As shown in Fig. 1(d), measurements were performed in a low-speed open-jet wind tunnel. Specifically, the wind tunnel is powered by a 5.5-kW centrifugal blower. Between the blower and the nozzle there is a cubic plenum connected, the size of which is 800 mm×800 mm×800 mm. In order to uncouple the flow from the blower and ensure a low turbulence intensity, a series of honeycomb layers are subsequently installed inside the plenum. As for the nozzle, length is 1000 mm and the square outlet size is 75 mm×75 mm. One horizontal end-plate is mounted and flush with the lower side of the nozzle, providing a three-quarter open test section for the measurement.

The plane jet blowing system includes a 2.2-kW centrifugal blower and a cubic plenum as well as a nozzle, shown in Fig. 1(c). In order to facilitate movement of the plenum, a hose is linked from the blower outlet to the plenum bottom, on which there is a hole for connection. Height of the plenum is 540 mm and size of the horizontal section is 424 mm×424 mm. Likewise, two layers of honeycomb with hexagonal grid (6-mm edge length) are installed to uncouple the flow from the blower and acquire a low turbulence intensity inside the plenum. In addition, baffles are added to minimize recirculation. As for shape of the nozzle on top of the plenum, it is fixed to be rectangular with a constant length of 100 mm and changeable width. In the test, the nozzle was set to be flush with the wind tunnel end-plate, and the width was chosen to be 10 mm. In addition, velocity of both main-flow coming from the wind tunnel and the plane jet could be controlled by adjusting the blower inlet area.

As mentioned earlier, tandem rods were used as the test object. Radius of the rods, D is 4 mm and both of them are polished (Fig. 2) to eliminate effects from roughness of the surface. For gap width L, particular values of L/D may lead to different flow regimes. For example,^[14] when 1.1–1.3< L/D < 3.5–3.8, free shear layers that get separated from the front rod may reattach alternately, permanently or intermittently onto the rear rod. Therefore, vortex shedding takes place mainly behind the rear rod. In this study, this configuration was adopted. More precisely, values of L/D in the study were 2, 2.5, 3, and 3.5.

Fig. 2.

	Figure Option ViewDownloadNew Window
	Fig. 2. Tandem rods used in the test.

As shown in Fig. 1(e), tandem rods were supported by tow blocks with a series of tubes, which could help adjust height to the end-plate and gap width between tandem rods. As for the height and the gap, both would be chosen based on PIV results hereinafter.

Beam-forming is an array-based measurement technique for sound source localization. It has become a standard method for spatial noise mapping. In this study, a planar semi-circular array with a camera was built, shown in Fig. 1(a) and Fig. 1(b). The array has a diameter of 600 mm and an irregular pattern of 25 microphones. On account of the beam-forming camera, the noise localization can be superimposed on a real photograph background. Moreover, this pattern can reduce typical spatial aliasing of the regular array. In this study, beam-forming in the time domain with diagonal deletion was used with 1/3 octave band filter. Mathematical theory can be found in Refs. [15] and [16]. Signals from microphones were amplified before being simultaneously sampled using the National Instrument DAQ system. Sampling rate and sampling time were set to be 80 kHz and 6 s respectively. All microphones were calibrated before being used. To validate the noise localization, an array calibration technique using one small speaker was conducted. Explicitly, one small speaker was differently positioned on the end-plate, generating broadband noise and therefore localized by the beam-forming array. Results are shown in Fig. 3, noise contour superimposed on background pictures visualized the sound pressure level (SPL) on the end-plate from the top view and the contour was normalised by subtracting the maximum SPL of each case to highlight source localization. As can be seen, the peak localises the speaker movement very well.

Fig. 3.

	Figure Option ViewDownloadNew Window
	Fig. 3. Beam-forming calibration.

Additionally, as the microphone array was placed out of main-flow, the refraction due to density gradient caused by the shear layer could affect the directivity of sound propagation as well as noise localization accuracy. However, since no directivity measurement was conducted and the beam-forming array was only 493 mm high to the end-plate in this study, no shear layer refraction corrections were performed.

During the test, speed of the main-flow and the plane jet were controlled to be 50 m/s (U_∞) and 55 m/s (U_j). Three flow regimes were chosen to facilitate comparison. As schematically shown in Fig. 6, “background noise” represents the regime that main-flow is actively switched on without the rods and the plane jet. Meanwhile, the plane jet orifice is sealed to avoid possible cavity noise induced by main-flow; “NoPJ” represents the regime that main-flow is on with rods, but no plane jet; “PJ” represents the regime that the plane jet is turned on compared to “NoPJ”.

Fig. 6.

	Figure Option ViewDownloadNew Window
	Fig. 6. Schematic diagram of the test campaign.

Previous studies^[19–21] demonstrated that the flow pattern behind the jet has complicated structures, and there is a low speed area behind it. Since the aerodynamic noise is in direct ratio to the 6th power of the local flow speed around the bluff-body components, it is best to put the rods in the low speed area. In this study, PIV experiment was performed to seek the location for rods.

