Hybrid Wing–Body Aircraft Slat Noise

Yueping Guo∗

Boeing Research and Technology, Huntington Beach, California 92647

Leon Brusniak† and Michael Czech†

Boeing Commercial Airplane, Everett, Washington 98124

and

Russell H. Thomas‡

NASA Langley Research Center, Hampton, Virginia 23681

DOI: 10.2514/1.J052540

This paper presents an analysis of the slat noise for hybrid wing–body aircraft. It is shown that the hybrid wing–

body slat noise is characterizedby its broad spectral shapeswith frequencies depending onboth themean flowvelocity

and the aircraft angle of attack, with the former following the conventional Strouhal number scaling and the latter

explainable by the dependence of the coherence length of the unsteady flows on the angle of attack. Although the

overall noise levels approximately follow the fifth power law in Mach number, the Mach number effects manifest

themselves spectrally in both amplitudes and spectral shapes. The noise amplitude is shown to also depend on the

angle of attack, assuming aminimum in the range of 3 to 5 deg. These features are allmodeled and incorporated in slat

noise-prediction methodologies, extending the prediction capability from conventional to hybrid wing–body

configurations. Comparisons between predictions and data show very good agreements in both various parametric

trends and the absolute levels. The hybrid wing–body aircraft is designed to operate at angles of attack higher than

those of conventional aircraft. This is shown to significantly increase the hybrid wing–body slat noise. To further

illustrate, the test data are extrapolated to full scale and compared with the slat noise of the Boeing 777 aircraft,

showing that the former is higher than the latter.

Nomenclature

M = Mach numberx = coordinate in flow directiony = coordinate in wing normal directionz = coordinate in vertical directionα = aircraft angle of attackθ = polar angle

I. Introduction

H YBRID wing–body (HWB) aircraft have been attracting muchattention in recent years because of the potential for reducing

aircraft engine noise bymounting the engines on the upper surface ofthe airframe structure, shielding the engine noise from the ground(e.g., [1–5]). This may significantly reduce the engine noisecomponents, but the effectiveness in reducing the total aircraft noisewill also depend on the levels of the non-propulsion-related com-ponents, including the noise from the HWB airframe structure andthat from the landing gears. These components may form a noisefloor that holds up the total noise levels. In this case, it may not haveadded value to trade aerodynamic performance and/or propulsionefficiency to increase the noise shielding, by shifting the engineinstallation further upstream of the trailing edge, for example, beforethe airframe noise is reduced. Thus, it is important to reliably assessthe levels of HWB airframe noise, which is necessary not only todemonstrate the total acoustic benefit of the HWB design but also toguide future developments ofHWBacoustic technologies, especiallyin the area of airframe-propulsion integration.

To this end, this paper presents an analysis of the HWB slat noise,one of the major airframe noise components for the HWB aircraft.Because of their designed functionality of providing high lift inaircraft takeoff and landing operations, leading-edge slats in bothconventional and HWB aircraft generate noise by similar mech-anisms: namely, the cove region flow separation and the high-speedunsteady flows in the gap between the slat trailing edge and the mainwing (e.g., [6–9]). Thus, the basic characteristics of the noise can beexpected to follow similar patterns for both. However, the designdetails and the operating conditions of the two types of aircraft aredifferent, potentially leading to differences in the parametric trendsand amplitudes of the noise.For the spectral features, it will be shown that the slat noise is

broadband, with the frequency of the spectral peak scaling on themean flow velocity. At fixed mean flow velocity, the peak frequencywill be shown tovarywith the aircraft angle of attack, shifting upwardas the angle of attack increases from zero to the slat deploymentangle. This does not follow the Strouhal number scaling based on theslat chord length, whichwould determine the peak frequency entirelyby the mean flow velocity and the chord length. This scaling hasproven useful for conventional aircraft slat noise, as discussed inprevious studies (e.g., [6–9]), because the variations in the angle ofattack in that case are small, typically about from 4 to 7 deg atapproach conditions. The HWB aircraft, however, may operate atangles of attack usually above 10 deg and can be as high as 14 deg.At such large angles of attack, the Strouhal number scaling based onthe slat chord does not seem to collapse the peak frequency to theusual value of about 2, and when considering the whole range ofangles of attack, from 0 to 15 deg, for example, it becomes apparentthat the peak frequency is not invariant for fixedmean flow velocityand chord length. This will be argued to result from the deficiencyof using the slat chord as the characteristic length scale of theunsteady flows. This characteristic length scale should depend onother flow conditions such as the angle of attack, instead of a fixedvalue based solely on the geometry such as the chord length.When the flow-dependent length parameter is used in the Strouhalnumber scaling, the dependence of the peak frequency on the angleof attack then becomes clear. The use of flow-dependent lengthscales in Strouhal number scaling has previously been used in otherstudies [10,11].

Received 4 January 2013; revision received 30 May 2013; accepted forpublication 31 May 2013; published online 11 October 2013. Copyright ©2013 by the American Institute of Aeronautics and Astronautics, Inc. Allrights reserved. Copies of this paper may be made for personal or internal use,on condition that the copier pay the $10.00 per-copy fee to the CopyrightClearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923; includethe code 1533-385X/13 and $10.00 in correspondence with the CCC.

*Technical Fellow, Acoustics Technology, 5301 Bolsa Avenue. AssociateFellow AIAA.

†Engineer, Acoustics Technology, P.O. Box 3707, MC OR-JF.‡Senior Research Engineer, Aeroacoustics Branch, MS 461. Senior

Member AIAA.

2935

AIAA JOURNALVol. 51, No. 12, December 2013

Dow

nloa

ded

by b

aew

arto

n@ai

aa.o

rg o

n Ja

nuar

y 27

, 201

4 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/1.J

0525

40

The noise amplitudes will be shown to depend on both the flowMach number and the aircraft angle of attack. For the flow Machnumber, the features are very similar to those for conventional aircraftdiscussed in [7]. The overall sound pressure levels approximatelyscale on the fifth power law in Mach number, but the effects of theMach number on the noise spectra are more complicated. Thespectral amplitudes do not seem to scale on any simple power law,even though they are in general an increasing function of the Machnumber. This is due to the dependence of the spectral shapes on theMach number. The variations of the noise amplitudewith the angle ofattack will be shown to assume a minimum in the range of about 3 to5 deg, with increasing levels when the angle of attack moves awayfrom this range. It can be noted that the range of the angle of attackfor the minimum slat noise is close to the usual operation range ofconventional aircraft. This gives the HWB configuration, usuallydesigned to operate at high angles of attack above 10 deg, an acousticdisadvantage because the increase in slat noise due to the increase inthe angle of attack can be significant, as high as 8 dB. This, togetherwith the larger sizes and the larger sweep angles of the HWB slats,compared to conventional aircraft designs,makes the slat noise one ofthe dominant components of HWB airframe noise.Based on the parametric trends revealed by the data, the features

such as the effects of the angles of attack on the noise amplitudeand its peak frequency will be modeled for slat noise prediction.Methodologies for slat noise prediction have so farmostly been basedon conventional aircraft (e.g., [6,7]), and hence, have been focusedon operations with the angle of attack in the range of 4 to 7 deg.The model developed here will be incorporated into the methoddeveloped in [7], extending its prediction capability from con-ventional aircraft designs to HWB configurations. The predictionmodels will be used to predict the HWB slat noise, including theparametric trends in spectral shape, Mach number dependence, andvariations with the angle of attack. It will be shown that thepredictions agree well with the HWB test data in both the parametrictrends and the absolute levels.For conventional aircraft, the high-lift systems include trailing-

