Qualification of Fan-Generated Duct Rumble Noise Part 2: Results

2008, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). ESL-PA-08-06-09 SL-08-003 (RP-1219) Qualification of Fan-Generated Duct Rumble Noise Part 2: Results Joshua Kading J. Adin Mann, III, PhD Michael B. Pate, PhD Associate Member ASHRAE Member ASHRAE Member ASHRAE This paper is based on findings resulting from ASHRAE Research Project RP-1219. ABSTRACT Duct rumble noise in HVAC air distribution systems is commonly attributed to the poor aerodynamic discharge conditions of fan outlets. To date, qualitative descriptors published in ASHRAE Handbooks have been used as design guidelines for engineers to limit the amount of duct rumble noise. ASHRAE funded a project, RP1219, to quantify the rumble noise as the fan discharge orientation and distance from a duct is varied. A test system was built and verified to measure discrete frequency and one-third octave spectra in an adjoining room to the fan room. Tones, associated with vibration, were removed from the spectra, so that the one-third octave band data focused on analysis of the aerodynamic noise. Results for the fan discharge orientation and distance relative to a 90 degree elbow are presented. Some trends are identified; however, overall the results indicate a need for testing with additional fans and coupling the testing with measurements of the fan discharge air flow. INTRODUCTION The discharge configuration of a fan affects the amount of low frequency duct rumble noise downstream of the fan. Rumble noise, defined for this study as noise in the 16 Hz to 300 Hz one third octave bands, is considered by many practicing engineers as an indication of turbulence caused by poorly designed fan discharge conditions. With the goal of replacing qualitative design recommendations with quantitative design tools in the ASHRAE Applications Handbook (2003), ASHRAE funded a research project, RP1219, to obtain data on the correlation of rumble noise to the fan and duct configuration. If successful, the result would be a prediction tool that could be used to compare design alternatives such as weighing the cost of modifying the orientation and position of the fan compared to a noise control measure such as lagging or lining the duct. In Kading (2006) the design and verification of the test facility and procedures were presented. The specification on the design of the test system was that four different fan discharge conditions could be tested with four different lengths of duct from the fan outlet to the first duct transition in the system, which in all but one case was a 90 degree elbow. The measurements were to be performed at 8 operating points of the fan, to be accomplished by varying inlet restriction to the fan and the fan speed. A companion paper, Kading et.al. (2008) contains a summary of the test system and verification of the test system. In this paper, the focus is on the key results and suggestions for future work. TEST SYSTEM The system consists of three rooms shown in Figure 1: a fan room, a measurement room and an outlet room. A more detailed description of the test system is given in a companion paper (Kading.et. al. 2008 and Kading 2006). The fan room was the inlet to the duct system, the measurement room is acoustically isolated from the rest of the system and the outlet room was open to the rest of the building, including the fan room. The fan was a double wide/double inlet, forward curve 18-inch diameter wheel, scroll fan. The inlet to the fan room Joshua Kading is an engineer at Stanley Consultants, Inc., Muscatine, IA. J. Adin Mann III is an associate professor and Michael B. Pate is a professor in the Department of Mechanical Engineering, Iowa State University, Ames, IA. 28 2008 ASHRAE Published in Vol. 114, Part 2

