In the past decade, the e-hail service provided by transportation network companies (TNCs) gained popularity in major cities around the world. By allowing passengers to virtually “hail” vehicles on mobile phones, e-hail has revolutionized the matching process in ride-hail, drastically reducing the existing search friction. However, previous studies found the e-hail service may not only suffer from unexpected efficiency losses under certain market conditions, but also bring considerable pressure on already-congested streets in big cities. Partly motivated by these concerns, TNCs introduced pooling services to complement the regular e-hail service. Passengers who opt for pooling are paired in real-time and share a portion of their trips with others. Despite its great potential in theory, only a fraction of TNC passengers choose to pool. The objective of this dissertation is thus to understand (i) whether pooling could address the efficiency and sustainability concerns about e-hail service in current practice, (ii) under which conditions it could thrive in competition against other transportation modes, and (iii) how it should be better designed, operated and regulated in different market scenarios. To this end, this dissertation develops both analytical and numerical tools to investigate pooling in an urban transportation system with both ride-hail and other travel modes.This dissertation is divided into four parts. Part 1 proposes a physical model that describes the interaction between passengers and vehicles in the matching process of a pooling ride. The model supports the analysis of the main trade-off of pooling with respect to matching and serves as a building block of the following studies. Part 2 studies pooling in an aggregate ride-hail market, where passengers make mode choice between pooling and other travel modes, drivers make decisions on whether to join the market, and the service platforms determine the optimal pricing strategy. The analysis is first conducted for a market with a single platform and then extended to tackle the inter-platform competition. Part 3 is devoted to modeling a spatial ride-hail market with pooling. A stylized two-node model is first developed to investigate the congestion effect of ride-hail operations and evaluate different congestion mitigation schemes targeted at them. Then, a game-theoretic approach is proposed for modeling the dynamic routing of ride-hail vehicles in a spatial market. The model is first formulated as a non-cooperative game and then extended to accommodate cooperative routing. Finally, an agent-based ride-hail simulation is developed to support the analysis with more operational details. With the simulator, we demonstrate a metamodel-based simulation optimization approach that embeds an analytical model into the simulation and has it provide updating directions of design variables (e.g., the pricing scheme and matching interval) throughout the simulation. Empirical TNC and traffic data are utilized to construct numerical experiments throughout this dissertation in an attempt to demonstrate the theoretical findings and test the sensitivity of the system performance under various market conditions.

Creator

• Zhang, Kenan

DOI

• https://doi.org/10.21985/n2-ecfw-yf08

Subject

• Transportation

Language

• en

Alternate Identifier

• http://dissertations.umi.com/northwestern:15842
• etdadmin_upload_859795

Date created

• 2021-01-01

Resource type

• Dissertation

Rights statement

• In Copyright