Design of Freight Logistics Services in the Era of Autonomous Transportation
Advisor: Professor Yanfeng Ouyang
Abstract
The booming e-commerce has magnified the critical importance of efficient and sustainable freight logistics. The current logistics industry, heavily reliant on traditional transportation technologies, faces substantial challenges related to cost, efficiency, and safety. Concurrently, the advent of autonomous transportation technologies presents unprecedented prospects for transforming both the first-/last-mile and line-haul segments of freight logistics systems. For first-/last-mile parcel deliveries between distribution centers and individual customers, the use of unmanned aerial vehicles (UAVs, also known as drones) has emerged as a promising solution that can partially supplant human labor. Drones offer a host of advantages, including high cruising speed, enhanced delivery flexibility, extended service hours, and reduced carbon emissions. However, large-scale delivery services require operating a large fleet of drones simultaneously in low-altitude air, which poses the challenge of aerial congestion. For line-haul shipments of bulk cargos, which heavily rely on freight trucks over freeway networks, the concept of truck platooning (i.e., compactly operating trucks with very small inter-vehicle spacings and lateral displacements) on dedicated lanes appeals to the industry with promising savings in fuel cost, mainly due to the reduced air drag. Such a solution, however, accelerates pavement deterioration by repetitively channelizing large loads onto a few lateral positions within the lane, necessitating more frequent and intensive pavement rehabilitation. It is therefore crucial to strike a balance between these competing pros and cons for all the involved stakeholders. With these challenges and opportunities in focus, this thesis endeavors to develop integrated mathematical frameworks for designing freight logistics services that harness the potential benefits from autonomous transportation technologies, while at the same time, explicitly address their potential socioeconomic disbenefits.
We first focus on ways to utilize drones in first-/last-mile parcel delivery, including a collaborative delivery system that uses both trucks and drones. The framework simultaneously incorporates multiple truck types and strategies, allowing for flexible delivery operations to efficiently fulfill demands under various situations. In the simplest strategy, drones can be directly dispatched from the distribution center to service customers within reach. A collaborative delivery strategy uses trucks to haul goods and drones over most of the distance and serve as satellite depots, from which drones take off and complete deliveries within a reasonable distance from the customers. Such a collaboration could be made even more efficient by allowing a fleet of drones to depart from and return to the truck while both are in motion, in so doing streamlining the delivery process and better leveraging the strengths of both trucks and drones. In each strategy, the routes of drones are optimized with explicit considerations of endogenous aerial congestion. We investigated the drone traffic flow in the continuous airspace and formulated the equilibrium condition as a system of partial differential equations (PDEs), which is solved using customized numerical methods. The solution to the PDEs reveals key performance metrics, including the travel time of delivery trips originating from various customer locations. This information is then used to evaluate the operational cost of the delivery strategy, which further provides the basis for optimizing several service design decisions, such as truck speed, truck routing plan, and delivery frequency. Through numerical experiments conducted on generalized roadway networks and demand patterns, we demonstrate the applicability of the proposed modeling framework and derive valuable managerial insights for logistics carriers. The results clearly demonstrate that with an optimized design of the drone-based delivery system, there is a significant reduction in both the average delivery cost and the average delivery time when compared to traditional truck-only delivery strategies.
For the line-haul shipments, our focus is on designing the optimal strategy for platooning autonomous trucks on dedicated freeway lanes. The dual impacts of truck platooning (i.e., reduction in fuel consumption and acceleration in pavement deterioration) under any configuration of platoons (e.g., size, weight, speed, clearance) are studied via (i) computational fluid dynamics studies on aerial resistance on truck bodies, and (ii) mechanistic-empirical analysis of pavement deterioration under repeated loading, respectively. The results provide a basis for simultaneously optimizing pavement rehabilitation activities and truck platoon configuration throughout the pavement's life cycle, in so doing obtaining generalized link cost functions for platoon lanes. We then propose an integrated network design model that simultaneously optimizes the placement of dedicated truck platoon lanes and the pricing of user tolls, while considering the pooling and routing of autonomous truck traffic among various origins and destinations in the freeway network. The applicability and performance of our proposed design framework are demonstrated via a numerical case study based on the Illinois freeway network. The model output effectively channelizes freight traffic onto a few corridors of platoon lanes, and achieved significant system-wide improvements for both freight carriers, and highway agencies.
