Department
Mechanical and Aerospace Engineering
Major
Aerospace Engineering
Research Advisor
Du, Xiaosong
Advisor's Department
Mechanical and Aerospace Engineering
Abstract
Machine learning and optimization are the stepping blocks in broad engineering areas, such as computational fluid dynamics (CFO) and CFO-based design optimization. Using the CFO-based optimization framework, MACH-Aero, we were able to use the SLSQP optimizer and compared the results with a reference case completed by the SNOPT optimizer. Results verified that our SLSQP optimizer was working properly. The solver created showed that our objective function had a difference in magnitude of 0.000019 allowing us to conclude that our solver was working as intended. However, there were several failed attempts before reaching a successful run that gave us a satisfactory result. Further work was done to attempt to make a solver that can handle cases of similar types to the failed one. Efficient global optimization will eventually be paired with the CFO solver, which is expected to result in successful runs using the previous unsuccessful conditions while alleviating computational costs.
Biography
Aaron Spillars is a senior undergraduate student studying Aerospace Engineering. Last summer he joined Dr. Du's research team "Phlai Lab" and began working on airfoil optimization. While not working on research, he is a member, and secretary, of the volleyball club, and can be found working at the student rec center. Once graduated, Aaron would like to find a job involving failure analysis.
Research Category
Engineering
Presentation Type
Poster Presentation
Document Type
Poster
Location
Innovation Forum - 1st Floor Innovation Lab
Presentation Date
10 April 2024, 1:00 pm - 4:00 pm
Using Efficient Global Optimization with CFD for Aerodynamic Optimization
Innovation Forum - 1st Floor Innovation Lab
Machine learning and optimization are the stepping blocks in broad engineering areas, such as computational fluid dynamics (CFO) and CFO-based design optimization. Using the CFO-based optimization framework, MACH-Aero, we were able to use the SLSQP optimizer and compared the results with a reference case completed by the SNOPT optimizer. Results verified that our SLSQP optimizer was working properly. The solver created showed that our objective function had a difference in magnitude of 0.000019 allowing us to conclude that our solver was working as intended. However, there were several failed attempts before reaching a successful run that gave us a satisfactory result. Further work was done to attempt to make a solver that can handle cases of similar types to the failed one. Efficient global optimization will eventually be paired with the CFO solver, which is expected to result in successful runs using the previous unsuccessful conditions while alleviating computational costs.