2008, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). ESL-PA-08-06-09 was controlled with a hole in the door with adjustable area, thus acting as the volume damper. The system was built so that flow rate, fan speed, fan configuration, and duct length could be varied. The measured quantities were flow rate, average sound level (discrete frequency spectra and one-third octave band) in the measurement room, and sound levels in selected points inside the duct. The uncertainty in the measured one-third octave band sound level was determined to be ±1.7 db when retesting and rebuilding a configuration. FAN DISCHARGE CONDITIONS The four fan configurations are shown in Figure 2. In each configuration, the distance from the fan outlet to the transition piece was varied: Configuration 1: the fan was in front blast, blowing directly in to the expansion duct. Distances of 0D and 1D were tested. Configuration 2: the fan discharged air to a 90 degree elbow turned with the fan rotation. Distances of 0D, 1D, 2D, and 2.3D were tested. Configuration 3: the fan discharged air to a 90 degree elbow turned against the fan rotation. Distances of 0D, 1D, 2D, and 2.3D were tested. Configuration 4: the fan discharged air to a 90 degree elbow turned perpendicular to the fan rotation plane. In this configuration there was a 7 transition piece at the elbow that directed the air into the duct system. Distances of 0D, 1D, 2D, and 2.3D were tested. SELECTION OF OPERATING CONDITIONS The required operating points for the fan were stated in the research request for proposals (RFP). These populate the fan curve to give a representation of the effects on sound level at different typical operating points, Figure 3 and Table 1. The fan curve in Figure 3 was used as a basis for setting all of the operating points. Table 2 provides the AMCA 300 Predicted inlet sound levels using fan vendors modeling program. DATA REPRESENTATION Processing and visualizing this large data set was a challenge. Figure 4 gives an example of the measured one-third octave band sound levels for Configuration 1 with the fan one duct diameter (1D) from the transition duct. Since there are significant changes in the spectra as a function of the operating point, the changes by configuration would have not been clear if the measured spectra were used directly. Therefore, as shown in Figure 5, the difference of spectra compared to a reference was used. The reference could change, however, in most cases the reference was Configuration 1 at a distance of 1D. Within the space constraints of the test facility this case represented the closest to the ideal configuration. REMOVING TONES In the cases where the sound levels were low, tones entering the sound measurements from the fan and motor vibration were a concern. This issue was remedied by removing the tones from the discrete frequency spectra, and then calculating the one-third octave band levels from the modified spectra. Figure 6 shows an example of a discrete frequency spectrum with significant tones along with the spectrum with the tones removed. The tones were removed manually. The data file was modified by setting the values at the frequencies of the tones to the nearby broadband levels. A similar approach was used to separate the tone and broadband noise in a fan noise study, The 4th, 8th, and 12th harmonics were left since they were correlated to an aerodynamic noise source on the fan (Kading 2006 and Kading et. al 2008). Figure 7 shows one third octave spectra for one case. In the top plot, the original sound levels are presented and in the Figure 2 Figure 1 Test system layout. Four fan outlet configurations with distance D as the variable length (0 to 24 in configuration 1, 0 to 55 in Configurations 2 and 3, and 7 to 55 in Configuration 4). Published in Vol. 114, Part 2 29

2008, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). ESL-PA-08-06-09 Figure 3 Operating conditions setpoints along the fan curve. Table 1. Operating Points Studied for Every Configuration Flow Rate (% of wide open flow) Pressure Drop Across Fan (in water) 1.0 2.5 4.0 90% X 80% X X 60% X X X 35% X X Table 2. AMCA 300 Predicted Inlet Sound Levels Using Fan Vendors Modeling Program %WOF Flow Rate (cfm) Fan Pressure Drop (in. wt) Fan Speed (rpm) Inlet Sound Power Level in db Octave Band, Hz 63 125 250 500 1k 2k 4k 8k LwA 90 15147 1.0 986 110 109 103 100 99 95 92 86 104 80 60 35 9732 1.0 712 107 101 95 92 87 82 79 73 94 15388 2.5 1126 111 110 104 100 98 95 92 87 104 5939 1.0 580 98 92 86 82 75 71 67 61 84 9391 2.5 916 104 100 95 91 88 83 80 76 94 11878 4.0 1159 108 106 101 97 94 90 86 83 100 5550 2.5 929 102 100 94 90 87 81 77 73 93 7020 4.0 1175 107 105 101 96 93 88 84 80 99 30 Published in Vol. 114, Part 2

2008, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). ESL-PA-08-06-09 Figure 5 Figure 4 Example of sound levels and sound level differences. Average one-third octave band sound levels in db at Configuration 1, 1 duct diameter. (a) 1.0 inch water pressure drop. (b) 2.5 inch water pressure drop. (c) 4.0 inch water pressure drop. bottom plot, the one-third octave levels with the tones removed are presented. In general, removing the tones, as in the case of Figure 7, has the greatest effect in the 31.5 Hz onethird octave band. Since the spectra with tones removed is more indicative of the aerodynamic noise, the rest of the results will focus on the spectra with the tones removed. RESULTS: OBSERVED TRENDS The results overall showed more complexity than was originally expected, however, there were some trends that were identified. Figure 8 shows all configurations for the case of the fan outlet being 1D from the transition. Each plot contains data for one operating point and shows the sound level change for Configurations 2, 3, and 4 relative to Configuration 1. As a general statement, Configuration 4 is the loudest, followed by Configuration 3 and then Configuration 2. The noise levels increase by as much as 5 to 10 db. However, the difference varies with operating point. In all cases, the difference decreases with frequency. For Configurations 2 and 4 and all the operating points, Figures 9 and 10 show the changes in sound level as a function of distances between the fan discharge and the transition. Each case has the smaller distance as the reference, thus positive Figure 6 Example of a measured spectrum and the spectrum with tones removed. 60% WOF, 1inch Pressure drop, Configuration 2, 0 Duct diameters. numbers represent a lowering of the sound level as distance is increased. Figure 9 shows a sound reduction of 2 to 4 db when increasing the distance from 0D to 1D for Configuration 2. However, for Configuration 4, the noise level generally increases by up to 6 db for certain operating points as the distance increases. One can hypothesize that in the case of Configuration 2, there is an aerodynamic interaction between the fan discharge and the transition that is reduced as the distance between the fan discharge and transition is increased. In some cases for Configuration 4, one can hypothesize that being close to the transition may stop an aerodynamic effect from being generated. Figure 10 shows that for the case of Configuration 2, the noise is more greatly decreased when increasing the distance from the fan discharge to the transition from 1D to 2D. However, the changes are limited to the 20 to 63 Hz one third octave bands. In the case of Configuration 4, the sound level decreases with increasing distance from 1D to 2D, but the change is smaller than changes in other cases such as Configuration 2. This would Published in Vol. 114, Part 2 31

