Direct numerical investigation and reduced-order modeling of 3-D honeycomb acoustic liners

Aero-acoustics;Computational Fluid Dynamics;Turbulence;Boundary Layer;Reduced-order Modeling;Large-scale Simulation;Noise Reduction

Zhang, Qi

关键词: Aero-acoustics;    Computational Fluid Dynamics;    Turbulence;    Boundary Layer;    Reduced-order Modeling;    Large-scale Simulation;    Noise Reduction;   

Others  :  https://www.ideals.illinois.edu/bitstream/handle/2142/49603/Qi_Zhang.pdf?sequence=1&isAllowed=y

美国|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

Single degree-of-freedom conventional acoustic liners are widely installed in jet engines to reduce internal engine noise.They work by converting acoustic energy into vorticity-bound fluctuations.Despite being widely used, effective design-stage models of acoustic liners placed in high sound amplitude conditions, possibly with a turbulent grazing flow, are not available due to the near-liner flow complexity and diagnostic challenges.The work presented in this thesis uses direct numerical simulations (DNS) of a compressible, viscous fluid to understand the inherent fluid mechanics and guide reduced-order-model development. While there are numerous orifices and cavities in a general conventional acoustic liner sample, only the one orifice and one cavity is investigated in this work. Resolved simulations of the sound-induced flow through a circular orifice with a 0.99 mm diameter are examined. The detailed investigations are split into two steps: the first step neglects any grazing flow. The no-flow simulation data identify the role the orifice wall boundary layers play in determining the orifice discharge coefficient which is an important indicator ofliner non-linearity.It is observed that when the liner behavior is not well described by linear models, the orifice boundary layers contain secondary vorticity generated from its separation from the corner on the high-pressure side of the orifice.Quantitative comparisons of the simulation-predicted impedance match available data for incident sound amplitude of 130 dB at frequencies from 1.5 kHz to 3.0 kHz.At amplitudes of 140 -- 160 dB the simulation impedance are in agreement with analytical predictions when using simulation-measured quantities, including the discharge coefficient and root-mean-square velocity through the orifice, although no experimental data for this liner exist at these conditions.The simulation data are also used to develop two time-domain models for the acoustic impedance wherein the velocity profile through the orifice is modeled as the product of the fluid velocity and a presumed radial shape, ξV(r).The models perform well, predicting the in-orifice velocity and pressure, and the impedance, except at the most non-linear cases where it is seen that the assumed shape V(r) can affect the back-plate pressure predictions.These results suggest that future time-domain models that take the velocity profile into account, by modeling the boundary layer thickness and assuming a velocity profile shape, may be successful in predicting the non-linear response of the liner.The second step introduces a grazing flow where the detailed interaction of an incident acoustic field and a Mach 0.5 laminar and turbulent grazing flow with a cavity-backed circular orifice is studied.All results are for tonal excitation at 130 dB from 2.2 -- 3.0 kHz, or at 3 kHz with 130 -- 160 dB acoustic amplitude. The results suggest that the liner experiences a drag increase over the baseline geometry with acoustic excitation and that facesheet shear stress measurements, while dominant at low acoustic amplitudes, contribute less at higher acoustic amplitudes. The DNS data further show that the orifice discharge coefficient can be semi-empirically modeled effectively using an acoustic-hydrodynamic scaling. The results indicate that experimental in situ impedance measurements can be contaminated by microphone-orifice interaction.Finally, the time-domain model without grazing flow was extended to include grazing flow by properly modeling the discharge coefficient and the turbulent boundary layer effect. Reasonable agreement of the liner impedance prediction was found with the DNS data.Discrepancies of the prediction suggest the future improvement of the model development.