ACOUSTIC PROPERTIES OF THE VIRGINIA TECH STABILITY WIND TUNNEL

Transcription

1 ACOUSTIC PROPERTIES OF THE VIRGINIA TECH STABILITY WIND TUNNEL December 6th, 1999 Jon Vegard Larssen and William J. Devenport Department of Aerospace and Ocean Engineering Virginia Polytechnic Institute and State University Blacksburg, Virginia DEPARTMENTAL REPORT: VPI-AOE-263

2 ABSTRACT This report investigates some of the acoustic properties of the Virginia Tech Stability Wind Tunnel. We have documented that acoustic noise in the facility increases predictably with free stream velocity. The effect of test section configuration has been determined quantitatively, and we have shown that using solid panels in the test-section was a quieter option than the slotted panels. Additionally, noise levels are weak functions of streamwise position in the test section. The noise levels increase moving downstream in the test-section, particularly with slotted walls. i

3 TABLE OF CONTENTS Abstract Table of Contents List of Tables List of Figures Nomenclature i ii iii iii iv 1. Objective Facility Equipment Procedure Data Discussion Solid Wall Test-Section Partly Slotted Test-Section Walls Fully Slotted Test-Section Walls Summary Further Discussion Wind-Induced Microphone Noise Microphone Alignment Further Analysis Conclusions References ii

4 LIST OF TABLES 1) Microphone locations 2) Details for run 2a: solid walls; microphone: downstream on the floor 3) Details for run 2b: solid walls; microphone: upstream on the floor 4) Details for run 2d: solid walls; microphone: upstream in the ceiling 5) Details for run 1a: partly slotted walls; microphone: downstream on the floor 6) Details for run 1b: partly slotted walls; microphone: upstream on the floor 7) Details for run 3a: fully slotted walls; microphone: downstream on the floor 8) Details for run 3b: fully slotted walls; microphone: upstream on the floor LIST OF FIGURES 1) Schematic of the Virginia Tech Stability Wind Tunnel 2) Experimental setup 3) Dimensioned sketch of microphone mount 4) Microphone and mount 5) Frequency response calibration chart for B&K condenser microphone type ) Stability Tunnel coordinate system 7) Solid wall configuration 8) Fully slotted configuration 9) Pressure Spectrum Level vs. ƒ; solid walls; downstream; floor; lower Re 10) Pressure Spectrum Level vs. ƒ; solid walls; downstream; floor; higher Re 11) Pressure Spectrum Level vs. ƒ; solid walls; upstream; floor; lower Re 12) Pressure Spectrum Level vs. ƒ; solid walls; upstream; floor; higher Re 13) Pressure Spectrum Level vs. ƒ; solid walls; upstream; ceiling; lower Re 14) Pressure Spectrum Level vs. ƒ; solid walls; upstream; ceiling; higher Re 15) Pressure Spectrum Level vs. ƒ; partly slotted walls; downstream; floor; lower Re 16) Pressure Spectrum Level vs. ƒ; partly slotted walls; upstream; floor; lower Re 17) Pressure Spectrum Level vs. ƒ; fully slotted walls; downstream; floor; lower Re 18) Pressure Spectrum Level vs. ƒ; fully slotted walls; upstream; floor; lower Re 19) Pressure Spectrum Level vs. ƒ; all configurations at Re = ) Pressure Spectrum Level vs. ƒ; digitized data from Brüel & Kjær for 4138 microphone fitted with Nose Cone UA ) Pressure Spectrum Level PSL vs. ƒ; Calibrator Jet 22) Pressure Spectrum Level PSL vs. ƒ; solid walls; downstream; 3 microphone angles at Re = and Re = ) Relationship between G pp and free stream velocity 24) Normalized Pressure Spectrum Level vs. ƒ; solid walls; downstream; floor; lower Re 25) P r ms vs. free stream velocity for solid wall configuration, microphone: downstream iii