2008, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). ESL-PA-08-06-09 Figure 7 Difference between Configuration 3 and Configuration 2 at operating point 35%WOF and 2.5 inch pressure drop. A) original and B) After tones are removed. Figure 8 Four discharge configurations at the eight different operating points. The first columns of plots are the 1.0 inch pressure drop sound level differences (90%, 80%, and 60% WOF). The second is the 2.5 inch pressure drop sound level differences (80%, 60%, 35%). The third column is the 4.0 inch pressure drop sound level differences (60%, 35%). 32 Published in Vol. 114, Part 2

2008, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). ESL-PA-08-06-09 Figure 9 Sound Level Reduction at 1D compared to 0D. A positive value indicates a lower sound level at 1D than 0D. Figure 10 Sound Level Reduction at 2D compared to 1D. A positive value indicates a lower sound level at 2D than 1D. Published in Vol. 114, Part 2 33

2008, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). ESL-PA-08-06-09 suggest that at 1D, the aerodynamic effects are fully formed so that increasing the distance decreases the noise generation and that the aerodynamic noise is dominated by the fan and discharge flows rather than the flow interacting with the elbow. It is assumed that since the fan operating points are the same for all these comparisons that the sound level changes are caused by changes in the aerodynamic noise generated between the fan discharge and the elbow. Measurements of the air flow in this region could help correlate the sound changes to the aerodynamic phenomena. Figure 11 shows the one-third octave band changes and the discrete frequency spectra for the same cases. The point of these plots is to show that the low frequency changes in these cases is caused by a significant change in the broad band noise when increasing the distance from 0D to 2D. Assuming that the broadband noise is dominated by turbulent flow, it seems reasonable to conclude that the noise level changes are caused by changes in the aerodynamic noise sources when the distance between the fan discharge and elbow are increased. Comparison of the discrete frequency spectra and the one-third octave sound level changes does indicate several cases where the sound level change is dominated by the 4th, 8th, or 12 th harmonic tones changing. Figure 12 is a generalized representation of the operating points that are dominated by tone noise and those by broad band noise. It also indicates which operating points generally did not follow trends for the configuration. Finally, Table 3 shows a compilation of all the results but in octave bands. The cells are coded with dark gray indicating a sound level increase of 2 db or more and light gray indicated Figure 11 Comparison of one-third octave levels and discrete frequency spectra levels showing broad band changes. 34 a sound level decrease of 2dB or more relative to Configuration 1 at 1D. There is no clear pattern to the results. CONCLUSIONS The results show that varying the fan configuration and distance from the fan discharge to the elbow can cause changes in low frequency noise over a 14 db range. The greatest changes occur in the one-third octave bands from 20 Hz to 63 Hz. For most configurations and operating points, the change decreases to below 2 db above the 125 Hz onethird octave band. An overriding conclusion from the work was that there was no simple relationship between the configuration, duct length between the fan discharge and 90 degree elbow inlet, and fan operating condition. In general, the expected trends were seen for selected operating points; however, for some operating points and distances between the fan outlet and elbow, the configurations considered non-ideal (Configurations 3 and 4) were quieter by 5 to 10 db. A primary outcome of the work is conclusions on how future work could be done differently to answer the questions originally raised in the RFP. The data suggests that the aerodynamics near the fan have a complex behavior that requires additional diagnostic tools to be used and that the experimental setup be redesigned to separate confounding effects and tone noise. The recommendations for future work fall into four categories. A discussion of each follows. 1. Use diagnostic tools to study the fluid flow from the fan exit through the elbow. 2. Reduce potential confounding of the aerodynamic noise generation and the breakout noise from the duct in the measurement room. Figure 12 Schematic representation of regions on the fan curves where tones dominate the sound level changes and where the change did not closely follow identified trends for the configuration. Published in Vol. 114, Part 2

