Modeling Acoustics in FLUENT

In recent years, as engineering design of components andsystems has become increasingly sophisticated, a signif-icant amount of effort has been directed toward thereduction of aerodynamically generated noise. With theongoing advances in computational resources and algo-rithms, CFD is being used more and more to studyacoustic phenomena. Through detailed simulations offluid flow, CFD has become a viable means of gaininginsight into noise sources and basic sound productionmechanisms.

FLUENT offers fourapproaches for simulatingaeroacoustics. In order ofdecreasing computationaleffort, these are computa-tional aeroacoustics(CAA, or the directmethod), the coupling ofCFD and a wave-equa-tion-solver, integralacoustic models, andbroadband noise sourcemodels.

Computational AeroacousticsComputational aeroacoustics is the most comprehensiveway to simulate aeroacoustics. It does not rely on anymodel, so is analogous to direct numerical simulation(DNS) for turbulent flow. CAA is a transient simulationof the entire fluid region, encompassing the sources,receivers, and entire sound transmission path in between.By rigorously calculating time-varying flow structures,pressure disturbances in the source regions can be fol-lowed. Sound transmission is simulated by resolving thepressure waves traveling through the fluid. While CAA isthe most general and accurate theoretical approach forsimulating aeroacoustics, it is unrealistic for most engi-neering problems because of a number of practical limi-tations, including widely varying length and time scalescharacteristic of the sound generation and transmissionphenomena, and widely varying flow and acoustic pressures.

While these constraints render CAA unsuitable for mostpractical situations, there is a small class of engineeringproblems to which it can be successfully applied. This

includes cases where the frequency range of interest is fair-ly narrow, the sources and receivers are located close toeach other, and the sound to be captured is fairly loud. Thesound generated by an open car window (see page 5) is oneexample, and the sound produced by a side view mirror isanother [1, 2]. For both of these cases, the CAA results arein good agreement with experimental measurements.

CAA has also been used successfully to predict whistles(loud tones) produced byautomotive air intakesystems. The whistlingsound is caused by an airjet passing underneaththe throttle plate (Figure1). As it passes over asump cavity, a shearlayer is established. Ifresonance occursbetween the flappingshear layer and soundwaves bouncing off thesump bottom, a loudwhistle develops. Thesound spectrum predict-

ed by a CAA simulation (Figure 2) is in excellent agree-ment with the corresponding experimental measurement[3, 4]. The CAA simulation predicts almost the exactsame whistle frequency and sound pressure level (SPL)as measured in the experiments.

! 2

Figure 1: Instantaneous velocity magnitude contours in a CAA simulationof whistles generated in an automotive air intake system; the shearlayer flapping in the mouth of the sump cavity causes a loud whistle

Figure 2: Computationally predicted (using CAA) and experimentallymeasured sound spectrum showing a loud whistle generated by anautomotive air intake system [3, 4]

CFD-Sound Propagation Solver CouplingThe computational aeroacoustics approach is prohibitive-ly expensive for most practical problems due to the largedifference in time, length, and pressure scales involved insound generation and transmission. Computationalexpense can be greatly reduced by splitting the probleminto two parts: (1) sound generation and (2) sound trans-mission. With this approach, sound generation is mod-eled by a comprehensive transient CFD analysis, while awave equation solver, such as SYSNOISE from LMSInternational, or ACTRAN from FFT, is used for analyz-ing sound transmission. These software products solvethe wave equation using the boundary element method(BEM). In one recent example, FLUENT was used tosimulate the transient flow field around a generic side-view mirror. Time-varying static pressure was recordedon the mirror surfaces and base plate and exported toSYSNOISE. The output includes a spatial distribution ofthe sound level as a function of sound frequency (Figure 3).

