Flight-Test Evaluation of Landing Gear Noise Reduction Technologies

Flight - Test Evaluation of Landing Gear Noise Reduction

Technologies

1 2 3 Mehdi R. Khorrami , David P. Lockard , William M. Humphreys, Jr.

N ASA Langley Research Center, Hampton, VA, 23681, USA and Patricio A. Ravetta AVEC Inc., Blacksburg, Virginia 24060, USA Results from the third Acoustic s Research M easurements flight test , conducted under the NASA Flight Demonstrations and Capabilities project , are presented and discussed . The test evaluat ed landing gear and gear cavity noise mitigation technologies installed on a NASA Gulfstream G - III. Aircraft configurations with and without main landing gear treatments were flown at several flap deflections to determine the acoustic performance of the technolo- gies for aircraft equipped with conventional Fowler flap s . With the aircraft flying an approach path and engines at “ground - idle , ” extensive acoustic measurements were acquired with a phased microphone array system. Computed beamform maps wer e used to examine the effec- tiveness of the test ed technologies in redu cing the strength of the noise sources generated by the main landing gear. Various integration regions were devised to extract the farfield noise spectra associated with the treated and untreated landing gear configurations . A nalys e s of the gathered acous tic data demonstrate that significant noise reduction was achieved. How- ever , the full noise reduction potential of the technologies could not be determined because of contamination from flap inboard edge noise and other secondary sources .

Nomenclature AOA = Angle of attack f = Frequency ACTE = Adaptive Compliant Trailing Edge AFB = Air Force Base AFRC = Armstrong Flight Research Center AFRL = U. S. Air Force Research Laboratory ARMD = Aeronautics Research Mission Directorate EPNL = Effective Perceived Noise Level ERA = Environmentally Responsible Aviation FDC = Flight Demonstrations and Capabilities MLG = Main l anding g ear NR = Noise reduction PKF = Porous knee fairing PSD = Power s pectral d ensity SCRAT = SubsoniC Research Aircraft Testbed SPL = Sound p ressure l evel UF = Upper fairings Aerospace E ngineer, Computational AeroSciences Branch, A ssociate F ellow AIAA.

Aerospace E ngineer, Computational AeroSciences Branch, S enior M ember AIAA .

Aerospace E ngineer, Advanced Measurement and Data Systems Branch, A ssociate F ellow AIAA .

Co - owner, C hief R esearch E ngineer, S enior M ember AIAA .

I. Introduction The a vailability of a broad range of technologies with superior aeroacoustic performance is essential for the pro- duction of quiet , efficient aircraft that will foster continued growth [1, 2] in internationa l civil aviation while reducing community noise . Noise generated by the a irframe dominates the acoustic spectrum of most modern aircraft during landing, with the undercarriage and wing high - lift devices, such as flaps and slats, constituting the primary sources [3].

Development and maturation of viable technologies that substantially reduce the airframe component of aircraft noise is of importance to the NASA Aeronautics Research Mission Directorate (ARMD). A major effort on airframe noise reduction was initiated under the Environmentally Responsible Aviation (ERA) project. N umerous concepts for mitigating the noise produced by aircraft flaps and undercarriage were conceived and evaluated via extensive model - scale testing and high - fidelity s imulations [4 ]. After the ERA p roject ended in 2015, a full - scale evaluation of the most promising technologies in a relevant environment was pursued under the Flight Demonstrations and Capabilities (FDC) project. T he Acoustics Research Measurements (ARM) flights , a campaign that comprised three separate two - and - a - half - month long tests, were completed over a two - year period between 2016 and 2018. S elect landing gear and flap noise reduction (NR) technologies were integrated onto a NASA Gulfstream G - III airc raft to determine their individual and aggregated effectiveness.

The first campaign (ARM - I, 2016) evaluated the Adaptive Compliant Trailing Edge (ACTE) concept as a mecha- nism to reduce flap noi se [5, 6 ]. The second test (ARM - II, 2017) assessed the effectiv eness of various main landing gear (MLG) and gear cavity NR treatments in combination with the ACTE technology [5, 6] . For the third flight test (ARM - III, 2018), the ACTE flaps were removed and the original Fowler flaps were reinstalled on the G - III aircra ft to obtain baseline flap and landing gear data, and to assess the noise reduction capability of the landing gear technologies for conventional flaps. E xtensive acoustic measurements were acquired during the ARM tests with a NASA - developed phased microphone array comprised of 185 sensors [ 7 ] . Additionally, t hree pole - mounted microphones were used for Effective Perceived Noise Level ( EPNL ) measurements . Processed phased array data from the ARM - I and ARM - II tests, presented in Ref. [ 5 ], demonstrated that significant reductions in acoustic energy for the sources associated with the flaps and MLG were obtained when the ACTE concept and the gear treatments were installed. The effectiveness of the noise reduction technologies was observed over a wide rang e of forward directivity angles [ 5 ].

An in - depth analys i s of the acoustic data acquired during the ARM - III test is presented in this paper. Data quality and repeatability are assessed for various baseline and treated landing gear configurations , and the a coustic perfor- mance of landing gear noise reductions technologies is evaluated for conventional (Fowler) flaps.

II. Test Aircraft and Test Site All ARM flights were conducted with two Gulfstream G - III aircraft based at the NASA Armstrong Flight Research Center (AFRC). The primary G - III used during the ARM - I and ARM - II tests was the SubsoniC Research Aircraft Testbed (SCRAT), also known by its tail number as “804” (see Fig. 1a) [5]. The heavily instrumented 804 allows in - flight recording of aircraft parameters such as global position, angle of attack (AOA), and true airspeed (TAS). The second aircraft, also known by its tail number as “808” (Fig. 1b), was a stock G - III that served as the initial baseline configuration. For the ARM - III test, the ACTE flaps were removed from 804 and the original Fowler flaps were reinstalled. This conversion allowed 804 to serve both as the baseline, unmodified testbed and a vehicle for evaluating the acoustic performance of the MLG and cavity NR technologies in conjunction with conventional (Fowler) flaps.

