PhD thesis was titled “Applications of machine learning to finite volume methods”.

Continuing work in the propulsion analysis group. Nonlinear thermal-structural analysis, nonlinear buckling, modal analysis, random vibrations, fatigue, fracture, and fluid systems analysis.Major Achievements: Developed criteria to bypass a $50000 test typically performed every launch. Based on statistical analysis of past tests, expect ~87% of such tests to be bypassed going forward, which will save on the order of $1 million per year.Developed improved procedures for leakage analysis/testing that will streamline testing and provide better understanding of system capabilities.Improved standard repair procedures of various systems to streamline hardware production and reduce number of failed parts.Analyzed hardware to NASA-STD 5012 and NASA-STD 5019 to demonstrate flight capabilities.

I worked with the propulsion analysis group. I mainly focus on structural analysis. My work generally involves developing static and dynamic simulations of engine systems or components, and processing the results to determine whether or not the hardware will operate nominally under flight conditions.Other responsibilities include:1. Assessing parts produced outside of tolerances to determine whether they can still be flown safely.2. Processing test data to develop boundary conditions for simulations, or to validate simulation resultsI worked remotely, part-time from school during the fall and winter quarters.

Worked with a team to design and prototype a diesel fuel nozzle that offered significant advantages over currently available nozzles.

My research involves optimization of rocket engines.The first goal of my research is to derive a system of partial differential equations that governs three dimensional flow around an aerospike nozzle.After I have derived these equations, I will implement a computational solution in C++ that uses the method of characteristics and finite difference methods to solve the equations and assess the performance of a given aerospike contour.The CFD software I write to complete this task must be efficient enough to quickly and accurately iterate over millions of aerospike contours and select the best one based on specific impulse, thrust, and nozzle weight.

This research project involved theoretical/computational modeling of shockwaves in compressible flows. In engineering, shockwaves are approximated as infinitesimally thin. However, in reality, they occur over a finite width. When shock width is modeled, the most common approach is to use statistical methods. This project aims to develop an accurate deterministic/continuum model for shock structure and thickness.

I conducted a study on how telemetry data is ingested into ground systems for various subsystems on various missions, focusing on what differences existed between missions. I then used the results of the study to determine which parts of the data ingestion process could be universalized and how to do so. I specifically looked at the dictionaries that define telemetry data and spacecraft commands. Then came the main part of the task, which was designing the concept of a software tool that can translate mission-specific dictionaries to a multimission format. Because I finished this task early, I also built a dictionary translator prototype based on my design in Java. I was also assigned the task of studying classified mission dictionaries and translating them in a way that obfuscated any classified information while complying with the required format. I was able to use an adaptation of my own dictionary translator tool to accomplish this task. These sample dictionaries have since been released outside of the lab to others wishing to try AMPCS, a telemetry processing software package maintained by JPL.