- Email: [email protected]
Experimental study of vortex–structure interaction noise radiated from rod–airfoil configurations
Journal of Fluids and Structures 51 (2014) 313–325
Contents lists available at ScienceDirect
Journal of Fluids and Structures journal homepage: www.elsevier.com/locate/jfs
Experimental study of vortex–structure interaction noise radiated from rod–airfoil configurations Yong Li a,b,n, Xun-nian Wang a, Zheng-wu Chen a,b, Zheng-chu Li a,b a b
State Key Laboratory of Aerodynamics, China Aerodynamics Research and Development Centre, Mianyang 621000, China Key Laboratory of Aerodynamic Noise Control, China Aerodynamics Research and Development Centre, Mianyang 621000, China
a r t i c l e i n f o
abstract
Article history: Received 24 July 2013 Accepted 24 August 2014 Available online 27 October 2014
Vortex–structure interaction noise radiated from an airfoil embedded in the wake of a rod is investigated experimentally in an anechoic wind tunnel by means of a phased microphone array for acoustic tests and particle image velocimetry (PIV) for the flow field measurements. The rod–airfoil configuration is varied by changing the rod diameter (D), adjusting the cross-stream position (Y) of the rod and the streamwise gap (L) between the rod and the airfoil leading edge. Two noise control concepts, including “air blowing” on the upstream rod and a soft-vane leading edge on the airfoil, are applied to control the vortex–structure interaction noise. The motivation behind this study is to investigate the effects of the three parameters on the characteristics of the radiated noise and then explore the influences of the noise control concepts. Both the vortex–structure interaction noise and the rod vortex shedding tonal noise are analysed. The acoustic test results show that both the position and magnitude of the dominant noise source of the rod–airfoil model are highly dependent on the parameters considered. In the case where the vortex– structure interaction noise is dominant, the application of the air blowing and the soft vane can effectively attenuate the interaction noise. Flow field measurements suggest that the intensity of the vortex–structure interaction and the flow impingement on the airfoil leading edge are suppressed by the control methods, giving a reduction in noise. & 2014 Elsevier Ltd. All rights reserved.
Keywords: Rod–airfoil model Vortex–structure interaction noise Noise control Vortex shedding Air blowing and soft vane
1. Introduction Vortex–structure interaction noise is a main concern in several aeronautical and industrial applications. Two important devices involving such interaction noise are the rotor configurations of turbo-engine and helicopter rotors, in which the downstream airfoil blades lie in the wake of upstream blades. At any given time, vortices shed from the upstream blades interact with and impinge upon the downstream blades, giving rise to a host of noise and vibration issues. The rod–airfoil configuration consisting of an airfoil located in the near wake of a rod in a flow is believed to be a benchmark well-suited for numerical predictions of such vortex–structure interaction noise and noise generation mechanisms (Casalino et al., 2003; Jacob et al., 2005). Jacob et al. (2005) performed PIV measurement combined with proper orthogonal decomposition (POD) reconstructions to identify the coherent structures in the rod wake and compare the experimental results with the numerical simulations using Reynolds average Navier–Stokes (RANS) and large eddy simulation (LES). These analyses highlighted that strong three dimensional effects were responsible for spectral broadening around the rod vortex shedding n
Corresponding author. E-mail address: [email protected] (Y. Li).
http://dx.doi.org/10.1016/j.jfluidstructs.2014.08.014 0889-9746/& 2014 Elsevier Ltd. All rights reserved.
314
Y. Li et al. / Journal of Fluids and Structures 51 (2014) 313–325
Nomenclature D H C L Y ReD ReC x, y, z
cylindrical rod diameter square rod width airfoil chord length streamwise gap between the rod and the airfoil leading edge cross-stream position of the rod Reynolds number based on rod diameter D Reynolds number based on airfoil chord length C Cartesian coordinates
U1 U V Ma Ωz f fvs St Stvs ΔLd PIV SPL
wind speed streamwise (x) velocity cross-stream (y) velocity Mach number spanwise vorticity ð∂V=∂x ∂U=∂yÞ frequency vortex shedding frequency Strouhal number fD/U1 vortex shedding Strouhal number (fvsD/U1) sound pressure level difference particle imaging velocimetry sound pressure level
frequency in the subcritical regime, and identified that the airfoil leading edge was the main contributor to the noise emission in a rod–airfoil configuration due to vortex–structure interaction. The flow physics of the rod–airfoil model have been further studied by means of experimental methods. In a similar manner to Jacob et al. (2005), Takagi et al. (2006) investigated the influence of the cylinder transverse location on the turbulent development of the flow around the airfoil. Their results showed that the transverse location of the cylinder might have an impact on the radiated acoustic field. In order to identify the flow features that are responsible for the noise generation, Henning et al. (2008, 2009) performed dual PIV and simultaneous far-field microphone measurements. By means of a correlation-technique, they identified the regularities in the near-field fluctuations that are related to the radiated sound field. Lorenzoni et al. (2009, 2010) applied time-resolved PIV combined with a pressure reconstruction procedure to characterise the noise sources. The comparison with simultaneous microphone measurements revealed the ability of this method to predict reasonably well the magnitude of the tonal peak of emission and the narrow band spectrum around it. Giesler and Sarradj (2009) firstly performed microphone array measurements to investigate the influences of the rod diameter and the streamwise gap between the rod and the airfoil leading edge on the broadband noise generation at the frequencies higher than the vortex shedding frequency. Their results showed that the noise generation for lower frequencies depended more strongly on the cylinder diameter than on the streamwise gap. In addition to these experimental studies, further understanding of the details of the rod–airfoil interactions can be gained from numerical simulations. Casalino et al. (2003) developed an aeroacoustic code based on porous Ffows-Williams and Hawking (FW–H) in combination with unsteady RANS simulations for the acoustic prediction. In a similar manner, Magagnato et al. (2003), Sorguven et al. (2003) and Boudet et al. (2005) coupled a FW–H acoustic analogy with LES calculations which allowed capturing of the sub-Karman vortical structures responsible for noise emission at higher frequencies. Greschner et al. (2008) combined a detached eddy simulation (DES) with the FW–H analogy to evaluate far-field pressure spectra. These investigations presented results which were in agreement with the experiments, but the applied numerical methods decouple the aerodynamic and acoustic fields, which makes it difficult to determine the details of the flow physics that are responsible for the noise generation. Recently, Berland et al. (2010, 2011) performed a direct noise calculation (DNC), based on compressible LES, to predict the sound radiation from the rod–airfoil configuration. The DNC method does not require any modelling of the sound sources and computes aerodynamic and acoustic fluctuations within a single run. The good agreement between calculation and experiment demonstrated the promise of the DNC method on noise prediction and generation mechanisms. The dominant noise source on the rod–airfoil model was demonstrated to be caused by the impingement of the Karman vortices upon the airfoil leading edge. Therefore, a modification of the leading edge aerodynamics can be expected to modify both the turbulent flow and the sound emission of the rod–airfoil configuration. Siller et al. (2005) applied a blowing and suction at the airfoil leading edge to modify the flow structure and showed that the peak sound pressure level (SPL) at vortex shedding frequency in the far-field could be significantly reduced by air blowing, while air suction might increase the main peak level. Most of the aforementioned studies were performed on one rod–airfoil configuration with fixed streamwise gap and/or rod diameter on the centreline. However, different gaps and rod diameters may have effects on the generation of the dominant noise. Giesler and Sarradj (2009) demonstrated experimentally the effect of the rod diameter and the streamwise gap on the broadband noise of high frequencies far above the vortex shedding frequency. In addition, Berland et al. (2011) performed DNC on the influence of the streamwise gap. The current paper follows Giesler and Sarradj (2009) by investigating further the effects of these two parameters on the radiated noise, not only at the broadband frequencies but also at the vortex shedding frequency, especially the influence of the gap/diameter ratio. In addition, the effect of the rod cross-stream position has also been investigated. Based on the gained knowledge, two noise control concepts using “air blowing” on the upstream rod and a downstream soft-vane airfoil leading edge are explored here. Acoustic and flow field measurements are taken in an anechoic wind tunnel using a phased microphone array and the PIV technique respectively. The use of the microphone array combining with an advanced data processing of CLEAN algorithm is to accurately localise noise sources and determine the dominant noise at different configurations, whereas the PIV technique reveals the corresponding flow physics of the noise generation and control.
Y. Li et al. / Journal of Fluids and Structures 51 (2014) 313–325
315
2. Experimental details 2.1. Test model and wind tunnel set-up A sketch of the rod–airfoil test model and the set-up is shown in Fig. 1. A symmetric NACA0012 airfoil with chord length of C ¼150 mm is located downstream of a rod, both extending by 400 mm in the spanwise z-direction. The airfoil chord is aligned with the centre of the nozzle and parallel to the streamwise x-direction. The origin of the Cartesian coordinate system is located at the mid-span position of the airfoil leading edge which is located 300 mm downstream of the nozzle exit. Eight cylindrical rods with diameters of D¼3 mm, 5 mm, 8 mm, 11 mm, 15 mm, 18 mm, 21 mm and 25 mm are used, and also a H ¼15 mm square rod. The streamwise gap between the rod and the airfoil leading edge is adjusted between L ¼30 mm and 170 mm at 10 mm intervals and the cross-stream position of the rod varies between Y ¼0 mm (i.e. on the centreline) and 40 mm at 5 mm intervals. Measurements are performed at wind speeds of U1 ¼30 m/s and 60 m/s with incoming turbulence level of 0.5% at the centreline, yielding the corresponding Reynolds number ReD between 6.3 103 and 1.06 105 based on the rod diameter D, and ReC between 4.5 105 and 9.0 105 based on the airfoil chord length C. Fig. 2 shows the photographs of the test set-up that is implemented in the anechoic wind tunnel of China Aerodynamics Research and Development Centre. The wind tunnel is an open-jet closed-circuit test facility with a rectangular 0.55 m 0.4 m nozzle exit. The anechoic chamber dimension is 5.5 m 5 m 4 m and its six walls are covered by 0.5 m long fibreglass wedges, yielding 99% acoustic absorption above 200 Hz. The rod–airfoil model is vertically installed between two metal endplates that are directly attached to the nozzle exit. The model is placed into the potential core of the nozzle jet. The lower endplate has an insertion made of Plexi-glass to ensure optical access for PIV recording. The model and the endplates are rigidly fixed to ensure no vibration during all the tests. 2.2. Measurement techniques The noise level and location are measured by a 0.72 m-diameter ring microphone array which consists of 32 omnidirectional GRAS type 40PH 1/4 in. microphones with a frequency response range of 4 Hz–70 kHz (Fig. 2a). All the microphones are covered by wind shields to eliminate the wind effects. A camera mounted in the centre of the array is used to capture the model image. The array scan plane is aligned with the airfoil chord and 0.75 m away from the surface of the microphone array.
Fig. 1. Sketch of the rod–airfoil configuration and the test set-up (not to scale).