Detailed inf ormation for the 804 and 808 aircraft is provided in Ref. [ 6 ].

Both ARM - I and ARM - II flight tests were conducted on the Rogers dry lakebed at Edwards Air Force Base (AFB) in S outhern California. The microphone array and pole - mounted certification microphon es were deployed at the North end of runway 18L. This location provided ample flat terrain for the disposition of various elements of the ground operations and data collection hardware. To avoid possible delays/cancellations due to rainy spring weather, th e mi- crophone array and certification microphones were deployed on the overrun section of inactive runway 24 of North base at Edwards AFB during ARM - III. Although t he new deployment site was relatively close to the lakebed site used during ARM - I and ARM - II , it introduced several notable differences that affected the acoustic measurements. Fore- most among these differences were a) placement of the microphone array on a paved , concrete surface versus the compact dirt surface of the lakebed during ARM - I and ARM - II , and b) close r proximity of the inactive runway to a major highway just outside the AFB ground s that resulted in higher background noise levels.

a) NASA 804 (SCRAT) b) NASA 808 Fig . 1 Gulfstream G - III aircraft with Fowler flaps .

A. Noise Reduction Technologies The landing gear NR technologies evaluated during the ARM flight tests were a collection of fairings installed on the main landing gear and two cavity treatments. The G - III main landing gear with and without acoustic treatment is shown in Fig. 2 . The fairings comprise the porous knee fairing (PKF) covering the front post plus an assortment of smaller fairings that are collectively referred to as upper fairings (UF). The two cavity treatments consist ed of a stretchable mesh and a concept that combines chevrons at the leading edge of the cavity with an acoustic foam treat- ment on the downstream side of the cavity. A close - up view of the se two concept s, as installed on the 804 aircraft, is shown in Fig. 3 . A full description of these technologies can be found in Refs. [5, 6].

a) Baseline b) With fairings installed Fig . 2 Gulfstream G - III main landing gear. Baseline gear is shown in its compressed , on the ground state ; the faired gear is depicted in its stretched , in - flight state (f rom Ref. [5] ) .

a) Stretchable mesh b) Chevron s and acoustic foam Fig . 3 Flight tested main gear cavity treatments as installed on 804 aircraft (f rom Ref. [5] ) .

III. Test Procedure and Measurements A full account of the flight and ground operations in support of the airframe noise test campaign is provided in Ref. [6 ]. Recordi ng of l ocal meteorological conditions , their correspon ding acoustic data corrections, and microphone array data processing techniques used in the present acoustic analyses are the same as those employed for ARM - I and ARM - II test data [5, 8 ]. Background noi se levels at the new location were higher than those encountered during the first two tests, as noted earlier.

Adhering to the procedure established for ARM - I and ARM - II, all array flyovers were executed with the aircraft engines operating at “ground - idle” to minimize contamination of the acoustic measurements by propulsion noise. As in the previous ARM tests, to ensure that the gathered data were statistically meaningful, multiple passes du ring each flight and multiple flights on different days were executed for most aircraft configurations. To determine velocity scaling, acoustic measurements for most configurations of interest were obtained at 140, 150, and 165 kts with the middle value re presenting the speed at which the majority of the measurements were taken. With the completion of ARM - III, data collection s panned multiple years for a few, select configurations . As a result, pass - to - pass, day - to - day, and year - to - year variations in the me asured noise signature s and the resulting uncertainties can be determined and assessed.

A . Array Data Processing A description of the processing approach applied to the microphone array data collected during the ARM tests is provided in Ref. [5 ]. An overview highlighting the salient aspects of the method ology is provided here for complete- ness.

A time - domain, delay - and - sum beamformer with diagonal removal coupled with CLEAN deconvolution , both available within AVEC’s Phased Array S oftware S uite [ 9] , w as used to process the microphone array data collected during the ARM test s . B eamform (source localization) maps were generate d on a square, 100 ft by 100 ft (30.5 m by 30.5 m) planar grid covering the entire aircraft. A g rid size of 201 × 201 points , representing a spatial resolution of 6 in . (15.25 cm), was deemed adequate for generation of the maps that provid e d the position and sound pressure level (SPL) of the sources . T he results presented here were scaled to an altitude of 394 ft . (120 m) under the assum ption of 2 2 spherical spreading for pressure ( p  ~ 1/r ).

T he maps provide valuable insight in to the effect that the NR technologies have on the intended noise sources. To generate the farfield spectra, the deconvolved CLEAN maps were integrated with a cutoff of 10 dB from the peak level in the map for each frequency . For illustrative purposes, a sample conventional beamform map at a frequency of 2 , 250 Hz is displayed in Fig. 4 for the 808 aircraft with the Fowler flaps defle cted at 39º and the MLG deployed .

To isolate relative source strengths, the maximum SPL has been subtracted from all levels in the map , and a range of 10 dB below the maximum has been used. This convention is also followed throughout th e paper , including i nstances when beamform maps from different datasets are compared . T he maximum SPL within a set of maps for a given frequency was subtracted from all levels in the set, and only sources within 10 dB of the maximum are displayed.

