Fred Mendonça

CFD Prediction of Narrowband and Broadband Cavity Acoustics at M=0.85

Two rectangular cavity configurations at Mach 0.85 are investigated with the objective of assessing the extent to which 3D CFD with advanced turbulence modeling is capable of predicting narrowband and broadband flow noise. A non-linear, two-equation, eddy-viscosity model run in unsteady mode (URANS) is compared with Detached Eddy Simulation (DES) on a cavity with a L/D ratio of 5, representing cavity flow in so-called shear layer mode. Detailed experimental data for this cavity, configured with and without doors, provides a valuable opportunity to compare predictions of the spectra at many points along the cavity ceiling and band limited amplitude along the cavity length.

DES Validations of Cavity Acoustics over the Subsonic to Supersonic Range

The present work follows on from a successful validation study into the use of 3D CFD with advanced turbulence modeling to predict narrowband and broadband flow noise in a rectangular cavity (L/D=5,W/D=1) at M=0.85. In the present study, this body of work is extended in three ways: first, from the transonic case to a range of flow speeds, from subsonic through to supersonic, M=0.6 to 1.35. Secondly, comparison is made between three DES variants; two of which are frequently referenced in the literature, namely SpalartAllmaras and k-ω-SST, and our own variant based on the standard k-e model as used in the previous work. Thirdly, we reduce uncertainties due to mesh and discretisation scheme dependence by comparing results from coarse and fine meshes, and blended discretisation schemes. The Mach number sweep predictions are then compared with the M219 cavity measured point spectra along the cavity ceiling, and RMS pressure along its length. The results are in good agreement with the measurements. Nomenclature L, D, W = Length, Depth, Width cavity dimensions (inches) M = Mach Number (dimensionless)

Simulation of Radial Compressor Aeroacoustics Using CFD

This paper focuses on the application of CFD to the prediction of radial compressor aeroacoustics. It concentrates mainly on automotive turbocharger operations in the low mass-flow range where blade leading-edge and tip separation reduce the compressor performance and induce transient flow behaviour. Whereas the blade-passing is tonal and at high frequency, usually beyond the human hearing range, transience in the flow are turbulence-dominated, broad-band in nature, and in magnitude a significant source of aeroacoustics which appears well within the range of peak human hearing (1–5kHz). Other noise sources occur due to distortions in the flow upstream of the compressor face, and rotating stall.The simulation methodology enumerated here pays attention to all the above flow-induced aeroacoustics. Due consideration is given to turbulence modelling, to ensure that both the narrow-band and broad-band sources are directly resolved in the CFD. Appropriate discretisation practices are adopted, so as to capture both turbulent-convection and sound-propagation mechanisms. Pressure-wave non-reflective boundary conditions are applied to the computational boundaries to remove any artificial resonances in the domain.STAR-CCM+, the commercial CFD code used here, was previously benchmarked against experimental data for the same compressor under ideal installation conditions, then the compressor performance assessed under real installation conditions [1]. The main foci of the studies reported here are to exploit possible improvements in modelling of the device performance and efficiency curves using more detailed wall modelling, comparing low-y+ versus high-y+ wall resolution, and to explore the viability for transient CFD calculations to capture the noise sources in the compressor at the challenging low mass flow end of the performance characteristic.© 2012 ASME

RANS and DES turbulence model predictions of noise on the M219 cavity at M=0.85

Two rectangular cavity configurations, with and without bay doors, at Mach 0.85 are investigated with the objective of assessing the ability of 3D CFD with advanced turbulence modelling to predict narrowband and broadband flow noise. A non-linear, two-equation, eddy-viscosity model run in unsteady mode (URANS) is compared with Detached Eddy Simulation (DES) on a cavity with a L/D ratio of 5. In a thorough evaluation, comparisons are made between DES variants published in the literature, namely Spalart-Allmaras, k-ε and k-ω-SST. We also assess the effect on the noise spectra of using different CFD prediction time-samples of approximately 100 flow passes compared with 250 flow passes. Detailed experimental data for both cavity configurations provide a valuable opportunity to compare the predicted spectra at many points along the cavity ceiling and band-limited amplitude along the cavity length. We conclude that for such cavity flows, all DES models perform similarly well and are superior to unsteady RANS due to their inherent ability to resolve broadband structures.

Aeroacoustic Simulation of Double Diaphragm Orifices in an Aircraft Climate Control System

Transient CFD acoustic source prediction and CA propagation studies of a double diaphragm orifice illustrate complex flow and aeroacoustics phenomena, combining shedding structures (tonal noise) and large eddy structures (broadband noise). Such studies are necessary for aircraft climate control systems because such components introduce nonlinearities in the system characterization. In this paper we investigate, through modeling, the noise signature of a given double diaphragm configuration comparing against microphone measurements the predicted noise spectra in the near field (source region) and the predicted propagated sound in the far field, some distance downstream of the source region. A second configuration is then assessed, with the spacing between the orifices doubled, so as to derive confidence that the modeling is accurate in both absolute and differential terms. A novel method is presented for estimating the mesh frequency cut-off from a steady-state CFD calculation; such an estimate gives valuable insight a priori into the local mesh refinement required for a transient CFD calculation to resolve the frequency range of interest. The work described in this paper is part of a wider evaluation into the use of CAA methods for aircraft climate control systems.

