Roland Ewert

RPM - the fast Random Particle-Mesh method to realize unsteady turbulent sound sources and velocity fields for CAA applications.

The simulation of broadband noise with CAA techniques is discussed. Unsteady broadband sound sources are modeled in the time-domain with a highly efficient computational method. The generated fluctuations reproduce very accurately autocorrelations and integral length-scales such as that provided by a RANS simulation of the time-averaged turbulent flow problem. It is argued that an approach based on synthetically generated sources has to be seen as an algorithmic extension of traditional statistical broadband methods in the frequency domain. Fluctuating quantities are generated by spatially filtering convective white-noise. The discrete realization of convective white-noise is based on random particles that are advanced using an area-weighted mean of the mean-flow from neighboring mesh points. The spatial filtering is realized by interpolating the random values with a particle shape function onto the neighboring mesh points and applying subsequently a sequence of 1D filtering operations. Sample results for different aeroacoustic applications of the method are presented.

CAA Slat Noise Studies Applying Stochastic Sound Sources Based On Solenoidal Digital Filters

The present paper studies the application of a low-cost CAA approach to a slat noise problem. A new fast and cheap stochastic approach is introduced to model the unsteady turbulent sound sources in the slat-cove. It is based on the spatial filtering of a random white-noise field and incorporates information about the integral length scale and the turbulent kinetic energy from a steady RANS computation. The stochastic method, which stems from the LES inflow boundary condition proposed by Klein et al. that reproduces firstand second-order one-point statistics, is extended in this paper for aeroacoustic applications. The extended formulation yields a perfectly solenoidal velocity field that is capable to reproduce exactly the complete second-order two-point correlation tensor of homogeneous isotropic turbulence. Results for the sound generation at the slat are given for the underlying RANS mean-flow field being based on a Menter SST turbulence model with and without Kato-Launder modification, respectively. The results for the modeled turbulent flow-field and the radiated acoustic field exhibit physical meaningful characteristics.

RANS/CAA based prediction of NACA 0012 broadband trailing edge noise and experimental validation

The prediction quality of a fast Computational Aeroacoustics (CAA) approach is studied for noise generated at a NACA 0012 trailing edge. Broadband spectra, the power law underlying their Mach number scaling, and the effect of Reynolds Number on the spectra are juxtaposed against published data from measurements and results from a semi-empirical prediction tool. The CAA method rests on the use of Reynolds Averaged Navier-Stokes (RANS) solutions to describe the turbulent flow problem around the airfoil. Acoustic Perturbation Equations (APE) are solved in the time domain, using a vortex source term, which is a function of turbulent field quantities. The acoustic approach was sccessfully validated in other works by utilizing turbulence data from Large Eddy Simulation (LES) to prescribe the unsteady sound sources. For the fast CAA approach applied in this work unsteady vortex sound sources are determined by a stochastic method, which generates 4D spatio-temporal synthetic turbulence that very accurately accompishes a local realization of all statistical and mean-flow features provided by steady RANS. Based on these prerequisites it becomes now feasible to make an assessment as to which acoustic accuracy can be achieved with such a hybrid RANS / CAA prediction method, bearing in mind the approximative nature and limited turbulence resolution of RANS:

Sweeping Sound Generation in Jets Realized with a Random Particle-Mesh Method

the sound spectrum in a manner that cannot be explained based on the classical Kolmogorov-theory of turbulence. But because of the lack of appropriate methods these findings could not be incorporated into the common statistical modeling approaches for realistic aeroacoustic problems using for example the acoustic analogy. In this article we develop an ecient Random Particle-Mesh method which enables us to model successfully the sweeping mechanism. We use it to simulate the noise radiation of a realistic subsonic jet. The results show that sweeping dynamics of turbulence is an important noise production mechanism in jets. Noise from the fine-scale turbulence can be predicted. There are furthermore indications that also Mach waves are generated.

Slat Noise Trend Predictions using CAA with Stochastic Sound Sources from a Random Particle Mesh method (RPM)

Slat Noise simulations are carried out for a high-lift airfoil. The high-lift airfoil geometry is a two-element slat-airfoil configuration without deployed flap in order to avoid additional sound sources at the flap. A low-cost CAA approach is applied, which is based on acoustic perturbation equations (APE) in the time-domain that are forced by stochastic sound sources. The stochastic model was introduced in AIAA 2005-2862 and is based on the spatial convolution of white-noise with a filter kernel and can reproduce target distributions of turbulent kinetic energy and length scales, e.g., provided by a steady RANS computation of the turbulent flow problem. The Mach number scaling law of the broadband slat noise component is evaluated based on three different freestream velocities (M=0.088, 0.118, 0.165). The capability of the stochastic method to reproduce a target turbulence kinetic energy topology is studied. The effect of model parameter variations on the quality of the reproduced turbulent kinetic energy distribution and its effect on the predicted far-field spectrum is evaluated. Only a small dependence on the model parameter is found. The effect of slat gap variation on the broadband acoustic far-field is studied for three different slat configurations, which encompass beside the reference slat position one configuration with reduced, and one with increased slat gap. Previous experimental findings at a full three-element high-lift airfoil indicated an effective dependence of the slat gap width on the acoustic far-field with typically 5dB reduction for a slat gap reduction of about 15\% due to a decrease of the velocity of about 25\%. The current study investigates this effect for the two-element high-lift configuration.