As outlined in Ref. [5], t wo integration regions were used to isolate flap and MLG contributions to the farfield noise and to assess the noise reduction performance of the tested landing gear technologies. The se regions are high- lighted in Fig. 4 : (1) a delta - shaped region named “Wings N g” that excludes the contributions from the nose gear, wing tips and leading edges , and engines ; and (2) a small , rectangular region called “MLG” that contain s the main landing gear s . The “ WingsNg” region allowed us to determine the acoustic benefit s of the landing gear technologies when the noise produced by the aircraft Fowler flap system was also included in the farfield spectrum . The initial “ MLG” region (the smaller rectangular area enclosed by the dash - dot black line) was used to assess the iso lated effectiveness of the main gear fairings and cavity treatments by excluding a significant portion of the flap noise . The se integration regions were devised to either remove or substantially diminish secondary source contribution s to the farfield signature. Nev- ertheless, as can be seen in Fig. 4, the “MLG” region contains the inboard flap tips , which are known to be prominent noise sources at low - to mid - frequenc ies , especially when the flap s are highly deflect ed . Use of the small “MLG” region to partially isolate the main gear noise sources produc ed inconsistent results : changes in the size and location of the inboard tip noise sources within the beamform maps yielded significantly different integrated spectral levels for repeated flyover passes o f the same aircraft configuration. Thus , proper interpretation of the acoustic performance of the tested landing gear technologies became elusive . To circumvent the problem, the initial “MLG” integration region was expanded in the downstream direction (hig hlighted by the dash - dot red line) to fully enclose the inboard flap tip sources. T he integrated levels obtained from the enlarged “MLG” region produced very consistent trends with good data repeatability , as will be shown in the following section. However , this consistency precluded the ability to ascer- tain the full performance of the tested technologies due to noise contribution from the inboard tip sources. T he reasons for choosing the shape and extent of these two integration regions are discussed in R ef. [ 5 ]. Unless specified otherwise, the “MLG” - based integrated spectra shown in this paper were obtained from this enlarged integration zone.

Fig . 4 P rimary integration regions used to include or exclude contributions of measured sources to the far- field noise spectrum . A s ample acoustic map using conventional beamforming is shown (f rom Ref. [5] ) .

IV. Results and Discussion T he highest - quality acoustic results were obtained during the time period that correspond ed to the aircraft being located along a streamwise segment ± 50 ft from the array center (overhead position) . Since the aircraft was closest to the array microphones at this time , atmospheric attenuation and wind - induced decorrelation of high - frequency sound were minimal . Therefore, resu lts correspond ing to the overhead position a re prominent in the data and analys e s pre- sented here . In those instances when the performance of various NR technologies is being examined , data at other angles are also presented to illustrate that the acoustic benefits are maintained over a wide range of forward directivity angles.

A. Data Quality and Repeatability The extensive acoustic measurements obtained during approximately 1100 flyover passes performed during the 2016 (ARM - I), 2017 (ARM - II), and 2018 ( ARM - III) tests permit a careful evaluation of data quality and repea ta bility in a manner similar to that used during model - scale wind tunnel test s . Although the number of aircraft passes for most configurations w as less than that needed for a typical statistical analysis, it was sufficient to ascertain pass - to - pass, day - to - day, and even year - to - year data repeatability. The g eneration and resultant quality of the beamform maps are discussed in detail in Ref. [5]. Because of the vast amount of data coll ected during the tests, only a very limited number of beamform maps are presented to demonstrate repeatability of the extracted noise sources or to illustrate the noise reduction capability of the tested technologies . Thus, the analysis is confined to dis cussing mostly the integrated farfield spectra to highlight the salient trends and features of the collected a coustic data.

1. Velocity Scaling Velocity scaling is very important during the reduction of data obtained from airframe noise flight test s . Unlike model - scale tests in wind tunnel facilities, exact repetition of target airspeeds is virtually impossible to obtain during actual aircraft flyby passes. Therefore, the resulting spectral levels must be scaled properly for comparison purposes .

As in Ref. [5 ], throughout this paper we have used V velocity scaling to adjust the measured levels. To illustrate the suitability of this scaling, integrated power spectral density (PSD) values at three vastly different speeds for the base- line 804 aircraft with flaps deflected at 20° and landing gear retracted are shown in Fig. 5a. The data were taken during the ARM - III test and the WingsNg region was used to integrat e the beamform maps. The corresponding V - scaled levels are displayed in Fig. 5b. Good data collapse is observed for frequencies up to 5 kHz , confirming the dependency of flap noise on the assumed velocity scaling . As ex plained in Ref. [5], the larger variation at frequencies above 5 kHz is caused mainly by residual engine noise, significant atmospheric attenuation of higher - frequency sound waves, and a lower signal - to - noise ratio (SNR) resulting from background/electroni c noise. The tight data collapse demonstrated in Fig. 5b unequivocally indicates that V scaling applies uniformly across the entire frequency range of interest. This observation differs from the assumptions used in the semi empirical model of Guo for the prediction of aircraft flap 5 6 side - edge noise [10 ], which employs a V dependency at low frequencies and V scaling at high frequencies. The strong V dependency observed in the ARM - III data also differs from the findings of Rossignol [11, 12 ], who applied 5.5 6.5 G uo’s velocity scaling to his model - scale experimental data but obtained a better collapse using V and V scaling at low and high frequencies, respectively.

An even bett er data collapse was observed for the landing configuration with flaps deflected 39  and landing gear deployed (Figs. 6a and 6b) , corroborating similar results presented in Ref. [5] for the 808 aircraft. As has been pointed out in Ref. [5], the 950 Hz tone present at 150 kts was caused by the main gear hollow front post . The intermitten cy of this tone depends strongly on the aircraft speed and wind direction encountered during a pass.

The main reason for attaining such remarkable data collapse for the landing configuration is the dominance of flap and MLG noise over secondary contributions to the farfield spectrum that may not scale with V (e.g., residual engine noise).

a) unscaled b) V scaling applied Fig. 5 . N oise produced by the 804 aircraft in approach configuration with flaps deflected at 20° and landing gear retracted . Spectra obtained from WingsNg integration of ARM - III test data for o verhead position.

a) unscaled b) V scaling applied Fig. 6 . N oise produced by the 804 aircraft in landing configuration with flaps deflected at 39° and landing gear deployed. Spe ctra obtained from WingsNg integration of ARM - III test data for overhead position.

