Enhanced Modeling of Aircraft Taxiway Noise, Volume 1: Scoping (2009)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

45 CHAPTER 4. OPERATIONAL MODELING 4.1. Single Event Analysis 4.1.1. Source Modeling (simulation comparisons using various fidelity spheres) Analyses were performed using Wyle’s acoustic simulation tools with 3D spectral sources for the various taxi throttle settings. The creation of these noise spheres was described in Chapter 3. One example, a comparison of the noise from constant speed taxi operations at different thrust settings, is shown here. An overlay of the various noise directivity contours is given in Figures 37 and 38. Shown are contours for six different thrust settings, from 480 lbs to 4800 lbs for a single taxi operation heading east at a speed of 15 knots over a distance of 2000 ft. The contours displayed are SEL values and indicate based on the assumed source sensitivity in the table above, the single event difference for this one particular element of a taxi operation. A contour comparison between the 480 and 4800 lbs setting, obtained by subtracting the SEL grids is provided in Figure 39. Similar comparisons have been made for changes in speed, spectra and directivity (lateral and longitudinal). These elements serve as the building blocks of the sensitivity study from which increasingly complex combinations can be examined to determine what parameters are needed to be modeled for taxi noise assessments. 35 35 35 45 45 50 55 60 80 35 35 35 40 40 50 45 55 55 65 70 35 35 35 40 4045 45 50 50 55 60 65 7580 85 90 35 35 35 40 40 45 45 50 50 55 60 65 75 80 85 95100 35 35 35 40 40 40 45 45 55 55 60 70 0 35 35 35 40 40 4045 45 50 50 55 60 75 80 X-Ft Y -F t -10000 -5000 0 5000 10000 -10000 -5000 0 5000 10000 SEL-A 100 95 90 85 80 75 70 65 60 55 50 45 40 35 B737-700 Taxi Operation Heading East Full Directivity T=480 (2%),1285 (5.25%),1680 (7%), 2400 (10%),3600 (15%),4800 (20%) C:\MyDocuments\Project_Docs\TRB\TaxiwayNoise\Data\AAM\SensitivityAAMruns\ConstSpeed\TaxiB7377T0480.PLT C:\MyDocuments\Project_Docs\TRB\TaxiwayNoise\Data\AAM\SensitivityAAMruns\ConstSpeed\TaxiB7377T1285.PLT C:\MyDocuments\Project_Docs\TRB\TaxiwayNoise\Data\AAM\SensitivityAAMruns\ConstSpeed\TaxiB7377T1680.PLT C:\MyDocuments\Project_Docs\TRB\TaxiwayNoise\Data\AAM\SensitivityAAMruns\ConstSpeed\TaxiB7377T2400.PLT C:\MyDocuments\Project_Docs\TRB\TaxiwayNoise\Data\AAM\SensitivityAAMruns\ConstSpeed\TaxiB7377T3600.PLT C:\MyDocuments\Project_Docs\TRB\TaxiwayNoise\Data\AAM\SensitivityAAMruns\ConstSpeed\TaxiB7377T4800.PLT Figure 37. B737-700 with full directivity, thrust: 480 - 4800 lbs.

46 45 45 50 50 50 50 55 50 55 60 65 60 70 75 8080 90 95 100 100 40 40 40 40 45 45 45 50 55 50 60 55 65 65 70 75 70 75 8585 100 40 40 40 40 45 45 45 55 50 50 55 55 65 60 70 75 75 8585 95 40 40 40 4045 45 45 45 50 55 50 60 60 60 60 65 65 70 75 80 859095 100 40 40 40 45 45 45 45 50 50 50 55 55 60 60 65 65 70 75 80 90 40 40 45 45 45 50 50 50 55 55 60 60 65 70 75 0 90 X-Ft Y -F t -5000 0 5000 -4000 -2000 0 2000 4000 SEL-A 100 95 90 85 80 75 70 65 60 55 50 45 40 35 B737-700 Taxi Operation Heading East Full Directivity T=480 (2%),1285 (5.25%),1680 (7%), 2400 (10%),3600 (15%),4800 (20%) C:\MyDocuments\Project_Docs\TRB\TaxiwayNoise\Data\AAM\SensitivityAAMruns\ConstSpeed\TaxiB7377T0480.PLT C:\MyDocuments\Project_Docs\TRB\TaxiwayNoise\Data\AAM\SensitivityAAMruns\ConstSpeed\TaxiB7377T1285.PLT C:\MyDocuments\Project_Docs\TRB\TaxiwayNoise\Data\AAM\SensitivityAAMruns\ConstSpeed\TaxiB7377T1680.PLT C:\MyDocuments\Project_Docs\TRB\TaxiwayNoise\Data\AAM\SensitivityAAMruns\ConstSpeed\TaxiB7377T2400.PLT C:\MyDocuments\Project_Docs\TRB\TaxiwayNoise\Data\AAM\SensitivityAAMruns\ConstSpeed\TaxiB7377T3600.PLT C:\MyDocuments\Project_Docs\TRB\TaxiwayNoise\Data\AAM\SensitivityAAMruns\ConstSpeed\TaxiB7377T4800.PLT Figure 38. B737-700 with full directivity, thrust: 480 - 4800 lbs. – zoom.