Fig. 2. Experimental set-up in the anechoic wind tunnel: (a) close-up view showing the rod–airfoil model between two endplates and the microphone array; and (b) PIV measurement set-up.
316
Y. Li et al. / Journal of Fluids and Structures 51 (2014) 313–325
The acoustic data are captured using National Instruments data acquisition cards at a sampling rate of 48 kHz and analysed using a block size of 4096, yielding a frequency resolution of about 12 Hz. To reduce spectral leakage, a Hanning window function is applied to each single block before performing fast Fourier transform. A total of 100 blocks are averaged for statistical confidence and the uncertainty in the noise level is determined to be within 2%. The data are processed by the advanced CLEAN–SC algorithm based on spatial coherence sources (Sijtsma, 2007). This algorithm is chosen following a classical beamforming technique (Dougherty, 2002) to improve the spatial resolution and reduce the sidelobe level. With this processing scheme and the measurement set-up, the spatial resolution of the current microphone array is about 0.41λ (λ: acoustic wavelength). The result of the post-processing is a two dimensional map showing the noise source distribution and the sound pressure level, in the following referred to as noise source map. To remove the effects of the convective flow and the free shear layer on the noise positioning the noise maps are corrected using the average-Mach number (Ma) method of Sijtsma (2010), i.e., the uniform flow Ma is replaced by the average Macor between the test model and the microphone array. The corrected flow Macor is given by
ζ ξ
M acor ¼ M a ð1Þ
Y ξ ; mic where ζ, ξ and Ymic are the model position, shear layer position and microphone array position, respectively. As a result, the uncertainty in noise source positioning is determined to be within 5%. To investigate the sound pressure spectra in the whole frequency range, the results are obtained by averaging the data from all the 32 channel microphones. The flow velocity data are acquired using a planar PIV system of Beamtech Vlite500 with a New Wave Gemini Nd: YAG dual laser. The PIV system operates at a repetition rate of 2 Hz. A special ES11000 4008 2672 high resolution CCD camera used to record a two-dimensional view of the flow field is set up perpendicular to the laser sheet on the scan plane and fixed on a support structure underneath the endplates (Fig. 2b). The flow is seeded with diethylhexylsebacate (DEHS) tracer particles with mean particle diameter of approximately 1 μm. The laser light sheet thickness at the model area is about 1 mm. The flow field area of interest is confined to a small window depicted in Fig. 1 and marked as thick dashed line. A total of 250 images of the flow field are taken for each experimental condition. Each image set is processed using a 32 32 pixels cross-correlation area and a 50% 50% overlap to improve the resolution of the vector map. The accuracy of the instantaneous velocity fields can be estimated by assuming an accuracy in the correlation of 0.1 pixel displacement (Raffel et al., 2000), which corresponds to a maximum error in the velocity of 0.4 m/s. The uncertainty in the instantaneous velocities is estimated to be about 2% for the present set up, while the corresponding spanwise vorticity Ωz has an uncertainty of about 5%.
2.3. Noise control concepts Two noise control concepts are developed, i.e., the air blowing and the soft vane leading edge as presented in Fig. 3. The air blowing rod (Fig. 3a) has 300 mm (length) 3 mm (width) slots on the sides of a square rod from which the air blows out in the cross-stream y-direction. The width of the air blowing rod is H¼ 15 mm. A square rod is chosen, rather than a circular one, only because it is relatively easy to construct the air blowing slots. The size of the upstream air entrance to the blowing slots is 300 mm (length) 9 mm (width). Unlike air blowing devices used in other studies, where an external air pump may have to be used, the current design allows the incoming main flow to directly provide the blowing air. Therefore, no external mass flow is introduced into the main flow. In such a way, the air-blowing speed and its flow momentum may be highly dependent on the main flow speed (U1) and the slot size. The idea behind this control concept is to reduce the flow impingement speed and push away the rod shed vortices from the airfoil leading edge by providing an “air curtain” in front of the airfoil. The concept of “air curtain” was previously introduced by Oerlemans and Bruin (2009) on landing gear main strut noise reduction using external air blowing and by Huang and Zhang (2010) on noise control using plasma actuators. The second concept is a soft vane, shown in Fig. 3b, wherein the airfoil leading edge is firstly made slotted and then covered with a porous woven cloth approximately 0.2 mm-thick in this case. The leading edge slot is 20 mm (depth) 2 mm (width). This soft vane concept was investigated by Jones et al. (2009) on engine noise reduction. As indicated by Jones et al. the slotted leading edge may allow communication of pressure fluctuations on both sides of the surface and the chamber formed within the cloth should provide pressure release (relative to the solid surface it replaced) on the leading edge surface.
3. Results and discussion This section summarises the results of several tests using the aforementioned microphone array and PIV technique to identify the governing mechanisms that lead to the noise generation and reduction. The parametric study (including the effects of D, L and Y) and the assessment of the soft vane are made using the cylindrical rod–airfoil configuration, while the assessment of the air blowing is made by comparing the air-blowing rod with a solid square rod of same size. Table 1 lists these test configurations and their characteristics.
Y. Li et al. / Journal of Fluids and Structures 51 (2014) 313–325
317
Fig. 3. Schematic of the noise control concepts (not to scale): (a) air blowing rod; and (b) soft-vane airfoil leading edge (slotted leading edge covered by 0.2 mm-thick cloth). Table 1 Overview of the test cases conducted on the rod–airfoil model. C-rod: cylindrical rod; S-rod: square rod. Tests
Rod type
D or H (mm)
L [Step] (mm)
Y [Step] (mm)
U1 (m/s)
Effects of D Effects of L Influences of L/D Effects of Y Air blowing Soft vane
C-rod C-rod C-rod C-rod S-rod C-rod
3–25 15 3–25 15 15 15
150 30–170 [10] 30–170 [10] 150 150 150
0 0 0 0–40 [5] 0 0
30, 30, 30, 30, 30, 30,
60 60 60 60 60 60
3.1. Acoustic measurements 3.1.1. Effects of rod diameter (D) To study the effects of the rod diameter, the cylindrical rod is first located on the centreline and 150 mm upstream of the airfoil leading edge. As expected, it is found from sound spectra (not shown here) that the dominant frequency is close to each rod vortex shedding frequency fvs and the corresponding Strouhal number Stvs is about 0.2. The noise source maps at fvs shown in Fig. 4 indicate that both location and level of the dominant noise vary with changing the rod diameter. For the small rods of 3 mm and 8 mm diameter (Fig. 4a and b), it can be seen clearly that the dominant noises at fvs ¼3.9 kHz and 1.5 kHz are concentrated on the upstream rods and the peak levels are 82 dB and 90 dB respectively. The array dynamic range is 10 dB in both source maps. This suggests that the radiated noise from the test model with a small rod is generated mainly by the upstream rod due to the unsteady loads induced by the vortex shedding. In the case of the 15 mm-diameter rod (Fig. 4c), however, the noise map shows that the mainlobe centre for fvs ¼760 Hz is located at the downstream airfoil leading edge, indicating that the noise is now radiated largely from the vortex–structure interaction on the surface of the airfoil leading edge. The dynamic range is reduced to 7 dB in Fig. 4c to show more clearly the noise source position at relatively low frequency. To see how the airfoil installed changes the level and location of the main source, a source map for the 15 mm-diameter rod-only is also given in Fig. 4d. It can be seen obviously from Fig. 4c and d that the peak level is
318
Y. Li et al. / Journal of Fluids and Structures 51 (2014) 313–325
Fig. 4. Noise source maps at the vortex shedding frequency (fvs) for different diameter cylindrical rods and airfoil. Flow (U1 ¼60 m/s) is from left to right: (a) D¼ 3 mm, fvs ¼ 3.9 kHz (b) D ¼8 mm, fvs ¼1.5 kHz, (c) D ¼15 mm, fvs ¼ 760 Hz and (d) D ¼ 15 mm, rod-only, fvs ¼760 Hz.
increased by 12 dB and the dominant noise position moves downstream to the airfoil leading edge once the airfoil is installed. The source location difference of 160 mm between the two maps is slightly less than the array spatial resolution of 0.41λ ¼180 mm at 760 Hz. As the rod diameter is increased further, the current microphone array is unable to distinguish the position of the vortex shedding noise from the vortex–structure interaction noise due to the poor array spatial resolution at lower frequencies. For instance, the array spatial resolution at fvs ¼490 Hz for the 21 mm-diameter rod is about 280 mm which is far more than the streamwise distance between the rod and the airfoil, making it difficult to locate precisely the noise source. The peak sound pressure level (SPL) difference ΔLd at fvs (obtained in 1/3 octave band) between the rod–airfoil and the rod-only for all the eight rods are shown in Fig. 5. For Dr8 mm, the rod–airfoil has almost the same SPL as the rod-only, suggesting that there is almost no interaction noise. As the rod diameter is increased, the SPL difference grows. At DZ15 mm, the interaction noise at the airfoil leading edge increases by about 12 dB compared to that of the rod-only, which is in good agreement with the finding of Jacob et al. (2005). These results indicate that the intensity of vortex–structure interaction at the airfoil leading edge is reduced as the rod diameter is decreased. This is somewhat equivalent to noise reduction when the airfoil thickness is increased. Olsen and Wagner (1982), Gershfeld (2004), Giesler and Sarradj (2009) found that for low Mach number flows a bigger and/or blunter shape of the airfoil leading edge is generally responsible for a lower leading edge interaction noise. This can be explained by the rapid distortion theory (Goldstein and Atassi, 1976) that the small rod shed vortices are less deformed when passing through the leading edge and their potential field is less extended. As a result, the pressure fluctuation on the airfoil that is responsible for the noise generation is smaller. It also can be seen that the behaviour of the noise change is highly similar at the two test wind speeds of U1 ¼30 m/s and 60 m/s.
3.1.2. Effects of streamwise gap (L) The effects of the streamwise gap are assessed for the case of 15 mm-diameter rod. The gap is within the range of 30–170 mm. It has previously been demonstrated in Fig. 4c that with the rod diameter of 15 mm the dominant noise is radiated from the airfoil leading edge due to the vortex–structure interaction. However, Fig. 6 shows that both the vortex shedding Strouhal number Stvs and the corresponding peak SPL difference ΔLd vary with the change of the gap.
Y. Li et al. / Journal of Fluids and Structures 51 (2014) 313–325
319
Fig. 5. Peak SPL differences between the rod–airfoil model and the rod-only versus the rod diameter (D) at the two wind speeds. L ¼ 150 mm.
Fig. 6. Vortex shedding Strouhal number Stvs (a) and corresponding peak SPL difference (b) versus the streamwise gap (L). U1 ¼ 60 m/s, D¼ 15 mm.