2. Data Consistency th C onsistency of the acoustic measurements from repeat aircraft passes is discussed in this section . Sample 1/12 - oc t ave beamform maps from passes of the 804 aircraft with flaps deflected 39  and landing gear deployed are shown in Fig. 7 at three select frequencies represent ative of low - , medium - , and high - frequenc y noise content . This configu- ration was used as a baseline for determining the performa nce of the main gear NR technologies. The sound level s have been scaled to 150 kts based on V . The noise maps have also been normalized for each figure by subtracting the maximum SPL at each frequency from the corresponding contours, thus allowing a direc t comparison for the location, strength, and shape of the noise sources. The maps in the left and middle columns correspond to two consecutive aircraft passes (no. 4 and 5) executed during March 20, 2018 and those on the right column belong to pass no. 9 e xecuted on March 30, 2018. At the frequency of 375 Hz, all three maps indicate that the main gear cavity a nd flap inboard tips are the prominent sources. At 2,000 Hz, the major sources are the main and nose gear s a s well as the inboard and outboard flap ti ps. Notice that, at medium frequencies, the flap brackets are also important secondary sources. At the high frequency of 4,500 Hz, the flap inboard tips appear as the dominant source s, followed by the main and nose landing gear s as much weaker sources. Ove rall, the maps show that all three passes produce a consistent identification of the airframe noise sources with peak sound levels in the maps that fall within 1 - 2 dB of each other.

Although not shown here , careful inspection of the maps at most other frequencies indicate s a similar ly close agree- ment. I nspection of the maps reveal s minor differences in the shape and peak level of the noise source s on either side of the aircraft. Such subtle discrepancies ar e to be expected and are mostly attributed to the uncontrolled nature of the test environment during the aircraft flyover s .

a ) f = 375 Hz b) f = 375 Hz c) f = 375 Hz d ) f = 2 , 000 Hz e ) f = 2 , 000 Hz f) f = 2,000 Hz g ) f = 4 , 500 Hz h ) f = 4 , 500 Hz i) f = 4,500 Hz I) P ass no. 4, March 20, 2018 II) P ass no. 5, March 20, 2018 III) P ass no. 9, March 30, 2018 Fig. 7 . Beamform maps for the 804 aircraft in landing configuration with flaps deflected at 39° and landing gear deployed at overhead position.

3. Data Repeatability To assess the repea ta bility of the acquired acoustic data , we utilize the integrated PSD plots associated with the 808 aircraf t. Despite the lack of measured true air speed (TSA) and reliance on IAS (as called out by the pilots) for scaling purpose s , this aircraft was extensively tested during all three ARM deployment s . As such, the gathered meas- urements provide a wealth of data for examining pass - to - pass repeatability, consistency of the observed trends, and data quality that spans multiple years and test sites. In Fig. 8 , we present the integrated spectra based on the “ WingsNg ” region for the 808 configuration with flaps deflect ed 20° and landing gear retracted. The plotted dataset contains 3 passes from ARM - I ( 2016 ) , 9 passes from ARM - II ( 2017 ) , and 5 passes from ARM - III ( 2018 ) . The figure illustrates that all the passes produce d consistent spectral levels that approximately fall within a 3 dB band . To isolate the trends, averaged spectr a based on p averaging of all the passes from each year were generated and the results are plotted in Fig. 9. Also included in Fig. 9 is the averaged spectrum for all shown multi year passes. Except for a narrow frequency band between 700 Hz < f < 1,000 Hz, there is very good agreement among all averaged spectra up to 5 – 6 kHz , indi- cating that the measurements are highly consistent and repeatable. The underlying factors causing the discrepancies at high er frequencies are the same as those mentioned previously. T he differences occurring in the frequency band 700 Hz < f < 1,000 Hz are puzzling , as no changes were made to the aircraft n or the test procedure. Moreover, the averaged spectra from the 20 16 and 2018 passes are almost coincident, with the 2017 aver aged spectrum being no- ticeably higher. Therefore, differences in the spectral levels cannot be attributed to the change in test site n or to the microphone array being deployed on a concrete - paved surface . We will examine this issue further when results from the 39° flap deflection are presented. For ease in visualization of the results, only the averaged spectra are presented for the other 808 aircraft configurations.

Fig. 8 . Passes spanning multip le years for the 80 8 aircraft in approach configuration with flaps deflected 20° and landing gear retracted. Spectra from WingsNg integration for overhead position.

Fig. 9 . Averaged spectra for the 80 8 aircraft in approach configuration w ith flaps deflected 20° and landing gear retracted. Spectra from WingsNg integration for overhead position.

Multi year averaged spectra for the 808 aircraft with flaps deflected 20  and landing gear deployed are shown in Fig. 10. The spectra display very good agreement with each other and with the averaged levels obtained from all the aircraft passes. Apart from the expected differences at high frequencies, the most noticeable discre pancy occurs in the 6 00 Hz < f < 1,000 Hz range , where the averaged spectrum from 2017 possesses higher levels. The averaged spectra for the 808 aircraft in the landing configuration of 39 ° flap deflection with landing gear retracted and deployed are p rese nted in Figs. 11a and 11b, respectively. Remarkable agreement between the 2016, 2017, and 2018 results is observed. A gain, the m ost not able discrepancy occurs in the 6 00 Hz < f < 1,000 Hz range. Co mpared to the 20° flap deflection, the differences in spect ral levels at this higher flap deflection coincide with the appearance of a moderate th tonal hump centered around 785 Hz. Examination of sample 1/12 - octave beamform maps from 2017 and 2018 passes in the 750 Hz < f < 900 Hz frequency band indicates that the tonal noise is generated at the flap inboard edges. The only geometric feature capable of producing such a noise source is the 0.5” (1.25 cm) deep cavity at both inboard and outboard tips. The shallow cavity begins at the flap leading edge and extends ove r most of the flap chord . W ith in- creasing flap deflection, flow angularity at the flap side - edge causes a more normal, rather than parallel, impinge ment of incoming flow on th e cavity. As a result , the separated flow at the bottom edge of the cavity interacts with the top edge and generat es the tone. Although the presence of the cavity explains the appearance of the tone , it does not answer why it would consistently produce higher - amplitude noise during the 2017 passes than similar passes in 2018 or 2 016 .

