Home Page Family Images OurHome Norway Apollo 1 Apollo 2 HamRadio
RETURN TO INTRODUCTION AND NOTES

APOLLO "DC" COOLING FAN NOISE
Impact of the noise at our home located a few Kilometers northwest of Apollo

The ESKOM "DC" converter station Apollo is located on the southern border of the Rietvlei Game Reserves about a Kilometer south east of our farm Salbu. 

    Note: ESKOM purchased a portion of the farm Witkoppies and constructed the station Apollo a number of years after we had established our home in this area. Eskom has undertaken to rectify the matter before the DC section of the station Apollo goes back online.
Information about the fans:  

Valve cooling fans, 48x 2.5 meter diameter fans each with 6x Blades, 37KW motor 1470 RPM, pulley ratio 710/160 = 4.43, fan speed 331 RPM, or 5.52 revs per second, Base Noise or Blade Repetition Rate (BLR) will be 6x5.52= 33.1Hz. 
 

Why is this incessant low level "throbbing sound" so debilitating?  

We are coming to the conclusion that it is the random beat note (circa 4 Hz per second) - arising out of the multiple (circa 33 Hz) pure tones generated by the battery of large cooling fans - that is causing us distress and that the impact of this noise cannot be simply defined by the level of the tones and/or the spectral balance of the noise. It is the long term nighttime exposure to this incessant "throbbing sound", with an irregular beat, that has caused us to become "sensitised" to this debilitating, low level, "modulated" VLF noise. 

Low frequency pure tones, just at or above the threshold of hearing, have been found to cause extreme annoyance. (see:- Broner, 1978b: Tempest 1979: Chatterton, 1979: Leventhall, 1980.) 


1- Three in One Image - DC fans On: These three images show the wave form and the spectrum of the LF noise from 9 of the 48 "DC" cooling fans as measured indoors at Salbu. There are 48x 2.5 meter diameter "DC" cooling fans at Apollo. These cooling fans are vertical facing and are arranged in 6 banks. Each bank has 8 fans positioned under oil filled radiators. Each fan is driven by a 37KW motor and had a blade repetition rate of 33Hz. The total fan power is 1.8 Megawatts and the cooling capacity of the fans is in excess of 10 Megawatts. Only 9 of the 48 fans were running when this test was conducted. With the B&K Sound Level Meter set to "fast" and using the 31Hz filter the 33Hz noise pulses were peaking to a level of 55dB. The noise pulses were peaking 20dB above the office ambient noise level of 35dB. 

2- Waterfall Image - DC fans On : This image illustrates the "pulsating" or "throbbing" characteristic of the Apollo noise. There are 48x 2.5 meter diameter "DC" cooling fans at Apollo. 

3- Waterfall Image - DC fans Off: This image illustrates how the characteristic "pulsating" or "throbbing" noise stops when the "DC" fans at Apollo are switched off. The residual noise is at a level of 35dB and is "shaped" by the passband of the B&K 31Hz filter. 

4- Spectrum Image - DC fans On: This image shows the 33Hz Apollo "DC" fans noise in relation to the Low Frequency Noise Rating (LFNR) Curves - SEE PAPER BY BRONER AND LEVENTHALL. This spectrum was recorded indoors with all doors and windows closed. The B&K sound level meter was set for linear. 

5- Spectrum Image - DC fans On: This image shows the 33Hz Apollo "DC" fans noise at a level of 20dB above the low frequency ambient noise. This spectrum was recorded using the B&K sound meter "linear" setting. 


A number of photographs follow this text - the images may take a few minutes to download. 