The Strouhal number plotted in Fig. 6a shows that Stvs is 0.192 for the rod-only, and that for the rod–airfoil model Stvs gradually decreases from 0.19 at L ¼170 mm down to 0.163 at L¼50 mm. This phenomenon, as mentioned by Jacob et al. (2005), may be due to the modification of the vortex shedding through its interaction with the airfoil, suggesting a feedback of the airfoil onto the rod. With a further decrease in the gap to 40 mm, Stvs increases again to reach the value of 0.183. Concerning the peak SPL difference at Stvs shown in Fig. 6b, the noise increases as the gap is narrowed from 170 mm to 50 mm. At L¼50 mm, the noise reaches the highest value which is about 15 dB higher than the noise for the rod-only. At L ¼40 mm, the noise drops dramatically by about 32 dB from the highest value at L ¼50 mm, and is even 17 dB lower than that of the rod-only. This sharp decrease is likely due to the airfoil in the vicinity of the rod preventing the formation of Karman vortices.
3.1.3. Influences of gap/diameter ratio (L/D) It is useful to understand the noise characteristics with regards to appropriate dimensionless quantities of the gap/ diameter ratio (L/D). Fig. 7 shows the overall SPL differences between the rod–airfoil and the rod-only at the different ratios. Due to the finite gap range of L ¼30–170 mm, the L/D radio is relatively low for big rods and high for small rods. Nevertheless, as already indicated in Fig. 6b, it can be seen in Fig. 7 that there is a critical value of L/D ¼3.3 below which the overall SPL undergoes a sharp decrease of more than 30 dB. This critical ratio is in close agreement with the numerical finding of Berland et al. (2011) where the ratio is 3.5 in the case of one diameter rod. This phenomenon suggests that at this situation the embedded airfoil may work as a splitter plate which prevents the formation of the rod Karman vortices (Spiteri et al., 2008; You et al., 1998), contributing to the drastic noise reduction. As the L/D ratio is increased, the noise level gradually decreases. It would be expected that at L/DZ20 the noise of the rod–airfoil almost collapses to that of the rod-only, indicating that at larger L/D the vortex shedding tonal noise is dominant and only weak interaction happens at the airfoil leading edge. In such a case, to reduce the noise the focus should be concentrated on the upstream vortex suppression.
320
Y. Li et al. / Journal of Fluids and Structures 51 (2014) 313–325
Fig. 7. Overall SPL differences between the rod–airfoil and the rod-only against the gap/diameter ratio L/D. U1 ¼ 60 m/s.
Fig. 8. Sound pressure spectra (a) and overall SPL (b) at different cross-stream positions (Y) of the rod. U1 ¼60 m/s, D ¼15 mm, L ¼150 mm.
Fig. 9. Sound pressure spectra versus Strouhal number for the air blowing and the square rod. H¼ 15 mm and L ¼ 150 mm.
3.1.4. Effects of rod cross-stream position (Y) To study the effects of the rod cross-stream position, the 15 mm-diameter rod is located 150 mm upstream of the airfoil leading edge. Fig. 8 depicts a set of sound spectra and the peak SPLs at the different cross-stream positions. The sound spectra for the rod-only and the airfoil-only are also plotted in Fig. 8a. Compared with the other cases, the noise generated by the airfoil-only is low and can thus be ignored. For the rod-only, one can observe discrete frequency noise consisting of pure tones at fundamental Strouhal number (Stvs) and its harmonics. When the airfoil is embedded directly in the wake of the rod, i.e., at Y¼0 mm, the peak SPL at Stvs is increased by about 12 dB and the broadband noise around is also increased.
Y. Li et al. / Journal of Fluids and Structures 51 (2014) 313–325
321
Fig. 10. Noise source map comparison showing the noise reduction at vortex shedding frequency (fvs) using slotted airfoil leading edge and soft-vane airfoil leading edge respectively: (a) solid airfoil leading edge; (b) slotted airfoil leading edge; and (c) soft vane airfoil leading edge. Flow (U1 ¼ 60 m/s) is from left to right, D¼ 15 mm, L ¼ 150 mm.
The impingement of these vortical structures upon the airfoil leading edge generates sound sources, which is similar to the discrete frequency tones and broadband noise observed in turbomachinery: rotor blades undergo unsteady pressure fluctuations and one may observe discrete frequency radiation consisting of pure tones at harmonics of the blade passing frequency and broadband noise produced by the interaction of the blade surfaces with the turbulent wakes. The overall SPL gradually decreases as the cross-stream position (Y) increases, as shown in Fig. 8b. This noise reduction is due to some of the shed vortices missing the airfoil leading edge, thus reducing the intensity of the vortex–structure interaction. At the Y¼ 40 mm position, the noise level collapses to the same value of the rod-only, indicating that the rod shed vortices may totally miss the airfoil leading edge and there is almost no vortex–structure interaction. Thus, any method that can push away the shed vortices from the airfoil leading edge would have potential in reducing the interaction noise. 3.1.5. Assessment of noise control concepts This section aims to investigate the benefits of the air blowing rod and the soft vane on the vortex–structure interaction noise. For assessment of the air blowing rod, comparison is made between the air blowing rod and a square rod with same size. The rods are positioned on the centreline and 150 mm upstream of the airfoil leading edge. Fig. 9 shows the sound spectra comparison at the wind speeds of U1 ¼30 m/s and 60 m/s respectively. At the small Strouhal numbers below Stvs ¼0.19 there is a slight discrepancy on the spectra, suggesting that the air blowing does not affect the low frequency noise. At U1 ¼ 30 m/s, the peak SPL at Stvs is reduced by about 4 dB, but at U1 ¼60 m/s there is not much difference. At the Strouhal numbers between 0.19 and 1.0, the broadband noise is clearly reduced at the two wind speeds. The large noise increase at the higher Strouhal numbers (St41) are due to the self-noise of the air blowing. The overall SPL in the whole frequency range is reduced by about 3.8 dB and 1.8 dB at U1 ¼30 m/s and 60 m/s respectively. The noise reduction in the broadband frequencies suggests that the air blowing rod may act as an “air curtain” in front of the airfoil to reduce the impingement speed and the vortex–structure interaction. The effects of the “air curtain” concept on noise reduction were previously investigated by Li et al. (2010) on bluff-body using plasma technique and Oerlemans and Bruin (2009) on simplified landing gear main strut using upstream air blowing. In those studies, however, additional power was needed to generate plasma and air blowing, for example high voltage device for plasma and air pump respectively. They showed that
322
Y. Li et al. / Journal of Fluids and Structures 51 (2014) 313–325
the plasma and the air blowing reduced the impingement speed upon the downstream components, resulting in significant broadband noise reduction. In this paper, the blowing air is directly supplied from the incoming main flow, making the control device simple. However, the reduced performance at U1 ¼60 m/s suggests that this air blowing control concept gradually loses its control authority as the main flow speed increases. The effects of the soft vane are clearly demonstrated on the noise source maps shown in Fig. 10. The 15 mm-diameter rod is positioned 150 mm upstream of the airfoil leading edge. The noise source maps at vortex shedding frequency for the solid, slotted and soft vane airfoil leading edge are presented. Compared with the solid leading edge, the slotted leading edge gives a noise reduction of about 4 dB, whereas an improved noise reduction of 6 dB can be obtained with the soft vane by covering the slotted leading edge with 0.2 mm-thick cloth. A similar behaviour using the soft vane concept was also reported by Jones et al. (2009) for the experimental study of the engine fan noise reduction. The corresponding sound pressure spectra are shown in Fig. 11. It can be seen that, in addition to the noise reduction at Stvs, both the slotted leading edge and the soft vane leading edge also significantly suppress the broadband noise at the Strouhal numbers above Stvs. The low-frequency parts of the spectra for the cases with the slotted leading edge and the solid leading edge are similar, whereas the spectrum for the soft vane exhibits high levels with one peak centred around St¼0.055. The noise reduction by the slotted leading edge is most likely due to the direct pressure-fluctuation communication at the airfoil leading-edge surfaces that reduces the vortex–structure interaction area. The pressure-release surface provided by the soft vane is believed to reduce noise radiation efficiency. To improve the noise reduction, both the internal structure and the porosity of chamber in the soft vane may have to be optimised to achieve maximum sound absorption.
3.2. Flow field measurements Presenting PIV results from all the measured rod–airfoil configurations goes beyond the scope of this section. Therefore, only the results for representative rod–airfoil configurations measured at the wind speed of U1 ¼30 m/s are presented. Fig. 12 illustrates the contours of instantaneous spanwise vorticity Ωz at three different rod–airfoil configurations, i.e., at different streamwise gaps and cross-stream position. When the cylindrical rod of D ¼15 mm is positioned 150 mm directly upstream of the airfoil (Fig. 12a), the large scale vortices can be seen to emanate from the upper and lower shear layers. The shed vortices move downstream and gradually lose intensity and coherence. In the region around the airfoil leading edge,
Fig. 11. Comparison of sound pressure spectra between the three different airfoil leading edges.
Fig. 12. Contour plots of instantaneous spanwise vorticity Ωz (mm s 1) in the rod (D¼ 15 mm) wake region under three rod positions: (a) x¼ 150 mm, y¼ 0; (b) x ¼ 150 mm, y¼ 40 mm; and (c) x ¼ 40 mm, y¼0.
Y. Li et al. / Journal of Fluids and Structures 51 (2014) 313–325
323
it is hard to recognise the vortex structure. The observations are consistent with the results obtained by Jacob et al. (2005), Henning et al. (2009) and Lorenzoni et al. (2009). The downstream convective vortices directly impinge upon the airfoil leading edge, giving rise to high vortex–structure interaction noise. Although in this configuration the strongest vorticity is seen to be located in the area close to the rod, the dominant noise has been demonstrated by the microphone array to be the interaction noise at the airfoil leading edge. At the cross-stream position of Y¼ 40 mm (Fig. 12b), the shed vortices pass over the upper side of the airfoil surface and totally miss the airfoil leading edge. As a result, there is no interaction noise and the dominant noise is shown to be radiated from the rod shed vortices, and the noise is almost the same as that of the rod-only. Moving the rod close to the airfoil leading edge at the streamwise gap of L¼ 40 mm in Fig. 12c, no visible vortices between the rod and the airfoil can be observed, suggesting that the presence of the airfoil in the vicinity of the rod prevents the formation of the Karman vortices. In this situation the noise undergoes a sharp decrease, as already indicated in Fig. 6b. Fig. 13 presents the instantaneous vorticity Ωz for the rod–airfoil model at three different rod diameters. In the case of 3 mm-diameter rod (Fig. 13a), there is a well organised vortex street in the vicinity of the rod up to 50 mm downstream, indicating that the airfoil has no effect on the upstream rod shed vortices. With increasing rod diameter, the size of the shed vortices increases and due to the feedback effect of the airfoil it is hard to recognise the vortex structure before the airfoil leading edge in the case of 21 mm-diameter rod (Fig. 13c). The effect of the air blowing on the flow field has also been illustrated by PIV tests. To quantify the differences in the downstream wake, the streamwise U-velocity profiles are obtained at two x positions (x¼ 100 mm and 10 mm) upstream of the airfoil leading edge. The results comparing the square rod and the air-blowing rod are shown in Fig. 14. At the streamwise location of x¼ 100 mm, the local velocity in the central area is significantly reduced by the air blowing on both sides of the square rod. The velocity on the centreline is reduced by about 15 m/s. At the far downstream position of x ¼ 10 mm, i.e., at 10 mm upstream of the airfoil leading edge, the wake width is increased and the flow speed is reduced by about 5 m/s on the centreline. The reduced velocity can alleviate the flow impingement on the airfoil leading edge, hence attenuates the vortex–structure interaction. In such a way, the air blowing forms an “air curtain” in front of the airfoil, which
Fig. 13. Contour plots of instantaneous spanwise vorticity Ωz (mm s 1) in the rod wake region under three rod diameters: (a) D ¼3 mm; (b) D ¼8 mm; and (c) D ¼ 21 mm.