Fig. 10 . Averaged spectra for the 80 8 aircraft in approach configuration with flaps deflected 20° and landing gear deployed. Spectra from WingsNg integration for overhead position.

a) L anding gear retracted b) L anding gear deployed Fig. 11 . Averaged spectra for the 80 8 aircraft for flaps deflected 39 ° with landing gear retracted and de- ployed . Spectra from WingsNg integration for overhead position.

The 804 a ircraft in its conventional Fowler flap configuration with out treatments on the gear or wheel cavity (baseline configuration) was only tested during the ARM - III test (2018) . Therefore, a similar year - to - year data repeat- ability analysis cannot be conducted for this testbed. However, there are two 804 configurations that can be used to assess yearly data consistency . In Fig. 12 a , integrated spectra based on the WingsNg region for passes of the 804 aircraft conducted in 2016 and 2018 are compared. The 2016 pas ses (dash - dot blue lines) used ACTE flaps set at 0° (cruise configuration) and landing gear deployed. The 2018 passes (solid red lines) were conducted with 804 Fowler flaps set at 0 ° deflection (cruise wing) and landing gear deployed. Notice that the two sets of data collapse nicely within a narrow amplitude band , indicating that the measurements were consisten t and repeatable. The spectra ob- tained from averaging the passes for each year and the passes for the two years are presented in Fig. 12b. G ood correspondence among the spectra is observed. The slightly higher jitter in the averaged levels for the 2018 data is attributed to the low number of passes available for averaging.

a) all acceptable passes b) averaged spectra Fig. 12 . Passes spanning multiple years for the 80 4 aircraft for flaps deflected 0° and landing gear deployed.

Spectra from WingsNg integration region for overhead position.

We now demonstrate that good consistency in the gathered data should be expected when NR technologies for the main gear are applied. S pectra from integration of data within the MLG region obtained during flyovers of the 804 aircraft with ACTE (2017) and Fowler (2018) flaps set at 0° , with fairings and cavity mesh installed on the main landing gear during both flights, are plotted in Fig. 13 . Data from 2017 (804 with ACTE) passes are represented by dash - dot blue lines; data from 2018 (804 with Fowler) passes are identified by solid red lin es . Highly consistent and repeatable trends for the two sets of passes are observed, producing averaged spectra that are very similar in shape and levels.

a) all acceptable passes b) averaged spectra Fig. 1 3 . Passes spanning multiple years for the 80 4 aircraft with 0° deflected flaps and fairings and cavity mesh applied to main landing gear . Spectra from MLG integration region for overhead position.

4. Impact of Local Meteorological Conditions T he impact of local meteorological conditions on the quality and repeatability of the measurements is assessed in this section . The results presented here complement earlier analys e s of ARM - I and ARM - II data regarding this im- portant issue [8]. For this discussion, we rel ied exclusively on ARM - III ( 2018 ) data for two critical configurations with the most repeat flights. MLG region integrated spectra for all acceptable passes of the 804 aircraft with flap s deflected 39° and fairings plus cavity chevrons and foam applied to the main landing gear are shown in Fig. 14 . The aircraft passes were performed on different days with good combinations of temperature and relative humidity. Low winds (1.9 m/s) were encountered on April 3, relatively high winds (4.5 m/s) on April 5, and moderate winds on April 30 (2 to 3.5 m/s) . Figure 14 shows a tight collapse of the spectra that fall within 2 dB of each other. Excellent d ay - to - day repeatability of the acquired acoustic measurements is demonstrated by the averaged spectra for each day , plotted in Fig. 15. Also included in this figure are the average of all acceptable passes collected during the three days and the average of five select passes from the same days. These five spectra were chosen because aircraft AOA, glide slope, fuel weight, and position above microphone array closely matched. For frequencies below 5 kHz, there are hardly any differences among the averaged spectra , indicating that good local meteor ological conditions alone could be es- senti al to obtaining good quality measurements. The fact that the averaged spectrum from the five select passes coincides with the other averages provides insight on the dependencies of airframe noise on aircraft param eters – during approach or landing, the farfield noise footprint is very weakly dependent on aircraft AOA, glide slope, and weight. The extremely weak dependency on AOA observed here corroborate s the mode l - scale test results of Ref. [13 ].

Fig. 14 . Integrated spectra for 804 aircraft with flaps deflected at 39 ° and fairings plus cavity chevrons and foam applied to main landing gear . Spectra from MLG integration region for overhead position.

Fig. 15 . Averaged spectra for the 80 4 aircraft with flaps deflected at 39 ° and fairings plus cavity chevrons and foam applied to mai n landing gear. Spectra from MLG integration region for overhead position.

See Ref. [5] for definition of pass acceptance.

The quality of acoustic data obtained on days with good local meteorological conditions was demonstra ted in Figs.

14 and 15. We will now explore the behavior of farfield noise when acoustic data were acquired under less than optimal meteorological conditions . I ntegrated spectra based on the WingsNg region are presented for the 804 aircraft in the landing configuration (flaps deflected 39° and landing gear deployed) in Fig. 16 . Measurements for this config- uration were obtained during five days. The temperature - relative humidity conditions were excellent for the three days in March. The winds were low to moderate, and more importantly, steady during these three days with March 30 being one of the best test days of the ARM - III de ployment. The temperature - relative humidity combination s for the two test days in April were marginal , combined with fluctuati ng winds (relatively high atmospheric turbulence) even though overall wind magnitude remained moderate (1 minute averages less tha n 2.5 m/s) and within the acceptable range.

