Session F06: Aeroacoustics

Data-informed dynamic acoustic source modeling in high-speed jets

We build on work by Zare, Jovanovic, and Georgiou (JFM, vol. 812, 2017) to develop dynamic noise models that account for the far-field acoustics of a Mach 0.9 subsonic turbulent jet. Given far-field time-averaged correlations of pressure we formulate an inverse problem to determine the forcing statistics to the linearized model that provide consistency with high-fidelity large-eddy simulation. Fluctuations in the near-field of the jet are projected to far-field sound via the Ffowcs Williams-Hawkings method. To reduce computational effort, we utilize the method of snapshots to obtain a reduced-order model of the input-output behavior of the jet, and design filters to provide an explicit dynamical representation of acoustic sources. Our data-driven approach reconciles linearized models of near-field fluctuations to acoustics at various observer angles by introducing dynamical modifications to the linearized operator. We conduct a frequency response analysis to examine the predictive capability of our model in capturing near-field coherent flow structures and far-field acoustic radiation.   

Delineation and evolution of aerodynamic and acoustic components of pressure in turbulence

To aid in jet noise source identification, several studies have attempted to isolate the propagating component of pressure from that which convects with structures. While the former contributes to farfield sound of the jet (acoustic pressure), the latter decays rapidly in the nearfield (aerodynamic or hydrodynamic pressure). Fourier- and wavelet-analyses usually used to split pressure in this manner rely on various pre-defined parameters such as wave-speed and statistics of coherent events, which introduce ambiguity in the analysis. The availability of accurate spatio-temporal data from high-fidelity simulations offers an opportunity to perform this separation in a more rigorous manner.  We propose a first-principle-based approach predicated on the availability of pure acoustic and hydrodynamic components of fluctuations in any turbulent medium. Specifically, the momentum potential theory is extended to derive Poisson equations which separate the acoustic and hydrodynamic components of pressure from the momentum equations. The splitting is successfully applied to a turbulent jet and yields insights into the spatio-temporal segregation that prevails among the constituent components of pressure fluctuations.   

Navier-Stokes Equations based Acoustic Analogy with Shock-Noise Prediction Example

We propose a new acoustic analogy that retains the Navier-Stokes equations as both the propagator and source terms. This approach relies on a new decomposition involving the base flow, aerodynamic fluctuations, and acoustic fluctuations. The aerodynamic fluctuating quantities are further decomposed into large-scale highly spatially coherent turbulence and fine-scale spatially incoherent turbulence. The resultant sources of sound are written as two-point cross-correlations involving each term of the Navier-Stokes equations. The linearized Navier-Stokes equations are solved for the vector Green’s function for the purpose of predicting the acoustic propagation. A closed-form equation for the spectral density of the fluctuating acoustic quantities is derived by a convolution integral involving the vector Green’s function and the two-point cross-correlation source term from Navier-Stokes equations. The closed-form equation is valid for acoustic fluctuations. We model the source terms using a combination of experimental data, source term analysis, and numerical simulation. We show a sample prediction for shock-associated noise using the dominant term in the source model for an off-design supersonic jet flow.   

Prediction of broadband trailing edge noise from a NACA0012 airfoil using DNS and WMLES

Motivated by the need to predict the broadband noise generated by the main propulsive fan of modern gas turbine engines, we study the broadband noise due to the turbulent flow on a NACA0012 airfoil at zero degree angle-of-attack, a chord-based Reynolds number of 408,000 and a Mach number of 0.1 using direct numerical simulation (DNS) and wall-modeled large-eddy simulation (WMLES).  We investigate the flow hydrodynamics and sound generation from the WMLES and examine its predictability compared with our DNS results, as well as available datasets for the same flow conditions. Verification of WMLES for such a canonical problem is crucial since it provides useful insight about the WMLES approach before using it for broadband fan noise prediction.  We discuss the resolution requirements necessary for the accurate prediction of wall surface pressures using WMLES.   

Reduction of Aeroacoustic Noise Generated from a Flow past a Cylinder by Porous Materials

It is important to reduce the aerodynamic noise generated from high-speed trains in order to achieve their further speed-up in the near future. In recent years, a new method for reducing aerodynamic noise by using porous materials has been proposed, and confirmed to be effective experimentally. In this study, by modeling the porous material by a large number of small cylinders, the flow around the cylinder covered by the porous materials is simulated by direct numerical simulation based on the modified volume penalization (VP) method. It is confirmed that the aerodynamic noise from the cylinder can be reduced by using porous materials. In particular, the aerodynamic sound reduction effect is significant when installing only one row of the small cylinders in the radical direction. In this case, depending on the number of the cylinders, the reduction rate of acoustic power becomes about  of the case of a bare cylinder. In contrast to the case of a bare cylinder, the Karman vortex shedding is delayed downstream by installing the small cylinders. In other words, vortex shedding in the wake is suppressed by using the porous material. The essential mechanism of the reduction of aerodynamic sound is that placing small cylinders makes the wake behind the cylinder nearly steady.   

Global response of cavity-flow shear layer to low and high amplitude acoustic forcing

Aeroacoustic instabilities in deep cavities subject to grazing flows are governed by the nonlinear response of the shear layer to transverse acoustic velocity forcing. We characterise experimentally the shear layer dynamics with and without acoustic forcing using time-resolved particle image velocimetry (PIV) and acoustic measurements. Considering the Navier-Stokes equations linarized around the mean flow from the PIV, we predict the harmonic response of the flow for a range of acoustic forcing amplitudes and frequencies.  A global stability analysis is also carried out, showing that the modes, which are marginally stable, define the hydrodynamic feedback responsible for the aeroacoustic instabilities. These analysis are performed for low Mach and highly turbulent conditions, for different cavity opening geometries, with sharp and round corners. These results allow us to explain the self-sustained aeroacoustic oscillations for a range of grazing flow velocities and predict the associated limit cycle amplitudes.