Search

This paper addresses the numerical simulation of the acoustic propagation through an aerodynamic flow and its interaction with solid bodies. First, a general conservative formulation is derived from Euler's equations using a small perturbation hypothesis. In a second time, a discrete form is obtained with the help of high-order explicit schemes. Then, boundaries treatments and curvilinear or filtering aspects are examined. Finally, as an illustration and a validation of the E3P code that has been developed on these theoretical basis, an extensive set of numerical simulations is presented. It should be showing the wide range of possible applications that this tool, taken alone or coupled with others, could offer.

Marine propellers, operating al the aft end of vessels, interact with the turbulent boundary layer developed along the whole length of the hull. This interaction results in fluctuations of blade angle-of-attack, inducing lift fluctuations which are known to be sources of radiated broadband noise. The prediction of this noise requires the knowledge of turbulence features like velocity integral length scales or wavenumber-frequency spectra. The aim of the present experimental/numerical study is to complete such a broadband aeroacoustic prediction in the case of a propeller operating at the conical end of an axisymmetric body. A formulation of the propeller broadband radiated noise was derived from the dipolar term of the Ffowcs Williams-Hawkings equation. It uses the free field, uniform subsonic flow, acoustic Green’s function developed on its angular modes. It also includes the lift fluctuations induced on the blade sections, which may be classically written using a 2D Sears function, the propeller design (pitch, chord, skew, rake), and the wavenumber-frequency spectrum of the velocity component normal to the blade section. To address this velocity wavenumber-frequency spectrum, the body was mounted in the ONERA F2 windtunnel for turbulence characterization based on 2-point measurements using a cross hot-wire probe and a Laser Doppler Anemometer (LDA). This specific instrumentation provided cross-correlations for various spacings in the axial, radial and tangential directions, at several locations in the turbulent field ingested by the rotating propeller. These data were used to adjust an analytical model of the velocity cross-correlation and of the corresponding velocity wavenumber-frequency spectrum. In the last step, the same body and propeller were tested in the ONERA CEPRA 19 anechoic wind tunnel for acoustic characterization. The measured noise spectra were compared to the spectra predicted using the above turbulence measurements and formulation applied to the CEPRA 19 experiment conditions. Measured and predicted spectra are in good qualitative agreement but predicted noise levels slightly overestimate the measured values.

This paper describes an experimental program conducted by ONERA to study the wavenumberfrequency spectrum /Cf/2 , boundary layer thickness, displacement thickness and momentum thickness, Ax! hydrophones spacing, P (co) wall pressure frequency spectral density, P(oo) wall pressure frequency spectral density filtered by an hydrophone, AIAA96-1758 Op(co) acoustic component of OP(co), OP (k, co) wavenumber-frequency spectrum, OPlk,a>) acoustic component of Oplk,coj, \JL dynamic viscosity, cinematic viscosity v = }i/p, frequency, Strouhal numbers coSj/U^, co8/UM , hydrophone spatial sensitivity, fluid density, v CO co*,Q wall shear stress T_ = ju. —V2 x2=0

In the context of the on-going European project JTI-SFWA, the modal acoustic content inside a realistic fan duct, designed by Dassault-Aviation, must be characterized experimentally. The objective is to use Onera's CAA solver sAbrinA-V0 to design a non-intrusive sensor distribution inside the nacelle, upstream/downstream the fan plane, accounting for the heterogeneous mean flow and the complex internal geometry. By computing the propagation of several interacting modes in the duct, a region in which the acoustic pattern is close to the infinite duct analytical solution was characterized. Then several 3D distributions of pressure sensors were tested in this zone. Radial and azimuthal sensor arrays were located at the nacelle wall, whereas a radial sensor array was placed along the bifurcation leading edge, without additional intrusivity. The response of the 3D arrays to a given acoustic field, either analytical or propagated through CAA, was projected on an analytical modal basis. For several array configu...

Airframe noise, or noise generated by flow around the airframe (high lift devices, landing gears and cavities), is a major contributor to the overall noise of large commercial aircraft in the approach phase. For several years, ONERA has been involved in the development of methodologies for the numerical simulation of airframe noise. These hybrid methods generally combine (i) CFD computations for the near field flow, including unsteady CFD techniques such as LES, DES and NLDE, and (ii) acoustic (CAA) propagation techniques in the mid‐ and far‐field. A large range of acoustical methods is available for the propagation of sound waves in uniform or heterogeneous mean flows, either from the family of integral methods (Lighthill, FWH, BEM) or in the domain of (non‐linear or linearized) Euler equations in perturbation. The key challenges of these hybrid methodologie are (i) the ad‐hoc coupling between individual techniques and (ii) the availability of reliable and dedicated experimental database for the validati...

In the present work, a volume source model based on the SNGR approach, first proposed by Bailly, has been implemented and developed for jet noise prediction purposes. This method presenting the advantage of providing unsteady three‐dimensional turbulent data from a RANS computation, it shall be compatible with industrial development processes. First, the theoretical aspects of the original SNGR model were revisited. More particularly, analytical developments showed that considerable modifications were needed to take local flow effects into account while generating unsteady turbulent data. Moreover, the way to include anisotropy effects in the model, as originally proposed by Billson, was improved by using a nonlinear extension of Boussinesq approximation to recover the local Reynolds stress tensor. In a second step, the potentialities of the stochastic tool for jet noise simulations have been evaluated against existing numerical and experimental databases. In addition to that concern, its sensitivity to a...

Cylindrical cavities are mounted at some positions on fuselage and wings of commercial transport aircraft for various service functions. Such cavities were identified as potential airframe noise sources generating high intensity whistles at frequencies situated in the audible range of 0.5‐0.8 kHz. The present paper focuses on the numerical simulation of the far field noise generated by the unsteady flow passing over a cylindrical cavity with an aspect ratio (diameter/height) equal to unity, using a numerical methodology combining a LES and a Kirchhoff integration method. The LES computation was done using ONERA's cell‐center/finite‐volume/CFD code named FLU3M. A laminar boundary layer profile was applied upstream the computational domain, without forcing any turbulence. The LES results were coupled with a 3D Kirchhoff method in the frequency domain. The computed far field noise was compared to dedicated experimental data, showing very good agreement, especially regarding the emission frequency. The final ...

Most CAA applications rely on hybrid methods providing the radiated sound (using the Lighthill analogy, Kirchhoff integral or linearized Euler equations) from unsteady CFD. In that context, LES is widely considered as a compromise between full DNS and unsteady RANS. For two years, ONERA has undertaken studies aiming at understanding and applying LES in CAA. Former works focused on coupling LES to the Lighthill analogy. One typical application was the prediction of the noise generated by vortex shedding at the trailing edge of a thick flat plate. The incompressible LES was performed on a Cartesian mesh using the ONERA ‘‘pegase’’ code. The far‐field noise was computed using the Curle formulation and favorably compared to airfoil noise experiments. More recently, a theoretical study analyzed the contribution of the LES subgrid scale model, or cutoff wave number, to the noise production computed from the Lighthill tensor, concluding that modeling this contribution may be important to correctly estimating the ...