The passes conducted on March 20, 29, and 30 produced very similar spectral content , as seen in Fig. 16. S pectra from the April 21 passes (dark blue dash - dot lines) are consistent but show noticeably lower amplit udes than the spectra collected during the three days in March. The differences in amplitude incre ase with frequenc y because of higher atmospheric attenuation a nd decorrelation of the sound waves caused by wind gusts. As clearly shown in Fig. 16, the wors t results were obtained during April 24 and show even larger differences with the spectra from data acquired during March. The corresponding daily averages are presented in Fig. 17. T he low number of passes available f or March 20 and 29 prompted considerat ion of the March data under a single grouping. T here are significant differences in the daily averaged spectral levels , with the April 24 averaged spectrum showing 2 – 4 dB lower levels at medium and high frequencies. Also included in Fig. 17 is the averaged spectrum composed of passes from all days. Although this spectrum lies closer to the averaged spectrum for the March passes, differences on the order of 1 to 1.5 dB are still present . Although these differences may seem reasonable, they have a devastating effect when trying to assess the performance of landing gear noise abatement technologies , where the maximum reductions that can be obtained are on t he order of 3 – 4 dB at most frequencies. As the results in Fig. 17 clearly demonstrate, t he availability of data from multiple flight days is essential for critical aircraft configurations of interest , in particular for those configurations that will serve as baselines. Since there is no viable correction for the severely adverse effect s that gust y winds have on medium and high frequency noise , testing should be performed during days when both steady and fluctuating winds are low even if the temperature - relative humidity combinations are marginal. Based on the results shown in Fig. 17, only the averaged spectrum obtained from the March passes was used as the baseline to determine the noise reduction performance of various gear technologies discussed in section IV - B .

Fig. 16 . Integrated noise spectra for 804 aircraft with flaps deflected 39° and landing gear deployed.

Spectra from WingsNg integration region for overhead position.

Fig. 17 . Averaged spectra for the 80 4 aircraft with flaps deflected 39° and landing gear deployed.

Spectra from WingsNg integration region for overhead position.

5. Effect of Baseline Aircraft on Farfield Noise T his section addresses whether the 804 and 808 aircraft in their conventional Fowler flap (baseline) configuration produce similar farfield noise signature s . Due to the unavailability of baseline 804 data at the time, the noise reduction performance of the ACTE flap with and without MLG noise abatement technologies w as evaluated in Ref. [5] using data acquired during the ARM - I and ARM - II flights wit h the 808 aircraft as the baseline. I ntegrated spectra from the WingsNg region for three select passes for each aircraft from the ARM - III ( 2018 ) campaign are plotted in Fig. 18a .

Since the 808 aircraft is not instrumented , fu el weight was used to match 804 and 808 passes. As seen clearly in Fig.

18, correspondence between the farfield noise signature s of the two aircraft is very good for frequencies up to 5 kHz.

Beyond 5 kHz, the 804 aircraft is nois ier . As demonstrated in Ref. [5], the extra noise is produced by the 804 engines , which are older and use a different generator cooling system than the 808 engines. The observed agreement become s more remarkable when consider ing that only IAS values (as called out by the pilots) were used to scale the 80 8 spectral levels. A veraged spectra for the two aircraft, obtained from all acceptable passes executed during ARM - III , are shown in Fig. 18b. Also plotted in this figure are the averaged spectra for the select passes of Fig. 18 a. For frequencies up to 5 kH z, the averaged spectra fall within a very tight band. Beyond 5 kHz, 804 engine noise causes higher sp ectral levels. However, as described in Refs. [5, 8], the quality of the measured signal deteriorate s very rapidly for frequen- cies above 5 kHz. The cause s are atmospheric attenuation, poorer SNR due to background and electronic noise, and residual engine noise. Thus , for the purpose s of evaluating the performance of the airframe technologies tested, one can assume that the noise signature s of both aircraft in their baseline configuration are equivalent. As a result, there is no need to re evaluat e the airframe technologies presented in Ref. [5] using baseline 804 data collected in 2018.

a) select passes b) averaged spectra Fig. 18 . Individual and averaged spectra for 80 4 and 808 aircraft with 39 ° deflected flaps , l anding gear de- ployed . Spectra from WingsNg integration region for overhead position.

B . Performance of Landing Gear N oise R eduction Technologies Since a vast amount of data was gathered during the ARM - III test, we include here only a few representative be am form maps that demonstrate how the applied technologies affe cted the intended noise sources. Because of the prominence of the inboard tip source at high flap deflection angle s , the 20° deflection was used to highlight the true noise reduction potential of the gear t reatments . Sample CLEAN beamform maps for the 804 aircraft with its Fowler flaps deflected 20º without (baseline) and with MLG NR technologies are displayed in Fig. 19 at low, medium, and high frequencies for the overhead position. For the treated gear configuration, the installed NR technologies comp rised the MLG fairings plus cavity chevrons and sound absorbing foam. This combination represents the quietest configu- ration tested. For frequencies below 600 Hz, the MLG cavity is the dominant source with the main gear acting as a secondary source [5]. Co mparison of the maps at 450 Hz (shown in Figs. 19 a and 19 b) clearly demonstrates that the gear fairings and cavity treatment significantly reduce the noise generated by the MLG, lowering peak sound levels by approximately 6 to 7 dB. At mid - frequencies (Fig s. 19 c and 19 d), the noise benefit is slightly less, approaching 4 to 5 dB, since cavity noise is no longer a major contributor and reductions are produced mostly by the gear fairings.

As shown in Figs. 19 e and 19 f, the fairings also perform exceptionally well at higher frequencies. The corresponding impact on integrated noise levels is presented later in this section.

