Ph.D. Defense

Brenden Oates

(Faculty Advisor: Professor Marilyn Smith)

"A Dynamic Model of Unpiloted Aerial Systems for Complex Ship Airwake Environments"

Thursday, April 9

1:00 p.m.

Price Gilbert 4222

Abstract: 

The complex airwake generated by naval ships poses a number of challenges for aircraft operations at sea. The bluff-body geometry of the ship itself creates troublesome turbulent structures that convect downstream over the flight deck where most vehicle launch and recovery takes place. In recent years there has been an increase in the procurement and application of small unpiloted aerial vehicles (UAVs) in many different fields. Shipboard operations of small UAV have great potential for a number of applications, including rapid natural disaster relief operations, systems support intelligence and reconnaissance, geo-sensing and mapping, and more. These smaller, novel UAVs are not constrained by the same design factors that go into traditional, full-scale aircraft and there are a number of uniquely designed systems that do not have a abundance of historical data to inform their operational limits in complex airwake environments. At-sea experimental testing is the most informative and accurate way to determine vehicle aerodynamic characteristics in these environments, but it is extremely costly and time-consuming. Advanced computational fluid dynamics (CFD) methods have been developed to resolve much of the turbulent flow structures and the interactions with the vehicle, but require significant allocations or investments in high-performance computing (HPC) clusters all while still requiring notable wall-clock time for converged results. Shipboard operations require fast, computationally efficient, and physically realistic modeling capabilities to supplement the traditional, resource-intensive experimental approaches. A tool of this nature must balance cost and accuracy while capturing the essential physics of the ship airwake and the corresponding unsteady aerodynamic response of the vehicle.

This effort focuses on developing an integrated machine learning (ML) -based modeling and simulating framework for predicting complex ship airwake quantities and the resulting small UAV responses during launch and recovery; and assessing the sensitivity, relative cost, and the accuracy of these predictions. First, metrics need to be defined for what constitutes a realistic reduced-order model of the complex ship airwake with respect to small UAV shipboard operations. To accomplish this, low-order flight dynamics simulations are employed to investigate the unsteady, dynamic vehicle response along the approach path towards the flight deck. The unsteady behavior of the vehicle through time is correlated to the specific structures in the airwake that induce energetic responses. These are used as benchmark quantities to capture in the reduced-order model for functionally realistic ship airwake predictions. Finally, a new model is introduced that incorporates the ML-based airwake predictions coupled with vehicle flight dynamics and an autonomous trim routine to predict the realistic trajectory of the small UAV given a prescribed approach path to help inform safe operating limits based on the given ship-UAV combination.

Committee:
Dr. Marilyn Smith (advisor), School of Aerospace Engineering
Dr. Juergen Rauleder, School of Aerospace Engineering
Dr. Beckett Zhou, School of Aerospace Engineering
Dr. Karen Feigh, School of Aerospace Engineering
Dr. Daniel Prosser, US Navy