Séparation des composantes compressibles et incompressibles dans un champ de pression pariétale - Separability of the aeroacoustic and aerodynamic components of a wall pressure field

Séparation des composantes compressibles et incompressibles dans un champ de pression pariétale

Separability of the aeroacoustic and aerodynamic components of a wall pressure field

Jérôme I. Mars Jocelyn Guillon  Sebastian Miron  Vincent Grulier  Christophe Picard 

GIPSA-Lab, Dept. Images-Signal, UMR5216, 961 rue de la houille blanche,Domaine universitaire,BP 46, 38402 Saint Martin d'Heres Cedex 9

2CRAN - Faculté des sciences et techniques BP 239 - 54506 Vandoeuvre Cedex

3PSA PEUGEOT CITROËN, Route de Gizy 78943 Vélizy-Villacoublay Cedex

Corresponding Author Email: 
14 March 008
31 August 2008
| Citation



Research in car industry keeps trying to improve comfort in the passenger cells of vehicles by reducing noise pollution induced by wind turbulences in contact with the surface of the vehicle. Noise can be classified into three categories.

The first concerns noise generated by the engine and the gearbox. The second is the sounds produced by the wheels on the road. The third is related to the aerodynamic noise generated by the airflow. In recent years, several studies have been made to identify and to reduce noise and vibrations associated to the two first categories by including shock absorbers and insulating materials for example.

The most annoying noises for users are aerodynamic noise caused by airflow around the vehicle, visible from 100 km/h and dominant from 130 km/h. This flow generates noise in the cabin in many ways. It causes resonance phenomena from roof and/or doors [1], [2]. Noise can be produce from outside objects such as wipers, antennas or mirrors.

To reduce it, many studies have been proposed as experimental studies [3], [4], digital studies [5] and/or design forms [6]. Noise may come also from an external flow entering directly into the car. Apart from these effects due to the presence of obstacles in the airflow, there is a noise disturbance called aerodynamic noise form [7]. It represents a large part of the total aerodynamic noise. It is produced directly into the car by the fluctuating pressures generated by the boundary layer on the vehicle surface. These pressure fluctuations are transmitted inside the vehicle by different parts radiating in the car. Production of this noise on surfaces has a double origin. On the one hand, the vortex flow near glasses generates excess and depression of local non stationary pressure. This fluctuation is called thereafter aerodynamic component or turbulent field. On the other hand, local pressure fluctuations are spreading as an acoustic field through walls of the vehicle. This excitement is called acoustic component. Both are usually coupled excitations [8].

It is essential to identify and / or to characterize each magnitude.

Recent work has highlighted the involvement of acoustic waves created by turbulent flow. Studies on the identification of these two types of noise are mainly conducted through experimental prototype vehicles wind tunnel [9], [10], [11]. In order to take into account the noise from the design phase of a vehicle, efforts will now focus on modeling the aero acoustics mechanisms noise and its transmission inside the vehicle. The recent work cited above is based either on the pressure field analysis of turbulent component (incompressible field) or on the analysis of far-field acoustic (sound propagation).

In order to describe the noise, this work shows the necessity to explore the sound field near the wall. To do that, with the Aero acoustics Department at PSA Peugeot Citroen, a network of sensors has recorded the two components of aerodynamic and aeroacoustic pressure field on a wall as the glass door or the windshield. This acquisition is close to real experiment included door vehicle with its rear-view mirror.

The purpose of this study is to propose some multidimensional signal processing methods to analyze and separate the two fields. From previous works done in the field [11], we propose some separation filters taking into account the spatial dimension of the fields. After presenting the context of the study, an adaptation of 3D FK filter is first performed to separate each fields. We show that filter is efficient only on high frequencies content.

In order to take advantage of the geometry of the acoustical field, a spatial transformation is applied to the signal. Then to allow multicomponent filter methods based on singular value decomposition [12], we propose this spatial transformation to produce plane wave field. This type of multicomponent filter is typically applied on 2D data. Many applications exist in various topics as biomedical imaging, geophysics, underwater acoustics, remote sensing, etc.. [13], [14], [15], [16].

