Andrej Neifeld

Jet Mixing Noise from Single Stream Jets using Stochastic Source Modeling

This work deals with the simulation of jet mixing noise in the time domain using stochastic source modeling. The sound sources are generated by means of the RPM (Random Particle M esh) method, which uses turbulence statistics gained from RANS data. The generated stochastic sound sources closely realize the two-point cross-correlation function proposed by Tam & Auriault (T&A) to describe the statistics of a fine-scale jet mixing noise source. By modeling the sound source in the time domain a direct (primal) prediction method of the T&A approach is realized. The methodology followed in this work allows to evaluate noise spectra at any position in the computational domain based on one CAA computation. The aspired goal is to prove numerically the ability of our time-domain model to give similar predictions as the genuine T&A approach. The realization of an eight power Mach number scaling law for the emitted sound is verified. Furthermore, it is demonstrated that jet similarity spectra are obtained with a time marching CAA code. At 90◦ to the jet axis good agreement with the G-Spectrum is found. The stochastic approach is slightly modified to enable Strouhal similarity for the peak level of the jet-noise spectra at different jet velocities. The appropriate scaling behavior is demonstrated numerically. The spectra are realized in a Mach number range between 0.3 and 0.9 for Strouhal numbers ranging from 0.01 to 10. The RANS solution to a single stream cold jet configuration is obtained for a set of subsonic Mach numbers using the DLR solver TAU.

A 3-D modal stochastic jet noise source model

In previous work the jet mixing noise model of Tam & Auriault was realized as a time marching method in 2-D using a stochastic sound source, generated by means of the RPM (Random Particle M esh) method. The generated stochastic sound sources closely realize the two-point cross-correlation function proposed by Tam & Auriault to describe the statistics of a fine-scale jet mixing noise source. In a first attempt to extend the prediction capability to 3-D, a simplified radial scaling function was introduced to use the 2-D sound sources for axisymmetric computations of azimuthal mode order m=0. In the current work a strictly derived 3-D stochastic noise source is introduced, based on an azimuthal mode decomposition. For azimuthal mode order m=0 clear differences to the previously used simplified scaling model are obtained for sources with small relative distance to the jet axis. The azimuthally decomposed 3-D stochastic noise sources are shown to be generated by mutually uncorrelated 2-D stochastic sources for each mode order. Since for jet noise only the first few azimuthal modes are essential to describe the acoustic far-field radiation, a highly efficient 3-D broadband jet noise model for jet mixing noise is obtained at the computational price of a 2-D computation. For a 2-D simulation the capability of RPM to properly prescribe sound spectra is verified with an analytical solution. Sample results that highlight the proper functioning of the 3-D modal source model are presented.

Linear- and Non-Linear Perturbation Equations with Relaxation Source Terms for Forced Eddy Simulation of Aeroacoustic Sound Generation

Turbulence related sound is generated by the dynamics of fluctuating vorticity. For example, trailing edge noise is caused by vorticity traveling past the trailing edge. To excite fluctuating vorticity by forcing the linearized Euler equations (LEE) with right-hand side source terms, one peculiar problem is observable: while the rise of vorticity levels by external sources poses no problem, to properly lower them, the right-hand side terms must act as a sink, being exactly in anti-phase to the vorticity levels as present in the LEE solution. However, the accurate prediction of vorticity in terms of phase cannot be guaranteed, especially for approximately modeled sources e.g. using stochastic methods. Thus in general there will be a mismatch between actual induced and intended levels of vorticity. In this paper a new class of relaxation source terms is introduced that enables the proper excitation of vorticity levels in linear and non-linear perturbation equations and as such enables an accurate control over the vorticity magnitudes. The source can be formulated to act selectively in wave-number space, i.e. without directly affecting the dynamics of resolved low wave-number vorticity components whereas the resolved high wave-number part is piloted by the fluctuating vorticity imposed as a reference solution. The reformulation of the Navier-Stokes equations in primitive variables and non-linear perturbation form is presented. Direct noise computation of sound radiated from a vortex shedding cylinder in laminar cross flow verify their implementation. The relaxation source term without forcing is applied to the unstable jet problem of the 4th CAA Workshop on Benchmark Problems. The forcing of frozen and decaying stochastic turbulence in conjunction with the relaxation source term is studied. First results for high-lift noise prediction with forced eddy simulation are presented.