5 NOMENCLATURE Symbol Description Unit SPL Sound Pressure Level db or dba PSL Pressure Spectrum Level - SPL in a frequency band of 1 Hz db SPL 1/3 Pressure Band Level (PBL) in a frequency band of 1/3 octave db P Sound Pressure Pa P ms Sound Pressure (mean square) Pa 2 P rms Sound Pressure (RMS) Pa P ms Total Integrated P ms Pa 2 P r ms P ms Pa Re Reynolds Number per Meter: ρ U /µ P atm Atmospheric Pressure Pa V Voltage V T Temperature K G pp P ms /ƒ Pa 2 /Hz W Power W WC Water Column at Standard Conditions RPM Revolutions Per Minute 1/min X, Y, Z Test Section Coordinate System ρ Density kg/m 3 µ Viscosity kg/ms U Velocity (streamwise direction) m/s q Dynamic Pressure (½ ρ U 2 ) in WC ƒ Frequency Hz λ Wavelength m k x Constants (x ) iv

6 1. OBJECTIVE The purpose behind this experiment was to map out the acoustic characteristics of the Virginia Tech Stability Wind Tunnel. We were interested in documenting how the noise intensity levels vary as a function of velocity, streamwise location, and test section wall configuration (slotted vs. solid). 2. FACILITY The Virginia Tech Stability Wind Tunnel (figure 1) is a closed-loop wind tunnel with an airexchange tower so that temperature stabilization is allowed for. The 14 propeller consists of 8 custom made constant pitch blades, and is powered by a 600 hp Westinghouse Model No motor generator which rotates at a maximum speed of 900 rpm. 1 Turbulence levels for the empty tunnel have been measured to be extremely low: less than 0.1% for flow speeds up to 38.1 m/s. 2 In front of the test section there are seven stainless steel anti-turbulence grids. The 24 long constant 6 by 6 test section can be configured in several ways using solid steel panels or slotted walls. These slotted walls are made up from numerous 3" and 6" (width) longitudinally mounted fiberboard beams. The transverse spacing between each beam is 3". See figure 8 for a picture of the test section in fully slotted configuration. Figure 1: Schematic of the Virginia Tech Stability Wind Tunnel 3 1 Choi, K and R. L. Simpson: Some Mean Velocity, Turbulence, and Unsteadiness Characteristics of the VPI & SU Stability Wind Tunnel, December It must be noted, however, that this study was done before the stability wind tunnel was fitted with new fan blades, which are likely to have reduced the turbulence levels even more. 2 Ibid. 3 Website: 1

7 3. EQUIPMENT To measure and monitor the factors governing each run speed we used the equipment already present in the Stability Wind Tunnel. This includes a Validyne DB-99 Digital Barometer (resolution: 0.01 Hg), 4 an Omega Thermistor type (accuracy: ±0.2 C), 5 and a Pitot-Static Tube connected to a Setra 239 Pressure Transducer (accuracy: ±0.14%). 6 The additional experimental setup utilized in this experiment is depicted in figure 2. In order to mount the microphone into the tunnel, a microphone mount made out of a swept 2' 4" long symmetrical steel airfoil section welded to a steel base was used. A piece of steel tubing was modified so that it would fit the tip of the airfoil section and mounted in place with two screws. Two setscrews with vinyl tips served to hold the microphone preamplifier in place (figure 3). Duct tape was used to affix the microphone chord down the trailing edge of the airfoil section, and on the floor in the test section (figure 4). All acoustic measurements were made using a 1/8" condenser microphone (Brüel & Kjær type 4138 fitted with nose cone UA 0355). This was mounted on a 1/4" preamplifier (B&K type 2760) using the B&K UA 0160 Adapter and connected to the B&K conditioning amplifier (NEXUS 2690) positioned outside the test section. The 4138 is a vented microphone that has a dynamic range of 56dB-168dB and sensitivity equivalent to mv/pa. The calibration chart of the frequency response is shown in figure 5. The amplified microphone signal was monitored using a Tektronix digital oscilloscope and recorded using a Hewlett Packard HPE1432A 16-bit A/D converter. Finally, the digital signal reached a PC running HP VEE software programmed in house. Figure 2: Experimental setup: A: B&K 4138 Microphone with preamplifier, adapter and nose cone (2760, UA 0160, and UA 0355); B: Microphone mount; C: B&K NEXUS 2690 Conditioning amplifier; D: Tektronix digital oscilloscope; E: HP VXI A-D converter; F: Data acquisition system 4 Specifications on model DB-99 from Validyne Engineering Company 5 Specifications on type from Omega Engineering Company 6 Specifications and calibration data on Model 239 from Setra 2