2008, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). ESL-PA-08-06-09 Table 3. Compiled Octave Band Data for All Cases* Configuration 1 Configuration 2 Configuration 3 Configuration 4 0D 1D 0D 1D 2D %D 0D 1D 2D %D 0D 1D 2D %D 16 1.3 81.5 3.5 1.2 1.9 1.0 1.2 4.2 4.1 1.6 1.2 0.7 1.9 2.9 31.5 0.4 81.0 4.1 1.4 2.5 1.4 1.7 5.6 4.4 2.3 2.4 4.6 4.7 3.8 63 1.3 81.1 5.2 0.1 1.2 1.0 1.6 4.0 3.5 1.7 1.5 2.9 2.9 2.8 125 0.4 80.6 2.2 1.1 0.2 0.5 1.3 3.4 2.6 0.4 0.5 0.8 0.7 0.7 16 0.1 74.2 2.8 1.7 1.9 1.5 1.0 2.6 2.4 1.4 1.1 2.8 3.2 3.8 31.5 1.2 72.7 2.1 1.4 0.4 0.5 3.1 3.0 1.4 1.4 3.5 5.6 3.0 3.7 63 2.1 73.1 1.8 0.8 0.3 0.7 5.6 3.7 1.2 1.0 4.3 1.8 0.9 1.8 125 0.2 70.3 1.6 1.4 1.2 0.2 2.3 2.5 1.3 0.2 0.2 0.9 0.6 0.9 16 0.7 67.4 2.2 0.2 1.4 0.6 1.4 2.7 3.3 3.0 0.3 4.3 4.2 4.6 31.5 7.0 65.6 6.0 1.1 0.7 3.0 3.2 0.7 0.6 1.3 2.5 3.6 2.1 1.3 63 2.4 65.4 1.7 1.1 0.5 0.7 2.2 0.8 0.3 0.5 2.0 0.4 0.3 0.5 125 2.9 60.8 2.1 0.5 1.7 1.4 2.1 1.5 1.6 0.9 0.4 0.6 0.2 0.2 16 0.8 83.5 1.5 0.8 1.5 0.1 0.5 0.2 2.5 1.0 0.2 1.6 2.2 4.6 31.5 0.4 80.0 2.9 2.9 1.5 0.8 3.9 4.1 3.2 0.6 2.2 5.7 4.6 3.7 63 0.6 81.8 4.3 2.1 1.5 0.4 3.4 4.3 2.4 1.0 2.6 4.1 3.7 4.0 125 0.4 80.2 2.2 2.0 1.6 0.1 2.0 3.8 2.1 0.6 1.2 2.1 2.4 2.5 16 0.3 77.1 0.6 0.2 0.2 0.3 0.1 1.7 1.2 2.0 1.2 3.5 2.8 5.0 31.5 1.4 72.1 1.7 2.1 0.3 1.5 3.4 4.8 2.1 1.0 2.8 6.5 4.6 2.6 63 2.5 75.3 2.9 0.6 0.5 1.7 4.2 1.8 2.2 0.9 2.4 2.2 2.0 3.6 125 0.2 73.6 0.1 1.0 0.7 0.8 1.2 0.4 0.7 0.6 0.5 0.3 0.2 0.1 16 0.3 80.8 1.2 5.0 6.3 5.1 1.3 1.2 1.1 4.0 0.1 0.3 0.5 2.7 31.5 2.3 73.2 1.5 0.5 5.9 2.0 5.4 3.1 3.7 5.1 4.2 4.3 3.0 1.3 63 3.4 72.6 1.9 0.7 1.7 2.3 3.5 0.7 0.2 0.2 3.0 1.1 1.0 0.1 125 1.5 70.6 0.3 0.5 2.0 0.3 2.0 1.1 0.2 0.1 0.7 0.6 0.7 1.1 16 0.2 80.7 0.9 0.2 0.6 0.9 0.8 1.3 1.9 2.5 1.6 3.1 3.9 6.2 31.5 0.6 77.0 1.2 2.3 0.3 0.8 3.0 5.1 2.1 0.7 4.0 6.8 6.5 4.9 63 1.6 78.2 2.1 0.9 0.9 0.6 2.7 2.7 1.7 0.9 2.9 2.8 3.4 4.0 125 0.4 79.3 1.2 1.3 1.4 0.8 0.3 0.4 0.2 1.7 1.0 0.3 0.6 0.3 16 0.2 84.1 1.7 5.2 6.5 5.4 1.1 0.6 2.1 4.8 0.2 0.9 0.6 3.4 31.5 2.6 79.8 0.1 3.3 8.4 4.1 5.2 2.2 2.4 5.0 2.3 1.1 2.5 1.0 63 3.5 75.5 1.7 1.4 3.0 1.7 4.1 0.6 0.6 0.6 2.3 0.8 0.8 0.4 125 1.1 75.8 3.3 0.5 0.5 0.2 1.7 1.8 0.5 0.5 0.1 1.2 1.0 0.6 4.0 Inch Pressure Drop 2.5 Inch Pressure Drop 1 Inch Pressure Drop 35% WOF 60% WOF 35% WOF 60% WOF 80% WOF 60% WOF 80% WOF 90% WOF * Configuration 1, 1D, is given as the reference. All levels are db difference. (Positive value indicated increases sound level.) Dark gray cells are noise increases greater than 2 db and light graycells are noise decreases by greater than 2 db. Published in Vol. 114, Part 2 35

