EPA-450/3-85-023
         ADDITION OF METHODS 3A, 6C, AND 7E
          TO APPENDIX A OF 40 CFR PART 60,
AND REVISIONS TO SUBPARTS D AND DA OF 40 CFR PART 60

     (Proposed February 28, 1985, 50 FR 08290)
         SUMMARY OF COMMENTS AND RESPONSES
            Emission Measurement Branch
    Emission Standards and Engineering Division
        U.S. Environmental  Protection Agency
    Research Triangle Park, North Carolina 27711
                 February 1986

-------
                             TABLE OF CONTENTS

                                                                       Page

Chapter 1.  Introduction 	    2


Chapter 2.  Summary of Changes Since Proposal. . .	    3


Chapter 3.  Summary of Comments and Responses	    8

                                                         i
     Table 1.  List of Acronyms Used in Summary of Comments

               and Responses	    1


     Table 2.  List of Commenters	   46

-------
                      Table 1.  LIST OF ACRONYMS USED

BAAPCD - Bay Area Air Pollution Control District
CARB   - California Air Resources Board
CD     - Calibration drift
CEM    - Continuous emission monitor
CEMS   - Continuous emission monitoring system
EPA    - Environmental Protection Agency
NBS    - National Bureau of'Standards
NDIR   - Nondispersive infrared
NSPS   - New Source Performance Standards
PS     - Performance specification (e.g., Performance Specification 2 or PS 2)
QA     - Quality assurance
QC     - Quality control
RA     - Relative accuracy
RM     - Reference method
SRM    - Standard reference material
TCEMS  - Transportable continuous emission monitoring system
UV     - Ultraviolet

-------
                                 Chapter 1
                                INTRODUCTION

     On February 28, 1985, the U.S. Environmental Protection Agency (EPA)
published in the federal jtegi_st_e_r_ (50 FR 8290) "Addition of Methods 3A, 6C.
and 7E to Appendix A of 40 CFR Part 60, and Revisions to Subparts D and Da
of 40 CFR Part 60."  These methods and subpart revisions were proposed
under the authority of Sections 101, 111, 114, 116, and 301 of the Clean
Air Act, as amended (42 U.S.C. 7401, 7411, 7414, 7416, and 7601).
     Public comments were solicited at the time of proposal.  To provide
interested persons the opportunity for oral presentation of data, views, or
arguments concerning the proposed methods, a public hearing was to be held,
if requested, on April 5, 1985, beginning at 10:00 a.m.  The hearing was
not held because no one requested to speak.  The public comment period
began February 28, 1985, and was originally scheduled to end May 6, 1985,
but was extended to June 6, 1985, at the request of one commenter to allow
additional time to develop comments.
     Twelve comment letters on the proposed methods were received from
industry, trade associations, a State air pollution control agency, and an
engineering firm.  The comments that were submitted along with EPA's
responses are summarized in this document.  The summary of comments and
responses serves as a basis for the revisions that have been made to the
methods and to the subpart revisions between proposal and promulgation.

-------
                                 Chapter 2
                     SUMMARY OF CHANGES SINCE PROPOSAL
Method 3A
     1.  Section 2.  Wording has been added to specify that the monitoring
system span is to be established so that the average Og or C02
concentration is not less than 20 percent of span.
     2.  Section 5.1.2.  It is now specified that a heated sample line is
not required for systems that measure the 02 or C02 concentration on a dry
basis.
     3.  Section 5.1.3.  It is now specified that the use of stainless
steel, Teflon, or nonreactive glass filters is not required.
     4.  Section 5.1.4.  It is now specified that the requirements for
measuring and controlling the analyzer flow rate are not applicable if data
are presented that demonstrate the analyzer is insensitive to the flow rate
variations over the range encountered during the test.
     5.  In Section 5.2, the number of calibration gases required has been
reduced from four to three.
     6.  In Section 9, an equation has been added to calculate the effluent
gas concentration for 02 analyzers that use a low-level calibration gas
in place of a zero gas.
Method 6C
     1.  Sections 1.2 and 5.1.3.  The specific types of S02 analyzers that
may be acceptable have been identified as ultraviolet (UV), nondispersive
infrared (NDIR), and fluorescence.
     2.  Section 2.1.  Wording has been added to specify that the monitoring
system span be established so that the average S02 concentration equivalent
to the emission standard is not less than 30 percent of span.

-------
     3.  Section 2.2.  The data recorder resolution requirement
specification has been moved to a new Section 5.1.11.
     4.  Sections 4.2 and 6.5.2.  The sampling system bias specification
has been relaxed to +5 percent.
     5.  Section 4.3.  The zero drift .specification has been relaxed to
_+3 percent.
     6.  Section 4.4.  The calibration drift specifications has been
relaxed to ^3 percent.
     7.  Section 5.1.  The requirement that the sample line be heated
has been clarified by separately identifying the portion of sample line
downstream of the moisture removal system in a new Section 5.1.3, and in
Figure 6C-1.
     8.  Section 5.1.4 (previously numbered 5.1.3).  A note has been added
to describe the circumstances whereby the determination of sample moisture
content is not necessary.
     9.  Section 5.1.5 (previously numbered 5.1.4).  The section has been
revised to clarify the filter heating requirements and the acceptable types
of filter material.
    10.  Sections 5.1.6 and 5.1.8.  These sections have been renumbered from
5.1.5 and 5.1.7, respectively.
    11.  Section 5.1.7 (previously numbered 5.1.6).  A note has been added
to assist in the prevention of overpressurization of the analyzer.
    12.  Section 5.1.9 (previously numbered 5.1.8).  A note has been added
to assist in the prevention of analyzer drift.
    13.  Section 5.1.10 (previously numbered 5.1.9).  A specification has
been added to allow for manual measurements in place of a chart recorder,
and the minimum frequency and number for such measurements are given.
                                     4

-------
    14.  Section 5.3.  A specification for the Q£ concentration of
calibration gases for fluorescence analyzers has been added.
    15.  Section 5.3.  Provision has been made for the use of S02/C02
in N2, S02/02 in N2, or S02/C02/02 in N2 gas mixtures.
    16.  Section 5.3.  Section 5.3.3 has been deleted, as the number of
required calibration gases has been reduced from four to three.
    17.  Section 6.1.  A note to recommend the use of calibration gases
prepared according to Protocol 1 has been added.
    18.  Section 6.4.  Section 6.5 has been renumbered 6.4.  The proposed
Section 6.4 has been deleted, as has Figure 6C-5.  The response time
determination has been simplified by adding a specification in
Section 6.4.1 that the tester determine the response time by observing the
time required to achieve a stable response when zero and upscale gases are
introduced to the system.  The bias check specifications have been changed
to allow either the mid-range or high-range calibration gas for the
upscale sampling system bias check.
    19.  Section 7.2.  Provision for omission of the interference check
has been replaced by a mandatory one-time check for each individual
analyzer.
    20.  Section 7.4.  Language has been added to clarify the method for
determining drift.
    21.  Section 8.  Corrections have been made to Equation 6C-1.
    22.  Figure 6C-2.  The impingers are now illustrated as immersed in
an ice bath.

-------
Method 7E
     1.  Sections 5.1.3 and 6.4.  Clarification has been  added  that  an  N02
to NO converter is required, unless data are presented to demonstrate that
the N02 concentration is less than 5 percent of the NOX concentration.
     2.  Section 5.2.  The number of calibration gases required has  been
reduced from four to three.
Subpart D
     1.  Section 60.45(c)(l).  A sentence has been added  to specify  that
Methods 6C and 7E are to be used only at the sole discretion of the  source
owner or operator.
     2.  Section 60.46(a)(2).  A sentence has been added  to specify  that
Method 3A is to be used only at the sole discretion of the source owner or
operator.
     3*  Section 60.46(a)(4).  A sentence has been added  to specify  that
Method 6C is to be used only at the sole discretion of the source owner or
operator.
     4.  Section 60.46(a)(5).  A sentence has been added  to specify  that
Method 7E is to be used only at the sole discretion of the source owner or
operator.
     5.  Section 60.46(f)(3).  A sentence has been added  to specify  that
Method 3A is to be used only at the sole discretion of the source owner or
operator.              j  ,
                       !  I
Subpart Da             !
     1.  Section 60.47a(h)(l).  A sentence has been added to specify that
Methods 3A, 6C, and 7E are to be used only at the sole discretion of the
source owner or operator.

