1、OR-05-14-3 (RP-1010) Model 36-14-1780 36- 17-1 780 36-26- 1780 Observations on the Acoustic Data Housing Hub Diameter Diameter 36 14 36 17.5 36 26.5 John A. Murphy, PhD Member ASHRAE RPM fbp - HZ ABSTRACT Although the primary purpose of ASHRAE research project I O1 O was to investigate the efects of
2、 inlet disturbances on the air andsoundperformance ofaxial fans, two interesting phenomena were revealed in the narrow band (1/24th octave) acoustic data for undisturbed inlets. These phenomena are discussed in this paper. The influence of operating point on the shape of the acoustic spectrum, espec
3、ially on the tonali, is discussed. The second interesting phenomenon is the vortex shedding, which occurs for most of the fans tested. This behav- ior is analyzed and explained. INTRODUCTION Although the primary purpose of ASHRAE research project 1010 was to determine the inlet system effects on the
4、 performance and noise of axial fans, an analysis of the data from this project, especially the 1/24th octave band data, can provide significant insight into the mechanism of the sound generation. It is the purpose of this paper to provide such an analysis. BACKGROUND The project consisted of testin
5、g three axial fans of the same tip diameter but with three different hub diameters. Since these fans were of the adjustable pitch (at rest) variety, each fan was tested at four blade angles. Thus, a total of twelve fans were tested. A full performance curve (a minimum of eight pressure/volume points
6、) was tested for each fan. Sound perfor- mance was measured at every point and was archived as 1/24th octave band data. These data provide the opportunity to exam- ine the characteristics of the noise generation mechanism. The three fans were supplied by New Philadelphia Fan Company. The description
7、s are: Blade Angle - Degrees 24 16 9 1 1779-1771 1786-1782 1796-1791 1798-1796 296.5 - 295.1 297.7 - 297.0 299.3 - 298.5 299.7 - 299.3 Blade Angles Degrees 1,9,16,24 3,10,18,25 15,22,30,37 Motor Speed 1780 ANALYSIS 36-1 4 Figures 1 through 4 show the 1/24th octave band sound pressure levels at the f
8、our blade angles tested. Even though no stall occurred for blade angles 9 and 1 degree, only 5 points are shown. The remaining test points represent performance normally not utilized. The blade passing frequencyhp calcu- lated from the actual RPM is shown in the table. John A. Murphy is vice preside
9、nt, JOGRAM, Inc., New Philadelphia, Ohio. 1 O08 02005 ASHRAE. Figure 1 1/24 octave band SPL vs. operating point for 36- 14 at 24“ blade angle. Figure 2 1/24 octave band SPL vs. operating point for 36- 14 at 16“ blade angle. RPM 1 1771 - 1759 1 1784- 1776 1 1792-1786 Figure 3 1/24 octave band SPL vs.
10、 operatingpoint for 36- 14 at 9“ blade angle. Figure 4 1/24 octave band SPL vs. operating point for 36- I4 at 1 “ blade angle. 1794- 1793 The octave band data in general showed an increased sound level as the fan moves up the curve, with a bigger increase in band 4 (500 Hz), which contains the first
11、 harmonic of the blade passing frequency, than in band 3. The 1/24th octave data indicate a significant change in the character of the sound over this performance range. For the 24 degree blade angle, the blade passing tone stands up higher above the adja- cent bands as the performance point is move
12、d from free flow to the top of the curve. For the other three blade angles, the reverse is true. For all four blade angles, the first harmonic (second blade passing) is less pronounced at the top than at free flow. Another interesting feature seen in the 1/24th octave band data is the “hump“ at abou
13、t 8 kHz seen at all blade angles except 24 degrees. This result is due to an interesting phenom- enon called “vortex shedding.“ This will be discussed in a later section of this paper. fbp - Hz I 354.2 - 351.8 1 356.8 - 355.2 1 358.4 - 357.2 36-1 7 358.8 - 358.6 Figures 5 through 8 show the 1J24th o
14、ctave band sound pressure levels at the four blade angles tested. Even though no stall occurred for blade angles 10 and 3 degrees, only 5 points are shown. The remaining test points represent performance not normally utilized. The blade passing frequency fbp calcu- lated from the actual RPM is shown
15、 in the table. Blade Angle - Degrees The fbp falls near the boundary of the third and fourth full octave bands (center frequencies = 250 and 500 Hz) for al1 blade angles. In the case of the 1/24th octave band data, the ASHRAE Transactions: Symposia 1 O09 Figure 5 1/24 octave band SPL vs. operating p
16、oint for 36- 17 at 25“ blade angle. Figure 6 1/24 octave band SPL vs. operating point for 36- 1 7 at 18“ blade angle. RPM f&, - HZ Fgure 7 1/24 octave band SPL vs. operatingpoint for 36- 1 7 at 1 O“ blade angle. Figure 8 1/24 octave band SPL vs. operating point for 36- 17 at 3“ blade angle. Blade An
17、gle - Degrees 31 30 22 15 1787-1783 1791-1786 1796-1793 1796-1794 476.5 - 475.5 477.6 - 476.3 478.9 -478.1 478.9 - 478.4 blade passing frequency falls in either band 72 (350 Hz) or band 73 (360 Hz) for all blade angles. The change in the sound pressure levels (in the octave bands containing the blad
18、e pass- ing frequency) as the operating point is moved from free flow to the top is generally an increase, with a pattern that does not appear to be a reasonable function of blade angle. The smallest increase was at 25 degrees, but the largest increase was at 18 degrees. The height of the blade pass
19、ing tone (taken from the 1/24th octave data) in general decreased as the operating point was moved from free flow to the top. The remarkable decrease for the 18 degree blade angle is almost entirely due to the large increase in the background levels (see Figure 6). It is interesting to note that no