8 Figure 3: Dimensioned sketch of microphone mount Figure 4: Microphone and mount in the Stability Wind Tunnel using the solid wall configuration and downstream microphone position on the floor (2a). 3

9 Figure 5: Frequency response calibration chart for B&K condenser microphone type 4138 Figure 6: Stability Wind Tunnel coordinate system Table 1: Microphone locations 4

10 4. PROCEDURE Measurements were made for two locations in the wind tunnel for all different test section configurations. One set of measurements downstream in the test section and one upstream. With the solid wall panels installed, it was also possible to affix the microphone mount in the ceiling and consequently this upstream position was added in the run. Table 1 gives the location of each set of measurements. The tunnel coordinate system used (figure 6) has its origin at the center of the upstream end of the test section. Two additional sets of measurements were made at the downstream position for the solid wall configuration. The microphone mount was here pitched upwards to form an angle of 4 with the free stream velocity. This procedure was then repeated for an angle of 6.5. For every test section configuration and microphone location the signal was recorded for several different speeds, including measurements with the fan off. The speed was increased in increments of approximately 5 m/s starting at a speed of about 10 m/s. Using the solid wall configuration (figure 7), we were able to measure noise levels for speeds up to 16" WC ( 80 m/s; max speed). However for the partly and fully slotted (figure 8) configurations we could not go past 4" WC ( 40 m/s) due to structural limitations of the test section and the surroundings. Utilizing a digital oscilloscope we were also able, in some cases, to measure dominant low-frequency unsteadiness, and the corresponding sound pressure amplitude. 5

11 5. DATA The microphone signal was recorded at a sampling rate of 51.2 khz. For each measurement, 1000 records - each of 8192 points in length - were measured and Fourier transformed using a Hanning window. From this, the accumulated averaged power spectra were computed. Total sampling time for each measurement was 160 seconds. This sampling scheme limited the lowest measurable frequency to 6.25 Hz. The highest readable frequency was 20 khz, limited by the antialiasing filter of the A/D converter. The averaged power spectra were converted into Pressure Spectrum Levels: PSL = 10 Log 10 (G pp / Pa 2 ) 1) They were also integrated to obtain the total RMS fluctuations, P r ms, which can be expressed as a Sound Pressure Level (SPL): SPL = 20 Log 10 (P r ms / Pa) 2) Figure 7: Solid wall configuration; microphone in the Figure 8: Fully slotted configuration; microphone in upstream location mounted in the ceiling. the downstream location mounted on the floor. 6