Integral Acoustics MethodsThe approach of splitting the flow and sound fields fromeach other and solving for them separately can be simpli-fied further if the receiver has a straight, unobstructedview of each individual point that is a source of noise.Sound transmission from a point source to a receiver canbe computed by a simple analytical formulation. TheLighthill acoustic analogy [5] provides the mathematicalfoundation for such an integral approach. The Ffowcs-Williams and Hawkings (FW-H) method [6] extends theanalogy to cases where solid, permeable, or rotating sur-faces are sound sources, and is the most complete formu-lation of the acoustic analogy to date. The FW-H methodis implemented in FLUENT.

The FW-H approach has been used to study the soundgenerated by flow over a cylinder of diameter D. Usingthe LES model, the 2D unsteady flow solution (Figure 4)is characterized by a predominant frequency Strouhalnumber of 0.19, compared to a measured value of 0.187.At an observer distance of 35D, the predicted sound pres-sure level is 114 dB, compared to an experimental valueof 117 dB [7]. At a distance of 128D, the predicted andexperimental SPL values are 102 and 100 dB, respectively.

The FW-H approach has also been applied to the genericside view mirror mentioned earlier. The LES model wasused for the 3D flow calculation in the region surround-ing the mirror. An iso-surface of vorticity, colored byvelocity (Figure 5) illustrates the complex, transientnature of the flow. Using this solution, the sound pressurelevels were computed at several microphone locations.Figure 6 shows the spectrum at one receiver for the FW-H calculation and the corresponding CAA calculation.Both are in good agreement with data [1].

3 "

Figure 3: SYSNOISE pre-diction of sound pressurelevel for an automotiveside-view mirror, basedon flow-induced sourcespredicted in FLUENT

Figure 4: Line contours of vorticity magnitude at one instant during theunsteady flow past a 2D cylinder, modeled using LES

Figure 5: Contours of velocity are plotted on an iso-surface of vorticitymagnitude, while line contours of pressure are shown on the plate at left

! 4

Broadband Noise Source ModelsThe three methods described so far require well-resolvedtransient CFD simulations, since they aim to determinethe actual time-varying sound-pressure signal at thereceiver, and from that, the sound spectrum. In severalpractical engineering situations, only the locations andrelative strengths of sound sources, rather than the soundspectra at the receivers, need to be determined. If thesound is broadband (without any prominent tones charac-terized by sharp peaks in the spectrum), the sourcestrengths can be evaluated with reasonable accuracyfrom the time-averaged structure of the turbulent flow inthe source regions.

Turbulence is the primary cause of sound in aeroa-coustics, so in a broad sense, regions of the flow fieldwhere turbulence is strong produce louder sources ofsound. FLUENT 6.2 includes a number of analyticalmodels referred to as broadband noise source modelswhich synthesize sound at points in the flow field fromlocal flow and turbulence quantities to estimate localsound source strengths. The key advantage of these modelsis that they require very modest computational resourcescompared to the methods described in the previous sections.

Broadband noise models only need a steady state flowsolution, whereas the other methods require well-resolved transient flow solutions. One example recentlystudied involves the prediction of prominent soundsources around a simplified sedan (Figure 7), using Lilley’sacoustic source strength broadband noise model [8].

In summary, FLUENT offers four ways for simulatingaeroacoustics. These range from highly accurate, butexpensive methods to quick and approximate approach-es. All of these methods are included in the standardFLUENT software; no add-on modules are necessary. !

References:1. R. Siegert, V. Schwarz and J. Reichenberger, AIAA

Paper No. 99-1895.2. B.S. Lokhande, S.D. Sovani and J. Xu, SAE Paper

No. 2003-01-1698.3. V. Kannan, J. Seifert, T. Golletti and D. Hanner,

SAE Paper No. 2004-01-0395.4. V. Kannan, S.D. Sovani, D. Greeley and A.D. Khondge,

Submitted to SAE NVH Conference, May 2005.5. M.J. Lighthill, Proc. Royal Society A 211, p. 564 (1952).6. J.E. Ffowcs-Williams and D.L. Hawkings, Proc.