The principle of this filter is, to find a space in which data are presented in a dimension as low as possible. The SVD is classically used to decompose the initial dataset into two complementary sub-spaces called signal space and noise space. Due to the multidimensionality of data (x, y, and time), in the last part of this article, we show that SVD can be extended to three-dimensional data through of 3D-SVD [17]. These data resorted in cube are obtained after a transformation from an Euclidean coordinates (x, y, time) to polar coordinates (ρ, θ, time). After applying this transformation, the acoustic field is expressed as planes wave while the turbulent wave field has no particular spatial coherency. The effectiveness of this method 3D-SVD will be discussed on the separation of the two fields (aerodynamic and aero acoustic).


Dans l'industrie automobile, les recherches actuelles intégrent de plus en plus la notion de confort dans l'habitacle du véhicule. Des études sont ainsi réalisées pour améliorer l'environnement du chauffeur et de ses passagers en limitant notamment les nuisances sonores dues à l'écoulement de l'air à la surface du véhicule. Dans ce contexte, et dans le but de faciliter l'analyse de ce type de nuisance, nous étudions la séparabilité des composantes d'origine aéroacoustique d'une part et aérodynamique d'autre part, présentes dans un champ de pression pariétale (au contact d'une paroi). Les deux composantes étant enregistrées sur un réseau de capteurs, nous montrons que les méthodes classiques de filtrage de type FK - 3D sont inefficaces dans les basses fréquences pour séparer les deux composantes. Nous proposons donc deux méthodes de filtrage matriciel multicomposante (3D-SVD notamment) permettant d'obtenir une séparabilité de ces deux composantes après une transformation spatiale des données.



Acoustical field, Turbulent wavefield, Wavefield separation, SVD-ICA method, 3D-SVD filtering

Mots Clés

Champ acoustique et turbulent, Séparation de champ d'ondes, filtrage SVD-ICA, filtrage 3D-SVD

1. Introduction
2. Présentation Des Données
3. Étude Multicapteur
4. Présentation Des Traitements 3D
5. Conclusions

[1] D. RICOT, Simulation numérique d’un écoulement affleurant une cavité par la méthode Boltzmann sur réseau et application au toit ouvrant de véhicules automobiles, PhD thesis, École Centrale de Lyon, 2002.

[2] J. MILBANK and S. WATKINS, Tonal noise from small cavities on motor vehicles and a case study of a whistling mirror, Proc. 7th International Congress on Sound and Vibration, GarmischPartenkirchen, Germany, 2000.

[3] G.S. KIM, H.K. PARK, S.G. JUNG, and K.D. IH, Development of acoustic holography and its application in hyundai aeroacoustic wind tunnel, Society of Automotive Engineers International Noise and Vibrations Conference and Exposition, Grand Traverse, MI, USA, (Paper 1497), 2001.

[4] V. LINET, J.J. LASSERRE, C. VARTANIAN, et J.C. BOUEILH, Exploitation des mesures piv pour l’aérodynamique et l’aéroacoustique automobile, Proc. 8ème Congrès Francophone de Vélocimétrie Laser, 2002.

[5] S. OGAWA and T. KAMIOKA, Review of aerodynamic noise prediction using cfd, Society of Automotive Engineers International Congress and Exposition, Detroit, MI, USA, pages Paper 1999-01-1126, 1999.

[6] W. WEI and R. WALKER, Optimization of mirror angle for front window buffeting and wind noise using experimental methods, Society of Automotive Engineers International Congress and Exposition, Saint Charles, Il, USA, pages Paper 2007-01-2401, 2007.

[7] J.Y. HER, M. LIAN, J.J. LEE, and J. MOORE, Experimental assessment of wind noise contributors to interior noise, Society of Automotive Engineers International Congress and Exposition, page Paper 971922, 1997.

[8] P.G. BREMMER and J.F. WILBY, Aero-vibro-acoustics : Problem and statement and methods for simulation-based design solution, 23th American Institute of Aerodynamics and Astronautics Conference, Breckenridge, Co, USA, June 2002, page 2251, 2005.

