[Federal Register Volume 63, Number 22 (Tuesday, February 3, 1998)]
[Notices]
[Pages 5664-5687]
From the Federal Register Online via the Government Publishing Office [www.gpo.gov]
[FR Doc No: 97-33934]
[[Page 5663]]
_______________________________________________________________________
Part V
Department of Labor
_______________________________________________________________________
Mine Safety and Health Administration
_______________________________________________________________________
Department of Health and Human Services
_______________________________________________________________________
Centers for Disease Control and Prevention
_______________________________________________________________________
Mine Shift Atmospheric Conditions; Respirable Dust Sample; and Coal
Mine Respirable Dust Standard Noncompliance Determinations; Correction
and Republication; Notices
Federal Register / Vol. 63, No. 22 / Tuesday, February 3, 1998 /
Notices
[[Page 5664]]
DEPARTMENT OF LABOR
Mine Safety and Health Administration
DEPARTMENT OF HEALTH AND HUMAN SERVICES
Centers for Disease Control and Prevention
RIN 1219-AA82
Mine Shift Atmospheric Conditions; Respirable Dust Sample
Correction and Republication
Note: For the convenience of the user, notice document 97-33934
is being reprinted in its entirety because of numerous errors in the
document originally appearing at 62 FR 68372-68395, December 31,
1997. Those wishing to see a listing of corrections, please call
Patricia Silvey, Mine Safety and Health Administration, 703-235-
1910.
AGENCIES: Mine Safety and Health Administration, Labor, National
Institute for Occupational Safety and Health, Centers for Disease
Control and Prevention, HHS.
ACTION: Final notice of joint finding.
-----------------------------------------------------------------------
SUMMARY: This notice announces that the Secretary of Labor and the
Secretary of Health and Human Services (the Secretaries) find, in
accordance with sections 101 and 202(f)(2) of the Federal Mine Safety
and Health Act of 1977 (Mine Act), 30 U.S.C. 811 and 842(f)
respectively, that the average concentration of respirable dust to
which each miner in the active workings of a coal mine is exposed can
be accurately measured over a single shift. This notice should be read
in conjunction with the notice published separately by the Mine Safety
and Health Administration (MSHA) elsewhere in today's Federal Register.
The Secretaries are rescinding the previous finding, which was proposed
on July 17, 1971 and issued on February 23, 1972, by the Secretary of
the Interior and the Secretary of Health, Education and Welfare.
EFFECTIVE DATE: This notice will be effective on March 2, 1998.
FOR FURTHER INFORMATION CONTACT: Patricia W. Silvey, Director, Office
of Standards, Regulations and Variances; MSHA; 703-235-1910.
SUPPLEMENTARY INFORMATION: In accordance with section 202(f)(2) and
section 101 of the Mine Act, this notice is published jointly by the
Secretaries of the Departments of Labor, and Health and Human Services.
I. Introduction
For as long as miners have taken coal from the ground, the presence
of respirable dust in coal mines has been a source of health problems
for miners. Coal workers' pneumoconiosis, one of the most insidious of
occupational diseases, is caused by deposits of coal mine dust in the
lung and is known as ``black lung disease.'' The disability that may
result from these deposits can range from slightly impaired lung
function to significant decreases in lung function resulting in
breathlessness, recurrent chest illness, and even heart failure. In
addition, the disease may progress even after the miner is no longer
exposed to coal mine dust.
The Federal Coal Mine Health and Safety Act of 1969 (Coal Act)
established the first comprehensive dust standard for underground U.S.
coal mines by setting a limit of 2.0 milligrams of respirable coal mine
dust per cubic meter of air (mg/m3). The 2.0 mg/
m3 standard sets a limit on the concentration of respirable
coal mine dust permitted in the mine atmosphere during each shift to
which each miner in the active workings of a mine is exposed. Congress
was convinced that the only way each miner could be protected from
black lung disease or other occupational dust disease was by limiting
the amount of respirable dust allowed in the air that miners breathe.
The Coal Act was subsequently amended by the Federal Mine Safety
and Health Act of 1977 (Mine Act), 30 U.S.C. 801 et seq. The standard
limiting respirable dust in the mine atmosphere to 2.0 mg/m3
was retained in the Mine Act, which also required that ``each operator
shall continuously maintain the average concentration of respirable
dust in the mine atmosphere during each shift to which each miner in
the active workings of such mine is exposed at or below 2.0 milligrams
of respirable dust per cubic meter of air.'' Section 202(b)(2). (Other
provisions in the Mine Act, sections 205 and 203(b)(2), provide for
lowering the applicable standard when quartz is present and when miners
with evidence of pneumoconiosis have elected to work in a low-dust work
environment.)
Today, dust levels in underground U.S. coal mines are significantly
lower than they were when the Coal Act was passed. Federal mine
inspector sampling results during 1968-1969 show that the average dust
concentration in the environment of a continuous miner operator was 7.7
mg/m3. Current sampling indicates that the average dust
level for that occupation has been reduced by 83 percent to 1.3 mg/
m3. Despite this progress, the Secretaries believe that
occupational lung disease continues to present a serious health risk to
coal miners. In November 1995, the National Institute for Occupational
Safety and Health (NIOSH) issued a criteria document which concluded
that coal miners in our country continue to be at risk for developing
black lung disease.
The Secretary of Labor believes that miners' health can be further
protected from the debilitating effects of black lung disease by
improving their workplace conditions through more effective assessment
of respirable dust concentrations during individual, full shifts. On
February 18, 1994, the Secretary of Labor and the Secretary of Health
and Human Services published a notice in the Federal Register proposing
to find that the average concentration of respirable dust to which each
miner in the active workings of a coal mine is exposed can be
accurately measured over a single shift in accordance with section
202(f)(2) of the Mine Act (56 FR 8357). Additionally, the Secretaries
proposed to rescind the previous finding, which was proposed on July
17, 1971 (36 FR 13286) and issued on February 23, 1972 (37 FR 3833), by
the Secretary of the Interior and the Secretary of Health, Education
and Welfare.
II. General Discussion
The issues related to this finding are complex and highly
technical. The Agencies have organized this final notice to allow
interested persons to first consider pertinent introductory material on
the Agencies' 1972 notice and its recision, and a short overview of the
NIOSH mission and assessment of this finding, as well as those aspects
of MSHA's coal mine respirable dust program relevant to this finding.
Following this introductory material is a discussion of the
``measurement objective,'' or what the Secretaries intend to measure
with a single, full-shift measurement, and the use of the NIOSH
Accuracy Criterion for determining whether a single, full-shift
measurement will ``accurately represent'' the full-shift atmospheric
dust concentration. Next, the validity of the sampling process is
addressed, including the performance of the approved sampler unit,
sample collection procedures, and sample processing. The concept of
measurement uncertainty is then addressed, and why sources of dust
concentration variability and various other factors are not relevant to
the finding. Finally, the notice explains how the total measurement
uncertainty was quantified, and how the accuracy of a single, full-
shift measurement was shown to meet the NIOSH Accuracy Criterion.
Several Appendices, which
[[Page 5665]]
contain relevant technical information, are attached and incorporated
with this notice. The Agencies have additionally included references to
the Appendices throughout this notice.
A. The 1971/1972 Joint Notice of Finding
In 1971 the Secretary of the Interior and the Secretary of Health,
Education and Welfare proposed, and in 1972 issued, a joint finding
under the Coal Act. The finding concluded that a single shift
measurement would not, after applying valid statistical techniques,
accurately represent the atmospheric conditions to which the miner is
continuously exposed. For the reasons that follow, the Secretaries
believe that the 1972 joint finding was incorrect.
Section 202(b)(2) of the Coal Act provided that ``each operator
shall continuously maintain the average concentration of respirable
dust in the mine atmosphere during each shift to which each miner in
the active workings of such mine is exposed at or below [the applicable
respirable dust standard].'' In addition, the term ``average
concentration'' was defined in section 202(f) of the Coal Act as
follows:
* * * the term ``average concentration'' means a determination
which accurately represents the atmospheric conditions with regard
to respirable dust to which each miner in the active workings of a
mine is exposed (1) as measured during an 18 month period following
the date of enactment of this Act, over a number of continuous
production shifts to be determined by the Secretary of the Interior
and the Secretary of Health, Education and Welfare, and (2) as
measured thereafter, over a single shift only, unless the Secretary
of the Interior and the Secretary of Health, Education and Welfare
find, in accordance with the provisions of section 101 of this Act,
that such single shift measurements will not, after applying valid
statistical techniques to such measurement, accurately represent
such atmospheric conditions during such shift.
Therefore, 18 months after the statute was enacted, the ``average
concentration'' of respirable dust in coal mines was to be measured
over a single shift only, unless the Secretaries found that doing so
would not accurately represent mine atmospheric conditions during such
shift. If the Secretaries found that a single shift measurement would
not, after applying valid statistical techniques, accurately represent
mine atmospheric conditions during such shift, then the interim
practice of averaging measurements ``over a number of continuous
production shifts'' was to continue.
On December 16, 1969, the U.S. Congress published a Conference
Report in support of the new Coal Act. The Report refers to section
202(f) by noting that:
At the end of this 18 month period, it requires that the
measurements be over one production shift only, unless the
Secretar[ies] * * * find, in accordance with the standard setting
procedures of section 101, that single shift measurements will not
accurately represent the atmospheric conditions during the measured
shift to which the miner is continuously exposed [Conference Report,
page 75].
This Report is inconsistent with the wording of the section 202(f),
which seeks to apply a single, full-shift measurement to ``accurately
represent such atmospheric conditions during such shift.'' Section
202(f) does not mention continuous exposure. The Secretaries believe
that the use of this phrase is confusing, and to the extent that any
weight of interpretation can be given to the legislative history, that
the Senate's Report of its bill provides a clearer interpretation of
section 202(f) when read together with the statutory language. The
Senate Committee noted in part that:
The committee * * * intends that the dust level not exceed the
specified standard during any shift. It is the committee's intention
that the average dust level at any job, for any miner in any active
working place during each and every shift, shall be no greater than
the standard.
Following passage of the Coal Act, the Bureau of Mines (MSHA's
predecessor Agency within the Department of the Interior) expressed a
preference for multi-shift sampling. Correspondence exchanged during
that time period of 1969 to 1971 reflected concern over the
technological feasibility of controlling dust levels to the limits
established, and the potentially disruptive effects of mine closure
orders because of noncompliance with the respirable dust limits. Both
industry and government officials feared that basing noncompliance
determinations on single, full-shift measurements would increase those
problems. In June 1971, the then-Associate Solicitor for Mine Safety
and Health at the Department of the Interior issued a legal
interpretation of section 202(f), concluding that the average dust
concentration was to be determined by measurements that accurately
represent respirable dust in the mine atmosphere over time rather than
during a shift. On July 17, 1971, the Secretaries of the Interior and
of Health, Education and Welfare issued a proposed notice of finding
under section 202(f) of the Coal Act. The finding concluded that, ``a
single shift measurement of respirable dust will not, after applying
valid statistical techniques to such measurement, accurately represent
the atmospheric conditions to which the miner is continuously exposed''
(36 FR 13286).
In February, 1972, the final finding was issued (37 FR 3833). It
concluded that:
After careful consideration of all comments, suggestions, and
objections, it is the conclusion of the Secretary of the Interior
and the Secretary of Health, Education, and Welfare that a valid
statistical technique was employed in the computer analysis of the
data referred to in the proposed notice [footnote omitted] and that
the data utilized was accurate and supported the proposed finding.
Both Departments also intend periodically to review this finding as
new technology develops and as new dust sampling data becomes
available.
The Departments intend to revise part 70 of Title 30, Code of
Federal Regulations, to improve dust measuring techniques in order
to ascertain more precisely the dust exposure of miners. To
complement the present system of averaging dust measurements, it is
anticipated that the proposed revision would use a measurement over
a single shift to determine compliance with respirable dust
standards taking into account (1) the variation of dust and
instrument conditions inherent in coal mining operations, (2) the
quality control tolerance allowed in the manufacture of personal
sampler capsules, and (3) the variation in weighing precision
allowed in the Bureau of Mines laboratory in Pittsburgh.
The proposed finding, as set forth at 36 F.R. 13286, that a
measurement of respirable dust over a single shift only, will not,
after applying valid statistical techniques to such measurement,
accurately represent the atmospheric conditions to which the miner
under consideration is continuously exposed, is hereby adopted
without change.
As explained in the 1971 proposed finding, the average
concentration of all ten full-shift samples (from one occupation)
submitted from each working section under the regulations in effect at
the time (these were the ``basic samples'' referred to in the proposed
notice of finding) was compared with the average concentration of the
two most recently submitted samples, then to the three most recently
submitted samples, then to the four most recently submitted samples,
etc. In discussing the results of these comparisons the Secretaries
stated that `` * * * the average of the two most recently submitted
samples of respirable dust was statistically equivalent to the average
concentration of the current basic samples for each working section in
only 9.6 percent of the comparisons.''
The title of the 1971/1972 notice and the conclusion it reaches are
clearly inconsistent. The title states that it is a ``Notice of Finding
That Single Shift Measurements of Respirable Dust Will Not Accurately
Represent Atmospheric Conditions During Such Shift.''
[[Page 5666]]
However, the conclusion states that, ``* * * a single shift measurement
* * * will not, after applying valid statistical techniques * * *
accurately represent the atmospheric conditions to which the miner is
continuously exposed'' (emphasis added).
The Secretaries have determined that section 202(f) requires a
determination of accuracy with respect to ``atmospheric conditions
during such shift,'' not ``atmospheric conditions to which the miner is
continuously exposed'' (37 FR 3833). The statistical analysis
referenced in the 1971/1972 proposed and final findings simply did not
address the accuracy of a single, full-shift measurement in
representing atmospheric conditions during the shift on which it was
taken. For this and other reasons set forth in the notice, the
Secretaries hereby rescind the 1972 joint final finding.
III. NIOSH Mission Statement and Assessment of the Joint Finding
The National Institute for Occupational Safety and Health (NIOSH)
was created by Congress in the Occupational Safety and Health Act in
1970. The Act established NIOSH as part of the Department of Health and
Human Services to identify the causes of work-related diseases and
injuries, evaluate the hazards of new technologies, create new ways to
control hazards to protect workers, and make recommendations for new
occupational safety and health standards. Under section 501 of the Mine
Act, Congress gave specific research responsibilities to NIOSH in the
field of coal or other mine health. These responsibilities include the
authority to conduct studies, research, experiments and demonstrations,
in order ``to develop new or improved means and methods of reducing
concentrations of respirable dust in the mine atmosphere of active
workings of the coal or other mine,'' and also ``to develop techniques
for the prevention and control of occupational diseases of miners * *
*.''
When the initial finding, issued under section 202(f) of the Coal
Act, was published in 1972, both the Secretary of the Interior and the
Secretary of Health, Education and Welfare (the predecessor to the
Department of Health and Human Services) indicated that the finding
would be reassessed as new technology was developed, or new data became
available. The Secretary of Health and Human Services, through
delegated authority to the National Institute for Occupational Safety
and Health, has reconsidered the provisions of section 202(f) of the
Mine Act, reviewed the current state of technology and other scientific
advances since 1972, and has determined that the following innovations
and technological advancements are important factors in the
reassessment of the 1971/1972 joint finding.
In 1977 NIOSH published its ``Sampling Strategies Manual,'' which
provided a framework for the statistical treatment of occupational
exposure data [DHEW (NIOSH) Publication No. 77-173; Sec. 4.2.1].
Additionally, that year, NIOSH first published the NIOSH Accuracy
Criterion, which was developed as a goal for methods to be used by OSHA
for compliance determinations [DHEW (NIOSH) Publication No. 77-185; pp.
1-5]. In 1980, new mine health standards issued by the Secretary of
Labor (30 CFR parts 70, 71, and 90) improved the quality of the
sampling process by revising sampling, maintenance, and calibration
procedures. Prior to 1984, filter capsules used in sampling were
manually weighed by MSHA personnel using semi-micro balances, making
precision weights to the nearest 0.1 mg (100 micrograms). In 1984, a
fully-automated, robotic weighing system was introduced along with
state-of-the-art electronic microbalances. In 1994, the balances were
further upgraded, and in 1995 the weighing system was again improved,
increasing weighing sensitivity to the microgram level. Also, in 1987,
electronic flow-control sampling pump technology was introduced in the
coal mine dust sampling program with the use of MSA
FlowLiteTM pumps. 1 These new pumps compensate
for the changing filter flow-resistance that occurs due to dust
deposited during the sampling period. The second generation of
constant-flow sampling pumps was introduced in 1994, with the
introduction of the MSA Escort ELF pump. The automatic
correction provided by these new pumps improves the stability of the
sampler air flow rates and reduces the inaccuracies that were inherent
in the 1970-1980s vintage sampling pumps. One further improvement was
made in 1992 with the introduction of the new tamper-resistant filter
cassettes. Because of these evolving improvements to the sampling
process, a better understanding of statistical methods applied to
method accuracy, and a reconsideration of the requirements of section
202(f) of the Mine Act, the Secretary of Health and Human Services has
determined that the previous joint finding should be reevaluated.
---------------------------------------------------------------------------
\1\ Reference to specific equipment, trade names or
manufacturers does not imply endorsement by NIOSH or MSHA.
---------------------------------------------------------------------------
IV. MSHA Mission Statement and Overview of the Respirable Dust
Program
With the enactment of the Mine Act, Congress recognized that ``the
first priority and concern of all in the coal or other mining industry
must be the health and safety of its most precious resource--the
miner.'' Congress further realized that there ``is an urgent need to
provide more effective means and measures for improving the working
conditions and practices in the Nation's coal or other mines in order
to prevent death and serious physical harm, and in order to prevent
occupational diseases originating in such mines.'' With these goals in
mind, MSHA is given the responsibility to protect the health and safety
of the Nation's coal and other miners by enforcing the provisions of
the Mine Act.
