Multidisciplinary Design Optimization of Delivery Uncrewed Aerial Vehicles Considering Operations

Abstract

Uncrewed aerial vehicles (UAVs) enable rapid and environmentally friendly delivery of lightweight payloads, such as commercial packages, food, and medical supplies. Designing an energy-efficient UAV is the key to achieving the full potential of UAV delivery and lowering the environmental impact. To do so, multidisciplinary design optimization (MDO) is a powerful tool that assists engineers with design tasks and accelerates design space exploration by automatically finding the optimal design given the requirements and assumptions.

This dissertation focuses on conceptual design optimization of delivery UAVs considering flight operations. Incorporating flight operational analysis and optimization in the UAV conceptual design process is important because the expected flight operations drive the design requirements, hence the UAV design. At the same time, the optimal flight operations are dependent on the UAV design parameters, meaning that the optimal UAV design and flight operations are mutually coupled. I apply MDO to capture this design-operation coupling, which ultimately improves the design and performance of delivery UAVs.

In the first part of this dissertation, I present design optimization of a UAV fleet considering delivery operations. This problem simultaneously optimizes the fleet size and composition, UAV design variables, and delivery routing to design a fleet that fulfills expected delivery demands. To solve this mixed-integer nonlinear optimization problem, I propose an effective sequential heuristic algorithm that combines gradient-based design optimization and vehicle routing heuristics. I then demonstrate the accuracy, computational efficiency, and scalability of the proposed algorithm by comparing it to a commercial branch-and-cut solver. The results show that simultaneous fleet design and routing optimization reduces the fleet acquisition cost and delivery energy consumption by up to 20% compared to conventional uncoupled optimization.

In the second part, I perform UAV conceptual design optimization considering takeoff flight operations, focusing on a single vehicle design in higher resolution. This problem yields simultaneous optimization of UAV design and takeoff trajectory, and I apply gradient-based optimization to solve it efficiently. To reduce the computational cost of gradient-based optimization, I propose new hierarchical linear solution strategies to accelerate derivative computations. I also perform benchmark studies of monolithic MDO architectures and design-trajectory coupling strategies to identify the best approach in terms of computational cost. The results of this dissertation show that simultaneous optimization can find a more energy-efficient UAV design compared to uncoupled optimization. This is achieved by capturing the system-level trade-offs between the efficiency in different flight phases, which uncoupled optimization cannot fully address. This demonstrates the importance of incorporating takeoff trajectory optimization in the conceptual design process of delivery UAVs.