-------
     2.  Section 60.47a(h)(5)(i)(1).  A sentence has been  added to specify
that Methods 6C and 7E are to be used only at the sole discretion of the
source owner or operator.
     3.  Section 60.48a(a)(l).  A sentence has been added  to specify that
Method 3A is to be used only at  the sole discretion of the source owner or
operator.

-------
                                 Chapter 3
                     SUMMARY OF COMMENTS AND RESPONSES
     Comments were addressed in chronological order as they were received.
Where a similar comment was subsequently made by another commenter, a
statement to that effect was added.

Commenter IV-D-1
     1-1  Comment:  The quality control (QC) procedure of Method 3A should
not be based on a comparison to an Orsat analysis.  Either rely on the
calibration procedure in the method, or have audit samples analyzed and
document the results.
     A similar comment was made by Commenter IV-D-11.
     Response:  The QC procedures of the proposed Method 3 are recommended
but are not required.  Two options are presented.  The first involves the
determination of the F0 value and is applicable when both 03 and C02 are
measured using Method 3A.  The second relies on comparison of results
provided by Method 3A with measurements obtained with either an Orsat or
Fyrite analyzer.  Although not addressed within the method, the tester
could choose to analyze additional calibration gases as a QC procedure.
     1-2  Comment:  The interference check required in Method 6C should not
consist of a parallel Method 6 run, as this requirement makes the use of
Method 6C superfluous.  An interference check should be made in the
laboratory before going into the field, according to a procedure similar to
Section 5.4 of Method 20.
     Similar comments were made by Commenters IV-D-4, IV-D-5, IV-D-10, and
IV-D-12.

-------
     Response:  It was our intention that the tester continue the
interference check procedure only until  he had obtained a body of  data  that
would demonstrate that for his particular instrument and source category
application, no low bias in the test data could occur.   The tester could
then apply for exemption from further interference checks, as previously
described in the method, for those particular situations.  The difficulty
with this approach (not specifying particular instruments, but relying
instead on individual testers to gather their own performance data),  is
that it is difficult to establish the quantity of data  and application
description that would be sufficient.  On the other hand, a laboratory
interference check for all types of SOg analyzers is impractical because  of
the number and types of potential analytical interferences that may occur.
As a compromise, Method 6C has been revised to require  the use of  specific
analytical techniques, all of which have been shown to  be accurate on at
least one source category, and the interference check requirement  has been
reduced to the first field test for each analyzer as it is applied to each
source category.
     1-3  Comment:  Section 5.3 of Method 6C is too general.  While S02 in
air and S02 in N2 are acceptable calibration blends for NDIR and UV
analyzers, they wil] not be satisfactory for pulsed fluorescence analyzers
because the fluorescence efficiency is dependent upon 02 concentration.
It is therefore suggested that Section 5.3 be revised to include proper
calibration gases for pulsed fluorescence analyzers.
     A similar comment was made by Commenter IV-D-12.
     Response:  We agree.  Section 5.3 of Method 6C that deals with
calibration gases has been expanded for fluorescence analyzers to  require
that (1) the 02 and C02 concentrations of the calibration gases, as

-------
introduced to the analyzer, be within 1 percent (absolute) of their
respective concentrations of the effluent gas, or (2) that the calibration
gases consist of S02 in air, and that the suppressed response of the
effluent due to Og be corrected with nomographs supplied by the instrument
manufacturer.
Cbmtnenter IV-D-2
     2-1  Comment:  The commenter fully supports the use of the proposed
test methods.
     Response:  No response is required.
Commenter IV-D-3
     3-1  Comment:  Method 3A is set up to be a continuous sampling/
analyzing method, which far exceeds Method 3 in equivalency.  While we
support an instrumental equivalent to the Orsat, we never envisioned a
continuous analyzer/recorder methodology.  We merely wish to substitute
a state-of-the-art analyzer for the Orsat apparatus, leaving the remainder
of the method largely unchanged, except for the addition of necessary
calibration and quality assurance (QA) requirements.
     A similar comment was made by Commenter IV-D-5.
     Response:  Method 3A is meant to be used with Methods 6C and 7E, all
of which are continuous real-time measurement methods.  If the commenter
wishes to use an instrumental  Og analyzer in place of an Orsat for the
analysis of grab or integrated bag samples collected according to Method 3,
he should submit that procedure to EPA for consideration as a separate
alternative procedure.
     3-2  Comment: Section 2.1 of Method 6C states that the span of the
monitoring system is to be selected so that the mean gas concentration is
                                    10

-------
between 20 and 90 percent of the span.  However,  if the  mean  concentration
approaches the 20 percent level, then three of the four  required
calibration gases (as per Section 5.3) could be substantially above  the
emission level.  Three calibration gases are suggested in  place of four,
as follows: zero calibration, mid-range calibration, and high-range
calibration.  It is further suggested that if the instrument  calibration
error does not exceed 2 percent of span, then the instrument  should  be
considered calibrated over the entire measurement range.  The comment
also applies to Methods 3A and 7E.
     A similar comment was made by Commenter IV-D-5.
     Response:  We agree with the comment.  The requirements  of
Methods 3A, 6C, and 7E have been revised to require the  use of zero,
mid-range, and high-range calibration gases.  As  suggested, the calibration
error specification remains +2 percent of span.
     3-3  Comment:  Section 2.2 of Method 6C requires a  measurement  system
resolution capability of jH).5 percent of span, but this  proposed
requirement does not seem reasonable in view of the calibration error
tolerance of £2 percent of span.  The sensitivity of the measurement
system should be ^2 percent of span to remain consistent with performance
specifications (PS's) described in Appendix B of  Part 60.   The same  comment
also applies to Method 3A and 7E.
     A similar comment was made by Commenter IV-D-5.
     Response:  Our intention was to require that the recorder be capable
of being read to within 0.5 percent of span or 1/2 of a  scale division
assuming 100 divisions, or 1/4 of a scale division assuming 50 divisions.
Including this requirement in Section 2.2 of Method 6C has caused confusion,
                                     11

-------
As a clarification, the requirement has been deleted from Section 2.2 and
included instead as a new Section 5.1.11.
     3-4  Comment:  The interference check proposed for Method 6C requires
the use of a modified Method 6 sampling train and analysis, thereby
negating many potential advantages of Method 6C.
     A similar comment was made by Commenter IV-D-5.
     Response:  The interference check has been revised from once per test
to the first field test for each analyzer for each source category.  Refer
also to Response 1.2.
     3-5  Comment:  The zero and CD PS's proposed for Methods 3A, 6C, and
7E should be verified both preceding and following each run.
     Response:  We intended that the sampling system bias checks be
performed before and after each sampling run in order to quantify both
the sampling system bias and zero and calibration drifts.  The wording of
Section 7.4 has been changed to clarify this requirement.
     3-6  Comment:  The description of the sample line required for
Methods 6C and 7E should be revised because a heated sample line is not
necessary except for instruments that analyze on a wet basis.
     Similar comments were made by Commenters IV-D-5 and IV-D-12.
     Response:  It was our intention that Method 6C require only that
portion of the sample line that transports the sample from the probe to
the moisture removal  system to be heated.  Method 6C has been clarified
by identifying an additional  sample line component in both Figure 6C-1
and Section 5.1.
     3-7  Comment:  In Section 5.1.4 of Method 6C, the letter ^ in
parentheses should be a b.
                                    12

-------
     A similar comment was made by Commenter IV-D-5.
     Response:  We agree.  The error has been corrected.
     3-8  Comment:  The heating requirement for out-of-stack  filters  as
proposed in Section 5.1.5 of Method 6C should be quantified.
     A similar comment was made by Commenter IV-D-5.
     Response:  Method 6C has been revised to clarify that the
out-of-stack filter be heated sufficiently to prevent water condensation
on the filter.
     3-9  Comment:  A potential alternative to the measurement system
performance test procedures described in Methods 3A,  6C,  and  7E would be
to allow the instruments' manufacturers to certify their  performance.
     A similar comment was made by Commenter IV-D-5.
     Response:  It is unclear how the manufacturer would  develop the
necessary data to demonstrate that a specific analyzer is completely
free of analytical interferences, or how we could verify  that such  a
demonstration was accomplished.  We have assumed that the manufacturers
have designed analyzers to be as interference free as possible, and thus
a field check of each analyzer on a particular source category is the
most logical QA procedure.
                                          i
     3-10  Comment:  Omission of the interference check preparation
described in Section 7.2 of Method 6C needs further clarification.   Is it
necessary for separate approvals by the Administrator prior to each test?
     Response:  Provision for possible omission of the interference
check has been dropped in view of other changes in the method.
                                    13