Similar maps for the same two aircraft configurations at a 39° flap deflection are presented in Fig. 2 0 . F or this higher deflection angle, at low and medium frequencies , the inboard flap tips are prominent sources comparable in strength to the cavity and MLG sources (Figs. 2 0 a, 2 0 c, and 2 0 e). A s was the case for the lower flap deflection , t he maps vividly illustrate that at 39° the cavity tre atments and gear fairings successfully either eliminate or greatly di- minish the strength of the MLG noise sources (Figs. 2 0 b, 2 0 d, and 2 0 f) with out affecting the inboard tip sources. F or frequencies greater than about 3 kHz, the flap outboard tips become t he dominant airframe source s and mask the noise reduction potential of the gear technologies.

a ) f = 450 Hz b) f = 450 Hz c) f = 2,000 Hz d) f =2,000 Hz Fig. 19 . N oise produced by 804 with Fowler flaps deflected 20  , without (left column) and with (right column) NR technologies installed – MLG fairings plus cavity chevrons and foam. Overhead position.

e) f = 4,500 Hz f) f = 4,500 Hz Fig. 19 . Concluded .

a ) f = 450 Hz b) f = 450 Hz c) f = 2,000 Hz d) f =2,000 Hz Fig. 20 . N oise produced by 804 with Fowler flaps deflected 39  , without (left column) and with (right column) NR technologies installed – MLG fairings plus cavity chevrons and foam. Overhead position.

e) f = 2,800 Hz f) f = 2,800 Hz Fig . 20 . Concluded .

To accurately extract the noise reduction attained with the MLG technologies, the “MLG” region was used for integrating the CLEAN beamform maps. The averaged (for multiple passes) , integrated , overhead farfield spectra corresponding to the untreated (baseline) and several configurations with treated MLG are p resented in Fig. 21 for a flap deflection of 20  . T he plotted spectra demonstrate and quantify the acoustic benefits achieved with the installed ML G technologies – significant noise reduction that is maintained over the entire frequency range. Notice that signif- icant jitter is present in the spectra at frequencies above approximately 5 kHz. The highest noise reduction was achieved with the combinatio n of the gear fairings plus cavity chevrons and foam. Due to the much lower magnitude of noise generated at the inboard tips for this flap deflection, the integrated spectra using the MLG region captures a substantial portion of the noise reduction benefit s attained with the applied technologies. This combination of the technologies produces close to 5 dB reduction at frequencies below 400 Hz and 3 - 4 dB for higher frequencies. The reduction levels reported here are slightly less than the values reported in Ref. [5] for an ACTE - equipped 804 aircraft when the flap noise was virtually eliminated. As seen i n Fig. 2 1 , the combination of the gear fairings plus cavity mesh provides less noise benefits for frequencies below 2 kHz. The removal of chevrons from the ca vity ( i.e. , fairings plus foam installed ) results in less noise reduction only in the 200 Hz < f < 400 Hz range . Otherwise, the integrated spec- trum is nearly the same as that produced by the fairings plus chevrons and foam combination. The configuration in volving only gear fairings lacks the full reduction benefits at lower frequencies. Nevertheless, the fairings still de- liver in excess of 1 dB reduction at frequencies below 300 Hz. At higher frequencies, the fairings produce close to 3 dB reduction that is about 1 dB less than the fairings plus foam combination. As alluded to in Ref. [5], this difference suggest s that the beneficial effects of placing sound absorbing foam on the back wall of the gear cavity is not limited to the very low frequency end of the spectrum. Surprisingly, good noise reduction was delivered by the combination of cavity chevrons and foam. Without the gear fairings, we anticipated a rapidly diminishing performance for this combination at frequencies above 900 Hz to 1 kHz. However, as the averaged spectrum in Fig. 2 1 indicates, substantial reduction was achieved over most of the frequency range of interest. T he flyover passes for this configuration were executed during a sing le test day when local meteorological conditions were marginal with moderate winds. Therefore, one must assume that the integrated levels at m edium to high frequencies come with a higher degree of uncertainty in the collected data.

Airframe noise is known to be dominant in the forward directivity angles. T o demonstrate that the performance of the gear technologies is maintained over a large segment of the forward quadrant, integrated averaged spectra at di- rectivity angles of ~ 67° and ~ 45° are presented in Fig. 2 2 a and 2 2 b , respectively . The data show similar , or even larger , reduction s at these additional directivity angles. G ood performance over a wide range of directivity angles typically translates into a substantial reduction in e ffective perceived noise levels (as demonstrated in Ref. [14]), a key metric for acceptance of new technologies by the industry. The averaged spectra for the same combinations of the gear NR technologies for the 804 aircraft with Fowler flaps set at 39° ar e presented in Fig. 2 3 . As expected, noticeably lower reductions are observed for all tested technology combinations due to the fact that integration of the MLG region now contains contribution s from inboard flap tip sources that mask the true performance of the gear and cavity tech- nologies. Despite this contamination, a reduction on the order of 2 to 3 dB was indicated by the measured data. As was the case for the flap s deflected 20°, the noise benefits become somewhat larger at forward directivity angles with the combination of gear fairings plus cavity chevrons and foam still providing the best performance.

Fig. 2 1 . Averaged s pectra from 804 passes for Fowler flap 20 °, gear deployed without/ with various combina- tions of MLG treatments installed . Spectra from MLG integrat ion region for overhead (90° directivity angle).

a) Directivity ≈ 67° b) Directivity ≈ 45° Fig. 2 2 . Averaged s pectra from 804 passes for Fowler flap 20 °, gear deployed without/ with various combina- tions of MLG treatments installed . Spectra fr om MLG integration region at forward directivity angles.

a) O verhead ( directivity = 90°) Fig. 23 . Averaged s pectra from 804 passes for Fowler flap 39°, gear deployed without/with various combina- tions of MLG treatments installed . Spectra from MLG integration region at overhead and forward d irectivity angles .

b) Directivity ≈ 67° c) Directivity ≈ 45° Fig. 23 . Concluded .

V. Concluding Remarks Detailed aeroacoustic analys e s of phased microphone array data obtained during flyover passes conducted for the 2018 ARM - III flight test were presented in this paper . The goal of the test was to evaluate the acoustic performance of main landing gear and cavity noise abatement technologies installed on a Gulfstream G - III aircraft flying in its conventional ( Fowler ) flap configuration. Microphone array measurements from the 2018 test , combined with acoustic data collected during the 2016 (ARM - I) and 2017 ( ARM - II) tests were used to study the quality and consistency of the acquired measurements. Through careful examination of data from numerous aircraft passes obtained on multiple flight days and spanning three years, good collapse of the integrated farfield noise spectra was demonstrated , estab- lishing excellent year - to - year repeatability of the measured trends. The good collapse highlighted the fact that airframe noise is only mildly dependent on the aircraft angle of attack , glide slope and fuel weight, and atmospheric corrections for temperature/humidity are effective even for marginal conditions. However, high winds or elevated wind fluctua- tions produced uncorrectable a nd unacceptable data variability. The analyses also indicated that , for any of the tested configurations, the spectrum obtained from p averaging all the acceptable repeat passes is representative of t he overall spectral content levels , but this requires s ufficient passes to identify and exclude outliers.