Figure 7 is PIV results on the flow speed field with vectors illustrating flow direction. The Y axis represents the height referred to the end-plate and normalized by D. The X axis represents the stream-wise length normalized by W. In addition, the origin was located in the centre of the slot. As shown in Fig. 7(a), a stream-wise low velocity area exists due to the boundary layers, which has a height of approximately 2h/D. More studies on boundary layers of cross-flow can be found in Refs. [22] and [23]. Meanwhile, over the 18h/D, there is a shear layer with high speed gradient, which can result in noise refraction mentioned earlier. Rods should be lower than the shear layer flow and higher than the boundary layer, both of which do not exist when the landing gear interacts with the coming flow, in order to approach real flight conditions as much as possible. More importantly, when the plane jet is turned on, a low speed area is highly preferred as it will contribute to the noise reduction. Therefore, the primary rod was simply chosen to be centred at (10,2.5) in terms of 2D view. Figure 8 shows the velocity profile normalized by U_∞ at the section x = 10, in which the two circles represented the rods (the profile was acquired without rods). The front rod was centered at (10,2.5). As is clearly depicted, velocity can be significantly lowered with the plane jet. In other words, the plane jet has great potential to reduce the landing gear noise.

Fig. 7.

	Figure Option ViewDownloadNew Window
	Fig. 7. Flow velocity field for case (a) without plane jet and (b) with plane jet.

Fig. 8.

	Figure Option ViewDownloadNew Window
	Fig. 8. Velocity profile for the impinging velocity.

To validate the application for noise reduction, spectrum analysis was performed.

For spectrum analysis, one microphone in the beam-forming array was chosen, located over the tandem rods. The block size was 4096 and data were averaged over 110 blocks. The sound pressure level was referred to 2 × 10⁻⁵ Pa.

To begin with, an example was given to pre-analyse the acoustic performance, and more importantly, to find the background noise, e.g. power noise from the main-flow and the plane jet. Figure 9 is an example of the noise spectrum for L/D = 2. It is obvious that the three curves overlapped each other within most of the frequency range marked by I, which could be explained by power noise of the wind tunnel. However, within the frequency range marked by II and III, “PJ” case is noisier than others. Likewise, this noise increase can be explained by power noise from the plane jet generator. Because the acoustic tests were not performed in an anechoic chamber and background noise existed as discussed above, the low frequency part will not be considered in the spectrum analysis below.

Fig. 9.

	Figure Option ViewDownloadNew Window
	Fig. 9. Background noise range.

As shown in Fig. 9, “NoPJ” in each case is considerably more noisy than “background noise” and therefore it is concluded that the noise increase is caused by tandem rods. A main tone exists in each regime, approximately in the range of 2 kHz to 2.5 kHz, which is caused by vortex shedding behind the rear rod. The shedding frequency as well as SPL is dependent on L/D. More theory on tandem rods noise generation related to L/D can be found in Refs. [24]–[26]. Additionally, a secondary tone co-exists with the main tone, which can be caused by vortex shedding from the front rod, whose shear layer reattaches alternately, permanently or intermittently onto the rear rod and is weaker. After the plane jet is turned on in “PJ”, overall speaking, noise is notably reduced, especially at both vortex shedding frequencies. It is obvious that the tones are removed by the plane jet. However, for each regime, in some frequency range, the noise in the “PJ” is higher than that in “NoPJ”. This noise increase can be from the plane jet. As is known to all the plane jet can emit jet noise, especially when it interacts with the main-flow. More relevant conclusions would be obtained from noise localization.

Figure 11 is an example of the noise localization by using the beam-forming for L/D = 3.5. The contour corresponds to the SPL distribution on the background photo. Data were filtered with the 1/3 octave bands centred at 2 kHz, 4 kHz, 10 kHz, and 16 kHz respectively. Pictures in each column shows comparison between “NoPJ” and “PJ” regimes in the same band, normalized by subtracting the maximum SPL of “NoPJ”. Overall, SPL decreases dramatically after the plane jet is turned on. To be more specific, all the noise localization results for “NoPJ” clearly illustrate that rods are the main noise source. Nonetheless, the main noise source in “PJ” varies over bands. For bands centered at 2 kHz and 16 kHz, SPL peaks around the plane jet nozzle, which means that the plane jet is the main contributor to noise emission of these bands. By contrast, the main noise source is still tandem rods in bands centered at 4 kHz and 10 kHz. This proves that even with the existence of the plane jet, rods can still make notable noise, which indicates that there could be further noise reduction potentially.

Fig. 10.

	Figure Option ViewDownloadNew Window
	Fig. 10. Noise spectral of different cases: (a) L/D = 2; (b) L/D = 2.5; (c) L/D = 3; (d) L/D = 3.5.

Fig. 11.

	Figure Option ViewDownloadNew Window
	Fig. 11. Results (case L/D = 2) of time domain beam-forming with 1/3 octave band filter centred at: (a) 2 kHz; (b) 4 kHz; (c) 10 kHz; (d) 16 kHz.