-------
     3-11  Comment:  No real incentive exists for testers to use
Methods 3A, 6C, and 7E as proposed due to their over-emphasis on QA
requirements.  The methods as proposed far exceed the point of equivalency
when compared to the existing RM's.
     A similar comment was made by Commenter IV-D-5.
     Response:  Revisions to the methods made since proposal will
significantly reduce the cost of performing tests using Methods 3A, 6C,
and 7E.  These revisions include (1) requiring the Method 6C interference
check for only the first use of each analyzer on a particular source
category, (2) reducing the number of calibration gases required for the
analyzer calibration error test, (3) eliminating separate calibration
gas injections used to quantify the measurement system response time,
(4) relaxing the sampling system bias check specification from j^3
to jn5 percent of span, and (5) relaxing the zero and calibration drift
specifications from +2 to *3 percent of span.
CommehteK1V-D-4
     4-1  Comment:  If the proposed methods are promulgated as proposed,
EPA will have made continuous monitoring so expensive that it will  be more
cost effective to use the wet chemical methods (Methods 3, 6, and 7).  This
will devastate the commenter's business due to his large investment in
continuous monitors.  It is felt, however, that certain changes to correct
this situation can be made to the proposed methods that will not impair
their accuracy and precision.
     Response:  Revisions made to the methods since proposal, as described
in Response 3-11, will  significantly reduce their applied cost.
                                    14

-------
     4-2  Comment:  The EPA claims to have reviewed instrumental  test
methods currently in use, yet they are not included in the method
bibliographies.
     A similar comment was made by commenter IV-D-12.
     Response:  We reviewed a number of available instrumental  methods
during the development of the proposed methods, including Bay Area Air
Pollution Control District (BAAPCD) Source Test Methods ST-13A and ST-19A,
and "Transportable Continuous Emission Monitoring System (TCEMS)
Operational Protocol", EPA 340/1-83-016.  However, only those documents
that contain additional useful information are listed in the bibliography.
     4-3  Comment:  It is very inconvenient to have one method reference
another when certain specifications or procedures are repeated, and we
would prefer to see each method as a complete document.
     A similar comment was made by commenter IV-D-12.
     Response:  The practice of having one method reference another is
due to the need to reduce Federal Register and Code of Federal  Regulations
printing costs.  The commenter may wish to prepare an unabridged  copy of
the method(s) for his own use by the "cut and paste" technique if cross
referencing continues to be a problem for him.
     4-4  Comment:  The proposed requirement in Methods 3A, 6C, and 7E
that the instrument span be selected so that the mean gas concentration
of each run is between 20 and 90 percent of span may be misinterpreted
by some of the less knowledgeable regulatory types to say that if the
mean gas concentration is less than 20 percent or more than 90 percent
of span (but <100 percent of span), then the data will be considered
invalid.  As an example, certain steam generators for enhanced oil recovery
                                    15

-------
have NOX limits of 225 ppm at 3 percent 03.  It is not unusual  to measure
230 ppm at some lower Og level during a compliance test.  This  level  is
above 90 percent of span (250 ppm range on TECO Model 10A).  The 20 to
90 percent rule is meant to apply to instruments with a nonlinear response,
but really is not a factor for a linear response instrument such as a
chemiluminescence analyzer.   Is there some way to reword this requirement
so that someone does not reject data less than 20 percent or more than
90 percent of span?
     Response:  It was originally our intention to reject data  when the
mean sample value was greater than 90 percent or less than 20 percent of
the span to ensure that an appropriate measurement range was used.  We
have been convinced by the comments received that sample means  above 90
percent of the span are acceptable, provided that no measurements exceed
the measurement range.  Our concern is that without such a limit, a tester
may increase the span excessively in order to relax the analyzer
calibration error and sampling system bias specifications, which are both
expressed as a percent of span.  If an excessive span is used,  the
accuracy of the measurement results may be significantly reduced.
Therefore, Methods 6C and 7E have been changed to require that  the
measurement range (i.e., measurement range displayed by the data recorder)
be selected so that the pollutant concentration equivalent to the emission
standard is not less than 30 percent of the span.  Method 3A has been
changed to require that the measurement range be selected so that the
average Q£ or C02 concentration is not less than 20 percent of  the span.
     4.5  Comment:  Is the proposed instrument sensitivity capability of
+0.5 percent of span as proposed in Section 2.2 of Method 6C really
                                     16

-------
justified or has EPA assumed you can measure 1/2  of  the  smallest division
on a chart?
     Response:  The 0.5 percent of span resolution specification of
Method 6C has been clarified.  (See Response 3-3).   The  0.5 percent of span
resolution is justified in order to determine if  the measurement system
meets the +2 percent analyzer calibration error specification.  We have
assumed that a tester can read a properly maintained 4-inch chart recorder
to 1/4 of a scale division for 50 divisions.
     4-6  Comment:  In reference to Section 3.7 of Method  6C, we see  no
reason to use a mid-range gas to determine CD. The  CD can be determined
using the 80-percent span gas used to calibrate the  analyzer before and
after a run.  Using a mid-range gas does not provide improved CD data, but
does increase the cost of using continuous monitors  because an  additional
tank of calibration gas must be purchased.
     Response:  We essentially agree with this comment.  The requirements
of the sampling system bias check have been changed  to require  the tester
to use zero gas and either the mid-range or the high-range gas  (whichever
most closely approximates the effluent concentration; see  Response 4-20).
The tester must also use the same calibration gases  for  the drift tests.
Therefore, the revised methods require that the tester use the  upscale
calibration gas that is closest to the effluent concentration for the
CD determinations.  We believe that these requirements will result in the
most accurate effluent measurement results.
     4-7  Comment:  In reference to Section 3.9 of Method  6C, we are
opposed to any sort of interference check on a routine basis.   What is the
point of using a continuous monitoring method for SC>2 which must be checked
                                     17

-------
each time with Method 6?  Why not just run Method 6 and forget about the
continuous monitor?  It is not cost effective to perform an interference
check in addition to performing the continuous monitoring.  If the
state-of-the-art of S02 monitors is such that they are subject to
interferences and must be checked routinely, then S02 continuous monitor
methods should not be implemented.  Most interferences with an SOg analyzer
(e.g., water absorption at that wavelength) will result in positive
interferences which should be acceptable to an agency anyway.  The local
agencies in California actually specify the measurement principle which is
acceptable for a given pollutant to avoid using inadequate instrumentation.
The EPA should take this to heart and eliminate these all  encompassing
requirements from the method.
     Response:  The intent of the interference check was not to require it
on a routine basis, but rather to require sufficient data for a particular
analyzer and application to show that there would be no interference.
(Refer also to Response 1-2).  The revisions made to the methods since
proposal, including relaxation of the interference check to once per
analyzer rather than once per test should satisfy this comment.
     4-8  Comment:  In reference to Methods 3A, 6C, and 7E, the
requirement for using three span gases is one of the major reasons this
proposed method will actually increase the cost of testing.  A single EPA
Protocol 1 gas (approximately 80 percent of span) is all  that is required
to determine analyzer calibration error.  The EPA could have adopted the
rules currently in use in California which require a multipoint (e.g., low,
medium, high) linearity check on a quarterly basis (this  is too stringent
as well).   These quarterly multipoint linearity checks should also be done
                                     18

