Philippe Druault

Generation of Three-Dimensional Turbulent Inlet Conditions for Large-Eddy Simulation

A method for generating realistic (i.e., reproducing in space and time the large-scale coherence of the flows) inflow conditions based on two-point statistics and stochastic estimation is presented. The method is based on proper orthogonal decomposition and linear stochastic estimation. This method allows a realistic representation with a minimum of information to be stored. Most of the illustrations of this method are given for a plane turbulent mixing layer that contains most of the basic features of organized turbulent flows. Examples of the application of the method are given first for the generation of inflow conditions for direct numerical simulation (DNS) and for large-eddy simulation from experimental results. Second, DNS results are used to generate realistic inflow conditions for two- and three-dimensional DNS, retaining only a minimum size of relevant information.

Proper orthogonal decomposition of in-cylinder engine flow into mean component, coherent structures and random Gaussian fluctuations

A snapshot proper orthogonal decomposition (POD) is performed from 2D time-resolved PIV measurements obtained in the tumble plane of a Spark Ignition engine flow. Based on this filtering approach, the in-cylinder flow field is decomposed into a mean part, a coherent part and a turbulent incoherent part. The analysis of the one-point statistical moments of orders 3 and 4 (skewness and flatness coefficients) as well as the analysis of the probability density function of the velocity field show that the POD extracts an incoherent random velocity field having Gaussian distribution properties. These small amplitude background fluctuations are also homogeneous. We then demonstrate the ability of the POD procedure, based on an energy criterion, in separating the coherent part and random Gaussian fluctuations. Based on this POD flow partitioning, providing a new triple decomposition of the instantaneous turbulent flow, the flow engine cycle-to-cycle variations associated with each velocity field contribution are then accessed. Thus, one of the first full field quantitative analyses of the cyclic variabilities associated with each part of the in-cylinder flow field is provided from the analysis of the corresponding POD temporal coefficients.

A coupled time-reversal/complex differentiation method for aeroacoustic sensitivity analysis: towards a source detection procedure

Defining and identifying the aeroacoustic sources in a turbulent flow is a great challenge especially for noise control strategy. The purpose of the present Study consists in proposing a new methodology to localize regions associated with sound generation. These regions are associated, in the present work, with those of high sensitivity of the acoustic field, using the heuristic argument that modifying the flow in these regions would lead to a very significant change in the radiated noise. The proposed method relies on the efficient coupling between the time-reversal theory applied to the Euler equations and the complex differentiation method to compute the sensitivity variable. To the knowledge of the authors, tills is the first time that the time-reversal technique is applied to vectorial hydrodynamic equations, in place of the classical scalar wave equation. Subsequently, regions associated with sound generation are related to spatiotemporal events which exhibit the maximum of sensitivity to acoustical disturbances measured in far field. The proposed methodology is then successively tested on three cases for which the nature of the Source is different: injection of mass, vibrating surfaces and flow instabilities arising in a plane mixing layer flow. For each test case, the two-dimensional Euler equations are solved using a numerical solver based on a pseudo-characteristies formulation. During these computations flow, variables are stored only at the Computational boundaries. These variables are time reversed and relevant information concerning the acoustical disturbances is tagged using complex differentiation in order to lead the sensitivity analysis. The same numerical solver is used to access the evolution of the time-reversed variables. In each test case, the proposed methodology allows to localize successfully zones associated with noise generation.

About the POD application for separating acoustic and turbulent fluctuations from wall pressure synthesised field

A turbulent flow developing over rigid surface generates two types of excitation loadings: a turbulent flow component that directly impacts the plate and an acoustic component that originates in turbulent flow. The analysis of these loadings is of particular interest especially for future noise reduction strategy implementation in an automobile context. To reproduce acoustic and turbulent fields, a synthesis of random pressure field is proposed thanks to Corcoss model and a diffuse acoustic field. One then proposes to evaluate the ability of proper orthogonal decomposition to discriminate both parts of this synthesised field. Such POD application demonstrates that it acts as a filtering technique in the wavenumber space. But in present study, the acoustic component is shared by multiple POD modes, and therefore cannot be extracted directly. It is also emphasised that such application depends on the frequency resolution and the energy repartition of the synthesised field.