47 0.5 1 1 1.5 1.5 1.5 1.5 2 2 2 2.5 2.5 2.5 3 X-Ft Y -F t -10000 -5000 0 5000 10000 -10000 -5000 0 5000 10000 SEL-A 3 2.5 2 1.5 1 0.5 B737-700 Taxi Operation Heading East Full Directivity Spheres Difference Plot T=4800 (20%) - 480 (2%) C:\MyDocuments\Project_Docs\TRB\TaxiwayNoise\Data\AAM\SensitivityAAMruns\ConstSpeed\TaxiB7377T0480.PLT C:\MyDocuments\Project_Docs\TRB\TaxiwayNoise\Data\AAM\SensitivityAAMruns\ConstSpeed\TaxiB7377T4800.PLT Figure 39. B737-700 difference plot: thrust: 4800 (20%) – 480 (2%) lbs. An additional comparison provided here illustrates the need to obtain realistic throttle operation and corresponding noise data. This example illustrates the importance of getting the timing and duration of the events set properly. The total operation duration is 80 seconds. Of that only 3 seconds are spent at the bumped up throttle setting. The operation is a “Go-Stop- Bump-Go” 80 second operation, 20 second hold, 3 second throttle bump. The first operation moves at 15 knots using 1285 lbs thrust, bumps up to 3600 (15%) lbs. The second bumps up to 7200 (30%) lbs. The simulation trajectory is given in Figure 40. Figure 40. Simulation trajectory Thrust Cumulative T X Y Z Speed Yaw Attack Roll Power Angle Distance (sec) (feet) (feet) (ft AGL) (knots) (degree) (degree) (degree) (degree) (feet) 0.000 -730.0 0.0 5.5 15.0 0.0 0. 0. 1285. 0. 0. 27.255 -40.0 0.0 5.5 15.0 0.0 0. 0. 1285. 0. 690. 30.324 -1.0 0.0 5.5 0.1 0.0 0. 0. 1285. 0. 729. 50.307 1.0 0.0 5.5 0.1 0.0 0. 0. 3600. 0. 731. 53.375 40.0 0.0 5.5 15.0 0.0 0. 0. 1285. 0. 770. 80.630 730.0 0.0 5.5 15.0 0.0 0. 0. 1285. 0. 1460.

48 One can see from the SEL contour plot (Figure 41) and the difference (Figure 42), not a lot of difference from the 3 second throttle bump is discernable in the contour. This suggests that breakaway thrust applied over a short duration may not have a significant impact. However recall that this suggestion is based on the assumption that the difference in source noise characteristics is 0.70 dB / klb Thrust and the breakaway thrust value was 3600 lbs and was applied for 3 second duration. Differences over the computation range were generally of the order of 0.15 dBA within 5000 ft of the operation. The discretization and precision of the contour levels on the receiver mesh is responsible for the jagged contours in the difference plot (Figure 42). 35 35 35 40 40 45 45 50 55 55 60 65 70 95 35 35 35 40 40 40 45 45 50 60 65 85 X-Ft Y -F t -10000 -5000 0 5000 10000 -10000 -5000 0 5000 10000 SEL-A 100 95 90 85 80 75 70 65 60 55 50 45 40 35 B737-700 Taxi Operation Heading East Full Directivity: Go-Stop-Bump-Go C:\MyDocuments\Project_Docs\TRB\TaxiwayNoise\Data\AAM\SensitivityAAMruns\Go-Stop-Bump-Go\BumpB7377T1285-7200.PLT C:\MyDocuments\Project_Docs\TRB\TaxiwayNoise\Data\AAM\SensitivityAAMruns\Go-Stop-Bump-Go\BumpB7377T1285-3600.PLT Figure 41. Overlay of both go-stop-bump-go operations.

49 0.04 0. 04 0. 06 0.0 6 0.0 6 0.06 0. 08 0. 08 0.08 0.08 0.08 0.1 0.1 0. 1 0.1 0.1 0. 12 0. 12 0.12 0.1 2 0.14 0.1 4 0.14 0.14 0.14 0.1 6 0.16 0.16 0.1 6 0. 18 0. 18 0.1 8 0.180.1 8 0.18 0.1 80 .18 0.20.22 0 .2 4 0.26 0. 04 0.04 0. 04 0.040.06 0.06 0. 08 0.08 0.08 0.0 8 0.1 0. 1 0.1 0.12 0.12 0.12 0.12 0.14 0. 14 0. 14 0.1 4 0. 14 0.1 6 0.1 6 0.16 0. 16 0.18 0.18 0.1 8 0.1 8 0.18 0.18 0.18 X-Ft Y -F t -10000 -5000 0 5000 10000 -10000 -5000 0 5000 10000 difference 0.28 0.26 0.24 0.22 0.2 0.18 0.16 0.14 0.12 0.1 0.08 0.06 0.04 0.02 B737-700 Taxi Operation Heading East Full Directivity: Go-Stop-Bump-Go C:\MyDocuments\Project_Docs\TRB\TaxiwayNoise\Data\AAM\SensitivityAAMruns\Go-Stop-Bump-Go\BumpB7377T1285-7200.PLT C:\MyDocuments\Project_Docs\TRB\TaxiwayNoise\Data\AAM\SensitivityAAMruns\Go-Stop-Bump-Go\BumpB7377T1285-3600.PLT Figure 42. Difference: 7200 – 3600 bump thrust levels for go-stop-bump-go operations. Analysis of the FDR dataset indicated that for some aircraft type, taxiing operations are performed with one or more engines shut down as much as 11 % of the time for moving taxi operations (Chapter 2, Table 2). While the 3D simulation model could analyze such asymmetric source directivity and incorporate such effects into the noise predictions, future research is needed to determine the source characteristics. One would need to either obtain empirical data simultaneously from both sides of the aircraft operating under asymmetric taxi engine conditions or exercise a shielding / diffraction model to account for the aircraft wing and fuselage shielding effects. An asymmetric source analysis was not performed during this study. In summary, detailed simulation techniques were conducted with AAM using 3D spectral noise sources to assess the sensitivity of various source features on noise contours. AAM treats the lateral and longitudinal directivity as a source modeling input, not as propagation corrections as is the case within INM. Modeled changes in the source noise characteristics tended to become diluted for propagation distances over soft ground approaching 1 mile, (Fig 4-3 and Fig 4-6) supporting again the inclusion in the short term of a nominal taxi operating state, while leaving a higher fidelity source model as an item for longer term development.