Jet Noise Prediction with Eddy Relaxation Source Model

Previously, a hybrid CFD/CAA approach has been applied for jet noise prediction utilizing a stochastic realization of the Tam & Auriault (T&A) and the Tam, Pastouchenko and Viswanathan (TPV) source models. These models describe two-point cross-correlation functions of a mixing noise source in the jet shear-layer. All input data needed for the modeling can be derived from RANS. The uctuating acoustic sources are generated stochastically by the Fast Random Particle-Mesh (FRPM) method. For sound propagation a CAA code PIANO is applied with linearized or non-linearized Euler equations in perturbed form. In comparison to measurements, CAA results with the T&A and TPV source models have proven a relatively high accuracy for jet mixing noise prediction of dierent isolated nozzle congurations (single/dual stream jets, static and forward-ight conguration, hot/cold).

Towards Prediction of Jet Noise Installation Effect using Stochastic Source Modeling

Previously, the prediction of jet mixing noise from hot and cold isolated nozzle configurations has been studied with the means of RPM (Random Particle M esh) method using the Tam & Auriault or rather Tam, Pastouchenko and Viswanathan source model. The acoustical sources are generated as a reproduction of experimentally observed two-point cross-correlation functions using as input the turbulence statistics from RANS data. The sound propagation and extrapolation are performed with the linearized Euler equations in the near-field and with the Ffowcs-Williams & Hawkings method in the far-field. Efficient CAA computations are achieved exploiting the advantage of Fourier series decomposition for isolated axisymmetric and quasi-axisymmetric nozzle geometries. For investigation of installation effects the azimuthal-modal approach is however not applicable due to highly three-dimensional flow interactions, e.g. if a jet engine is mounted below an airfoil. Therefore, in contrast to the isolated nozzle configurations, in this case the resolution of computational domain in all three spatial dimensions is needed. Next to the RPM method, which is solely capable to generate two-dimensional acoustical sources, the FRPM method is available that is able to generate both, 2-D and 3-D sources. First test computations with FRPM method were conducted for a single stream jet, whereas the azimuthal-modal results of RPM were used as reference solution. The installed jet configurations are subsequently studied building on the FRPM computations of the single stream jet.

Prediction of Hot Jet Mixing Noise Using Extended Stochastic Source Correlations

The prediction of jet mixing noise from a stochastic realization based on the Tam & Auriault model was previously studied for cold jet configurations. To generate the acoustical sources the Random Particle M esh method (RPM) was applied, which uses turbulence statistics from RANS data. The generated stochastic sound sources closely realize twopoint cross-correlation functions of the fine-scale jet mixing noise model. In this work the RPM method is extended to realize besides the cross-correlations of the cold jet noise model of Tam & Auriault also those of the hot jet noise model proposed by Tam, Pastouchenko and Viswanathan. For the stochastical realization of the latter model, a recently proposed three-parameter Langevin procedure is utilized to reproduce the cross-correlation functions. The RPM code is combined with the DLR CAA solver PIANO. Similarly to the cold jet noise computations, for hot jet noise computations we use the azimuthal-modal decomposed Linearized Euler Equations (LEE). The combination of stochastic source modeling with an azimuthal formulation allows to predict efficiently in the near-field the fine-scale jet noise spectra at any position. For the prediction of far-field noise spectra the modal Ffowcs-Williams and Hawkings method is conducted subsequently to the CAA computations. In the second part of this paper, the ability to simulate large scale noise by means of linearized Euler equations, including mean-flow gradient terms that allow for the onset of hydrodynamic instabilities, will be discussed. Finally, the CAA results of dual-stream jet computations with nozzle geometry variations are discussed.

Hybrid RANS/CAA Computation and Validation of A320 V2527 engine at ground operation

In the framework of SAMURAI project the DLR’s research aircraft ATRA (Advanced Technology Research Aircraft), which is an Airbus A320 with the V2527 engine, has been studied at ground operation. Several DLR institutes have been involved in this project dealing with experimental and numerical investigations on structure, aerodynamics and aeroacoustics. This paper is presenting the aeroacoustical results of jet noise measurements in the first part that are used subsequently for the validation of numerical prediction with CAA (Computational AeroAcoustics) computation, which are discussed in the second part. The numerical prediction of radiated sound has been conducted in time domain with a hybrid RANS/CAA approach. This approach requires a precomputed RANS solution (solved at least with two-equation turbulence model) as input, which is provided in this case as a half-model of A320 aircraft with all relevant airframe details (except landing gears) above solid ground. The experimental data describes acoustical radiation of only one operating engine, while the second is deactivated. Hence, the RANS solution of a half-model is representing adequately the experimental setup, whereby the CAA computation is considering also only the spatial domain in the direction of placed microphones. In the post-processing step, the microphone array in the near-field and single microphones in far-field measurements have been used for the validation of CAA results. The evaluation of the numerical results revealed a relatively good agreement with the experimental data. In general, the both data sets are matching in terms of spectra form, the position of spectra in frequency region and in sound pressure level magnitudes. At certain frequency bands, some deviations are observable due to several assumptions in the numerical setup. Altogether, we can state that the principal acoustical effects of this configuration are numerically reproduced. The acoustical properties such as the ground reflections, jet flow refraction and boundary layer influence from the ground are considered and deliver meaningful results in the polar angle range 60◦ − 90◦. Since the applied source model is just holding for jet mixing noise, the large scale noise radiation is not predicted, which appears at polar angles below 60◦.