edge flaps, whose side edge noise is known to be a major airframenoise component [10–13] with amplitude comparable to or higherthan the slat noise, making the slat noise component a less importantpriority in airframe noise technologies. For HWB aircraft, however,trailing-edge flaps are not used, and hence, flap side edge noise isabsent, increasing the relative importance of slat noise in the rankingorder of HWB noise components. To further demonstrate theimportance of HWB slat noise and to quantify the levels of thiscomponent, the data from the small-scale model test are extrapolatedto full scale, to an aircraft comparable to the Boeing 777 aircraft insize and functionality. It will be shown that the HWB aircraft slatsgenerate substantiallymore low-frequency noise then theBoeing 777aircraft, in the frequency domain around 100 Hz, by as much as10 dB, because the HWB slats have larger chord length. In thefrequency range fromabout 100Hz to 1000Hz, theHWBslat noise ishigher than that of the Boeing 777 aircraft by about 2 dB. The higherlevels of the HWB slat noise may seem to be a surprise because itoperates at a lowerMach number, 0.2 for the HWB aircraft comparedwith 0.25 for the Boeing 777, which would lead to lower HWB slatnoise according to the fifth power law on Mach number. However,this simple reasoning, based solely on the flightMach number, fails toaccount for features such as the slat sweep angle and the aircraft angleof attack, for both of which, the HWB aircraft has larger values.Obviously, larger values of both the slat sweeping angle and the angleof attack may lead to more intense flow separation in the slat flowregion (e.g., [6–9]), and hence, lead tomore slat noise generation. Thetest data and analysis presented here confirm this and show that theacoustic benefit of lower flight Mach number for the HWB aircraft ismore than offset by its large angles of attack and large sweep angle.

II. Boeing Low Speed Aeroacoustic FacilityExperimental Description

The HWB airframe noise experiment was performed in May of2009, one of two back-to-back experiments in the Boeing LowSpeed

Aeroacoustic Facility (LSAF) with the objective of understandingkey aspects of hybrid wing–body (HWB) acoustics and forsupporting system noise assessments of low noise designs of HWBaircraft concepts. The second experiment included the jet noiseshielding aspect of a hybrid wing–body and the development ofnozzle configuration technology to increase the shielding effec-tiveness for the jet noise source [4]. Data from both experiments havebeen used in system noise assessments by NASA [3] and by Boeing[5], showing the potential of the HWB to reach the NASA N � 2noise goal of 42 dB cumulative below Stage 4 given the technologyselections described.The airframe model is a 3% scale representation of the Boeing

blended wing–body (BWB) 450-1L design by Liebeck [14]. The450-1L was designed as a state-of-the-art configuration, and at thetimewhen the HWB airframe noise experiment was planned in 2008,it still represented one of the best configuration and aerodynamicdesigns for a hybrid wing–body aircraft concept. A realistic, highfidelity design of the airframe is essential for realistic airframe noisecharacteristics. Themodel was built by NASA in the year 2000 basedon the Boeing design and was originally built for low-speedaerodynamic and flight dynamics testing. The model was used inseveral NASAwind-tunnel campaigns including the NASA Langley14 by 22 ft SubsonicWind Tunnel. In fact, this 3%model was used todevelop the bulk of the low-speed aerodynamic database obtained byNASA and Boeing for this 450-1L design [15]. In the 2008 planningof this HWB airframe noise experiment, this model was in excellentcondition and available, had considerable aerodynamic data associ-ated with it, and represented a state-of-the-art design with whichNASA and Boeing had developed a large amount of experience overthe prior 10–15 years of joint research.The model is 89.02 in. in span and is a 3% scale of a 247 ft span

full-scale design of the 450-1L. This full-size concept had threeengines mounted with the nozzle exhausts downstream of the trailingedge. Although this 450-1L design represents a state-of-the-artHWB, it was not designed with a low aircraft system noise goal inmind and, therefore, has this one key feature of the position of theengines as a difference from subsequent low noise designs.With verylittle other HWB airframe noise information available at the time ofthis research, the use of this model was an excellent choice eventhough the position of the flow through engine nacelles was incon-sistent with the subsequent low noise designs. The flow-throughnacelles are insignificant as an airframe noise source. Figure 1 showsthe identification of the aerodynamic control surfaces that areincluded on the model from the full-scale design. Inboard andoutboard control elevons cover the length of the trailing edge. Theouter most two elevons can also split to form a drag rudder. Thewinglets also include a rudder. Leading-edge slats are incorporatedfor high lift. Landing gears were not available for this model.The experimental facility is schematically shown in Fig. 2. The

LSAF was configured with a 9 by 12 ft open-jet test section to havethe largest test section possible given that the 3% model is a largemodel for this test section. Free-field Brüel & Kjær Type 4939

Leading Edge Slat

Inner Elevon

Outer Elevon Clam Shell Elevon

Winglet

Fig. 1 Aerodynamic control surfaces and leading-edge slats of the 3%HWB model.

2936 GUO ETAL.

Dow

nloa

ded

by b

aew

arto

n@ai

aa.o

rg o

n Ja

nuar

y 27

, 201

4 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/1.J

0525

40

microphones were set up to measure a polar angle range from 50 to150 deg. Three different arrays were deployed simultaneously to alsomeasure the azimuthal variation of the sound field at angles of 90, 60,and 30 deg. The array at 90 degwas laid out in a polar arc at radius (R)of 25 ft. The 30 and 60 deg azimuthal arrays were at a constantsideline distance of 17.6 and 7.9 ft respectively. The atmosphericattenuation coefficients were obtained from the method of Shieldsand Bass [16]. The spectral data presented in this paper are modelscale data unless otherwise noted.The BWB model was supported from above the test section by

an overhead structure with an aerodynamic, smooth blade strutattachment to the airframe using a standard location for this model.In this way, the model could be pitched to angle of attack as wellas be traversed in two dimensions in a plane normal to the airframemodel, as shown in Fig. 3. The precise positioning of the airframewas monitored with photogrammetry and was determined to bewithin 0.05 in.Figure 3 also shows the phased array panel that can traverse

downstream of the tunnel nozzle exit plane. When not deployed, thephased array panel is stowed upstreamof the nozzle exit plane so as tonot interfere with the far-field microphones, which are also shown inFig. 3. Figure 4 is a photograph taken from the opposite direction ofthe installed model and shows the underside or pressure side of theHWB airframe. Tape is seen on the model and was used to smoothover cover attachments and to eliminate some gaps that would not bepresent in a full-scale application.The test was done to cover the basic operating conditions of the

HWB aircraft, including variations in the aircraft angle of attack, themean flow Mach number, and the high-lift system configurations.Both conventional free-field microphones and phased microphonearrays were used for the noise measurements, with various polar and(for free-field microphones) azimuthal angles, most of which weredesigned to mimic aircraft noise certification test measurements. The

test conditions and measurement locations are summarized inTable 1.Though most of the analysis presented in this paper will be based

on the phased microphone array measurements because of theadvantage of being able extract component noise, the far-fieldmicrophone measurements will be used to calibrate the array results.It is also of interest to use the far-field microphone measurements todemonstrate the background noise levels of the test facility. To thisend, comparisons are given in Fig. 5 between the cases of slatdeployed, slat retracted, and wind-tunnel background without theHWB model. For all cases, the tunnel Mach number is 0.16, and themeasurements are at the overhead location with 90 deg emissionangle. For the caseswith theHWBmodel in the tunnel, themodel is at10 deg of angle of attack, and for the slat deployed case, the slat is at30 deg. Clearly, the noise generated by the HWBmodel is well abovethewind-tunnel background levels, especially in the frequency rangeof practical interests, which is above about 1000 Hz, considering thescale of themodel. For this frequency range, the noise from the HWBmodel is at least 10 dB above the background level. It should also bepointed out that the noise from the model is not entirely due to theslats, even though it is labeled as slat deployed. As will be discussedin the next section, the model had to be installed with part of its winginteracting with thewind-tunnel shear layer, inducing extra noise andcontaminating the far-field noise measurements.