Royal Society of London A 264, pp. 321-342 (1969).7. Revel, Lockheed Report 28074.8. G.M. Lilley, The Radiated Noise from Isotropic

Turbulence Revisited, NASA Langley Research Center ICASE Report 93-75; NASA CR-191547.

Figure 6: The Direct (CAA) and FW-H approaches are both in goodagreement with experiment for a receiving point not far from the mirrorData courtesy of DaimlerChrysler

Figure 7: An iso-surface ofLilley’s acoustic sourcestrength shows prominentwind noise sources on ageneric sedan

Wind buffeting, the noise and pulsating forces that areexperienced when driving a car with the side windowsopen, has become a significant factor in the overall pas-senger experience in recent years. It is caused by anunstable shear layer that is established at the upstreamedge of the window opening. Disturbances are shed fromthis location and travel along the side of the vehicle.When they reach the rear edge of the window opening, apressure wave is generated that propagates both insideand outside the passenger compartment. Outside thevehicle, this wave propagates both forward and backwardalong the side of the car. When the forward travelingwave reaches the front edge of the opening, it triggersanother disturbance that moves back to the rear edge.This process is repeated many times every second andcauses the shear layer to develop a characteristic buffet-ing frequency, which depends on the speed of the auto-

mobile and the geometry of the opening. Often the frequencyis below the range that can be heard by human ears but itstill can be felt by passengers as a pulsating wind force.Wind buffeting can be detected using microphones, butthe complicated pressure waves that are its cause are verydifficult to measure. As a result, engineers in the pasthave had to wait until relatively late in the design processwhen prototypes become available to measure this phe-nomenon. These measurements typically give them littleor no information about what areas of the design areaffecting wind buffeting and what could be done toreduce it. The only option is to modify and test the proto-types to see whether individual changes have any effect.This process is so costly and time-consuming that it is dif-ficult to identify changes that will improve the design.

At DaimlerChrysler, engineers have been using the com-putational aeroacoustics (CAA) approach in FLUENT tosimulate wind buffeting. In two recent studies [1, 2], themodeling process began by importing a surface model ofthe outer shape of the vehicle and a CAD model of thevehicle interior (Figure 1) into a CFD preprocessor. Thesimulation domain was defined to include the entire pas-senger cabin, which is connected to the external flowdomain through an open window. Dummies representingthe passengers were included in the model to correctlyrepresent the volume of the passenger compartment. Thevehicle surface was modeled to a significant degree ofdetail to capture flow development from the vehicle frontend to the window opening. Several levels of local meshrefinement were used. Two refinement levels wereapplied outside the vehicle to capture the wake behindthe vehicle. One refinement level was applied inside thepassenger compartment to capture wave propagationinside the cabin. The finest refinement level was appliedat the area of the opening to capture the shear layer. Aclose-up view of the surface mesh near the open windowis shown in Figure 2.

The turbulent flow was captured using the RNG k-ε andLES turbulence models, both of which have been shownin the past to provide good results for a range of turbu-lent conditions. Interior surfaces of the vehicle wereassumed to be solid walls instead of soft surfaces such ascarpeting or fabric. Actual car surfaces are less reflectiveand more absorbent than solid walls, giving the model a

5 "

The Sound of Side Window Buffeting

Figure 1: The model geometry of the car exterior and interior withthe front left window open

Figure 2: The surface mesh detail for the car exterior and interiorin the vicinity of the open front window

tendency to overpredict the windbuffeting phenomena.

A steady-state solution was firstobtained for each case, and theseresults were used as initial condi-tions for a subsequent series oftransient simulations that wereused to capture the pressure fluc-tuations in the vicinity of theopen window. Monitors were setup at the driver’s and front seatpassenger's ear locations andstatic pressure was recorded atthese locations at every timestep. The initial transients dieddown and the pressure tracesreached a dynamically steadysolution in roughly 300 to 500time steps. Subsequently, pressuretraces were recorded for timeperiods between 1.0 and 2.0 sec-onds - long enough to obtain asound pressure spectrum. Thepressure signals were convertedto the sound frequency spectrumby taking a discrete Fouriertransform using a Hanning window filter. The soundpressure level (SPL) was finally converted to dB units.