50 4.2. Multiple Operations In the ideal scenario, detailed ground operations will be modeled as individual tracks where the position and state of the aircraft is known as a function of time. Using INM’s point to point tracks and user defined profiles to represent the movement of the aircraft as low altitude overflights in conjunction with stationary run up operations, represents the highest fidelity geometric modeling possible in INM. This section will begin with a high fidelity analysis of a realistic taxi scenario for a given day. Intermediate analyses spanning from high fidelity modeling to lumped time-in-mode assessments and the impacts of the modeling scenarios on the predicted taxi noise contours will be presented. This analysis leverages a concurrent taxi assessment being performed for a major United States international airport. The acoustic predictions made with INM7 will be utilized as the framework for this sensitivity study assessing the geometric coverage of taxi aircraft movements. The standard INM7 code including the default NPD curves will be used. Since the intent of this section is to assess the applicability of different spatial modeling techniques, specific variations were not made to the acoustic source data within INM. As long as a consistent set of noise parameters are used across the entire sensitivity study, the impact on the output contours of the input modeling will depend purely on the selected input geometric fidelity. The full airport analysis presented here utilizes the following thrust values: Jets: 10% of Maximum Static Thrust; Props: 50% of maximum in appropriate units; Piper Cub: 1000 rpm. 4.2.1. Operational Modeling using INM7 INM7 fully supports the capability to include complex combinations of stationary and moving aircraft operations which can be applied to the prediction of taxi noise. Within INM7, a user can utilize multiple profile segments to model changes in aircraft speed and thrust such as when stopping and subsequently accelerating to cross over runways or intersecting taxi ways or to model aircraft holding queues, also holding queues and locations where aircraft accelerate to cross runways utilizing breakaway thrust can be modeled, provided of course one has the input taxi operational information available. Given what appears to be a wide choice of modeling options, we pose the following question: “How much fidelity of motion must be modeled in order to properly predict the noise from aircraft taxi operations?” This question will be explored by first examining a realistic dataset containing a full set of detailed taxi operations. Simplifications to the modeling of the location and orientation of the operations will be made, by assigning operations to a reduced number of taxi tracks, by combining run up operations and lumping them using varied orientations and locations. From this, insight can be obtained from the acoustic impact changes due to varying the fidelity of the taxi operations modeling. The INM User's Guide (1) specifies a method for modeling taxi noise in INM, specifying that "a taxi path can be approximated by an over-flight track and a fixed-point over-flight profile." The specific INM analysis techniques employed in this Section are those based on the Federal Aviation Administration (FAA) accepted Noise Analysis Protocol Version 6 (11), for the Boston Logan Airport Noise Study (BLANS). The following steps which apply to each individual taxi operation:

51 • Step 1. Create an over-flight track consisting of the necessary number of segments to model straight segments, runway crossing points, and turns. • Step 2. Create a corresponding fixed-point profile. Set the following parameters for each segment of the taxi profile: - Profile distance (the measured taxi track segment length). - Set altitude equal to engine height. - Set speed equal to average taxi speed along the segment. - Set thrust setting equal to the average taxi thrust along the segment. - Set operation mode to approach. • Step 3. Create a taxi operation that combines the taxi track (step 1) with the taxi profile (step 2) and assign the proper numbers of daytime and nighttime operations. • Step 4. Taxi track points will be added to represent runway intersections (where aircraft slow or wait to cross the runway). Low average speed and high average thrust will be used at track segments at these points represent acceleration across the runway. • Step 5. For queue times longer than five minutes, a low-power run up operation will be defined in INM at the appropriate point and orientation on the taxiway. 4.2.2. Geometric Modeling Fidelity A high fidelity analysis of a US International Airport was conducted using INM based on modeling specific gate to runway taxi tracks (Figure 43) as predicted by the Total Airspace and Airport Modeler (TAAM®) (16). This analysis considered 1205 individual taxi operations, spanning a fleet mix of 71 different aircraft for a typical busy day and included the impact of aircraft delay and queuing. Standard day and acoustically soft ground was considered without any acoustic considerations due to nearby bodies of water, terrain, buildings or other forms of shielding. In this segment of the report, taxi noise only contours are displayed down to the 55 dB DNL level in order to better illustrate the effect of taxi noise at a distance of several miles. The FAA policy is only to publish airport noise contours down to a level of 60 dB DNL.

52 Figure 43. Complex taxi configuration with high-fidelity taxi movements. Four high fidelity analyses, itemized below as #1 through #4, were conducted in INM. For these INM runs, detailed point to point tracks with user defined profiles were modeled and operations across the fleet mix were assigned. 1. High fidelity analysis – point to point tracks, user defined profiles, arrival NPDs for taxi, 5 min duration threshold run up assessment. 2. High fidelity analysis – point to point tracks, user defined profiles, departure NPDs for taxi, 5 min duration threshold run up assessment. 3. High fidelity analysis – point to point tracks, user defined profiles, departure NPDs for taxi, 1 min duration threshold run up assessment. 4. High fidelity analysis – point to point tracks, user defined profiles, departure NPDs for taxi, no run ups included. 5. High fidelity analysis, 1 min duration threshold run up operations only.

53 As shown in Section 4.3.4, run up operation data is consistent with departure NPDs in INM, therefore only departure NPDs will be utilized in this analysis. It is the differential in the DNL contours that is of importance, not the absolute value of the contour level. For reference, a comparison of the moving portion of the operations was made with arrival NPDS (#1) and with departure NPDs (#2) and are displayed in Figure 44. It is no surprise that selecting the departure-NPD operation type (solid line) are bigger than the arrival-NPD operation type (dashed line) due to the fact that the departure-NPD SEL curves are greater than the arrival-NPD SEL curves for the same thrust. Figure 44. Arrival and departure NPD for moving taxi operations (#1-dashed vs. #2-solid).

