September 2024 - Present

The BYU Hypersonics Lab is actively working on a project from the University Consortium of Applied Hypersonics in conjunction with the Joint Hypersonics Transition Office.

In my work, I am exploring different surrogate modeling techniques of high-fidelity hypersonic CFD data of the Orion Reentry Vehicle. The Orion is simulated ranging from Mach 10 to Mach 30 and at varying angles of attack. Hypersonic CFD is very difficult with highly complex physics to be captured. Once a successful simulation is converged, this data can be used to trian surrogate models.

Traditional surrogate modeling techniques use dimensionality reduction to tranform high dimensional data (like CFD with millions of degrees of freedom at times) into a simpler latent representation with only tens of degrees of freedom. There are many techniques to reduced the dimensionality. In my work, I am employing machine learning methods to reduce the CFD data by using a Convolutional Autoencoder. This technique leverages the powerful nonlinear capabilities of deep learning methods and allows it to be utilized with traditional surrogate modeling techniques that have been used for decades.