A. The Coal Mine Respirable Dust Program
In 1970, federal regulations were issued by MSHA's predecessor
agency that established a comprehensive coal mine operator dust
sampling program, which required the environment of the occupation on a
working section exposed to the highest respirable dust concentration to
be sampled--the ``high risk occupation'' concept. All other occupations
on the section were assumed to be protected if the high risk occupation
was in compliance. Under this program, each operator was required to
initially collect and submit ten valid respirable dust samples to
determine the average dust concentration (across ten production
shifts). If analysis showed the average dust concentration to be within
the applicable dust standard, the operator was required to submit only
five valid samples a month. If compliance continued to be demonstrated,
the operator was required to take only five valid samples every other
month. The initial, monthly, and bimonthly sampling cycles were
referred to as the ``original,'' ``standard,'' and ``alternative
sampling'' cycles, respectively. When the average dust concentration
exceeded the standard, the operator reverted back to the standard
sampling cycle.
In addition to sampling the high risk occupation at specified
frequencies, each miner was sampled individually at different
intervals. However, these early individual sample results were not used
for enforcement but were provided to NIOSH for medical research
purposes.
MSHA revised these regulations in April 1980 (45 FR 23990) to
reduce the
[[Page 5667]]
operator sampling burden, to simplify the sampling process, and to
enhance the overall quality of the sampling program. The result was to
replace the various sampling cycles with a bimonthly sampling cycle and
to eliminate the requirement that each miner be sampled. These are the
regulations that currently govern the mine operator dust sampling
program, and which continue to be based on the high risk occupation
concept, now referred to as the ``designated occupation'' or ``D.O.''
sampling concept.
It should be noted that the preamble to the final rule amending the
regulations in April 1980 (45 FR 23997), explicitly refers to the use
of single versus multiple samples as it applies to the operator
respirable dust sampling program.
Compliance determinations will generally be based on the average
concentration of respirable dust measured by five valid respirable
dust samples taken by the operator during five consecutive shifts,
or five shifts worked on consecutive days. Therefore, the sampling
results upon which compliance determinations are made will more
accurately represent the dust in the mine atmosphere than would the
results of only a single sample taken on a single shift. In
addition, MSHA believes the revised sampling and maintenance and
calibration procedures prescribed by the final rule will
significantly improve the accuracy of sampling results.
At the time of these amendments, MSHA examined section 202(b)(2) of
the Coal Act, which was retained unchanged in the 1977 Mine Act. The
Agency stated in the preamble to the final rule that:
Although single-shift respirable dust sampling would be most
compatible with this single-shift standard, Congress recognized that
variability in sampling results could render single-shift samples
insufficient for compliance determinations. Consequently, Congress
defined ``average concentration'' in section 202(f) of the 1969 Coal
Act which is also retained in the 1977 Act.
MSHA believes that this interpretation merely recognized the two
ways of measurement authorized in section 202(f), and expressed the
preference on the part of MSHA in 1980 to retain multi-shift sampling
in the operator sampling program. The phrase used in the preamble to
the final rule reflects that MSHA understood that the 2.0 mg/
m3 limit was a single-shift standard, which was not to be
exceeded on a shift. The preamble referenced the continuous multi-shift
sampling and single-shift sampling conducted by the Secretary of the
Interior and the Secretary of Health, Education, and Welfare, and noted
that in the 1971/1972 proposed and final findings,
It had been determined after applying valid statistical
techniques, * * * that a single shift sample should not be relied
upon for compliance determinations when the respirable dust
concentration being measured was near 2.0 mg/m3.
Accordingly, the [Secretaries] prescribed consecutive multi-shift
samples to enforce the respirable dust standard.
The preamble provides no further explanation for the statement that
single-shift samples should not be relied on when the respirable dust
concentration being measured was near 2.0 mg/m3. Thus, the
1980 final rule, which reduced the number of samples that operators
were required to take for compliance determinations, merely reiterated
the rationale behind the 1971/1972 proposed and final findings
concerning single-shift samples, and did not address the accuracy of a
single, full-shift measurement.
MSHA continues to take an active role in sampling for respirable
dust by conducting inspections annually at each surface and underground
coal mine. During these inspections, MSHA inspectors collect samples on
multiple occupations to determine compliance with the applicable
standard, assess the effectiveness of the operator's dust control
program, quantify the level of crystalline silica (quartz) in the work
environment, and identify occupations other than the ``D.O.'' which may
be at risk and should be monitored by the mine operator.
Depending on the concentration of dust measured, an MSHA inspector
may terminate sampling after the first day if levels are very low, or
continue for up to five shifts or days before making a compliance or
noncompliance determination. MSHA inspection procedures require
inspectors to sample at least five occupations, if available, on each
mechanized mining unit (MMU) on the first day of sampling. The operator
is cited if the average of those measurements exceeds the applicable
standard. However, if the average falls below the standard, but one or
more of the measurements exceed it, additional samples are collected on
the subsequent production shift or day. The results of the first and
second day of sampling on all occupations are then averaged to
determine if the applicable standard is exceeded. Additionally, when an
inspector continues sampling after the first day because a previous
measurement exceeds the standard, MSHA's procedures call for all
measurements taken on a given occupation to be averaged individually
for that occupation. If the average of measurements taken over more
than one day on all occupations is equal to or less than the applicable
standard, but the average of measurements taken on any one occupation
exceeds the value in a decision table developed by MSHA (based on the
cumulative concentration for two or more samples exceeding 10.4 mg/m
\3\, which is equivalent to a 5-measurement average exceeding 2.0 mg/m
\3\), the operator is cited for exceeding the applicable standard.
B. The Spot Inspection Program (SIP)
In response to concerns about possible tampering with dust samples
in 1991, MSHA convened the Coal Mine Respirable Dust Task Group (Task
Group) to review the Agency's respirable dust program. As part of that
review, MSHA developed a special respirable dust ``spot inspection
program'' (SIP).
This program was designed to provide the Agency with information on
the dust levels to which underground miners are typically exposed.
Because of the large number of mines and MMUs (mechanized mining units)
involved and the need to obtain data within a short time frame,
respirable dust sampling during the SIP was limited to a single shift
or day, a departure from MSHA's normal sampling procedures. The term
``MMU'' is defined in 30 CFR 70.2(h) to mean a unit of mining
equipment, including hand loading equipment, used for the production of
material. As a result, MSHA decided that if the average of multiple
occupation measurements taken on an MMU during any one-day inspection
did not exceed the applicable standard the inspector would review the
result of each individual full-shift sample. If any individual full-
shift measurement exceeded the applicable standard by an amount
specified by MSHA, a citation would be issued for noncompliance,
requiring the mine operator to take immediate corrective action to
lower the average dust concentration in the mine atmosphere in order to
protect miners.
During the SIP inspections, MSHA inspectors cited violations of the
2.0 mg/m \3\ standard if either the average of the five measurements
taken on a single shift was greater than or equal to 2.1 mg/m \3\, or
any single, full-shift measurement exceeded or equaled 2.5 mg/m \3\.
Similar adjustments were made when the 2.0 mg/m \3\ standard was
reduced due to the presence of quartz dust in the mine atmosphere.
The procedures issued by MSHA's Coal Mine Safety and Health
Division during the SIP were similar to those used by the MSHA Metal/
Nonmetal Mine Safety and Health Division and
[[Page 5668]]
the Occupational Safety and Health Administration (OSHA) when
determining whether to cite based on a single, full-shift measurement.
That practice provides for a margin of error reflecting an adjustment
for uncertainty in the measurement process (i.e., sampling and
analytical error). The margin of error thus allows citations to be
issued only where there is a high level of confidence that the
applicable standard has been exceeded.
Based on the data from the SIP inspections, the Task Group
concluded that MSHA's practice of making noncompliance determinations
solely on the average of multiple-sample results did not always result
in citations in situations where miners were known to be overexposed to
respirable coal mine dust. For example, if measurements obtained for
five different occupations within the same MMU were 4.1, 1.0, 1.0, 2.5,
and 1.4 mg/m \3\, the average concentration would be 2.0 mg/m \3\.
Although the dust concentration for two occupations exceeds the
applicable standard, under MSHA procedures no citation would have been
issued nor any corrective action required to reduce dust levels to
protect miners' health. Instead, MSHA policy required the inspector to
return to the mine the next day that coal was being produced and resume
sampling in order to decide if the mine was in compliance or not in
compliance.
The Task Group also recognized that the results of the first full-
shift samples taken by an inspector during a respirable dust inspection
are likely to reflect higher dust concentrations than samples collected
on subsequent shifts or days during the same inspection. MSHA's
comparison of the average dust concentration of inspector samples taken
on the same occupation on both the first and second day of a multiple-
day sampling inspection showed that the average concentration of all
samples taken on the first day of an inspection was almost twice as
high as the average concentration of samples taken on the second day.
MSHA recognized that sampling on successive days does not always result
in measurements that are representative of everyday respirable dust
exposures in the mine because mine operators can anticipate the
continuation of inspector sampling and make adjustments in dust control
parameters or production rates to lower dust levels during the
subsequent sampling.
In response to these findings, in November 1991, MSHA decided to
permanently adopt the single shift inspection policy initiated during
the SIP.
C. The Keystone Decision
In 1991, three citations based on single, full-shift measurements
were issued under the SIP to the Keystone Coal Mining Corporation. The
violations were contested, and an administrative law judge from the
Federal Mine Safety and Health Review Commission (Commission) vacated
the citations. The decision was appealed by the Secretary of Labor to
the Commission because the Secretary believed that the administrative
law judge was in error in finding that rulemaking was required under
section 202(f) of the Mine Act for the Secretary to use single, full-
shift measurements for noncompliance determinations. In addition, the
Secretary contended that the 1971/1972 finding pertained to operator
sampling and that the SIP at issue involved only MSHA sampling. The
Commission, which affirmed the decision of the administrative law
judge, found that:
Title II [of the Mine Act] applies to both operator sampling and
to MSHA actions to ensure compliance, including sampling by MSHA.
Section 202(g) specifically provides for MSHA spot inspections.
Nothing in Sec. 202(f) or Sec. 202(g) suggests that Sec. 202(f)
applies differently to MSHA sampling. Thus, the 1971 finding, issued
for purposes of Title II, applies broadly to both MSHA and operator
sampling of the mine atmosphere.
The Commission also held that the revised MSHA policy was in
contravention of the 1971/1972 finding and could only be altered if the
requirements of the Mine Act and the Administrative Procedure Act, 5
U.S.C. 550, were met.
V. Executive Order 12866 and Regulatory Impact Analysis
MSHA has designated this joint finding as a significant action; it
has been reviewed by OMB under E.O. 12866. MSHA estimates that the
total annual costs associated with the implementation of this finding
will be $707,950, of which $446,125 will be incurred by underground
coal mines and $261,825, incurred by surface coal operations. MSHA
projects that this finding will result in reductions of future cases of
occupational lung disease and attendant cost savings. MSHA has prepared
a separate regulatory impact analysis which is available to the public
upon request.
VI. Procedural History of the Current Notices
As a result of the innovations and technological advancements
described earlier, and the decision in Keystone Coal v. Secretary of
Labor, 16 FMSHRC 6 (January 4, 1994), the Secretary of Labor and the
Secretary of Health and Human Services published a proposed joint
notice in the Federal Register on February 18, 1994 (59 FR 8357),
pursuant to sections 101 and 202(f)(2) of the Mine Act. The notice
proposed to rescind the 1971/1972 proposed and final findings by the
Secretaries of the Interior and Health, Education and Welfare, and find
that a single, full-shift measurement will accurately represent the
atmospheric conditions with regard to the respirable dust concentration
during the shift on which it was taken.
Concurrently, MSHA published a separate notice in the Federal
Register announcing its intention to use both single, full-shift
respirable dust measurements and the average of multiple, full-shift
respirable dust measurements for noncompliance determinations (59 FR
8356). That notice was published to inform the mining public of how the
Agency intended to implement its new enforcement procedure utilizing
single, full-shift samples, and to solicit public comment on the new
procedure.
The comment period on the proposed joint finding was scheduled to
close on April 19, 1994, but was extended to May 20, 1994, in response
to requests from the mining community (59 FR 16958). Subsequently,
public comments were received, including comments from both labor and
industry.
On July 6, 1994, in response to requests from the mining community,
a public hearing was held on both notices in Morgantown, West Virginia
(59 FR 29348). Also, in response to additional requests from the mining
community, a second hearing was held on July 19, 1994, in Salt Lake
City, Utah. To allow for the submission of post-hearing comments, the
record was held open until August 5, 1994.
The hearings on the proposed joint notice were conducted by a joint
MSHA/NIOSH panel. Presenters at the Morgantown hearing included
international and local representatives of the United Mine Workers of
America (UMWA), several mine operators, and a panel presentation from
the American Mining Congress (AMC) and the National Coal Association
(NCA). Presenters at the Salt Lake City hearing included the Utah
Mining Association, several mine operators, and another joint AMC/NCA
panel. The joint MSHA/NIOSH panel received prepared remarks from the
presenters and asked questions as well. The joint agency panel also
responded to questions from the presenters.
To ensure that all issues raised were fully considered, MSHA and
NIOSH conducted a thorough review of existing data, engaged in an
extensive literature
[[Page 5669]]
search, sought an independent analysis of the scientific validity of
single, full-shift measurements, and conducted additional testing.
These efforts resulted in the collection of a significant amount of
information, which was made a part of the public record on September 9,
1994 (59 FR 50007). To allow interested parties the opportunity to
review and comment on the supplemental material, the Agencies extended
the comment period from September 30 to November 30, 1994.
After the close of the comment period, the Agencies reviewed all of
the comments, data and other information submitted into the record.
Some of the commenters raised questions regarding the accuracy of
single, full-shift measurements and challenged the Agencies' estimate
of measurement imprecision inherent in sample collection and analysis.
While reviewing these issues, the Agencies concluded that the term
``accurately represent'' as used in section 202(f) needed to be defined
because of the issues which commenters raised. In response, the
Agencies reopened the record on March 12, 1996, to provide a criterion
for ``accuracy'', to supply new data and statistical analytical
analyses on the precision of coal mine respirable dust measurements
obtained using approved sampling equipment, and to allow the public to
review and submit comments on the supplemental information (61 FR
10012). In addition, the March 12 notice identified certain refinements
in MSHA's measurement process as applied to inspector samples. These
modifications, currently in place, involve the measurement of both pre-
and post-exposure filter weights to the nearest microgram on a scale
calibrated using the established procedure in MSHA's laboratory, and
discontinuing the practice of truncating the recorded weights used in
calculating the dust concentration (that is, MSHA no longer ignores
digits representing hundredths and thousandths of a milligram).
The new comment period was scheduled to close on April 11, 1996,
but was extended until June 10, 1996, in response to requests from the
mining community. Additionally, on April 11, 1996, the Agencies
announced their intention to conduct a second public hearing on the
content of the March 12 notice (61 FR 16123). On May 10, 1996, a public
hearing conducted by a joint MSHA/NIOSH panel was held in Washington,
DC. One scheduled presenter, representing the UMWA, appeared at this
hearing.
Some commenters expressed concern for the procedures used by the
Agencies in making a new finding, asserting that MSHA and NIOSH were
not complying with the rulemaking provisions of the Mine Act. These
commenters contended that the recision of the final finding and
implementation by MSHA of single, full-shift sampling can only be
effectuated through notice and comment rulemaking. These commenters
argue that because MSHA failed to appeal the Keystone case, MSHA was
bound by the Commission decision in that case which mandated notice and
comment rulemaking to rescind the prior finding and authorize use of
single samples by the Agency.
MSHA and NIOSH have considered these comments, but believe that the
process they have chosen to follow is consistent with the requirement
of section 202(f) of the Mine Act, which provides that a finding shall
be made ``in accordance with the provisions of section 101'' of the
Mine Act. Section 101 contains the procedural requirements for
promulgation of mandatory health and safety standards, including
provision for notice and comment. All interested parties were given
ample opportunity for notice and comment at every stage of
consideration of the proposed joint finding. The Agencies are not
developing, promulgating, or revising a mandatory health standard in
this notice, nor is the 2.0 mg/m \3\ respirable dust standard being
revised. Moreover, the Agencies have made a finding that the average
concentration of respirable dust in the mine atmosphere to which each
miner in the active workings of a coal mine is exposed during a shift
can be accurately measured with a single, full-shift sample. This is a
scientific finding contemplated by section 202(f) of the Mine Act.
While one commenter asserted that the Secretaries were not following
proper notice and comment procedures in section 101 [e.g., sections
101(a)(1) through (9)], the only example given by the commenter is the
fact that the notice was published in the ``Notice'' section, rather
than the ``Proposed Rules'' or ``Rules and Regulations'' section of the
Federal Register. Because this is not a mandatory safety and health
standard, there is no need for the Secretaries to publish the finding
as a proposed rule, or to address feasibility, for example, which would
be required under section 101(a)(6)(A) when a mandatory safety or
health standard is promulgated. The Secretaries have properly complied
with all the procedural elements of section 101 which apply to this
notice.
