CO
              United States
              Environmental Protection Air and Radiation  EPA/400/1-91/008.C
              Agency         (ANR-445)     April 1991
ve/EPA        Acid Rain Advisory
              Committee Meeting:
              March 20-22, 1991
               Emissions Monitoring
               Issue Papers
HEADQUARTERS LIBRARY
ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460

-------

-------
                         INDEX

        Emissions Monitoring Subcommittee Papers
                  March 20  - 22,  1991
Document       Title                                    Page
number

£•14      Minutes from February 20*21, 1991 Subcommittee  1
           Meeting
£-15      Roberson Paper on Relative Accuracy Test Audits 14
£-16      Shigehara Slides (hardcopy) on Wet Chemistry    24
£-17      Wrona ARAC Briefing Slides (Hardcopy)           27
£-18      Poirot Paper "Time Resolution for Reporting     44
           CEM Data"
£-19      Central and South West Services, Inc Issue      53
           Paper "Continuous Emission Monitoring
           Exemption for Gas-Fired Utilities"
E-20      Houston Lighting & Power Co., and Texas'        60
          Utilities  Electric  Co.  Issue  Paper
           "Alternative Emission Monitoring Approaches"
£-21      Alternate Methods by EPA Reporting and          81
           Recordkeeping Requirements by EPA
           (Distributed prior to or at meeting)
£-22      Reporting and Recordkeeping Requirements by EPA 92
£-23      State Agency Experience in Data Recordkeeping   104
          and Reporting
£-24      Baseline Tons of S02 (1985)  for  Selected       108
           Categories of Affected Units Under Title IV
£-25      Coal Sampling and Analysis as an Alternative    109
           to Continuous Emission Monitoring in
           Pennsylvania (Presented at 1983 Engineering
           Foundation Conference Pocono Hershey Resort,  PA)

-------

-------
                                                                               I
                                                                            EH
                       ACID RAIN ADVISORY COMMITTEE

                      Subcommittee on Emissions Monitoring

                                    MINUTES

                               February 20-22, 1991
                                 Washington, DC
ATTENDANCE

The following Subcommittee members and U.S. EPA staff attended the February 20-22,

1991 meeting:

Ms. Nancy Wrona
Director, Office of Air Quality
Arizona Department of Environmental Quality
Phoenix, AZ
Acting Subcommittee Chair

Mr. Henry Beal
Vice President for Strategic Planning
Research-Cot trell Companies
 ranch burg, NJ

Mr. Robert Bergstrom, Jr.
Chief Counsel
Iowa Southern
Centerville, IA

Mr. Jerry Golden
Manager, Clean Air Program
Tennessee Valley Authority
Chattanooga, TN

Mr. David Hawkins
Senior Attorney
Natural Resources Defense Council
Washington, DC

Mr. Larry Kertcher
Source Control Branch Chief
Acid Rain Division (ARD)
Office of Atmospheric and Indoor Air Programs (OAIAP)
U.S. Environmental Protection Agency
Washington, DC                                              .
EPA Lead
                                                          Preliminary Draft for AHAC
                                              Discussion Does Not Represent the Position
                                             of the U.S. Environmental Protection Agency

-------
                                                                              E14
Mr. Robert McWhorter
Senior Vice President
Ohio Edison
Akron, OH

Mr. Daniel Plum ley
Director of Park Protection
The Adirondack Council
Elizabethtown, NY

Mr. Richard Poirot
Air Quality Planner
Vermont Department of Environmental Conservation
Waterbury, VT

Also attending the meeting were:

Ms. Doris Price, Emissions Monitoring Section Chief
Ms. Margaret Shepherd, Environmental Scientist
U.S. Environmental Protection Agency
Washington, DC

Mr. Joe Nazzaro
Chief, Continuous Emissions Monitoring Unit
Department of Environmental Resources
Harrisburg, PA


The meeting was facilitated by Dr. William Warren-Hicks of KUkelly Environmental
Associates and Dr. Joan Baker of Western Aquatics, Inc.


GENERAL

The Acid Rain Advisory Committee (ARAC) Continuous Emissions Monitoring (CEM)

Subcommittee met at the Ramada Renaissance Hotel, Herndon, VA on February 20,

1991, for a full day of discussions and February 21 for a half day of discussions. During

the afternoon of February 21 and for the entire day of February 22, 1991, the
Subcommittee members attended the meeting of the full ARAC, which was also

convened at the Ramada Renaissance Hotel.  The following minutes summarize the

Subcommittee meeting and the presentations made by the Emissions Monitoring

Subcommittee to the full ARAC. These minutes are organized principally by topic area
under the following major categories: (1) the Pennsylvania experience, (2) missing data
periods, (3) utility of Subpart Da regulations for Title IV of the Clean  Air Act

Amendments (CAAA), (4) COz monitoring, and (5) assignments for the March meeting.
                                                           Preliminary Draft fur ARAC
                                               Discussion Docs Nut Represent the Position
                                              of the U.S. Environmental Protection Agency

-------
                                                                              £14
 1. THE PENNSYLVANIA EXPERIENCE

 The following is a summary of a presentation by Mr. Joe Nazzaro, Pennsylvania

 Department of Environmental Resources (PaDER), concerning the experience of the

 State of Pennsylvania with the use of continuous emission monitors (CEMs) for meeting

 state air quality regulatory requirements.

 •    PaDER has encouraged CEMs for measuring SOz since the early 1970s, with formal
      regulations beginning in 1979.  Currently over 200 CEMs are operating in
      Pennsylvania with another ZOO CEMs expected to be operating in the near future.