III. Boeing Low Speed Aeroacoustic Facility PhasedArray Data-Reduction Process

Phased microphone arrays were used to locate and quantify thelevels of the acoustic sources for a large fraction of the testing. Thephased array consisted of 416 Brüel & Kjær in: type 4938-W-001

Air IntakeSettling Chamber

Ventilation Air Intake

Exhaust Collector with Silencers

Exhaust Deflector

65 ′×75 ′×30 ′Anechoic Chamber

(200 Hz ~ 80,000 Hz)

Control Room9′×12 ′ (M = 0.25) or7′×30 ′ (M = 0.25)

Test Section

Fig. 2 Boeing’s LSAF.

Fig. 3 Photograph of the LSAF experimental setup for the HWBairframe noise experiment.

Fig. 4 Underside view of the 3% HWB airframe in the Boeing LSAF

facility.

Table 1 Test matrix and measurement locations

Parameter Value

Angle of attack 0, 4, 8, 10, 12, 13, 15Mach number 0.12, 0.16, 0.20, 0.24Slat angle 30, retractedInner elevon angle 0, 2, −5, −3.6Outer elevon angle 0, 60, −60Array position polar angle 50, 70, 90, 110, 130, 150Free-field microphone polarangle

50, 60, 70, 80, 90, 100, 110, 120, 130,140, 150

Free-field microphone sidelineangle

30, 60, 90

GUO ETAL. 2937

Dow

nloa

ded

by b

aew

arto

n@ai

aa.o

rg o

n Ja

nuar

y 27

, 201

4 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/1.J

0525

40

microphones, with B&K 2670-W-001 preamplifiers, flush mountedin a flat plate positioned 131.6 in. laterally from the streamwise-oriented vertical plane passing through the far-fieldmicrophone arrayorigin in the polar angle plane, as illustrated in Fig. 6. The plate wasattached to a mobile cart, which was mounted on a horizontaltraversing track (which traversed parallel to the wind-tunnel flowdirection), allowing for a wide range of phased array positionsrelative to the polar array origin. Phased array measurements wereacquired largely at geometric angles 50, 70, 90, 110, 130, and150 deg, and a limited set of measurements were acquired from 50to 150 deg in 10 deg increments. When acquiring far-field acousticdata, the cart was traversed to a stow location behind the tunnelcontraction.The BWB model rotation axis was located at �x; y; z� ��81.94; 15.0; 11.75 in:�, in wind-tunnel coordinates. For a phasedarray angle of 90 deg, the phased array originwas located at an xvalueof 100.439 in., approximately 18 in. downstream of the modelrotation point. The relative locations of the phased array, the model,and the tunnel exit are sketched in Fig. 6. The phased array consistedof four subarrays of various sizes, where the subarrays providedoverlapping coverage over the frequency range of interest. The foursizes are referred to as small (S), medium (M), large (L), and extralarge (XL), with the baseline subarrays containing, respectively, 170,199, 110, and 170 microphones. The horizontal/vertical subarray

apertures were approximately 12 × 9 in: (S), 26.2 × 19.8 in: (M),57.6 × 43.6 in: (L), and 126 × 95.6 in: (XL) in size. A sharing ofmicrophones between the various subarrays was used to reduce theoverall microphone count. Conventional beamforming was used forall of the phased array processing. To enhance the spatial resolution inthe beamforming maps, the diagonal of the cross-spectral matrixwas replaced by the row average. This is a technique similar todeleting the diagonal elements in the cross-spectral matrix toreduce background noise. The properties of the four subarrays aresummarized in Table 2.Owing to the presence of thewind-tunnel shear layer, a ray-tracing

algorithm was used [17]. In the implemented algorithm, the shear-layer half-velocity distribution at any given x location cross sectionwasmodeled as an ellipse. This model worked fairly well over awiderange of phased array geometric measurement angles. At the extremeangles (approximately 50 deg, and larger than ∼130 deg), shifts inthe measured source locations from the expected locations aroseand are attributed to increased differences between the modeled andactual shear layer location and shape.The beamforming grid was a rectangular grid with 0.5 in.

separation between grid points in both the x and z dimensions. Thegrid center was fixed in space and located at the model rotation point,illustrated in Fig. 7. In this configuration, the gridwas relatively closeto the primary noise sources of interest (which extended both aboveand below the grid plane). To account for the various model angles ofattack (0, 4, 8, 10, 12, 13, and 15 deg), the grid was likewise pitchedwith the model. The lower plot in Fig. 7 shows a side view schematicof the grid for a 15 deg angle of attack.For the subcomponent spectra extractions, a variety of sub-

component regions were defined as sketched qualitatively in Fig. 8.The regions for the component integration need to be definedaccording to the test configurations, even though the definitions ofthe components such as leading-edge slats and trailing-edge elevonsremain unchanged. In this definition, regions 1 to 4 are not related tothe airframe noise components; regions 1 accounts for the contami-nation noise source due the interaction between the upper wing of themodel and the test-section shear flow; regions 2 and 3 are used totrack some flow noise sources that appear at some test conditions; andregion 4 is used to provide information on noise floor levels. Theregions labeled 5 and 6 encompass part of the model center body andcenter trailing edge, respectively. For the slat noise sources, theregions are defined as region 13 and 16, as shown in Fig. 8. It shouldbe noted that region 16 for the upper wing is not truly slat noise; it isdue to the interactions at the top of the model between the wing andthe test-section shear layer flows, which is an artifact resulting fromthe wind-tunnel size limitation. In this case, the slat noise from theupper wing can be derived from region 13, as a mirror image.Extracted subcomponent spectra include the peak levels in each of the18 defined regions and the integrated levels. See [17] for a descriptionof the beamforming map integration method. For producing totalairframe noise levels, the spectral levels from regions 5 and 6 can betaken in their entirety and summedwith a doubling of the values fromthe lower half of the airframe.

Frequency (kHz)0 20 40 60 80 100

SPL

(dB

)

0

20

60

40

Slat Deployed

Slat Retracted

Background

Fig. 5 Comparison of far-field noise levels between the cases of slat

deployed, slat retracted and background without the HWB model.

x

y

Phased Array

Tunnel Contraction

FlowPolar Array

Origin +

Model Rotation

Point131.6′′

Fig. 6 Illustration ofBoeing’s LSAF test setup forHWBairframenoise.

Table 2 Properties of phased microphone arrays used

in the LSAF test

Size Symbol Frequencyrange, Hz

Aperturesize, in.

Number ofmicrophones

Extra large XL 700–3,330 126 × 95.6 170Large L 3,330–6,670 57.6 × 43.6 110Medium M 6,670–13,300 26.2 × 19.8 199Small S 13,300–33,300 12 × 9 170

Fig. 7 Phased array beamforming grid plane in reference to the HWBairframe model; forward view (upper) and side view (lower).