Some commenters referenced section 101(a)(9) of the Mine Act, 30
U.S.C. 811(a)(9), which provides that no mandatory standard shall
reduce the protection afforded miners by an existing standard under the
Mine Act. As stated previously, this scientific finding does not
constitute rulemaking and is not a promulgation of a mandatory health
standard. Rather, it is a ``finding'' under the Mine Act, established
in the same manner as the initial finding, in 1972, the effect of which
is to increase health protection for miners by allowing single, full-
shift measurements to be used to determine average concentrations
during a single work shift instead of continuing to rely solely on
averaging the results of several days of sampling or sampling across
various occupations on the same shift.
In MSHA's notice published on February 18, 1994 (59 FR 8356), the
Agency specifically noted that any change to the substantive procedure
for mine operator respirable dust sampling governed by MSHA regulations
would require rulemaking by MSHA.
VII. Issues Regarding Accuracy of a Single, Full-Shift Measurement
Some commenters questioned the accuracy of single, full-shift
measurements, and challenged the Secretaries' assessment of measurement
accuracy. Some commenters questioned the Secretaries' interpretation of
section 202(b) of the Mine Act, while others agreed with the
interpretation. The following issues were generally raised: the
measurement objective as defined by the Mine Act; the definition of the
term ``accurately represent'', as used in section 202(f); the validity
of the sampling process; measurement uncertainty and dust concentration
variability; and the accuracy of a single, full-shift measurement.
A. Measurement Objective
Some comments reflected a general misunderstanding of what the
Secretaries intend to measure with a single, full-shift measurement,
i.e., the measurement objective. For example, some commenters asserted
that the dust concentration that should be measured is dust
concentration averaged over a period greater than a single shift. Some
commenters noted that dust concentrations can vary during a shift and
that dust concentration is not uniform throughout a miner's work area.
In order to clarify the intent of the Secretaries, the explanation that
follows describes the elements of the measurement objective and how the
measurement objective relates to the requirements of section 202(f).
To evaluate the accuracy of a dust sampling method it is necessary
to specify the airborne dust to be measured, the time period to which
the
[[Page 5670]]
measurement applies, and the area represented by the measurement. Once
specified, these items can be combined into a measurement objective.
The measurement objective represents the goal of the sampling and
analytical method to be utilized.
1. The Airborne Dust to be Measured
Section 202(f) of the Mine Act states that ``average
concentration'' means `` * * * a determination [i.e., measurement]
which accurately represents the atmospheric conditions with regard to
respirable dust to which each miner in the active workings of a mine is
exposed.'' Later in section 202(f), the phrase ``atmospheric
conditions'' is used to refer to the concentration of respirable dust.
Therefore, the airborne dust to be measured is respirable dust. Section
202(e) defines respirable dust as the dust measured by an approved
sampler unit.
2. Time Period to Which the Measurement Applies
Section 202(b)(2) provides that each mine operator ``* * * shall
continuously maintain the average concentration of respirable dust in
the mine atmosphere during each shift to which each miner * * * is
exposed'' at or below the applicable standard. In section 202(f)
``average concentration'' is defined as an atmospheric condition
measured ``over a single shift only, unless * * * such single shift
measurement will not, after applying valid statistical techniques,
accurately represent such atmospheric conditions during such shift.''
For the purpose of this notice, the Secretaries have determined that
``atmospheric conditions'' mean the fluctuating concentration of
respirable coal mine dust during a single shift. These are the
atmospheric conditions to which a sampler unit is exposed. Therefore,
the present finding pertains only to the accuracy in representing the
average of the fluctuating dust concentration over a single shift.
3. Area Represented by the Measurement
The Mine Act gives the Secretary of Labor the discretion to
determine the area to be represented by respirable dust measurements
collected over a single shift. As articulated by the United States
Court of Appeals for the 10th Circuit in American Mining Congress (AMC)
versus Marshall, 671 F.2d 1251 (1982), the Secretary of Labor may place
the sampler unit in any area or location ``* * * reasonably calculated
to prevent excessive exposure to respirable dust.'' Because the
Secretary of Labor intends to prevent excessive exposure by limiting
dust concentration at every location in the active workings, the area
represented by any respirable dust measurement must be the sampling
location.
Some commenters identified the dust concentration to be estimated
as either the mean dust concentration over some period greater than an
individual shift, the mean dust concentration over some spatially
distributed region of the mine, or a ``grand mean'' consisting of some
combination of the above. These comments were based on the false
premise that the measurement objective in section 202(f) is something
other than the average atmospheric conditions during a single shift at
the sampling location. It is true that these mean quantities described
by some commenters cannot be accurately estimated using a single, full-
shift measurement, but the Secretaries make no claim of doing so, nor
are they required to make such considerations.
Some commenters argued that Congress intended that the measurement
objective be a long-term average. Specifically, some commenters stated
that because coal dust exposure is related to chronic health effects,
the exposure limit should be applied to dust concentrations averaged
over a miner's lifetime. These commenters identified the measurement
objective as being the dust concentration averaged over a long, but
unspecified, term and argued that a single, full-shift measurement
cannot accurately estimate this long-term average.
If the objective of section 202(b) were to estimate dust
concentration averaged over a lifetime of exposure, then the
Secretaries would agree that a single, full-shift sample, or even
multiple samples collected during a single inspection, would not
provide the basis for an accurate measurement. Section 202(b) of the
Mine Act, however, does not mention long-term averaging, rather it
explicitly requires that the average dust concentration be continuously
maintained at or below the applicable standard during each shift
(emphasis added). Furthermore, in Consolidation Coal Company versus
Secretary of Labor 8 FMSHRC 890, (1986), aff'd 824 F.2d 1071, (D.C.
Cir. 1987), the Commission found that each episode of a miner's
overexposure to respirable dust significantly and substantially
contributes to the health hazard of contracting chronic bronchitis or
coal workers' pneumoconiosis, diseases of a fairly serious nature.
Some commenters submitted evidence that dust concentrations can
vary significantly near the mining face, and that these variations may
extend into areas where miners are located. That is, the average dust
concentration over a full shift is not identical at every point within
a miner's work area. These commenters submitted several bodies of data
purporting to show significant discrepancies between simultaneous dust
concentration measurements collected within a relatively small distance
of one another. Several commenters maintained that the measurement
objective is to accurately measure the average concentration within
some arbitrary sphere about the head of the miner, and that multiple
measurements within this sphere are necessary to obtain an accurate
measurement. The Secretaries recognize that dust concentrations in the
mine environment can vary from location to location, even within a
small area near a miner. As mentioned earlier, the Mine Act does not
specify the area that the measurement is supposed to represent, and the
sampler unit may therefore be placed in any location reasonably
calculated to prevent excessive exposure to respirable dust.
Several commenters suggested that the measurement objective should
be a miner's ``true exposure'' or what the miner actually inhales. The
Secretaries do not intend to use a single, full-shift measurement to
estimate any miner's ``true exposure,'' because no sampling device can
exactly duplicate the particle inhalation and deposition
characteristics of a miner at any work rate (these characteristics
change with work rate), let alone at the various work rates occurring
over the course of a shift. Section 202(a) of the Mine Act, however,
refers to ``the amount of respirable dust in the mine atmosphere to
which each miner in the active workings of such mine is exposed''
measured ``* * * at such locations * * *'' as prescribed by the
Secretary of Labor. It is sufficient for the purposes of the Mine Act
that the sampler unit accurately represent the amount of respirable
dust at such locations only.
Accordingly, the Secretaries define the measurement objective to be
the accurate determination of the average atmospheric conditions, or
concentration of respirable dust, at a sampling location over a single
shift.
B. Accuracy Criterion
A ``single shift measurement'' means the calculated dust
concentration resulting from a valid single, full-shift sample of
respirable coal mine dust. In reviewing the various issues raised by
commenters, the Agencies found that the term ``accurately represent,''
as used
[[Page 5671]]
in section 202(f) in connection with a single shift measurement, was
not defined in the Mine Act. Therefore, in their March 12, 1996 notice,
the Secretaries proposed to apply an accuracy criterion developed and
adopted by NIOSH in judging whether a single, full-shift measurement
will ``accurately represent'' the full-shift atmospheric dust
concentration. This criterion requires that measurements come within 25
percent of the corresponding true dust concentration at least 95
percent of the time [1].
One commenter opposed the application of the NIOSH Accuracy
Criterion since it ignores environmental variability. For reasons
explained above, the Secretaries have restricted the measurement
objective to an individual shift and sampling location. Therefore,
environmental variability beyond what occurs at the sampling location
on a single shift is not relevant to assessing measurement accuracy.
For over 20 years, the NIOSH Accuracy Criterion has been used by
NIOSH and others in the occupational health professions to validate
sampling and analytical methods. This accuracy criterion was devised as
a goal for the development and acceptance of sampling and analytical
methods capable of generating reliable exposure data for contaminants
at or near the Occupational Safety and Health Administration's (OSHA)
permissible exposure limits.
OSHA has frequently employed a version of the NIOSH Accuracy
Criterion when issuing new or revised single substance standards. For
example, OSHA's benzene standard provides: ``[m]onitoring shall be
accurate, to a confidence level of 95 percent, to within plus or minus
25 percent for airborne concentrations of benzene''(29 CFR
1910.1028(e)(6)). Similar wording can be found in the OSHA standards
for vinyl chloride (29 CFR 1917), arsenic (29 CFR 1918), lead (29 CFR
1925), 1,2-dibromo-3-chloropropane (29 CFR 1044), acrylonitrile (29 CFR
1045), ethylene oxide (29 CFR 1047), and formaldehyde (29 CFR 1048).
Note that for vinyl chloride and acrylonitrile, the accuracy criteria
for the method is 35 percent at 95 percent confidence at
the permissible exposure limit.
Some commenters contended that the NIOSH Accuracy Criterion does
not conform with international standards recently adopted by the
European Committee for Standardization (CEN) [2]. Contrary to these
assertions, the NIOSH Accuracy Criterion not only conforms to the CEN
criterion but is, in fact, more stringent. The CEN criterion requires
that 95 percent of the measurements fall within 30 percent
of the true concentration, compared to 25 percent under the
NIOSH criterion. Consequently, any sampling and analytical method that
meets the NIOSH Accuracy Criterion will also meet the CEN criterion.
The NIOSH Accuracy Criterion is relevant and widely recognized and
accepted in the occupational health professions. Further, commenters
proposed no alternative criteria for accuracy. Accordingly, for
purposes of section 202(f) of the Mine Act, the Secretaries consider a
single, full-shift measurement to ``accurately represent'' atmospheric
conditions at the sampling location, if the sampling and analytical
method used meets the NIOSH Accuracy Criterion.
Several commenters suggested that method accuracy should be
determined under actual mining conditions rather than in a laboratory
or in a controlled environment. Although the NIOSH Accuracy Criterion
does not require field testing, it recognizes that field testing ``does
provide further test of the method.'' However, in order to avoid
confusing real differences in dust concentration with measurement
errors when testing is done in the field, ``precautions may have to be
taken to ensure that all samplers are exposed to the same
concentrations'' [1]. Similarly, the CEN criterion for method accuracy
specifies that ``testing of a procedure shall be carried out under
laboratory conditions.'' To determine, so far as possible, the accuracy
of its sampling and analytical method under actual mining conditions,
MSHA conducted 22 field tests in an underground coal mine. To provide a
valid basis for assessing accuracy, 16 sampler units were exposed to
the same dust concentration during each field test using a specially
designed portable chamber. The data from these field experiments were
used by NIOSH in its ``direct approach'' to determining whether or not
MSHA's method meets the long-established NIOSH Accuracy Criterion. (See
section VII.E.2. of this notice).
In response to the March 12, 1996 notice, a commenter claimed that
the supplementary information and analyses introduced into the public
record by that notice addressed the precision of a single, full-shift
measurement rather than its accuracy. According to this commenter, by
focusing on precision, important sources of systematic error had been
overlooked. The Secretaries agree with the comment that precision is
not the same thing as accuracy. The accuracy of a measurement depends
on both precision and bias [1,3]. Precision refers to consistency or
repeatability of results, while bias refers to a systematic error that
is present in every measurement. Since the NIOSH Accuracy Criterion
requires that measurements consistently fall within a specified
percentage of the true concentration, the criterion covers both
precision and uncorrectable bias.
Since the amount of dust present on a filter capsule used by an
MSHA inspector is measured by subtracting the pre-exposure weight from
the post-exposure weight determined in the same laboratory, any bias in
the weighing process attributable to the laboratory is mathematically
canceled out by subtraction. Furthermore, as will be discussed later, a
control (i.e., unexposed) filter capsule will be pre-and post-weighed
along with the exposed filter capsules. The weight gain of the exposed
capsule will be adjusted by the weight gain or loss of the control
filter capsule. Therefore, any bias that may be associated with day-to-
day changes in laboratory conditions or introduced during storage and
handling of the filter capsules is also mathematically canceled out.
Moreover, the concentration of respirable dust is effectively defined
by section 202(e) of the Mine Act and the implementing regulations in
30 CFR parts 70, 71, and 90 to be whatever is measured with an approved
sampler unit after multiplication by the MRE-equivalent conversion
factor prescribed by the Secretary of Labor. Therefore, the Secretaries
have concluded that the improved sampling and analytical method is
statistically unbiased. This means that such measurements contain no
systematic error. It should also be noted that since any systematic
error would be present in all measurements, measurement bias cannot be
reduced by making multiple measurements. Other comments regarding
measurement bias are addressed in Appendix A.
For unbiased sampling and analytical methods, a standard
statistic--called the coefficient of variation (CV)--is used to
determine if the method meets the NIOSH Accuracy Criterion. The CV,
which is expressed as either a fraction (e.g., 0.05) or a percentage
(e.g., 5 percent), quantifies measurement accuracy for an unbiased
method. An unbiased method meets the NIOSH Accuracy Criterion if the
``true'' CV is no more than 0.128 (12.8 percent). However, since it is
not possible to determine the true CV with 100-percent confidence, the
NIOSH Accuracy Criterion contains the additional requirement that there
be 95-percent confidence that measurements by the method will come
within 25 percent of
[[Page 5672]]
the true concentration 95 percent of the time. Stated in mathematically
equivalent terms, an unbiased method meets the NIOSH Accuracy Criterion
if there is 95-percent confidence that the true CV is less than or
equal to 0.128 (12.8 percent).
C. Validity of the Sampling Process
A single, full-shift measurement of respirable coal mine dust is
obtained with an approved sampler unit, which is either worn or carried
by the miner directly to and from the sampling location and is operated
portal to portal. The unit remains operational during the entire shift
or for eight hours, whichever time is less. A portable, battery-powered
pump draws dust-laden mine air at a flow rate of 2 liters per minute
(L/min) through a 10-mm nylon cyclone, a particle-size selector that
removes non-respirable particles from the airstream. Non-respirable
particles are particles that tend to be removed from the airstream by
the nose and upper respiratory airways. These particles fall to the
bottom of the cyclone body called the ``grit pot,'' while smaller,
respirable particles (of the size that would normally enter into the
lungs) pass through the cyclone, directly into the inlet of the filter
cassette. This airstream is directed through the pre-weighed filter
leaving the particles deposited on the filter surface. The collection
filter is enclosed in an aluminum capsule to prevent leakage of sample
air around the filter and the loss of any dust dislodged due to impact.
The filter capsule is sealed in a protective plastic enclosure, called
a cassette, to prevent contamination. After completion of sampling, the
filter cassette is sent to MSHA's Respirable Dust Processing Laboratory
in Pittsburgh, Pennsylvania, where it is weighed again to determine the
weight gain in milligrams, which is the amount of dust collected on the
filter. The concentration of respirable dust, expressed as milligrams
per cubic meter (mg/m\3\ ) of air, is determined by dividing the weight
gain by the volume of mine air passing through the filter and then
multiplying this quantity by a conversion factor (discussed below in
Appendix A) prescribed by the Secretary.
Some comments generally addressed the quality and reliability of
the equipment used for sampling. Specific concerns were expressed about
the quality of filter cassettes and the reliability, due to their age
and condition, of sampling pumps used by MSHA inspectors. Other
commenters questioned the effect of sampling and work practices on the
validity of a sample.
The validity of the sampling process is an important aspect of
maintaining accurate measurements. Since passage of the Coal Act, there
has been an ongoing effort by MSHA and NIOSH to improve the accuracy
and reliability of the entire sampling process. In 1980, MSHA issued
new regulations revising sampling, maintenance and calibration
procedures in 30 CFR parts 70, 71, and 90. These regulatory provisions
were designed to minimize human and mechanical error and ensure that
samples collected with approved sampler units in the prescribed manner
would accurately represent the full-shift, average atmospheric dust
concentration at the location of the sampler unit. These provisions
require: (1) Certification of competence of all individuals involved in
the sampling process and in maintaining the sampling equipment; (2)
calibration of each sampler unit at least every 200 hours; (3)
examination, testing, and maintenance of units before each sampling
shift to ensure that the units are in proper working order; and (4)
checking of sampler units during sampling to ensure that they are
operating properly and at the proper flow rate. In addition,
significant changes, such as robotic weighing using electronic balances
were made in 1984, 1994, and 1995 that improved the reliability of
sample weighings at MSHA's Respirable Dust Processing Laboratory. These
changes are discussed below in section C.3.
All of these efforts improved the accuracy and reliability of the
sampling process since the time of the 1971/1972 proposed and final
findings. A discussion follows concerning the three elements which
constitute the sampling process: sampler unit performance, collection
procedures, and sample processing.
1. Sampler Unit Performance
In accordance with the provisions of section 202(e) of the Mine
Act, NIOSH administers a comprehensive certification process under 30
CFR part 74 to approve dust sampler units for use in coal mines. To be
approved for use, a sampler unit must meet stringent technical and
performance requirements governing the quantity of respirable dust
collected and flow rate consistency over an 8-hour period when operated
at the prescribed flow rate. NIOSH also conducts annual performance
audits of approved sampler units purchased on the open market to
determine if the units are being manufactured in accordance with the
specifications upon which the approval was issued.