Towards Forced Eddy Simulation of Airframe Induced Noise Radiation from Coherent Hydrodynamic Structures of Jet Flow

The application of Forced Eddy Simulation (FES) is studied in the context of Direct Noise Computation of jet installation noise from nozzle-wing configurations with small relative distance between jet axis and wing trailing edge of the order of the nozzle diameter. Direct noise computations are performed solving the Navier-Stokes equations in perturbation form over a given background RANS flow to realize a zonal RANS/LES simulation approach in the Non-Linear Disturbance Equation (NLDE) framework. The direct noise computation is tackled with a modified version of the full compressible Navier-Stokes equations suitable for moderately compressible flow problems as often present in Computational Aeroacoustics. The formulation enables a formulation of the viscous stress and subfilter contributions in terms of a vector force model that supports an efficient computational treatment. The presented FES method carries on with the development of a stochastic backscatter model for technical applications. The subfilter forcing model provides a combination of dissipation and forcing that yields an additional driving mechanism of the turbulent energy cascade–otherwise not present in purely dissipative models – which enables a rapid onset of fluctuations in the simulation without gray areas. First, simulations of two isolated single stream jets in static condition with nozzle exit Mach number 0.6 and 0.9 are performed to verify the simulation method for the simpler test case of an isolated jet. A second configuration concerns an installed Ultra High Bypass Ratio (UHBR) nozzle below a wing with deployed flap in forward flight at approach condition with a flap deflection of 25°. The simulation considers the entire wind tunnel setup of a related experiment conducted in the Acoustic Windtunnel Braunschweig (AWB).

CAA Prediction of Jet-Wing Interaction Noise Using an Eddy Relaxation Source Model

This contribution presents a numerical study of jet noise installation effect for a generic configuration. The geometry definition consist of a single stream nozzle which is mounted closely under a NACA0012 wing. Such constellations produce significant interaction effects especially in terms of aeroacoustics. The applied methodology is a novel CAA approach which can be classified as a partly scale resolving simulation. In conclusion, the computed results are validated with measurements for a comparable constellation of a flat plate over a single stream nozzle with the same flow conditions. The validation of the CAA computation delivered a good agreement which indicates that all principal effects are numerically well reproduced.

Simulation of Cold Jet Installation Noise using a Stochastic Backscatter Model

This work presents results of Computational Aeroacoustics simulation for two different installation noise problems involving a cold jet interacting with a wing. Similar to very large-eddy simulation (VLES), the resolvable very large scales of turbulent fluctuations are directly calculated and the dissipation of the non-resolved scales is accounted for by a subfilter scale stress model. In addition, stochastic forcing in space and time is applied to model turbulent backscatter. The paper presents and discusses the rationale to explicitly realize turbulent backscatter along with details of the proposed stochastic backscatter model and its calibration. As a novel approach, the entire subfilter forcing function is modeled by means of an eddy-relaxation source term that provides forcing and dissipation as an entangled compound. The relaxation parameter defines the amount of correlation of the subfilter forcing with resolved quantities. Its proper calibration is achieved using decaying homogeneous isotropic turbulence. Further characteristics of the backscatter forcing are analyzed from synthetic turbulence data. The first jet-wing interaction problem studied is based on a generic static jet interacting with a non-inclined rectangular wing. The second problem deals with a dual-stream nozzle installed at a high-lift wing with deployed flap and slat in wind tunnel flow under approach conditions. For both problems installation noise from the airframe yields higher peak levels than the jet-noise contribution alone. For the first problem, relative to the corresponding jet spectrum a low-frequency narrow-band contribution is observed that can be attributed to coherent jet structures interacting with the airfoil trailing edge. Very good agreement with measured spectra is obtained. For the second problem a broadband airframe installation contribution to the overall spectrum is predicted with peak frequency above the jet contribution.