54 Within this analysis any time an aircraft holds at a fixed position and exceeds a specified duration threshold, it can be modeled as a run up operation within INM as a static operation with the appropriate duration, location and heading. A comparison of a holding duration threshold of 1 and 5 minutes were considered (Figure 45). For this analysis departure-NPDs were used (solid lines: 1 minute threshold, dashed lines: 5 minute threshold). A total of 725 run up operations met the 1 minute holding threshold, whereas only 31 operations met the 5 minute holding threshold. The difference in the energy from these events can be considerable, as evidenced by the approximate 4 dB displacement of the south eastern portion of the 55 dB DNL contour. The importance of a holding threshold will be most important to the airport modeler in regions where queuing is most likely to occur. Figure 45. Holding duration threshold of 5 and 1 minutes, (#2-dashed and #3-solid).

55 4.2.2.1. Moving vs. Holding Modeling In order to visualize the relative importance of moving versus stationary operations, a buildup of the contours from the various modes of operations was created. Figure 46 shows a comparison of only the flight operations modeled as a departure-NPD operation type (dashed) and the run up operations using a 1 minute duration threshold (red) with the overall contour representing the sum of the moving and stationary operations (solid). It is apparent that the run up operations drive the bulge in the contour along the southern and northern edges of the DNL contour. Figure 46. Contour build up #3-all (solid), #4-moving (dashed), #5-run ups (red). 4.2.2.2. Lumped Time in Mode Analysis One simplification of taxi modeling is to consider all the operations combined as a lumped time in mode. This style of modeling is currently employed in emissions inventory analyses as with EDMS (See Appendix D). Continuing with this example study, an assessment of the total time the aircraft spend on the ground was computed. On average independent of operation type the time spent on the ground for the entire taxi event (ramp to runway) was 374 seconds (6 minutes, 14 seconds). This does not include pushback or runway time. An alternate technique would be to utilize the known queuing delays at the airport and simply lump the operations at one location. For this analysis all segments of the taxi, whether the aircraft is moving or stationary, will be computed by modeling the total duration as a run up operation.

56 Lumped Time in Mode Analysis: All operations at one location. At one extreme, each run up could be modeled at the individual locations of the aircraft with the appropriate orientation. The other extreme is all operations lumped at one location. Both analysis techniques have the same total duration. The sensitivity of the contours to modeling scenario is shown in Figure 47 which displays the #5 run ups (red) with a 1 minute duration threshold, the full contour build up #3 (solid) with high-fidelity modeling and #6 Lumped Time in Mode (dashed). For convenience the location picked for this simplified analysis was the airport reference point. Had this been a ‘real’ study the modeler could improve the analysis by computing the geometric center of the taxi operations, or by taking guidance from runway usage statistics. Figure 47. Contour build up #3-all (solid), #5-run ups (red), #6-lumped mode (dashed). Clearly the impact of the geometric distribution (or lack thereof) of the time in mode operations can be seen. This suggests another simple improvement. Consider the primary holding locations at the airport. In this case there are three obvious holding locations driving the shape of the contours, one at the end of each of the primary runways. The high fidelity modeling including both one minute threshold stationary run up and moving operations was simplified to stationary operations at three locations, coincident with the end of the runway activity zones. The total duration of the operations remained constant as the modeling fidelity changed. Lumped Time in Mode Analysis. All operations modeled at three run up locations. Figure 48 compares the geometric level of modeling of the high fidelity combined operations (solid) with the simplified three locations lumped (red) time in mode operations. Since the purpose of this study is to assess the modeling sensitivity, no attempt was made to

57 determine a broadly applicable modeling philosophy for selection of the three run up centroids. However, it was felt that at each location operations would be represented best by pointing the aircraft the 4 cardinal directions (East, West, North, South) because while taxiing, the aircraft will have potentially faced all directions during ground operations. Figure 48. Run up operations: #3-high-fidelity all ops (black) and #7-lumped mode (red). 4.2.2.3. Nominal Taxi Track Modeling When considering the moving portion of the taxi operations, one can perform a high fidelity analysis (#4) or simplify the analysis by utilizing a fewer number of taxi tracks. In this case the overall taxi motions were assessed and six individual tracks were selected from the 1205 detailed tracks. The goal of the selection process was to cover the full extent of the spatial coverage over the airport grounds. Operations were then assigned to these 6 nominal taxi tracks randomly ensuring that the total number and type of aircraft represented in the study was preserved. Only the tracks were modified and assigned from the original detailed track to one of the 6 nominal tracks. Low Fidelity Analysis. Nominal taxi tracks, user defined profiles, departure NPDs for taxi, no run ups included. A side by side comparison of the high fidelity taxi tracks and the selected 6 nominal taxi tracks can be seen in Figure 49. Figure 50 provides a comparison of the community noise impact from the high fidelity tracks #4 (black) and the six nominal taxi tracks #8 (red).

58 Figure 49. All taxi tracks and 6 nominal tracks. Figure 50. High fidelity tracks (#4-black) and nominal taxi tracks #8 (red).