The system of technical and quality assurance checks currently in
place is designed to prevent a defective sampler unit from being
manufactured and made commercially available to the mining industry or
to MSHA. In the event these checks identify a potential problem with
the manufacturing process, the system requires immediate action to
identify and correct the problem.
In 1992, NIOSH approved the use of new tamper-resistant filter
cassettes with features that enhanced the integrity of the sample
collected. A backflush valve was incorporated into the outlet of the
cassette, preventing reverse airflow through the filter cassette, and
an internal flow diverter was added to the filter capsule, reducing the
possibility of dust dislodged from the filter surface falling out of
the capsule inlet.
Several commenters questioned the quality of the filter cassettes
used in the sampling program, expressing concern about whether the
cassettes always meet MSHA specifications. These concerns primarily
involve filter-to-foil distance and floppiness of the filters, which
are manufacturing characteristics not related to part 74 performance
requirements. The Secretaries believe that such characteristics have no
effect on the accuracy of a single, full-shift measurement because,
unlike the part 74 requirements, they would not affect the amount of
dust deposition.
Commenters also questioned the condition of sampling pumps used by
MSHA inspectors, stating that many of the pumps are 10 to 20 years old
and are not maintained as well as they could be. They claimed that the
age and condition of these pumps call into question not only whether
the sampling equipment could meet part 74 requirements if tested, but
also the accuracy of the measurement.
This concern is unwarranted. In 1995, MSHA replaced all pumps in
use by inspectors with new constant-flow pumps that incorporate the
latest technology in pump design. These pumps provide more consistent
flow throughout the sampling period. In addition to using new pumps,
MSHA inspectors are required to make a minimum of two flow rate checks
to ensure that the sampler unit is operating properly. The sample is
voided if the proper flow rate was not being maintained during the
final check at the conclusion of the sampling shift. Units found not
meeting the requirements of part 74 are immediately repaired, adjusted,
or removed from service. Nevertheless, MSHA recognizes that as these
pumps age, deterioration of the performance of older pumps could become
a concern. However, there is no
[[Page 5673]]
evidence that the age of the equipment affects its operational
performance if the equipment is maintained as prescribed by 30 CFR
parts 70, 71, and 90.
Some commenters suggested that the accuracy of a dust sample may be
compromised when a miner is operating equipment, due to vibration from
the machinery. The potential effect of vibration on the accuracy of a
respirable dust measurement was recognized by NIOSH in 1981. An
investigation, supported by NIOSH, was conducted by the Los Alamos
National Laboratory which found that vibration has an insignificant
effect on sampler performance [4].
2. Sample Collection Procedures
MSHA regulations at 30 CFR parts 70, 71, and 90 prescribe the
manner in which mine operators are to take respirable dust samples. The
collection procedures are designed to ensure that the samples
accurately represent the amount of respirable dust in the mine
atmosphere to which miners are exposed on the shift sampled. Samples
taken in accordance with these procedures are considered to be valid.
Several commenters questioned the effects of sampling and work
practices on the validity of a sample. Instances were cited where the
sampling unit was accidentally dropped, with the potential for the
sample to become contaminated. Commenters also pointed out that work
activities requiring crawling, duck walking, bending, or kneeling could
cause the sampling hose to snag. Such activities could also cause the
sampling head assembly to be impacted or torn off a person's garment,
possibly contaminating the sample. These commenters stated that sampler
units are sometimes treated harshly while being worn by miners,
mishandled when being transferred from one miner to another, or handled
casually at the end of a work shift.
These commenters maintained that it is impossible for MSHA
inspectors or mine operators to continuously observe collection of a
sample in order to ensure its validity, and that, for this reason, the
reliability and accuracy of the sampling equipment, when used under
actual mining conditions, is not the same as when tested and certified
in a laboratory. Averaging multiple samples would, according to these
commenters, provide some ``leeway'' in the system, by reducing the
impact of an aberrant sample.
While MSHA and NIOSH agree that it is not possible to continuously
observe the collection of each sample, MSHA inspectors are normally in
the general vicinity of the sampling location, and therefore have
knowledge of the specific conditions under which samples are taken. In
addition, MSHA inspectors are instructed to ask miners wearing the
sampler units whether anything that could affect the validity of the
sample had occurred during the shift.
Other commenters expressed concern that, if special dust control
measures are in effect during sampling, a single, full-shift
measurement may fail to represent atmospheric conditions during shifts
when samples are not collected. The Secretaries believe that this
concern is beyond the scope of this notice, which, as described in the
discussion of measurement objective, deals solely with the accuracy of
a measurement in representing atmospheric conditions on the shift being
sampled. One commenter recommended that MSHA, NIOSH, or the Bureau of
Mines (now a part of NIOSH) should evaluate the need for standardizing
the MSHA respirable dust sampling procedures. In fact, the procedures
for respirable dust sampling are already standardized under the revised
1980 MSHA regulations codified at 30 CFR parts 70, 71 and 90.
MSHA inspectors will also begin using control filter capsules to
eliminate any bias that may be associated with day-to-day changes in
laboratory conditions or introduced during storage and handling of the
filter capsules. A control filter capsule is an unexposed filter
capsule that was pre-weighed on the same day as the filter capsules
used during a sampling inspection. These control filter capsules will
be carried by the inspector, but will remain plugged and not be exposed
to the mine environment.
3. Sample Processing
Sample processing consists of weighing the filter capsules,
recording the weight gains, and examining certain samples in order to
verify their validity. Sample processing also includes electronic
transmission of the results to MSHA's computer center where dust
concentrations are computed. The results are then distributed to MSHA
enforcement personnel and to mine operators.
(a) Weighing and recording procedures. One commenter cited a
personal experience in which anomalies were noted in the pre-exposed
weights recorded by the dust cassette manufacturer. The commenter was
concerned that such anomalies indicated poor quality control in the
manufacturer's weighing process, implying that this would cause a
significant number of single, full-shift measurements to be inaccurate.
The procedures and analytical equipment used by MSHA to process
respirable coal mine dust samples have improved since 1970. From 1970
to 1984, samples were manually weighed using semimicro balances. In
1984, the process was automated with a state-of-the-art robotic system
and electronic balances, which increased the precision of sample weight
determinations. Weighing precision was further improved in 1994, when
both the robotic system and balance were upgraded.
The full benefit of the 1994 improvements of the weighing system
for inspector samples was, however, not fully attained until mid-1995,
when MSHA implemented two modifications to its procedures for
processing inspector samples. One modification involved measuring both
the pre- and post-exposed weights to the nearest microgram (0.001 mg)
on a balance calibrated using the established procedure within MSHA's
laboratory. Prior to mid-1995, filter capsules had been weighed in the
manufacturer's laboratory before sampling, and then in MSHA's
laboratory after sampling. MSHA is now pre-weighing all such filter
capsules in its own laboratory, which will significantly reduce the
potential for anomalous pre-exposed weights of filter capsules used by
inspectors. To maintain the integrity of these pre-exposed weights,
eight percent of all capsules are systematically weighed a second time.
If a significant deviation is found, the balance is recalibrated and
all filter capsules with questionable weights are reweighed.
The other modification was to discontinue the practice of
truncating the recorded weights used in calculating dust concentration.
This means that MSHA no longer ignores digits representing hundredths
and thousandths of a milligram when processing inspector samples. These
modifications improved the overall accuracy of the measurement process.
To eliminate the potential for any bias that may be associated with
day-to-day changes in laboratory conditions or introduced during
storage and handling of the filter capsules, MSHA will use control
filter capsules in its enforcement program. Any change in weight of the
control filter capsule will be subtracted from the change in weight of
the exposed filter capsule.
(b) Sample validity checks. All respirable dust samples collected
and submitted as required by 30 CFR parts 70, 71, and 90 are considered
valid unless a questionable appearance of the filter capsule or other
special circumstances are noted that would
[[Page 5674]]
cause MSHA to examine the sample further. Several commenters expressed
concern about the potential contamination of samples with ``oversize
particles.'' Such contamination, according to one commenter, can result
in aberrational weight gains. These commenters noted that current
procedures do not systematically ensure that samples collected by MSHA
contain no oversize particles. It was recommended that MSHA analyze,
for the presence of oversize particles, any dust sample that exceeds
the applicable dust standard. Also suggested for such an analysis was
any sample with a weight gain significantly different from other
samples taken in the same area.
Standard laboratory procedures, involving visual, and microscopic
examination as necessary, are used to verify the validity of samples.
Samples weighing 1.4 milligrams (mg) or more are examined visually and
microscopically, as necessary, for abnormalities such as the presence
of large dust particles (which can occur from agglomeration of smaller
particles), abnormal discoloration, abnormal dust deposition pattern on
the filter, or any apparent contamination by materials other than
respirable coal mine dust. Also examined are samples weighing 0.1 mg or
less for insufficient dust particle count. Similar checks are also
performed in direct response to specific inspector or operator concerns
noted on the dust data card to which each sample is attached.
The commenters' concerns about the contamination of samples with
oversize particles are based on the assumption that all oversize
particles, defined as dust particles greater than 10 micrometers in
size, are not respirable and therefore should be totally excluded from
any sample taken with an approved sampler unit. In fact, it has long
been known that particles greater then 10 micrometers in size can be
inhaled, and that some of these particles can reach the alveoli of the
lungs [5]. According to the British National Coal Board, ``particles as
large as 20 microns (i.e. micrometers) mean diameter may be deposited,
although most ``lung dust'' lies in the range below 10 microns
diameter'' [6]. Furthermore, it is known that, due to the irregular
shapes of dust particles, the respirable dust collected by the MRE
instrument (the dust sampler used by the British Medical Research
Establishment in the epidemiological studies on which the U.S. coal
dust standard was based) may include some dust particles as large as 20
micrometers [6]. Moreover, MSHA studies have shown that nearly all
samples taken with approved sampler units, even when operated in the
prescribed manner, contain some oversize particles [7]. Since section
202(e) of the Mine Act defines concentration of respirable dust to be
that measured by an approved sampler unit, and because the approved
sampler unit will collect some oversize particles, the Secretaries do
not consider a sample to be ``contaminated'' because it contains some
oversize particles.
The Secretaries recognize that there are occasions when oversize
particles can properly be considered a contaminant. For example, an
excessive number of such particles could be introduced into the filter
capsule if the sampling head assembly is accidentally or deliberately
turned upside down or ``dumped'' (possibly causing some of the contents
of the cyclone grit pot to be drawn into the filter capsule), if the
pump malfunctions, or if the entire sampler unit is dropped. When MSHA
has reason to believe that such contamination has occurred, the suspect
sample is examined to verify its validity.
Contrary to the assertions of some commenters, checking for
oversize particles is not standard industrial hygiene practice.
Nevertheless, MSHA checks any dust sample suspected of containing an
excessive number of oversize particles. MSHA's laboratory procedures
require any sample exhibiting an excessive weight gain (over 6 mg) or
showing evidence of being ``dumped'' to be examined for the presence of
an excessive number of oversize particles. Samples identified by an
inspector or mine operator as possibly contaminated are also examined.
If this examination indicates that the sample contains an excessive
number of oversize particles according to MSHA's established criteria,
then that sample is considered to be invalid, and is voided and not
used. In fiscal year 1996, only 83 samples or 0.4 percent of the 20,331
inspector samples processed were found to contain an excessive number
of oversize particles and thus were not used.
While rough handling of the sampler unit or an accidental mishap
could conceivably cause a sample weighing less than 6 mg to become
contaminated, as claimed by some commenters, studies show that short-
term accidental inclinations of the cyclone will not affect respirable
mass measurements made with currently approved sampler units [8].
Sampler units currently used are built to withstand the rigors of the
mine environment, and are therefore less susceptible to contamination
than suggested by some commenters. In any event, the Secretaries
believe that the validity checks currently in place, as discussed
above, will detect such samples.
D. Measurement Uncertainty and Dust Concentration Variability
Overall variability in measurements collected on different shifts
and sampling locations results from the combination of errors
associated with the measurement of a particular dust concentration and
variability in dust concentration. Variability in dust concentration
refers to the differing atmospheric conditions experienced on different
shifts or at different sampling locations. Measurement uncertainty, on
the other hand, refers to the differing measurement results that could
arise, at a given sampling location on a given shift, because of
potential sampling and analytical errors.
Numerous commenters identified sources of measurement uncertainty
and dust concentration variability that they believed should be
considered when determining whether or not a measurement accurately
represents such atmospheric conditions. Because the measurement
objective is to accurately represent the average dust concentration at
the sampling location over a single shift, it does not take into
consideration dust concentration variability between shifts or
locations. Sources of dust concentration variability will not be
considered by the Secretaries in determining whether a measurement is
accurate. Consequently, the Secretaries have concluded that the only
sources of variability relevant to establishing accuracy of a single,
full-shift measurement for purposes of section 202(f) of the Mine Act
are those related to sampling and analytical error.
1. Sources of Measurement Uncertainty
Filter capsules are weighed prior to sampling. After a single,
full-shift sample is collected, the filter capsule is weighed a second
time, and the weight gain (g) is obtained by subtracting the pre-
exposure weight from the post-exposure weight, which will then be
adjusted for the weight gain or loss observed in the control filter
capsule. A measurement (x) of the atmospheric condition sampled is then
calculated by Equation 1:
[GRAPHIC] [TIFF OMITTED] TN31DE97.000
where: x is the single, full-shift dust concentration measurement (mg/m
\3\);
1.38 is a constant MRE-equivalent conversion factor;
[[Page 5675]]
g is the observed weight gain (mg) after adjustment for the control
filter capsule;
v is the estimated total volume of air pumped through the filter
during a typical full shift.
The Secretaries recognize that random variability, inherent in any
measurement process, may cause x to deviate either above or below the
true dust concentration. The difference between x and the true dust
concentration is the measurement error, which may be either positive or
negative. Measurement uncertainty arises from a combination of
potential errors in the process of collecting a sample and potential
errors in the process of analyzing the sample. These potential errors
introduce a degree of uncertainty when x is used to represent the true
dust concentration.
The statistical measure used by the Secretaries to quantify
uncertainty in a single, full-shift measurement is the total sampling
and analytical coefficient of variation, or CVtotal.
CVtotal quantifies the magnitude of probable sampling and
analytical errors and is expressed as either a fraction (e.g., 0.05) or
as a percentage (e.g., 5 percent) of the true concentration. For
example, if a single, full-shift measurement (x) is collected in a mine
atmosphere with true dust concentration equal to 1.5 mg/m \3\, and the
standard deviation of potential sampling and analytical errors
associated with x is equal to 0.075 mg/m \3\, the uncertainty
associated with x would be expressed by the ratio of the standard
deviation to the true dust concentration: CVtotal = 0.075/
1.5 = 5 percent.
Based on a review of the scientific literature, the Secretaries in
their March 12, 1996 notice, identified three sources of uncertainty in
a single, full-shift measurement, which together make up
CVtotal:
(1) CVweight--variability attributable to weighing
errors or handling associated with exposed and control filter capsules.
This covers any variability in the process of weighing the exposed or
control filter capsules prior to sampling (pre-weighing), assembling
the exposed and control filter cassettes, transporting the filter
cassettes to and from the mine, and weighing the exposed and control
filter capsules after sampling (post-weighing).
(2) CVpump--variability in the total volume of air
pumped through the filter capsule. This covers variability associated
with calibration of the pump rotameter,2 variability in
adjustment of the flow rate at the beginning of the shift, and
variation in the flow rate during sampling. It should be noted that
variation in flow rate during sampling was identified as a separate
component of variability in MSHA's February 18, 1994, notice. Here, it
is included within CVpump.
---------------------------------------------------------------------------
\2\ The rotameter consists of a weight or ``float'' which is
free to move up and down within a vertical tapered tube which is
larger at the top than the bottom. Air being drawn through the
filter cassette passes through the rotameter, suspending the
``float'' within the tube. The pump is ``calibrated'' by drawing air
through a calibration device (usually what is known as a bubble
meter)at the desired flow rate and marking the position of the float
on the tube. The processes of marking the position on the tube
(laboratory calibration) and adjusting the pump speed in the field
so that the float is positioned at the mark are both subject to
error.
---------------------------------------------------------------------------
(3) CVsampler--variability in the fraction of dust
trapped on the filter. This is attributable to physical differences
among cyclones. This component was introduced in the material submitted
into the record in September 1994.
These three components of measurement uncertainty can be combined
to form an indirect estimate of CVtotal by means of the
standard propagation of errors formula:
[GRAPHIC] [TIFF OMITTED] TN31DE97.001
These three components are discussed in greater detail, along with
responses to specific comments, in Appendix B.
2. Sources of Dust Concentration Variability
Numerous commenters also raised issues related to sources of dust
concentration variability. Some of these commenters maintain that the
Secretaries should include in CVtotal additional components
representing the effects of shift-to-shift variability and variability
related to location (spatial variability). These comments reflect a
misunderstanding of the measurement objective as intended by the Mine
Act (see section VII.A. of this notice).
Exposure variability due to job, location, shift, production level,
effectiveness of engineering controls, and work practices will be
different from mine to mine, and is under the control of the mine
operator. The sampler unit is not intended to account for these
factors.
(a) Spatial variability. Several commenters stated that
CVtotal should account for spatial variability, or the
differences in concentration related to location. The Secretaries agree
that dust concentrations vary between locations in a coal mine, even
within a relatively small area. However, real variations in
concentration between locations, while sometimes substantial, do not
contribute to measurement error. As stated earlier, the measurement
objective is to accurately measure average atmospheric conditions, or
concentration of respirable dust, at a sampling location over a single
shift.