The CFD predictions were validated by comparing themto experimental measurements conducted in a wind tun-nel. The CFD simulations accurately predicted buffetingfrequency and sound pressure level, and matched SPLand frequency variation trends observed in the experi-ments. Contours of pressure were examined on two hori-zontal planes in the critical front window area (Figure 3).The results indicated that a vertical vortex occurs behindthe A-pillar (the structural member between the wind-screen and the front window). An animation of the tran-sient solution showed vortex movement with the localflow, with impingement on the B-pillar (the structuralmember between the front and rear windows). The wavegenerated at the B-pillar was shown to propagate into thepassenger compartment. Visualization of the formation

and propagation of these wavescan help engineers understandexactly how wind buffeting occursin a particular design, and can helpthem iterate quickly to animproved design.

The frequency spectra and soundpressure level were in good agree-ment for all locations studiedwithin the vehicle (Figure 4).Additional simulations correlatedwell with experimental measure-ments in predicting reductions inthe SPL and frequency of thesound from an open front windowas the vehicle speed is reducedfrom 60 mph to 50 mph.Simulations with the left front win-dow wide open and the right rearwindow open 1 inch were also per-formed. These showed that buffet-ing was substantially reduced.Simulations with a modified sidemirror design reduced buffeting by13 dB, which also correlated wellwith experimental measurements.

DaimlerChrysler engineers are making use of these resultsby simulating other vehicles, evaluating the influence ofmore parameters, and evaluating different modeling tech-niques. As simulation is more fully integrated into thedesign process, this approach should make it possible tosubstantially reduce wind buffeting in the future. !Courtesy of DaimlerChrysler

Reference:1. D. Hendriana, S.D. Sovani, M.K. Scheimann, “On

Simulating Passenger Car Window Buffeting,” SAE Paper No. 2003-01-1316 (2003).

2. C.-F. An, S.M. Alaie, S.D. Sovani, M.S. Scislowicz, K. Singh, “Side Window Buffeting Characteristics ofa SUV,” SAE Paper 2004-01-0230 (2004).

! 6

Figure 3: Pressure field at a speed of 60 mphand yaw angle of 5 degrees, showing the wakesbehind the A-pillar and mirror

Figure 4: Spectrum of the side window buffetingsound heard by a car driver

Cavity flows have been the subject of research since the1950’s. Although geometrically simple, the fluid dynamicsin such flows is complicated, involving shear layer insta-bility, flow induced resonance, and turbulence. Flow overa cavity causes large pressure oscillations to develop, andthese can lead to structural damage. Suppression tech-niques have been applied with varying degrees of success.

The flow generated by an open cavity with transonic flowacross the top combines several clearly identifiable flowphenomena, such as a mixing layer with its train of largestructures, recirculation flow in the cavity, pressure wavesgenerated by the crossing of the structures, and strongacoustic coupling between all these phenomena. Such flowsoccur in landing gear wells and bomb bays on aircraft,where sonic fatigue and the reduction in pressure fluctua-tions and noise are of prime concern. The pressure vents onthe space shuttle cargo bay have also been observed to causehigh internal noise levels during ascent.

In this example, a shallow rectangular cavity, 20 inchesin length and 4 inches square in cross-section, is studied.A bulk Mach number of 0.85 outside the cavity is speci-fied. The turbulent flow is modeled using the detachededdy simulation (DES) approach, whereby a RANS cal-culation (in this case, using the Spalart-Allmaras model)is performed in the near-wall region, and an LES calcu-lation is performed in the free stream. A hexahedral meshof 1.4 million cells is used. Using the non-iterative timeadvancement (NITA) solver in FLUENT, a period of 0.6seconds is simulated, and noise calculations are performedusing the computational aeroacoustics (CAA) approach.Time averaged results from the last 0.2 seconds of the sim-ulation are used to compute the frequency spectrum of thenoise produced for comparison to experiment.

