Reduced-order modeling of aeroelastic phenomena

Traditional methods to identify the aeroelastic stability of turbomachinery have been focused on the use of either experimental investigations of the device or fully coupled fluid-structural simulation techniques. While both methods provide accurate depictions of the underlying stability of the device in question, the time or computational cost associated with obtaining the stability analyses becomes excessive when the aeroelastic stability of the device must be evaluated over multiple operating regimes. In this dissertation, a reduced-order modeling method is developed and presented in order to greatly diminish the numerical expense associated with evaluating aeroelastic stability at a defined operating condition. The method differs from existing low-order approaches in that it leverages the use of piston theory in conjunction with steady-state simulation data from computational fluid dynamics simulations in order to predict the fluid loading that arises in response to the structural deformation. The model is applied to previously-studied, canonical panel flutter configurations to demonstrate the accuracy of the method. The application of the method on the high-pressure turbine of a dual-stage turbocharger is then demonstrated and the stability predictions compared against experimental observations conducted independently by scientists at the Army Research Laboratory. The reduced-order modeling method is confirmed to accurately diagnose the qualitative stability properties of the device with respect to aeroelastic flutter and a discussion regarding methods to properly diagnose the susceptibility of the device to forced response is presented. To address the shortcomings of aerodynamic piston theory in subsonic flow regimes and in modest supersonic flow regimes, a stability method incorporating spatial pressure fluctuation modes learned using dynamic mode decomposition is developed. This method is first applied to two- and three-dimensional flows over beams and panels exhibiting a harmonic response consistent with aeroelastic flutter. The manner in which to learn the leading spatial modes that dominate the pressure response for each configuration is presented and these spatial modes are compared against the spatial modes computed by a boundary element method in each scenario to confirm that the dynamic mode decomposition algorithm indeed learns the correct pressure response. These leading spatial modes are then used to compute the stability of the structures in modest supersonic regimes. The results are compared against separate investigations, both numerical and experimental, that have been previously presented in the literature to confirm that the stability method incorporating approximate pressure fluctuation modes learned from data can accurately predict the onset of aeroelastic flutter.

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Aerospace Engineering

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2023

Author dc:creator

• Fellows, David William

Contributors dc:contributor

• Kang, Sang-Guk
• Bodony, Daniel J
• Goza, Andres
• Vakakis, Alexander

• Fluid-structural Interaction
• Aeroelasticity
• Computational Fluid Dynamics
• Data-driven Modeling
• Unsteady Aerodynamic Modeling

Statement dc:rights

• Copyright 2023 David William Fellows

Language dc:language

en, eng

Handle dc:identifier

https://hdl.handle.net/2142/121400

OAI record oai:identifier

https://www.ideals.illinois.edu/oai-pmh?verb=GetRecord&metadataPrefix=etdms&identifier=oai%3Awww.ideals.illinois.edu%3A2142%2F121400