12 6. RESULTS AND DISCUSSION 6.1 Solid Wall Test-Section Figures 9 and 10 show the acoustic pressure spectra for the solid wall configuration with the microphone positioned downstream (run 2a) for free stream velocities ranging from 0 to approximately 43 m/s (fig. 9) and 48 m/s to 82 m/s (fig. 10). Some peaks can be explained: At the lower speeds we see peaks close to 60 Hz. These peaks are expected electronic noises. We see however that as the speed increases these peaks drown in the additional generated noise. At a Re of (per meter) we see a large peak at approximately 40 Hz. This corresponds to the blade passing frequency of the fan. At a setting of 306 RPM the eight blades generate a blade passing frequency of 40.8 Hz. The same pronounced peaks can also be found at corresponding frequencies for higher Re. Most of the traces, with the exception of the ones for very small Re, have peaks at the corresponding blade passing frequencies. For all the different speeds we observe a peak at about 185 Hz. This could be the resonant sound from a standing wave across the test section. (λ=1.83m). Especially since it is present in all plots of the solid test section and does not appear in any of the slotted configurations (figs ). Peaks appearing around 350 Hz for lower Re are consistent for every configuration without much variation, but disappear in the presence of surrounding noise for higher Re. This could be a sound generated by the tunnel itself or by the motor. At higher Re we see two sets of peaks appearing. The first starts at 700 Hz (Re: ), moving up to approximately 4000 Hz for a Re of Another starts at 4000 Hz (Re: ) and moves past Hz for the highest Re measured. These peaks may be acoustic manifestations of the vibrations that take place at the corresponding speed. The vibration of the panels and hence the microphone mount will generate sound waves, which are picked up by the microphone. As the speed increases these vibrations grow faster and therefore the peaks tend to move higher and higher up into the frequency range. 7

13 Figure 9: Pressure Spectrum Level vs. ƒ; solid walls; downstream on the floor; lower Re Figure 10: Pressure Spectrum Level vs. ƒ; solid walls; downstream on the floor; higher Re 8

14 Table 2: Details for run 2a: solid walls; microphone: downstream on the floor Figures 11 and 12 show PSL vs. Frequency for the same configuration (solid), but with the microphone positioned upstream on the floor in the test-section (run 2b). 7 The characteristics of these spectra are almost identical to the ones for run 2a. However they are shifted slightly downwards. In other words, for each Re, the sound intensity levels are 1 db to 3 db lower than when the microphone was in the rear position. Conclusion: The test-section gets noisier as we move downstream. In addition low-frequency unsteadiness was recorded for the first free stream velocities of this run. Sinusoidal signals appeared on the oscilloscope for very low frequencies (between 6 Hz and 18 Hz), which had corresponding amplitudes exceeding any other frequency in the measured spectrum. The large levels seen at the very low frequencies in the Pressure Spectrum Level Vs. Frequency figures can be looked upon as components of this feature since the resolution is very limited at these frequencies. Hence we can conclude that there is a significant amount of ambient low-frequency noise in the facility. See table 3 for details on run 2b. When positioning the microphone in the ceiling instead of on the floor, we get the noise characteristics described in figures 13 and 14 (see also table 4 for details on run 2d). We can easily see when comparing figure 13 with figure 11 that there is no significant difference between positioning the microphone in the ceiling apart for the very pronounced peak at a Re of This tone could even be heard from the control room. The peak could be a resonant frequency of the microphone mount itself, which was not nearly as firmly attached in place as it was when mounted on the floor. This could also be the reason why, for high Re (fig. 14), the noise levels are a few db higher than in figure 12 (microphone on the floor) for the peaks that are assumed to be caused by vibrations. Hence, we can conclude that in a steady flow, the noise intensity is not dependent on the microphone positioning in other than the streamwise direction. Some low-frequency unsteadiness was again recorded for selected speeds, and the same trend as in 2b was evident. The frequencies were not as well distinguishable, but clearly took place below the audible region. Again the figures suffer from poor resolution at such low frequencies. Nevertheless, the facility clearly displays high ambient noise levels in this lowfrequency band (0-20Hz), and the corresponding noise levels increase with free-stream velocity. 7 See table 1 and figure 6 for explicit location in the tunnel 9

15 Figure 11: Pressure Spectrum Level vs. ƒ; solid walls; microphone: Upstream on the floor; lower Re Figure 12: Pressure Spectrum Level vs. ƒ; solid walls; microphone; upstream on the floor; higher Re 10

16 Figure 13: Pressure Spectrum Level vs. ƒ; solid walls; microphone: upstream in the ceiling; lower Re Figure 14: Pressure Spectrum Level vs. ƒ; solid walls; microphone: upstream in the ceiling; higher Re 11

