Series
Doctor of Philosophy with a Major in Aerospace Engineering

Permanent Link
https://hdl.handle.net/1853/72952
Series Type
Degree Series
Description
Associated Organization(s)
Associated Organization(s)

Filters
1
2022 1
1
1
1
1
Reset filters

Settings
Start date: 2020 × End date: 2023 × Author: Baier, Dalton ×

Publication Search Results

Now showing 1 - 1 of 1
  • Thumbnail Image
    Item
    A Surrogate Modeling Approach for High-Speed Aerodynamics of Scramjet Inlets & Isolators
    (Georgia Institute of Technology, 2022-07-30) Baier, Dalton
    Several unique challenges exist in the design & analysis of airbreathing hypersonic systems. One of the greatest challenges is caused by inclusion of an airbreathing propulsion system as it introduces multiple strong couplings and interactions with other disciplines in a system already characterized as highly coupled. Additionally, the propulsion system is a key driver in the design & analysis of these vehicles. Many current approaches for early stage design & analysis of scramjets rely on quasi-1D physics or empirical models due to low computational costs required for rapid/efficient design space exploration or conceptual design studies. In these approaches, computational accuracy is lost due to assumptions regarding physics of the problem in order to yield computationally efficient predictions, however these predictions typically do not capture important complex phenomena (e.g. SBLI). Viscous phenomena have been shown to exert first order effects on scramjet inlet performance, which makes it imperative to include viscous flow considerations in the final design; however these necessary physics-based analyses for scramjet systems are prohibitively expensive in early stage vehicle design. Current design methods typically focus on predicting overall vehicle performance scalars or transferring scalar quantities between disciplines, nevertheless inherent complexity of airbreathing hypersonic systems also requires field data to assess overall system performance. Surrogate modeling techniques present a possible solution to reducing the high expense associated with the required CFD analysis of scramjet systems; however none have been explored for field surrogate models of flows relevant to scramjets. Scramjet flowfields are complex and characterized by the presence of nonlinear phenomena such as shocks, viscous effects, SBLI, etc. These observations formulate the aim of this thesis: to identify and explore a surrogate modeling approach tailored to accurately and efficiently model the complex, nonlinear flowfields present in scramjet inlets & isolators. This research aims to investigate a neural network surrogate modeling approach to improve the information available earlier in the design process for the scramjet inlet/isolator, by yielding field data at a reduced cost compared to the full-order (CFD) models. Two loss functions are investigated with respect to representative single and multiple condition/geometry problems: 1) data-driven and 2) physics-informed, where the governing equations are included in the loss function. Additionally, the scalability of the approach is investigated and compared to full-order (CFD) solutions through a parametric scramjet mission relevant problem. Finally, the proposed neural network surrogate modeling approach is demonstrated on an inward turning inlet with bleed/vents at varying flow conditions. Distribution Statement A. Approved for public release. Distribution is unlimited. Abstract cleared: PA# AFRL-2022-2663 The views expressed are those of the author and do not necessarily reflect the official policy or position of the Department of the Air Force, Department of Defense, or the U.S. Government.