2938 GUO ETAL.

Dow

nloa

ded

by b

aew

arto

n@ai

aa.o

rg o

n Ja

nuar

y 27

, 201

4 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/1.J

0525

40

IV. Comparison with Free-Field Microphone Data

In addition to the phased array measurements, free-fieldmicrophones were also used in the test, deployed in a line in theflyover plane. The free-field microphone data were used to acquirequantitative far-field noise levels to provide an accurate assessment ofthe HWB airframe noise. However, the HWB model, which is apreexisting model from previous studies, is larger than the wind-tunnel test section. A compromise was made to resolve this issue byinstalling the model off the test-section center so that one wing of theHWB model is in the test-section uniform mean flow but the otherprotrudes into the shear layer of the open jet. This inevitablyintroduces an unwanted noise component due to the shear layerimpinging onto the HWB wing. The model setup and the potentialnoise sources are illustrated in Fig. 9. As a result, the free-fieldmicrophone data are contaminated by this unwanted noise com-ponent. This was considered in the test planning as a necessarycompromise with reduced expectations for the free-field microphonedata; the data will not be able to give an accurate level for the HWBairframe noise. The free-field microphone data, however, can stillbe used to calibrate the phased microphone array measurements bycomparing the free microphone data that include noise from allsources with the phased array data integrated over the entirebeamforming map and hence also including contributions from all

sources. The phased array can then be used to extract various noisecomponents, by integration of the phased array beamforming resultsin individual source regions, which are free from the contaminationdue to the wing-shear layer interaction and are relevant to HWBcomponent airframe noise. It is also recognized that, even with asmaller model that would not interact with the upper/lower shearlayers, a spurious noise sourcewould have still been generated by theinteraction of the model support with the jet shear layer, probably atlower frequencies. In that case, the phased arraymeasurementswouldstill provide a valuable tool to reject the contamination.To illustrate the accuracy and consistency of the phased array

measurements, when used to derive far-field spectral data, someexamples are given in Figs. 10 and 11, respectively, for 0 and 13 degangle of attack. The data plotted in these two figures are the 1∕3octave sound pressure levels at 90 deg emission angle in the flyoverplane. The results shown in these two figures are for the case ofMachnumber 0.2, which is the designed flight Mach number of the HWBaircraft. The results are very representative of other Mach numbers.The HWB configuration for these data has the leading-edge slatsdeployed at 30 deg. The phased array results in the comparison are theintegration of the beamforming maps over all the source regions. Inthe test setup, the array and the free-fieldmicrophones are at differentdistances from the test model, respectively at 11 and 25 ft fromthemodel center at the angle of 90 deg. The comparisons shown in thetwo figures are for array measurements that are normalized to themicrophone location according to the inverse square of the distance.The comparisons show that the phased array measurements

capture the spectral featureswell in the low- andmidfrequency range,but with noticeable differences for high frequencies. These trendsare also representative of other operational conditions. The good

Fig. 8 Definitions of surface regions for component noise integration.

Slat Noise

oiseElevon N

Shear Layer/Wing Interaction Noise

Fig. 9 Installation of HWB model in test section bounded by shear

layers.

Frequency (Hz)103 104 105

Phased Array

Microphone

1/3

Oct

ave

Ban

d SP

L (

dB)

10230

40

50

60

70

80

90

α = 0 degθ = 90 deg

M = 0.20

Fig. 10 Comparison between phased array measurements and free-field microphone data at 0 deg angle of attack.

Frequency (Hz)

1/3

Oct

ave

Ban

d SP

L (

dB)

102 103 104 10530

40

50

60

70

80

90Phased Array

Microphone

α = 13 degθ = 90 deg

M = 0.20

Fig. 11 Comparison between phased array measurements and free-

field microphone data at 13 deg angle of attack.

GUO ETAL. 2939

Dow

nloa

ded

by b

aew

arto

n@ai

aa.o

rg o

n Ja

nuar

y 27

, 201

4 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/1.J

0525

40

agreement in the midfrequency domain is understandable from thestandard beamformingmethod used to derive the array data; the high-frequency beamforming can suffer from decorrelation effects.Because no rigorous deconvolution is applied to the beamformingresults, the integrated spectra from the phased array data are notexactly the true far-field noise levels. Thus, the good agreementsbetween the phased array measurements and the free-fieldmicrophone data should only be considered as an added confidencefor using the results in the midfrequency domain for relativecomparisons and parametric studies.Thus, the focus of the data analysis in the rest of the paper will be in

the frequency domain between about 1000 Hz to about 30,000 Hz,model scale as measured. For a HWB model of 3% scale, thefrequency range at full scale would then be between about 30 Hz toabout 900 Hz. Obviously, this is on the low side of the frequencydomain for aircraft noise certification, but it is the only frequencydomain inwhich consistent parametric trends can be derived from thedatabase, as will be seen in the following sections. Furthermore, thefull-scale frequency range between 30 and 900Hz is relevant becauseslat noise is known to peak around 200 Hz for aircraft sizes scaled upby the factor of 3 percent. The scale factor can of course be eliminatedin the analysis by the use of Strouhal number, based on the slat chordlength and themean flow velocity. In this case, the frequency range of1000 to 30,000 Hz for model scale and 30 to 900 Hz for full-scaleaircraft corresponds to a range in Strouhal number between 0.2 and10. This clearly covers the peak Strouhal number between 1 and 2 forslat noise (e.g., [6–9]).

V. Effects of Angle of Attack on Slat Noise

Slat noise has a complicated dependency on the aircraft angle ofattack. This has been observed for conventional aircraft and has beenattributed to the degree of flow separation behind the slat that isinfluenced jointly by the aircraft angle of attack and the slatdeployment angle.With the slats deployed at a fixed angle, the case ofzero angle of attack usually causes large-scale flow separationbecause the slat chord line is at an angle to the inflow that isapproximately equal to the slat angle. Thus, the flow in this case seesthe slat as a bluff body with nonstreamlined shapes, causing large-scale separation behind it. This is usually not a configuration used inaircraft operations, but it serves as a good reference configuration fordiscussion. From this reference configuration, an increase in theaircraft angle of attack effectively reduces the angle between the slatchord line and the inflow direction and thus aligns the slat chord linebetter with the flow. As the angle of attack increases, the flow aroundthe slat transits from bluff-body separations to flow patterns that areusually designed for high-lift slats, namely a vortical flow in the slatcove region, a high-speed jet-like flow in the gap between the slattrailing edge and the main wing, and localized flow separation at theslat cusp. Because of the differences in the flow features of these twotypes of flow, the noise-generation mechanisms can also be expectedto differ in the two cases, which is why complex dependencies of slatnoise on aircraft angle of attack are usually observed in airframe noisestudies for conventional aircraft.This complex dependence similarly manifests itself for HWB

aircraft, as illustrated in Figs. 12 and 13, which plot the HWB slatnoise levels respectively for Mach numbers 0.12 and 0.2, each atvarious values of angle of attack, all measured at the overheadlocation of 90 deg of emission anglewith the slats deployed at 30 degand the elevon at 2 deg, which is the normal approach configurationfor this aircraft design. Clearly, the variations of the noise levels withincreasing angle of attack are notmonotonic and thevariations are notuniform in frequency either, due to the source mechanism changesdiscussed in the previous paragraph. For both cases shown in thesetwo figures, the lowest noise levels are for the case of 4 deg of angle ofattack. This trend is representative of other cases with different Machnumbers, different elevon settings, and other measurement angles, aswill be shown in subsequent discussions in the paper. Because theincrement in the angle of attack is not small, the minimum noiseconfigurationmay not be at precisely 4 deg; it is probably in the rangeof 3 to 5 deg.