(b) Shift-to-shift variability. Several commenters stated that
CVtotal should take into account the differences or
variations in dust concentration that occur shift to shift. Although
the Secretaries agree that dust concentrations vary from shift to
shift, the measurement objective is to measure average atmospheric
conditions on the specific shift sampled. This result is consistent
with the Mine Act, which requires that concentrations of respirable
mine dust be maintained at or below the applicable standard during each
shift.
3. Other Factors Considered
(a) Proportion of oversize particles. Several commenters expressed
concern that respirable dust cyclones are handled in a rough manner in
normal use and occasionally turned upside down. According to one
commenter, this type of handling would cause more large particles to be
deposited on the filter in the mine environment than when used in the
laboratory. This commenter knew of no data that could be used to
evaluate the error associated with such occurrences and recommended
that a study be commissioned to measure the proportion of non-
respirable particles on the filters after they are weighed to MSHA
standards.
After considering this recommendation, the Secretaries have
concluded that the available evidence shows that short-term
inclinations of the cyclone, as might frequently occur during sampling,
will not affect respirable dust measurements made with approved sampler
units [8]. The weight of the sampler head assembly makes it extremely
unlikely that a
[[Page 5676]]
sampler unit could be turned upside down in normal use. Furthermore,
with a field study of the type recommended, variability in the field
measurements due to normal handling would be confounded with
variability due to real differences in atmospheric conditions.
Therefore, the Secretaries believe that such a study would not be
useful in establishing variability in measurements due to differences
in handling of the sampler unit.
(b) Anomalous events. Several commenters asserted that
unpredictable, infrequent events, such as a ``face blowout'' on a
longwall (a violent expulsion of coal together with large quantities of
coal dust and/or methane gas) or high winds at a surface mine, can
cause rapid loading of a filter capsule and thereby distort a
measurement to show an excessive dust concentration based on a single,
full-shift sample when, they argue, the dust standard had not been
exceeded. In fact, if such an occurrence were to cause a measurement
above the applicable standard, the dust standard would in fact be
violated. No evidence was presented to demonstrate that short-term high
exposures can overload a dust sampling filter or cause the sampling
device to malfunction. Nor was evidence presented to demonstrate that
miners are not also exposed to the same high dust concentrations as the
sampler unit when such events occur. The Secretaries conclude that such
events are results of the dynamic and ever-changing mine environment--
an environment to which the miner is exposed. The sampler unit is
designed to measure the atmospheric condition at a specific sampling
location over a full shift. If such events occur, the sampler unit will
accurately record the atmospheric condition to which it is exposed.
(c) Conversion factor used in the dust concentration calculation.
Several commenters questioned the 1.38 MRE-conversion factor used in
Equation 1. This factor is used to convert a measurement obtained with
the type of dust sampler unit currently approved for use in coal mines
to an equivalent concentration as measured with an MRE gravimetric dust
sampler. The term ``MRE instrument'' is defined in 30 CFR Sec. 70.2(I).
The conversion factor is necessary because the coal mine dust standard
was derived from British data collected with an MRE instrument, which
collects a larger fraction of coal mine dust than does the approved
dust sampling unit [9]. The 1.38 constant has been established by the
Secretaries as applying to the currently approved dust sampler unit
described in 30 CFR part 74.
Some commenters contended that variability involved in the data
analysis used in establishing the conversion factor should be taken
into account in determining CVtotal. This suggestion
demonstrates a misunderstanding of the difference between measurement
imprecision and measurement bias. The 1.38 factor applies to every
sampler unit currently approved under part 74. Since the same
conversion factor is applied to every measurement, any error in the
value used would cause a measurement bias but would have no effect on
measurement imprecision. Since Congress defined respirable dust in
section 202(e) of the Mine Act as whatever is collected by a currently
approved sampler unit, a measurement incorporating the 1.38 factor is
unbiased by definition. Further discussion is provided in Appendix A on
why use of the 1.38 factor does not introduce a bias. Appendix A also
addresses comments relating to other aspects of the 1.38 conversion
factor; comments regarding the fact that MSHA's sampler unit does not
conform to other definitions of respirable dust; and questions
concerning the effect of static charge on sampler unit performance.
(d) Reduced dust standards. One commenter pointed out that in
estimating CVtotal, MSHA and NIOSH did not take into account
any potential errors associated with silica analysis. The commenter
argued that since silica analysis is used to establish reduced dust
standards, MSHA and NIOSH had failed to demonstrate ``* * * accuracy
for all samples `across the range of possible reduced dust standards.'
''
This commenter confuses the accuracy of a respirable dust
concentration measurement with the accuracy of the procedure used to
establish a reduced dust standard. MSHA has a separate program in which
silica analysis is used to set the applicable respirable coal mine dust
standard, in accordance with section 205 of the Mine Act, when the
respirable dust in the mine atmosphere of the active workings contains
more than 5 percent quartz. As shown by Equation 1, no silica analysis
is used in a single, full-shift measurement of the respirable dust
concentration. Therefore, the Secretaries do not agree with the comment
that CVtotal should include a component representing
potential errors in silica analysis.
(e) Dusty clothing. Several commenters pointed out that local
factors such as dusty clothing could cause concentrations in the
immediate vicinity of the sampler unit to be unrepresentative of a
larger area. Dust from a miner's clothing nevertheless represents a
potential hazard to the miner. No evidence was presented to demonstrate
that miners are not also exposed to dust originating from dusty
clothing.
E. Accuracy of a Single, Full-Shift Measurement
1. Quantification of Measurement Uncertainty
Several commenters argued that MSHA underestimated
CVtotal in its February 18, 1994 notice and suggested
alternative estimates ranging from 16 to 50 percent. These commenters
cited several published studies and submitted five sets of data in
support of these higher estimates. Statistical analyses of the data
were also submitted.
MSHA and NIOSH reviewed all of the studies referenced by the
commenters. The review showed that all of the estimates of measurement
variability were from studies carried out prior to improvements
mandated by the 1980 MSHA revisions to dust sampling regulations,
discussed earlier in ``Validity of the Sampling Process.'' For example,
the General Accounting Office (GAO) 3 and the National
Bureau of Standards (NBS, now the National Institute of Standards and
Technology) studies were conducted in 1975. The National Academy of
Sciences report, which analyzed the same data as the NBS and GAO
reports, was issued in 1980. The review further showed that the
measurement variability quantified in these studies included effects of
spatial variability--a component of variability the Secretaries
deliberately exclude when determining the accuracy of a sampling and
analytical method as discussed in section D.2.(a). Additionally, since
past studies frequently relied on combining estimates of variability
components obtained from different bodies of data, some of them also
suffered from methodological problems related to combining individual
sources of uncertainty. For example, in 1984, a NIOSH study identified
several conceptual errors in earlier studies that had led to double-or
even triple-counting of some variability components [10].
---------------------------------------------------------------------------
\3\ Many of the recommendations in the GAO report were later
adopted and implemented by MSHA.
---------------------------------------------------------------------------
Although all the data and analyses submitted by commenters included
effects of spatial variability, one of these data sets, consisting of
paired sample results, contained sufficient information to indicate
that weighing imprecision
[[Page 5677]]
was less than what MSHA had assumed in its February 18, 1994 notice.
However, without an independent estimate of spatial variability
applicable to these samples, it is not mathematically possible to
utilize this data set to estimate variability attributable to the
sampler unit or the volume of air sampled. A second data set consisted
only of differences in dust concentration between paired samples,
making it impossible to use it even for evaluating weighing
imprecision. The remaining three data sets included effects of shift-
to-shift variability, which, like spatial variability, is not relevant
to the measurement objective. Therefore, none of these data could be
used to estimate overall measurement imprecision. Further details are
provided in Appendix C.
One of the commenters particularly questioned the value MSHA used
in its February 18, 1994 notice to represent variability in initially
setting the pump flow rate. In response to this commenter's suggestion,
MSHA conducted a study to verify the magnitude of this variability
component. This study simulated flow rate adjustment under realistic
operating conditions by including a number of persons checking and
adjusting initial flow rate under various working situations [11].
Results showed the coefficient of variation associated with the initial
flow rate adjustment to be 30.5 percent, which is less than
the 5-percent value used by MSHA in the February 1994 notice. In
addition, based on a review of published results, the Secretaries have
concluded that the component of uncertainty associated with the
combined effects of variability in flow rate during sampling and
potential errors in calibration is actually less than 3 percent. As
explained in Appendix B, these two sources of uncertainty can be
combined to estimate CVpump. After reviewing the available
data and the comments submitted, the Secretaries have concluded that
the best estimate of CVpump is 4.2 percent. Additional
details regarding CVpump, along with the Secretaries'
responses to comments, are presented in Appendix B.
Intersampler variability, represented by CVsampler,
accounts for uncertainty due to physical differences from sampler to
sampler. Most of the commenters ignored this source of uncertainty. As
explained in Appendix B, the Secretaries have adopted a 5-percent
estimate of CVsampler.
To address commenters' concerns that the Agencies had
underestimated CVtotal, MSHA conducted a field study to
directly estimate the overall measurement precision attainable when
dust samples are collected with currently approved sampler units and
analyzed using state-of-the-art analytical techniques. The study
involved simultaneous field measurements of the same coal mine dust
cloud using sampling pumps incorporating constant flow technology.
Using a specially designed portable dust chamber, 22 tests were
conducted at various locations in an underground coal mine. Each test
consisted of collecting 16 dust samples simultaneously and at the same
location. No adjustments in the flow rate were made beyond what would
routinely have been done by an MSHA inspector.
Prior to the field study, two modifications to MSHA's sampling and
analytical method had been considered by MSHA and NIOSH: (1) Measuring
both the pre-and post-exposure weights to the nearest microgram
(g) on a balance calibrated using the established procedure
within MSHA's Respirable Dust Processing Laboratory; and (2)
discontinuing the practice of truncating the recorded weights used in
calculating the dust concentration. These modifications were
incorporated into the design of the field study.
One commenter characterized the field study as being ``woefully
incomplete'' because it was conducted ``in a tightly controlled
environment * * * not subject to normal environmental variation.''
While it is true that the samples within each test were not subject to
normal environmental variability, this was because the experiment was
deliberately designed to avoid confusing spatial variability in dust
concentration with measurement error. However, pumps were handled and
flow rates were checked in the same manner as during routine sampling.
Furthermore, the sampler units were disassembled and reassembled in the
normal manner to remove and replace dust cassettes.
Commenters also questioned the value that MSHA used in the February
1994 notice to represent uncertainty due to potential weighing errors.
In September 1994, MSHA submitted into the record an analysis based on
replicated weighings for 300 unexposed filter capsules, each of which
was weighed once by the cassette manufacturer and twice in MSHA's
laboratory [12]. An estimate of weighing imprecision derived from this
analysis was used by NIOSH in its September 20, 1995 assessment of
MSHA's sampling and analytical procedure (discussed in more detail
later).
In the March 12, 1996 joint notice, MSHA described the results of
an investigation into repeated weighings of the same capsules made over
a 218-day period using MSHA's automatic weighing system. It was noted
that after approximately 30 days, filter capsules left exposed and
unprotected gained a small amount of weight--an average of 0.8
g (micrograms) per day. Neither NIOSH nor MSHA considered this
a problem, since all dust samples are analyzed within 24 hours of
receipt and are not left exposed and unprotected. However, more recent
data collected to quantify weighing variability between the MSA and
MSHA laboratories showed that filter capsules tend to gain a small
amount of weight even when stored in plastic cassettes [13]. To check
this result, 75 unexposed filter cassettes that had been distributed to
MSHA's district offices were recalled and the filter capsules were
reweighed. On average, the weight gain was about 40 g over a
time period of roughly 150 days. Statistical analyses of these data
performed by MSHA and NIOSH confirmed the previous result [13,14].
While the cause has not been established, it is hypothesized that at
least some of the observed weight gain may be the result of outgassing
from the plastic cassette onto the filter capsule. If uncorrected, any
systematic change in weight not due to coal mine dust would introduce a
bias in dust concentration measurements.
One commenter had previously stated that the Secretaries were
addressing only precision, thereby implying that potential biases were
being ignored. To eliminate the potential for any bias due to a
spurious gain or loss of filter capsule weight, MSHA will use control
filter capsules in its enforcement program. Any change in weight
observed for the control filter capsule will be subtracted from the
measured change in weight of the exposed filter capsule. Each control
filter capsule will be pre-weighed with the other filter capsules, will
be stored and transported with the other capsules, and will be on the
inspector's person during the day of sampling. This modification to
MSHA's inspector sampling and analytical procedure will assure an
unbiased estimate of the true weight gain [14].
2. Verification of Method Accuracy
With its field study, MSHA exceeded the usual requirements for
determining the accuracy of a sampling and analytical method, as
described by NIOSH [1] and the European Community [2]. Both of these
require only a laboratory determination of method accuracy. NIOSH's
independent analysis of the study data determined, with 95-percent
confidence, that the
[[Page 5678]]
true CVtotal for MSHA's sampling and analytical method is
less than the target maximum value of 12.8 percent for concentrations
ranging from 0.2 mg/m3 to greater than 2 mg/m3
[3]. In other words, NIOSH demonstrated that, with two recommended
modifications, MSHA's sampling and analytical method for collecting and
processing single, full-shift samples would meet the NIOSH Accuracy
Criterion at dust concentrations greater than or equal to 0.2 mg/
m3.
NIOSH also applied an indirect approach for assessing the accuracy
of MSHA's sampling and analytical method. The indirect approach
involved combining independently derived estimates, previously placed
into the public record, of intra-laboratory weighing imprecision, pump-
related variability, and variability associated with physical
differences between individual sampler units. This indirect approach
also indicated that MSHA's sampling and analytical method meets the
NIOSH Accuracy Criterion at concentrations greater than or equal to 0.2
mg/m3, thereby corroborating the analysis of MSHA's field
data.
These NIOSH analyses predate MSHA's more recent data indicating a
correctable weight gain bias (discussed above). As explained in
Appendices A and B, the use of control filter capsules will eliminate
this bias but also affect the precision of a single, full-shift
measurement. Consequently, NIOSH reassessed the accuracy of MSHA's
sampling and analytical method, taking into account the effect of using
a control filter capsule on the measurement process [14]. After
accounting for the effects of control filter capsules on both bias and
precision, NIOSH concluded, based on both its direct and indirect
approaches, that a single, full-shift measurement will meet the NIOSH
Accuracy Criterion at dust concentrations greater than or equal to 0.3
mg/m3.
One commenter claimed that the Secretaries ``have not addressed the
`accuracy' of a single sample collected from an environment where the
concentration is unknown''. The purpose of any measurement process is
to produce an estimate of an unknown quantity. Since the Secretaries
have concluded that MSHA's sampling and analytical method for
inspectors meets the NIOSH Accuracy Criterion for true concentrations
ranging from 0.3 mg/m3 to greater than 2 mg/m3,
it is possible to calculate the range of measurements for which the
Accuracy Criterion applies. Since CVtotal increases at the
lower concentrations, it is important to determine the lowest
measurement at which the NIOSH Accuracy Criterion is met. If the true
concentration exactly equaled the lowest concentration at which MSHA's
sampling and analytical method meets the Accuracy Criterion (i.e., 0.3
mg/m3), no more than 5% of single, full-shift measurements
would be expected to exceed 0.36 mg/m3 [14]. Conversely, if
a measurement equals or exceeds 0.36 mg/m3, it can be
inferred, with at least 95% confidence, that the true dust
concentration equals or exceeds 0.3 mg/m3 [14].
Consequently, the Secretaries conclude that MSHA's improved sampling
and analytical method satisfies the NIOSH Accuracy Criterion whenever a
single, full-shift measurement is at or above 0.36 mg/m3.
As a result of the prior analyses, MSHA's existing inspector sample
processing procedures were changed to reflect the modifications that
were incorporated into MSHA's field study. MSHA is now pre- and post-
weighing inspector samples in the same laboratory, and reporting the
pre- and post-exposure weights of inspector samples to the nearest
microgram (g). As a result of NIOSH's latest analysis, MSHA
will now require its inspectors to use control filter capsules during
sampling. In addition, MSHA is now using only constant-flow control
pumps in the inspector sampling program. MSHA believes that exclusive
use of constant-flow pumps, as in the field study, further enhances the
quality of the Agency's sampling program.
The Secretaries recognize that future technological improvements in
MSHA's sampling and analytical method may reduce CVtotal
below its current value. Also, as additional data are accumulated,
updated estimates of CVtotal may become available. However,
so long as the method remains unbiased and CVtotal remains
below 12.8 percent, at a 95-percent confidence level, the sampling and
analytical method will continue to meet the NIOSH Accuracy Criterion,
and the present finding will continue to be valid.
VIII. Finding
The Secretaries have concluded that sufficient data exist for
determining the uncertainty associated with a single, full-shift
measurement; rigorous requirements are in place, as specified by 30 CFR
parts 70, 71, and 90, to ensure the validity of a respirable coal mine
dust sample; and valid statistical techniques were used to determine
that MSHA's improved dust sampling and analytical method meets the
NIOSH Accuracy Criterion. For these reasons the Secretaries find that a
single, full-shift measurement at or above 0.36 mg/m3 will
accurately represent atmospheric conditions to which a miner is exposed
during such shift. Therefore, pursuant to section 202(f) and in
accordance with section 101 of the Mine Act, the 1972 joint notice of
finding is hereby rescinded.
Appendix A--Why Individual Measurements are Unbiased
The accuracy of a measurement depends on both precision and bias
[1,3]. Precision refers to consistency or repeatability of results, and
bias refers to an error that is equally present in every measurement.
