Marlene Sanjose

Subsonic Jet Noise Simulations Using Both Structured and Unstructured Grids

For the last 10 years, large-eddy simulations have become a major tool for investigating jet noise sources because of their intrinsic ability to capture broadband turbulent features. However, many challenges still arise when dealing with complex geometries in terms of method accuracy and computational costs. Two different approaches to compute jet noise in an industrial context are here validated and compared. Both approaches are based on a hybrid methodology combining large-eddy simulation of jet flows for sources computations and Ffowcs Williams and Hawkings’s analogy for far-field noise prediction, but they differ on their grid topologies. The first approach uses classical block structured grids. The numerical scheme is a low-dispersive, low-dissipative finite-volume compact scheme. The second approach uses fully unstructured tetrahedral grids with a low-dispersive, low-dissipative Taylor–Galerkin finite-element scheme. Both approaches are used to compute a Mach 0.9 cold jet at the moderate Reynolds nu...

Noise Source Analysis of a Rod-Airfoil Configuration Using Unstructured Large-Eddy Simulation

The noise generated by a rod–airfoil configuration is investigated using unstructured Large-Eddy simulation coupled with a Ffowcs-Williams and Hawkings technique. The detailed experimental database and several numerical simulations available enable an extensive validation of the proposed methodology. Similar or improved results are obtained both in the near field (surface coefficient distributions and velocity profiles) and in the acoustic far field (power spectral densities obtained with both porous and solid surfaces, and directivities) compared with the best and most recent simulations. Dipolar noise radiation at the rod vortex-shedding frequency and its harmonics are found to be dominant. Some additional quadrupolar source contributions are seen at low and high frequencies. The rod is shown to have a significant contribution to the far-field noise at grazing angles. Constructive interferences occur normal to the rod–airfoil axis, whereas destructive interferences appear at grazing angles.

Jet noise simulation with realistic nozzle geometries using fully unstructured LES solver

Different jet simulations with realistic nozzle geometries for single and dual jets are performed using the fully unstructured Large Eddy Simulation solver AVBP. Two single jet nozzle geometries from NASA are considered: the Single Metal Chevron 000 which is the base line of the series and the Acoustic Reference Nozzle 2 (ARN2). Both are convergent nozzle but with different section variations. Isothermal Mach 0.9 jets at moderate Reynolds number are simulated using these nozzles. The operating condition corresponds to the set point 7 of Tanna matrix. Grids with different resolutions are used and perturbations are injected in boundary layers to get a fully turbulent mixing layer at the nozzle exit. Apart from a laminar to turbulent transition at the nozzle exit, results show a good agreement with experimental data in terms of aerodynamics and acoustics. A co-axial jet nozzle geometry experimentally studied at the P’ Institute of Poitiers is also simulated. It is a isothermal configuration with a primary jet Mach number of 0.5 and a secondary jet Mach number of 0.35. This low Mach configuration makes it more challenging as the noise level reduces and the acoustic propagation is slower. For the mean flow quantities, an overall good agreement with experimental measurements is obtained. However the flow dynamics is dominated by a strong vortex pairing phenomenon due to the late laminar to turbulent transition of the mixing layers and the discretization of their interactions on the present grid might be unsufficient.

Direct self-noise simulation of the installed Controlled Diffusion airfoil

Broadband noise produced by the trailing-edge of a controlled di usion (CD) airfoil is directly simulated using a Lattice-Boltzmann method (PowerFlow) that resolves both the aerodynamic and acoustic eld around the airfoil. The proper DNS resolution is rst achieved in the vicinity of the airfoil on a quasi-2D slice of the mock-up. It is then extended to a 3D slice with a span of 12 % chord length. Two numerical setups of the anechoic openjet facility where both aerodynamic and acoustic data have been collected are investigated to capture the installation e ects: in a rst numerical setup (called free), the CD airfoil is set in an uniform ow, while in the second setup (lips) the real jet nozzle geometry is considered. While in the freeeld con guration the boundary layer rapidly detaches on the suction side, in the lips the jet shear layers modify the pressure load on the airfoil and the boundary layer keeps attached in the con guration with nozzle. In both setups a laminar recirculation bubble is captured on the suction side near the leading edge which triggered the development of the boundary layers along the suction side. The wall-pressure and noise spectra for the free con guration are spread over a large frequency band and agree with similar measurements at higher angle of attack for which the ow is detached. The spectra for the lips con guration better agree with the experimental data and has been selected for the 3D simulation. The vortex stretching along the span-wise direction that was missing in the previous investigated set-ups, allows to nely capture the turbulence length scales and accurately reproduce the experimental measurements. Both the boundary layer pro le and the wall pressure spectra near the trailing edge are nicely and accurately captured. The predicted fareld sound pressure levels also provide satisfactory agreement with noise measurements in the anechoic wind tunnel.