It can also be seen from the figures that the spectral shapes atdifferent angles of attack differ from each other, with those at smallangles of attack dropping more rapidly above the spectral peak thanthose at large angles of attack. This seems to indicate noise sourcemechanisms that are narrow-banded in frequency for the small anglesof attack, consistent with the analysis of bluff-body flow separationthat is dominated by vortex shedding, andmechanisms at large anglesof attack, which are more broadband. It can also be clearly seen fromthe two figures that the spectral peak increases with the angle ofattack. For the datasets in each figure, the flowMach number and slatchord length is fixed. Thus, the Strouhal number scaling based onthese two parameters would predict a fixed peak frequency for eachfigure. This is clearly not the case, and the discrepancy results fromthe use of the length scale in the Strouhal number scaling. For flow-generated noise, the Strouhal number scaling should be based on alength scale that characterizes the unsteady flows: their coherencelength, for example [10,11]. For slat noise, the unsteady flows in thecove region and the slat-wing gap, and hence the characteristic lengthof the flows, are strongly affected by the angle of attack, which bringsin the dependence of the peak frequency on the angle of attack, asshown by the data. The Strouhal number scaling based on the slatchord length has been used for conventional aircraft with reasonableresults because the variations of the angle of attack for conventionalaircraft are not large,mostly confined between 4 and 7 deg, leading tosmall variations in the peak frequency. For a much larger range of theangle of attack, as is the case considered here from 0 to 15 deg, thevariations in the peak frequency also become larger. Evidently, thepeak frequency is not invariant even for fixed flowMach number andfixed slat chord length.The increase of the peak frequencywith the angle of attack, shown

in Figs. 12 and 13, can be explained by the changes that the flows inthe slat region experience as the angle of attack changes. The flowseparation and its coherence length scale depend on how theincoming flow is disturbed and blocked. With the slat deployed at30 deg, the flow experiences significant blockage at 0 deg angle of

Frequency (Hz)

3 104 105

α = 0 degα = 4 degα = 8 degα = 12 degα = 13 degα = 15 deg

SP

L(d

B)

1030

40

50

60

70M = 0.12

Fig. 12 Dependence of HWB slat noise on aircraft angle of attack for

M � 0.12.

Frequency (Hz)

3 104 105

α = 0 degα = 4 degα = 8 degα = 12 degα = 13 degα = 15 deg

SP

L(d

B)

1040

50

60

70

80M = 0.20

Fig. 13 Dependence of HWB slat noise on aircraft angle of attack forM � 0.2.

2940 GUO ETAL.

Dow

nloa

ded

by b

aew

arto

n@ai

aa.o

rg o

n Ja

nuar

y 27

, 201

4 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/1.J

0525

40

attack because the relative angle between the flow and the slat is30 deg.As the angle of attack increases, the relative angle between theflow and the slat decreases, until the angle of attack reaches the slatdeployment angle when the flow is aligned with the slat chord lineand only sees the thickness of the slat. This indicates that thecharacteristic length of the unsteady flow is reduced as the angle ofattack increases, corresponding to an increase in the characteristicfrequency.

VI. Dependence on Flow Mach Number

The dependence of slat noise spectra on the mean flow Machnumber is known to follow complex patterns; instead of collapsing ona single integer power law, the spectral shapes as well as theamplitudes vary with the Mach number, as documented anddiscussed in previous studies for conventional aircraft slat designs(e.g., [7,8]). This is found also to be the case for the HWB slat noiseand is illustrated in Figs. 14 and 15, respectively for 0 and 12 degangles of attack. The data are for the normal approach configurationwith the slats deployed at 30 deg and the elevons at 2 deg. The data arefrom the array located at the polar angle of 90 deg. In these twofigures, the slat noise levels at four Mach numbers are normalized bythe fifth power law and are plotted as a function of the Strouhalnumber based on the flow velocity and the slat chord length. The slatchord length is used for the Strouhal number scaling here because, foreach of the figures, the angle of attack is fixed; a different lengthwould only shift the location of the peak Strouhal number withoutany changes to the relative value, namely the collapse, of the peak

frequencies. This is sufficient for the discussions in this section only,which is to normalize both the frequencyparameter and the amplitudeby the flow velocity to discuss the dependence on the flow Machnumber, where a variable length scale should be considered ingeneral.It is clear that the fifth power law of flow Mach number does not

collapse the spectral data uniformly for all the frequencies, and thisnonuniformity varies with the angle of attack. At small angles ofattack (0 deg, for example), the collapse seems to be satisfactory forStrouhal number between about 2 to 10, but the results showsignificant scatter below the Strouhal number 2, as shown in Fig. 14.At large values of angle of attack (12 deg, for example), as shown inFig. 15, the trend seems to be the opposite; the data collapse well inthe low-frequency region around unity Strouhal number but shownoticeable scatter in the high-frequency region, above the Strouhalnumber of about 3.The deviation of the noise levels from the simple fifth power law of

the mean flow Mach number has been discussed in detail in [7]. Themain reason for this is the dependence of the spectral shape, inaddition to the spectral amplitude, on the flow Mach number. Thewell-known fifth power law is derived for the overall sound level,which would also be applicable to the spectral levels only if thespectral shape is invariant with the flowMach number. This is not thecase for slat noise, as shown by various test data for conventionalaircraft and discussed in detail in [7]. The data in Figs. 14 and 15 showthat this is also the case for slat noise for HWB aircraft, and the trendsin these figures are consistent with those observed in conventionalaircraft. For instance, Figs. 14 and 15 show that the deviation of thespectral levels from the fifth power law is mainly in the high-frequency domain, and the normalized datasets show higher levelsfor smaller Mach numbers. This means that the high-frequency noiseshould be scaled on a power law less than the fifth, again consistentwith data from conventional aircraft slats.

VII. Hybrid Wing–Body Slat Noise Modelingand Prediction

Slat noise modeling and prediction has been an active researchtopic in aeroacoustics because slats are a prominent feature inconventional aircraft design and are known to be one of the majorairframe noise components (e.g., [6–10]). The efforts, however, havelargely been focused on conventional slat designs and operatingconditions, characterized by smaller sweep angles, smaller chordlengths, and smaller angles of attack, in comparison with HWBdesigns. In principle, the prediction methodology developed in [7]should be applicable to slat noise of HWB designs because theprediction model is based on the component noise source mech-anisms independent of the aircraft type. To demonstrate this, pre-dictions will be made in this section for the HWB configuration, andthe results will be compared with the experimental data.Figures 16 and 17 show the sound pressure level (SPL) comparison

respectively for the flow Mach numbers 0.2 and 0.24, at variousangles of attack for the normal approach configuration with the slatsdeployed at 30 deg and the elevons at 2 deg. The data are from thearray located at the polar angle of 90 deg and the predictions are forthe noise at the emission angle of 90 deg. The levels are plotted as afunction of the measured small-scale frequency, but the amplitudesare shifted by the same amount for both the prediction and the data fora fixed angle of attack, so that the curves are clearly separated fromeach other for easy visualization. The amount of shifting for eachangle of attack is indicated in the plots. The predictions are plotted bythe solid curves and the data by the dashed curves. It is clear that thepredictions capture the overall spectral trends very well; both thespectral peaks and overall shapes are well predicted, and the absolutelevels are within a few dB between prediction and data.Both the data and the predictions in these two figures show that the

spectral peaks occur at about 3000 to 4000 Hz. If this is scaled to afull-scale aircraft the size of a Boeing 777, the scaling factor would be3%, which gives the peak frequency at about 100Hz for the full-scaleaircraft. This is lower than the peak slat noise frequency of about 200to 300 Hz for the Boeing 777 aircraft, an expected outcome because

Strouhal Number (fL/U)100 101

SPL

-50l

og(M

)(d

B)

70

80

90

100

110

M = 0.12M = 0.16

M = 0.20M = 0.24

120

LE Slatsα = 0 deg

θ = 90 deg

Fig. 14 Dependence of HWB slat noise on mean flow Mach number atzero angle of attack.