Fig. 14. Comparison of the streamwise U-velocity profiles between the square rod and the air blowing rod at two x positions: (a) x¼ 100 mm; and (b) x¼ 10 mm.
324
Y. Li et al. / Journal of Fluids and Structures 51 (2014) 313–325
has a similar effect as the passive control fairing used to reduce the landing gear noise (Dobrzynski et al., 2002; Li et al., 2012) by reducing the flow speed impinging upon the downstream components. Fig. 15 illustrates the instantaneous vorticity Ωz for the square rod and the air-blowing rod respectively. For the square rod (Fig. 15a), most of the vortices concentrate on the central area within 720 mm and impinge directly upon the airfoil leading edge. With the air blowing on both sides (Fig. 15b), however, a series of scattered vortices are observed. Some of the vortices reach as high as 40 mm and then pass over the airfoil leading edge because the air blowing pushes them away from the airfoil, contributing to the interaction noise reduction. As far as the slotted leading edge and the soft vane are concerned, the spanwise vorticity Ωz in the wake of the rod shown in Fig. 16 does not have much difference with that of solid leading edge, at least in the area of interest by the PIV view in this paper. The phenomena can be understood since the noise reduction should mainly be due to the relief of the pressure fluctuations on the surface of the airfoil leading edge, not the change of the upstream shed vortices. 4. Conclusions This paper investigates the effects of different rod–airfoil configurations on the vortex–structure interaction noise and addresses the techniques of upstream air blowing and a soft-vane airfoil leading edge as methods for reducing interaction noise. Acoustic and aerodynamic measurements are taken on a representative rod–airfoil model at two wind speeds with varies of the model configuration by changing the rod diameter (D), adjusting the streamwise gap (L) between the rod and the airfoil leading edge, and the rod cross-stream position (Y). The sound pressure level and position of the dominant noise are highly dependent on the three parameters. Generally speaking, the radiated noise increases with increasing rod diameter and/or shortening the streamwise gap due to the strong vortex–structure interaction at the airfoil leading edge. There exists a critical gap/diameter ratio of L/D ¼3.3 below which the noise level undergoes a sharp decrease of more than 30 dB due to the suppression of the rod's Karman vortices by the downstream airfoil. At about L/DZ20, there is almost no vortex–structure interaction and the noise is mainly radiated from the upstream rod vortex shedding. In this case, attention should be paid to suppress the upstream vortices to reduce the noise. However, once the vortex–structure interaction is dominant, control strategies should concentrate on suppressing both the vortices and the interaction. The noise of the rod–airfoil model gradually decreases with increasing the rod cross-stream position (Y), i.e., with moving the rod out of alignment with the airfoil. This noise decrease is due to the reduced vortices impinging upon the
Fig. 15. Comparison of the instantaneous spanwise vorticity Ωz (mm s 1) between (a) the square rod and (b) the air blowing rod.
Fig. 16. Contour plots of instantaneous spanwise vorticity Ωz (mm s 1) in the rod wake region with different airfoil leading edge: (a) solid airfoil leading edge; (b) slotted airfoil leading edge; and (c) soft-vane airfoil leading edge.
Y. Li et al. / Journal of Fluids and Structures 51 (2014) 313–325
325
airfoil leading edge. Once the shed vortices totally miss the airfoil leading edge at larger Y positions, the noise of the rod– airfoil model drops to that of the rod-only. Thus, any method that can prevent the vortices impinging upon the airfoil leading edge would have a potential on noise reduction. The noise control results demonstrate that the vortex–structure interaction noise can be attenuated by air blowing out of the sides of the upstream rod and by the soft-vane airfoil leading edge. The air blowing shows an effective noise reduction at U1 ¼30 m/s since it reduces the flow impingement speed and vortices impinging upon the airfoil leading edge. In such the case, the air blowing seems to create an “air curtain” in front of the airfoil that partly shields the airfoil from the incoming flow. At U1 ¼60 m/s, however, the air blowing loses it control authority, which puts a limitation on practical applications. As the applied blowing air is directly from the main flow, how to increase the air blowing speed and the flow momentum would be a challenge and needs further fundamental study. The soft vane airfoil leading edge which is formed by covering the slotted leading edge with a porous cloth shows significant noise attenuation at the two wind speeds. The control mechanism is the pressure release at the leading edge surface of the soft vane which reduces the transformed noise energy from the interaction. It would be expected that more noise reduction may be obtained if the internal structure and the porosity of the soft vane undergo an optimisation process. Acknowledgement The experiments are made possible through the support provided by State Key Laboratory of Aerodynamics in China (Project no. 4071506). The authors like to thank Dr. Malcolm Smith from Institute of Sound and Vibration Research (ISVR) at University of Southampton for helpful discussion and comments on the manuscript. References Berland, J., Lafon, P., Crouzet, F., Daude, F., Bailly, C., 2010. Numerical insight into sound sources of a rod–airfoil flow configuration using direct noise calculation. AIAA-2010-3705. Berland, J., Lafon, P., Crouzet, F., Daude, F., Bailly, C., 2011. A Parametric study of the noise radiated by the flow around multiple bodies: direct noise computation of the influence of the separating distance in rod–airfoil flow configurations. AIAA-2011-2819. Boudet, J., Grosjean, N., Jacob, M.C., 2005. Wake–airfoil interaction as broadband noise source: a large-eddy simulation study. International Journal of Aeroacoustics 4 (1), 93–115. Casalino, D., Jacob, M.C., Roger, M., 2003. Prediction of rod–airfoil interaction noise using the Ffowcs-Williams and Hawkings analogy. AIAA Journal 41 (2), 182–191. Dobrzynski, W., Chow, L.C., Guion, P., Schiells, D., 2002. Research into landing gear airframe noise reduction. AIAA-2002-2409. Dougherty, R.P., 2002. Beamforming in acoustic testing. In: Mueller, TJ (Ed.), Aeroacoustic Measurements, Springer, Berlin. Gershfeld, J., 2004. Leading edge noise from thick airfoils in turbulent flows. Journal of the Acoustical Society of America 116, 1416. Giesler, J., Sarradj, E., 2009. Measurement of broadband noise generation on rod–airfoil-configurations. AIAA-2009-3308. Goldstein, M.E., Atassi, H.M., 1976. A complete second order theory for the unsteady flow about an airfoil due to a periodic gust. Journal of Fluid Mechanics 74, 741–765. Greschner, B., Thiele, F., Jacob, M.C., Casalino, D., 2008. Prediction of sound generated by a rod–airfoil configuration using EASM DES and the generalised Lighthill/FW–H analogy. Computers & Fluids 37 (4), 402–413. Henning, A., Kaepernick, K., Ehrenfried, K., Koop, L., Dillman, A., 2008. Investigation of aeroacoustic noise generation by simultaneous particle image velocimetry and microphone measurements. Experiments in Fluids 45, 1073–1085. Henning, A., Koop, L., Ehrenfried, K., Lauterbach, A., Kroeber, S., 2009. Simultaneous multiplane PIV and microphone array measurements on a rod–airfoil configuration. AIAA, 2009–3184. Huang, X., Zhang, X., 2010. Plasma actuators for noise control. International Journal of Aeroacoustics 9 (4), 679–704. Jacob, M.C., Boudet, J., Casalino, J., Michard, M., 2005. A rod–airfoil experiment as benchmark for broadband noise modelling. Journal of Theoretical and Computational Fluid Dynamics 19 (3), 171–196. Jones, M.G., Parrott, T.L., Sutliff, D.L., Hughes, C.E., 2009. Assessment of soft vane and metal foam engine noise reduction concepts. AIAA-2009-3142. Li, Y., Zhang, X., Huang, X., 2010. The use of plasma actuators for bluff body broadband noise control. Experiment in Fluids 49 (2), 367–377. Li, Y., Smith, M., Zhang, X., 2012. Measurement and control of aircraft landing gear broadband noise. Aerospace Science and Technology 23, 213–223. Lorenzoni, V., Moore, P., Scarano, F., Tuinstra, M., 2009. Aeroacoustic analysis of a rod–airfoil flow by means of time-resolve PIV. AIAA-2009-3298. Lorenzoni, V., Violato, D., Scarano, F., 2010. Characterization of noise sources in a rod–airfoil configuration by means of time-resolve Tomographic PIV. AIAA -2010-3898. Magagnato, F., Sorg̈ uven, E., Gabi, M., 2003. Far-field noise prediction by large-eddy simulation and Ffowcs–Williams and Hawkings analogy. AIAA-20033206. Oerlemans, S., Bruin, A., 2009. Reduction of landing gear noise using an air curtain. AIAA-2009-3156. Olsen, W., Wagner, J., 1982. Effect of thickness on airfoil surface noise. AIAA Journal 20 (3), 437–439. Raffel, M., Willert, C., Kompenhans, J., 2000. Particle Image Velocimetry. a Practical Guide, 3rd ed. Springer, Berlin. Sijtsma, P., 2007. CLEAN based on spatial source coherence. AIAA, 2007–3436. Sijtsma, P., 2010. Phased Array Beamforming Applied to Wind Tunnel and Fly-Over Tests. Report no. NLR-TP-2010-549. Siller, H.A., Jacob, M.C., Michel, U., 2005. Flow and noise modification by suction and blowing on a rod–airfoil Configuration. AIAA-2005-3029. Sorguven, E., Magagnato, F., Gabi, M., 2003. Acoustic prediction of a cylinder and airfoil configuration at high Reynolds numbers with LES and FWH. ERCOFTAC Bulletin 58, 47–50. Spiteri, M., Zhang, X., Molin, N., Chow, L.C., 2008. The use of a fairing and split plate for bluff body noise control. AIAA-2008-2817. Takagi, Y., Fujisawa, N., Nakano, T., Nashimoto, A., 2006. Cylinder wake influence on the tonal noise and aerodynamic characteristics of a NACA0018 airfoil. Journal of Sound and Vibration 297, 563–577. You, D., Choi, H., Myung-Ryul, C., Shin-Hyoung, K., 1998. Control of flow-induced noise behind a circular cylinder using splitter plates. AIAA Journal 36 (11), 1961–1967.
