Characterizing Variability in Underground Coal Mine Dust Exposure
1. CHARACTERIZING THE VARIABILITY IN RESPIRABLE
DUST EXPOSURE AND EXAMINING 2010 PROPOSED
CHANGES TO THE UNDERGROUND COAL MINE
DUST STANDARD
Al Imran Khan
Graduate Student (MS)
Thesis Supervisor: Dr. Thomas Novak
Department of Mining Engineering
University of Kentucky
3. Dust
“Dust: Small, dry, solid particles projected into the air
by natural forces, such as wind, volcanic eruption, and
by mechanical or man-made processes such as
crushing, grinding, milling, drilling, demolition, shovelin
g, conveying, screening, bagging and sweeping. Dust
particles are usually in the size range from about 1 to
100 micron in diameter, and they settle slowly under
the influence of gravity.”
(IUPAC, 1990)
4. Dust Particle Type and Particle Deposition
• Inhalable fraction (less than 100 micron AED)
• Thoracic fraction (less than 25 micron AED)
• Respirable fraction (less than 10 micron AED)
Breysse [2006]
5. Respirable Fraction of Dust
Aerodynamic Respirable
Diameter, μm Fraction
2.0 1.0
2.5 0.75
3.5 0.50
5.0 0.25
10.0 0
Los Alamos Group [1962]
6. Motivation Respirable Coal Dust Study
• Respirable dust is the reason
for development of chronic
disease such as Coal Workers’
Pneumoconiosis (CWP)
• Respirable dust particles are
insoluble in water and can not
be absorbed into the blood
stream Normal lung (left) and a lung from a
• Respirable dust particles are miner diagnosed with CWP (right).
invisible to the human eye
Best Practices for Dust Control, NIOSH [2010]
7. Prevalence of CWP in Recent Years
Best Practices for Dust Control, NIOSH [2010]
9. Sampling Dust Exposure
COMPARISION
Personal Dust Monitor
Gravimetric Sampler
(PDM)
Care must be taken to ensure the
Ergonomic Design
upright position of cyclone
assembly
No Mass Sensor
Mass Sensor
Real Time Measurement,
Measurement obtained after the
Continuous Monitoring throughout
end of the shift
the shift
Low measurement bias
High measurement bias
12. Sampling Dust Exposure
PDM Output
Primary Measurements
• MC0: Primary current mass concentration for
the last 30 minute
• CUM0: Primary cumulative mass
concentration since the beginning of the
working shift.
• PROJ: Projected End-of-Shift (EOS)
concentration.
Volkwein [2006]
14. Existing Standard VS Proposed Standard
PARAMETER EXISTING STANDARD PROPOSED STANDARD
1.0 mg/m3 (24-month
Threshold Limit 2.0 mg/m3 period following the
effective date of final rule)
Average of five sample Measuring Exposure Every
Sampling Method
measurements (Bi-monthly) Shift
Production rate as a
percentage of average 50% 100%
production of last 30 shifts
Sampling Device Gravimetric Sampler Personal Dust Monitor
Converted to 8-hr
Sampling Time Standard 8-hr Sampling
equivalent sample
15. Respirable Dust Exposure Study
METHODOLOGY
• The data was collected from three mines one after
another
• The data was standardized to a mean of 1.00 mg/m^3
in order to assess variability
• Fitting the data to a suitable distribution
• Determining confidence bounds and exceedance
fractions
• Determining the mean exposure if the data is not
expected to exceed the permissible limit
16. Variability in Dust Exposure
•Relative standard deviation (RSD) is a useful tool to
statistically inspect sets of data and is commonly used
for scientific studies.
•RSD allows the variability of different sample groups to
be compared more meaningfully.
•The RSD is the ratio of standard deviation to mean.
Criteria Mine A Mine B Mine C Total
Sample Size 197 206 197 600
Mean 1.00 1.00 1.00 1.00
RSD (Relative
0.45 0.32 0.60 0.47
Standard Deviation)
18. Fitting Data to Distribution
• Lognormal: The occupational exposure in
some environments follows a lognormal
distribution.
If y is the lognormally distributed exposure
measurement of an employee, then x = ln(y) is
distributed normally
The collected mine exposure data did not fit
the lognormal
Lyles and Kupper [1996]
19. Fitting Data to Distribution
• Johnson Transformation fit the data very well.
• Johnson Transformation
Z is a standard normal random variable, γ and δ
are shape parameters, σ is a scale parameter and
θ is a location parameter. Without loss of
generality, it is assumed that δ>0 and θ > 0.