In Figure 1, an iso-surface of vorticity, colored by velocity,illustrates the complex, transient nature of the cavity flowfield. The vorticity is being generated principally in the freeshear layer, although the figure only shows the cavity gen-

eration. The growth of the boundary layer along the entryplate is visible in the form of spreading oil film lines.Following the separation of the boundary layer from theleading edge of the cavity, Kelvin-Helmholtz instabilitiesdevelop and pulsate during the transient flow, causingregions of localized shocking to appear and disappear with-in the unstable shear layer.

Rossiter first developed an empirical formula for predict-ing cavity-flow resonant frequencies, today referred to asRossiter modes. For this configuration the first threeRossiter modes (peaking at 145Hz, 350Hz and 590Hzrespectively) are of a similar strength across the ceilingof the cavity. Each modal band is calculated by process-ing the power spectral density, using frequencies thatbracket the peak. The RMS pressure along the cavityceiling, PRMS, for the first mode is in very good agreementwith experimental data [1], as shown in Figure 2.

The sound pressure level at one of ten microphone loca-tions is compared to experimental data [1] in Figure 3.The DES approach provides a frequency spectrum that isin very good agreement with experiment, while anunsteady RANS calculation does not. This result is con-sistent at every monitor point considered in the study. !

Reference:1. Experimental data provided by QinetiQ, funded by

UK MOD Applied Research Program. 7 "

Cavity Noise Generation

Figure 1: The complex flowinside the cavity

Figure 2: PRMS along the cavity ceiling for the first mode; FLUENTDES predictions are compared to experiment [1]

Figure 3: Sound pressure level (SPL) 7 inches downstream of thecavity, computed using DES, URANS, and measured by experiment [1]

A recent survey by J.D. Power and Associates [1] indicatesthat excessive wind noise is a major concern for automo-bile passengers. Wind noise is generated by features onthe outer body of a car, such as roof-racks, door-gaps,and side-view mirrors. One prominent feature contribut-ing to the overall wind-noise level is the A-pillar rain gut-ter. This rain gutter serves the purpose of collecting anddraining rain water that would otherwise be swept fromthe windshield, past the A-pillar, and onto the side win-dows, thereby reducing the visibility through them. It hasthe shape of a long narrow channel spanning the lengthof the A-pillar and facing the wind. From an aerodynam-ics perspective, the rain-gutter acts as a turbulator, creat-ing a highly turbulent flow in its wake. Pressure fluctua-tions created by such turbulence contribute to wind noise.

With recent advances in CFD models and algorithms andwith increases in computational power, it is now possibleto study wind noise generation and transmission usingCFD simulations. The present work focuses on the windnoise produced by an idealized rain gutter. The results arecompared with experimental data and other CFD simula-tions reported in the literature.

The idealized rain gutter is a backward facing elbowmounted on a flat plate. The flat plate and gutter areplaced in a virtual wind tunnel with a rectangular cross-section. The free stream air speed is 22.35 m/s, corre-sponding to a Reynolds number of 40,000, based on theheight of the rain gutter. The width and height of the rain-gutter are both 0.0127 m. The large eddy simulation(LES) model is used for the simulation. This transientturbulence model can be used to predict both the sourcesand transmission of sound, but at a high computational

cost. A more economical approach makes use of LES tocompute the time-varying pressure field (the noisesources), and a simple acoustic analogy to compute thesound transmission. This second approach has beenapplied to the rain gutter, using the Ffowcs-Williams andHawkings [2] acoustics model to compute the sound trans-mission. The goals of the simulation are to determine: 1. the transient flow structure around the rain-gutter 2. the distribution of the pressure coefficient, Cp, along

the base plate 3. the frequency spectrum of the static pressure at a point

on the base plate downstream of the rain gutter, and 4. the frequency spectrum at a point above the rain gutter.