17 Table 3: Details for run 2b: solid walls; microphone: upstream on the floor Table 4: Details for run 2d: solid walls; microphone: upstream in the ceiling 12

18 6.2 Partly Slotted Test-Section Walls This part of the experiment was conducted with both walls and the ceiling in slotted configuration while solid paneling remained on the floor. Due to strong structural vibrations throughout the facility caused by the gusts of the wind flowing through the slotted test section, we could not run the Stability Tunnel beyond 4" WC ( 40 m/s). The wind gusts were a dominant feature of the pressure signals as well, producing very large amplitude oscillations visible on the oscilloscope. These pressure fluctuations were experienced as an unpleasant sensation by personnel present in the facility. They appeared to be a consequence of instability of the flow exiting and re-entering the slotted walls, perhaps coupled with a Helmhotz-type resonance of the entire wind-tunnel control room. Since the frequency of this instability (3 to 11Hz, increasing with speed) was too low to be accurately recorded in the pressure spectra, measurements of the gust amplitude were made directly from the oscilloscope screen and then converted into SPL. These results are listed as Low Freq. SPL in tables 5 and 6. In all cases this single frequency oscillation carried more energy than that accounted for in the entire resolved range of the pressure spectra (listed as SPL in the tables). Comparing these low-frequency SPL values with the ones for the solid wall configuration (tables 3 and 4) we see that the increase with free stream-velocity is much greater for the slotted walls, especially in the downstream position. Figure 15 and 16 show the spectra for the downstream (run 1a) and upstream (run 1b) microphone location respectively. For the downstream microphone location (fig. 15), we notice that for each Re the intensity levels are much higher than those of the solid test section. This is especially the case for frequencies below 100 Hz where the noise intensity exceeds 100dB for the higher speeds. These high levels presumably represent those parts of gusting broadened to higher frequency. The above-mentioned spikes at 350 Hz are still present, but the peaks assumed to be from a standing wave across the test section have disappeared, which supports our theory since nodes would not form on the fiberboard. We also notice the electronic noise at 60 Hz. The blade passing frequency can still be seen, but is not as evident as in the solid test-section configuration. Also noticeable is the absence of the high-frequency spikes that were observed with the solid panels at higher speeds. Apart from some small pressure changes at the end of the spectrum, the traces decay steadily from 1,000 Hz to 25,000 Hz. This supports the hypothesis that the spikes in the solid wall spectra are associated with panel vibrations, since the fiberboard absorbs a considerable amount of vibration. For the microphone upstream (fig. 16) we see the same trend as for the downstream location. However the levels are lower. This is due to the smaller amplitude of pressure fluctuations associated with the gusts at this station. Again the blade passing frequencies appear strongly for the three highest Re, and the electronic noise is also evident. Most of the spectra are about 1dB lower than with the microphone in the rear position. However, beyond 10,000 Hz it is actually a little bit higher, although the same trend of little vibration is followed. 13

19 Figure 15: Pressure Spectrum Level vs. ƒ; partly slotted walls; microphone: downstream on the floor; lower Re Figure 16: Pressure Spectrum Level vs. ƒ; partly slotted walls; microphone: upstream on the floor; lower Re 14