59 Potential improvements to this modeling scenario would be to model the direction along the taxi ways tangent to the taxi heading, but modeling the run ups in the center of the taxi turning locations with all possible heading directions. This leads to the obvious parallels with detailed flight track analyses, namely the creation of backbone tracks based on statistical analysis, the assignment of operations to individual taxi segments (over flights in INM) and perhaps even the examination of ground radar data to obtain taxi movements. In summary, one of the most surprising and unanticipated findings was the impact of the taxi geometry on the contours. It was very instructive to learn that a mere six taxi segments, selected to span the geographic area of interest, was comparable (within a few dB DNL) with the benchmark high-fidelity contours for an annual busy day at a major U.S. airport. This finding pointed strongly to the fact that the key element for taxi modeling (taxi operation geometric positioning) is already in INM, so a short term taxi modeling need can be met without requiring massive INM changes. 4.3. INM Modeling Considerations In order to properly assess taxi modeling techniques targeting inclusion within the framework of INM 7, one must explore the capabilities of the current INM7 taxi operation modeling process. This section will explain and discuss the current recommended and accepted techniques for modeling taxi operations. In previous sections in this report, various INM modeling features have been used without detailed examination. The technical basis upon which INM has been built is described in Reference (31). This Section will cover those elements in INM7 which impact the modeling of taxi noise. An in depth examination of the two sets of INM NPD data curves (approach and departure) will be made. The current INM extrapolation procedures for obtaining NPD data at lower thrust settings, typical of taxi operations, will be exercised. The impact of duration effects from static run up operations, the consistency between combining static and moving flight operations and alternative techniques using other modeling options currently available within INM will be examined. The use of flight profile altitudes in order to activate or deactivate specific propagation and source directivity features will be explored. In the following examples a Boeing 737-700 aircraft was selected based on the availability of empirical data for taxi operations. Based on standard emissions ground idle recommendations (14), 7% thrust was used to model the taxi operations. For a Stage 4 operation with takeoff thrust at 24,000 lbs, this corresponds to an INM modeled thrust setting of 1680 lbs. Other sources of taxi thrust include recent ICAO / CAEP findings (18) indicating approximately a 5% idle thrust setting for the next generation 737s (-700, -800 and -900) and a slightly higher 5.5% - 6% setting for classic 737s (-200, -300, -400 and -500). For the purposes of the next few sections all INM modeling utilizes 1680 lbs for ground idle thrust for both stationary and moving taxi operations. 4.3.1. Duration and Noise Exposure from Static Run up Operations Static run up operations and the associated duration equivalence between static and moving operations have already been employed in the high fidelity assessment in Section 4.2. The physical basis behind this equivalence will be explained. The effect of duration on the noise exposure metric (SEL) is straightforward. As provided in the INM Technical Manual, sound

60 exposure is directly proportional to the run up duration time. In terms of sound exposure level, this may be represented by Equation 1. SEL = 10 log10(t) + Level (1) Here, “Level” is the adjusted maximum level of concern and “t” is the run up duration time in seconds. This fundamental relationship between exposure and duration is important in the modeling of taxi noise because INM can not model stationary aircraft, requiring instead the substitution of a run up operation for a taxiing aircraft in a stop and hold operation. The fundamental premise of the SEL NPD curves implies integrated noise from an infinite line segment. As a result, static operations occupying only a single point can not be computed from the SEL data. Rather, static operations require Lmax source data, to which a duration adjustment applied, in order to obtain SEL contours. The consistency between SEL NPD and Lmax NPD curves is therefore important, specifically when considering a single source at a constant thrust level which can either remain still or move merely by the release of the brakes. The impact of using the Approach or the Departure NPD curve becomes important when combining moving and static operations, or using a time assessment with one mode or the other to model taxi noise. 4.3.2. Selection of NPD Curves for a Moving Taxi Operation The key item to be examined here is the multiplicity and consistency of data between the INM static run up data and the INM approach NPD and departure NPD data. The previous section detailed the duration relationship for stationary operations. For moving operations the time element has already been incorporated into the SEL values in the NPD curves. When modeling moving taxi operations, one treats the taxi segments as level over flights, at the altitude of the engines and flight speed of the taxiing aircraft. When modeling such an operation within INM, one may choose between arrival and departure NPD curves. The INM manual recommends using approach NPD curves, because they generally have data at lower thrust settings than the departure NPD curves, thereby requiring less data extrapolation. Two important facts must be considered: 1) the flight condition from which this data was obtained does matter and 2) consistency with static operations could be an issue depending on the modeling fidelity employed. In exploring the impact of using a departure-NPD curve instead of an approach-NPD curve, it is important to note the fact that the INM GUI only allows run up operations to use Departure NPD curves and are modeled as ‘on the ground.’ No altitude variation is allowed for the run up pad. To show the difference of using a Departure NPD curve instead of an Approach NPD curve, a custom aircraft was created that had the same Departure NPD curves as the 737- 700 Approach NPD curves. Figure 51 shows the contours from 100 sec duration of each aircraft for 1680 lbs thrust. The black line contour is a 737-700 (using the standard Departure NPD curves) and the red contour lines are from the custom 737-700 (using the standard Approach NPD curves). As can be seen the SEL levels from the custom 737-700 are less than those of the standard 737-700 farther from the aircraft and greater closer in. Of interest is the fact that the 50 dB contours approximate one another. At higher contour levels the arrival NPD curve results in larger contours than the departure NPD curve. However at lower contour levels (further away from the run up operation) the reverse is true – the Departure NPD curve run up operation contours are larger than the arrival NPD curve run up contours. There is a consequence of which set of NPD curves are used to extrapolate down to typical taxi thrust settings, notably the result of two primary features: the available data levels