Contents lists available at ScienceDirect
Journal of Fluids and Structures journal homepage: www.elsevier.com/locate/jfs
Experimental study of vortex–structure interaction noise radiated from rod–airfoil configurations Yong Li a,b,n, Xun-nian Wang a, Zheng-wu Chen a,b, Zheng-chu Li a,b a b
State Key Laboratory of Aerodynamics, China Aerodynamics Research and Development Centre, Mianyang 621000, China Key Laboratory of Aerodynamic Noise Control, China Aerodynamics Research and Development Centre, Mianyang 621000, China
a r t i c l e i n f o
abstract
Article history: Received 24 July 2013 Accepted 24 August 2014 Available online 27 October 2014
Vortex–structure interaction noise radiated from an airfoil embedded in the wake of a rod is investigated experimentally in an anechoic wind tunnel by means of a phased microphone array for acoustic tests and particle image velocimetry (PIV) for the flow field measurements. The rod–airfoil configuration is varied by changing the rod diameter (D), adjusting the cross-stream position (Y) of the rod and the streamwise gap (L) between the rod and the airfoil leading edge. Two noise control concepts, including “air blowing” on the upstream rod and a soft-vane leading edge on the airfoil, are applied to control the vortex–structure interaction noise. The motivation behind this study is to investigate the effects of the three parameters on the characteristics of the radiated noise and then explore the influences of the noise control concepts. Both the vortex–structure interaction noise and the rod vortex shedding tonal noise are analysed. The acoustic test results show that both the position and magnitude of the dominant noise source of the rod–airfoil model are highly dependent on the parameters considered. In the case where the vortex– structure interaction noise is dominant, the application of the air blowing and the soft vane can effectively attenuate the interaction noise. Flow field measurements suggest that the intensity of the vortex–structure interaction and the flow impingement on the airfoil leading edge are suppressed by the control methods, giving a reduction in noise. & 2014 Elsevier Ltd. All rights reserved.
Keywords: Rod–airfoil model Vortex–structure interaction noise Noise control Vortex shedding Air blowing and soft vane
1. Introduction Vortex–structure interaction noise is a main concern in several aeronautical and industrial applications. Two important devices involving such interaction noise are the rotor configurations of turbo-engine and helicopter rotors, in which the downstream airfoil blades lie in the wake of upstream blades. At any given time, vortices shed from the upstream blades interact with and impinge upon the downstream blades, giving rise to a host of noise and vibration issues. The rod–airfoil configuration consisting of an airfoil located in the near wake of a rod in a flow is believed to be a benchmark well-suited for numerical predictions of such vortex–structure interaction noise and noise generation mechanisms (Casalino et al., 2003; Jacob et al., 2005). Jacob et al. (2005) performed PIV measurement combined with proper orthogonal decomposition (POD) reconstructions to identify the coherent structures in the rod wake and compare the experimental results with the numerical simulations using Reynolds average Navier–Stokes (RANS) and large eddy simulation (LES). These analyses highlighted that strong three dimensional effects were responsible for spectral broadening around the rod vortex shedding n
Corresponding author. E-mail address: [email protected] (Y. Li).
http://dx.doi.org/10.1016/j.jfluidstructs.2014.08.014 0889-9746/& 2014 Elsevier Ltd. All rights reserved.
314
Y. Li et al. / Journal of Fluids and Structures 51 (2014) 313–325
Nomenclature D H C L Y ReD ReC x, y, z
cylindrical rod diameter square rod width airfoil chord length streamwise gap between the rod and the airfoil leading edge cross-stream position of the rod Reynolds number based on rod diameter D Reynolds number based on airfoil chord length C Cartesian coordinates
U1 U V Ma Ωz f fvs St Stvs ΔLd PIV SPL
wind speed streamwise (x) velocity cross-stream (y) velocity Mach number spanwise vorticity ð∂V=∂x ∂U=∂yÞ frequency vortex shedding frequency Strouhal number fD/U1 vortex shedding Strouhal number (fvsD/U1) sound pressure level difference particle imaging velocimetry sound pressure level
frequency in the subcritical regime, and identified that the airfoil leading edge was the main contributor to the noise emission in a rod–airfoil configuration due to vortex–structure interaction. The flow physics of the rod–airfoil model have been further studied by means of experimental methods. In a similar manner to Jacob et al. (2005), Takagi et al. (2006) investigated the influence of the cylinder transverse location on the turbulent development of the flow around the airfoil. Their results showed that the transverse location of the cylinder might have an impact on the radiated acoustic field. In order to identify the flow features that are responsible for the noise generation, Henning et al. (2008, 2009) performed dual PIV and simultaneous far-field microphone measurements. By means of a correlation-technique, they identified the regularities in the near-field fluctuations that are related to the radiated sound field. Lorenzoni et al. (2009, 2010) applied time-resolved PIV combined with a pressure reconstruction procedure to characterise the noise sources. The comparison with simultaneous microphone measurements revealed the ability of this method to predict reasonably well the magnitude of the tonal peak of emission and the narrow band spectrum around it. Giesler and Sarradj (2009) firstly performed microphone array measurements to investigate the influences of the rod diameter and the streamwise gap between the rod and the airfoil leading edge on the broadband noise generation at the frequencies higher than the vortex shedding frequency. Their results showed that the noise generation for lower frequencies depended more strongly on the cylinder diameter than on the streamwise gap. In addition to these experimental studies, further understanding of the details of the rod–airfoil interactions can be gained from numerical simulations. Casalino et al. (2003) developed an aeroacoustic code based on porous Ffows-Williams and Hawking (FW–H) in combination with unsteady RANS simulations for the acoustic prediction. In a similar manner, Magagnato et al. (2003), Sorguven et al. (2003) and Boudet et al. (2005) coupled a FW–H acoustic analogy with LES calculations which allowed capturing of the sub-Karman vortical structures responsible for noise emission at higher frequencies. Greschner et al. (2008) combined a detached eddy simulation (DES) with the FW–H analogy to evaluate far-field pressure spectra. These investigations presented results which were in agreement with the experiments, but the applied numerical methods decouple the aerodynamic and acoustic fields, which makes it difficult to determine the details of the flow physics that are responsible for the noise generation. Recently, Berland et al. (2010, 2011) performed a direct noise calculation (DNC), based on compressible LES, to predict the sound radiation from the rod–airfoil configuration. The DNC method does not require any modelling of the sound sources and computes aerodynamic and acoustic fluctuations within a single run. The good agreement between calculation and experiment demonstrated the promise of the DNC method on noise prediction and generation mechanisms. The dominant noise source on the rod–airfoil model was demonstrated to be caused by the impingement of the Karman vortices upon the airfoil leading edge. Therefore, a modification of the leading edge aerodynamics can be expected to modify both the turbulent flow and the sound emission of the rod–airfoil configuration. Siller et al. (2005) applied a blowing and suction at the airfoil leading edge to modify the flow structure and showed that the peak sound pressure level (SPL) at vortex shedding frequency in the far-field could be significantly reduced by air blowing, while air suction might increase the main peak level. Most of the aforementioned studies were performed on one rod–airfoil configuration with fixed streamwise gap and/or rod diameter on the centreline. However, different gaps and rod diameters may have effects on the generation of the dominant noise. Giesler and Sarradj (2009) demonstrated experimentally the effect of the rod diameter and the streamwise gap on the broadband noise of high frequencies far above the vortex shedding frequency. In addition, Berland et al. (2011) performed DNC on the influence of the streamwise gap. The current paper follows Giesler and Sarradj (2009) by investigating further the effects of these two parameters on the radiated noise, not only at the broadband frequencies but also at the vortex shedding frequency, especially the influence of the gap/diameter ratio. In addition, the effect of the rod cross-stream position has also been investigated. Based on the gained knowledge, two noise control concepts using “air blowing” on the upstream rod and a downstream soft-vane airfoil leading edge are explored here. Acoustic and flow field measurements are taken in an anechoic wind tunnel using a phased microphone array and the PIV technique respectively. The use of the microphone array combining with an advanced data processing of CLEAN algorithm is to accurately localise noise sources and determine the dominant noise at different configurations, whereas the PIV technique reveals the corresponding flow physics of the noise generation and control.
Y. Li et al. / Journal of Fluids and Structures 51 (2014) 313–325
315
2. Experimental details 2.1. Test model and wind tunnel set-up A sketch of the rod–airfoil test model and the set-up is shown in Fig. 1. A symmetric NACA0012 airfoil with chord length of C ¼150 mm is located downstream of a rod, both extending by 400 mm in the spanwise z-direction. The airfoil chord is aligned with the centre of the nozzle and parallel to the streamwise x-direction. The origin of the Cartesian coordinate system is located at the mid-span position of the airfoil leading edge which is located 300 mm downstream of the nozzle exit. Eight cylindrical rods with diameters of D¼3 mm, 5 mm, 8 mm, 11 mm, 15 mm, 18 mm, 21 mm and 25 mm are used, and also a H ¼15 mm square rod. The streamwise gap between the rod and the airfoil leading edge is adjusted between L ¼30 mm and 170 mm at 10 mm intervals and the cross-stream position of the rod varies between Y ¼0 mm (i.e. on the centreline) and 40 mm at 5 mm intervals. Measurements are performed at wind speeds of U1 ¼30 m/s and 60 m/s with incoming turbulence level of 0.5% at the centreline, yielding the corresponding Reynolds number ReD between 6.3 103 and 1.06 105 based on the rod diameter D, and ReC between 4.5 105 and 9.0 105 based on the airfoil chord length C. Fig. 2 shows the photographs of the test set-up that is implemented in the anechoic wind tunnel of China Aerodynamics Research and Development Centre. The wind tunnel is an open-jet closed-circuit test facility with a rectangular 0.55 m 0.4 m nozzle exit. The anechoic chamber dimension is 5.5 m 5 m 4 m and its six walls are covered by 0.5 m long fibreglass wedges, yielding 99% acoustic absorption above 200 Hz. The rod–airfoil model is vertically installed between two metal endplates that are directly attached to the nozzle exit. The model is placed into the potential core of the nozzle jet. The lower endplate has an insertion made of Plexi-glass to ensure optical access for PIV recording. The model and the endplates are rigidly fixed to ensure no vibration during all the tests. 2.2. Measurement techniques The noise level and location are measured by a 0.72 m-diameter ring microphone array which consists of 32 omnidirectional GRAS type 40PH 1/4 in. microphones with a frequency response range of 4 Hz–70 kHz (Fig. 2a). All the microphones are covered by wind shields to eliminate the wind effects. A camera mounted in the centre of the array is used to capture the model image. The array scan plane is aligned with the airfoil chord and 0.75 m away from the surface of the microphone array.
Fig. 1. Sketch of the rod–airfoil configuration and the test set-up (not to scale).