20 Table 5: Details for run 1a: partly slotted walls; microphone: downstream on the floor Table 6: Details for run 1b: partly slotted walls; microphone: upstream on the floor 6.3 Fully Slotted Test-Section Walls We ran the same experiment with fiberboard on all four walls of the test section. Again, due to the structural vibrations throughout the facility, we could not go beyond a dynamic pressure of 4" WC. The gusts that appeared in the partly slotted configuration were now even stronger. This can be seen in the spectra in figures 17 and 18, where the intensity levels are very high in the low frequency range, and the Low Freq. SPL s measured from the oscilloscope. The amplitudes for the downstream location reach well up in the 130-decibel range for almost all test speeds (see table 7). Frequencies for this configuration ranged from about 5 Hz at the lowest speeds to 11 Hz for 40 m/s. For the upstream location the amplitudes are much smaller (see table 8). Frequencies were harder to read from the oscilloscope which signifies that the great contribution from the low frequencies to the instability levels were less of an issue than for the downstream location. Still the levels are significantly higher than those measured for the corresponding position in partly slotted wall configuration. An interesting fact that needs to be pointed out in comparing slotted and solid walls is that the ambient low-frequency noise when the tunnel is not running, is significantly lower when utilizing the slotted walls (see tables 3, 6, and 8). This suggests that a large portion of the ambient low-frequency noise is generated in the tunnel itself and not by the immediate surroundings. Blade passing frequencies are still visible in the spectra (although they tend to drown in surrounding noise for the downstream position). Apart from the peaks at 350 Hz, there are very 15

21 few big fluctuations in these spectra. The intensity is high with the upstream position being about one db quieter than downstream, with the exception of frequencies lower than 30 Hz where the difference is much greater as discussed above. The general noise level is about four db higher than partly slotted for most frequencies over 1000 Hz, and almost 10 db higher than the corresponding frequencies for the solid test section. This time we see that the fiberboard seems to absorb practically all of the vibrations that appeared in the high frequency range when solid wall panels were used. This leaves us with uniform spectra with fewer pressure fluctuations (but much higher intensity levels) than in any other configuration. Table 7: Details for run 3a: fully slotted walls; microphone: downstream on the floor Table 8: Details for run 3b: fully slotted walls; microphone: upstream on the floor 16

22 Figure 15: Pressure Spectrum Level vs. ƒ; fully slotted walls; microphone: downstream on the floor; lower Re Figure 16: Pressure Spectrum Level vs. ƒ; fully slotted walls; microphone: upstream on the floor; lower Re 17

23 6.4 Summary To sum up this section we can conclude that the solid walls are quieter than partly slotted, which again are less noisy than fully slotted. This is the overriding factor. But within any configuration, downstream in the test section is noisier than upstream. However, for very low frequencies (less than 20 Hz), downstream, partly slotted is noisier than the upstream, fully slotted. Figure 19 shows this graphically at a dynamic pressure of 1.0" WC or corresponding Re of ( 21 m/s). Figure 19: Pressure Spectrum Level vs. ƒ; all configurations at Re = Going back to the instability issue, we see the same trend as for general noise levels, but the effect is even more dramatic. Slotted walls create very large low frequency pressure fluctuations that downstream override the rest of the entire spectrum. This effect decays as we go upstream in the test section, but are still significantly higher than with the solid walls. Even though there are advantages in using slotted walls for acoustic measurements (noise absorption etc.), the adverse effects are too significant to overlook. By using the solid wall configuration the instability effects are eliminated in the low frequency regime and the overall Sound Pressure Level is lowered significantly. This also diminishes the streamwise position effects. 18

24 7. FURTHER DISCUSSION 7.1 Wind-Induced Microphone Noise In order to measure the amount of pure tunnel noise we needed to have some knowledge of the amount of microphone noise induced by the wind. The microphone manufacturer (B&K) supplied us with wind induced noise levels using the same equipment as we utilized in our experiments. A digitized copy of this data for two speeds cross plotted with the data from the fully slotted wall configuration with the microphone in the downstream position (the most noisy combination we measured) can be seen in figure As we can see, the B&K supplied noise levels are significantly higher than what was measured in the Virginia Tech Stability Wind Tunnel for the same velocities. The B&K data were made using a rotating microphone boom, however the atmospheric conditions under which the data was acquired are not known. If the B&K data was measured at standard conditions, the two sets of data move closer to each other, but still the difference is unacceptable. The only reasonable explanation is that the B&K data in addition to the windinduced noise is also picking up external noise in the test facility, which exceeds the noise generated in the Virginia Tech Stability Wind Tunnel. However, this noise is thought to come from the apparatus driving their microphone boom since the B&K data is significantly lower than our tunnel data for a velocity of 0 m/s (but nowhere close to 0dB either). Figure 20: Pressure Spectrum Level vs. ƒ; B&K wind induced noise data for the 4138 microphone fitted with nose cone UA 0355, cross plotted with the noisiest VT Stability Tunnel configuration (3a) at the same speeds for reference. 8 B&K supplied plot converted into PSL instead of the original SPL 1/3 19