-------
with certified calibration gas, not EPA Protocol  1  gas,  to  reduce  the  cost
of this questionable check.  All  our instruments  are linear by  design  and
have never failed this quarterly check.  The only reason for requiring a
zero and three-span gas check is because the instrument  response is  not
linear.  This check should not be required for linear response  instruments.
An EPA Protocol 1 gas costs $289 (not including tax and  shipping)  and
requires 6 weeks to deliver and is valid for 6 months.  Our calibration gas
bill for 1984 (using one Protocol 1 span gas for  the analyzer calibration
error check and three certified gases for the quarterly  linearity  check)
was more than $25,000.  If we will be required to use an additional  two
Protocol 1 gases, the bill for calibration gases  will  increase  to
approximately $75,000 and will make the use of continuous monitors for
compliance prohibitively expensive.  This does not  even  take into  account
the extra gas regulators, plumbing and valving necessary for the additional
gases.
     The EPA has gone to all the trouble of developing the  National  Bureau
of Standards (NBS) traceable Protocol 1 standard.  This  gas is
unquestionably accurate and any analyzer calibrated with such a gas  should
produce accurate data.  Either EPA does not believe the  Protocol  1 standard
is accurate, or they do not believe the linear response  instruments  are
linear; both of which are nonsense.
     Response:  The analyzer calibration error requirements of  Methods 3A,
6C, and 7E have been revised to require the use of  zero, mid-range,  and
high-range calibration gases (i.e., a total of three gases  rather  than four
as originally proposed).  The concentrations of the calibration gases  may
                                     19

-------
be determined using either Alternative 1 (-i.e.,  Protocol  1)  or
Alternative 2 (RM analysis as described in  Section 6.1.2).
     Almost all  analyzers provide for adjustment of the zero response  and
the gain of the analyzer output.  Thus, virtually any analyzer may  be
adjusted to provide the correct response at two  measurement  points. We
believe that verification of proper analyzer response at  a third  point is
necessary to ensure that the analyzer (including linearizer  circuits)  is
properly calibrated over the entire measurement  range.   We also believe
that the results of a quarterly linearity check  performed in the
laboratory are not necessarily representative because (1) repairs and
adjustments to the analyzer might be made immediately prior  to performing
the test or after the quarterly check is completed and (2) problems with
the actual analyzer and linearizer circuits that affect the  accuracy of the
instrument calibration occur in the field when the analyzer  malfunctions.
Thus, a quarterly linearity check would not necessarily ensure the  accuracy
of the analyzer calibration during a specific test.
     4-9  Comment:  The proposed Method 6C  sampling system bias
specification is too restrictive.  A +5 percent  limit is  achievable on a
routine basis.  A +3 percent limit is so tight that other parameters,  such
as instrument warmup time and stability, become  extremely critical. Most
work done in California is on the 0-100 ppm range for NOX, S02, and CO and
+3 percent means +3 ppm.  If the instruments have not been set up the  night
before a test and allowed to thermally stabilize in a temperature-controlled
atmosphere, it will be impossible to meet the +3 ppm limit on  a regular
basis.  A regulation should not be so strict as  to make it unusable for
routine use.  We can supply EPA with data to show that  a  +5  percent limit
                                     20

-------
can be met without extensive (and expensive)  overnight  stabilization of
the instruments, but we question if EPA can provide us  with  their  data to
show that +_3 percent is routinely achievable.  We know  of no consultants
in California (where most of the continuous monitor work  is  done)  who have
contributed such data to E.PA.
     Response:  The sampling system bias check specification of  the
proposed methods was +3 percent of span relative to the response of the
instrument in the analyzer calibration error check.  Thus, for the proposed'
methods, the responses obtained during the sampling system bias  check may
differ by 5 percent of span from the actual calibration gas  values (i.e.,
+2 percent of span for the analyzer calibration error check  plus _+3 percent
of span for the sampling system bias check).  The proposed specifications
were less restrictive than the +_3 percent of the calibration gas
concentration specification of the system calibration requirement  within
the "TCEMS Operational Protocol."  Nevertheless, we believe  that relaxing
the sampling system bias check specification to +5 percent of span will  not
detract from the accuracy of the effluent measurement results provided that
the calculation procedures of the proposed methods are  used  in determining
the effluent measurement results.  Based on the comments received, we
believe that such a change may reduce the costs of performing tests using
Methods 6C and 7E.  Therefore, the specification has been relaxed  to
_+5 percent of the span.
     4-10  Comment:  The proposed Method 6C zero drift  tolerance of
+2 percent of span over the test period is very restrictive  for  mobile
continuous monitoring systems.   It can be met, but not  on a  consistent
daily basis, unless the overnight stabilization technique is implemented,
                                     21

-------
which again increases the cost of the tests.  A _+3 percent of span zero
drift limit would be more reasonable based on our experience with our
systems.
     Response:  We agree with the comment because the sampling system bias
check results, rather than the CD check results, determine whether the
sampling run is valid.  (If the sampling system bias check results before
and after the run are within the specification but the drift is exceeded,
then the tester is required to repeat the analyzer calibration error check
and sampling system bias check before conducting additional  sampling runs.)
The zero drift specification has now been relaxed to +3 percent of span.
     4-11  Comment:  The proposed Method 6C CD specification of +2 percent
of span limit is achievable but not on a routine basis.  Regulations in
California allow +5 percent.  Any data with over +5 percent but less than
+10 percent can be corrected for both zero and span drift.  Any data over
^10 percent drift is rejected.  We would suggest to EPA that the California
regulation be adopted which allows for a data correction to be made, rather
than invalidating the data because the drift exceeded a strict limit.
     Response:  Method 6C does not require invalidating a sample run when
the CD specification is exceeded (see Response 4-10).  A run is invalidated
only if the results of the sampling system bias checks performed before or
after the run exceeded the sampling system bias limit.  The CD
specification has also been relaxed to +3 percent of span.  The calculation
procedures of the proposed methods are retained in the revised methods to
correct for drift during the sampling run and the error quantified by the
sampling system bias check.
                                     22

-------
     4-12  Comment:  The proposed Method 6C interference check  should  not
be required.  In fact, we are opposed to making an interference check  at
all.  If one has to run a Method 6 check for each continuous  monitoring
run, why do continuous monitoring at all?  We can provide EPA with  a great
deal of data showing less than +7 percent variation between Method  6 or
Method 8 data and continuous monitor data, providing the continuous
monitoring system is operated strictly in accordance with the California
Air Resources Board (CARB) or BAAPCD procedures.  We suspect  that the
+7 percent value was chosen because this is the best that Method 6  can
achieve (based on EPA audit data) and is not related to the performance of
the continuous monitors at all.
     Response:  It is not now necessary, nor was it intended  to be
necessary, that a Method 6 check be performed for each continuous
monitoring run.  Refer to Response 1-2.  It is necessary that an initial
inteference check be performed on each source category to show  that the
instrument does not have any otherwise unknown problems.  The interference
check specification of +7 percent of the modified Method 6 value is the
level of agreement that should be obtained if the instrumental  analyzer  is
properly calibrated and operated and the modified Method 6 measurements  are
performed correctly.  The difference between the results of the two
determinations is expected to occur because of measurement errors
arising in both determinations.
     4-13  Comment:  Why is a different type of interference  test being
required for the NOX and 03 or C0£ analyzers compared to the  S02 analyzer
(interference check versus interference response)?  The interference checks
being suggested in this group of methods proposed by EPA are  not consistent
                                     23

-------
from analyzer to analyzer and do not address all the interferences that can
occur in common sampling situations.  This requirement needs to be
completely reworked.
     Response:  A laboratory interference response determination for S02
analyzers is impractical because of the number and type of potential
analytical interferences that may occur.  In contrast, a laboratory
interference response determination for chemiluminescent NOX analyzers is
acceptable, since this particular analytical technique is relatively free
from interferences over the measurement range used for NOX emissions from
stationary sources.  A laboratory interference response determination is
also acceptable for 02 and CC>2 analyzers because interferences are not
significant when the 02 and C02 concentrations being measured are
considered.  Refer to Response 1-2.  The interference check for S02
analyzers has been changed from once per test to once per analyzer per
source category (e.g., fossil-fuel-fired steam generators).
     4-14  Comment:  The proposed Method 6C particulate filter description
needs further clarification.  An expanded statement of acceptable types
similar to the description in Method 5 would be appropriate, since some
glass fiber filters can react with  S02.
     A similar comment was made by  Commenter IV-D-12.
     Response:  The Method 6C filter requirements have been clarified to
specify borosilicate or quartz glass wool  fiber mat for the probe filter.
This is equivalent to the filter specifications for Methods 6, 6A, and 6B.
     4-15  Comment:  In addition to the flow valves specified in the
proposed Method 6C, the entire sample manifold pressure should be
regulated to a constant value (we use 5 psi) to insure that the analyzers
                                     24