Application of a Discontinuous Galerkin Method to Discretize Acoustic Perturbation Equations

Airframe noise of complex geometries, like e.g. high-lift airfoil configurations, may conveniently be computed on unstructured grids. A Discontinuous Galerkin Method (DGM) provides a robust, high-order accurate discretization of systems of partial differential equations like for example the Acoustic Perturbation Equations (APE) even on this type of grid. In particular, a DGM based on Lagrange polynomials may be used, since it enables simple and cheap truncation of flux terms. The goal of the work reported herein is to verify such a DGM. Therefore, the sound field of a monopole situated in a laminar boundary layer is computed. Computations are stable, free from spurious numerical oscillations, and show very good agreement with other computations and with theoretical results.

Spectral broadening of jet engine turbine tones

The process of turbulent scattering is studied for the generic experiment conducted by Candel et al.1 applying an analytic weak scattering model and CAA computations. For the analytic weak scattering model an approximate form of the Lilley equation is used. The source terms of this equation are in terms of the turbulence and the incident acoustic field. In the CAA simulations the wave equation proposed by Pierce for sound in fluids with unsteady inhomogeneous flow is integrated. The unsteady turbulent base-flow is modeled using a stochastic method to generate turbulence with locally varying turbulence features as provided by time-averaged RANS. To study the spectral broadening effect analytically and computationally, the experimental set-up of Candel is considered, which involves an omnidirectional sound source, located on the axis of a round jet. The analytical predictions show very good agreement with the general trends as measured by Candel for an observer position normal to the jet axis. The computations reveal a spectral shape, which is in good agreement with those found in the experiments.

Trailing-Edge Noise Data Quality Assessment for CAA Validation

The paper provides a detailed reexamination of previously published trailing-edge (TE) noise experimental data as acquired in DLRs low-noise open-jet Acoustic Wind Tunnel Braunschweig (AWB). The objective is to set up a parametric reference data base to be later used for CAA validation purposes. A modular plate airfoil with variable chord length was used to investigate Reynolds number effects on both near field quantities and farfield sound. The original data base, including corrected farfield TE noise spectra as well as turbulent boundary-layer (TBL) characteristics, has been extended by unsteady surface pressure measurements in the TE region. Limitations and weaknesses of the applied measurement techniques are pinpointed by comparisons with independent data sets using alternative measurement approaches. Comparisons include also available semi-empirical prediction models for surface pressure spectra as well as for farfield TE sound emission. First RANS-based CAA prediction results using stochastic source models show essential agreement with the measurement data.

A CAA Based Approach to Tone Haystacking

The far-field noise spectra of jet engines show for certain jet configurations and turbine tones a characteristic spectral broadening effect, causing a reduction of tone peaks in favor of a more distributed spectral hump around each tone frequency. This haystacking effect likely occurs due to the interaction of the turbine tones with the unsteady turbulent jet shear layer. A better understanding of this effect may help to utilize it for noise reduction purposes. Furthermore, the effect is of interest for the measurement of tone sources in an open acoustic wind tunnel test section, since the tone will be scattered in the open jet shear layer. A correction for this measured broadening effect is desirable. A non-empirical computational approach to predict tone haystacking as a function of Reynolds number/jet shear layer characteristics is currently missing. This paper reports about ongoing work to utilize Computational Aeroacoustics (CAA) methods for the prediction of haystacking. In a first step CAA techniques are applied to simulate the propagation of tones through the time averaged steady exhaust of a jet engine. To simulate the haystacking effect with CAA, the unsteady turbulent base-flow is modeled with a 4D synthetic turbulence method. The employed RPM approach generates turbulence with all local statistical features as predicted by time-averaged RANS. To study the spectral broadening effect computationally, the experimental set-up of Candel is considered first, which involves an omnidirectional sound source located on the axis of a round jet. The analytical predictions show very good agreement with the general trends as measured by Candel for an observer position normal to the jet axis. The computations reveal a spectral shape, which is in good agreement with those found in the experiments. In a next step the methodology is combined with the exhaust problem to simulate sound propagation through the unsteady turbulent exhaust.

Broadband Trailing-Edge Noise Predictions - Overview of BANC-III Results

The Third Workshop on Benchmark Problems for Airframe Noise Computations, BANC-III, was held on 14-15 June 2014 in Atlanta, Georgia, USA. The objective of this workshop was to assess the present computational capability in the area of physics-based prediction of different types of airframe noise problems and to advance the state-of-the-art via a combined effort. This documentation summarizes the results from workshop category 1 (BANC-III-1) which focuses on the prediction of broadband turbulent boundary-layer trailing-edge noise and related source quantities. Since the forerunner BANC-II workshop identified some room for improvements in the achieved prediction quality, BANC-III-1 relies on the same test cases, namely 2D NACA0012 and DU96-W-180 airfoil sections in a uniform flow. Compared to BANC-II particularly the scatter among predictions for the DU96-W- 180 test case could be significantly reduced. However, proposed adaptations of previously applied computational methods did not systematically improve the prediction quality for all requested parameters. The category 1 workshop problem remains a challenging simulation task due to its high requirements on resolving and modeling of turbulent boundary-layer source quantities.