Validation tests for flow induced excitation and noise radiation from a car window

Experiments to simulate the flow induced vibration and noise radiation from a car window are described. Two cases are considered: flow over a backwards-facing step and flow over a half-cylinder placed on a flat plate. Thefirst configuration represents the flow separation and subsequent re-attachment into a turbulent boundary layer generated by the A-pillar of a car. The second configuration is representative of the turbulent flow over a side window produced by a car wing mirror. Measurements were carried out in an open jet anechoic wind tunnel at the ISVR, where a very low noise flow of up to 40 m/s can be achieved. Wall pressure fluctuations were measured using a streamwise array of microphones, from which the Power Spectral Density, coherence and cross-correlation of the pressure fluctuations were obtained. It was also possible carry out a spatial Fourier transform to give the wavenumber-frequency spectra of the wall pressure fluctuations. The vibrational response of the test panel was measured on a grid of points and noise radiation was measured using sound intensity mapping and a fixed microphone in the acoustic farfield.

COMPUTATIONAL AEROACOUSTIC ANALYSIS OF A ¼ SCALE G550 NOSE LANDING GEAR AND COMPARISON TO NASA & UFL WIND TUNNEL DATA

In conjunction with the NASA-Gulfstream and Gulfstream-University of Florida wind tunnel testing of a generic nose landing gear model, Computational Aero-Acoustics research was conducted at Gulfstream using the commercial Navier-Stokes Finite Volume CFD solver STAR-CCM+. The simulation modeled the ¼ scale G550 nose landing gear mounted in the NASA Basic Aerodynamics Research Tunnel (BART) wind tunnel test section. The model was run at the 0.166 Mach Number condition on the simplified nose landing gear with the gear cavity closed. Prior to simulation execution, dynamic pressure probes were inserted onto the surface of the model at the exact locations of the dynamic pressure transducers on the nose landing gear wind tunnel model. Delayed Detached Eddy Simulation (DDES) (1) was run with a 58 million cell unstructured grid on 47 3.0 GHz quad-core processors for approximately two weeks to obtain a total of 0.05 seconds of statistically limit-cycled pressure data was processed. Results of the simulation were compared to data from the NASA BART and University of Florida anechoic wind tunnel results at both surface mounted steady and unsteady pressure transducers. The simulation showed good agreement in comparisons of power spectral density up to 5 kHz at mesh/time step dependent frequencies at all locations captured. In addition, the simulation was in somewhat good agreement with the measured mean turbulence levels at the wheel hub height and good agreement at static pressure taps located on the starboard wheel.

Aeroacoustic Simulation of the Noise radiated by an Helmholtz Resonator placed in a Duct

The quintessential aeroacoustic test case of an Helmholtz resonator was studied, deploying the recently released coupling between the Engineering design tools STAR-CD (a CFD code) and Actran/LA (a CA code). The objective, to accurately model both the frequency and magnitude of the aeroacoustic resonance phenomenon in the geometry as compared with both experimental and analytical data, was achieved. This paper presents the methodology used, demonstrates the importance of accurate representation of compressible effects in CFD, and presents methods to optimise the data transfer between the codes by focusing on the dominant source regions.

Comparison between measured and predicted tonal noise from a subsonic fan using a coupled Computational Fluid Dynamics (CFD) and Computational Acoustics (CA) approach

The flow through an HVAC blower fan, typical to the automotive industry, was solved using STAR-CD. Time varying pressure data was then exported to ACTRAN/TM where the CFD results were decomposed into acoustic duct modes, using a multiple plane matching method. These duct modes were then propagated to far field locations and compared with experimental data at the blade passing frequency, obtained in the semi-anechoic chamber at Denso Thermal Systems. Good agreement was attained for microphones located downstream of the inlet and outlet ducts. At the other microphone locations there is doubt that the dominant noise generation mechanism is purely aeroacoustic: vibroacoustic noise sources are present. Since aeroacoustic noise sources alone were modeled, there was a deterioration in the simulation/experimental comparison.

Optimal Sunroof Buffeting Predictions with Compressibility and Surface Impedance Effects

Computational Fluid Dynamic (CFD) studies of sunroof buffeting on production vehicles demonstrate accurate prediction of the main buffeting frequency and its harmonics. For production vehicles, none to date has illustrated the phenomenon of buffeting intensity maximization over a vehicle speed range, at a frequency related to the volume of the passenger compartment. All assume that the interior surfaces of the vehicle are rigid, potentially overestimating in-cabin noise intensities by failing to account for surface impedance from non-acoustically rigid trims and linings. In this paper, a modelling study of both effects is presented. Advanced LES-type turbulence modelling, in the form of Detached Eddy Simulation (DES), is used. The hybrid approach, linking acoustic source generation in CFD to acoustic pressure wave propagation with Boundary Element Methods (BEM), is adopted. The former is used to predict the surface pressure excitations induced by the flow, the latter to interpret these as equivalent dipoles to be propagated. The results confirm both the necessity of accounting for compressibility effects in the CFD solver when predicting the buffeting intensity maximization, and surface impedance affects on noise levels perceived by the driver and passengers at higher frequencies closer to the peak audibility range.