Planning Fleet Deployment and Charging Operations for Shared, Integrated, and Electrified Urban Mobility Systems
Advisor: Professor Yanfeng Ouyang
Abstract
This dissertation aims at developing overarching aspatial queuing network models to improve the strategic planning of a spectrum of shared mobility services, with main focuses on: (i) integration of on-demand mobility services (e.g., taxi, ride-pooling) with conventional public transit; (ii) vehicle deployment and passenger matching under spatially heterogeneous demand; and (iii) charging infrastructure planning and operations (e.g., for shared dockless electric bikes and scooters).
On-demand mobility services are expected to complement conventional transit services, and further to facilitate a modal shift from private-owned vehicles to sustainable public transportation. However, they may also pose as competitors. There is numerous real-world evidence showing that they cause demand diversion away from transit, partly because they offer more convenient travel options. Yet, conventional transit is indispensable in major cities, providing efficient, affordable, and sustainable transportation services. To better integrate these services with conventional transit, this dissertation proposes a zone-based scheme, where on-demand mobility provides door-to-door service for intra-zonal trips, while inter-zonal trips use on-demand services as a feeder to access the conventional transit. Aspatial queuing network models are developed to capture taxi and ride-pooling operations of serving intra- and inter-zonal trips simultaneously. A system of equations is formulated to quantify system performance based on the planning decisions, including zone partition, fleet size and deployment, and rebalancing operations for the local shared mobility services, as well as the spacing and headway of a grid transit network.
Aspatial queuing network models traditionally ignore the spatial heterogeneity in demand, vehicle deployment, and rebalancing operations. To fill this gap, this dissertation proposes a multi-zone queuing network model to analyze steady-state ride-pooling operations under heterogeneous demand. The service region is partitioned into a set of relatively homogeneous zones. A set of criteria is imposed to avoid significant detours when matching passengers. A large system of equations is formulated to describe passenger-vehicle matching, conservation, and vehicle movement within each zone and across neighboring zones. These equations analytically evaluate steady-state system performance based on planning decisions, such as zone-level vehicle deployment, vehicle routing paths, and vehicle rebalancing operations.
Shared electric micro-mobility services have become increasingly popular in recent years, but they also impose additional challenges related to their limited service ranges and frequent charging needs. This dissertation studies the operations and planning of charging options for shared e-scooters, first using local charging stations as an example. Riders can pick up e-scooters at both random locations and at a set of charging stations, and they are encouraged to drop off e-scooters at charging stations with incentives. An aspatial queuing network model is developed to capture the steady-state e-scooter reservation, battery consumption and charging processes, and the associated pricing and management mechanisms. The analysis on e-scooter charging stations is further extended to other charging options, such as charging at a central off-site depot, or at isolated charging hubs, swapping batteries, and crowd-sourcing independent contractors.
For each service system, the queuing network model is integrated into a constrained non-linear program, to optimize the planning decisions, and near-optimal solutions are obtained based on customized algorithms.
All results from the proposed models and algorithms are verified via agent-based simulations and tested under a series of hypothetical scenarios. The numerical results demonstrate the advantages of the proposed mobility systems that integrate on-demand services and conventional transit services, highlight the importance of addressing spatially heterogeneous demand for the strategic planning, and recommend optimal charging options for electric micro-mobility services under different service scenarios.
