Brenton J. Willier
(Advisor: Prof. Dimitri Mavris)
will defend a doctoral thesis entitled,
A Methodology For Quantifying High-Dimensional Unsteady Field Uncertainty With An Entry Vehicle Application
On
Wednesday, August 6 at 10:00 a.m. EDT
Collaborative Visualization Environment (CoVE) Weber SST II
And
Abstract
As planetary missions grow more complex, future blunt-body entry vehicles must accommodate larger payloads and tighter landing constraints. To support this evolution, NASA has identified key gaps in current entry, descent, and landing aerodynamic modeling capabilities, most notably, the need for improved low Mach number models that incorporate uncertainty. Traditional uncertainty quantification (UQ) methods, which rely on conservative scalar adders and multipliers in aerodynamic look-up databases, inflate vehicle mass margins, reduce the accuracy of landing predictions, and hinder mission optimization. This is particularly detrimental in the low supersonic to transonic regime, where the blunt-body's expansive unsteady wake significantly contributes to the uncertainty of aerodynamic force and moment coefficients.
As an alternative to utilizing aerodynamic databases to predict trajectories, emerging state-of-the-art tools, such as Computational Fluid Dynamics (CFD)-in-the-loop frameworks, enable high-fidelity 6-DOF simulations and provide a rich source of unsteady aerodynamic data. However, these tools are too computationally expensive to use with traditional landing accuracy prediction methods, such as Monte Carlo-based trajectory dispersion analysis. Motivated to find a compromise between databases and CFD-in-the-loop, contemporary research efforts seek to develop innovative surrogate-based approaches, such as reduced order models (ROMs), to leverage high-fidelity CFD data while simultaneously enabling rapid aerodynamic predictions.
Upon reviewing the literature for aerodynamic prediction methods, surrogate modeling, and UQ, it was apparent that to construct a ROM that can predict aerodynamic force and moment coefficients and propagate the associated unsteady uncertainty, a new methodology was needed. To compile this method, four key technical areas were investigated: (1) how to sample unsteady uncertainty from CFD, (2) how dimensionality reduction would affect the uncertainty captured in high-dimensional fields, (3) how to predict uncertain latent coordinates, and (4) how to back-map latent uncertainty to integrated aerodynamic quantities. Together, the successful experimentation of these four research areas enabled the compilation of the finalized uncertain parametric ROM methodology.
An overarching hypothesis was posed and tested with a high-fidelity CFD-in-the-loop demonstration to evaluate the compiled uncertain parametric ROM methodology. The ROM-predicted aerodynamic coefficient distributions show strong agreement with the high-fidelity CFD results, accurately capturing both the marginal and joint distributions. Therefore, the effectiveness of the uncertain parametric ROM methodology was substantiated, and the research objective was successfully achieved.
Committee
- Prof. Dimitri Mavris – School of Aerospace Engineering (advisor)
- Prof. Lakshmi Sankar – School of Aerospace Engineering
- Prof. Graeme Kennedy – School of Aerospace Engineering
- Dr. Christian Perron – School of Aerospace Engineering
- Dr. Kenneth Decker – SpaceWorks Enterprises