1、OR-05-6-1 Assessment and Regulation of Low Frequency Noise Geoff Levent hall, P h D Member ASHRAE ABSTRACT The statistics of the low frequency hearing threshold indi- cate that about 10% of the population may have a threshold that has 1 O dB or greater sensitivity than the median (50%) threshold, in
2、dicating that low frequency noise may cause difi- culties for a small number ofpeople. Low frequency noise has been accepted in some European countries as a problem for which regulation is required and about one-fih of the enlarged EU total of 25 have national methodsfor assessment of low frequency
3、noise, some ofwhich are usedby other coun fries. The regulations are mostly limited to low frequency noise in the home, although one countty does adapt the limits to commercial premises. Limits for low frequency noise levels in homes can be an indicator of what is required in ofices and other workpl
4、aces. CAN YOU HEAR IT? Low frequency noise presents a problem of perception. Perhaps you dont hear it, but others do. Perhaps you hear it, but others dont. In order to prevent irritation and annoyance, with consequent adverse effects on comfort and productivity, a very strict noise regulation might
5、require that the low frequency noise should be below threshold. But what thresh- old? An obvious choice is the standard IS0 tone threshold (IS0 226 2003). Although the lowest frequency in IS0 226 is 20 Hz, there is good correlation in an overlapping region between IS0 226 and work on low frequency t
6、hresholds down to 4 Hz (Watanabe and Mdler 1990b), as shown in Figure 1. Experience is that the median 50% threshold does not protect everybody. This is not surprising, as the same number of people will have greater and lower sensitivity than this threshold, leading to consideration of the statistic
7、s of thresh- old determinations. An early version of IS0 226 relied heavily on Robinson and Dadson (1956) in which the standard devi- ation of the threshold (minimum audible field) measurements varied from 4.5 dB at 200 Hz to 8 dB at 25 Hz, say a typical value of 6 dB. A similar standard deviation w
8、as found by Watanabe and Mdler (1 990a) in the range 25 Hz to 1 kHz and also at lower frequencies (Watanabe and Mdler 1990b). Other workers give a comparable figure. In Japan, the stan- dard deviations of the low frequency pure tone threshold of a group of subjects was found to be about 6 dB (Kitamu
9、ra and Yamada 2002). In the Netherlands, a criterion to protect 90% of the 50- to 60-year-old age group from disturbance by low frequency noise puts the level at 10-12 dB below the 50% threshold for the group (van den Berg and Passchier-Vermeer 1999). That is, 10% of people are more sensitive than t
10、his criterion level. Assuming, a normal distribution, the 10% level is given by 1.3 standard deviations, leading to a standard devi- ation of 7-9 dB for 10% levels of 10-12 dB below the median level. Support for this is given by Zwicker and Fastl (1999). A plot of the tone threshold and 10% and 90%
11、values for their measurements on young adults (20 to 25 years old) indicates that, at the lower frequencies, these percentages are at about *l OdB from the 50% threshold, as shown in Figure 2 (Zwicker and Fastl 1999). Note that the low frequency thresholds given by Zwicker and Fastl are lower than t
12、he current IS0 standard . The 50% level is the hearing threshold for which 50% ofthe popu- lation are more sensitive and 50% less sensitive and is also referred to as the median. The 10% level is that hearing threshold, lower than the 50% threshold, for which 10% are more sensitive and 90% less sens
13、itive. The 90% level is that hearing threshold, above the 50% threshold, for which 90% are more sensitive and 10% less sensitive. Geoff Leventhall is a noise and vibration consultant, Ashtead, UK. 02005 ASHRAE. 655 -201 1 I I Qo2 Wi 2 Q5 1 2ke5 1020 frequency of test tone Figure I Low frequency thre
14、shold. threshold. The reasons for this are not known, but the IS0 threshold is a consensus developed from the results of a number of laboratories. Additionally, an IS0 standard (IS0 7029 2000) gives age- related statistics of the tone threshold for the range 18 to 70 years old, but the lowest freque
15、ncy considered is 125 Hz. At this frequency, the standard deviations for 18 year olds are 5 to 7 dB, depending on sex and whether above or below the 50% threshold. There is also an indication that the standard devia- tion is increasing into the lower frequencies. Thus, the pointers are to a standard
16、 deviation of around 6 dB, perhaps more, in the individual pure tone thresholds.* The distribution curve is two sided, as there are likely to be more people with hearing loss than those with acute hearing, although hearing loss does not occur at low frequencies as readily as it does at high frequenc
17、ies. This is taken into account in IS0 7029 by defining different standard deviations for above and below the 50% threshold. For example, for 18-year- old males at 125 Hz, the standard deviation is 7.23 dB when the threshold is above the 50% level and 5.78 dB when below it. IS0 7029 also gives infor
