- Facilities & Capabilities
- Auditory Hazard Assessment Algorithm for Humans (AHAAH)
- Uniqueness of the Albuquerque data set
The uniqueness of the Albuquerque data set and "Evaluation of impulse noise criteria using human volunteer data"
[J. Acoust. Soc. Am 110, 1967 - 1975 (2005)] by Chan et al.
G. Richard Price
U.S. Army Research Laboratory,
Human Research and Engineering Directorate,
Aberdeen Proving Ground, MD 21005-5425
The data set of hearing loss and impulse noise exposure from tests sponsored by the US Army Medical Research and Development Command has been used by Chan et al. [J. Acoust. Soc. Am 110, 1967-1975 (2001)] to evaluate current impulse noise criteria. They correlated peak pressure of impulses in the free field with changes in hearing sensitivity in subjects wearing a circum-aural muff whose seal that had been compromised with eight tubes to reduce low frequency attenuation (to simulate fit that might be achieved in the field). If such an analysis is to be useful in revising damage risk criteria (DRC) for impulse noise, one must implicitly presume that the muff is broadly representative of hearing protectors, which are normally linear with respect to amplitude. The unanticipated effect of the tubes, not noted in the article by Chan et al., was that the muff became nonlinear, attenuating much better at higher intensities, by between 4 and 11 dB depending on the exposure condition (the range of peak pressures was just 18 dB in any one exposure condition). In contrast, an intact muff (measured for one condition) became worse by 5 dB. The uniqueness of the muffs in these tests creates a problem for any method of rating hazard that uses the characteristics of the incident waveform. The calculation by Chan et al. is therefore not generalizable outside the specific data set and should not be used to guide the revision of impulse noise DRCs.
PACS numbers: 43.50.Qp, 43.50.Pn, 43.66.Vt
In their analysis of impulse noise hazard using human volunteer data, Chan et al (2001) stated that the purpose of their work was to guide future revision of the world's damage risk criteria (DRCs). This letter is written to point out unique conditions in the data set that they used, which are not noted in their article. The effect of these conditions is non-trivial and anyone intending to make a similar use of this data set or the specific analysis by Chan et al. should be aware of the limits imposed by the acoustic peculiarities in the particular data set(s) they used.
II. THE TESTS.
Chan et al. analyzed the data from tests with human volunteers sponsored by the US Army Medical Research and Materiel Command (Johnson, 1994; 1997 and Patterson, Mozo and Johnson, 1994). Because the studies were conducted on a site near Albuquerque, they have become known as the Albuquerque studies. In these tests the subjects were exposed from one to as many as 100 impulses from explosive sources (peak pressures in the high 170's to 195 dB). Subjects wore hearing protection for their exposure to one of four different exposure conditions: free field impulses with A-durations of about 0.8, 1.5, and 2.8 msec and an impulse in a reverberant space (B-duration about 300 msec). For each of the exposure conditions, pressures were raised in six steps of 3 dB, giving a total range of exposure of 18 dB in the incident waveform. Somewhat different pressure ranges were used for the different impulses.
The analytical technique employed by Chan et al. used logistic regression to establish the relationship between peak incident pressure and onset of change in hearing sensitivity. Implicit in this approach is the assumption that no other important conditions systematically influenced the test results. It is commonly assumed that hearing protection is essentially an attenuator that may vary in its effect at different frequencies but which is essentially linear with respect to amplitude. If this were in fact true, then analyses using pressure measured outside the protector would prove useful in predicting hazard for impulses that were similar to those used in the tests they analyzed. It is also implicit in such an approach to hazard analysis that the hearing protector worn during the tests being analyzed must be generally representative of hearing protective devices (HPDs) likely to be worn in the hazardous conditions.
III. THE PROBLEM OF THE PROTECTOR
The central issue raised in this letter is that the attenuation of the HPD worn by the subjects (a circum-aural muff) was not in any sense representative of HPDs. In order to simulate 'field fit,' the experimenters compromised the seal on the right muff (the one at normal incidence to the impulse source) by inserting eight plastic tubes entirely through the seal (2.3 mm inside diameter, four open to the front and four open to the rear). In the exposures, the right ear faced the incident waveform, so the wave front crossed the openings in the seal. The leaks in the seal were intended to simulate a bad fit and perhaps more closely replicate fit during field conditions. The result, as anticipated, was that the muff provided essentially no low-frequency attenuation. The unanticipated problem was that although the holes installed in the seal did eliminate low frequency attenuation, they also created a protector that was nonlinear with respect to amplitude. Furthermore, the size of the nonlinearity was different for each type of impulse in the experiment.