The results from the CFD simulations are compared toexperimental and computational data reported byKumarasamy and Karbon [3].

A diagram of the solution domain is shown in Figure 1.The rain gutter is located a distance a from the inlet sur-face AEHD. The distance a is approximately 7.4b, whereb is the height of the rain gutter (0.0127 m). The lengthof the rain gutter, c, is about 0.64b. Surface ABCD is asymmetry plane.

Figure 2 illustrates the turbulent flow structures generatedby the rain gutter. Iso-surfaces of vorticity are colored byvelocity magnitude to illustrate the flow. Contours of pres-sure are shown on the base plate and perpendicular wall.

Instantaneous velocity vectors on the symmetry plane areshown in Figure 3. The flow separates upstream of therain gutter, and two very distinct flow regions develop.The free stream flow outside the separated region issteady, while that inside is complex and unsteady.

! 8

Automotive Rain Gutter Noise

Figure 1: The outline of the domain, showing the rain gutter, inlet,and microphone

Figure 2: Iso-surfaces of vorticity behind the rain gutter, colored byvelocity magnitude, and pressure contours on the base plate andsymmetry plane

9 "

The distribution of the pressure coefficient, Cp, along theflat plate at the symmetry line is shown in Figure 4. Theposition of the rain gutter is shown. The experimental datashown is from Kumarasamy and Karbon [3]. The FLUENTresults are in very good agreement with the data.

The spectrum of time-varying pressure recorded at theintersection of the base plate and symmetry plane, 0.0254 mdownstream of the rain gutter’s vertical surface, is shownin Figure 5. The FLUENT results are in good agreementwith the experimental measurements [3]. The soundspectrum at the far-field microphone, located 0.10795 mabove the rain gutter on the symmetry plane, is shown inFigure 6. The results are again in good agreement withpublished values.

In summary, FLUENT has been used to simulate the flowfield around an idealized automotive A-pillar rain gutterand the sound radiated from it. The LES turbulencemodel was used to compute the transient flow field, andsound radiation was calculated with the Ffowcs-Williamsand Hawkings integral method. Numerical results werecompared with corresponding experimental measure-ments reported in the literature. The time-averaged pres-sure coefficient values predicted by the FLUENT simula-tions were found to match (within the bounds of experi-mental uncertainty) the measured steady-state values.The spectra of pressure at a point on the base plate and ata far-field microphone were within a few dB of the cor-responding experimental measurements over a wide fre-quency range. Overall, the results demonstrate that theLES turbulence model coupled with the Ffowcs-Williams and Hawkings method are well suited toacoustics simulations of this type. !

Reference:1. 2000 Vehicle Acoustic Study, J.D. Power and

Associates, Westlake Village, CA 91361.2. J.E. Ffowcs-Williams and D.L. Hawkings,

Proceedings of the Royal Society of London A264, p. 321-342 (1969).

3. A. Kumarasamy and K. Karbon , Aeroacoustics of anAutomobile A-Pillar Rain Gutter: Computational andExperimental Study, SAE Paper 1999-01-1128 (1999).

Figure 3: Velocity field on the symmetry plane

Figure 4: Pressure coefficient as a function of position along the baseplate, with comparisons to experiment [3]

Figure 5: Sound spectrum at a microphone located behind the raingutter on the base plate

Figure 6: Sound spectrum at the microphone located above the rain gutter

! 10

Landing gear noise is not the first thing that comes tomind when thinking about noise pollution at a busy air-port. The once dominant jet engine noise has beenreduced significantly over the past thirty years, primarilythrough the introduction of high bypass turbofan engines.As a result, airframe noise has emerged as a leading com-ponent of aircraft noise during the final approach phaseof a landing. Environmental concerns and noise certifica-tion regulations are therefore causing aircraft manufac-turers to take a closer look at this phenomenon.