Strouhal Number (fL/U)

SPL

-50l

og(M

) (d

B)

100 10170

80

90

100

120 M = 0.12M = 0.16

M = 0.20M = 0.24110

LE Slatsα = 12 degθ = 90 deg

Fig. 15 Dependence of HWB slat noise on mean flow Mach number at12 deg of angle of attack.

GUO ETAL. 2941

Dow

nloa

ded

by b

aew

arto

n@ai

aa.o

rg o

n Ja

nuar

y 27

, 201

4 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/1.J

0525

40

of the larger slat chord length for the HWB aircraft. More detailedcomparisons between the two will be given in the next section. ForStrouhal number scaling based on the slat chord length and the meanflow velocity, the spectral peaks occur at about Strouhal number 2,which is known from previously published data and clearly seenin Fig. 15.As discussed in the previous section and more clearly shown in

Figs. 16 and 17, the peak frequencies of the spectra vary with theangles of attack, clearly shown by the data and correctly modeled inthe predictions. It can be seen that the peak frequency shifts upward asthe angle of attack increases. At first glance, this may seem to be asurprise because the peak frequencies are usually considered scalableby the Strouhal number based on the slat chord and the flow velocityand thus only depend on these two parameters. Although this can be afirst-order estimate for the peak frequencies, it actually misses animportant feature of the flow physics, namely the characteristiclength scale of the flow fluctuations. This length scale defines thecoherence length of the unsteady flows that furnish the noise sources.The slat chord length is probably of the same order as the coherencelength, but the two can be easily seen to be not equal. Thus, toaccurately model and scale the peak frequencies by the Strouhalnumber scaling, the length scale should be the flow characteristiclength instead of simply the chord length.Because the characteristic length scale of the unsteady flows in the

slat region is likely a function of other flow and operating parameters,with the angle of attack being one of them, it is then not a surprise thatthe peak frequencies of the slat noise spectra also vary with otherparameters. The effects of the angle of attack on the peak frequency

shift can be understood from the flow separation in the slat coveregion, which determines the coherence length scale. The intensity ofthe separation and its coherence length are affected by how theslats are aligned with the incoming mean flow. With the slats fixedat 30 deg, the mean flow at 0 deg of angle attack sees the mostblockages, and hence the flow separation in this case has the largestlength. As the angle of attack increases toward the slat deploymentangle, the flow becomesmore alignedwith the slat chord line, and thedimension that blocks the flow decreases until that dimensionbecomes the slat thickness when the angle of attack is about the sameas the slat deployment angle. The decrease in the length scale as theangle of attack increases from zero corresponds to an increase in thepeak frequency, as shown in Figs. 16 and 17. This feature is clearlyseen in the data and is also correctly modeled in the prediction,leading to the agreement in the comparison.The dependence of the slat noise on the flow Mach number is

shown in Figs. 18 and 19, with the overall sound pressure level(OASPL) plotted as a function of the flowMach number,with variousangles of attack for each diagram in the figure. Both figures are for thearray location at 90 deg and slats deployed at 30 deg, with Fig. 18 forthe elevon setting of 2 deg and Fig. 19 for −5 deg, as indicated ineach diagram. The case of negative elevon angle means that theelevons are deployed upward. The predictions are plotted as curves,and the data are shown by the symbols. For all cases, good agreementbetween prediction and data can be seen from the figure, in both thetrends in Mach number and the absolute levels of the OASPL.From the data discussed in previous sections, especially in Figs. 12

and 13, it is clear that the slat noise levels are a function of the angle ofattack, and the effects of the angles of attack on the noise levels varywith flow Mach number. The data consistently show that the lowestslat noise comes from the case of 4 deg of angle of attack, for variousflow Mach numbers, different elevon deployment settings, andvarious measurement locations. Thus, even though the sparseparametric variations in the angle of attack from the test data preventsthe minimum noise angle of attack from being definitely identified, it

Frequency (Hz)104 105

SP

L(d

B)

10340

60

80

100

120

α = 4 deg

8 deg (+5 dB)

12 deg (+15 dB)

13 deg (+25 dB)

M = 0.24

15 deg (+35 dB)

Fig. 17 Comparison of SPL between prediction (solid curves) and data(dashed curves) forM � 0.24.

Mach Number0.15 0.2 0.25 0.3

α = 0 degα = 8 degα = 10 degα = 12 degα = 13 degα = 15 deg

OA

SPL

(dB

)

0.160

70

80

90

100Elevon at -5 DegArray at 90 Deg

Fig. 18 Comparison of Mach number dependence between prediction

(curves) and data (symbols) with elevons deployed at −5 deg.

Elevon at +2 DegArrayat 90 Deg

α = 0 degα = 4 degα = 8 degα = 12 degα = 13 degα = 15 deg

Mach Number

OA

SP

L(d

B)

0.1 0.15 0.2 0.25 0.360

70

80

90

100

Fig. 19 Comparison of Mach number dependence between prediction(curves) and data (symbols) with elevons deployed at 2 deg.

Frequency (Hz)104 105

15 deg (+35 dB)

α = 4 deg

8 deg (+5 dB)

12 deg (+15 dB)

13 deg (+25 dB)

M = 0.20

SP

L(d

B)

10340

50

60

70

80

90

100

110

120

Fig. 16 Comparison of SPL between prediction (solid curves) and data(dashed curves) forM � 0.2.

2942 GUO ETAL.

Dow

nloa

ded

by b

aew

arto

n@ai

aa.o

rg o

n Ja

nuar

y 27

, 201

4 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/1.J

0525

40

is certain that there is a minimum slat noise range for the angle ofattack, in the range between about 3 to 5 deg.This feature is modeled in the prediction methodology with some

results shown in Figs. 20 and 21, where the OASPL is plotted as afunction of the angle of attack, for four flowMach numbers in each ofthe plots. The results are for the measurement location of 90 deg withthe two figures respectively for two elevon settings,−5 and�2 deg,as indicated in each diagram. The predictions are plotted by thecurves and the data are shownby the symbols. Clearly, the predictionscapture the effects of the angle of attack very well, for both theparametric trends and the absolute amplitudes.