APOLLO NOISE SPECTRUM - RECORDED AT SALBU
ApolloL1.gif
IMAGE - ApolloL1 : Three in One Image - DC fans On: These three images show the wave form and the spectrum of the LF noise from 9 of the 48 "DC" cooling fans as measured indoors at Salbu. There are 48x 2.5 meter diameter "DC" cooling fans at Apollo. These cooling fans are vertical facing and are arranged in 6 banks. Each bank has 8 fans positioned under oil filled radiators. Each fan is driven by a 37KW motor and had a blade repetition rate of 33Hz. The total fan power is 1.8 Megawatts and the cooling capacity of the fans is in excess of 10 Megawatts. Only 9 of the 48 fans were running when this test was conducted. With the B&K Sound Level Meter set to "fast" and using the 31Hz filter the 33Hz noise pulses were peaking to a level of 55dB. The noise pulses were peaking 20dB above the office ambient noise level of 35dB. 
ApolloL2.gif
IMAGE - ApolloL2 : Waterfall Image - DC fans On: This image illustrates the "pulsating" or "throbbing" characteristic of the Apollo noise. 
ApolloL3.gif
IMAGE - ApolloL3: Waterfall Image - DC fans Off: This image illustrates how the characteristic "pulsating" or "throbbing" noise stops when the "DC" fans at Apollo are switched off. The residual noise is at a level of 35dB and is "shaped" by the passband of the B&K 31Hz filter. 
ApolloL4.gif
IMAGE - ApolloL4 : Spectrum Image - DC fans On: This image shows the 33Hz Apollo "DC" fans noise in relation to the Low Frequency Noise Rating (LFNR) Curves - SEE PAPER BY BRONER AND LEVENTHALL. This spectrum was recorded indoors with all doors and windows closed. The B&K sound level meter was set for linear. Clearly any attempt to balance the spectrum by the introduction of a "shaped" noise spectrum would necessitate increasing the noise level by at least 20dB and as much as 30dB in parts of the spectrum. The resultant "balanced" noise level would be very high indeed and totally intolerable in the bedroom area.
ApolloL5.gif
IMAGE - ApolloL5 : Spectrum Image - DC fans On: This image shows the 33Hz Apollo "DC" fans noise at a level of 20dB above the low frequency ambient noise. This spectrum was recorded using the B&K sound level meter set for linear. 
ApolloL6.gif
IMAGE - ApolloL6 : xxxxxxxxx 
ApolloL6B.gif
IMAGE - ApolloL6B: xxxxxxxxx 
ApolloL7.gif
IMAGE - ApolloL7 : xxxxxxxxx 
ApolloL8.gif
IMAGE - ApolloL8 : xxxxxxxxx 
ApolloL9.gif
IMAGE - ApolloL9 : These two images shows the spectrum of the noise as recorded outside the Salbu office on a morning when the tonal noise could not be heard - the Apollo fan noise was present but was below the threshold of hearing. The 33 Hz and 66 Hz fan (BRR) noise components can be seen. There other "unknown" VLF noise signals present. A 127 Hz signal can be seen on the spectrogram image is off-scale in this particular spectrum analyser image but the 127 Hz signal can be seen in the image headed "Spectrum Image - DC fans On". As one would expect - due to the multiple fans and propagation conditions - the level of the fan noise components changes all the time. 
ApolloB1.gif
IMAGE - ApolloB1 : xxxxxxxx
The above B&W "all in one" image is suitable for printing and for sending by fax
LFNRandMODULATION.gif
IMAGE - R1 : RC Mark II NOISE RATING CURVES FOR SALBU 

The above diagram illustrates the spectrum imbalance in our home caused by the low frequency tonal noise radiated by the bank of large cooling fans at the ESKOM station Apollo. The curves are plots of the noise level measured in the B&K sound level meter's octave band filters.  

The yellow and green curves are examples of a reasonably balanced noise spectrum - the blue and purple curves illustrate an imbalance spectrum. A perfectly balanced noise spectrum would follow the profile of one of the RC curves. Click to view typical Salbu noise profiles: A - when the R21 Freeway is Busy and B - when the R21 Freeway Not Busy. In both examples the "33 Hz pulsating tonal noise" from the Apollo DC Fans is the dominant noise signal.

Note: The red RC curves have been added for guidance purposes only are not included in the RC Mark II paper as the RC Mark II method does not address very-low ambient noise conditions typical of performing arts and assembly spaces for worship, or special facilities such as recording studios..  

DVL Notes:  

1_ The Apollo noise is tonal 
2_ The Apollo noise fluctuates dramatically in level. (100% modulated at a rate of 2-12 Hz) 

Question: What correction factor needs to be applied to take these factors into account? 

