Optimization of operational and planning problems for shared transportation services

Abstract

As a critical component of the sharing economy, Shared Transportation Services (STS) have emerged as a promising paradigm for addressing transportation challenges by promoting the shared use of both vehicles and infrastructure. This thesis focuses on two emerging forms of STS: (1) co-modality, which involves the shared use of vehicles for both passenger and freight transportation, and (2) infrastructure sharing within integrated parking and charging lots (IPCLs), where parking and charging spaces are jointly utilized by both parking and charging users. Centered around these two STS, the thesis investigates decision optimization at both the operational and planning levels, structured into four key modules. The first module addresses the design of freight service networks within a bus-integrated logistics system, where excess capacity on scheduled bus services is utilized for freight transport. A time-space network-based optimization model is proposed to jointly determine truck fleet sizing, routing and scheduling, and the allocation of freight between trucking and bus transport, aiming to minimize total operating costs. A column generation-based two-stage solution method is developed to solve the problem efficiently. Numerical results indicate that co-modality can lead to substantial economic and social benefits. The second module explores the integration of modular vehicles into shared passenger and freight transportation. Each modular vehicle comprises identical modules that can be dynamically assigned to either passenger or freight service. A mixed-integer programming formulation is established to optimize vehicle dispatch times, (un)docking operations, and the allocation of modules for passenger and freight services. A two-stage algorithm is developed to solve the problem efficiently. A case study validates the potential cost reductions achieved through station-wise (un)docking and co-modality operations. The third module proposes a reservation system for jointly managing parking and charging demand in IPCLs through admission and allocation controls. The system allows users to request parking or charging spaces in advance, while the operator decides on reservation admission and space allocation. This problem is formulated as a Markov decision process, and two decomposition-based strategies are proposed to enable prompt operational decisions. Numerical experiments show substantial improvements in revenue and facility utilization compared to current practices. The fourth module extends the analysis by incorporating vehicle relocation operations, enabled by emerging autonomous valet parking technologies. This relocation capability allows vehicles to vacate charging spaces once charging is complete and temporarily park in non-charging spaces until a charging spot becomes available. Three online allocation strategies are proposed within a rolling horizon framework. A case study demonstrates the significant advantages of non-myopic strategies over myopic ones, and highlights the considerable profit improvements achieved by incorporating vehicle relocation operations. In sum, this thesis provides an in-depth investigation into the planning and operational challenges of co-modality and shared parking and charging services. The insights gained not only inform the formulation of relevant policies but also offer robust support for the design, planning, and daily operations of real-world implementations.