The main contributors to airframe noise in a landing con-figuration are high-lift devices, such as slats and deployedflaps, and surprisingly, the landing gear. Measurementshave shown that these components are not equally importanton all aircraft.While the high-liftdevices are noisieron medium size air-craft, the landinggear is becoming thedominant source onlarge airplanes, suchas the Boeing 777.

Landing gear sys-tems have complex, non-streamlined geometries, and gen-erate highly turbulent wakes. Vortices shed from one com-ponent impinge on other elements, generating noise with abroad spectrum, from a few hundred Hz to several kHz. Ifnoise can be predicted using engineering software, modi-fications such as fairings and streamlining can be intro-duced during the design phase.

At Fluent, engineers have recently analyzed a 1/10thscale landing gear model, representative of the gear usedon a Boeing 757 aircraft. The same configuration hasbeen studied using CFD by researchers at Penn State [1]and NASA Langley [2, 3], and will also be tested in awind tunnel at the Quiet Flow Facility at the LangleyResearch Center. The four-wheel landing gear assemblycontains all of the major components, including the oleo-strut, axles, connecting blocks, diagonal struts, a door,and additional parts that hold the configuration together.A flat plate simulates the wing surface.

Predicting aeroacoustic noise is not a trivial matter. Onlya minute fraction of the kinetic energy present in the pri-mary flow is converted into acoustic energy and radiated.To correctly capture the acoustics, the turbulent flowmust be calculated with high fidelity. Since turbulence isan inherently unsteady phenomenon, a time-consumingtransient simulation is required. The large eddy simulation(LES) turbulence model with the Smagorinsky subgridscale model was used for the landing gear calculation.

Integral techniques that predict the far-field acoustic signalusing source data input from a near-field CFD simulationhave emerged as a promising and economic way to computesound levels. The Ffowcs-Williams and Hawkings (FW-H)approach [4], the most universal and complete integral

method availabletoday, was used forthe simulation.

The CAD (STEP)model providedby NASA Langleywas cleaned up inGAMBIT, and ac o m p u t a t i o n a lgrid was built

using GAMBIT and TGrid. Boundary layer prisms weregrown in TGrid, so that the prism cap surface mesh couldbe used to control the growth and continuity of the tetra-hedral elements away from the boundary layer. Size func-tions were used to cluster elements in the vicinity and wakeof the landing gear, resulting in a 5.3 million cell mesh, suit-able for an LES simulation.

The landing gear case was run incompressibly (a validapproximation for compact sound sources) for nearly10,000 time steps, or one flow-pass through the domain,before the turbulence statistics were sufficiently stabi-lized and the acoustic source data sampling could be start-ed. The acoustic source data was extracted directly on thelanding gear surface over approximately one additionalflow-pass, and then processed with the FW-H solver.

The FW-H tool is ideal for predicting far-field radiation inthe absence of external scattering surfaces. The necessarysource data can be extracted from permeable (interior) or

Low Noise Landing

Figure 1: Vortical structures visualizedusing iso-surfaces of the second invari-ant of the deformation tensor, colored byvelocity magnitude

11 "

solid (wall) surfaces, and the method is not very sensitiveto the actual source surface placement. The direct outputincludes the far-field pressure signals at user-specifiedreceiver locations. Postprocessing tools are available toperform spectral analyses of these signals, including over-all sound pressure level (OASPL) outputs. Also availableis the local dipole source strength, which can be used toassess contributions from different source locations.

Surprisingly, but in good agreement with other studies per-formed on the same configuration [2, 3], flow visualizationrevealed that the two diagonal struts shed nearly as muchvorticity as the big wheels (Figures 1 and 2). A very shortdistance downstream of the landing gear, it is difficult to dif-ferentiate the flow structures originating from different com-ponents. Persistent flow separation due to an asymmetricflow was observed at the gear door leading edge. Animationsof unsteady surface pressure showed more complex patternson the rear wheels and rear strut, as expected.

