1--;:::=:;:'
:;7 WindowsL:J
-c::::~=~~::;:::::;:=~:::i::;::====::::J1 Ceiling
,
~
, ~ FlowC::::===========::JI Floor
Fig. 3.29. NASA Langley Low Turbulence Pressure Tunnel test section access for aero-acoustic phased array installation. Top part of diagram shows the test section ceiling as viewed from above the test section. Instrumentable areas include locations of existing wdows and lights. Lower part of diagram shows the position of the model relative to the in-strumentable areas
Case Study II
-Array Design for The NASA Langley Low-Turbulence Pressure Tunnel
The NASA Langley Low-Turbulence Pressure Tunnel (LTPT) is a single return, closed-circuit tunnel that can be operated at stagnation pressures from 0.1 to 10 atmospheres. The rectangular test section is 3-feet (0.91 m) wide by 7.S-feet (2.29 m) high by 7.S-feet longll • This case study for phased array design relates to airframe noise tests conducted in the LTPT in 1997, 1998, and 1999 (Choud-hari et al. 2002). In the 1997 test, a two-dimensional wing with a part-span flap was suspended between the sidewalls of the tunnel. Acoustic array access was limited to three openings in the ceiling of the test section as shown in Figure 3.29. The data acquisition system was limited to 60 simultaneous channels.
Since there was a requirement to simultaneously acquire data from eight fluc-tuating pressure transducers mounted in the model, the array was limited to 52 microphones.
11 NASA Langley Research Center Wind Tunnel Enterprise Low-Turbulence Pressure Tun-nel brochure, Document Version 1.0, Hampton, Virginia.
The design strategy used a multi-arm spiral array design covering a circu-lar region that encompassed the three openings in the ceiling. Only the design locations that corresponded to instrumentable areas were retained. With this approach and using an equal-aperture-area-per-sensor spiral sampling strat-egy, the approximate number of sensors to be used in the design, ND , can be calculated using
(3.38) where N is the number of sensors to be used in the array, AI is the area of the instrumentable region of the design space, and AD is the area of the design space. This approach took advantage of the non -redundancy of multi -arm spi-ral arrays to design (nominally) N-element arrays where the N elements are all in instrumentable areas. However, it was difficult to come up with an array with worst-case sidelobes comparable to what can be obtained with an unen-cumbered region. A first attempt, shown in Figure 3.30, placed the center of the multi-arm spiral array far enough downstream in the large opening so that nearly two full circles of microphones would be retained after positions not falling in instrumentable regions were eliminated. Several designs using this approach had poor sidelobe characteristics, probably due to the absence of certain spatial lags caused by the large non-instrumentable regions of the test
40 30 20 10
c 0
=
>--10 -20 -30
-40L-~--~--~--~--~~--~--~
-40 -30 -20 -10 0 10 20 30 40 x (in.)
Fig. 3.30. Phased array design approach for an aeroacoustic array installation in the NASA Langley Low Turbulence Pressure Tunnel. Only microphone positions falling within the three rectangular regions (instrumentable areas of the test section ceiling) were included in the design. This approach was abandoned because of unacceptable worst -case sidelobe levels
40 .---, 20
~
~
2010 P. 0 0 0 0 0 0
~ >. 0 0°0000 ~oOo
=
c >. 0-10 ~
o o
0 0~
- 20- 20
- 10 0 10 20 30 - 40
x ( in.) - 40 - 20 0 20 40
x (in.)
Fig. 3.31. Array design (left) from approach shown in Figure 3.30 and associated coarray (right). Note the clustering in the coarray in some areas and the relative thinness in other areas
30 20 10
§. 0
>.
-1 0 - 20 - 30
- 30 - 20 - 10 0 10 20 30 x (in.)
Fig. 3.32. Phased array design approach for an aeroacoustic array installation in the NASA Langley Low Turbulence Pressure Tunnel. Only microphone positions falling within the three rectangular regions (instrumentable areas of the test section ceiling) were included in the design. This approach was used because of improved worst-case sidelobe levels over the approach shown in Figure 3.30
section ceiling. The coarray for the design of Figure 3.30 is shown in Figure 3.31. Note the clustering in the coarray in some regions and the relative thin-ness in other regions.