18、mation on how to calculate age- related standard deviations, which increase with age. Assuming a normal distribution and a conservative figure of 6 dB standard deviation, 16% of us will be greater than 6 dB more sensitive than the median, 10% will be greater than 8 dB more sensitive, and 2% will be
19、greater than 12 dB more sensi- tive.3 That is, one in fifty people are expected to have a hearing 2. Note that threshold determinations by earphone listening or in small pressure chambers, using experienced subjects, have shown a lower standard deviation than 6dB. Whittle, L. S., Collins, S. J., and
20、 Robinson, D. W. 1972. The audibility of low frequency sounds. Jnl Sound fibn, 21(4): 431-448. Yeowar, N. S. and Evans, M. J. 1974. Thresholds of audibility for very low frequency pure tones. JAcoust Soc Am, 55(4): 814-818. Figure 2 Median hearing threshold ofyoung adults (50%) and 10% and 90% thres
21、holds (test tone level is dB re 20 pa). threshold 12 dE3 or more below the median. The one in fifty will include some at the tail of the distribution, who are very sensitive. Additionally, it is known that the equal loudness contours become closer together at low frequencies, so that the mid-frequen
22、cy rule of thumb that a 10 dB change corresponds with a doubling of loudness does not apply. At very low frequencies, about 3 dB increase may result in a doubling of loudness, leading to a rapid growth in annoyance (Mdler 1987). While the statistics may not be precise, they do serve to give an indic
23、ation of the range of sensitivities and explain why there may be only one complainant in a group of office work- ers. Perception is an essential, but not the only, precursor to complaint. Personal, situational, and other factors are also involved, while a persistent complainant may make others aware
24、 of a low-level noise that they had not previously perceived. If, say, 2% of people are very sensitive to low frequency noise, one might ask why there are not more complaints. This is because a complaint anses only when a sensitive individual is presented with a low frequency noise in a situation, s
25、uch as work or home, from which they cannot remove themselves. In developing protection criteria, a decision has to be made on what percentage of a population to protect, taking into account that half the population will be more sensitive than the median. But before criteria can be implemented, it i
26、s necessary to measure the noise. PROBLEMS OF MEASUREMENT Noise levels vary in different locations in an office. In a small office, higher fkequencies decrease in level with distance In the normal distribution, one standard deviation below the 50% level includes 34% of those below this level. Two st
27、andard devi- ations includes 48%. 3. 656 ASHFiAE Transactions: Symposia Table 1. Room Modes Up to 1OOHz Hz 28.6 42.9 51.5 57.2 68.6 71.5 74.3 80.9 85.7 85.7 85.8 89.3 nx 1 O 1 2 O 2 1 O O 3 1 2 O 1 1 O O 1 O 1 2 O 1 O nY nz O O O O 1 O 1 1 O O 1 1 90.4 95.9 1 3 2 1 O O from a terminal, while lower f
28、requencies have the potential to excite room modes. For example, a room 6 m by 4 m by 2.5 m has low frequency room modes, shown in Table 1, where the x dimension of the room is 6 m, the y dimension 4 m, and the z dimension is 2.5 m. Excitation of these resonances leads to spatial variations in the s
29、ound level. Conversely, there are fluc- tuations in level at a fixed point in a room as the frequency is varied. If the low frequency excitation is due to breakout from a duct run above an office, this is a line source, in contrast to a terminal unit that approximates to a point source. The line sou
30、rce leads to a fairly uniform level in the room at the low frequency breakout. Where it is required to determine the sound level in aroom at low frequencies, it is necessary to make a number of measurements at different positions in order to obtain an aver- age. Procedures for carrying out room aver
31、aging are given in IS0 140, Part 4 (IS0140 1998)and in IS0 16032 (IS016032 2000), although these may need to be adapted for lower frequencies, Alternatives are to measure at the work position of a listener or find the highest noise level in the room The choice of measurement method depends, to an ex
32、tent, on the purpose of the measurement: whether it is to check contract conformity or to identi a problem. PROBLEMS OF ANALYSIS Any analysis is always an average over time. It may be a short time, say 100 ms, or a long time, say 10 minutes. Measurements of HVAC noise are normally averaged over minu
33、tes, rather than milliseconds, and the result related to a criterion. This assumes that human response can be quantified by an average over time, rather than by what is experienced instant by instant, and that short-term fluctuations in the noise level are consequently not important. This is the “no
