Droneport airside operations

Abstract

There is a growing need for Unmanned Aerial Vehicles (UAVs, or drones) in commercial, civil, and military applications. Tens of millions of annual flights are expected to fly in the airspace by 2050. To address potential safety and airspace congestion issues caused by the rise in drone operations, this research presents the concept of the droneport and the dynamic carousel circuit. The droneport is a facility that accommodates and manages drones taking off and landing in a protected space under air traffic control. The dynamic carousel circuit, a traffic pattern, with a changing radius, is designed to accommodate the growing demand for future drones. This thesis presents a few contributions to the concept of the droneport: (1) Future delivery drone demand is forecasted using historical online retailer data and the Holt-Winters’ seasonal method; (2) Optimum number and distribution of droneports are determined by a multi-objective optimization model considering both costs and societal values from six aspects: maximizing e-commerce demand coverage, airtaxi demand coverage, subzone coverage, and area coverage, and minimizing service distance for both parcel and passenger delivery drones; (3) Optimization model integrates Gaussian noise to make the measurement of service distance more practical; (4) Future capacity of each droneport is estimated based on the number of droneports and their placement. A real-world case study is carried out for Singapore to find the optimum distribution of droneports and forecast capacity of each droneport. Additionally, in order to manage a large number of drones around the droneport, the dynamic carousel circuit is conceived and integrated with the droneport operation. This thesis also develops a simulation model as well as an optimization algorithm to determine the optimum circuit radius, the moving speed of drones on the circuit, and the circuit altitude with forecasted drone demand. A more realistic simulation with the residual endurance estimation model and the cubic trajectory planning model is applied. The former model uses a practically applicable method to calculate the drone’s endurance with the remaining battery level as input. The latter model generates a smooth landing trajectory and thus estimates the travel time of the drone more accurately. Numerical tests are conducted to analyze the performance of the dynamic carousel circuit. The findings from this study show that the dynamic carousel circuit has the potential to increase droneport capacity, improve aviation safety, and enhance droneport operational efficiency. Furthermore, two uncertainties, weather and dynamic incoming flow are taken into consideration to validate the efficiency and effectiveness of the dynamic carousel circuit. A comparison of prediction accuracies for first-order and second-order Markov chain models with simple weather states or realistic weather states is presented in this thesis. Besides, a novel approach, two-layer simulation optimization, is introduced to solve the optimization problem for large-scale stochastic simulation efficiently. This proposed approach is composed of a segmental simulation optimization algorithm with a Genetic Algorithm, which is designed to obtain the optimum radius for the dynamic carousel circuit of each interval in parallel. Then a ranking and selection process to find the best candidate for a whole-day simulation duration among these optimum results efficiently. The finding of this study shows that such approach can be successfully applied to obtain the optimum radius for the dynamic carousel circuit with stochastic inputs. Results from the Monte Carlo simulation prove the stability of the dynamic carousel circuit under weather uncertainty and changing demand. Lastly, to provide a more realistic analysis of the carousel circuit performance under weather uncertainty, wind disturbance is modeled and integrated with quadrotor's six Degrees of Freedom dynamics, and an adaptive neural-fuzzy controller is applied to the real-time simulation. This controller performs well to compensate for the wind effect and provides meaningful insights into the influence of wind on safe time separation.