See section 5 headed "LIMITATIONS IN APPLICATION OF THE RC RATING METHODOLOGY" in the following section. 
 

Extracted from Warren E. Blazier, Jr's paper: 

 "............. LF frequency range, there is evidence that people are more sensitive to spectrum unbalance in this region than in others.  One reason may be that the large amplitude fluctuations characteristic of aerodynamically-generated low-frequency noise adds a dimension to the subjective assessment process that is inadequately accounted for in time-averaged RMS data.  For example, the equal-loudness contours are bunched closer together at low-frequencies, and the perception of this amplitude fluctuation would be quite varied and this draws attention!  Furthermore, at low sensation levels the loudness growth exponent is greater than at higher sensation levels, and thus the perception change would be even greater.  In some situations at only moderate levels, it is also possible that the negative reaction to excess low-frequency noise may be subliminal and not noticeable during a short-term exposure period." 
 


Conclusion - Salbu 

We are coming to the conclusion that it is the random beat note (circa 4 Hz per second) - arising out of the multiple (circa 33 Hz) pure tones generated by the battery of large cooling fans - that is is causing us distress and that the impact of this noise cannot be defined only by the level of the tones and/or the spectral balance of the noise. It is the long term nighttime exposure to this incessant "throbbing sound", with an irregular beat, that has caused us to become "sensitised" to this debilitating, low level, "modulated" VLF noise. 

NB: Low frequency pure tones, just at or above the threshold of hearing, have been found to cause extreme annoyance. (see:- Broner, 1978b: Tempest 1979: Chatterton, 1979: Leventhall, 1980.) 
 

 
NOISE RATING CURVES 
 

RC Mark II;  a refined procedure for rating the noise of heating, venti-lating and air-conditioning (HVAC) systems in buildings - (Submitted July 1997) - Warren E. Blazier, Jr.) 

Since the introduction in 1981 of the RC procedure for rating heating, ventilating and air-conditioning (HVAC) system-related noise in buildings, sixteen years of practical experi-ence in applying the methodology to develop a relationship between objective measure-ments and the subjective response of room occupants has shown that certain refinements are needed to improve the correlation.  These refinements include a modification to the shape of the RC reference curves in the octave-band centered at 16 Hz, an improvement in the procedure for assessment of sound quality and the development of a scale to estimate the magnitude of subjective response as a function of spectrum imbalance.  The refined methodology discussed in this paper is based on these refinements, and is identified as the RC Mark II procedure for rating the noise of HVAC systems in buildings.  The American Society of Heating, Refrigerating and Air-Conditioning Engi-neers (ASHRAE) has adopted this refined methodology, and plans to recommend it as the preferred procedure for HVAC system sound rating in the next edition of the ASHRAE Applications Handbook to be published in 1999. 

Primary subject classification:  69;  Secondary subject classification:  51 
 

1 INTRODUCTION: 

Since the introduction in 1981 of the RC procedure for rating heating, ventilating and air-conditioning (HVAC) system-related noise in buildings1, sixteen years of practical experience in applying the methodology to develop a relationship between objective measurements and the subjective response of room occupants has shown that certain refinements are needed to im-prove the correlation.  Four shortcomings have been identified in this respect: 
1)  The RMS sound pressure level corresponding to the octave-band centered at 16 Hz in the family of RC reference curves is too permissive in that it underestimates the subjective re-sponse of a typical room occupant to the large fluctuations in low-frequency noise level ob-served in many HVAC systems in buildings today. 
2)  Hissy-sounding spectra can not be predicted reliably from objective measurements if the levels in the low- and mid-frequency range lie below those of the RC reference curve.  (The original methodology addresses only those levels equal to, or in excess of the reference RC curve). 
3)  A spectrum with a “roar” sound quality can not be predicted from objective measurements because the assessment of spectral deviations is based upon only two grouped frequency ranges, which overlap in the region corresponding to the roar subjective attribute. 
4)  The original assessment of spectrum character was principally qualitative and no basis was established for estimating the magnitude of subjective tolerance for imbalanced spectra. 
 The refinements made to the original methodology, which are the subject of this paper, are based upon not only the recognition of the shortcomings listed above, but also upon new re-search sponsored by the American Society of Heating, Refrigerating and Air-Conditioning Engi-neers (ASHRAE)2.  The refined methodology is identified as the RC Mark II procedure for rating HVAC system-related noise in buildings.  Furthermore, ASHRAE has endorsed this refined methodology, and plans to recommend it as the preferred procedure for HVAC system sound rating in the next edition of the ASHRAE Applications Handbook to be published in 1999. 
 