1. Measures of Attenuation
Fortunately, physical ear attenuation tests had also been performed as part of the Army's experiments. The pressures in the free field and the pressures inside the muff as worn by the subjects had been measured at the levels actually used in the exposures (Patterson, Mozo, Gordon, Canales and Johnson; 1997; representative waveforms have also been made available on a CD). The pressure data could be analyzed a number of ways; but if we simply take the A-weighted energy in the incident waveform and subtract from it the A-weighted energy under the muff, we can arrive at an attenuation for the muff, for that impulse and at that peak pressure. The calculated attenuations are presented in Fig. 1.
Figure 1. Attenuation of the muffs used in the Albuquerque studies for three different exposure conditions displayed as a function of the test level (in dB with respect to the lowest exposure level in the tests). (The base level(s) were 178 dB (1M and 3M) and 173 dB (5M). By way of comparison, there is also a line for a muff with an intact seal for the 5M condition. The designation of conditions by "M" is a convention from the Albuquerque studies and refers to the distance from the impulse source. The A-durations of the impulses increased with distance from the source.)
There are several interesting points in Fig. 1. First, rather than having constant attenuation as a function of level, the altered muffs all attenuated better as the pressure rose. During the experiments, head position was rigidly controlled, so the attenuations in Fig.1 are good estimates of the attenuation received during the experiments. For the 1M condition the attenuation rose 11 dB, for the 3 M condition, 5 dB and for the 5M condition, 4 dB. The implication is that at the level of the ear canal entrance under the muff, the real point of input to the ear, the exposures did not increase 18 dB, as might have been expected, but only between 7 and 14 dB.
Secondly, the intact muff, properly fitted, lost 5.4 dB of attenuation at the highest levels tested for the 5M condition. In this case, the energy under the muff would have increased 23.4 dB instead of the expected 18 dB, if the muffs had been linear. In the Albuquerque experiments, then, depending on the specific conditions, the pressure under the muff rose anywhere between 7 and 23 dB, for a nominal 18 dB change in the peak pressure of the incident waveform. The variation in attenuation of the muffs is almost as large as the range of exposures in the Albuquerque studies!
The unusual behavior of the earmuffs in these tests is counter to the normal expectation that HPDs will be linear. It should be noted that the pressures used in these studies were very high indeed (195 dB peak for the 1M and 3M conditions). With impulses at these levels, the forces imposed by the acoustic wave are considerable, with the result that the muffs can and did move during the exposure (Johnson, 1994). It is still reasonable to assume that at levels below 180 dB or so, most protectors will probably behave linearly. However, it would be prudent to verify attenuations with physical measures rather than making assumptions about them.
We can speculate that the mechanism responsible for the nonlinearity may have been that the acoustic flow through the tubes inserted to defeat the seal of the muff may have become non-linear as the pressure rose, much as the non-linear elements in protectors that are designed to become nonlinear. (Hamery, Dancer, Evrard, 1997).
The Albuquerque data set includes a nonlinear hearing protector, which is not characteristic of hearing protectors in general. Hence, any analysis, such as the one performed by Chan et al (2001) in which only the incident pressures were examined, would be insensitive to sizable anomalies in attenuation of the HPDs and hence the stimulation actually reaching the ear. It follows that it would be inadvisable to attempt to use the values from that paper in the revision of a DRC for impulse noise in spite of the fact that the logistic regression analysis produced a good fit the data set. Any analytical technique aimed at being used for the establishment or revision of a DRC needs to take specific account of the pressure history actually reaching the ear. Given the (potentially) complex behavior of hearing protectors at very high pressures, it also follows that HPD performance needs to be evaluated for the specific conditions of stimulation.
Chan, P. C., Ho, K. C., Kan, K.K., Stuhmiller, J. H. and Mayorga, M. M. (2001) "Evaluation of impulse noise critieria using human volunteer data", J. Acoust. Soc. Am. 110, 1967-1975.
Hamery, P., Dancer, A., and Evrard, G. (1997). "Etude et realization de bouchons d'orielles perfores non lineares" Rapprt R 128/97 Institut Franco-Allemand de Recherche de St. Louis 18 Novembre 1997
Johnson, D. L. (1997). "Blast overpressure studies", Final Task Report for US Army Medical Research and Materiel Command, Contract No. DAMD17-93-C-3101.
Johnson, D. (1994). "Blast overpressure studies with animals and men: A walk-up study", USAARL Contract Report No. 94-2, U.S.Army Aeromedical Research Lab, Ft. Rucker, AL 36362-0577
Patterson, J. H. Jr., Mozo, B. T., and Johnson, D. L. (1994). "Actual effectiveness of hearing protection in high level impulse noise", USAARL Report No. 94-48, U.S. Army Aeromedical Research Laboratory, Fort Rucker, AL 36362-0577
Patterson, J. H., Jr., Mozo, B. T., Gordon, E., Canales, J. R. and Johnson, D. L. (1997) "Pressures measured under earmuffs worn by human volunteers during exposure to freefield blast overpressure", USAARL Report No. 98-01, U.S. Army Aeromedical Research Laboratory, Fort Rucker, AL 36362-0577