Strategic planning for service-based advanced air mobility systems under uncertainty : models, methods and applications

Abstract

National Aeronautics and Space Administration's Advanced Air Mobility (AAM) initiative is set to transform our communities by enabling the movement of people and goods from the ground to the sky. This new transportation system will revolutionize both urban mobility and cargo delivery services. In the context of urban transportation, this concept is commonly known as Urban Air Mobility (UAM), which focuses on enabling low-altitude aerial travel for passengers using electric vertical take-off and landing vehicles. UAM aims to alleviate traffic congestion, shorten commute times, and offer an environmentally sustainable alternative to traditional ground transportation. In the cargo delivery domain, particularly in humanitarian logistics, AAM provides critical advantages in disaster response through the use of drones. Drones can rapidly deliver goods to isolated or hard-to-reach regions, bypassing blocked roads and ensuring timely aid. This ability to quickly and flexibly deliver supplies to disaster-stricken areas provides an indispensable tool for humanitarian organisations. This thesis focuses on the strategic planning of service-based AAM systems under uncertainty, with a particular emphasis on the application of UAM and drone-supported last-mile humanitarian logistics. The first study introduces an integrated optimisation problem aimed at addressing decision-making processes related to the strategic planning and service operations of UAM systems, considering both demand uncertainty and spatial equity. We introduce a spatial equity metric and establish a bi-objective optimisation model to balance the trade-off between service profitability and spatial equity considerations. We employ a scenario-based robust optimisation framework that incorporates the interval robust method to capture the demand uncertainty, enhancing resilience against uncertain factors. We evaluate the model performance through numerical experiments and a practical case study based on a megalopolis in southern China and propose valuable policy recommendations for UAM service providers. The second study focuses on drone-supported last-mile humanitarian logistics planning under demand uncertainty, specifically in scenarios where the uncertainty is realised sequentially. We propose a drone-supported last-mile humanitarian logistics system planning problem. To capture the demand uncertainty, we establish a multistage stochastic programming model incorporating nonanticipativity constraints to make decisions at each stage without knowledge of the demand information in future time periods. The Benders decomposition algorithm is then employed to derive exact solutions. We validate the proposed optimisation models and solution methods through a case study of the Lushan earthquake in China. This research contributes to the field of humanitarian logistics by providing a comprehensive framework for planning drone-supported last-mile humanitarian logistics systems. In the third study, we examine a scenario in which historical demand data is lacking and distribution information is only partially available. We investigate a novel drone-supported relief facility location problem and apply a distributionally robust optimisation (DRO) framework, utilising box, polyhedral, and ellipsoidal ambiguity sets to address demand uncertainty. To overcome the computational challenges, we reformulate the DRO models into computationally tractable forms. The DRO approach ensures superior out-of-sample performance in the face of partial demand information, thereby supporting effective decision-making in humanitarian logistics. Finally, we also validate the proposed optimisation models through a case study of the Lushan earthquake in China and offer valuable managerial implications to support decision-making processes for humanitarian organisations. The models and methods proposed in this thesis pave the way for innovative practices in strategic planning for service-based AAM systems.