Fig. 2. Experimental set-up in the anechoic wind tunnel: (a) close-up view showing the rod–airfoil model between two endplates and the microphone array; and (b) PIV measurement set-up.
316
Y. Li et al. / Journal of Fluids and Structures 51 (2014) 313–325
The acoustic data are captured using National Instruments data acquisition cards at a sampling rate of 48 kHz and analysed using a block size of 4096, yielding a frequency resolution of about 12 Hz. To reduce spectral leakage, a Hanning window function is applied to each single block before performing fast Fourier transform. A total of 100 blocks are averaged for statistical confidence and the uncertainty in the noise level is determined to be within 2%. The data are processed by the advanced CLEAN–SC algorithm based on spatial coherence sources (Sijtsma, 2007). This algorithm is chosen following a classical beamforming technique (Dougherty, 2002) to improve the spatial resolution and reduce the sidelobe level. With this processing scheme and the measurement set-up, the spatial resolution of the current microphone array is about 0.41λ (λ: acoustic wavelength). The result of the post-processing is a two dimensional map showing the noise source distribution and the sound pressure level, in the following referred to as noise source map. To remove the effects of the convective flow and the free shear layer on the noise positioning the noise maps are corrected using the average-Mach number (Ma) method of Sijtsma (2010), i.e., the uniform flow Ma is replaced by the average Macor between the test model and the microphone array. The corrected flow Macor is given by
ζ ξ
M acor ¼ M a ð1Þ
Y ξ ; mic where ζ, ξ and Ymic are the model position, shear layer position and microphone array position, respectively. As a result, the uncertainty in noise source positioning is determined to be within 5%. To investigate the sound pressure spectra in the whole frequency range, the results are obtained by averaging the data from all the 32 channel microphones. The flow velocity data are acquired using a planar PIV system of Beamtech Vlite500 with a New Wave Gemini Nd: YAG dual laser. The PIV system operates at a repetition rate of 2 Hz. A special ES11000 4008 2672 high resolution CCD camera used to record a two-dimensional view of the flow field is set up perpendicular to the laser sheet on the scan plane and fixed on a support structure underneath the endplates (Fig. 2b). The flow is seeded with diethylhexylsebacate (DEHS) tracer particles with mean particle diameter of approximately 1 μm. The laser light sheet thickness at the model area is about 1 mm. The flow field area of interest is confined to a small window depicted in Fig. 1 and marked as thick dashed line. A total of 250 images of the flow field are taken for each experimental condition. Each image set is processed using a 32 32 pixels cross-correlation area and a 50% 50% overlap to improve the resolution of the vector map. The accuracy of the instantaneous velocity fields can be estimated by assuming an accuracy in the correlation of 0.1 pixel displacement (Raffel et al., 2000), which corresponds to a maximum error in the velocity of 0.4 m/s. The uncertainty in the instantaneous velocities is estimated to be about 2% for the present set up, while the corresponding spanwise vorticity Ωz has an uncertainty of about 5%.
2.3. Noise control concepts Two noise control concepts are developed, i.e., the air blowing and the soft vane leading edge as presented in Fig. 3. The air blowing rod (Fig. 3a) has 300 mm (length) 3 mm (width) slots on the sides of a square rod from which the air blows out in the cross-stream y-direction. The width of the air blowing rod is H¼ 15 mm. A square rod is chosen, rather than a circular one, only because it is relatively easy to construct the air blowing slots. The size of the upstream air entrance to the blowing slots is 300 mm (length) 9 mm (width). Unlike air blowing devices used in other studies, where an external air pump may have to be used, the current design allows the incoming main flow to directly provide the blowing air. Therefore, no external mass flow is introduced into the main flow. In such a way, the air-blowing speed and its flow momentum may be highly dependent on the main flow speed (U1) and the slot size. The idea behind this control concept is to reduce the flow impingement speed and push away the rod shed vortices from the airfoil leading edge by providing an “air curtain” in front of the airfoil. The concept of “air curtain” was previously introduced by Oerlemans and Bruin (2009) on landing gear main strut noise reduction using external air blowing and by Huang and Zhang (2010) on noise control using plasma actuators. The second concept is a soft vane, shown in Fig. 3b, wherein the airfoil leading edge is firstly made slotted and then covered with a porous woven cloth approximately 0.2 mm-thick in this case. The leading edge slot is 20 mm (depth) 2 mm (width). This soft vane concept was investigated by Jones et al. (2009) on engine noise reduction. As indicated by Jones et al. the slotted leading edge may allow communication of pressure fluctuations on both sides of the surface and the chamber formed within the cloth should provide pressure release (relative to the solid surface it replaced) on the leading edge surface.
3. Results and discussion This section summarises the results of several tests using the aforementioned microphone array and PIV technique to identify the governing mechanisms that lead to the noise generation and reduction. The parametric study (including the effects of D, L and Y) and the assessment of the soft vane are made using the cylindrical rod–airfoil configuration, while the assessment of the air blowing is made by comparing the air-blowing rod with a solid square rod of same size. Table 1 lists these test configurations and their characteristics.
Y. Li et al. / Journal of Fluids and Structures 51 (2014) 313–325
317
Fig. 3. Schematic of the noise control concepts (not to scale): (a) air blowing rod; and (b) soft-vane airfoil leading edge (slotted leading edge covered by 0.2 mm-thick cloth). Table 1 Overview of the test cases conducted on the rod–airfoil model. C-rod: cylindrical rod; S-rod: square rod. Tests
Rod type
D or H (mm)
L [Step] (mm)
Y [Step] (mm)
U1 (m/s)
Effects of D Effects of L Influences of L/D Effects of Y Air blowing Soft vane
C-rod C-rod C-rod C-rod S-rod C-rod
3–25 15 3–25 15 15 15
150 30–170 [10] 30–170 [10] 150 150 150
0 0 0 0–40 [5] 0 0
30, 30, 30, 30, 30, 30,
60 60 60 60 60 60
3.1. Acoustic measurements 3.1.1. Effects of rod diameter (D) To study the effects of the rod diameter, the cylindrical rod is first located on the centreline and 150 mm upstream of the airfoil leading edge. As expected, it is found from sound spectra (not shown here) that the dominant frequency is close to each rod vortex shedding frequency fvs and the corresponding Strouhal number Stvs is about 0.2. The noise source maps at fvs shown in Fig. 4 indicate that both location and level of the dominant noise vary with changing the rod diameter. For the small rods of 3 mm and 8 mm diameter (Fig. 4a and b), it can be seen clearly that the dominant noises at fvs ¼3.9 kHz and 1.5 kHz are concentrated on the upstream rods and the peak levels are 82 dB and 90 dB respectively. The array dynamic range is 10 dB in both source maps. This suggests that the radiated noise from the test model with a small rod is generated mainly by the upstream rod due to the unsteady loads induced by the vortex shedding. In the case of the 15 mm-diameter rod (Fig. 4c), however, the noise map shows that the mainlobe centre for fvs ¼760 Hz is located at the downstream airfoil leading edge, indicating that the noise is now radiated largely from the vortex–structure interaction on the surface of the airfoil leading edge. The dynamic range is reduced to 7 dB in Fig. 4c to show more clearly the noise source position at relatively low frequency. To see how the airfoil installed changes the level and location of the main source, a source map for the 15 mm-diameter rod-only is also given in Fig. 4d. It can be seen obviously from Fig. 4c and d that the peak level is
318
Y. Li et al. / Journal of Fluids and Structures 51 (2014) 313–325
Fig. 4. Noise source maps at the vortex shedding frequency (fvs) for different diameter cylindrical rods and airfoil. Flow (U1 ¼60 m/s) is from left to right: (a) D¼ 3 mm, fvs ¼ 3.9 kHz (b) D ¼8 mm, fvs ¼1.5 kHz, (c) D ¼15 mm, fvs ¼ 760 Hz and (d) D ¼ 15 mm, rod-only, fvs ¼760 Hz.
increased by 12 dB and the dominant noise position moves downstream to the airfoil leading edge once the airfoil is installed. The source location difference of 160 mm between the two maps is slightly less than the array spatial resolution of 0.41λ ¼180 mm at 760 Hz. As the rod diameter is increased further, the current microphone array is unable to distinguish the position of the vortex shedding noise from the vortex–structure interaction noise due to the poor array spatial resolution at lower frequencies. For instance, the array spatial resolution at fvs ¼490 Hz for the 21 mm-diameter rod is about 280 mm which is far more than the streamwise distance between the rod and the airfoil, making it difficult to locate precisely the noise source. The peak sound pressure level (SPL) difference ΔLd at fvs (obtained in 1/3 octave band) between the rod–airfoil and the rod-only for all the eight rods are shown in Fig. 5. For Dr8 mm, the rod–airfoil has almost the same SPL as the rod-only, suggesting that there is almost no interaction noise. As the rod diameter is increased, the SPL difference grows. At DZ15 mm, the interaction noise at the airfoil leading edge increases by about 12 dB compared to that of the rod-only, which is in good agreement with the finding of Jacob et al. (2005). These results indicate that the intensity of vortex–structure interaction at the airfoil leading edge is reduced as the rod diameter is decreased. This is somewhat equivalent to noise reduction when the airfoil thickness is increased. Olsen and Wagner (1982), Gershfeld (2004), Giesler and Sarradj (2009) found that for low Mach number flows a bigger and/or blunter shape of the airfoil leading edge is generally responsible for a lower leading edge interaction noise. This can be explained by the rapid distortion theory (Goldstein and Atassi, 1976) that the small rod shed vortices are less deformed when passing through the leading edge and their potential field is less extended. As a result, the pressure fluctuation on the airfoil that is responsible for the noise generation is smaller. It also can be seen that the behaviour of the noise change is highly similar at the two test wind speeds of U1 ¼30 m/s and 60 m/s.
3.1.2. Effects of streamwise gap (L) The effects of the streamwise gap are assessed for the case of 15 mm-diameter rod. The gap is within the range of 30–170 mm. It has previously been demonstrated in Fig. 4c that with the rod diameter of 15 mm the dominant noise is radiated from the airfoil leading edge due to the vortex–structure interaction. However, Fig. 6 shows that both the vortex shedding Strouhal number Stvs and the corresponding peak SPL difference ΔLd vary with the change of the gap.
Y. Li et al. / Journal of Fluids and Structures 51 (2014) 313–325
319
Fig. 5. Peak SPL differences between the rod–airfoil model and the rod-only versus the rod diameter (D) at the two wind speeds. L ¼ 150 mm.
Fig. 6. Vortex shedding Strouhal number Stvs (a) and corresponding peak SPL difference (b) versus the streamwise gap (L). U1 ¼ 60 m/s, D¼ 15 mm.