25 An additional experiment was conducted measuring the noise level at the same velocities in a jet calibrator. Unfortunately, the jet generated very high noise levels and exceeded the levels in the Stability Wind Tunnel for most frequencies and velocities (fig. 21). However, for frequencies lower than 200 Hz (for speeds over 40 m/s), the data from the tunnel reaches higher levels than that produced from the calibrator. This suggests that for at least these frequencies, wind induced noise is of little relevance. In addition, we can notice the peaks that occur for every speed. These peaks move to higher frequencies as we increase the speed, and they could also be heard as whistling sounds as the jet stream exited the nozzle in the calibrator. According to this data, the G pp (P ms /ƒ) tends to go up with velocity to the fifth power, which corresponds to an increase in 15dB with a doubling in the speed. Jet noise is known to increase with velocity to the eighth power, which suggests that the microphone should have been positioned further upstream in this experiment for a higher quality jet stream. Taking this additional data into account we must conclude that the intensity levels that are documented throughout this report should be looked upon as upper levels of the acoustic noise in the Stability Tunnel since we have been unsuccessful in subtracting the wind induced microphone noise. However, both figure 20 and 21 support the fact that the peaks that appear around 350 Hz for lower Re in the Stability Wind Tunnel are not a characteristic of the microphone but rather of the facility itself. Figure 21: Pressure Spectrum Level vs. ƒ; calibrator jet data using the 4138 microphone with UA Tip of microphone positioned ½" in front of the nozzle. (* Refers to data collected with the nozzle cap off). The two last traces are included from tunnel run 3a for reference. 20

26 7.2 Microphone Alignment In addition, we wanted to investigate whether the alignment of the microphone would have an effect on the data. This was done using the downstream position in the solid panel configuration. Data collected with the microphone aligned with the free stream was compared with data from tilting the microphone mount upwards (this way the airfoil section of the mount would still be at 0 angle of attack) first 4 and then 6.5. The results for two different Re can be seen in figure 22. As we would expect there is, for the most part, very little difference between the corresponding traces. However, the aligned traces seem to have higher spikes than the unaligned ones. This could be wind-induced noise, or just that the noise itself is aligned with the free stream. Figure 22: Pressure Spectrum Level vs. ƒ; solid walls; microphone: upstream; three angles with respect to free stream at Re = (lower curve) and Re = (upper curve) 21