61 and hence the extrapolation procedure, and also the difference in spectral classes and hence absorption rates for the two different sets of NPD curves. Figure 51. Run up SEL single event contours comparing departure NPD (black) vs. arrival NPD (red). 4.3.3. Combining Static and Moving Portions of Taxi Operations Analyses were performed in order to investigate the effectiveness of the current taxi noise modeling technique - a combination of run up and near ground level over flight operations - as recommended in the INM 7.0 User Guide (1). Several elements come into play, including the source modeling, operational characterization and the importance of spatial distribution. This section will utilize a single event comprised of both static and moving portions, in order to investigate the impact of decisions the noise modeler must make. The effects of the combination of run ups and over flights near ground altitude and their relative spatial distribution will be considered from the perspective of the surrounding community. The community perspective is reflected in the contours by providing three levels of comparisons, over a large geographic area (~ 15 miles), an intermediate area (~8 miles) and finally a smaller area close to the airport property (~4 miles). A single recommendation for modeling technique can not be made; rather modeling decisions must consider the location of the communities of interest relative to the geographic scales of the modeling features. Given a total taxi operation time and a percentage of time the plane is moving, what can be modeled? Consider the following case: a 737-700 taxiing down a one-half nautical mile straight taxiway. It has a total taxi time of 240 seconds (6 minutes): during 50% of the time the plane is moving, and during the remaining 50% of time, the vehicle is stationary. For this example an average taxi speed of 15 knots and 7% thrust (1680 pounds) is utilized. Within INM, a flight track representative of the moving portions of the taxi operation is defined, and locations at which static operations (treated as run ups) are defined. Both thrust level and speed are prescribed in the INM profile for the moving flight segments. A thrust level and run up duration

62 are used to define static operations. An important exception is made here: in order to more closely compare run up versus over flight operations, the over flight will be modeled with the previously developed user defined SEL NPD extrapolated departure thrust settings in order to more closely match the Run up conditions. An alternative technique to enact a closer comparison between the operational types, would match the LAmax level of the run up and over flight at some distance, it would also have introduced yet another sensitivity parameter, namely the distance at which the LAmax predictions are matched, be them in the far, intermediate or close in community. This comparison based on LAmax run up and Departure SEL NPD data will suffice to point out the major features and modeling implications for distributed operations. Figure 52 a) shows a B737-700 run up modeled in a single location at 1680 pounds thrust with a duration of 120 seconds. Within the INM study three run up operations, each with a 40 second duration are modeled. This is the acoustical equivalent of a single 120 second event. The run up location is in the middle of the taxi segment. a) b) Figure 52. SEL run up operation a) center run up, b) west, center, east run up locations. Figure 52 b) shows three run up locations modeled at the beginning, midpoint and end of the taxiway, each as a single operation with 40 second duration and 0.25 nm apart. A comparison of the single and triple location run up modeling effects is provided in Figure 53 over several regions of interest. If the community area of interest is over 5 nm away from the location of the run up operation, (Figure 53 a) modeling can likely be represented by a single run up location in lieu of three separate locations 0.25 nm apart. For community regions of interest at intermediate distances from the airport taxi operation (Figure 53 b) a judgment call must be made depending on the orientation of the community. For example if the point of interest are abeam the operations (north and south) contour differences are not as substantial and a single location with lumped operations might suffice. If the community area of interest is significantly closer to the airport, within a few nautical miles (Figure 53 c), then simplified equivalent duration modeling is not appropriate.

63 a) b) c) Figure 53. SEL run up contour, center only (black) and left+center+right (red) at different scales. 4.3.4. Comparison of INM7 Extrapolated NPD data with Taxi Measurement Data Included here are the results from one sensitivity study, in which we compared INM7 predictions with measurement data. It is important to note that the actual engine operating state (% thrust) during the measurements was not known since flight data recorder information was not available during any of these measurements. Different thrust levels were modeled in INM in order to determine if a change in the modeled thrust would better match the measured taxi noise

64 levels. If such a simple change in INM would bring predictions and measurements in alignment, it would be the simplest and least expensive way to model taxi noise in INM. Unfortunately, as will be shown in the remainder of this section, this simple thrust change in INM did not match measurement data. In order to determine the sensitivity to thrust level in INM, a simple track was created to model a 737-700 aircraft during a single taxi operation. As suggested in the INM manual, a flight profile for the track was created with an altitude equal to aircraft's engine height. The speed was kept at a constant 15 knots for all profiles. The thrust of the profile was varied from 1 to 40% of the maximum static thrust, and the noise levels at detailed grid points were plotted. The standard INM extrapolation procedures were applied. The distances from the track to the grid points were the same as those used in the NPD curves plus an additional distance (235 ft) in order to mimic the location of noise measurements during taxi operations at Providence Rhode Island for ongoing activities at the T. F. Green State Airport (33). In order to better visualize the relationship between the different thrust settings, the difference level was calculated from the 7% thrust numbers. A comparison of the INM7 taxi predictions (lines) with the measurements (circles) is given in Figure 54 for several aircraft types, while Figure 55 focuses on the 737-700. The sensitivity to thrust setting of predicted taxi noise using default INM7 NPD data and extrapolation procedures is given in Figure 56. Measurement Comparison with INM NPDs 70 75 80 85 90 95 100 105 0 100 200 300 400 500 600 700 Distance (Ft) Lm ax (d B A ) B737-700 INM7 Lmax CFM567B 3000 lbs A319-131 INM7 Lmax V2522A 2000 Lbs MD83 (sub for MD88) INM7 Lmax 2JT8D2 4000 Lbs GV Sub for CRJ INM7 Lmax BR710 1830 Lbs B737 300 CRJ MD88 A319 G4 Figure 54. Taxi noise comparison for several aircraft types (INM7 and measurements).

65 737-700 Taxi Thrust Sensitivity & comparison with Measurements 10 20 30 40 50 60 70 80 90 100 100 1000 10000 100000 Distance (ft) LA M A X (d B ) 24 240 480 960 1200 1320 1680 2400 4800 7200 MEAS 1 MEAS 2 Thurst (fn/delta Figure 55. Taxi noise comparison for the B737-700 (INM7 and measurements). 737-700 Taxi Thrust Sensitivity - LAMAX -2 -1 0 1 2 3 4 5 6 100 1000 10000 100000 Distance (ft) D iff er en ce L ev el (d B fr om 7 % tr ac e) 1 2 4 5 5.5 7 10 20 30 40 Thrust (%) Figure 56. Difference between measurements and standard INM7 taxi noise predictions.