Trial and error led to a slightly different approach shown in Figure 3.32 that yielded reasonably good sidelobe characteristics. The array was centered up-stream of the large opening such that only a portion of the innermost circle of sensors ended up in the large opening. An additional circle was added to the central part of the array with a radius very close to that of the innermost cir-cle of the multi-arm spiral to add some very short spacings to the co array. This
40
Fig. 3.33. Array design (left) from approach shown in Figure 3.32 and associated coarray (right). Note the relative uniformity in the co array as compared with Figure 3.31
34
Fig. 3.34. Six dB down isosur-face for the array shown in Figure 3.33. The isosurface was created by beamforming for a postulated point source corre-sponding to a location on the model flap edge (15,0,43) rela-tive to the phase center of the array (0,0,0)
approach tended to distribute the coarray more evenly as shown in Figure 3.33. However, after the array was deployed it was discovered that there was an undesirable characteristic in sidelobe performance. Typically, the array evalu-ation process postulates a source at some distance from the array surface nom-inally corresponding to where sources are expected to exist on the device to be tested. Characteristics of the array (resolution and sidelobes) are then exam-ined on a plane that cuts through the postulated source position and is paral-lel to the array. It turns out that this approach is inadequate for evaluating ar-rays where part of the aperture is encumbered such that sensor positions must be excluded from the array design. In the present case sidelobes existed in the third dimension that were higher in level than those in the evaluation plane.
For the present array, an example of these "out-of-plane sidelobes" is shown in Figure 3.34. Fortuitously, the out-of-plane sidelobes were above-and-aft and below-and-forward of the important source near the flap edge, and therefore
Fig. 3.35. Array design approach for small upstream window in the NASA Langley Low turbulence Pressure Tunnel. Only microphone positions falling within the rectangular re-gion (instrumentable area of the test section ceiling) were included in the design. Asterisks show locations of existing microphones from the array of Figure 3.33. Microphone posi-tions for the small-window design that were too close to the existing posiposi-tions from the larger array were excluded from the small-window array
did not significantly interfere with the flap side-edge measurements. The directive to be taken from this case study is to evaluate aperture-encumbered arrays in three dimensions to insure that out-of-plane sidelobes are not detrimental to array performance.
Tests in 1998 and 1999 added a leading-edge slat to the model used in 1997.
The model was also inverted so the array was looking at the pressure side. In the 1998 test and through subsequent analytical work (Singer et al. 1999) the slat noise was shown to be highly directional, impacting only a fraction of the array microphones and thus rendering the array's noise source level estimates suspect. Since the dominant slat-radiated noise was at a high frequency, the relatively large array was not required to achieve good source resolution. In the 1999 test entry, the model was positioned such that the radiated slat noise would impact the small upstream windows in the ceiling and a new smaller ar-ray with 60 microphones was designed for one of the upstream windows. The array design approach is shown in Figure 3.35. The design process was rela-tively straightforward since the opening was unencumbered. A multi-arm spi-ral array strategy was employed and sensors with design positions outside the rectangular opening or that conflicted with existing sensor locations (from the large array) were excluded. The resulting array and co array are shown in Figure 3.36.
The 1999 test entry also included landing gear. The results from the large ar-ray illuminate another important consideration in arar-ray design. Recall that the depth resolution (perpendicular to the array) of a planar array is not
particu-5 r;;;===;::;::::;:;~;;l o
0 0
10 .---.
c· 0000 0 0 8<9 0'*
O 0 0 0 0 0 0 c~_ O
V. 0° t:P~Qo 00
->. ofl.oO 0 0 0
at
00 0 >.- 5 - 10 ~---~
- 10 - 5 0 5 10 - 20 - 1 0 0 10 20
x (in.) x (in.)
Fig. 3.36. Array design (left) from approach shown in Figure 3.35 and associated co array (right). Asterisks (left) show locations of existing microphones from the array of Figure 3.33
Test Section Side View
C:::::::;:7=~-;r?==<::::=:;::====:::::J1 Ceiling Large Array
Center 3 dB Contours
- Gear Source
~I Sla::ce
Flow
C::============:=:JI
FloorFig. 3.37. Side view of NASA Langley Low Turbulence Pressure Tunnel with phased ar-ray and airframe model installed. Note that the phase center of the large phased arar-ray (Figure 3.33) is collinear with the landing gear and slat. Three dB down contours for noise sources at the landing gear and slat overlap making resolution of the two independent sources impossible (compare with Figure 3.38)
larly good. An isosurface some-number-of-dB down from the peak has its ma-jor axis oriented along a line passing through the phase center of the array and the source location. As shown in Figure 3.37, the landing gear, slat, and phase center of the large array are co-linear. Given the resolution of the large array this made it impossible to clearly distinguish between slat and landing gear generated noise sources. However, the smaller array was able to distinguish be-tween slat and gear sources as illustrated in Figure 3.38.