Since the amount of dust present on a filter capsule is measured, for
MSHA inspector samples, by subtracting the pre-exposure weight from the
post-exposure weight observed in the same laboratory, any bias in the
weighing process attributable to the laboratory is mathematically
canceled out by subtraction. A control filter capsule will be pre- and
post-weighted along with the exposed filter capsules. The weight gain
of each exposed capsule will be adjusted by subtracting the weight gain
or loss of the control filter capsule. Consequently, any bias
introduced during storage and handling of the filter capsules is also
mathematically canceled out. Therefore, since respirable dust is
defined by section 202(e) of the Mine Act to be whatever is measured by
an approved sampler unit, the Secretaries have concluded that a single,
full-shift measurement made with an approved sampler unit provides an
unbiased representation of average dust concentration for the shift and
sampling location sampled. Some commenters, however, suggested that
MSHA's sampling and analytical method is subject to systematic errors
that would have the same effect on all measurements. These comments are
addressed in this appendix.
I. The Value of the MRE Conversion Factor
The current U.S. coal mine dust standard is based on studies of
British coal miners. In these studies, full-shift dust measurements
were made using a sampler employing four horizontal plates which
removed the large-sized particles by gravitational settlement
(simulating the action of the nose and throat) and collecting on a pre-
weighed filter those particles which are normally deposited in the
lungs [6]. This instrument, known as the Mining Research Establishment
(MRE) sampler, was designed to collect airborne dust according to a
collection efficiency curve, developed by the British Medical Research
Council (BMRC) to approximate the deposition of inhaled
[[Page 5679]]
particles in the lung. Because the MRE instrument was large and
cumbersome, other samplers using a 10-mm nylon cyclone were developed
for taking samples of respirable dust in U.S. coal mines. However,
these cyclone-based samplers collected less dust than the MRE
instrument. Therefore, a factor was derived (1.38) to convert
measurements obtained with the cyclone-based samplers to measurements
obtained with the MRE instrument.
Two commenters noted that the 1.38 conversion factor was derived
from a comparison of MRE measurements to measurements obtained using
pumps made by two manufacturers [Mine Safety Appliances Co. (MSA) and
Unico]. These commenters noted that there was some variability in these
comparisons that MSHA and NIOSH did not consider in estimating
CVtotal, and noted that MSHA and NIOSH should therefore make
allowances for any error or uncertainty in the conversion factor. It
was also noted that the report deriving the conversion factor showed
that MSA pumps more closely approximated MRE concentrations than Unico
pumps, indicating that the 1.38 conversion factor (derived empirically
using both types of pumps) may systematically overestimate the MRE-
equivalent dust concentration for MSA samplers specifically. This
commenter argued that such potential bias in the conversion factor
should be addressed in order to account for the possibility of a
systematic error in the conversion.
The study referred by these commenters involved collecting side-by-
side samples using MRE and cyclone-based samplers [9]. The data showed
that multiplying the cyclone sample concentrations by a constant factor
of 1.38 gave values in reasonable agreement with MRE measurements.
Consequently, a conversion factor of 1.38 was adopted for use with
approved sampler units equipped with the 10-mm nylon cyclone.
Variability in the operating characteristics of individual sampler
units is expressed by CVsampler. In response to the comment
on potential bias, MSHA and NIOSH reviewed the original report
recommending the 1.38 MRE conversion factor. This report contained both
an empirical determination, using side-by-side comparison data
collected in underground coal mines, and a theoretical determination of
the conversion factor. Two sets of field data were collected: one set
was collected by mine inspectors who visited 200 coal mines across the
U.S.; the other set was collected by investigators from MSHA's
Pittsburgh laboratory at 24 coal mines. Linear regression was used to
analyze both sets of data, with the slope of the regression line
representing the conversion factor. The theoretical determination
suggested that the conversion factor should be close to a value of
1.35. Analysis of the district mine inspector data resulted in a
conversion factor of 1.38, while analysis of the laboratory
investigator data suggested a greater conversion factor of 1.45.
Because the conversion factor derived from the inspector data came
closer to the theoretical value, the former U.S. Bureau of Mines'
Pittsburgh Technical Support Center (in the Department of Interior)
recommended that 1.38 be the value adopted for any approved sampler
unit operating at 2.0 L/min and equipped with a 10-mm nylon cyclone.
This recommendation was subsequently accepted. The 1.38 conversion
factor was not, as implied by the commenters, meant to represent the
average value to be used with two different types of sampler unit, one
of which is no longer in use. Instead, based largely on the theoretical
value, it was meant to represent the appropriate value to be used with
any approved sampler unit operating at 2.0 L/min and equipped with a
10-mm nylon cyclone. No data or analyses were submitted to suggest that
this conversion factor, which has been accepted and used for over
twenty years, should be any other value.
II. Conforming to the ACGIH and ISO Standard
One commenter implied that the respirable dust cyclone
specifications used by MSHA result in a different particle collection
efficiency curve than that specified by the American Conference of
Governmental Industrial Hygienists (ACGIH) and the International
Organization for Standardization (ISO) for a respirable dust sampler.
Other commenters questioned whether the 2.0 L/min flow rate used by
MSHA was appropriate, since a NIOSH study recommended using a 1.7 L/min
flow rate when conforming to the recently adopted ACGIH/ISO
specifications for collecting respirable particulate mass.
It is true that MSHA's respirable dust cyclone specifications
result in a different particle size distribution than that specified by
ACGIH and ISO. However, this fact has no bearing on the conversion to a
respirable dust concentration as measured by an MRE sampler, which is
the basis of the respirable dust standard. The 1.38 factor used to
obtain an MRE-equivalent concentration was derived for a cyclone flow
rate of 2.0 L/min. If a flow rate of 1.7 L/min were used, then this
would correspond to some other factor for converting to an MRE-
equivalent dust concentration. Therefore, the particle size
distribution obtained at 2.0 L/min governs the relationship derived
between an approved respirable coal mine dust sampler and an MRE
sampler. The appropriate dust fraction (i.e., the fraction
corresponding to the 1.38 conversion factor) is sampled so long as the
specified 2.0 L/min flow rate is maintained.
III. Effects of Other Variables
The effects of any other variables on the sampled dust fraction are
covered by the 1.38 conversion factor, so long as these effects were
present in the data from which the conversion factor was obtained. For
example, one commenter expressed concern that nylon cyclones are
subject to performance variations due to static charging phenomena. Any
systematic effect of static charging on the performance characteristics
of the nylon cyclone is implicitly accounted for in the conversion
factor, because the same static charging effect would have been present
when the comparative measurements were obtained for deriving the
relationship between an approved sampler unit and an MRE instrument.
Random effects of static charging, i.e., effects that vary from sample
to sample, are included in CVtotal.
Appendix B--Components of CVtotal
I. Weighing Uncertainty
(a) Derivation of CVweight
The weight of a dust sample is determined by weighing each filter
capsule before and after exposure and then determining the weight gain
by subtraction. This weight gain is adjusted by subtracting any change
in weight observed for the unexposed, control filter capsule. This
practice eliminates potential biases due to any possible outgassing of
the plastic cassette or other time-related factors but introduces two
additional weighings. The weighing process is designed to control
potential effects of temperature, humidity, and contamination. However,
because the initial and final weighings of both the exposed and the
control filter capsules are each still subject to random error, there
is some degree of uncertainty in the computed weight of dust collected
on the filter.
For both the control and the exposed filter capsule, the error in
the weight-gain measurement results from combining two independent
weighing errors. For example, suppose that the true pre- and post-
exposure weights of
[[Page 5680]]
a filter capsule are W1=392.275 mg and W2=392.684
mg, respectively. The true weight gain (G) would then be:
G=W2-W1=0.409 mg.
If, due to weighing errors, pre- and post-exposure weights were
measured at w1=392.282 mg and w2=392.679 mg,
respectively, then the measured weight gain (g) would be:
g=w2-w1=0.397 mg.
The error (e) in this particular weight-gain measurement, resulting
from the combination of a 7 g error in w1 and a -5
g error in w2, would then be:
e=g-G=(w2-w1)-(W2-W1)=(w2
-W2) -(w1-W1)=-5-7=-12
g.4
\4\ Prior to mid-1995 there were two additional sources of
uncertainty in the weight gain recorded for MSHA inspector samples.
First, filter capsules were routinely weighed in different
laboratories before and after exposure, subjecting them to
interlaboratory variability. Second, the pre- and post-exposure
weights were both truncated down to the nearest exact multiple of
0.1 mg, below the weight actually measured, prior to recording
weight gain and calculating dust concentration.
---------------------------------------------------------------------------
Imprecision in the true weight gain is expressed by
e, the standard deviation of e. When a weight-gain
measurement (g) is converted to an MRE-equivalent concentration (in
units of mg/m3) based on a 480-minute sample at 2.0 L/min,
both the actual weight gain (G) and the weight-gain error (e) are
multiplied by the same factor:
[GRAPHIC] [TIFF OMITTED] TN31DE97.002
Therefore, the standard deviation of the propagated weighing error
component in a single, full-shift measurement (x=g1.438/m \3\ ) is
1.438e mg/m \3\, assuming no adjustment for weight
change in the control filter capsule.
Since a control filter capsule will be used to eliminate potential
bias, the weight gain measured for the exposed filter (g) will be
adjusted by subtracting the change in weight (which may be positive or
negative) observed for the control filter capsule (g'). Therefore, the
adjusted measurement of dust concentration is
[GRAPHIC] [TIFF OMITTED] TN31DE97.003
Any change in weight observed for the control filter capsule is
subject to the same measurement imprecision due to random weighing
errors, represented by e, as the weight gain
measurement for an exposed filter. In addition to the weight-gain error
for the exposed filter whose measured weight gain is g, x' will also
contain a weight-gain error contributed by the measured change in
weight of the control filter capsule (g'). Using a standard
propagation-of-errors formula, the imprecision in g-g' is represented
by
[GRAPHIC] [TIFF OMITTED] TN31DE97.004
Therefore, the standard deviation of the propagated weighing error
component in the adjusted measurement is
1.438e2 mg/m \3\.
To form an estimate of CVweight when control filter
capsules are used, the estimated value of 1.438e is
multiplied by 2 and expressed as a percentage of the true dust
concentration being measured (X):
[GRAPHIC] [TIFF OMITTED] TN31DE97.005
Since e is essentially constant with respect to
dust concentration, CVweight decreases as the dust
concentration increases.
(b) Values Expressing Weight-Gain Uncertainty
Table 1 summarizes six different values of
e that have been mentioned during the proceedings
related to this notice and two additional values for
e derived in this appendix from data introduced
during these proceedings. A ninth value for e is
derived from newly acquired data being placed into the record along
with this notice [14]. The nine values listed in Table 1 are not
inconsistent, but as explained below, represent estimates of weight-
gain imprecision during different historical periods or under different
sample processing procedures.
Table 1.--Standard Deviation of Error in Weight Gain
----------------------------------------------------------------------------------------------------------------
e
Description Reference (g)
----------------------------------------------------------------------------------------------------------------
MSHA's historical estimate of upper bound............ 59 FR 8356, [15].......................... 97.4
1981 Measurement Assurance Estimate (older [16,17]................................... 81
technology, truncation of weights).
Experiment on 300 unexposed, tamper-resistant filter [12]...................................... 29
capsules (pre- and post-weighing in different labs;
no truncation).
Inspector samples processed between late 1992 and mid Appendix B................................ 51.7
1995 (truncation of weights; pre- and post-exposure
weighing in different labs; adjusted for differences
between labs).
NMA Data (obtained from samples collected by Skyline Appendix C................................ 76
Coal, Inc.).
Value used in NIOSH ``indirect approach'' (pre- and 61 FR 10012, [12]......................... 5.8
post-exposure weighing on same day and in the same
lab; derived from Kogut [12]).
MSHA Field Study..................................... [18,3].................................... 9.1
1996 Measurement Assurance Estimate.................. 61 FR 10012, [19]......................... 6.5
1997 field data (75 unexposed capsules).............. [14]...................................... 8.2
----------------------------------------------------------------------------------------------------------------
In MSHA's February 1994 notice, 1.438e
(identified as ``variability associated with the pre- and post-weighing
of the filter capsule'') was presented as 0.14 mg/m3, or 7
percent of 2.0 mg/m3, as described in Kogut [15]. It follows
that the value of e implicitly assumed in MSHA's
February 1994 notice (obtained by dividing 0.14 by 1.438) was 0.0974 mg
(97.4 g). Seven percent of 2.0 mg/m3 had been used
by MSHA from the inception of its dust enforcement program to represent
an upper bound on weighing imprecision in a dust concentration
measurement.
After publication of the February 1994 notice, several other
candidate values for e were placed into the public
record. In 1981, based on data collected to implement a measurement
assurance program in MSHA's weighing laboratory, e
was estimated using a method developed by the NBS to be 0.0807 mg (80.7
g) [16]. The published NBS estimate reflected weighing
technology in place at the time the article was published (1981), as
well as the practice (no longer in effect for MSHA inspector samples)
of truncating both the pre- and post-exposure weights down to an exact
multiple of 0.1 mg. This estimate was used to calculate
CVweight by Bartley [17], in September 1994.
Some commenters misread or misunderstood the published NBS
estimate. One of these commenters claimed that ``the only published
report of the weighing error in MSHA's
[[Page 5681]]
laboratory * * * was 0.16 mg of variation, which would convert to a
concentration of 0.20 mg/m3 compared to the 0.14 mg/
m3 * * * MSHA and NIOSH used.'' This is incorrect, since the
standard deviation of weight-gain errors (including the effect of
truncation) is actually identified as 0.0807 mg in the Appendix to
Parobeck et al. [16]. The 0.16-mg figure quoted by the commenter is
presented in that paper as defining a 2-tailed 95-percent confidence
limit, for use in establishing process control limits. It is derived by
multiplying e by 2.0. As explained above, the
published value of e = 0.0807 mg is multiplied by
1.438 to propagate an MRE-equivalent concentration error of 0.116 mg/
m3. Contrary to the commenters' assertion, this is less--not
more--than the quantity (0.14 mg/m3) assumed in the February
1994 notice.
In September 1994, a more recent analysis was placed into the
public record, based on repeated weighings of 300 unexposed filter
capsules, each of which was weighed once in the MSA laboratory and
twice in MSHA's laboratory using current equipment [12]. Based on this
analysis, e was estimated to be 29 g for
pre- and post-weighings on different days at different laboratories, or
5.8 g for pre- and post-weighings on the same day within
MSHA's laboratory. The 5.8-g value was used as part of the
NIOSH ``indirect approach'' in its 1995 accuracy assessment [3].
Neither of these two estimates, however, reflects the effects of
truncation or of a mean difference of about 12 g discovered
between weighings in the two laboratories. Combining these two
additional effects with the 29-g estimate results in an
adjusted estimate of e = 51.7 g for
weighings made in different laboratories and truncated to a multiple of
0.1 mg. MSHA and NIOSH regard this 51.7-g value to be the best
available estimate of e for inspector samples
processed between late 1992, when the current style of (tamper-
resistant) cassette was introduced, and mid-1995, when the most recent
changes in inspector sample processing were implemented.
Some commenters suggested that the estimates of
e, placed into the record in September 1994, did
not adequately account for potential errors in the weighing process as
it existed at that time. One of these commenters asserted that
truncation error was an additional source of uncertainty that had not
been accounted for. As explained above, however, e
accounts for uncertainty deriving from both the pre- and post-exposure
weighings. Both the 80.7-g NBS estimate and the 97.4-
g value assumed in the February 1994 notice included the
effects of truncating weight measurements to 0.1 mg. Truncation effects
are also included in the 51.7-g estimate.
Some commenters expressed special concern over the accuracy of pre-
exposure filter capsule weights as measured by MSA. One commenter
expressed ``grave concern'' with regard to the 12-g systematic
difference in weights found between MSA and MSHA weighings of the same
unexposed capsules, as described in MSHA's 1994 analysis [12]. These
concerns are moot, at least with respect to MSHA's inspector sampling
program, since all inspector samples are now pre- and post-weighed at
MSHA's laboratory. Furthermore, any potential bias resulting from
differences in laboratory conditions on the days of pre- and post-
exposure weighings should be eliminated by the use of control filter
capsules. However, contrary to this commenter's interpretation, the
analysis submitted to the record in September 1994 resulted in a
substantially lower estimate of e than that assumed
in the February 1994 notice--even after adjustment for the 12-
g systematic difference observed between weighing
laboratories. The 51.7-g estimate discussed above includes
this adjustment.
MSHA and NIOSH also analyzed data submitted by the NMA in
connection with these proceedings. An important result of that
analysis, described in Appendix C, was an estimate of
e equal to 76 g 15
g.5 This estimate is not significantly different,
statistically, from either the 97.4-g value assumed in the
February 1994 notice, the 80.7-g NBS estimate, or the 51.7-
g value estimated for samples collected between late 1992 and
mid-1995. Since the NMA data were obtained from samples collected by
Skyline Coal, Inc., prior to 1995, the Secretaries believe these data
confirm the 51.7-g value of e applicable
to the Skyline samples. The estimate of e obtained
from the Skyline data is, however, significantly greater than the value
estimated for weight-gain measurements under MSHA's current inspection
program. This is explained by the fact that when the Skyline samples
were collected, all samples were weighed in different laboratories
before and after sampling, and the weights were truncated to 0.1 mg.
before calculating the weight gain.
---------------------------------------------------------------------------
\5\ To construct a 90-percent confidence interval for
G, based on the Skyline data, the 15-
g ``standard error of the estimate'' must be
multiplied by a confidence coefficient of 1.64.