66 4.3.5. Specifying NPD Curves Section 4.3.6 demonstrated that INM NPD data extrapolation to taxi / idle thrust settings did not agree particularly well with measurement data. The objective of this section is to determine whether customizing the INM NPD curves to match the measurement data would improve the quality of noise predictions. The data source for this analysis is two taxiing 737-700 aircraft events, measured at T.F. Green airport (Appendix B). The measurements were made 235 ft from the taxiway centerline. Both aircraft were traveling in the same direction. Their average taxi speed was 21 knots. The average SEL from the two operations was 90 dB. The meteorological conditions during the measurement were: Temperature 46 degrees Fahrenheit, barometer 30.17 in Hg and relative humidity 24%. An INM7 study was created to mirror this scenario consisting of a 2 nm, straight-segment track oriented in the East / West direction. A 737-700 aircraft was added to the study together with a case and scenario. A user defined points profile was made for the aircraft using 3000 lb Approach thrust with speed 21 knots and altitude 5.5 ft (the composite engine height for this aircraft). A detailed grid was created 235’ from the mid-point of the taxi segment. One flight operation was made that joined this profile and track. To study the effect meteorological conditions, within the INM study the modify NPD curves feature was enabled. The effect of humidity on absorption is accounted for by using the spectral class to adjust the difference in absorption to the various distances which were pre- populated in the NPD curves (D: distance) for sea level standard day conditions. The measured data could have been corrected from the actual atmospheric conditions to sea level standard day, however we opted to instead engage the modify NPD curves option in INM thereby predicting INM results at the measurement atmospheric conditions. The INM7 run for this operation predicted an SEL of 100.2 dB for the grid point for the case without modifying the NPD curves and an SEL of 98.2 dB with modifying the NPD curves. To obtain a corrected NPD curve for the custom aircraft, the 3000 lb approach levels were normalized to the measured levels by subtracting the difference between the predicted and measured SEL at 235’. This was done for both cases – with and without weather considered. The default approach-NPD curves at various power settings (indicated as 3000A, 4000A etc…) are compared with the corrected NPD curve (solid dots) and the corrected and modified NPD curve (plus symbol) in Figure 57.

67 737-700 NPD SEL Curves 40.0 50.0 60.0 70.0 80.0 90.0 100.0 100 1000 10000 100000 Distance (ft) Le ve l ( dB re v 20 m ic ro Pa ) 3000 A 4000 A 5000 A 6000 A 7000 A Corrected Corr-ModNPD Figure 57. INM7 approach and user defined taxi NPD curves based on measurements. Two user-defined aircraft were then added to the INM7 study using these custom noise IDs with the corrected data as the 3000 lb approach power condition. Since INM7 requires one to define in the custom NPD dataset for at least two power settings, a 4000 lb approach data NPD curve was added to the custom noise ID with identical corrections. The custom aircraft were modeled as operations on as above. Both cases correctly predicted the measured SEL of 90 dB at 235’. The case for the corrected aircraft using the weather data to make the NPD curve had to be run checking the ‘modify NPD curves’ in the case study with the meteorological conditions noted above in order to match the 90 dB at 235’. The INM7 default spectral class, #203 for Approach, was used. The difference in the SEL contour for the original 737-700 and these customized versions can be seen in Figure 58. The area is a 16 nm by 16 nm square. The dashed contour lines represent the corrected predictions for the aircraft made without modifying the NPD curves, the dotted contour lines represent the corrected predictions for the aircraft made with the ‘modify NPD curves’ option, and the solid lines represent the predictions from the original 737-700 aircraft modeled with 3000 lbs thrust and traveling 21 knots. As can be seen, the effect of the slightly higher amplitude NPD curve results in the contour area of the Corrected 737-700 expanding by a proportional amount. The purpose of this simple exercise was to develop a process for obtaining a new INM NPD curve from a simple measurement data point, to create a user defined NPD curve within INM and successfully predict the measured noise value while correcting for atmospheric conditions using an INM spectral class. These objectives were accomplished, lending credence to one modeling concept whereby a new taxi-NPD class could be created in INM (in addition to approach-NPD and departure-NPD) and populated based on actual taxi noise measurement data.

68 Figure 58. Contour comparison taxi operation: INM7, user defined, user defined + modified 4.3.6. Source Height Modeling and INM Directivity The INM User’s Guide specifies that for moving taxi operations the flight profile should specify the flight altitude as the average engine installation height. As long as the engine height is greater than zero, INM is prevented from applying longitudinal directivity (Ground-Based Directivity Adjustment - as is applied to the region behind the take off roll), to the propagation computations. Lateral directivity, as specified by SAE (34) is applied only when the aircraft is above 0 ft. height. In this section, the effects of the aircraft altitude defined in the user profile, are investigated. The effect of varying the source height was explored using a single track operation. The taxi segment starts at the origin and heads ‘East.’ Two separate single event taxi operations were modeled, both using departure NPD curves at the same thrust setting, 1680 lbs. In one case the B737-700 aircraft was modeled at the composite engine height of 5.5 ft. In the second case, the aircraft was placed on the ground at 0 ft. Figure 59 shows SEL predictions for a single 15 knot taxi operation traveling from west to east, modeled in INM as an over flight the half-nautical mile length of the taxiway with 1680 pounds departure thrust at an altitude of 5.5 ft AGL. A half nautical mile distance traveled at a speed of 15 knots will take 120 seconds to traverse. As can be seen in Figure 59, the 40 dB SEL contour nearly reaches the 25,000 ft grid point (corresponding to the largest distance in the INM NPD curves). The maximum width of this contour is only slightly less in extent than that of the 120 second single and combined static run up operations, as seen in Figures 51, 52 and 53. The pinched in aspect of the contour lines at the sides of the figure are an artifact of how INM applies a noise fraction to an infinitely long track segment to approximate the half mile segment