Unstructured LES of the baseline EXEJET dual-stream jet

The EXEJET project is a large-scale jet noise collaborative program by Snecma, Onera and Airbus. In this framework, an exhaustive experimental database has been collected on a realistic dual-jet stream mounted around a central plug. The high-bypass ratio jet is operated at approach conditions in the CEPRA19 anechoic wind-tunnel. In the present paper, several Large-Eddy Simulations (LES) have been performed on the EXEJET baseline configuration using the unstructured flow solver AVBP developped at Cerfacs. The nozzle inlet boundary conditions are obtained by performing RANS simulations until the nozzle exit total pressure conditions match the experimental pressure probe survey. Three LES simulations have been performed following the numerical strategy developped by the present authors, the first one with a fully tetrahedral mesh, the second one uses a hybrid mesh including prismatic layers to better resolve the boundary layer development in the nozzles and finally, a finer mesh including an original tripping method near the fan jet exit to trigger the laminar to turbulent transition of the external mixing layer. The boundary layers developping in the nozzles have a major influence on the turbulent jet streams mixing. The mean velocity magnitude and turbulent kinetic energy evolution along the jet from the PIV measurements are well captured with the three simulations but a better agreement is obtained with the finer hybrid tripped simulation. The acoustic predictions obtained using a Ffowcs William and Hawking’s analogy are in good agreement with the experimental measurements especially at 30◦ degree in the downstream direction. The successive meshes tested bring similar acoustic results in the low and middle frequency ranges. The hybrid meshes recover the proper high-frequency decay of the sound levels and the cut-off frequency is increased with the mesh refeinement. Preliminary beamforming results have identified two main noise sources upstream and downstream the core-jet collapse.

Direct numerical simulation of acoustic reduction using serrated trailing-edge on an isolated airfoil

Broadband noise produced by the trailing edge of a controlled diffusion (CD) airfoil with a serrated trailing edge in the actual anechoic wind-tunnel test configuration is directly simulated using a Lattice-Boltzmann method at a high Reynolds number of 1.5× 10 typical of actual ventilation systems and low pressure turbomachinery stages. Such an unique simulation resolves both the aerodynamic and acoustic field around the airfoil. The comparison of the airfoil loading, the wall-boundary layer development, and the near-wake flow between the clean and serrated configurations is systematically performed and results are compared with the detailed experimental database available for the clean airfoil. A slight lift reduction is found with the serrations near the trailing-edge. The turbulent wallpressure fluctuations on the airfoil suction-side are only slightly modified. Because of the serrations the pressure and suction side boundary layer flows merge differently than in the clean case, modifying the near-wake evolution with larger coherent hairpin structures. The slight vortex shedding observed on the pressure side of the clean airfoil is also suppressed. Finally a noise reduction is captured between 1000 and 3500 Hz in all direction except in the wake direction where a slight noise increase is noticed.

Prediction of the sound generated by a rod-airfoil configuration using a compressible unstructured LES solver and a FW-H analogy

This paper aims at predicting the noise generated by flows interacting with airframe elements using unstructured LES coupled with a FW-H technique. The rod-airfoil canonical geometry 5 has been selected as a benchmark representative of such phenomena. The detailed experimental database 5 and several numerical simulations 3,6–11 available enable an extensive validation of the proposed methodology. Similar or improved results are obtained both in the near-field (velocity profiles) and in the acoustic far-field (power spectral densities obtained with both porous and solid surfaces) compared with the best, most recent simulations. The impact of two important numerical parameters is also assessed : the spanwise dimension of the computational domain, and the sensitivity of the rod/airfoil alignment. The former improves the low frequency content of the simulated acoustic pressure, the latter only the simulated near-field in the cylinder wake. Finally best practices are drawn from this test case for future airframe industrial configurations.