Figures 12 and 13 respectively illustrate peak reduction at the main shedding frequency and OASPL reduction in certain frequency range with A-weighting, which accounts for relative loudness perceived by human hearing. As is shown, the plane jet works very well for reduction in terms of both tonal and broadband noise.

Fig. 12.

	Figure Option ViewDownloadNew Window
	Fig. 12. Peak reduction.

Fig. 13.

	Figure Option ViewDownloadNew Window
	Fig. 13. OASPL reduction.

Figures 14 and 15 present the profiles in the test sections acquired by hot-wire measurement. Height, velocity, and Reynolds stress term are normalized and non-dimensionalised by D/2 radius of the rod, U_∞, and U_∞² respectively. As illustrated in Fig. 14(a), all “NoPJ” cases unexpectedly start with lower velocity than “PJ” cases, more precisely, in the range of roughly 0 < 2(h – h₀)/D < 0.7. The PIV demonstrated that flow speed in the area behind the plane jet can be significantly reduced. Therefore, the reason for slight increase of the flow speed inside the rod gap is unclear. A possible explanation could be recirculation in the wake behind and induced by the front rod.^[27] Hence, further noise generated by rods can be explained by flow coming from behind caused by the recirculation. Subsequently, though velocity profiles of “NoPJ” start from being lower than “PJ”, they soar from around h = 0.7 and shortly reach main-flow velocity level, especially around the height of 2(h – h₀)/D = 1, which is as high as the rod. Moreover, it is at exactly the same height that the vortex shedding happened on the rear rod. By contrast, “PJ” keeps velocity level and does not increase. As acoustic test results already showed, noise of the rods decreased on account of the plane jet introduction. Therefore, the conclusion can be made that the impinging velocity at 2(h – h₀)/D = 1 dominates over that of other height in terms of noise generation.

Fig. 14.

	Figure Option ViewDownloadNew Window
	Fig. 14. Profiles of the test section in front of the rear rod: (a) velocity; (b) Reynold stress.

Fig. 15.

	Figure Option ViewDownloadNew Window
	Fig. 15. Profile of the test section behind the rear rod: (a) velocity; (b) Reynold stress.

Apart from velocity, turbulence level also underlies noise generation. Figure 14(b) depicts Reynolds stress term, which is a sign of turbulence level. Obviously, for all “NoPJ”, u′u′ peaks around 0.9 < 2(h – h₀)/D < 1.3 as a result of shear layer from the front rod and those peaks are eliminated with the existence of the plane jet. Likewise, in the part with low and high height, Reynolds stress in “PJ” is higher than “NoPJ”, which could be possibly attributed to recirculation caused by the front rod as well. Therefore, a similar conclusion can be drawn from a perspective of Reynolds stress, namely impinging turbulence level around 2(h – h₀)/D = 1 plays a more important role than others in terms of noise generation. Moreover, the plane jet is able to reduce the impinging velocity as well as the turbulence level around this height.

In Fig. 15, i.e., wake of the rear rod, reduction of both u and u′u′ are notable with the plane jet on. Specifically, u is significantly reduced at all sampling points. In terms of the Reynolds stress, as shown in Fig. 15(b), in all cases it diminishes in the range of 0 < 2(h – h₀)/D < 1, which is firmly inside the wake of the rear rod. Apparently, it can be concluded that vortex shedding of the rear rod is diminished. With increasing, Reynolds stress in “NoPJ” gradually becomes lower than “PJ”. That is due to the fact that sampling points are outside the wake flow and for those points, turbulence level is higher than “NoPJ” due to the plane jet.

In order to further validate elimination of the vortex shedding induced by the rear rod with the plane jet, velocity fluctuation spectra for “NoPJ” and “PJ” were calculated. Since 2(h – h₀)/D = 1 is the main shedding place, the analysis was chosen to be at this height. Figure 16 presents the result. Likewise, for noise spectra, the block size was 4096 and the data were averaged over 110 blocks. Here all the fluctuations are referred to U_∞ by dB, i.e.

Fig. 16.

	Figure Option ViewDownloadNew Window
	Fig. 16. Velocity fluctuation spectral at 2(h – h0)/D = 1: (a) L/D = 2, (b) L/D = 2.5, (c) L/D = 3, (d) L/D = 3.5.

As shown in Fig. 16, fluctuation level stays high in low frequency range that is below around 400 Hz in the spectra of “NoPJ”, then declines with increase of the frequency. A characteristic here is that there are two fluctuation peaks. Those peaks firmly confirm that the vortex shedding exists behind the rear rod, and the peak in the L/D = 2 regime is higher than others, which means that vortex shedding is stronger in 8 mm than others. This is consistent with the noise spectra because tonal noise, i.e. the peak generated in “NoPJ” regime in L/D = 2 is also superior to others.

Once the plane jet is turned on, apparently the fluctuation significantly decreases in all frequencies, especially the peaks, as they are removed. These results further demonstrate that the plane jet can potentially be of great use to decrease vortex shedding induced by the rear rod so that noise can be reduced.