---------------------------------------------------------------------------
Truncation of weights, and also the practice of pre- and post-
weighing samples in different laboratories, were discontinued for
inspector samples in mid-1995. Under MSHA's revised procedures for
processing inspector samples, filter capsules are weighed both before
and after sampling in MSHA's laboratory. Furthermore, the results
recorded and used in calculating dust concentrations are expressed to
the nearest g. Therefore, the 5.8-g estimate of
e described above, applying to pre- and post-
exposure weighings in the same laboratory using current equipment and
no truncation, was used by NIOSH to calculate CVweight as
part of the NIOSH ``indirect'' evaluation of CVtotal, placed
into the public record on March 12, 1996.
Based on the results of MSHA's 1995 field study,
e was estimated to be 9.12 g [18]. In this
study, the filter capsules were used to collect respirable coal mine
dust samples in an underground mine between pre- and post-exposure
weighings in MSHA's laboratory, potentially subjecting them to unknown
sources of variability in weight gain not covered by the laboratory
estimates. Substituting the estimated value of e =
9.12 g into Equation 3 results in a corresponding estimate of
CVweight that declines as the sampled dust concentration
increases--ranging from 9.3 percent at dust concentrations of 0.2 mg/
m3 to less than one percent at concentrations greater than
2.0 mg/m3. This estimate of CVweight applies to
the procedure utilizing control filter capsules.
An updated estimate of e = 6.5 g was
also calculated using the published NBS procedure for filter capsules
processed with the current equipment and procedures for inspector
samples. This estimate, derived from weighing the same group of 55
unexposed filter capsules 139 times over a 218-day period, was
described in material placed into the public record on March 12, 1996
[19]. The 6.5 g estimate applies to filter capsules pre- and
post-weighed robotically on different days within MSHA's laboratory,
but it does not reflect any potential effects of removing the capsule
from the laboratory and exposing it in the field between weighings.
The estimate of imprecision in measured weight gain derived from
the MSHA's 1995 field study discussed earlier (9.1 g), falls
only slightly above the 6.5 g laboratory estimate. This
suggests that the process of handling and actually exposing the filter
capsule in a mine environment does not add appreciably to the
imprecision in measured weight gain.
[[Page 5682]]
In February 1997, 75 unexposed filter capsules that had been pre-
weighed in MSHA's laboratory and distributed to MSHA district offices
were recalled and reweighed [13]. After adjusting for variability
attributable to the date of initial weighing (i.e., variability that
would be eliminated by use of a control filter capsule), these data
provide an estimate of e equal to 8.2 g
[14]. This estimate, which is based on weighings separated by a span of
about four to five months, corroborates the 9.1 g estimate
obtained from MSHA's 1995 field study.
(c) Negative Weight-Gain Measurements
Some commenters pointed out that MSHA routinely voids samples when
the measured pre-exposure weight of a filter capsule is greater than
the measured post-exposure weight. According to these commenters, such
occurrences reflect an unacceptable degree of inaccuracy in weight-gain
measurements. One commenter asserted that such cases are ``of
particular significance when only one sample is relied upon.'' This
commenter attributed such occurrences solely to errors in the capsule
pre-weight and implied that they should not be expected to occur under
MSHA's quality assurance program. It was, therefore, implied that
negative weight-gain measurements are not consistent with the degree of
uncertainty being attributed to weighing error.
Prior to implementation of the 1995 processing modifications, a
significant fraction of samples with less than 0.1 mg of true weight
gain (i.e., G < 0.10="" mg)="" could="" be="" expected="" to="" exhibit="" negative="" weight="" gains="" (i.e.,="" g="">-0.1 mg). Contrary to the commenter's
implication, however, negative weight-gain measurements do not arise
exclusively from positive pre-exposure weighing errors (i.e.,
w1 > W1). They can also arise, with equal
likelihood, from negative post-exposure weighing errors (i.e.,
w2 <>2).
What is required for a negative weight gain (w2 <>1) is that e < -g.="" since="" the="" true="" weight="" gain="" (g)="" is="" always="" greater="" than="" or="" equal="" to="" zero,="" this="" means="" that="" a="" negative="" weight="" gain="" is="" observed="" when="" e="" is="" sufficiently="" negative.="" under="" standard="" assumptions="" of="" normally="" distributed="" errors,="">e fully accounts
for the probability of such occurrences. Naturally, this probability
becomes smaller as G increases and also as e
decreases.
The occasional negative weight-gain measurements that have been
observed are consistent with values of e estimated
for previous processing procedures. Table 2 contains the probability of
a negative weight-gain measurement for true weight gains (G) ranging
from 0.0 mg to 0.08 mg, assuming e = 51.7
g and the previous practice of truncation, which has now been
discontinued for inspector samples. Since the purpose here is to
evaluate the probability of negative weight gains under MSHA's previous
processing procedures, it is also assumed that no control filter
capsules are used to adjust weight gains.
Table 2.--Probability of Negative Weight-Gain Measurement, Assuming
Truncation and e=51.7 g
------------------------------------------------------------------------
Estimated probability of negative
True weight gain G=W2-W1 (mg) measurement, %
------------------------------------------------------------------------
0.00............................... 12.9
.01................................ 8.4
.02................................ 5.1
.03................................ 2.8
.04................................ 1.5
.05................................ 0.7
.06................................ .4
.07................................ .2
.08................................ .1
------------------------------------------------------------------------
Note: Tabled probabilities (in percent) were obtained from a simulation
of 35,000 weight-gain measurements at each value of G, assuming
normally distributed weighing errors and the now discontinued practice
of measurement truncation.
One commenter suggested the use of a test based on the frequency of
negative weight-gain measurements to check the magnitude of the MSHA/
NIOSH estimate of CVtotal. As proposed by the commenter, the
test of CVtotal would consist of comparing the observed
proportion of samples voided due to a negative recorded weight gain to
the proportion expected, given CVtotal equal to the MSHA/
NIOSH estimate. If the observed proportion were to exceed the expected
proportion, then this would constitute evidence that CVtotal
was being underestimated.
The commenter miscalculated the expected proportion, because he
mischaracterized the MSHA/NIOSH estimate of CVtotal as
constant over the continuum of dust concentrations. The MSHA/NIOSH
estimate of CVtotal increases as dust concentrations
decrease. This would cause a higher proportion of negative results than
what the commenter projected under the MSHA/NIOSH estimate, regardless
of what statistical distribution of dust concentrations is assumed.
The commenter's projection also neglected to take into account the
effects of truncating pre- and post-exposure weights to multiples of
0.1 mg. Although this practice has now been discontinued for MSHA
inspector samples, it is a factor in the available historical data.
In principle, if the statistical distribution of true dust
concentrations were known, the expected proportion of samples voided
for negative weight gain could be recalculated to reflect both a
variable CVtotal and, when applicable, truncation of
recorded weights. However, under the commenter's proposal, deriving the
expected proportion of negative measurements would involve not only
CVtotal, but also an estimate of the distribution of true
dust concentrations. Such an estimate would rely on the tenuous
assumption that a mixture of dust concentrations in different
environments is closely approximated by a lognormal distribution far
into the lower tail--i.e., even at concentrations extremely near zero.
Furthermore, valid estimation of the lognormal parameters, applicable
to dust concentrations near zero, would be complicated by measurement
errors, especially those resulting in negative or zero values.
Depending on the data used, truncation effects could also confound the
analysis.
Before truncation was discontinued, negative weight-gain
measurements were caused by various combinations of pre- and post-
exposure weighing and truncation error. Since truncation, and
especially interlaboratory variability, have now been removed as
sources of error in weight-gain measurements for inspector samples,
negative weight-gain measurements are expected to occur less frequently
than in the past.
(d) Comparing weight gains obtained from paired samples
Some commenters maintained that ``although there may be slight
differences between how the samples are dried * * *,'' differences
between the weight gain observed in MSHA samples and simultaneous
samples collected nearby (and processed at an independent laboratory)
indicated a greater degree of weighing uncertainty than what was being
assumed. In response to the Secretaries' request for any available data
supporting this position, results from paired dust samples were
provided by two coal companies.
In comparing measurements obtained from paired samples, there are
several important considerations that some commenters did not take into
account. First, if two different sampler units are exposed to identical
atmospheres for the same period of time, the difference between weight-
gain measurements g1 and g2 arises, in part, from
two independent weight-gain measurement errors, e1 and
e2. If uncertainty due to
[[Page 5683]]
each of these errors is represented by se, then the
difference between g1 and g2 has uncertainty due
to weighing error equal to se2. Consequently,
weight gains measured in the same laboratory, on the same day, for
different filter capsules exposed to identical atmospheres can be
expected to differ by an amount whose standard deviation is
1.41se.
Furthermore, if the two exposed capsules are processed at different
laboratories, the difference in weight gains contains an additional
error term arising from differences between laboratories. Evidence was
presented that this term ( in the notation of [12])
is far more significant than the intra-lab, intra-day weighing error in
MSHA's laboratory. Moreover, the additional uncertainty introduced by
use of a third laboratory also depends on unknown weighing imprecision
within that laboratory, which may differ from that maintained by MSHA's
measurement assurance process. (See Appendix C for analysis of paired
sample data submitted by NMA).
However, the most important consideration in comparing weight gains
from two different samples is that under real mining conditions, the
atmospheres sampled may not be identical--even if the sampler units are
located near one another. Differences in atmospheric dust
concentrations over relatively small distances have been documented
[20]. Such differences would be expected to produce corresponding
differences in weight gain that are unrelated to the accuracy of a
single, full-shift measurement as defined by the measurement objective
explained earlier in this notice.
II. Pump Variability
The component of uncertainty due to variability in the pump,
represented by CVpump, consists of potential errors
associated with calibration of the pump rotameter, variation in flow
rate during sampling, and (for those pumps with rotameters) variability
in the initial adjustment of flow rate when sampling is begun. The
Secretaries believe that CVpump adequately accounts for all
uncertainty identified by commenters as being associated with the
volume of air sampled.
In deriving the Values Table published in MSHA's February 1994
notice, MSHA used a value of 5 percent to represent uncertainty
associated with initial adjustment of flow rate at the beginning of the
shift and another value of 5 percent to represent flow rate
variability. The 5-percent value for variability in initial flow rate
adjustment was estimated from a laboratory experiment conducted by MSHA
in the early 1970s, while the value for flow rate variability was based
on the allowable flow rate tolerance specified in 30 CFR part 74. This
part requires that the flow rate of all sampling systems not vary by
more than 5 percent over a full shift with no more than two
adjustments. MSHA did not include a separate component of variability
for pump rotameter calibration because it was already included in the
5-percent value used to represent flow rate variability.
Based on a review of published results [10], the Secretaries
concluded that the component of uncertainty associated with the
combined effects of variability in flow rate during sampling and
potential errors in calibration is less than 3 percent. Therefore, as
proposed in the March 12, 1996 notice, the Secretaries are now
estimating uncertainty due to variability in flow rate to be 3 percent.
Because MSHA could not provide the experimental data supporting the
5-percent value used to represent uncertainty associated with the
initial adjustment of flow rate, one commenter recommended that MSHA
conduct a new experiment. In response to that request, MSHA conducted a
study to establish the variability associated with the initial flow
rate adjustment. The study, placed into the public record on September
9, 1994, attempted to emulate realistic operating conditions by
including a variety of sampling personnel making adjustments under
various conditions. Results showed the coefficient of variation
associated with the initial adjustment to be 3 0.5 percent
[11]. The Secretaries consider this study to provide the best available
estimate for uncertainty associated with the initial adjustment of a
sampler unit's flow rate. Therefore, as proposed in the March 12, 1996
notice, the Secretaries are now estimating uncertainty due to
variability in the initial adjustment to be 3 percent.
One commenter expressed concern regarding how representative MSHA's
study on initial flow rate adjustment was of actual sampling
conditions. The Secretaries consider the conditions under which the
study was conducted to have adequately mimicked conditions under which
the flow rate of a coal mine dust sampling system is adjusted. This was
more rigorous than the original study, from which MSHA estimated the 5-
percent value assumed in the February 12, 1994 notice. The tests were
conducted in an underground mine, using both experienced and
inexperienced persons to make the adjustments. Also, the only
illumination was supplied by cap lamps worn by the person making the
adjustments. Tests were conducted for adjustments made in three
different physical positions: standing, kneeling and prone. Inspection
personnel participating in the study provided guidance as to the
methods typically used by inspection personnel in adjusting pumps. In
fact, environmental conditions under which the test was conducted were
generally more severe than those normally encountered by inspection
personnel, since initial adjustment of the pumps normally occurs on the
surface just before the work shift begins.
The same commenter also questioned why only the variability
associated with initial adjustment of the flow rate was estimated and
not the variability associated with subsequent adjustments during the
shift. This is because the variability associated with the subsequent
flow rate adjustments of an approved sampler unit is already included
in the 3-percent value estimated for variability in flow rate over the
duration of the shift.
Since variability in the initial flow rate adjustment is
independent of calibration of the pump rotameter and variability in
flow rate during sampling, these two sources of uncertainty can be
combined through the standard propagation of errors formula:
[GRAPHIC] [TIFF OMITTED] TN31DE97.006
This estimate accords well with a more recent finding based on 186
measurements in an underground mine, using constant flow-control pumps
[18]. That study estimated CVpump = 4.0 percent and
concluded that CVpump was unlikely to exceed 4.4 percent.
Three commenters stated that there are reports of sampling pumps
being calibrated and used at altitudes differing by as much as 3000
feet and that, for many pumps, this could result in more than a 3-
percent change in flow rate per 1000 feet of altitude. MSHA recognized
this as a potential problem as early as 1975. As a result, MSHA
conducted a study to ascertain the effect of altitude on coal mine dust
sampler calibration [21]. The study showed that both pump performance
and rotameter calibration were affected by changes in altitude but that
an approved MSA sampling system, calibrated and adjusted at an altitude
of 800 feet to a flow rate of 2.0 L/min, would meet the requirement of
30 CFR 74.3(11) when sampling at an altitude of 10,000 feet, even if no
adjustment were made to the pump. The study also provided equations for
adjusting the calibration mark on the pump rotameter so that, when
sampling at an altitude
[[Page 5684]]
different from the one at which the rotameter was calibrated, the
appropriate flow rate would be obtained. These procedures are used by
MSHA inspectors in instances where the sampling altitude is
significantly different from the altitude where the sampling system is
calibrated.
Some commenters questioned the ability of the older MSA Model G
pumps to meet the same flow rate specifications as new pumps. MSHA has
discontinued the use of these older pumps in its sampling program and
will be using only flow-control pumps. More recent MSHA studies show
that these pumps continue to meet the flow rate requirement of 30 CFR
74.3(11) at altitudes up to 10,000 feet [22]. As a result, the flow-
control pumps currently used by inspectors can be calibrated at one
altitude and used at another altitude with no additional adjustments
made to the pumps. Furthermore, all sampler units used to measure
respirable dust concentrations in coal mine environments are required
to be approved in accordance with the regulatory requirements of 30 CFR
part 74, which require flow rate consistency to be within
0.1 L/min of the 2.0 L/min flow rate.6 MSHA's experience
over the past 20 years has demonstrated that flow rate consistency of
older sampling systems will continue to meet the requirements specified
in part 74, provided the systems are regularly calibrated and
maintained in approved condition. To ensure that sampling systems
continue to meet the specification of part 74, MSHA's policy requires
calibration and maintenance by specially trained personnel in
accordance with MSHA Informational Report No. 1121 (revised).
---------------------------------------------------------------------------
\6\ Section 74.3(13) requires that flow rate in an approved
sampler unit deviate from 2.0 L/min by no more than 5 percent over
an 8-hour period, with no more than 2 readjustments after the
initial setting. However, this is a maximum deviation, and the
uncertainty associated with pump flow rate, as quantified by its
coefficient of variation, is 3 percent.
---------------------------------------------------------------------------
III. Intersampler Variability
Intersampler variability, represented by CVsampler,
accounts for uncertainty due to physical variations from sampler to
sampler. Most of the commenters ignored this source of uncertainty. One
commenter, however, stated that 10-mm nylon cyclones are subject to
performance variations due to static charging phenomena (discussed in
Appendix A).
Intersampler variability was investigated by Bowman et al. [10],
Bartley et al. [17], and Kogut et al. [18]. Bowman et al. designed a
precision experiment to determine the contribution to
CVtotal from differences between individual coal mine dust
sampler units. Based on their experiment, they reported
CVsampler = 1.6 percent, which included variation in both
the 10-mm nylon cyclone and the MSA Model G pump. They concluded that
this low degree of component variability indicates there is excellent
uniformity in the mechanical components of dust sampler units. Bartley,
from his experimental investigation of eight 10-mm nylon cyclones,
estimated CVsampler to be no more than 5 percent for
aerosols with a size distribution typical of those found in coal mine
environments. Based on an analysis involving 32 different sampler
units, Kogut et al. found that CVsampler was unlikely to
exceed 3.1 percent. Unlike Bartley's study, however, this analysis
relied on new cyclones, which might be expected to exhibit less
variability than older, heavily used cyclones. Therefore, NIOSH used
the more conservative estimate of 5 percent, with an upper 95-percent
confidence limit of 9 percent, in its ``indirect approach'' for
estimating CVtotal and evaluating method accuracy [3].