-------
are maintained at the same pressure during sampling,  and  especially during
calibration (i.e., it is easy to overpressurize the manifold with  a
2000 psi calibration gas).
     Response:  We believe that the requirements of the proposed Method  6C
are adequate to ensure that accurate results are obtained.  Therefore, the
requirements have not been changed.  However, we agree that the use of a
back-pressure regulator to maintain a constant sample manifold pressure
both (1) protects the analyzer from overpressurization and  (2) makes  it
easier to maintain a constant sampling rate.  A note  has  been added to
Section 5.1.8 that indicates the advantages of installing a back-pressure
regulator.
     4-16  Comment:  Alternative Number 2 in Section  6.1.2  of the  proposed
Method 6C should not be allowed.  We doubt that EPA Method  6 is accurate
and precise enough to certify low span gas concentrations to +5 ppm.  The
thought of allowing anyone to change the manufacturer's cylinder tag
regardless of whether the gas value is right or wrong is  ridiculous.
     Response:  We disagree with the comment, although we understand  the
commenter's concern.  Method 6 is sufficiently accurate and precise to
allow verification of calibration gases to +5 percent or  +5 ppm, whichever
is greater, in many cases.  However, special precautions  must be taken in
the analysis of low concentration gases in order to satisfy the
requirements of Alternative Number 2.  In addition, the tester should be
aware that the error in the Method 6 results may make it  difficult to
comply with the analyzer calibration error and sampling system bias check
specifications.  Therefore, we suggest the use of Protocol  1 gases (i.e.,
                                     25

-------
Alternative Number 1).  The wording of Section 6.1 has been changed to
reflect this recommendation.
     4-17  Comment:  There is nothing said in Method 6C about housing the
instruments in a clean, thermally-controlled environment.   The EPA could at
least provide a comment to this effect, otherwise, it will  be impossible to
meet the drift specifications.
     Response:  Method 6C has been revised to incorporate  this comment.
     4-18  Comment:  In Section 6.2 of the proposed Method 6C, there is  no
statement regarding a simple leak check to insure there are no gross leaks
in the system.  We normally plug the end of the sampling probe when the
monitoring system is completely assembled and watch the rotameters in the
system fall to zero before proceeding with the sampling system bias check.
     A similar comment was made by Commenter IV-D-12.
     Response:  The sampling system bias check and drift checks of
Method 6C will detect leaks in the measurement system.  Thus, a leak check
per se is not required.  The appropriate procedures for performing a leak
check depend on the specific components and design of the  measurement
system.  The tester is free to conduct such checks during  the measurement
system preparation if he so wishes.
     4-19  Comment:  There is no need to repeat the response time tests
specified in Section 6.4 of the proposed Method 6C three times to get an
average value.  This simply wastes calibration gas and costs money.  One
upscale and one downscale measurement is sufficient to determine a response
time so that a sampling time can be calculated.  Three upscale and three
downscale measurements of response time is complete overkill.  In
situations where 300 feet of line is used to get the sample from the stack
                                     26

-------
to the trailer on the ground, the amount of time and calibration  gas
consumed doing six response checks would be outrageous even if a  cheaper
certified gas were used.
     Similar comments were made by Commenters IV-D-5 and IV-D-12.
     Response:  The Method 6C requirements for determination of response
time have been simplified.  The procedures of the proposed method in
Section 6.4 have been eliminated.  The procedure in Section 6.5 have  been
modified to require that (1) the zero and mid- or high-range gases be
introduced until a stable response is achieved and (2) the tester observe
the response time of the measurement system during the sampling system bias
check.  Reporting of the response time is no longer required.
     4-20  Comment:  In reference to Section 6.5.2 of the proposed
Method 6C, a zero and an 80 percent span gas (not mid-range gas)  would be
adequate to perform this check.  It should be stated that certified
calibration gas (+2 percent) or Protocol 1 gas would be acceptable for this
check.
     Response:  We believe that the sampling system bias check should be
performed using a zero gas and the calibration gas that most nearly
approximates the effluent concentration.  In the development of the
proposed methods, it was believed that the mid-range gas would be closest
to the effluent concentration.  However, based on the comments received, it
is apparent that in some instances the high-range calibration gas will more
closely represent the effluent concentration.  Therefore, the methods now
require the tester to use either the mid-range or high-range calibration gas
(whichever most closely approximates the effluent concentration)  for  the
sampling system bias check.
                                     27

-------
     Calibration gases with a manufacturer's tolerance not to exceed
+_2 percent are acceptable (in addition to "Protocol  1 gases") provided
Section 6.1.2 of Method 6C is followed.  However, use of calibration gases
without having verification of their concentration by direct analysis has
not proven to be acceptable.
     4-21  Comment:  The option to omit the check described in Section 7.2
of the proposed Method 6C and petition the Administrator is really not a
viable option for consultants such as we.  We would have to provide data
to the Administrator prior to each test on a different source to show that
our continuous monitoring data is not biased low.  The EPA knows this is
not economically possible.
     Response:  This concern should be satisfied in view of changes made in
the interference check requirements.
     4-22  Comment:  When a NOX analyzer is installed on a stationary
source, there is no interference response test required by Performance
Specification 2 (PS 2).  Why should it be continually required now?
     Response:  We disagree with the comment.  When a NOX CEMS is installed
at a stationary source subject to EPA regulations, PS 2 requires a relative
accuracy test to be performed.  This test checks for the presence of
interferents.  Refer also to Response 1-2.
     4-23  Comment:  When a Method 6 test is run on a stack, why is there
no interference check required?  There are other compounds besides NH3 that
may produce a low result.  No one examines all  the species present in the
stack gas in advance to see if one specie might interfere, so why require
it for continuous monitoring?  What is to say that a compound will not
interfere with the continuous monitor but will  with the RM?
                                     28

-------
     Response:  An interference check is not required for Method  6 because
it is the RM, and as such, it has already been subjected to inteference
considerations.
     4-24  Comment: It would be better if EPA were to specify that a
preliminary traverse be done with the continuous monitoring probe to insure
there was no stratification prior to conducting a test.   If no
stratification (no individual reading less than J^IO percent of average
reading) existed, then a single point sample would be sufficient.  This
would eliminate the need to have an extra man traverse the stack  with the
probe and reduce the cost of testing.
     A similar comment was made by Commenter IV-D-12.
     Response:  The requirements for selection of a sampling site and
sampling points for Methods 3A, 6C, and 7E are to use the same criteria  as
are applicable to Methods 3, 6, and 7, respectively.  These criteria are
found in the applicable subparts for performance tests and in Appendix B
for relative accuracy (RA) tests of Continuous Emission Monitoring Systems
(CEMS's).
     4-25  Comment:  In reference to Section 7.3 of the proposed  Method  6C,
why can sampling time not be made simple?  Simply state each run  must be
1 hour in length and forget about two times the average response  time plus
the Method 6 time.
     A similar comment was made by Commenter IV-D-12.
     Response:  The duration of a sampling run is specified in the
applicable Subparts for performance tests and in Appendix B for RA tests of
CEMS's.  The duration of a sampling run is not specified in Methods 6, 6A,
or 6C.  Therefore, Method 6C refers to the same criteria that are applicable
to Method 6.  Ignoring the data at the beginning of the sample run (i.e.,
                                     29

-------
twice the response time of the measurement system) ensures that the
measurement system is responding to the effluent sample rather than to the
effects of residual calibration gases within the measurement system.
     4-26  Comment:  In reference to the emission calculation procedures
of the proposed Methods 3A, 6C, and 7E, if the .measurement system meets the
specification for drift and bias, etc., then no corrections should be made
to the data.  For example, EPA Method 5 allows a leak rate of 0.02 cfm.  If
the leak rate exceeds 0.02 cfm, the data must be leak-corrected.  The same
principle should apply here.  No correction of the data should be required
unless the drift or bias specifications are exceeded.  Corrections should
not be allowed at all if the drift and bias specifications are exceeded by
more than 10 percent.  This will streamline data reduction costs but not
affect the accuracy or reliability of the data in any appreciable manner.
     A similar comment was made by Commenter IV-D-12.
     Response:  We disagree with the comment.  Calculating the effluent
concentrations relative to the results of the sampling system bias check
results provides the most accurate results.  For a linear analyzer, this
approach eliminates all  of the error quantified  by the analyzer calibration
error and sampling system bias checks.  Furthermore, if the response of the
analyzer drifts at a constant rate during a run  (but remains within
acceptable limits), the calculation procedure of the proposed methods also
eliminates any error associated with drift.  Finally, use of the proposed
calculation procedure allows for the sampling system bias check
specification to be set at a higher value than would otherwise be
appropriate.
                                     30