20、high frequency “hump“ occurs for this fan at any blade angle. More on this later. 36-26 Figures 9 through 12 show the 1/24th octave band sound pressure levels at the four blade angles tested. Even though no stall occurred for blade angle 15 degrees, only 5 points are shown. The remaining test points
21、 represent performance not normally utilized. The blade passing frequency fbp calculated from the actual RPM is shown in the table. 1010 ASHRAE Transactions: Symposia Figure 9 1/24 octave band SPL vs. operatingpointfor 36- 26 at 3 7“ blade angle. Figure 10 1/24 octave band SPL vs. operating pointfor
22、 36- 26 at 30” blade angle. Figure II 1/24 octave band SPL vs. operating point for 36- Figure 12 1/24 octave band SPL vs. operating point for 36- 26 at 22” blade angle. 26 at 15” blade angle. Thefb, falls in the fourth full octave band (center frequen- cies = 500 Hz) for all blade angles. In the cas
23、e of the 1/24th octave band data, the blade passing frequency falls in band 83 (480 Hz) for all blade angles. The change in the sound pressure levels in the octave band containing the blade passing frequency as the operating point is moved from free flow to the top is always an increase, with the am
24、ount of increase becom- ing larger as the blade angle is decreased. The height of the blade passing tone (taken from the 1/24th octave data) always decreased as the operating point was moved from free flow to the top. The height of the tone was always high enough to ensure that this tone would be si
25、gnificant at all blade angles and all operating points. This can be clearly seen in Figures 9 through 12. Note that the high frequency “hump” exists at all four blade angles. subject ofthis paper, it is interesting that for this fan (36-26 at 37 degrees), a sixth point on the stable part of the curv
26、e was found. It is clear from the narrow band (1/24th octave) data as well as the full octave data (not shown) that the shape of the sound spectrum is very different at this point than at any other. A discussion of the repercussions of this point on the perfor- mance data is the subject of another p
27、aper. VORTEX SHEDDING The flow at the trailing edge of a bluff body often produces a phenomenon referred to as a “Karman Vortex Street.” This flow pattern has long been known to be respon- sible for “galloping transmission lines” and was the underly- ing cause of the collapse of the Tacoma Narrows b
28、ridge in the mid-twentieth century. The shed vortices have a distinct Deri- Figure 9 shows an interesting artifact of the test results. Although the analysis of the pressure-volume data is not the odiciy and this can cause large motions if the period happens to coincide with a natural mode of the st
29、ructure in the airflow. ASHRAE Transactions: Symposia 1011 A discussion of the wake behind a cylinder was included in almost every elementary aerodynamics text. A rather large body of literature has been developed on predicting the frequency of vortex shedding behind bluff bodies of various shapes i
30、n the past several decades, driven by the needs of the turbomachinery industry. A good summary can be found in Chapter 9 of Handbook of the Acoustic Characteristics of Turbomachinery Cavities published by ASME in 1997. Bluff bodies shed structured wakes because a shear layer separates from both side
31、s of the body. The distance separating the shear layers, known as the wake width, d, is the character- istic length common to all bluff bodies. The vortex wake will have a frequency determined by the Strouhal number, which is defined as: St = fdN. The critical value ofthe Strouhal number is 0.18. Fo
32、r our purposes, we can assume that the wake width (at least for small angles of attack) is the trailing edge thickness of the blade. In a perfect world the trailing edge would be sharp, i.e., the upper and lower surfaces wouldmeet at an acute angle. The fans used for RP-1010 have blades that were ca
33、st of aluminum alloy using a permanent mold process. It is not feasible to produce castings that come to a point, so the blade design was altered to define the trailing edge as a semicircle. The mold parts at the leading and trailing edges. This joint and the lack of complete mold closure usually pr
34、oduce some flash- ing that must be removed during the subsequent manufactur- ing procedure. Ideally this would be done with a flexible sander, which would allow a skilled operator to produce a trail- ing edge profile approaching the ideal case. This cleaning is often done using a flat grinding wheel
35、 and this usually leads to a trailing edge, which is rather blunt. As the operating point is moved from free flow toward the stall point (if one exists), the separation point on the upper surface will begin to move forward and will, at the stall, move close to the leading edge. We can estimate that
36、the value of the distance between the shear layers, d, will also increase. Now consider Figures 1 through 4 (the 36-14 fan). The high frequency hump in Figure 1 (24 degrees) is quite small and is detectable only at free flow. The magnitude of the “hump” increases as the blade angle is decreased and
37、decreases as the operating point moves to lower flows. The peak moves to a lower frequency as the blade angle is decreased and when the operating point moves to lower flows. This behavior can be explained by using d = 0.0625 and assuming that the strength of the shed vorticity is proportional to the
38、 actual flow velocity at the trailing edge. For the 36-17 fan (Figures 5 through 8) using d = 0.05 (or smaller) predicts that the “hump” occurs beyond the range of the data. The behavior observed in Figures 9 through 12 (36-26) fan is explained by using d = O. 125. 1012 ASHRAE Transactions: Symposia
