Ph.D. Proposal
Brenton J. Willier
(Advisor: Prof. Dimitri Mavris)
"A Methodology For Quantifying High-Dimensional Unsteady Field Uncertainty With An Entry Vehicle Application"
Friday, July 19
9:00 a.m.
Collaborative Visualization Environment (CoVE) Weber SST II
And
Abstract
The design of blunt-body entry vehicles involves optimizing their shape to maintain a balance between payload deceleration limits, vehicle heating, and landing accuracy during the Entry, Descent, and Landing (EDL) phases. Simulations are critical for designing craft with minimal access to flight tests. NASA has identified multiple gaps in the existing modeling capabilities. This includes models with limited uncertainty knowledge, the need to evolve models for advanced vehicle designs, and limited low Mach number aerodynamic models.
Currently, aerodynamic models center around a lookup database comprised of force and moment coefficients compiled from physical and computational tests, such as wind tunnels and Computational Fluid Dynamics (CFD). At low Mach numbers, the aerodynamics of blunt body entry vehicles are driven by complex, chaotic behaviors of the unsteady recirculating wake. This unsteady behavior directly leads to uncertainty in the aerodynamic loading, which can be further propagated to the flight trajectory of the vehicle. The current state-of-the-art for quantifying the uncertainties found in EDL analysis within an aerodynamic database involves using conservative adders and multipliers on scalar coefficients based on derived quantities and historical engineering intuition. Limitations in grounding the aerodynamic uncertainty in physical phenomena impact the uncertainty quantification process for entry vehicles, leading to extraneous vehicle structure weight and lowered accuracy in landing predictions.
In recent years, advances in coupled CFD-Rigid Body Dynamics (RBD) frameworks have significantly improved flight simulation capabilities. In these systems, a high-fidelity CFD model computes vehicle forces and moments, which are passed to a trajectory module to compute the next step in the vehicle's flight. While CFD computational performance has improved significantly, the number of cases required to produce a meaningful sample for an uncertainty quantification analysis is still intractable. This time restriction has led researchers to implement various surrogate models, like Reduced Order Models (ROMs), to quickly generate force and moment information for trajectory simulations.
ROMs seek to reduce the complexity of high-dimensional data by identifying dominant features and projecting them into a smaller set of generalized latent coordinates. Parametric ROMs train data fit models to predict new latent coordinates for unseen states. These types of ROMs have been successfully applied to predict aerodynamic coefficients for blunt body entry vehicles in free flight; however, the uncertainty due to the unsteady data was not quantified. When training with unsteady data, a shift is required to advance the state-of-the-art parametric ROMs away from deterministic predictions of scalar quantities toward uncertain scalar predictions.
A methodology is presented to capture, encode, and propagate the uncertainty from unsteady high-dimensional fields using parametric reduced order models. Multiple gaps were identified in the fundamental steps used to construct the methodology. A set of research questions, hypotheses, and experiments are posed that will address the identified gaps. The methodology experimentation and demonstration are rigorously applied to unsteady entry vehicle free-flight data from high-fidelity CFD as well as an unsteady Lorenz model. Ultimately, this research will make significant contributions in the disciplines of high-dimensional field predictions, uncertainty quantification, and the capture of unsteady CFD effects.
Committee
- Prof. Dimitri Mavris – School of Aerospace Engineering (advisor)
- Prof. Lakshmi Sankar – School of Aerospace Engineering
- Prof. Graeme Kenndy – School of Aerospace Engineering
- Dr. Christian Perron – School of Aerospace Engineering
- Dr. Kenneth Decker – SpaceWorks Enterprises