27 7.3 Further Analysis Choosing the fully solid configuration with the downstream microphone (run 2a) for our further analysis, we plotted G pp against Velocity for three different frequencies: 600Hz, 1,000Hz, and 5,000Hz. All three of these traces in figure 23 reveal that G pp increases with velocity adhering to the following power relationship: G pp = k 1 v 6 3) Therefore, RMS sound pressure increases with velocity according to: P rms /ƒ = k 2 v 3 4) This again means that by doubling the free stream velocity we should see an increase in P rms by a factor of eight. Then the corresponding PSL levels should increase with 18dB when doubling the velocity, and all of the supplied pressure spectra (figs. 9 through 18) show this quite clearly. In other words, this power relationship is independent on test section configuration. The difference is contained within the constant. Figure 23 also reveals that at 18,000Hz the G pp does not behave according to equation 3). This is due to the many irregular pressure fluctuations that occur at high frequencies for this configuration. But for frequencies where the spectrum traces are relatively uniform for any Re, equation 3) holds sufficiently. By normalizing our G pp according to equation 3) we end up with normalized spectra vs. frequency plots like the one in figure 24. Using the run configuration 2a we see that for free stream velocities up to about 40 m/s most of the normalized spectra fall on top of each other, and the relationship gets better as free stream velocity increases For reference we have included corresponding A-weighted Sound Pressure Levels for every configuration and test speed (tables 2 through 8). Most applications use this weighting scale, which is tuned to the average human ear. Note that when using this weighting scheme, the energy in the lower frequency region is discriminated against. By this measure the Virginia Tech Stability Tunnel appears as a relatively quiet facility, considering that it was not designed for aero-acoustic measurements. In comparison, the Pininfarina Aero-acoustic Wind Tunnel in Italy recently made some modifications to their facility, which reduced the ambient noise level to 72 dba 9 when operating at a free stream velocity of 100 km/h. Using the solid wall downstream location in the Virginia Tech Stability Tunnel we get noise levels around 81 dba at the same free stream velocity. These levels are on the same order as in the Pininfarina Wind Tunnel before the effort in later years to reduce the noise down to its present day value

28 Figure 23: Relationship between G pp and free stream velocity at several frequencies. Solid wall configuration. Microphone positioned downstream on the floor Figure 24: Normalized spectra vs. frequency for run configuration 2a (solid walls; microphone downstream on the floor). Scaled with free stream velocity to the 6 th power and a reference velocity at Re = 9.73*10^5 23

29 Figure 25 shows the relationship between the integrated pressure fluctuation levels P r ms and velocity for the solid-wall configuration. Discarding the data points for low Re we see that it increases according to: P r ms = k 3 v 2.6 5) This relationship, however, can only be applied to the solid wall configuration. The slotted walls do not adhere to such a power law. Figure 25: P r ms vs. free stream velocity. Solid wall configuration; upstream and downstream microphone locations 24

30 8. CONCLUSIONS This report documents the acoustic noise characteristics of the Virginia Tech Stability Wind Tunnel. Measurements were made as a function of streamwise position for three-different test section configurations - solid walls, floor and ceiling, slotted walls, floor and ceiling, and slotted walls and ceiling with solid floor. Measurements were made in the free stream using a 1/8" B&K microphone fitted with a nose cone. Frequency spectra and overall sound pressure levels were computed. The measurements contain an unknown (but apparently not dominant) contribution from wind induced microphone noise and thus should be looked upon as upper bounds of the acoustic noise in the Stability Tunnel. We have found that the solid-wall test section configuration produces by far margin the quietest flow. The more slotted walls installed in the test section, the noisier it becomes. In particular, the slotted walls are responsible for intense low-frequency unsteadiness (or gusts ) in the wind-tunnel pressure field. These gusts carry much more energy than the rest of the spectrum put together. The further downstream measurements are taken, the stronger this low-frequency unsteadiness will be. Acoustic Noise is also a weak function of streamwise position. Noise levels increase going downstream for all configurations. In the low-frequency range for the slotted wall configurations this streamwise effect becomes very pronounced. Broadband acoustic spectral levels increase with free stream velocity to the 6 th power for all configurations and streamwise positions. This corresponds to an increase of 18dB with a doubling in free stream velocity. By using the A-weighting scheme the Virginia Tech Stability Tunnel in solid-wall configuration emerges as a relatively quiet facility for aero-acoustic measurements, especially considering that it was not built for this purpose. 25

31 9. REFERENCES 1. Chio, K and R. L. Simpson: Some Mean Velocity, Turbulence, and Unsteadiness Characteristics of the VPI & SU Stability Wind Tunnel, December Website: 3. Reynolds, G.A; Experiments on the Stability of the Flat Plate Boundary-Layer with Suction; VPI & SU Specifications on Equipment in the Virginia Tech Stability Tunnel from Omega, Setra and Validyne 5. Wind Induced Data and Microphone Calibration from Brüel & Kjær 6. Website: 26