Large-Eddy Simulation of Jet Noise Using the Energy Cascade
Open Access
- Author:
- Mikkelsen, Dana
- Graduate Program:
- Aerospace Engineering
- Degree:
- Master of Science
- Document Type:
- Master Thesis
- Date of Defense:
- November 01, 2024
- Committee Members:
- Philip John Morris, Thesis Advisor/Co-Advisor
Daning Huang, Committee Member
Amy Pritchett, Program Head/Chair
Eric Greenwood, Committee Member - Keywords:
- Large Eddy Simulation
LES
Computational Fluid Dynamics
CFD
Aeroacoustics
Jet Noise
Turbulent Energy Spectrum
Energy Cascade - Abstract:
- The use of large-eddy simulation (LES) techniques for jet noise problems has been a subject of discussion for many years due to their ability to capture the turbulent structures necessary for acoustic analysis. Historically, LES has been prohibitively computationally expensive for use beyond simple configurations; however, recent strides made in the area as well as the introduction of GPU-accelerated solvers, which improve the turn-around for these intensive calculations, are bringing the use of LES into the realm of studies in industry. This thesis investigates a method of grounding LES grid generation in the turbulent energy spectrum to eliminate the trial-and-error approach to the construction of a high-fidelity model. In this method, the mixing, Taylor microscale, and Kolmogorov microscale lengths were extracted from a Reynolds-averaged Navier Stokes (RANS) simulation conducted on an arbitrarily fine grid. These length scales were then used to calculate the grid size necessary to resolve 80% of the total kinetic energy of a LES. The velocity and turbulence statistics obtained from the LES developed using this method showed good agreement with experimental data. The acoustic data showed good agreement with experimental data as well, validating the grid developed using the energy spectrum. This thesis then suggests a method of normalizing acoustic data from two nozzles to an equal thrust basis to make such comparisons more practical for industry application. Using the same grid, a second RANS and LES calculation were run at a slightly lower nozzle pressure ratio (NPR) to simulate losses in the nozzle that could result from a change in geometry. The thrust values for both the original nozzle and the second nozzle were calculated from the RANS runs. Then, it was determined at which velocity the second simulation should be run to match the thrust value of the first simulation. Acoustic data were obtained from the second LES and adjusted using an acoustic scaling law to recover the decrease in noise levels due to the ‘losses’ in the nozzle. The adjusted acoustic data showed good agreement with the original acoustic data from the first LES, suggesting the validity of this method.