-------
     The analogy to Method 5 is not relevant.  In Method  5  there  is  no
way to estimate accurately the actual  bias due to leaks  (except for  using
the worst case assumption), because the sampling vacuum  varies during the
test and is typically less than the vacuum during the leak  check.  An
analogy to Method 7, where the "best fit curve" relative  to the four
standards used for determining the Kc value is employed,  is perhaps  more
relevant.  The calculation procedure of the proposed methods is
conceptually similar to the procedures included in the Method 7 analysis.
     4-27  Comment:  In reference to proposed Method 7E,  an N02 to NO
converter should be specified all the time since a test  firm does not
know if the N02 is less than 5 percent of the total  NOX.
     Response:  The requirements of Method 7E have been  revised to require
the use of an N02 to NO converter unless data are available to demonstrate
that the N02 concentration is less than 5 percent of the  effluent NOX
concentration.
Commenter'IV-D-5
     5-1  Comment:  The length of time over which both the  zero drift and
CD are measured is inadequately defined in the proposed  methods.
     Response:  Section 7.4 has been revised to clarify  that the  sampling
system bias checks must be performed immediately before  and after each
sampling run.  Section 7.4.2 has been clarified to state  that the drift  is
determined based on the difference between the sampling  system bias  check
results immediately before and after each sampling run.   The specific time
over which the drift checks are performed is not stated,  since the duration
of the sampling run is not specified in the proposed methods (see
Response 4-25).
                                     31

-------
     5-2  Comment:  The allowable analyzer calibration error of +2  percent
of span in the proposed methods could produce a significant doubt as  to  the
determination of source compliance in a marginal  situation.  This problem
would be particularly acute if the average emission rate approached
20 percent of the span value, while the calibration is referenced to  the
full span range.  The EPA docket material  requires calibration checks
within 3 percent of the zero and mid-range gases.  In addition, it  requires
a span gas with pollutant concentration close to  the stack gas being  measured
This is a more reasonable approach.
     Response:  For the proposed Methods 6C and 7E, the analyzer
calibration error specification is +2 percent of  span and the sampling
system bias check is +3 percent of span.  The EPA docket material referred
to by the commenter is the "TCEMS Operational Protocol" (EPA-340/1-83-106).
This document imposes a system calibration requirement of +3 percent  of  the
calibration gas value.  It is true that the system calibration check
requirement in the referenced document is  more restrictive than the
specifications included in the proposed Methods 6C and 7E.  However,  the
emission calculation procedure included in the proposed methods removes  the
errors quantified by the analyzer calibration error and sampling system
bias checks from the effluent measurements for a  linear analyzer (see
Response 4-26).  Therefore, tighter specifications are not justified  for
the proposed methods, and little, if any,  uncertainty will exist regarding
the accuracy of emission measurement results for  a source with emissions
close to the standard.  The EPA believes that the calculation procedure
included in the proposed methods will  effectively eliminate the errors
quantified by the sampling system bias check, provided that the zero  and
                                     32

-------
upscale bias check results are within 5 percent of span  of the responses
exhibited by the analyzer during the analyzer calibration error check.
     5-3  Comment:  A more practical CD PS in the proposed methods  would  be
+3 percent of calibration gas values.  It is unclear whether a calibration
device would be acceptable for determining CD.
     Response:  The CD limit has been increased to +3 percent of span
(refer to Response 4-11).  The CD check must be performed using the same
calibration gas used for the sampling system bias checks.  The calibration
gas concentration must be verified according to the requirements of
Section 6.1.  Other calibration devices cannot be used.
     5-4  Comment:  In reference to Section 6.1 of the proposed Method  6C,
the use of calibrated gas cells, if the monitor is so equipped, could be
another alternative to gas cylinders.
     Response:  We disagree with the comment.  We do not have information
that the monitor responses to calibration gas cells are representative  of
actual monitor performance.  Therefore, no change to the method has been
made.
     5-5  Comment:  In reference to Section 6.5.1 of the proposed
Method 6C, the use of a calibration gas which represents a value close  to
the stack effluent should be substituted for the mid-range gas.
     Response:  We agree that use of a calibration gas which most nearly
approximates the effluent concentration would be preferable for the
sampling system bias check.  The method has therefore been revised  to
allow either the mid- or high-range gas for this purpose.
                                     33

-------
Comrnenter IV-D-6
     6-1  Comment:  The proposed methods use three span levels of
Protocol 1 gases or Methods 3, 6, or 7 proven gases for analyzer
calibration error checks in the field.  These checks essentially verify
instrument linearity or calibration curve fit.  An alternative should  be
added allowing a zero and five-point span check to be performed within
30 days before and after field use of the instruments.   The calibration
should be performed utilizing an appropriate dynamic dilution system and  a
high-level span gas.  The span and zero gases should have a +_! percent
accuracy as certified by the gas manufacturer.  The five span levels should
be in the 10-25 percent, 30-45 percent, 50-65 percent,  70-85 percent,  and
90-100 percent level of the instrument range used.
     During field use of the instrument, a Protocol 1 zero and mid-level
span gas should be used to provide before-test run calibrations, and
after-test zero and CD tests.  The use of a gas manufacturer analyzed
+1 percent high-level gas should be allowed when performing the response
time tests and sampling system bias checks.
     The alternative allowing +1 percent accuracy non-Protocol  1 gases
for the multipoint calibration, response time tests, and sampling system
bias checks would be less expensive than using Protocol  1 gases.  This
would be particularly true because of the large amount  of gas required to
perform the response time test and sampling system bias checks (gas  is
introduced through the complete sampling system).   Diluting one calibration
gas to provide five span gas levels, if performed  properly, will  be  more
accurate than using three different span gas cylinders.   It is possible
                                     34

-------
for a 2 percent error to be Introduced when performing  a  linearity check
by using two different gas cylinders, since Protocol  1  gases  have an
accuracy of +1 percent.  A five point span check  would  also give greater
coverage over the instrument range than a three point check.
     We believe this alternative would not reduce the accuracy of the
proposed test methods and performing the five point  linearity calibration
in the laboratory would reduce the field testing  time.
     Response:  The analyzer calibration error requirements of Methods  3A,
6C, and 7E have been revised to require the use of zero,  mid-range, and
high-range calibration gases (i.e., a total of three gases rather than
four, as originally proposed).  In addition, the  requirements for
determining response time have been simplified so that  no additional gas
injections are necessary to quantify the response time  (see Response 4-19).
These changes will reduce the number and quantity of calibration gases
required by the methods.
     The commenter's suggestion of using a laboratory five-point
calibration check based on the dynamic dilution of a single calibration gas
is of interest.  However, we are not convinced that  the laboratory checks
would always be representative of the analyzer performance during a
specific field testing program (see Response 4-8).  The approach suggested
by the commenter could be approvable as an alternative  method for
supplying calibration gases, provided that sufficient data and procedural
description were submitted to demonstrate the accuracy  of the dynamic
dilution method.
                                     35