Appendix C--Data Submitted by Commenters
During the public hearings, several commenters indicated they had
data showing that MSHA and NIOSH had underestimated the overall
magnitude of uncertainty associated with a single, full-shift
measurement. These data and accompanying analyses were submitted to the
record and evaluated by MSHA and NIOSH. Some of the data sets consisted
of paired samples, where two approved sampler units were placed nearby
one another and operated for a full shift. One of the resulting samples
was analyzed in MSHA's laboratory and the other by an independent
laboratory. These data were represented as showing that single, full-
shift measurements cannot accurately be used to estimate dust
concentrations. Other data sets submitted consisted of unpaired
measurements collected from miners at intervals over varying spans of
time. These data sets were represented as showing that exposures vary
widely between shifts and between occupations.
I. Paired Sample Data Submitted by the NMA
The American Mining Congress and National Coal Association [AMC and
NCA have since merged into the National Mining Association, (NMA)]
submitted at the request of MSHA and NIOSH a data set consisting of 381
pairs of exposure measurements. These measurements had been obtained
from the ``designated occupations'' on two longwall and six continuous
mining sections belonging to Skyline Coal, Inc. Two sampling units were
placed on each participating miner and operated for the full shift.
After sampling, one sample cassette was sent to MSHA for analysis while
the other was analyzed at a private laboratory. All samples were
reported to be ``portal to portal'' samples as required by MSHA
regulations. Using these data, the NMA estimated an overall CV of 16
percent. Based on this 16-percent estimate, the NMA suggested that MSHA
had underestimated measurement uncertainty in its February 1994 notice
by 60 percent at dust concentrations of 2.0 mg/m\3\.
The NMA estimate of 16 percent for overall CV includes not only
sampling and analytical error, but also variability arising from two
additional sources: (1) Spatial variability between the locations where
the two samples were collected; and (2) interlaboratory variability
introduced by the fact that a third laboratory was involved in weighing
exposed filter capsules.
Since the two dust samples within each pair submitted were not
collected at precisely the same location, differences observed between
paired samples in the Skyline data are partly due to spatial
variability. The Secretaries fully recognize and acknowledge that, as
suggested by the Skyline data, spatial variability in mine dust
concentrations can exist, even within a relatively small area such as
the so-called breathing zone of a miner. Consistent with general
industrial hygiene practice, however, the Secretaries do not consider
such variability relevant to the accuracy of an individual dust
concentration measurement.
The NMA expressed sampling and analytical error as a single
percentage relative to the average of all dust concentrations that
happened to be observed in the data analyzed. Contrary to the NMA
analysis, sampling and analytical error cannot be expressed as a
constant percentage of the true dust concentration. Because
e is constant with respect to dust concentration,
CVweight declines with increasing dust concentration, as
explained in Appendix B. The value of CVtotal assumed by
MSHA and NIOSH for the period when the Skyline samples were collected
is approximately 7.5 percent when the true dust concentration
() is 2.0 mg/m \3\ and approximately 16.2 percent when
= 0.5 mg/m \3\. This is based on applying Equations 2 and 3
to
[[Page 5685]]
e = 51.7 g, CVpump = 4.2
percent, and CVsampler = 5 percent.
Even if the effects of spatial variability and the third laboratory
are ignored, and the overall CV is interpreted as an average over the
range of concentrations encountered, the 16-percent value reported by
the NMA makes no allowance for the paired covariance structure of the
data. Therefore, MSHA and NIOSH consider the 16-percent value to be
erroneous, even under NMA's assumptions.
MSHA and NIOSH re-analyzed the Skyline data in order to check
whether these data were consistent with the value of
e (i.e., 51.7 g) estimated for the time
when the Skyline samples were collected. To distinguish the NMA
interpretation of sampling and analytical error (including spatial
variability) from the Secretaries' interpretation (excluding spatial
variability), SAE will denote sampling and analytical error according
to the Secretaries' interpretation, and SAE* will denote
sampling and analytical error according to the NMA interpretation. If
CVspatial denotes the component of SAE*
attributable to spatial variability for each measurement, it follows
that
SAE* = (CV \2\total +
CV\2\spatial)1/2.
To estimate SAE* as a function of dust concentration
from the data provided, a least-squares regression analysis was
performed on the square of the difference between natural logarithms of
dust concentrations x1 and x2 observed within
each pair. Let * denote the true mean dust
concentration, not only over the full shift sampled, but also over the
two locations sampled. The expected value (E{}) of each squared
difference forms the ordinate of the regression line at each value of
the abscissa (1/*)\2\:
E{(Ln(X1)-Ln(X2)) 2}
2(SAE*) 2
= 2(CV 2total+CV 2spatial)
= 2[CV 2pump+CV 2sampler+CV
2 weight+CV 2spatial]
= 2(CV2pump+CV 2sampler+CV
2spatial)+
2(1.438e/*)2
=a0+a1(1/*) 2
Since no control filter capsules were used in processing the
Skyline dust samples, CV weight does not, in this analysis,
contain the 2 factor shown in Equation 3 of Appendix B. The
intercept of the regression line is
a0=2(CV\2\pump+CV2+sampler+C
V 2 spatial), and the slope is
a1=2(1.438e) 2. To carry out
the regression analysis, * was approximated by
(x1+x2)/2. Regression estimates of the parameters
a0 and a1 were used to generate corresponding
estimates of e and CV 2
spatial.
The least squares estimate of e obtained from
this analysis is 76.0 g, with standard error of 15
g. This is not significantly different, statistically, from
the 51.7-g value estimated for the time period when the
Skyline samples were collected. Assuming CVpump=4.2 percent
and CVsampler=5 percent, the value of CVspatial
obtained from the least squares estimate of a0 is 19.7
percent, with standard error of 2.9 percent.
II. Paired Sample Data Submitted by Mountain Coal Company
Mountain Coal Company submitted a data set consisting of the
difference (expressed in mg/m 3) between paired samples
collected from miners over roughly a one-year period. Two sampler units
were placed on each participating miner (presumably one on each collar
or shoulder) and operated for roughly a full shift. One sample cassette
was sent to MSHA for analysis (post-weighing) while the other was
analyzed at a private laboratory.
Mountain Coal Company provided only the differences between
measurements within each pair and not the concentration measurements
themselves. Since CVtotal varies with dust concentration,
and the dust concentrations were not provided, it was impossible to
form a valid estimate of measurement variability from these data, or to
determine what part of the observed differences could be attributed to
weighing error and what part to spatial variability or variability
attributable to operation of the pump and physical differences between
sampler units.
III. Exposure Data Submitted by Jim Walter Resources, Inc.
Jim Walter Resources, Inc. submitted a data set consisting of
exposure measurements collected from all miners working on two longwall
sections. Measurements were collected from each miner on five
consecutive days. This procedure was repeated during five sampling
cycles over a two-year period. During each sample cycle the five
measurements for each miner were averaged and compared to the
respirable dust standard. According to Jim Walter Resources, Inc., the
sampling plan ``eliminates the effect of the variability of the
environment and minimizes the error due to the coefficient of variation
of the pump because all miners [original emphasis] are sampled for five
shifts,'' and these data ``show the variability of the sample pump and
of the worker's exposure to respirable dust.''
In its submission, Jim Walter Resources, Inc. apparently assumed
that the quantity being measured is average dust concentration across a
number of shifts, rather than average dust concentration averaged over
a single shift at the sampling location. The Secretaries agree that
dust concentrations do vary from shift to shift and from job to job, as
these data illustrate. This variability, however, is largely under the
control of the mine operator and should not be considered when
evaluating the accuracy of a single, full-shift measurement.
IV. Exposure Data Submitted by the NMA
The NMA submitted data consisting of recently collected and
historical measurements collected from the designated occupations
(continuous miner operator for continuous mining sections and either
the headgate or tailgate shearer operator for longwall mining sections)
for three continuous mining sections and five longwall mining sections.
According to the NMA analysis, there is a 17-percent probability that
these mines would be cited, even though the long-term average is less
than the respirable dust standard.
The NMA failed to recognize that the quantity being measured is
dust concentration averaged over a single shift at the sampling
location. The Secretaries agree that exposures do vary from shift to
shift, as these data illustrate. This variability, however, is largely
under the control of the mine operator and should not be considered
when evaluating the accuracy of a single, full-shift measurement.
V. Sequential Exposure Data Submitted by Jim Walter Resources, Inc.
Jim Walter Resources, Inc. submitted data collected from several
longwall faces. For each longwall, seven dust samples were collected,
using sampler units placed on the longwall face at least 48'' from the
tailgate at the MSHA 061 designated location. Pumps were successively
turned off in one hour increments, resulting in samples covering
progressively longer time periods over the course of the shift, from
one to eight hours. This was repeated on a number of days at each
longwall.
Many of the samples showed either the same or less weight gain than
the previous sample (collected over a shorter time period) within a
sequence. In the cover letter and written comments accompanying these
data, it was claimed that the weight gains observed for samples within
each sequence should progressively increase, irrespective of variations
in air flow and production levels, and that the patterns observed
exemplify ``the variability of sample results with today's equipment
and weighing techniques.''
[[Page 5686]]
MSHA and NIOSH have concluded that these data cannot be used to
estimate or otherwise evaluate measurement accuracy for the following
reasons: First, a highly sensitive and accurate sampling device would
be expected to produce variable results when exposed to even slightly
different environments. Since the samples within each sequence of seven
were not collected at exactly the same point, they are subject to
spatial variability in dust concentration. It is well known that dust
concentrations can vary even within small areas along a longwall face.
Therefore, variability in sample results is attributable not only
to measurement errors but also to variations in dust concentration due
to spatial variability.
Second, even on a production shift, variations in air flow and
production levels over the course of the shift can result in periods
within the shift during which the true dust concentration to which a
sampler is exposed is low or near zero. If a sampler unit is exposed to
a relatively low dust concentration during the final hour in which it
is exposed, any difference between that sample and the previous sample
will tend to be dominated by spatial variability. In such cases the
increase in weight accumulated during the final hour would be
statistically insignificant as compared to variability in dust
concentration at different locations. Without detailed knowledge of the
airflow and production levels as they varied over each shift, it is
impossible to determine how many cases of this type would be expected.
However, approximately one-half of such samples would be expected to
exhibit less weight gain than the previous sample.
Further, because sample weights were truncated to 0.1 mg at the
time these data were collected, and because expected weight gains of
less then 0.1 mg are not uncommon over a one-hour period, there would
be no apparent increase in recorded weight gain in many cases where the
two sample results actually differed by a positive amount. Therefore,
some unknown number of cases showing no difference in successive weight
gains are attributable to truncation effects. Truncation has now been
discontinued for samples collected under MSHA's inspection program.
Finally, as has been shown in Appendix B, a certain percentage of
negative weight-gain measurements at low dust concentrations is
consistent with the weighing imprecision experienced at the time these
samples were collected. However, since these data were not collected in
a controlled environment, it is impossible to determine what that
percentage should be. Because the weight gain for each sample is
determined as the difference between two weighings, comparison of
weight gains between two samples involves a total of four independent
weighing errors. Therefore, variability attributable purely to weighing
error in the difference between weight gains in two successive samples
is greater (by a factor equal to 2) than variability due to
weighing error in a single sample. Furthermore samples collected over
less than a full shift are subject to more variability due to random
fluctuations in pump air flow and cyclone performance than samples
collected over a full shift. Both of these considerations increase the
likelihood that a sample will exhibit less weight gain than its
predecessor, as compared to the likelihood of recording a negative
weight gain for a single, full-shift sample.
References
1. Kennedy, E.R., T.J. Fischbach, R. Song, P.M. Eller, and S.A.
Shulman. Guidelines for Air Sampling and Analytical Method
Development and Evaluation. U.S. Department of Health and Human
Services, Public Health Service, National Institute for Occupational
Safety and Health, DHHS (NIOSH) Publication No. 95-117.
2. European Standard No. EN 482: Workplace atmospheres--General
requirements for the performance of procedures for the measurement
of chemical agents. European Committee for Standardization (CEN),
1994.
3. Wagner, G.R. Letter of October 13, 1995, from Gregory R.
Wagner, M.D., National Institute for Occupational Safety and Health,
to Ronald J. Schell, Chief, Division of Health, Coal Mine Safety and
Health, Mine Safety and Health Administration.
4. Gray, D.C. and M.I. Tillery. Cyclone vibration effects. Am
Ind Hyg Assoc J, 42(9):685-688, 1981.
5. Lippmann, M. and R.E. Albert. The Effect of Particle Size on
the Regional Deposition of Inhaled Aerosols in the Human Respiratory
Tract. Am Ind Hyg Assoc J, 30:257-275, 1969.
6. Goddard, B., K. Bower, and D. Mitchell. Control of Harmful
Dust in Coal Mines. National Coal Board, 6-12, 1973.
7. Tomb, T.F. Memorandum of August 31, 1981, from Thomas F.
Tomb, Chief, Dust Division, Pittsburgh Health Technology Center,
MSHA, to William Sutherland, Chief, Division of Health, Coal Mine
Safety and Health, MSHA, Subject: Evaluation of Criterion Used to
Select Respirable Coal Mine Dust Samples for Examination for
Oversize Particles.
8. Treaftis, H.N. and T.F. Tomb. Effect of Orientation on
Cyclone Penetration Characteristics. Am Ind Hyg Assoc J, 35(10):598-
602, 1974.
9. Tomb, T.F., H.N. Treaftis, R.L. Mundell, and P.S. Parobeck.
Comparison of Respirable Dust Concentrations Measured With MRE and
Modified Personal Gravimetric Sampling Equipment. BuMines RI 7772,
1973.
10. Bowman, J.D., G.M. Breuer, S.A. Shulman, and D.L. Bartley.
Precision of Coal Mine Dust Sampling. U.S. Department of Health and
Human Services, Public Health Service, Centers for Disease Control,
National Institute for Occupational Safety and Health, NTIS No. PB-
85-220-721, 1984.
11. Tomb, T.F. Memorandum of September 1, 1994, from Thomas F.
Tomb, Chief, Dust Division, Pittsburgh Safety and Health Technology
Center, MSHA, to Ronald J. Schell, Chief, Division of Health, Coal
Mine Safety and Health, MSHA, Subject: Determination of the
Precision of Setting the Rotameter Ball to a Calibration Mark on
Personal Respirable Dust Sampling Pumps.
12. Kogut, J. Letter of May 12, 1994, from Jon Kogut, MSHA, to
David Bartley, Division of Physical Sciences and Engineering, NIOSH.
13. Parobeck, P., J. Kogut, T. Tomb, and L. Raymond.
Investigation of Weighing Variability Between MSHA and MSA
Laboratories. Internal MSHA Report 1997.
14. Wagner, G.R. Letter of May 28, 1997, from Gregory R. Wagner,
M.D., Acting Associate Director for Mining, National Institute for
Occupational Safety and Health, to Ronald J. Schell, Chief, Division
of Health, Coal Mine Safety and Health, MSHA.
15. Kogut, J. Memorandum of September 6, 1994, from Jon Kogut,
Mathematical Statistician, Denver Safety and Health Technology
Center, MSHA, to Ronald J. Schell, Chief, Division of Health, Coal
Mine Safety and Health, MSHA, Subject: Coal Mine Respirable Dust
Standard Noncompliance Determinations.
16. Parobeck, P., T. Tomb, H. Ku, and J. Cameron. Measurement
Assurance Program for Weighings of Respirable Coal Mine Dust
Samples. J Qual Tech, 13(3):157-165, 1981.
17. Barley, D.L. Letter of September 7, 1994, from David L.
Bartley, Research Physicist, Division of Physical Sciences and
Engineering, NIOSH, to Ronald J. Schell, Chief, Division of Health,
Coal Mine Safety and Health, MSHA.
18. Kogut, J., T.F. Tomb, P.S. Parobeck, A.J. Gero, and K.L.
Suppers. Measurement Precision With the Coal Mine Dust Personal
Sampler. Internal MSHA Report, 1995.
19. Tomb, T.F. Memorandum of February 16, 1996, from Thomas F.
Tomb, Chief, Dust Division, Pittsburgh Safety and Health Technology
Center, MSHA, to Ronald J. Schell, Chief, Division of Health, Coal
Mine Safety and Health, MSHA, Subject: Investigation to Determine
the Precision of MSHA's Automatic Weighing System for Weighing
Respirable Coal Mine Dust Samples.
20. Kissell, F.N. and R.A. Jankowski. Fixed-Point and Personal
Sampling of Respirable Dust for Coal Mine Face Workers. Paper in
Proceedings of the 6th US Mine Ventilation Symposium. Society of
Mining, Metallurgy, and Exploration, Inc (SME), Littleton, CO, 281-
186, 1993.
21. Treaftis, H.N., T.F. Tomb, and H.F. Carden. Effect of
altitude on personal respirable dust sampler calibration. Am Ind Hyg
Assoc J, 37(3):133-138, 1976.
22. Gero, A.J., P.S. Parobeck, K.L. Suppers, B.P. Apel, and J.D.
Jolson. The Effect of
[[Page 5687]]
Altitude, Sample Port Inlet Loading, and Temperature on the
Volumetric Flow Rate of the MSA Escort Elf Constant Flow
Rate Pump. Pres. at Second International Conference on the Health of
Miners, Pittsburgh, PA, November 11-13, 1995.
Dated: December 19, 1997.
J. Davitt McAteer,
Assistant Secretary for Mine Safety and Health.
Dated: December 19, 1997.
Linda A. Rosentock,
Director, National Institute for Occupational Safety and Health.
Note: For the convenience of the user, notice document 97-33934
is being reprinted in its entirety because of numerous errors in the
document originally appearing at 62 FR 68372-68395, December 31,
1997. Those wishing to see a listing of corrections, please call
Patricia Silvey, Mine Safety and Health Administration, 703-235-
1910.
[FR Doc. 97-33934 Filed 12-30-97; 8:45 am]
BILLING CODE 4160-18-P 4510-43-P