2.  CONSIDERATIONS IN THE DEVELOPMENT OF NOISE RATING PROCEDURES 
     For complete text see paper by Warren E. Blazier, Jr
 

3.  THE RC MARK II NOISE RATING METHODOLOGY 
       For complete text see paper by Warren E. Blazier, Jr

4.  DIAGNOSTIC CONSIDERATIONS IN ESTIMATING SUBJECTIVE RESPONSE 
      For complete text see paper by Warren E. Blazier, Jr

5. LIMITATIONS IN APPLICATION OF THE RC RATING METHODOLOGY 

The RC Mark II rating procedure is based on the use of time-averaged RMS data measured in octave-bands.  Because of this, it is principally applicable to the analysis of only broad-band sounds that do not contain audible tonal or narrow-band components.  If only octave-band data are available, the bandwidth is generally too wide to identify and quantify the presence of tonal components that may significantly alter a subjective opinion with respect to sound quality. 
One-third-octave, or narrower band analyzers are generally necessary to supplement octave-band data when analyzing problems that may have been triggered by the presence of audible tonal components.  Unfortunately, even if such data are available, at the present time there is no consensus standard for assessing and rating HVAC-related sound containing audible tonal or narrowband components
The RC Mark II method does not address very-low ambient noise conditions typical of performing arts and assembly spaces for worship, or special facilities such as recording studios.  The acoustical design of such spaces generally involves the services of a specialized consultant whose views with regard to acceptable ambient noise criteria usually dictate the requirements to be satisfied. 

6.  CONCLUSION 

The revised methodology (RC Mark II) for rating HVAC-related noise in occupied spaces has corrected several shortcomings that became evident during some 16 years in practical application of the original method introduced in 1981.  The revisions are also based on new information revealed by recent ASHRAE-sponsored research.  It is hoped that these improvements in noise rating methodology, together with the suggested criteria for interpreting the ratings to estimate the subjective response of a room occupant, will become a useful and practical tool for the design professional concerned with the control of HVAC-system noise. 
 

7.  REFERENCES 

1  W.E. Blazier, Jr., “Revised noise criteria for application in the acoustical design and rating of HVAC systems”, Noise Control Eng. 16(2), 64-73 (1981 March-April). 

2 N. Broner (VIPAC), “Determination of the relationship between low-frequency HVAC noise and comfort in occupied spaces, Objective Phase” - ASHRAE 714-RP, April 1994. 

3 C.E. Ebbing and W.E. Blazier, Jr., “HVAC low-frequency noise in buildings”, Proceedings INTER-NOISE 92, Vol. 2, pp. 767-770. 

4 W.E. Blazier, Jr., “Room noise criteria:  The importance of temporal variations in low-frequency sounds from HVAC systems”, Proceedings Noise-Con 96,, Vol. 2, pp. 687-692. 

5 W.E. Blazier, Jr., “Sound quality considerations in rating noise from heating, ventilating and air-conditioning (HVAC) systems in buildings”, Noise Control Eng. J. 43 (3), 1995 May-Jun. 

6 Kerstin Persson-Waye et al, “Effects on performance and work quality due to low-frequency ventilation noise”, Journal of Sound and Vibration, 1997, 205 (4) 467-474. 

