of new transit lines (Mendes et al., 2017), or (iv) act as a transit feeder mode (Meyer et al., 2017, Scheltes and de Almeida Correia, 2017). Similarly, the modeling framework in the current study endogenously determines modal flows (demand) for SAMSs and public transit based on the performance of the two systems, rather than exogenously (e.g. Shen et al., 2018).
The remainder of the paper is structured as follows. Section 2 presents an overview of the bi-level mathematical programming modeling framework and reviews relevant literature. The Nomenclature section displays the mathematical notation used in the paper. Section 3 presents the mathematical formulation of the bi-level JTNR-SFSDP. Section 4 presents the heuristic solution procedure designed to solve the bi-level JTNR-SFSDP. Section 5 describes the experimental design and Section 6 presents the computational results. Section 7 concludes the paper.
Section snippets
Transit network design and frequency setting problems
Ceder and Wilson (1986) present the transit network design problem (TNDP) and the transit network frequency setting problem (TNFSP) as interdependent public transit service sub-problems. The TNDP is inherently a long-term strategic planning problem, wherein the network designer chooses transit lines and stops (i.e. transit routes). A significant volume of research exists on the TNDP (seminal work: Baaj and Mahmassani, 1995, Ceder and Wilson, 1986, Clarens and Hurdle, 1975). TNDP objective
Model
This section presents the joint transit network redesign and SAMS fleet size determination problem (JTNR-SFSDP) subject to user equilibrium at the mode and route choice level. The general framework of the bilevel model is depicted in Fig. 2 to facilitate the understanding of the mathematical model.
The mathematical formulation of the JTNR-SFSDP is based on a transit network frequency setting problem formulation (TNFSP), in which the set of transit service patterns is fixed. Two major changes to
Overview
To solve the bi-level problem in Eqs. (3), (4), (5), (6), (7), (8), (9), (10), (11), (12), (13), (14), (15), (16), this study employs a heuristic implementation of the approach outlined in Fig. 2. The upper-level problem is the JTNR-SFSDP, except that the lower-level decision variables (mode choice and route choice) are fixed. The outputs of the upper-level model are transit pattern headways and SAMS fleet size. This information is passed to the lower-level model, which is a dynamic combined
Experimental design
The modelling framework is demonstrated at scale using the actual network of the Greater Chicago metropolitan area (USA). The application is performed in the large-scale urban network for the transit system, with SAMS fleet coverage tested in a suburban area. Transit data is taken from the General Transit Feed Specification for services provided by the Chicago Transit Authority and Metra (operators of urban bus and heavy rail, and commuter rail respectively).
The SAMS fleet coverage test area
Computational results
The output is shown for lower level and upper level iterations. The lower level output represents the result of simulated experiences after 28 inner iterations of the dynamic transit traveler assignment-simulation. The upper level results show features associated with the recommended design (SAMS fleet size and pattern frequencies) across 30 upper iterations. An additional “iteration 0” is added to some of the charts to show conditions before any changes in the transit design and before an SAMS
Summary
Fully-autonomous vehicles (AVs) and the recent emergence of shared-use AV mobility services (SAMSs) are likely to significantly impact passenger transportation systems and the behavior of travelers using these systems. This study aims to provide a methodology and modeling framework to support the joint re-design of multimodal transit networks and SAMS fleets to explore and plan for likely future AV-enabled mobility scenarios. Accordingly, this study introduces the joint transit network redesign
Acknowledgements
Partial funding for this work was provided through a doctoral fellowship from the Brazilian National Council for Scientific and Technological Development (CNPq) for the first author. Partial funding was provided through an Eisenhower fellowship from the US Department of Transportation for the second author. Additional funding was provided by the Northwestern University Transportation Center. The computational experiments were supported in part through the computational resources and staff
References (41)
- et al.
Hybrid route generation heuristic algorithm for the design of transit networks
Transp. Res. Part C Emerg. Technol.
(1995) - et al.
Cost-based analysis of autonomous mobility services
Transp. Policy.
(2018) - et al.
Optimizing frequencies in a transit network: a nonlinear bi-level programming approach
Int. Trans. Oper. Res.
(1995) - et al.
A continuous equilibrium network design model and algorithm for transit systems
Transp. Res. Part B Methodol.
(2004) - et al.
Simultaneous optimization of transit line configuration and passenger line assignment
Transp. Res. Part B Methodol.
(2006) - et al.
Transit network design and scheduling: a global review
Transp. Res. Part A Policy Pract.
(2008) - et al.
Planning, operation, and control of bus transport systems: a literature review
Transp. Res. Part B Methodol.
(2015) - et al.
Optimizing the service area and trip selection of an electric automated taxi system used for the last mile of train trips
Transp. Res. Part E Logist. Transp. Rev.
(2016) - et al.
Autonomous vehicles: the next jump in accessibilities?
Res. Transp. Econ.
(2017) - et al.
Exploring the use of automated vehicles as last mile connection of train trips through an agent-based simulation model: an application to Delft, Netherlands
Int. J. Transp. Sci. Technol.
(2017)
Integrating shared autonomous vehicle in public transportation system: a supply-side simulation of the first-mile service in Singapore
Transp. Res. Part A Policy Pract.
A new dynamic traffic model and the existence and calculation of dynamic user equilibria on congested capacity-constrained road networks
Transp. Res. Part B Methodol.
A bi-level programming model and solution algorithm for the location of logistics distribution centers
Appl. Math. Model.
Exploring trade-offs in frequency allocation in a transit network using bus route patterns: methodology and application to large-scale urban systems
Transp. Res. Part B Methodol.
Gap-based transit assignment algorithm with vehicle capacity constraints: simulation-based implementation and large-scale application
Transp. Res. Part B: Methodol.
Performance analysis and fleet requirements of automated demand-responsive transport systems as an urban public transport service
Int. J. Transp. Sci. Technol.
Automated mobility-on-demand vs. mass transit: a multi-modal activity-driven agent-based simulation approach
Transp. Res. Rec. J. Transp. Res. Board.
User and operator perspectives in transit network design
Transp. Res. Rec. J. Transp. Res. Board.
Bus network design
Transp. Res. Part B.
An operating strategy for a commuter bus system
Transp. Sci.
Cited by (0)
- ☆
This paper has been accepted for a podium presentation at the 23rd International Symposium on Transportation and Traffic Theory (ISTTT23) July 24–26, 2019 in Lausanne, CH.