-------
Commenter IV-D-7
     7-1  Comment:  The commenter will be able to complete development of
and have adequate time to provide his comments to EPA with the granted
30-day extension of the comment period.
     Response:  No response is required.
Commenter IV-D-8
     8-1  Comment:  In Method 3A, Section 5.1.4 states that a means for
controlling analyzer flow rate and a device for determining proper sample
flow rate must be provided.  Since some oxygen analyzers are not sensitive
to gas flow rates, we question the need for these requirements in all
situations.
     Response:  Section 5.1.4 of Method 3A has been modified by including
the following statement:  "The requirements for measuring and controlling
the analyzer flow rate are not applicable if data are submitted that
demonstrate that the analyzer response is insensitive to flow rate
variations over the range encountered during the test."
     8-2  Comment:  In Method 6C, there appears to be a bar missing over
the C inside the brackets of Equation 6C-1.
     Response:  We agree.  The equation has been corrected.
Commenter IV-D-9
     9-1  Comment:  The proposed regulations to include instrumental  test
methods as an acceptable means of determining compliance can be an accurate,
reliable and inexpensive means of measuring pollutants on a continuous
basis.
     The commenter has used this methodology as an alternative to manual
sampling for the last two compliance tests.  By using continuous analyzers,
                                     36

-------
malfunctions or sampling errors are more readily detected;  consequently,
retesting (and associated cost) is substantially reduced or eliminated.
Results from the analyzers are consistent with both the manual  sampling
methods and the plant's CEMS.
     Calibration of these analyzers, using EPA Protocol  1 gas,  serves  as
additional QC in that the gas is analyzed according to RM's.
     Response:  We agree with these remarks.  However, it should  be  noted
that EPA Protocol 1 gases are not analyzed using EPA RM's.
     9-2  Comment:  The proposed regulations do not specifically  mention
applicability to 40 CFR Part 60, Appendix B.  Inasmuch as the potential
benefit for CEMS certification is equally as valuable, we recommend  that
specific language be included designating the proposed changes  as
applicable to Appendix B.
     Response:  We agree.  However, we have already listed the  methods as
acceptable for this purpose in those Part 60 subparts where they  are listed
as acceptable compliance test methods.
Commenter IV-D-10
     10-1  Comment:  In Section 5.2 of proposed Method 3A, and  Section 5.3
of proposed Method 6C, gas mixtures (S02/02 in Ng or S02/C02 in N2)  should
be allowed for calibration of S02/diluent systems in order to reduce the
number of cylinders required.
     Response:  Calibration gas mixtures of S02/C02 in N2 and S02/02 in  N2
are available as EPA Protocol 1 gases.  The three component gas mixtures
are convenient because the number of calibration gas injections,  as  well  as
the number of cylinders required for a test program, is decreased.   However,
                                     37

-------
the cost of the gas mixture relative to the gas volume may not  be cost
effective.  The regulation has been changed to allow the  tester to select
this option.
     10-2  Comment:  In Section 5.1.4 of proposed Method  6C,  for an
S02/C02 system that measures both components on a wet basis and is used
to determine Ibs 502/10° Btu, determining moisture content and  reporting
individual results on a dry basis should not be required, since it is
unnecessary in order to obtain the desired results.
     Response:  The following note has been added to Section  5.1.4:  "The
determination of sample moisture content is not necessary for pollutant
analyzers that measure concentrations on a wet basis when (1) a wet basis
C02 analyzer operated according to Method 3A is used to obtain
simultaneous measurements, and (2) the pollutant/C02 measurements are  used
to determine emissions in units of the standard."
     10-3  Comment:  In proposed Methods 3A, 6C, and 7E,  data recorders
should not be required if readings are recorded manually, similar to wet
chemistry RM testing.  A frequency for recording data should  be specified,
similar to the "cycle time" for CEMS's.
     Response:  We agree with the comment.  Therefore, the following
sentences have been added to Section 5.1.11:  "Alternatively, a digital or
analog meter having a resolution of 0.5 percent of span may be  used to
obtain the analyzer responses and the readings may be recorded  manually.
If this alternative is used, the readings shall be obtained at  equally
spaced intervals over the duration of the sampling run.  For  sampling  run
durations of less than 1 hour, readings shall  be obtained at  1-minute
                                     38

-------
intervals or a minimum of 30 measurements, whichever is less  restrictive,
shall be obtained.  For sampling run durations greater than  1 hour,
measurements at 2-minute intervals or a minimum of 96 measurements,
whichever is less restrictive, shall be obtained."  Similar  instruction  has
been added to Section 8 to ensure that the average concentration displayed
by the analyzer is determined in a consistent manner.
     10-4  Comment:  In proposed Methods 3A, 6C, and 7E, use  of gases
prepared following Protocol Number 1 should be allowed only  if N8S
Standard Reference Materials (SRM's) were used exclusively in the Protocol.
Use of gas manufacturer's primary standards lowers the confidence with
which the concentrations are known.
     Response:  We believe that analysis of calibration gases according  to
Protocol 1 is sufficient to ensure the accuracy of the calibration gas
values.  No change to the requirement has been made.
     10-5  Comment:  In proposed Methods 3A, 6C, and 7E, sources should  be
allowed to "certify" cylinder gases using a laboratory analyzer calibrated
with NBS SRM's.
     Response:  This option is already allowed, provided that the gas
analysis is performed according to the requirements of Protocol 1.  It is
believed that use of this option is unlikely because of the  cost associated
with NBS SRM's.
     10-6  Comment:  In proposed Methods 3A, 6C, and 7E, the response time
should be measured from the stack concentration to the zero  and high-range
calibration gas concentrations.  This would allow determination of the
95 percent response level easily without the difficulty of defining
"stable value."
                                     39

-------
     Response:  The difficulty of determining the 95 percent response level
for concentration changes between the zero or high-range calibration gas
and the effluent concentration is eliminated because of the simplification
of the response time determination requirements.  See Response 4-19.
     10-7  Comment:  Since the proposed method would expand the use of
instrumental methods in determining compliance with source emission
standards, EPA should seriously consider certifying the equivalency of
the analytical portions of such systems whether used continuously or for
compliance tests.  Many agencies do not have the resources necessary to
determine the applicability of a particular analytical technique to
various sources.  In addition, sources would be protected in the event
of discovery of a problem requiring modification of the analyzer to
maintain equivalency.
     Response:  The EPA does not certify equivalent methods for source
emission measurements.
Commenter IV-D-11
     11-1  Comment:  These instrumental methods are well conceived and
written, and we are extremely supportive of these instrumental techniques
which allow relatively simple accuracy evaluation of CEMS's.
     Response:  No response is required.
     11-2  Comment:  These are extremely important revisions, since
accuracy audits should be part of any monitoring program.  However, these
techniques should not be used for the initial certification of a CEMS.  It
is not good scientific design to certify one instrument using another.
After certification, I feel  all audit functions could be accomplished using
these methods.
                                     40

-------
     Response:  The EPA believes that the proposed  instrumental methods
will provide accurate effluent measurements.   Therefore,  these methods are
suitable for use in conducting RA tests of installed  CEMS's, since the
purpose of the PS tests is to determine whether  an  installed CEMS is
capable of providing valid effluent measurements.
Commenter IV-D-12
     12-1  Comment:  The commenter believes that EPA  has  not sufficiently
demonstrated the adequacy of the three proposed  methods for use in
determining compliance with S02 and NOX emission standards generally.  The
commenter does not disagree with the concept that  instrumental test methods
could be used to determine compliance, but believes that  the proposed
methods have not been adequately demonstrated by EPA  to fulfill the
necessary requirements of general compliance assessment procedures.  That
is, before a procedure can be used to assess compliance,  both  EPA and the
affected sources must have a quantitative knowledge of the precision,
accuracy, and reliability of the procedure.  The commenter believes that
EPA has not yet produced enough critical information  to establish the
adequacy of these techniques as generally applicable  methods.
     Response:  The EPA believes that Methods 3A,  6C, and 7E as promulgated
are adequate for use in conducting both source performance tests and CEMS
RA tests at fossil-fuel-fired steam generators in order to demonstrate
compliance with subparts D and Da of 40 CFR Part 60 for several  reasons.
     First, EPA is aware that instrumental test methods for S02, NOX, C02,
and 02 are already used widely by many source testing organizations.
Applications include the collection of critical  engineering data in
circumstances where large economic consequences are associated with the
                                     41

