Advisor

Sescu, Adrian

Committee Member

Janus, J Mark

Committee Member

Bhushan, Shanti

Committee Member

Doty, Michael

Date of Degree

12-1-2020

Document Type

Dissertation - Open Access

Degree Name

Doctor of Philosophy

College

James Worth Bagley College of Engineering

Abstract

The noise generated by jet engines represents a significant environmental concern that still needs to be addressed. Accurate and efficient numerical predictions are a key step towards reducing jet noise. The current standard in highidelity prediction of jet noise is large eddy simulation (LES), which resolves the large turbulent scales responsible for the low and medium frequency noise and models the smallest turbulent scales that correspond to the high frequency noise. While LES requires significant computational resources to produce an accurate solution, it fails to resolve the noise in the high frequency range, which cannot be simply ignored. To circumvent this, in this dissertation the Coupled LES-Synthetic Turbulent method (CLST) was developed to model the missing frequencies that relate to un-resolved sub-grid scale fluctuations in the flow. The CLST method combines the resolved, large-scale turbulent fluctuations from very large eddy simulations (VLES) with modeled, small-scale fluctuations from a synthetic turbulence model. The noise field is predicted using a formulation of the linearized Euler equations (LEE), where the acoustic waves are generated by source terms from the combined fluctuations of the VLES and the synthetic fields. This research investigates both a Fourier mode-based stochastic turbulence model and a synthetic eddy-based turbulence model in the CLST framework. The Fourier mode-based method is computationally less expensive than the synthetic eddy method but does not account for sweeping. Sweeping and straining of the synthetic fluctuations by large flow scales from VLES are accounted for in the synthetic eddy method. The two models are tested on a Mach 0.9 jet at a moderately-high Reynolds number and at a low Reynolds number. The CLST method is an efficient and viable alternative to high resolution LES or DNS because it can resolve the high frequency range in the acoustic noise spectrum at a reasonable expense.

URI

https://hdl.handle.net/11668/20824

Sponsorship

Center for Advanced Vehicular Systems, Tohoku University, Mississippi Space Grant

Share

COinS