69 modeled here. This example demonstrates INM’s use of finite track segments and SEL NPD curves for predicting noise contours from moving operations. In this particular example the flight segment was modeled using a departure SEL NPD curve rather than an approach SEL NPD curve to more closely match the static run up computation presented above in Sections 4.3.2 and 4.3.3. Figure 59. INM single segment moving taxi operation at 5.5 ft AGL. An interesting effect, the filling in of the contour area behind the aircraft, can be seen when modeling the 15 knot taxi operation on the ground at 0. ft AGL (Figure 60). The taxi operation starts at the West side of the track and heads in the easterly direction. The contours are filled in based on the behind takeoff roll longitudinal directivity algorithms in INM. These algorithms are designed to more accurately portray departure operations, but are only enabled within INM by placing the aircraft profile at 0 ft. The East side of the figure still shows the effect of INM modeling a finite track segment. It should be noted that this track segment connected to another track segment, rather than treated as an isolated segment, the first segment would be ‘filled in’ by the subsequent track segment. This is related to the reasoning behind the noise fraction adjustment used by INM to account for finite length track segments; though how accurate this portrayal is for individual taxi segments especially at longer propagation distances with low grazing angles remains to be seen. Figure 61 displays INM contours for moving taxi operations at both 5.5 ft and 0 ft with one another. There is a significant impact on the noise contours due to INM applying the behind takeoff roll algorithm for taxi operations.

70 Figure 60. INM single segment moving taxi operation at 0 ft AGL. Figure 61. Comparison of INM Single Segment Moving Taxi Operations at 0 and 5.5 ft AGL 4.3.7. Effects of Taxi Speed Of interest in the modeling of taxi noise is noting the fact that the speed assigned to a flight segment influences the exposure metric (SEL) as stated in the INM7 Technical Manual (Equation 2). DURadj = 10 log10(ASref/ASseg) (2)

71 Where ASref is 160 knots and ASseq is the speed of the aircraft on a track segment. An increase in speed will decrease the SEL. Using a 2-nmi straight track a set of detailed grid points were made at distances perpendicular to the midpoint of the track. User-defined profiles were made for a 737-700 traveling at 5.5’ altitude (the composite height of this aircraft’s engines) at 1680 pounds thrust (7% of max static) using INM extrapolations of the approach operation NPD curves as recommended in the INM manual for speeds noted. As expected, for a doubling of speed doubled, the SEL decreased by 3 dB (Figure 62). This is important information in defining the relationship between when an aircraft is stationary and when it is moving. SEL and Speed 35 45 55 65 75 85 95 105 100 1000 10000 100000Distance (ft) Le ve l ( dB re f 2 0 m ic ro Pa ) 15 knots 30 knots 45 knots 60 knots Figure 62. INM7 effect of taxi speed on SEL. 4.3.8. Single Event Standard INM7 Modeling of Time in Mode Operations It has been suggested that taxi operations can be more simply modeled using a “time in mode” approach, where operations are represented by an accumulation of time spent stationary and moving lumped together at key locations on the airfield. This example shows the dominance of the static operations over moving operations when employing the current INM manual suggested modeling technique: Approach-SEL for over flights and run up data for static operations. In this study, three different operations, each requiring 240 seconds, were considered and compared. Figure 63 shows a composite of three different computations by INM, with each operation a duration of 240 seconds. The red contour indicates only the taxiway as an over flight – the same operation considered in the generation of Figure 59, and the blue contour is the over flight taxi operation and three static run ups at one location. The red contour lines result from modeling a single over flight of a 737-700 at 5.5 feet altitude with 1680 pounds of (departure) thrust traveling 7.5 knots on the taxiway. The time it would take an aircraft to travel the 0.5 nm length of a taxiway at 7.5 knots is 240 seconds. The dashed, green contour lines result from the summation of two operations in INM: an over flight of a 737-700 at 5.5 feet altitude with 1680 pounds of (departure) thrust traveling 15 knots on the same taxiway segment with a duration of 120 seconds and three static operations (as depicted in Figure 51). The three individual static run up operations are at the beginning, midpoint, and end of the taxiway for a total duration of 120

72 seconds. The solid blue contour lines result from modeling two operations in INM. One operation is the same over flight of the 737-700 at 5.5 feet altitude with 1680 pounds of (departure) thrust traveling 15 knots on the taxiway. The other operation is that depicted in Figure 52, a single run up at the midpoint of the taxiway of a 737-700 running with 1680 pounds thrust for a duration of 40 seconds with 3 operations assigned to it. As expected, these contours overlay those made from the 3 separate run up locations of equal durations. One can see the dominance the static portion of the taxi operations have over the over moving portion of the taxi operations in most areas of the study. This is primarily due to the inconsistency between the approach-NPD curves and the Lmax data for static run up operations. With the current range of available thrust settings and INM’s extrapolation procedure, stationary operational noise (obtained from the run up Lmax NPD curve) rather than the moving operational noise (obtained from the extrapolated approach-SEL NPD curve) dominates the contours. This example indicates that modeling of taxiway segments in INM may be represented as single run up operations at the midpoint of the segment with the caveat that the levels need to match those that actually represent a taxiing aircraft. Figure 63. INM single event comparison from combined moving and stationary operations. In summary, within INM and the current AEDT noise computational module, the distinction between source modeling and propagation modeling is not completely separated. It is important that there is consistency between the NPD data used for moving operations and the Lmax data used for static operations. This consistency will permit taxi predictions from time in mode operational parameters if the geometric area of coverage is preserved. The impact of holding queues on the nearby community can be important for distances under one mile. Expansion of the taxi modeling capability within INM and in the future, AEDT will need to ensure consistency between propagation and source modeling effects and commonality between static and moving operations NPD source noise data.