The Strouhal number plotted in Fig. 6a shows that Stvs is 0.192 for the rod-only, and that for the rod–airfoil model Stvs gradually decreases from 0.19 at L ¼170 mm down to 0.163 at L¼50 mm. This phenomenon, as mentioned by Jacob et al. (2005), may be due to the modification of the vortex shedding through its interaction with the airfoil, suggesting a feedback of the airfoil onto the rod. With a further decrease in the gap to 40 mm, Stvs increases again to reach the value of 0.183. Concerning the peak SPL difference at Stvs shown in Fig. 6b, the noise increases as the gap is narrowed from 170 mm to 50 mm. At L¼50 mm, the noise reaches the highest value which is about 15 dB higher than the noise for the rod-only. At L ¼40 mm, the noise drops dramatically by about 32 dB from the highest value at L ¼50 mm, and is even 17 dB lower than that of the rod-only. This sharp decrease is likely due to the airfoil in the vicinity of the rod preventing the formation of Karman vortices.
3.1.3. Influences of gap/diameter ratio (L/D) It is useful to understand the noise characteristics with regards to appropriate dimensionless quantities of the gap/ diameter ratio (L/D). Fig. 7 shows the overall SPL differences between the rod–airfoil and the rod-only at the different ratios. Due to the finite gap range of L ¼30–170 mm, the L/D radio is relatively low for big rods and high for small rods. Nevertheless, as already indicated in Fig. 6b, it can be seen in Fig. 7 that there is a critical value of L/D ¼3.3 below which the overall SPL undergoes a sharp decrease of more than 30 dB. This critical ratio is in close agreement with the numerical finding of Berland et al. (2011) where the ratio is 3.5 in the case of one diameter rod. This phenomenon suggests that at this situation the embedded airfoil may work as a splitter plate which prevents the formation of the rod Karman vortices (Spiteri et al., 2008; You et al., 1998), contributing to the drastic noise reduction. As the L/D ratio is increased, the noise level gradually decreases. It would be expected that at L/DZ20 the noise of the rod–airfoil almost collapses to that of the rod-only, indicating that at larger L/D the vortex shedding tonal noise is dominant and only weak interaction happens at the airfoil leading edge. In such a case, to reduce the noise the focus should be concentrated on the upstream vortex suppression.
320
Y. Li et al. / Journal of Fluids and Structures 51 (2014) 313–325
Fig. 7. Overall SPL differences between the rod–airfoil and the rod-only against the gap/diameter ratio L/D. U1 ¼ 60 m/s.
Fig. 8. Sound pressure spectra (a) and overall SPL (b) at different cross-stream positions (Y) of the rod. U1 ¼60 m/s, D ¼15 mm, L ¼150 mm.
Fig. 9. Sound pressure spectra versus Strouhal number for the air blowing and the square rod. H¼ 15 mm and L ¼ 150 mm.
3.1.4. Effects of rod cross-stream position (Y) To study the effects of the rod cross-stream position, the 15 mm-diameter rod is located 150 mm upstream of the airfoil leading edge. Fig. 8 depicts a set of sound spectra and the peak SPLs at the different cross-stream positions. The sound spectra for the rod-only and the airfoil-only are also plotted in Fig. 8a. Compared with the other cases, the noise generated by the airfoil-only is low and can thus be ignored. For the rod-only, one can observe discrete frequency noise consisting of pure tones at fundamental Strouhal number (Stvs) and its harmonics. When the airfoil is embedded directly in the wake of the rod, i.e., at Y¼0 mm, the peak SPL at Stvs is increased by about 12 dB and the broadband noise around is also increased.
Y. Li et al. / Journal of Fluids and Structures 51 (2014) 313–325
321
Fig. 10. Noise source map comparison showing the noise reduction at vortex shedding frequency (fvs) using slotted airfoil leading edge and soft-vane airfoil leading edge respectively: (a) solid airfoil leading edge; (b) slotted airfoil leading edge; and (c) soft vane airfoil leading edge. Flow (U1 ¼ 60 m/s) is from left to right, D¼ 15 mm, L ¼ 150 mm.
The impingement of these vortical structures upon the airfoil leading edge generates sound sources, which is similar to the discrete frequency tones and broadband noise observed in turbomachinery: rotor blades undergo unsteady pressure fluctuations and one may observe discrete frequency radiation consisting of pure tones at harmonics of the blade passing frequency and broadband noise produced by the interaction of the blade surfaces with the turbulent wakes. The overall SPL gradually decreases as the cross-stream position (Y) increases, as shown in Fig. 8b. This noise reduction is due to some of the shed vortices missing the airfoil leading edge, thus reducing the intensity of the vortex–structure interaction. At the Y¼ 40 mm position, the noise level collapses to the same value of the rod-only, indicating that the rod shed vortices may totally miss the airfoil leading edge and there is almost no vortex–structure interaction. Thus, any method that can push away the shed vortices from the airfoil leading edge would have potential in reducing the interaction noise. 3.1.5. Assessment of noise control concepts This section aims to investigate the benefits of the air blowing rod and the soft vane on the vortex–structure interaction noise. For assessment of the air blowing rod, comparison is made between the air blowing rod and a square rod with same size. The rods are positioned on the centreline and 150 mm upstream of the airfoil leading edge. Fig. 9 shows the sound spectra comparison at the wind speeds of U1 ¼30 m/s and 60 m/s respectively. At the small Strouhal numbers below Stvs ¼0.19 there is a slight discrepancy on the spectra, suggesting that the air blowing does not affect the low frequency noise. At U1 ¼ 30 m/s, the peak SPL at Stvs is reduced by about 4 dB, but at U1 ¼60 m/s there is not much difference. At the Strouhal numbers between 0.19 and 1.0, the broadband noise is clearly reduced at the two wind speeds. The large noise increase at the higher Strouhal numbers (St41) are due to the self-noise of the air blowing. The overall SPL in the whole frequency range is reduced by about 3.8 dB and 1.8 dB at U1 ¼30 m/s and 60 m/s respectively. The noise reduction in the broadband frequencies suggests that the air blowing rod may act as an “air curtain” in front of the airfoil to reduce the impingement speed and the vortex–structure interaction. The effects of the “air curtain” concept on noise reduction were previously investigated by Li et al. (2010) on bluff-body using plasma technique and Oerlemans and Bruin (2009) on simplified landing gear main strut using upstream air blowing. In those studies, however, additional power was needed to generate plasma and air blowing, for example high voltage device for plasma and air pump respectively. They showed that
322
Y. Li et al. / Journal of Fluids and Structures 51 (2014) 313–325
the plasma and the air blowing reduced the impingement speed upon the downstream components, resulting in significant broadband noise reduction. In this paper, the blowing air is directly supplied from the incoming main flow, making the control device simple. However, the reduced performance at U1 ¼60 m/s suggests that this air blowing control concept gradually loses its control authority as the main flow speed increases. The effects of the soft vane are clearly demonstrated on the noise source maps shown in Fig. 10. The 15 mm-diameter rod is positioned 150 mm upstream of the airfoil leading edge. The noise source maps at vortex shedding frequency for the solid, slotted and soft vane airfoil leading edge are presented. Compared with the solid leading edge, the slotted leading edge gives a noise reduction of about 4 dB, whereas an improved noise reduction of 6 dB can be obtained with the soft vane by covering the slotted leading edge with 0.2 mm-thick cloth. A similar behaviour using the soft vane concept was also reported by Jones et al. (2009) for the experimental study of the engine fan noise reduction. The corresponding sound pressure spectra are shown in Fig. 11. It can be seen that, in addition to the noise reduction at Stvs, both the slotted leading edge and the soft vane leading edge also significantly suppress the broadband noise at the Strouhal numbers above Stvs. The low-frequency parts of the spectra for the cases with the slotted leading edge and the solid leading edge are similar, whereas the spectrum for the soft vane exhibits high levels with one peak centred around St¼0.055. The noise reduction by the slotted leading edge is most likely due to the direct pressure-fluctuation communication at the airfoil leading-edge surfaces that reduces the vortex–structure interaction area. The pressure-release surface provided by the soft vane is believed to reduce noise radiation efficiency. To improve the noise reduction, both the internal structure and the porosity of chamber in the soft vane may have to be optimised to achieve maximum sound absorption.
3.2. Flow field measurements Presenting PIV results from all the measured rod–airfoil configurations goes beyond the scope of this section. Therefore, only the results for representative rod–airfoil configurations measured at the wind speed of U1 ¼30 m/s are presented. Fig. 12 illustrates the contours of instantaneous spanwise vorticity Ωz at three different rod–airfoil configurations, i.e., at different streamwise gaps and cross-stream position. When the cylindrical rod of D ¼15 mm is positioned 150 mm directly upstream of the airfoil (Fig. 12a), the large scale vortices can be seen to emanate from the upper and lower shear layers. The shed vortices move downstream and gradually lose intensity and coherence. In the region around the airfoil leading edge,
Fig. 11. Comparison of sound pressure spectra between the three different airfoil leading edges.
Fig. 12. Contour plots of instantaneous spanwise vorticity Ωz (mm s 1) in the rod (D¼ 15 mm) wake region under three rod positions: (a) x¼ 150 mm, y¼ 0; (b) x ¼ 150 mm, y¼ 40 mm; and (c) x ¼ 40 mm, y¼0.
Y. Li et al. / Journal of Fluids and Structures 51 (2014) 313–325
323
it is hard to recognise the vortex structure. The observations are consistent with the results obtained by Jacob et al. (2005), Henning et al. (2009) and Lorenzoni et al. (2009). The downstream convective vortices directly impinge upon the airfoil leading edge, giving rise to high vortex–structure interaction noise. Although in this configuration the strongest vorticity is seen to be located in the area close to the rod, the dominant noise has been demonstrated by the microphone array to be the interaction noise at the airfoil leading edge. At the cross-stream position of Y¼ 40 mm (Fig. 12b), the shed vortices pass over the upper side of the airfoil surface and totally miss the airfoil leading edge. As a result, there is no interaction noise and the dominant noise is shown to be radiated from the rod shed vortices, and the noise is almost the same as that of the rod-only. Moving the rod close to the airfoil leading edge at the streamwise gap of L¼ 40 mm in Fig. 12c, no visible vortices between the rod and the airfoil can be observed, suggesting that the presence of the airfoil in the vicinity of the rod prevents the formation of the Karman vortices. In this situation the noise undergoes a sharp decrease, as already indicated in Fig. 6b. Fig. 13 presents the instantaneous vorticity Ωz for the rod–airfoil model at three different rod diameters. In the case of 3 mm-diameter rod (Fig. 13a), there is a well organised vortex street in the vicinity of the rod up to 50 mm downstream, indicating that the airfoil has no effect on the upstream rod shed vortices. With increasing rod diameter, the size of the shed vortices increases and due to the feedback effect of the airfoil it is hard to recognise the vortex structure before the airfoil leading edge in the case of 21 mm-diameter rod (Fig. 13c). The effect of the air blowing on the flow field has also been illustrated by PIV tests. To quantify the differences in the downstream wake, the streamwise U-velocity profiles are obtained at two x positions (x¼ 100 mm and 10 mm) upstream of the airfoil leading edge. The results comparing the square rod and the air-blowing rod are shown in Fig. 14. At the streamwise location of x¼ 100 mm, the local velocity in the central area is significantly reduced by the air blowing on both sides of the square rod. The velocity on the centreline is reduced by about 15 m/s. At the far downstream position of x ¼ 10 mm, i.e., at 10 mm upstream of the airfoil leading edge, the wake width is increased and the flow speed is reduced by about 5 m/s on the centreline. The reduced velocity can alleviate the flow impingement on the airfoil leading edge, hence attenuates the vortex–structure interaction. In such a way, the air blowing forms an “air curtain” in front of the airfoil, which
Fig. 13. Contour plots of instantaneous spanwise vorticity Ωz (mm s 1) in the rod wake region under three rod diameters: (a) D ¼3 mm; (b) D ¼8 mm; and (c) D ¼ 21 mm.