VIII. Comparison with Conventional Aircraft

The HWB slat noise may not only dominate over other com-ponents, except perhaps for the landing gear noise, but also mayhave levels comparable to or even higher than the slat noise ofconventional aircraft. To quantify this, the data from the small-scalemodel test are extrapolated to full scale, to an aircraft comparable tothe Boeing 777 aircraft in size and functionality, and compared to theslat noise levels of the 777 aircraft. The results are given in Fig. 22,with the squares for the HWB aircraft, extrapolated from small to fullscale, and the diamonds for the 777 slat noise, derived from a flighttest [18]. The extrapolation uses a dimension scaling factor of 3%,applied to the HWB aircraft, which scales up the noise amplitude andscales down the frequencies according to standard procedures. Thegeometry and operational parameters of the two aircraft are listed inTable 3. The dimensions of the slats for the HWB aircraft are scaledup by dividing the small-scale dimensions by 0.03. In addition, thefrequencies of the small-scale test data are scaled down by themultiplying factor 0.03. The amplitudes of the noise are also scaledup by the square of the scaling factor. All of the data are extrapolatedto the distance of 394 ft, which is the altitude for aircraft noise

certification at approach conditions, at the observation angle of90 deg, namely the overhead location.A striking conclusion from the data shown in Fig. 22 is that the

HWB aircraft generates more slat noise than the Boeing 777 aircraft,even though the former flies at a lower Mach number than the latter(0.2 for the HWB and 0.25 for the Boeing 777). It is known that thepeak spectral levels of slat noise scale on the fifth power of Machnumber. According to this, the HWB slats would have lower noiselevels, by about 4.8 dB, in comparison with the levels for the Boeing777 aircraft, due to the lower flight Mach number. This is obviouslynot the case shown by the data, which include other effects, as well asthe effects of Mach number. Instead, the HWB aircraft generatessubstantially more noise in the low-frequency region around about100Hz than theBoeing 777 does, by asmuch as 10 dB, and has levelsabout 2 dB higher than the Boeing 777 for frequencies from about100 to 1000 Hz. The higher noise levels at low frequencies from theHWB aircraft is partially due to the larger slat chord length, whichleads to lower peak frequency, and partially due to the increase ofnoise amplitude that is proportional to the slat chord length. Theoverall higher slat noise levels of the HWB aircraft are, however,largely due to the different slat operating conditions. From Table 3, itis clear that the source area, approximately the product of the slatchord and span length, is comparable between the two cases:38; 515.6 in:2 for the HWB and 31; 899.3 in:2 for the 777, cor-responding to a difference of about 0.8 dB in noise levels. Thus, thenoise advantage of theHWB(of about 4.8 dB) from theMach numberdifference must have been offset by an increase in the source densityor source strength. This source increase comes from the slat sweepangle and the aircraft angle of attack, for both of which the HWBaircraft has higher values. The data and prediction given in previoussections confirm this; for example, the difference in the angle ofattack can lead to a noise increase for the HWB aircraft of about 8 dBin a broad frequency range. This 8 dB noise increasewill of course bepartially offset by the 4.8 dB noise benefit due to the lower Machnumber, which is probably why the HWB slat noise in Fig. 22 is seento be about 2 to 3 dB higher than the slat noise of the Boeing 777aircraft for the high-frequency range.The result shown in Fig. 22 is a valuable conclusion on HWB slat

noise. It not only quantifies the noise levels but also illustrates theshortfall of simply using the Mach number scaling to infer a lower

AOA (Deg)

OA

SPL

(dB

)

0 5 10 15 2060

70

80

90

M = 0.12

0.20

0.24

Elevon at -5 DegArray at 90 Deg

100

0.16

Fig. 20 Comparison of the dependence on the angle of attack between

prediction (curves) and data (symbols) for slats at 30 deg with elevonsdeployed at −5 deg.

AOA (Deg)0 5 10 15 20

OA

SP

L(d

B)

60

70

80

90

M= 0.12

0.16

0.20

0.24

Elevon at +2 DegArray at 90 Deg

100

Fig. 21 Comparison of the dependence on the angle of attack betweenprediction (curves) and data (symbols) for slats at 30 deg with elevonsdeployed at 2 deg.

Frequency (Hz)102 103 104

80

1/3

Oct

ave

SPL

(dB

)

50

55

60

65

70

75

HWB

777

Fig. 22 Comparison of slat noise levels between HWB and 777 aircraft.

Table 3 Geometry and operation parametersof HWB and 777 aircraft

HWB 777

Chord length, in. 36.3 27.1Span length, in. 1062.5 1178.4Slat angle, deg 30 30Sweep angle, deg 37.8 34Angle of attack, deg 12 6Flight Mach number 0.2 0.25

GUO ETAL. 2943

Dow

nloa

ded

by b

aew

arto

n@ai

aa.o

rg o

n Ja

nuar

y 27

, 201

4 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/1.J

0525

40

noise level for the HWB aircraft. It should be recognized that theMach number scaling law, being approximately the fifth power forslat noise, is applicable only if all other conditions, including bothdesign parameters and operating conditions, are the same, which isclearly not the case when comparing the HWB aircraft withconventional aircraft. In this case, changes in other conditions mayalter the source characteristics, causing noise changes that are notgoverned by the Mach number scaling law. In the case of HWB slatnoise, the data show that the acoustic benefit of lower Mach numberfor the HWB aircraft is completely offset and taken over by thepenalty due to the increase in noise source strength, mostly due to thelarger angle of attack.

IX. Conclusions

An analysis has been presented for the slat noise of the HWBaircraft design, revealing various functional trends of the noise levelsfor parameters such as the flow Mach number and the aircraft angleof attack. It has been shown that the overall slat noise levelsapproximately follow the fifth power law in flow Mach number, butthe dependence of the noise spectra on the Mach number is muchmore complicated, with effects manifesting themselves not only inthe amplitudes but also in the spectral shapes, confirming featurespreviously discussed for slat noise of conventional aircraft. Theeffects of the aircraft angle of attack on the slat noise have been shownto be both in the noise levels and in the peak frequencies. For theamplitudes, the slat noise seems to assume aminimum in the range ofabout 3 to 5 deg of angle of attack, and increases in level as the angleof attack moves away from this minimum noise range. For the peakfrequencies, it has been shown that an increase in the angle of attack,from zero to the slat deployment angle, causes an upward shift in thepeak frequencies, a feature that can be explained by the decrease ofthe characteristic length scale of the flow separation in the slat flowregion as the angle of attack increases. All of these features have beenmodeled in slat noise-prediction methodologies, which have beenshown to agree well with the experimental data, not only capturingthe correct parametric trends but also accurately predicting theabsolute amplitudes, especially for the cases of large angles of attackfor which test data were not previously available.One of themain conclusions has been that the HWB slat noisemay

have levels comparable to or even higher than those of conventionalaircraft, even though the HWB aircraft usually has a lower flightMach number. This may seem to be a surprise because airframe noiseis usually scaled on either the fifth or the sixth power law on Machnumber; a difference between 0.2 for HWB aircraft and 0.25 forconventional aircraft would correspond to a noise reduction of about5 dB. However, this assumes that all the other conditions are thesame between the two aircraft being compared, including operatingconditions and aircraft designs, which is obviously not the case. Thetwo aircraft have drastically different designs and operate at differentconditions. Thus, the Mach number scaling laws should not beblindly applied. Instead, the individual noise components should beexamined to account for the unique features of each design detail thatmay impact the noise source mechanisms. In comparison with con-ventional aircraft designs, the slats of the HWB aircraft have largerdimensions and larger sweep angles. Furthermore, the HWB aircraftis designed to operate at large angles of attack, usually above 10 deg,compared with conventional aircraft that operate in the range of 4 to7 deg. All of these increase levels of the slat noise of the HWBaircraft, which more than offsets the benefit of its lower flight Machnumber. It has been shown by the test data that the large angles ofattack alone can increase the slat noise by as much as 8 dB from thelevels of conventional aircraft slat noise. By scaling the model testdata to a full-scale configuration that is comparable to the Boeing 777aircraft in dimensions and functionalities, it has been shown that theHWB has more slat noise than the Boeing 777 aircraft.It is important to point out that the study presented here does not

