Philemon C. Chan, Ph.D.

1
Evaluation of Impulse Noise Criteria Using Human
Volunteer Data
2997-17

• Philemon C. Chan, Ph.D.
• Kevin H. Ho
• Kit K. Kan, Ph.D.
• James H. Stuhmiller, Ph.D.
• JAYCOR
• Maria A. Mayorga, M.D.
• WRAIR

International Military Noise Conference Baltimore,
Maryland, USA April 24-26, 2001
2
Introduction

• Problem
• US military impulse noise exposure is governed by
  MIL-STD-1474D
• New weapons exceed the allowed exposure levels
  for single protection
• There is reason to believe current standard is
  overly conservative
• Background
• Criteria must use freefield pressure data
• Four major criteria have been proposed by NATO
  members since 1968
• USAMRMC human walk up study data were collected
  from 1989-1995 (Johnson, D., USAARL 94-2)
• Objective
• Assess four auditory criteria against data from
  human volunteers using single hearing protection

3
Criteria Definition (Single Protection)
With 15-dB attenuation assumed for modified
earmuff based on REAT data
4
Peak-Based Criteria
Auditory Criteria.xls

• Peak-based auditory criteria with single hearing
  protection and N1.
• No distinction of hearing protection for MIL-STD
• 15 dB global attenuation added to Pfander and
  Smoorenburg for modified muff based on averaged
  REAT from 500-2000 Hz (Brinkmann, 1994)

5
BOP Human Data (Modified Muff)
Free Field Pressure

• Modified muff to simulate poor fit
• Plastic tubes inserted through ear seals to
  introduce leak
• REAT reduced by about 10 dB compared to
  unmodified muff
• Free Field Test parameters
• Distances 5m, 3m, 1m
• Levels (1-7)
• Number of shots (6,12,25,50,100)
• Peak P 174-196 dB
• TA 0.8-2.9 ms
• Bunker Test
• 7 levels, N1, 2, 3
• Total data set
• 192 subjects
• 2000 exposures
• Injury TTS225 dB

6
Free Field Walk-up Test Matrix and Data
1-Meter Distance
3-Meter Distance
7
Walk-up Test Matrix and Data contd.
Bunker Test
5-Meter Distance Free Field Test
8
MIL-STD-1474D vs. Free Field Test Data
9
Statistical Analysis

• Data used at each test condition
• Results of all subjects actually tested
• Presumed failures from all subjects who failed at
  lower L
• Logistic Regression by maximum likelihood,
  (Hosmer Lemeshow, 1989)
• lnp/(1-p) b0 b1 L (pFailure probability)
• L L(P, T, N) (Continuous criteria variable)
• Population-averaged model with exchangeable
  autocorrelation to model walk-up test
• Confidence interval estimated using robust
  formula by Huber and Royall (1967)
• Computations using STATA Software

10
Data Comparison with Correlation
MIL-STD-1474D
Pfander
mf_ep_t35_subjid.xls
mf_ep_t35_subjid.xls
11
Data Comparison with Correlation contd.
Smoorenburg
LAeq8
12
Logistic Regression Results

• Observed thresholds are higher than all four NATO
  criteria by 9.6-21.2 dB
• Best-fit model obtained by backward stepwise
  elimination
• TC and TD eliminated
• Negative TB coefficient
• 3.44 log N trading instead of 10 or 5 log N
• Highest goodness of fit p-value

13
Data Comparison with Best Fit Correlation
mf_ep_t35_subjid.xls
14
Protection Analysis and Discussion

• Protection level has been successfully calculated
  for 95 protection with 95 confidence interval
  for each criterion
• Analysis indicates that all four criteria are
  overly conservative by 9.6-21.2 dB
• Assumption of global correction for protection
  may be inadequate
• Protection depends on intensity and spectral
  effects
• Study is still limited for high intensity noises
• Effects of variable presentation rate and
  intensity combination are not studied
• Wave types studied are limited
• Based on this population of subjects and
  collection of blast waves, then 95 protection
  against TTS25 dB with 95 confidence can be
  achieved for LMthan current MIL-STD threshold