2008, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). ESL-PA-08-06-09 3. Redesign the fan and motor support structure to reduce vibration. 4. Study additional fans and select fans with lower tone noise. An approach that has been successful in cooling fan noise is to use measure flow data to verify and tune computational fluid dynamic (CFD) models of the air flow and then use the CFD models to evaluate fan designs. CFD has been demonstrated for HVAC design (Herrin et. al. 2007). Measuring the complex three dimensional flow is a very difficult task and would require advances in some of the available full field diagnostic tools, however there has been some success with using Particle Image Velocimetry (PIV) (Laage 2005) to study the complex flow at the outlet of a blower fan. Combining CFD with measured flow data is an approach that generates confidence in the CFD results and then allows the three dimensional spatial detail available of the CFD results to be used to better identify and understand the key flow phenomena that are dominating the rumble noise. The second recommendation is that the experimental configuration be redesigned so that experiments are designed to study two issues separately: 1) the aerodynamic noise generated from the fan outlet to elbow, and 2) the transmission of low frequency noise down the duct and the breakout noise in a room. The impact of these two issues was likely confounded in the results presented for this study. Therefore, it is recommended that two experimental setups are considered. In one case measurements of the sound power inside the duct is measured, where the duct is rigid down stream of the elbow. And in a second configuration, a typical duct is used down stream of the elbow and the break out noise is measured, as was done in this study. The third recommendation is that the experimental setup be changed to reduce the tonal noise. One cause of the tone noise was the flexibility in the structure supporting the fan and motor over a range of heights from on the ground to over 5 feet in the air. This was done so that the outlet duct was kept fixed and in a typical configuration in the measurement room. Therefore it is recommended that the future configuration be with a system rotated 90 degrees, so that the motor and fan are mounted near the floor and are moved along the floor, rather than raised, to increase the distance between the fan outlet and elbow. Such a system could be designed with the fan and motor support secured rigidly to the floor at each position of the fan outlet to elbow distance. A successful design will likely reduce the need for the tone removal. The fifth recommendation is that all these changes be implemented in a study of additional fans to identify the variation between fans and the degree to which the data obtained here provides a general characterization of the fan discharge design issues. In particular, fans with low tonal noise components should be studied in order to further reduce or eliminate the need for the tone removal. REFERENCES Herrin D.W., Tao Z., Carter A.E., Liu J., and Seybert A.F., Using Numerical Methods to Analyze Multicomponent HVAC Systems (RP-1218), 2007, Vol. 113, pt 1 Kading, J, Qualification of Duct Rumble Noise Test System MS thesis, Iowa State University, 2006. Kading, J., Mann, J.A., and Pate, M.P., Qualification of Fan Generated Duct Rumble Noise: Part 2: Results (Rp 1219),, submitted for publication in 2008 Annual Meeting. Laage, J.W., Mahendra, P., Melzer, P.J., Olsen, M.G., Mann J.A., Yarbough, D.,Yu, D., Measurement Tools for Studying Fan Noise, Proceedings of. Noise-Con 2004, pp 593-601 Laage, J.W., Armstrong, A.J., Eilers, D.J., Olsen, M.G., and Mann J.A., Air Flow Measurement Techniques Applied to Noise Reduction of a Centrifugal Blower, Proceedings of Noise-Con 2005, Minneapolis, Minnesota, October 17-19, 2005. 36 Published in Vol. 114, Part 2

ESL-PA-08-06-09 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Published in Vol. 114, Part 2