20. Inverse Transform
• In order to present the results in actual
exposure, it is essential to convert the
transformed dust exposure to original value
using inverse transform.
• The inverse transform to the original value:
21. Fitting the Exposure Data
Lognormal Johnson Transformation
Probability Plot of Transformed Dust Exposure (Combined Data)
99.99
Mean 1.416665E-11
StDev 1.001
N 600
99 AD 0.312
P-Value 0.550
95
80
Percent
50
20
5
1
0.01
-4 -3 -2 -1 0 1 2 3 4
Transformed Dust Exposure
22. Confidence Upper Bound Calculation
• The 95% confidence bound defines that 95%
of population will be less than it
• Analysis was conducted with the transformed
data to determine upper confidence bounds
for a single (1) measurement, the average of
five (5) measurements, and the average of ten
(10) measurements
• The confidence bound was calculated using
equation: mean + z*(σ/√n)
23. Confidence Upper Bound Calculation
Sample Mine A Mine B Mine C Total
Size
1 1.82 1.54 2.13 1.85
5 1.29 1.21 1.31 1.29
10 1.18 1.14 1.17 1.17
24. Reliability of the Results
• Mine A
10 out of 197 measurements exceed 1.82 mg/m^3, which
is 5.1%
• Mine B
8 out of 206 measurements exceed 1.54 mg/m^3, which is
3.9%
• Mine C
12 out of 197 measurements exceed 2.13 mg/m^3, which
is 6.1%
• Combined Data
28 out of 600 measurements exceed 1.85 mg/m^3, which
is 4.6%
25.
26. Exceedance Fraction
• The exceedance fraction is the percentage of
dust exposure measurements that will be
above an occupational exposure limit for an
exposure group in a particular sampling
environment. The proposed standard
demands that the dust concentration should
not exceed 1.13 mg/m3 for any shift.
27. Exceedance Fraction Calculation
The algorithm to find the exceedance fraction:
• Determining the transformed value (Z) of
permissible limit, i.e. 1.13 mg/m^3
• From normal probability table, the probability of
exceeding Z was obtained
• For different sample sizes, Z was divided by the
respective sample size. Then probability of
exceeding this new value was found
• The number obtained from the normal
probability table indicates the exceedance
fraction for the respective sample size
29. Reliability of the Results
• Mine A
70 out of 197 measurements exceed 1.13 mg/m^3, which
is 35.5%. It was 34% according to the model
• Mine B
65 out of 206 measurements exceed 1.13 mg/m^3, which
is 31.5%. It was 31% according to the model
• Mine C
58 out of 197 measurements exceed 1.13 mg/m^3, which
is 29.4%. It was 32% according to the model
• Combined Data
193 out of 600 measurements exceed 1.13 mg/m^3, which
is 32%. It was 34% according to the model
30.
31. Standard Mean Exposure
• If the dust exposure is expected to be within a
permissible limit with a probability of
95%, then a desired mean exposure must be
determined. Therefore, a recommended mean
exposure value can be calculated where the
individual shift exposure does not exceed 1.13
mg/m3.
32. Standard Mean Exposure
The algorithm to find standard mean
exposure:
• In this case the X = 1.13 mg/m3 was fixed for Z
= 1.645 for the parameters obtained from
Johnson transformation
• Excel Solver was used to change the
parameters of Johnson transformation
• Then the inverse transformed value of zero is
the standard mean exposure
33. Standard Mean Exposure
Standard Mean
95% confidence Exposure
Mine
bound (mg/m^3) (mg/m^3)
Mine A 1.13 0.625
Mine B 1.13 0.725
Mine C 1.13 0.53
Total 1.13 0.6
34. Discussions
• The exposure data was not normally
distributed
• It did not fit the lognormal distribution
• Johnson transformation was the best fit
among the selected distributions
• The behavior of dust was similar when it
comes to exceeding permissible limit across
different mines
35. Discussions
• An RSD of 0.078 was used to calculate ECV
1.13 mg/m^3 in the proposed rule of MSHA
• The RSD found in this study is 0.47
37. Recommendations
• RSD is an important tool to compare dust across
different sample group of dust exposure. However,
the confidence bounds and exceedance fractions
need to be calculated
• The Johnson transformation is a good fit to the
dust exposure data
• An ECV of 1.85 mg/m^3 can still satisfy the mean
exposure of 1.00 mg/m^3
• RSD of 0.47 would be more appropriate to
determine exposure limits
Notes de l'éditeur
Statistical Distribution:Lognormal: This is a type of a parametric distribution where the log of data follows a normal distribution.