Abstract
This dissertation describes a series of real-time vehicle routing problems with the associate optimization and simulation modeling for flexible passenger transport systems such as the High Coverage Point-to-Point Transit (HCPPT) and shared-taxi, which involve a sufficient number of deployed small vehicles with advanced information supply schemes to match real-time passenger demands and vehicle position for passenger transportation over large areas. HCPPT is an alternate design for mass passenger transport developed in recent years at the University of California at Irvine. The designs rely on transfer hubs, trunk route connections between the hubs where the vehicles are non-reroutable, and local areas around the hubs where the vehicles are reroutable. First, we relax the restriction in the existing heuristic rules of HCPPT, expecting to yield higher efficiency for general cases. Optimization schemes are proposed for both trunk and local vehicle routing problems to consider global optimality for large-scale problems. Significantly, the new algorithms allow globally optimal vehicle movements over multiple-hubs, unlike the earlier designs that allowed travel only to the adjacent hubs. This in turn ensures that the scheme has scalability in large areas and has design flexibility in adjusting the distances between hubs. Second, for an efficient and productive taxi system of the conventional kind, a design of shared-taxi operation is proposed, which also can be potentially used for local area operations in HCPPT. Three algorithms are developed and compared with different objective functions. Another contribution of this research is the development of a simulation platform targeting large-scale flexible point-to-point transit systems with various vehicle operation schemes. Traditionally, real-time DRT operations are simulated with commercial traffic simulators such as mesoscopic or microscopic simulation models, which is cumbersome because the available software were not designed for such real-time routed vehicle simulation, and also because they include details of less relevance to large-scale real-time Demand Responsive Transit (DRT) systems. The simulation studies in this research evaluate the vehicle routing algorithms through the proposed platform for Orange County, U.S.A. and Seoul, Korea. Finally, this thesis studies two large-scale fleet applications of Electric Vehicles (EV) as a future transportation alternative, as the hub locations which are part of the designs developed in this research are particularly suitable as energy replenishment nodes. Since EVs have a limited driving range and need to visit charging stations frequently, this part mainly focuses on the vehicle charge replenishing schedules in conjunction with passenger pickup and delivery schedules and measures the benefits from combining EVs and DRT fleets.