The noise benefits of the tested landing gear and gear cavity technologies were determined from comparisons with baseline G - III data obtained during the ARM - III test . Data from flap deflection angles of 20° and 39° at multiple directivity angles were analyzed . For the 20° flap deflection, the combination of gear fairings plus cavity chevrons and foam produce d the quietest configuration , providing overhead noise reduction s of 5 dB at frequencies below 400 Hz and 3 - 4 dB at frequencies above 400 Hz. The removal of chevrons from the cavity (i.e. , fairings plus foam installed) result ed in decreased noise reduction only in the 200 Hz < f < 400 Hz range. Surprisingly, good noise reduction was delivered by the com bination of cavity chevrons and foam , despite marginal weather conditions that possibly affect ed the uncertainty at high frequencies. For the 39° flap deflection, lower reductions ( on the order of 2 to 3dB) were observed for all tested technology combinati ons – integration of the MLG region for this configuration contain ed contributions from inboard flap tip sources that mask ed the true performance of the gear and cavity technologies. T he noise benefits were slightly better at forward directivity angles fo r both flap deflections, with the combination of gear fairings plus cavity chevrons and foam providing the best performance. In general, t he results show ed that the tested landing gear fairings and gear cavity treatments are highly effective at mitigating the noise produced by undercarriage systems, even in the presence of conventional (Fowler) flaps.

Acknowledgments This work was supported by the Flight Dem onstrations and Capabilities project under the Integrated Aviation Systems Program of the NASA Aeronautics Research Mission Directorate. The ARM flight tests would not have been possible without the dedicated effort of a large group of people , especially t he NASA Armstrong Flight Research Center personnel. For the ARM - III test i n particular, we would like to express our sincere appreciation to Claudia Herrera (SCRAT Chief Engineer), Erin Waggoner, Angel Guilloty, and the lead test pilots Timothy Williams an d Troy Asher.

References [1] Adib, M., Catalano, F., Hileman , J., Huff , D., Ito , T., Joselzon , A., Khaletskiy , Y., Michel , U., Mongeau , L., and Tester B . J. , “Novel Aircraft - Noise Technology Review and Medium - and Long - Term Noise Reduction Goals,” International Civil Avia- tion Organization, Doc. 10017, 2014.

[ 2 ] https://www.faa.gov/data_research/aviation/aerospace_forecasts/media/FY2016 - 36_FAA_Aerospace_Forecast.pdf , accessed October 24, 2016 .

[ 3 ] Dobrzynski, W., “Almost 40 Years of Airframe Noise Research: What Did We Achieve,” J. Aircraft , Vol. 47, No. 2, March - April 2010, pp. 353 – 367.

[ 4 ] Khorrami, M. R., Humphreys, W. M. Jr., Lockard, D. P., and Ravetta, P. A., “Aeroacoustic Evalua t ion of Flap and Landing Gear Reduction Concepts,” AIAA Paper 2014 - 2478, June 2014.

[ 5 ] Khorrami, M. R., Locka rd, D. P., Humphreys, W. M. Jr., and Ravetta, P. A. , “ Flight - Test Evaluation of Airframe Noise Mitigation Technologies ,” AIAA paper 2018 - 2972 , June 2018.

[6] Baumann, E. and Waggoner, E., “ Flight and Ground Operations in Support of Airframe Noise Reduction Tests, ” 2018 AIAA/CEAS Aeroacoustics Conference, AIAA paper 2018 - 2970 , June 2018.

[ 7 ] Humphreys, W. M. Jr., Lockard, D. P., Khorrami, M. R., Culliton, W. G., McSwain, R. G., Ravetta, P. A., and Johns, Z., “Development and Calibration of a Field - Deployable Microphone Phased Array for Propulsion and Airframe Noise Flyover Measurements,” AIAA Paper 2016 - 2898, May - June 2016.

[8] Lockard, D. P. and Bestul, K. A., “The Impact of Local Meteorological Conditions on Airframe Noise Flight Test Data,” AIAA Paper 2018 - 2971, June 2018 .

[ 9 ] AVEC Time Domain Beamforming Software, Ver. 2. 85 , AVEC, Inc., Blacksburg, VA, URL: http://www.avec - engineer- ing.com/products.html , cited Ma rch 21 , 201 9 .

[10] Guo, Y. , “ Aircraft Flap Side Edge Noise Modeling and Prediction ,” AIAA Paper 2011 - 2711, June 2011 .

[11] Rossignol, K. S. , “ Development of an E mpirical P rediction M odel for F lap S ide - E dge N oise ,” AIAA Paper 2010 - 3836, June 2010.

[12] Rossignol, K. S. , “ Flow Field Measurements to Characterize Flap Side - Edge Noise Generation ,” AIAA Paper 2013 - 2061, May 2013 .

[ 13 ] Khorrami, M. R., Humphreys, W. M. Jr., and Lockard, D. P., “ An Assessment of Flap and Main Landing Gear Noise Abate- ment Concepts ,” AIAA paper 2015 - 2987, June 2015.

[ 14 ] Ravetta, P. A. , Wisda, D. M., Khorrami, M. R., and Van de Ven, T. , “ Assessment of Airframe Noise Reduction Technologies th based on EPNL from Flight Tests, ” Paper to be presented at the 25 AIAA/CEAS Aeroacoustics Conference in Del ft, The Netherlands, May 2018.