On the Use of POD for Identifying Acoustic and Turbulent Components of an Inhomogeneous Wall Pressure Field

In automotive industry, the acoustic comfort is become of primarily interest. Car window which is one of the relevant acoustic transmission paths, has to support two distinct pressure loadings: a Turbulent Boundary Layer excitation (turbulent component) and the acoustic component related to acoustic waves generated in the flow. In this study, Proper Orthogonal Decomposition is applied to two wall pressure databases in order to identify each component (acoustic and turbulent) of available databases. First, an inhomogeneous synthesized wall pressure field is considered. Second, Lattice Boltzmann Method is used to simulate the turbulent flow around a realistic car. Then wall pressure field is extracted on a specific area of the car window and used to perform POD application. In each test case, it is observed that at high frequencies, low order POD modes are associated with acoustic component of the aeroacoustic wall pressure field. Indeed, the spatial footprint of the first POD modes is dominated by the largest wavelengths, detecting at the smallest frequencies and associated with the most energetic turbulent part. To validate such separation, the frequency-wavenumber spectrum analysis as well as statistical analysis are performed.© 2013 ASME

A time reversal method coupled with complex differentiation for the study of aeroacoustic sources

An original methodology for the localization of aeroacoustic sources has been recently developped. 1 This method is based on a coupled time reversal / complex differentiation approach. In this paper, the method is recalled. Then it is illustrated with two original cases: noise generation by mass injection and noise generation by flow instabilities.

Aeroacoustic analysis of cavity flows using Quadratic Stochastic Estimation coupled with Proper Orthogonal Decomposition

New mathematical post-processing tools are proposed to investigate the noise emission in the well known cavity flow. Based on the causality principle, Proper Orthogonal Decomposition and Quadratic Stochatic Estimation are successively implemented to access the aerodynamic events which contribute to the main noise emission in such a flow. Proper Orthogonal Decomposition is firstly used to extract selected flow structure events. Based on the knowledge of these events, the Quadratic Stochastic Estimation of the far acoustic pressure field is performed. Such methodology is successively applied to 2-D and 3-D numerical database of a flow over a cavity. In the 2-D test case, it is then demonstrated that POD can extract relevant velocity data information which can be associated with selected frequencies in the far acoustic field. Conversely, in the 3-D test case, it appears that the three dimensional flow structures may have a signature in the whole POD basis. Then, some difficulties may appear in interpreting the QSE recontructed pressure field using as reference data selected flow events deduced from the POD basis. However, the whole results show the great potential of the QSE procedure coupling to POD to analyze experimental databases.

Sensibility analysis and wave tracking of sound

Identi cation and characterization of aeroacoustic sources in turbulent ows are important scienti c challenges. Indeed, it is important to know the exact position and type of acoustic sources in order to reduce the noise generated. Numerical analysis of aeroacoustic noise generation and propagation often requires huge amount of data. Here, a versatile method called Complex Variable Method (CVM) is proposed. It is a powerful tool dedicated to analyze numerical simulations. It is shown that the CVM allows to make sensibility analysis or to track chosen part of waves propagating through the ow. Those theoretical results are illustrated with numerical simulations of noise generated in a ow passing over a cavity separated from the ow by an elastic structure. They show that the method can be e ciently used to get the sensibility to various parameters or to distinguish and follow the acoustic part in the aeroacoustic eld, even coupled to elastic structures.

Analysis of Door Mirror Aeroacoustic Source Using Extended Proper Orthogonal Decomposition

Based on a 3D simulation of the air flow around a vehicle, the Extended Proper Orthogonal Decomposition (EPOD) is employed in order to identify the existing relation between aerodynamic events around the door mirror (Ω domain) and the near acoustic pressure field (S domain). The aerodynamic pressure field on a 2D plan of Ω is decomposed using a classical formulation of POD. These modes are then sorted according to their correlation to the acoustic pressure field. The modes for which this correlation value is higher than a given threshold are selected. The proposed threshold determination is detailed. The selected modes are then employed in an EPOD procedure in order to determine their contribution to the acoustic pressure field. The aerodynamic pressure field reconstructed by these selected modes has a good correlation with the acoustic pressure field. This correlation is higher than the correlation between the aerodynamic pressure field reconstructed by the most energetic modes and the acoustic pressure field.Copyright