-------
test results.  Many papers are found in the literature and are presented at
professional meetings describing instrumental  test methods and the
successful results provided by the application of such methods.  Some State
and local control agencies, including the CARB, the BAAQMD, and the South
Coast Air Quality Management District (also in California), have adopted or
approved such instrumental test methods.  In addition, EPA has already
proposed and promulgated several methods that are similar to Methods 3A,
6C, and 7E.  These methods include Method 20 for the measurement of NOX
emissions from both gas turbines and internal  combustion engines (i.e.,
NSPS Subpart 66 and the proposed Subpart FF, respectively), and Method 10
for the measurement of CO emission levels (i.e., NSPS Subpart J).
     Second, and most important, the technical adequacy of Methods 3A, 6C,
and 7E is based on a thorough and sound understanding of emission
monitoring instrumentation and the application of this technology to the
measurement of emissions from stationary sources.  Each of the methods
contains equipment specifications to ensure that appropriate
instrumentation is used.  Each of the methods includes operational PS's.
The EPA has carefully considered each specification and the corresponding
qualification that it places on the data provided by the method.  Based on
these fundamental principles, EPA is convinced that when the equipment and
operational PS's are met, the methods will  provide accurate and precise
data.  The fact that conformance with the applicable operational PS's can
be demonstrated during each field test provides a greater level  of
assurance of the quality of the results than heretofore achievable.
                                     42

-------
     Third, EPA considered many methods in the development of Methods 3A,
6C, and 7E, including the "TCEMS Operational Protocol" (EPA 340/1-83-016),
the BAAQMD Source Test Methods ST-13A and ST-19A, and others.  The EPA
recognizes that each of these methods reflect slightly different approaches
and applications of instrumental techniques.  Based on the review of the
various methods and in consideration of QA factors, EPA has developed a set
of procedures and PS's that will best ensure the quality of the data
obtained, while also providing for user flexibility.
     The EPA has previously included in the docket a summary of test
results obtained with the "TCEMS Operational Protocol" to demonstrate that
sufficiently accurate data can be obtained through the application of
instrumental methods for the measurement of effluent S02 concentrations.
The EPA recognizes that the equipment and PS's within "The TCEMS
Operational Protocol" differ from those of Methods 3A, 6C, and 7E.  These
differences reflect the fact that "The TCEMS Operational Protocol" and
Methods 3A, 6C, and 7E were developed for different purposes.  ("The TCEMS
Operational Protocol" applies to only one specific set of instruments,
rather than to a generalized method.)  Thus, the accuracy of those test
results included in the docket are not intended to represent the accuracy
of data that will be provided by Methods 3A, 6C, and 7E,- but instead
demonstrate, in general, that sufficiently accurate results can be
obtained when a well-defined procedure is followed.  The accuracy of test
results provided by Methods 3A, 6C, and 7E should in fact' surpass the
accuracy of data obtained in accordance with the "TCEMS Operational
Protocol."
                                     43

-------
     It should be noted that use of the Methods 3A,  6C,  and  7E  is  not
required by Subparts D and Da, in the promulgated revisions  to  those
subparts, but instead is an option to be employed solely at  the discretion
of the source.  Thus, if a particular source doubts  their ability  to
execute properly the instrumental test methods, then that source can
continue to rely on manual, wet chemical methods as  they have  in the past.
The EPA is satisfied that Methods 3A, 6C, and 7E will  provide data of
acceptable quality: each potential  user can arrive at their  own conclusion.
     12-2  Comment:  The proposal is deficient because it does  not
designate the proposed methods as either reference,  equivalent, or
alternative methods.  The commenter believes that the Agency must  designate
the proposed methods as one of these three types of  methods  and explain  the
significance of such a designation.
     Response:  Appendix A of 40 CFR Part 60 is meant to serve  as  a
repository of test methods that EPA may use or refer to  in specific
subparts of Part 60 (Subparts D and Da, for example) for specific
compliance determination purposes.   Except as they are cited for
application in a specific subpart or subparts to Part 60, no other
representation of a method is made or intended.  It  was  EPA's intention
that the proposed revisions to Subparts D and Da mean that the  proposed
methods could be alternative choices, and no mandatory use was  expected.
The revisions to Subparts D and Da  as promulgated have been  clarified  to
say specifically that Methods 3A, 6C, and 7E are to  be used  only at the
sole discretion of the source owner or operator.
     12-5  Comment:   The entire issue of instrumental  linearity should be  .
carefully considered.  By requiring the use of linear (or linearized)
                                     44

-------
instruments, EPA could significantly simplify  the  methods.  Moreover, the
use of four calibration gases is excessive and could  be  replaced by
requiring only zero and 80 percent of span gas.  In addition to reducing
costs associated with the procurement of gases and regulators, this
procedure would reduce the time required to move gas  cylinders to the
sampling location (when required) and the time used in the  calibration
process.
     Response:  Use of analyzers equipped with linearizers  requires the
use of at least three calibration gases to check (1)  the analyzer zero,
(2) the analyzer gain, and (3) the performance of  the linearizers, rather
than a two-point calibration check as the commenter suggests.  See
Response 3-2.
     12-4  Comment:  Reconsider the sensitivity, system  bias, CD,
calibration error, and zero drift requirements.  In each of these areas,
the values in the proposed methods appear to be in conflict with other
existing limits or procedures.  For example, the proposal calls for a
calibration error of "less than +2 percent of the  span," (50 FR 8291)
while EPA's key supporting report in the docket specifies a response
"within +3 percent of the calibration gas value."   Further, the proposal's
use of four calibration gases conflicts with most  established procedures,
including the procedures in EPA's report in the docket.
     Response:  We have reconsidered all of the applicable  specifications
and have made revisions to the number of calibration  gases, the sampling
system bias, zero drift, and CD specifications. We believe that the
revised specifications are adequate to ensure that applications of the
revised methods will provide data of sufficient accuracy and precision.
Refer to Responses 4-8 through 4-11.
                                     45

-------
                  Table 2.  LIST OF COMMENTERS
                     Docket Number A-84-35
Docket'Item Number                       Commehter/Affi1 i ati on

      IV-D-1                       R.R. Kienle, Manager
                                   Environmental Affairs
                                   Shell  Oil  Company
                                   One Shell  Plaza
                                   Houston, Texas 77002

      IV-D-2                       U.V. Henderson, Associate Director
                                   Environmental Affairs
                                   Texaco, Inc.
                                   Post Office Box 509
                                   Beacon, New York 12508

      IV-D-3                       Hopping Boyd Green and  Sams for
                                   Florida Electric Power
                                   Coordinating Group
                                   Post Office Box 6526
                                   Tallahassee, Florida  32314

      IV-D-4                       Jim Steiner
                                   Pape and Steiner Environmental
                                   Services
                                   5801 Norn's Road
                                   Bakersfield, California 93308

      IV-D-5                       W.J. Barrow, Jr., Manager
                                   Environmental Permitting Programs
                                   Florida Power and Light Company
                                   Post Office Box 14000
                                   Juno Beach, Florida 33408

      IV-D-6                       Lawrence J. Ogden, Vice President
                  i                 Construction and Operations
                  ;                 Interstate Natural Gas  Association
                                   of America
                                   1660 L Street, Northwest
                                   Washington, D.C. 20036

      IV-D-7                       Hunton and Williams for
                                   The Utility Air Regulatory Group
                                   Post Office Box 19230
                                   Washington, D.C. 20036
                               46

-------
                  Table 2.  LIST OF COMMENTERS
                          (Continued)
                     Docket Number A-84-35
Docket Item Number                  Commenter/Affi1 i ation

      IV-D-8                  John E. Pinkerton, Program Manager
                              Air Quality
                              National Council of the Paper Industry
                              for Air and Stream Improvement, Inc.
                              260 Madison Avenue
                              New York, New York 10016

      IV-D-9                  Jack L. Byron, Vice President
                              Engineering
                              Sierra Pacific Power Company
                              Post Office Box 10100
                              Reno, Nevada 89520

      IV-D-10                 Ben A. Brodoviczf, Chief
                              Division of Technical Services and
                              Monitoring
                              Commonwealth of Pennsylvania
                              Department of Environmental Resources
                              Bureau of Air Quality Control
                              Post Office Box 2063
                              Harrisburg, Pennsylvania 17120

      IV-D-11                 Vincent J. Brisini, Environmental
                              Scientist
                              Pennsylvania Electric Company
                              1001 Broad Street
                              Johnstown, Pennsylvania 15907

      IV-^D-12                 Hunton and Williams for
                              The Utility Air Regulatory Group
                              Post Office Box 19230
                              Washington, D.C. 20036
                               47

-------