A hybrid multimodal mobility system aims to deliver mobility as a service by integrating a broad range of existing and emerging public transportation services. The appeal of a hybrid system is twofold. First, it can strike a better balance between cost and level of service (accessibility, efficiency etc.). Second, it promises a seamless, user- and environmentally-friendly, and affordable travel experience. The thesis aims to advance the analysis on hybrid transit systems, with a focus on the optimal system design at the strategic level, and insights about the fundamental trade-offs that can inform decision-making. The first part of the thesis is concerned with the optimal design of hybrid transit systems. We first assume a nonlinear trip production density in the monocentric city, though it might not lead to more insights into the effect of spatial heterogeneity compared to a much simpler demand representation. Then, a simpler trip production and distribution model is introduced that differentiates the central business district (CBD) of a city from its periphery. To cope with the heterogeneous demand pattern, the hybrid transit system is also configured differently inside and outside the CBD. Allowing such supply heterogeneity complicates transit users' route choice modeling considerably. As a result, user costs must be estimated separately for six scenarios, each corresponding to a unique route choice pattern and a design model. Our main finding is that recognizing demand heterogeneity and responding to it with a tailored design can be highly beneficial. However, this benefit diminishes as the average demand density increases. The second part of the thesis examines the robustness of strategic transit design models. We first investigate the basic design trade-off in six distinctive transit systems, including two state-of-the-art fixed-route systems and four hybrid systems that use ride-pooling as an integrated feeder service. We find that ride-pooling changes little the fundamental laws inherent in transit design. Specifically, all six systems, despite their seemingly vastly different design features, display the following laws: (1) that the per capita agency cost correlates linearly with the per capita user cost, (2) that both costs are power functions of the demand density with an exponent close to -1/3 and -0.4 for agency and user costs, respectively, and (3) that the per capita agency cost is not significantly affected by city size but the user cost is. Furthermore, we examine the effect of route choice modeling on the strategic design models. Traditionally, passengers' route choice is greatly simplified in such models. We aim to understand whether this simplification would compromise qualitatively the results obtained from them. To this end, three transit systems, which all offer competitive alternative routes, are considered. We test what the impact on transit system performance (e.g. optimal designs and system costs) would be if travelers somehow split between these routes, rather than concentrate on the "best" one. A random utility model is employed to enable a probabilistic assignment of passengers to different routes according to the "perceived" utility. Analytical methods are then developed to estimate the aggregate share of each route in each system, based on which the user cost is obtained. Results show that, while stochastic route choice modestly increases the optimal user cost, it has a negligible effect on the agency cost. Furthermore, the actual system design is largely insensitive to route choice modeling. In the third part, we consider a hybrid system that integrates traditional transit with bike-sharing services. In this new system, bikes are treated as a feeder to transit. Travelers may ride bikes to the closest transit stop or a more distant one, depending on which option provides a higher utility. We model this route choice behavior using the continuous approximation (CA) approach, based on which the integrated design is formulated as a mixed integer program. We discuss analytical solutions to a simplified model, although the full model still has to be solved numerically. Results show that the agency cost enjoys a much greater economy of scale in an integrated transit bike-sharing system than in a transit system without shared bikes, whereas the user cost scales similarly.

Creator

• Luo, Sida

DOI

• https://doi.org/10.21985/n2-f7af-j227

Subject

• Transportation
• Civil engineering

Language

• en

Alternate Identifier

• http://dissertations.umi.com/northwestern:15206
• etdadmin_upload_758565

Keyword

• Network design
• Continuous approximation
• Hybrid mobility system
• Public transit

Date created

• 2020-01-01

Resource type

• Dissertation

Rights statement

• In Copyright