[9] P. IBARS, Contribution des petits nombres d’onde en champ pariétal de pression dans une couche limite turbulente bidimensionnelle, Comparaison de différentes techniques expérimentales, PhD thesis, École Centrale de Lyon, 1990.

[10] B.M. ABRAHAM and W.L. KEITH, Direct measurement of turbulent boundary layer wall pressure wavenumber-frequency spectra, Journal of Fluid Engineering, 120: 29-39, 1998.

[11] B. ARGUILLAT, Étude expérimentale et numérique de champs de pression pariétale dans l’espace des nombres d’ondes, avec application aux vitrages automobiles, PhD thesis, 2006.

[12] G.H. GOLUB and C.F. VAN LOAN, Matrix computation, 3rd Ed, Johns Hopkins, 1996.

[13] D.P. BERRAR, W. DUBITZKY, and M. GRANZOW, A Practical Approach to Micro-array Data Analysis, Springer ed., 2003.

[14] R. DE FRANCO and G. MUSACHIO, Polarization filter with singular value decomposition, Geophysics, 66(3): 932-938, 2001.

[15] V.C. KLEMA and A.J. LAUB, The singular value decomposition: its computation and some applications, IEEE. Trans on Auto. Control, 25(2): 164-176, 1980.

[16] T.B. NIELSEN and E.K. WESTWOOD, Extraction of acoustic normal mode depth functions using vertical line array data, JASA, 111(2): 748-756, 2002.

[17] L. DE LATHAUWER, Signal processing based on multilinear algebra, PhD thesis, Leuven Katholic University, 1997.

[18] J.R. CALLISTER and A.R GEORGE, Wind Noise, In Aerodynamics of Road Vehicles, 4th Edition, Edited by Wolf Heinrich Hucho (Society of Automotive Engineers International), 1995.

[19] F. PERROT, Calcul du rayonnement acoustique d’écoulements turbulents basé sur des analogies acoustiques aux simulations aérodynamiques instationnaires, PhD thesis, École Centrale de Lyon, 2004.

[20] B. ARGUILLAT and D. RICOT, Measurement of the wavenumberfrequency spectrum of wall pressure fluctuations under turbulent flows, 26th American Institute of Aerodynamics and Astronautics Conference, Monterey, Ca, USA, pages 1-18, 2005.

[21] V. GRULIER, S. DEBERT, J.I. MARS and M. PACHEBAT, Acoustic and turbulent wavenumbers separation in wall pressure array signals using emd in spatial domain, International Conference on Acoustics, Speech and Signal Processing, ICASSP, pages 1-4 Avril, Las Vegas, 2008.

[22] J. F. CARDOSO and A. SOULOUMIAC, Blind beamforming for non-gaussian signals, IEE Proc.-F, 140(6): 362-370, 1995.

[23] V. VRABIE, N. LE BIHAN and J.I. MARS, 3d-svd and partial ica for 3d arrays sensors, 72nd Meeting of Society of Exploration geophysicists, Salt Lake City, Expanded Abstract, 2002.

[24] V. VRABIE, J.I. MARS and J.L. LACOUME, Modified singular value decomposition by means of independent component analysis. Signal Processing, 84(3): 645-652, 2004.

[25] N. LE BIHAN, Traitement algébrique des signaux vectoriels. Application en séparation d’ondes sismiques, PhD thesis, Institut National Polytechnique de Grenoble, 2001.

[26] N. LE BIHAN and G. GINOLHAC, Three-mode dataset analysis using higher order subspace method: application to sonar and seismoacoustic signal processing, Signal Processing, 84(5) :919–942, 2004.

[27] N. LE BIHAN, V. VRABIE and J.I. MARS, Multidimensional seismic data decomposition by higher order SVD and unimodal-ICA, Signal and Image Processing for Remote Sensing by C. H. Chen. CRC Press (Taylor and Francis books), 2006.

[28] P.M. KROONENBERG, Three-mode principal component analysis, DSWO Press, 1983.

[29] L. ELDEN, Matrix methods in data mining and Pattern Recognition, Series fundamentals of algorithms. SIAM, Cambridge University Press, 2007.