include discussions on the effect of thewind-tunnel test configurationon the aerodynamics of the HWB model. It is known that free-jetmean flows may be bent by the presence of high-lift models,effectively increasing the mean flow angles of attack. This effect, of

course, depends on the relative dimensions of the test model and thetest section. Though there were no aerodynamic measurements inthe HWB tests, the model size was considered to be small enoughfor the LSAF facility, and significant flow changes were notexpected, based on experience with various previous tests in thefacility. The vertical mounting of the model in the test section alsohelps to alleviate any potential flow blockage effect because the largerof the two dimensions of the wind-tunnel cross section is in the wingtransverse direction. From the model size and the test-sectiongeometry, it can be estimated that the flow domains on both thepressure and the suction side of theHWBwing are about 5 times of theaveragewing chord length, which is considered sufficient for themeanflow to mimic free space conditions. It should be acknowledged,however, that the large angles of attack and the large lift of the HWBmodelmay induce blockage to themean flow; thus, the potential for theeffective increase of the aircraft angles of attack, which should bechecked out in future studies either by aerodynamic measurements orbycomputation.Should this turnout tobe the case, themain results andconclusions presented here would be further reinforced because theangles of attack would be even higher for the HWB model to becompared with flight configurations, which would correspond tofurther slat noise increase for the HWB model.

Acknowledgments

The work reported here was sponsored by the NASAEnvironmentally Responsible Aviation Project underNASA contractNNL04AA11B, task order NNL10AA71T. Dan Vicroy, NASALangley Research Center, is acknowledged for making available the3% blended wing–body model for the Boeing Low SpeedAeroacoustic Facility experiment.

References

[1] Clark, L.R., andGerhold,C.H., “InletNoiseReduction byShielding forthe Blended Wing Body Airplane,” 15th Aerodynamic Decelerator

Systems Technology Conference, AIAA Paper 1999-1937, June 1999.[2] Hill, G. A., Brown, S. A., Geiselhart, K. A., and Burg, C. M.,

“Integration of Propulsion Airframe Aeroacoustic Technologies andDesign Concepts for a Quiet Blended Wing Body Transport,” AIAA 1st

Intelligent Systems Technical Conference, AIAA Paper 2004-6306,Sept. 2004.

[3] Thomas, R. H., Burley, C. L., and Olson, E. D., “Hybrid Wing BodyAircraft System Noise Assessment with Propulsion AirframeAeroacoustic Experiments,” International Journal of Aeroacoustics,Vol. 11, Nos. 3–4, 2012, pp. 369–410.doi:10.1260/1475-472X.11.3-4.369

[4] Czech, M. J., Thomas, R. H., and Elkoby, R., “Propulsion AirframeAeroacoustic Integration Effects of a Hybrid Wing Body AircraftConfiguration,” International Journal of Aeroacoustics, Vol. 11,Nos. 3–4, 2012, pp. 335–368.doi:10.1260/1475-472X.11.3-4.335

[5] Bonet, J. T., Schellenger, H. G., Rawdon, B. K., Elmer, K. R.,Wakayama, S. R., Brown, D., and Guo, Y. P., “EnvironmentallyResponsible Aviation (ERA) Project—N � 2 Advanced VehicleConcepts Study and Conceptual Design of Subscale Test Vehicle(STV),” NASA CR-NND11AG03C, Dec. 2011.

[6] Guo, Y. P., “A Discrete Vortex Model for Slat Noise Prediction,” 7th

AIAA/CEASAeroacoustics Conference andExhibit, AIAAPaper 2001-2157, May 2001.

[7] Guo, Y. P., “Slat Noise Modeling and Prediction,” Journal of Sound

Vibration, Vol. 331, No. 15, 2012, pp. 3567–3586.doi:10.1016/j.jsv.2012.03.016

[8] Dobrzynski, W., and Pott-Pollenske, M., “Slat Noise Source Studies forFarfield Noise Prediction,” 7th AIAA/CEAS Aeroacoustics Conference

and Exhibit, AIAA Paper 2001-2158, May 2001.[9] Mendoza, J. M., Brooks, T. F., and Humphery, W. M., “Aeroacoustic

Measurements of a Wing/Slat Model,” 8th AIAA/CEAS Aeroacoustics

Conference and Exhibit, AIAA Paper 2002-2604, June 2002.[10] Guo, Y. P., Joshi, M. C., Bent, P. H., and Yamamoto, K. J., “Surface

Pressure Fluctuations on Aircraft Flaps and Their Correlation with FarFieldNoise,” Journal of FluidMechanics, Vol. 415, July 2000, pp. 175–202.doi:10.1017/S0022112000008740

2944 GUO ETAL.

Dow

nloa

ded

by b

aew

arto

n@ai

aa.o

rg o

n Ja

nuar

y 27

, 201

4 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/1.J

0525

40

[11] Guo, Y., and Joshi, M., “Noise Characteristics of Aircraft High LiftSystem,” AIAA Journal, Vol. 41, No. 7, 2003, pp. 1247–1256.doi:10.2514/2.2093

[12] Dobrzynski, W., Nagakura, K.,, Gehlhar, B., and Buschbaum, A.,“Airframe Noise Studies on Wings with Deployed High-Lift System,”4th AIAA/CEAS Aeroacoustics Conference, AIAA Paper 1998-2337,June 1998

[13] Brooks, T. F., and Humphreys, W. M., Jr., “Flap Edge AeroacousticMeasurements and Predictions,” Journal of Sound and Vibration,Vol. 261, No. 1, 2003, pp. 31–74.doi:10.1016/S0022-460X(02)00939-2

[14] Liebeck, R. H., “Design of the Blended-Wing-Body SubsonicTransport,” 40th AIAA Aerospace Sciences Meeting and Exhibit, AIAAPaper 2002-0002, Jan. 2002.

[15] Owens, D. B., Brandon, J.M., Croom,M. A., Fremaux, C.M., Heim, E.H., andVicroy, D.D., “Overview ofDynamicTest Techniques for FlightDynamics Research at NASA LaRC,” 25th AIAA Aerodynamic

Measurement Technology and Ground Testing Conference, AIAAPaper 2006-3146, June 2006.

[16] Shields, F. D., and Bass, H. E., “A Study of Atmospheric Absorption ofHigh Frequency and Application to Fractional Octave Bands of Noise,”NASA CR-2760, 1976.

[17] Dougherty, R. P., “Beamforming in Acoustic Testing,” Aeroacoustic

Measurements, editedbyMueller,T. S., Springer-Verlag,NewYork, 2002,pp. 1–25.

[18] Herkes, W.H., Olsen, R.F., and Uellenberg, S., “The Quiet TechnologyDemonstrator Program: Flight Validation of Airplane Noise-ReductionConcepts,” 12th AIAA/CEAS Aeroacoustics Conference, AIAA Paper2006-2720, May 2006.

D. PapamoschouAssociate Editor

GUO ETAL. 2945

Dow

nloa

ded

by b

aew

arto

n@ai

aa.o

rg o

n Ja

nuar

y 27

, 201

4 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/1.J

0525

40