The acoustic analysis indicated that the overall soundpressure levels at a distance of 10 wheel diametersupstream and downstream of the landing gear are about4dB lower than those measured in the two lateral direc-tions (Figure 3). Differences were also noticed in thesound pressure spectra (Figure 4). The lateral directionspeak at around 700 Hz, and the same frequency wasobserved to be dominant in the crossflow force response.The streamwise spectra peak at considerably higher fre-quencies. A total of 18 surface pressure probes werestrategically placed in the rear of the wheels, struts, andalong the wheel door. The recorded pressure traces con-firmed that the rear diagonal strut is one of the dominantnoise sources. Fluent engineers are anxiously awaitingthe experimental data expected from the wind tunnel toconfirm these findings. !

References:1. F.J. Souliez, L.N. Long, P.J. Morris and A. Sharma,

International Journal of Aeroacoustics 1, No. 2, p. 115-135, 2002.

2. F. Li, M.R. Khorrami and M.R. Malik, AIAA Paper 2002-2411, 8th AIAA/CEAS Aeroacoustics Conference, Breckenridge, CO, June 17-19, 2002.

3. D.P. Lockard, M.R. Khorrami and F. Li, AIAA Paper2004-2887, 10th AIAA/CEAS Aeroacoustics Conference, Manchester, UK, May 10-13, 2004.

4. J.E. Ffowcs-Williams and D.L. Hawkings, Proceedings of the Royal Society of London A264, p. 321-342 (1969).

Figure 3: Overall sound pressure levels (OASPL) for five receivers locat-ed 1 m from the landing gear

Figure 4: Sound pressure level spectra (dB) for four of the receiversshown in Figure 3

Figure 2: Dipole source strength,using contours of dp/dtRMS, showsa high source intensity at the reardiagonal strut and behind theoleo-strut

Fluent Worldwide

©2005 Fluent Inc. All rights reserved. FLUENT is a registered trademark of Fluent Inc. All other products or brands are trademarks of their respective holders.

Corporate HeadquartersFluent Inc.10 Cavendish CourtLebanon, NH 03766, USATel: 603 643 2600Fax: 603 643 3967Email: info@fluent.com

US Regional OfficesAnn Arbor, MI 48104Tel: 734 213 6821

Austin, TX 78746Tel: 512 306 9299

Evanston, IL 60201Tel: 847 491 0200

Morgantown, WV 26505Tel: 304 598 3770

Santa Clara, CA 95051Tel: 408 522 8726

European Regional OfficesFluent BeneluxWavre, BelgiumTel: 32 1045 2861Email: info@fluent.be

Fluent Deutschland GmbHDarmstadt, GermanyTel: 49 6151 36440Email: info@fluent.de

Fluent Europe Ltd.Sheffield, EnglandTel: 44 114 281 8888Email: info@fluent.co.uk

Fluent France SAMontigny le Bretonneux, FranceTel: 33 1 3060 9897Email: info@fluent.fr

Fluent ItaliaMilano, ItalyTel: 39 02 8901 3378Email: info@fluent.it

Fluent Sweden ABGöteborg, Sweden Tel: 46 31 771 8780Email: info@fluent.se

Asian Regional OfficesFluent Asia Pacific Co., Ltd.Tokyo, JapanTel: 81 3 5324 7301Email: info@fluent.co.jp

Osaka, JapanTel: 81 6 6445 5690

Fluent China Holdings Ltd.Shanghai, ChinaTel: 86 21 5385 5180Email: info_china@fluent.com

Fluent India Pvt. Ltd.Pune, IndiaTel: 91 20 414 2500Email: info@fluent.co.in

DistributorsFor a full listing of our worldwide distributor network, log on to:www.fluent.com/worldwide/dist