7 N. Broner (VIPAC), “Determination of the relationship between low-frequency HVAC noise and comfort in occupied spaces, Psychoacoustic Phase”, ASHRAE 879-RP, (In progress; completion in 1998)  

8 Birgitta Berglund et al, “Sources and effects of low-frequency noise”, J. Acoust. Soc. Am., Vol. 99, No. 5, May 1996 
 
 

LFNR1.jpg
IMAGE - R2 : LFNR CURVES FOR SALBU - APOLLO NOISE LEVEL IN OCTAVE BAND FILTERS 
(not a great image but it will do for now)

Click to view typical Salbu noise profiles: A - when the R21 Freeway is Busy and B - when the R21 Freeway Not Busy. In both examples the "33 Hz pulsating tonal noise" from the Apollo DC Fans is the dominant noise signal.

Low Frequency Noise Rating (LFNR) Curves - SEE PAPER BY BRONER AND LEVENTHALL 

Note: Complaints can be expected once the 33 Hz Tonal noise level approaches the threshold of hearing as shown by the curve marked with "black squares".  (It is interesting to note that whilst at 31Hz complaints could be expected circa the threshold of hearing at 16Hz complaints can be expected some 20dB below the threshold of hearing). 

The green line defines an acceptable indoor noise profile in the bedroom section of our home. See curves 55555, 3333333 and 4443333 measured under "fans off" conditions. 

Curves 66666 defines unacceptable indoor noise profiles due to the dominance of the low frequency noise components. 

After a period of exposure to low frequency noise people become sensitised to the noise and this can result result in allergic type reactions to low frequency noise sources. 

After a period of some six months of exposure to the Apollo noise my son Gernot was found to be "incredibly" sensitive to LF noise. The audiologist said "he has the most incredibly sensitive low frequency hearing - by far the most sensitive hearing that I ever measured in a patient". 

A neighbour, situated about 1km due east of Apollo, related his experience of this "silent noise" as follows: "I was working outdoors and become aware of a strange thumping sensation in my chest and a pulsating pressure in my ears. I thought I had a medical problem until a colleague said that he was experiencing the same sensations. On checking we found that the phenomena was not related to our heart beat rates." 

See curve CCCC - noise from the ESKOM station Apollo measured outdoors on a noisy night, the noise in the 31 Hz octave filter measures in excess of 60dB. It should be noted that domestic dwelling are totally transparent to very low frequency noise. 

Summary: Within a few kilometer radius of the ESKOM station Apollo the LFNR "complaints limit" as defined by the "green line" is being exceed by nearly 20dB. This means that people living in the area are being subjected to "indoor" low frequency sound intensity levels of 100 times more than they are expected to have to tolerate.  

Sound intensity is measured in terms of Watts per square meter. It is interesting to note that a sound intensity level (SL) of one Watt per square meter corresponds to the threshold of feeling and a sound intensity level of one trillionth of a Watt corresponds to the threshold of hearing - a range of 120 dB. A trillionth is equal to one millionth of a millionth

RETURN TO INTRODUCTION AND NOTES
Apollo1S.jpg
FAN LOCATION: There is a total of forty eight 2.5 meter diameter fans. The oil cooling radiators cover most of the roof area of each of the eight concrete sheds. The eight sheds are arranged in a straight line.
Apollo2S.jpg
FAN LOCATION: The fans blow upward through the oil cooling radiators. No mufflers or cowlings have been fitted to the fans. The fan blade design is very "basic".
Apollo3S.jpg
NO IMAGE AS YET 
Apollo4S.jpg
NO IMAGE AS YET 
Apollo5S.jpg
NO IMAGE AS YET 
Apollo6S.jpg
NO IMAGE AS YET
Apollo7S.jpg
NO IMAGE AS YET 
Apollo8S.jpg
NO IMAGE AS YET
Apollo9S.jpg
NO IMAGE AS YET
Apollo10S.jpg
NO IMAGE AS YET
Apollo11S.jpg
NO IMAGE AS YET
Apollo12S.jpg
NO IMAGE AS YET
ApolloV4S.jpg
NO IMAGE AS YET 
ApolloV5S.jpg
NO IMAGE AS YET 
ApolloV6S.jpg
NO IMAGE AS YET 
RETURN TO INTRODUCTION AND NOTES
Home Page Family Images OurHome Norway Apollo 1 Apollo 2 HamRadio