34、ise dose” concept, which is acceptable for fairly steady noises but which we may question when fluctuations are a major characteristic of a noise. It has been shown that fluctuations in level of a few times a second increase the annoyance of HVAC noise (Brad- ley 1994). EU CRITERIA Some EU countries
35、 have national criteria for low frequency noise, which give limits, authorized by the appro- priate authorities, for use in those countries. Assessment by non-authorized methods will undermine the credibility of the outcome. The criteria are based either on exceedance of a criterion curve at any one
36、-third octave frequency or on a single figure measurement from a low frequency weighting curve. A summary of the methods is given by Poulsen (2002) and Poulsen and Mortensen (2002). All methods, except the Danish one, are restricted to disturbance in dwellings. Denmark The noise is measured in one-t
37、hird octave bands, aver- aged over 10 minutes, and the A-weighted summation deter- mined for the 13 bands in the range 10 Hz to 160 Hz, giving a numerical value, Lpa, (Jakobsen 2001). Limit values for L pa,LF are: Dwelling: Day 25 dB. Eveninflight 20 dB Offices: 30dB Here, Day is 07.00 to 18.00 hour
38、s, and Evening /Night is 18.00 to 07.00. Limits are reduced by 5 ds if the noise has an impulsive character. The 20 dB limit will be exceeded if a single one-third octave band is greater than this limit. Alternatively, the limit will also be exceeded if the summation of all the one-third octave band
39、s in the assessment is greater than 20 dB. This will be given, for example, by a continuous spectrum that lies parallel to the limit values and has levels 11 dB below these values, since 101og13 = 11. The 30 dB limit for commercial premises will be exceeded, for example, at 3 1.5 Hz by a tone of 69
40、dB level, as the A-weighting correction at 3 1.5 Hz is -39 dB. The IS0 226 median tone threshold at 31.5 Hz is 59.5 dB, while RC 35 Mark II is 60 dB at the 3 1.5 Hz octave band. Poland This method uses the frequency range 10 Hz to 250 Hz (Mirowska, 2001). The sound pressure levels of the one-third o
41、ctave bands of the noise are compared with a reference curve, LA, ,derived from LA10 = 10 - kAi where kA is the value of the A-weighting for the center frequencies of the one-third octave bands. Thus, at 100 Hz where k, = -19.1 dB, the crite- rion level is 29.1 dB. Any single frequency that meets th
42、e crite- rion curve will have a level of 1 O dBA. Background noise is taken into account by determining the difference between the sound pressure levels of the noise and the background noise. Consequently there are two compo- nents in the assessment: ALl-the difference between the measured sound pre
43、s- sure level and the LALO curve. ASH RAE Transactions: Symposia 657 AL2-the difference between the sound pressure levels of the noise and the background noise. The noise is considered to be annoying when AL,O AL2 1 O dB for tonal noise or 6 dB for broadband noise Sweden A criterion curve is in the
44、range 3 1.5 Hz to 200 Hz, which is not to be exceeded at any one-third octave frequency by the averaged value of the noise (Socialstyrelsen-Sweden, 1996). Netherlands A criterion for audibility is defined in the range 20 Hz to 100 Hz, based on the hearing threshold of the 10% most sensi- tive, unsel
45、ected listeners in the 50 to 60 year age group, which, depending on frequency, is 2 to 5 dB lower than the 50% level for otologically selected young adults. The age group 50 to 60 years was chosen because this is the age that generates most complaints of environmental low frequency noise (NSG 1999;
46、van den Berg et al. 1999). Germany Initial assessment is by the C-weighted-A-weighted difference (DIN:45680 1997). If this is greater than 20 dB, the noise is investigated further as a potential low frequency noise problem. Tonal noise is defined as a one-third octave band, which is 5 dB or more abo
47、ve adjacent bands. The equivalent level (Leq) of the tone is compared with an assumed hearing threshold curve, which is about 2 dB lower than IS0 226 and extended to one-third octave bands down to 8 Hz by rising at 8 dB/third octave. The following limits are applied for exceed- ance of the threshold
48、 curve by either the Leq or maximum value of the noise: Day: 8 Hz to 63 Hz: Leq 5 dB. Maximum 10 dB 80 Hz: Leq 10 dB Maximum 15 dB 100 Hz: Leq 15 dB Maximum 20 dB Night. All levels reduced by 5 dB. For nontonal noise, the one-third octave levels of only those bands that exceed the threshold curve ar
49、e A-weighted and summed over the range 10 Hz to 80 Hz. Limits are Day: 35 dB, Night: 25 dB. The limits for maxima are 45 dB and 35 dB. 40 20 O Frequency Hz Figure 3 Comparison of criteria. Decibel levels are in one- third octave band re 20 pPa except for ISO, which is pure tone level. COMPARISON The criteria are compared in Figure 3. In this, the Danish criterion is shown as a “20dBA” curve, but this is a weighting curve, not a limit curve as for the other criteria. Additionally, the Danish criterion for office premises is 10 dB higher than the night criterion shown. Figure 3 also