Fig. 14. Comparison of the streamwise U-velocity profiles between the square rod and the air blowing rod at two x positions: (a) x¼ 100 mm; and (b) x¼ 10 mm.
324
Y. Li et al. / Journal of Fluids and Structures 51 (2014) 313–325
has a similar effect as the passive control fairing used to reduce the landing gear noise (Dobrzynski et al., 2002; Li et al., 2012) by reducing the flow speed impinging upon the downstream components. Fig. 15 illustrates the instantaneous vorticity Ωz for the square rod and the air-blowing rod respectively. For the square rod (Fig. 15a), most of the vortices concentrate on the central area within 720 mm and impinge directly upon the airfoil leading edge. With the air blowing on both sides (Fig. 15b), however, a series of scattered vortices are observed. Some of the vortices reach as high as 40 mm and then pass over the airfoil leading edge because the air blowing pushes them away from the airfoil, contributing to the interaction noise reduction. As far as the slotted leading edge and the soft vane are concerned, the spanwise vorticity Ωz in the wake of the rod shown in Fig. 16 does not have much difference with that of solid leading edge, at least in the area of interest by the PIV view in this paper. The phenomena can be understood since the noise reduction should mainly be due to the relief of the pressure fluctuations on the surface of the airfoil leading edge, not the change of the upstream shed vortices. 4. Conclusions This paper investigates the effects of different rod–airfoil configurations on the vortex–structure interaction noise and addresses the techniques of upstream air blowing and a soft-vane airfoil leading edge as methods for reducing interaction noise. Acoustic and aerodynamic measurements are taken on a representative rod–airfoil model at two wind speeds with varies of the model configuration by changing the rod diameter (D), adjusting the streamwise gap (L) between the rod and the airfoil leading edge, and the rod cross-stream position (Y). The sound pressure level and position of the dominant noise are highly dependent on the three parameters. Generally speaking, the radiated noise increases with increasing rod diameter and/or shortening the streamwise gap due to the strong vortex–structure interaction at the airfoil leading edge. There exists a critical gap/diameter ratio of L/D ¼3.3 below which the noise level undergoes a sharp decrease of more than 30 dB due to the suppression of the rod's Karman vortices by the downstream airfoil. At about L/DZ20, there is almost no vortex–structure interaction and the noise is mainly radiated from the upstream rod vortex shedding. In this case, attention should be paid to suppress the upstream vortices to reduce the noise. However, once the vortex–structure interaction is dominant, control strategies should concentrate on suppressing both the vortices and the interaction. The noise of the rod–airfoil model gradually decreases with increasing the rod cross-stream position (Y), i.e., with moving the rod out of alignment with the airfoil. This noise decrease is due to the reduced vortices impinging upon the
Fig. 15. Comparison of the instantaneous spanwise vorticity Ωz (mm s 1) between (a) the square rod and (b) the air blowing rod.
Fig. 16. Contour plots of instantaneous spanwise vorticity Ωz (mm s 1) in the rod wake region with different airfoil leading edge: (a) solid airfoil leading edge; (b) slotted airfoil leading edge; and (c) soft-vane airfoil leading edge.
Y. Li et al. / Journal of Fluids and Structures 51 (2014) 313–325
325
airfoil leading edge. Once the shed vortices totally miss the airfoil leading edge at larger Y positions, the noise of the rod– airfoil model drops to that of the rod-only. Thus, any method that can prevent the vortices impinging upon the airfoil leading edge would have a potential on noise reduction. The noise control results demonstrate that the vortex–structure interaction noise can be attenuated by air blowing out of the sides of the upstream rod and by the soft-vane airfoil leading edge. The air blowing shows an effective noise reduction at U1 ¼30 m/s since it reduces the flow impingement speed and vortices impinging upon the airfoil leading edge. In such the case, the air blowing seems to create an “air curtain” in front of the airfoil that partly shields the airfoil from the incoming flow. At U1 ¼60 m/s, however, the air blowing loses it control authority, which puts a limitation on practical applications. As the applied blowing air is directly from the main flow, how to increase the air blowing speed and the flow momentum would be a challenge and needs further fundamental study. The soft vane airfoil leading edge which is formed by covering the slotted leading edge with a porous cloth shows significant noise attenuation at the two wind speeds. The control mechanism is the pressure release at the leading edge surface of the soft vane which reduces the transformed noise energy from the interaction. It would be expected that more noise reduction may be obtained if the internal structure and the porosity of the soft vane undergo an optimisation process. Acknowledgement The experiments are made possible through the support provided by State Key Laboratory of Aerodynamics in China (Project no. 4071506). The authors like to thank Dr. Malcolm Smith from Institute of Sound and Vibration Research (ISVR) at University of Southampton for helpful discussion and comments on the manuscript. References Berland, J., Lafon, P., Crouzet, F., Daude, F., Bailly, C., 2010. Numerical insight into sound sources of a rod–airfoil flow configuration using direct noise calculation. AIAA-2010-3705. Berland, J., Lafon, P., Crouzet, F., Daude, F., Bailly, C., 2011. A Parametric study of the noise radiated by the flow around multiple bodies: direct noise computation of the influence of the separating distance in rod–airfoil flow configurations. AIAA-2011-2819. Boudet, J., Grosjean, N., Jacob, M.C., 2005. Wake–airfoil interaction as broadband noise source: a large-eddy simulation study. International Journal of Aeroacoustics 4 (1), 93–115. Casalino, D., Jacob, M.C., Roger, M., 2003. Prediction of rod–airfoil interaction noise using the Ffowcs-Williams and Hawkings analogy. AIAA Journal 41 (2), 182–191. Dobrzynski, W., Chow, L.C., Guion, P., Schiells, D., 2002. Research into landing gear airframe noise reduction. AIAA-2002-2409. Dougherty, R.P., 2002. Beamforming in acoustic testing. In: Mueller, TJ (Ed.), Aeroacoustic Measurements, Springer, Berlin. Gershfeld, J., 2004. Leading edge noise from thick airfoils in turbulent flows. Journal of the Acoustical Society of America 116, 1416. Giesler, J., Sarradj, E., 2009. Measurement of broadband noise generation on rod–airfoil-configurations. AIAA-2009-3308. Goldstein, M.E., Atassi, H.M., 1976. A complete second order theory for the unsteady flow about an airfoil due to a periodic gust. Journal of Fluid Mechanics 74, 741–765. Greschner, B., Thiele, F., Jacob, M.C., Casalino, D., 2008. Prediction of sound generated by a rod–airfoil configuration using EASM DES and the generalised Lighthill/FW–H analogy. Computers & Fluids 37 (4), 402–413. Henning, A., Kaepernick, K., Ehrenfried, K., Koop, L., Dillman, A., 2008. Investigation of aeroacoustic noise generation by simultaneous particle image velocimetry and microphone measurements. Experiments in Fluids 45, 1073–1085. Henning, A., Koop, L., Ehrenfried, K., Lauterbach, A., Kroeber, S., 2009. Simultaneous multiplane PIV and microphone array measurements on a rod–airfoil configuration. AIAA, 2009–3184. Huang, X., Zhang, X., 2010. Plasma actuators for noise control. International Journal of Aeroacoustics 9 (4), 679–704. Jacob, M.C., Boudet, J., Casalino, J., Michard, M., 2005. A rod–airfoil experiment as benchmark for broadband noise modelling. Journal of Theoretical and Computational Fluid Dynamics 19 (3), 171–196. Jones, M.G., Parrott, T.L., Sutliff, D.L., Hughes, C.E., 2009. Assessment of soft vane and metal foam engine noise reduction concepts. AIAA-2009-3142. Li, Y., Zhang, X., Huang, X., 2010. The use of plasma actuators for bluff body broadband noise control. Experiment in Fluids 49 (2), 367–377. Li, Y., Smith, M., Zhang, X., 2012. Measurement and control of aircraft landing gear broadband noise. Aerospace Science and Technology 23, 213–223. Lorenzoni, V., Moore, P., Scarano, F., Tuinstra, M., 2009. Aeroacoustic analysis of a rod–airfoil flow by means of time-resolve PIV. AIAA-2009-3298. Lorenzoni, V., Violato, D., Scarano, F., 2010. Characterization of noise sources in a rod–airfoil configuration by means of time-resolve Tomographic PIV. AIAA -2010-3898. Magagnato, F., Sorg̈ uven, E., Gabi, M., 2003. Far-field noise prediction by large-eddy simulation and Ffowcs–Williams and Hawkings analogy. AIAA-20033206. Oerlemans, S., Bruin, A., 2009. Reduction of landing gear noise using an air curtain. AIAA-2009-3156. Olsen, W., Wagner, J., 1982. Effect of thickness on airfoil surface noise. AIAA Journal 20 (3), 437–439. Raffel, M., Willert, C., Kompenhans, J., 2000. Particle Image Velocimetry. a Practical Guide, 3rd ed. Springer, Berlin. Sijtsma, P., 2007. CLEAN based on spatial source coherence. AIAA, 2007–3436. Sijtsma, P., 2010. Phased Array Beamforming Applied to Wind Tunnel and Fly-Over Tests. Report no. NLR-TP-2010-549. Siller, H.A., Jacob, M.C., Michel, U., 2005. Flow and noise modification by suction and blowing on a rod–airfoil Configuration. AIAA-2005-3029. Sorguven, E., Magagnato, F., Gabi, M., 2003. Acoustic prediction of a cylinder and airfoil configuration at high Reynolds numbers with LES and FWH. ERCOFTAC Bulletin 58, 47–50. Spiteri, M., Zhang, X., Molin, N., Chow, L.C., 2008. The use of a fairing and split plate for bluff body noise control. AIAA-2008-2817. Takagi, Y., Fujisawa, N., Nakano, T., Nashimoto, A., 2006. Cylinder wake influence on the tonal noise and aerodynamic characteristics of a NACA0018 airfoil. Journal of Sound and Vibration 297, 563–577. You, D., Choi, H., Myung-Ryul, C., Shin-Hyoung, K., 1998. Control of flow-induced noise behind a circular cylinder using splitter plates. AIAA Journal 36 (11), 1961–1967.