Aeroacoustic Prediction of the Tonal Noise Radiated by a Ring Fan in Uniform Inlet Flow

Aeroacoustic simulation of a ring rotor in presence of a uniform inlet ow is presented. An unsteady RANS simulation with a compressible solver is used to compute the ow eld and identify the acoustic sources on the rotor. The tip clearance recirculation shows upstream vortices that appear to impact the rotor blades creating the main source of unsteadiness. Since those vortices rotate at a lower speed than the rotor, the impact frequency implied is shown to be di erent than the blade passing frequency. Besides the aerodynamic simulation, the acoustic signature of this rotor is computed by propagating the noise sources located on the rotor surfaces using two methods: a Ffowcs-Williams & Hawkings analogy in the time-domain, as well as an analytical model in the frequencydomain based on the compact rotating dipole formulation. A comparison with experimental results con rms the aeroacoustic phenomena are well captured and properly propagated by the acoustic codes.

Dual-stream jet noise simulations with realistic nozzle geometries using a fully unstructured LES solver

Different jet simulations with realistic nozzle geometries for dual-stream jets are performed using the fully unstructured Large Eddy Simulation solver AVBP developed at Cerfacs. First, an isothermal low-Mach coaxial jet which has been experimentally investigated at the P Institute of Poitiers is simulated. The second investigated configuration is a tenth reduction of a generic model of a high-bypass ratio engine operated at take-off conditions experimentally investigated in the EXEJET project framework at the CEPRA19 Onera anechoic wind-tunnel. The third one is a 12-lobes internal mixer investigated at NASA Glenn Research Center. A similar meshing and numerical strategy has been followed for the three simulations in order to validate the methodology on different challenging jet noise configurations of increasing complexity. The aerodynamic results are in good agreement with the experimental measurements available for the three configurations. The sensitivity to the nozzle exit conditions and especially to the boundary layers resolution at the exit of the nozzle are demonstrated by the different meshes used for each configuration. These results validate the present numerical approach to capture the proper jet developments in all three configurations. For the three configurations, the acoustic results computed using the Ffowcs Williams and Hawking’s analogy are also in good agreement with the experimental measurements, particularly at 30 degrees in the downstream direction. The low Mach number of the coaxial jet configuration makes it more challenging as the noise level decreases. For all the simulations, proper control of the laminar to turbulent transition, to avoid large vortex pairing in the external mixing layers, seems to be the key for accurate acoustic predictions.

A Comparison between Galerkin and Compact Schemes for Jet Noise Simulations

Several test cases of single isothermal and hot jets at various Reynolds numbers have been run with a Large Eddy Simulation unstructured ow solver AVBP developed at Cerfacs that uses a third order Galerkin scheme on fully unstructured tetrahedral meshes. Comparisons have been made with a high-order structured code, HECTOR developed at McGill, dedicated to jet noise applications that uses sixth-order compact schemes on multiblock structured grids. The comparisons between the two codes and the experimental data on similar grids and using the same forcing at the inlet show that both codes predict the turbulent ow quite well and most of the features of the acoustic far eld. The fareld noise is obtained for both codes with the same Ffowcs-Williams and Hawkings code MCAAP developed at McGill. The dominating downstream radiation at about 30 and the additional mixing noise radiating mostly at 90 are well captured. Yet, on the one hand HECTOR shows some excessive noise at a Strouhal number of about 0.4 caused by the classical vortex pairing of the destabilizing laminar shear layer yielding an additional noise source radiating downstream at about 60 . On the other hand, AVBP does not show this extra noise source, but yields lower overall levels that are most likely caused by a larger dissipation of either the lower order discretization scheme or a too rapid grid stretching.