 •    All CEMs undergo a 3-step approval  process that includes (1) submission by the
      source of a proposed  monitoring plan, (2) performance testing of the CEM with a
      written report submitted to PaDER,  and  (3) PaDER review and approval of  the
      report.

      The source is required to establish accuracy of CEM measurements and  meet  the
      quality assurance audit criteria as established by PaDER.

      PaDER generally requires  more performance testing than EPA  does, and some
      audit procedures and criteria are different from those established by EPA in
      Subpart Da requirements.

 •    As part of the quality assurance/quality control (QA/QC) requirements, PaDER
.      requires the source to submit a quarterly emissions report including quarterly
      calibration error testing. All hourly SOz averages are reported to the state.  The
      PaDER experience shows that this type and amount  of data is manageable using
      personal computer data bases. The  burden of proof is on the source to
      demonstrate compliance with air quality operating permits. In addition to
      penalties for violations of  the emissions standards, sources are penalized for the
      amount of missing  data above prespecif led limits. The amount of the penalty is a
      function of the total amount of missing data.

 •    Out of control (OOC) periods are determined retrospectively for daily zero and
      calibration/drift checks. Generally, 90%-95% availability of SOz CEMs is
      observed. Some sources have backup CEMs  to avoid penalties.

 •    Pulverized coal sampling,  in accordance  with Pennsylvania design and performance
      specifications, and wet chemical methods have been approved as alternatives to
      primary CEMs.

 •    Some sources have attempted to use parameter monitoring as an alternative
      compliance demonstration during periods of invalid CEM data.  However, PaDER
      does not  believe that sufficient documentation is available to justify parametric
      monitoring as a viable alternative.

 *    PaDER does not require 100% SOz data availability on an hourly basis.  The
      Pennsylvania requirements (75% data collected during the hour) allow for a certain
                                                            Preliminary Draft for AKAC
                                                Discussion Docs Not Represent the Position
                                               of the U.S. Environmental Pndcclion Agency

-------
                                                                                £14
      amount of missing data to be. present during the hour because the short-term
      variability is not considered to be large enough to affect the monitoring results:

2. MISSING DATA PERIODS
The majority of the Subcommittee discussions on February ZO, 1991, concerned issues
associated with the presence of missing data periods in the continuous emissions
monitoring record.  Most of the discussions concerned monitoring of SO? emissions.
Title IV of the CAAA requires that in the absence of CEM data a unit is assumed to  be
operating in an uncontrolled manner unless information "satisfactory to the
Administrator* is presented.  The Subcommittee discussed various issues associated  with
missing data periods including the following: (1) appropriate time period for accounting
for missing data, (2) definition of a valid hour, (3) development of a threshold, and
{4} methods for filling in missing data. As the discussion of issues progressed, two
general perspectives on the role of continuous emissions monitoring evolved. In many
cases, views on specific CEM issues were related td Subcommittee members' general
perspectives on the role of emissions monitoring. These perspectives can be summarized
as follows:

Perspective I: The calculation of the annual tons of SOj emitted by a unit is
emphasized.  Those data necessary for the calculation of an annual average should be
collected, with less emphasis on the  measurement of emissions for smaller units of time
(e.g., aa hour).

Perspective XL The maintenance of a continuous record of emissions on a small time
scale (e.g., an hour) is emphasized.  Incentives which would minimize the amount of  "  -
missing data and provide rigorous estimates of hourly emissions are encouraged.

Appropriate Time Period for Accounting
The Subcommittee discussed the time units of concern for filling in missing data. For
CEMs, the Subcommittee agreed that an hour is the appropriate unit of time.  However,
the Subcommittee agreed that other time periods may be appropriate for evaluating
alternatives to CEMs.  The Subcommittee did not explicitly address which time periods
would be appropriate for various CEM alternatives.
                                                            Preliminary Draft for A It AC
                                                Discussion Docs Not Represent the Pusilion
                                              of the U.S. Environmental PruUrclion Agency

-------
                                                                               £14
                                                                                 J"
Definition of a Valid Hour
A key issue when establishing the number of missing data periods for CEMs is the
definition of a valid hour.  As the QA/QC and other requirements that define a valid hour
become more stringent, the number of missing data periods in the emissions record may

increase.  A summary, of the Subcommittee discussions follows:

•    Subpart Da design specification is 4 data points per hour (15 minute cycle).
     Capturing Z data points per hour is considered an exception.

•    Under Title IV, allowing Z data points per hour may be advantageous, especially for
     calibration periods. For many systems, daily calibration of the CEM can be
     accomplished in less than  30 minutes. Therefore, allowing Z data points for the
     calibration period would result in fewer missing data periods across the year.  In
     contrast,  requiring 4 data  points for a valid hour may result in better estimates of
     the hourly emissions.
           i       .
     The Subcommittee split on whether Z or 4 data points should be required for a valid
     hour.

*    The Subcommittee agreed that variation of measurements within an hour is an
     important consideration.  However, Subcommittee members were not aware of any
     studies on this issue.
Development of a Threshold

Several members of the Subcommittee thought that the use of an incentive- based

approach would be appropriate to encourage a smaller amount of missing data (more

complete record of emissions monitoring). As the discussions evolved, the concept of a
threshold for the number of missing data periods in a year was developed as a means to

simplify the selection of methods for filling in missing data periods as well as to provide

an incentive-based approach. A summary of the Subcommittee discussions follows:

      Most Subcommittee members agreed that  some  threshold is required. Examples
      included a minimum requirement of 90% or 95% data capture based on  the total
      number of plant operating hours per year.  The Subcommittee did not reach an
     * agreement on the exact threshold value.

•     Most Subcommittee members agreed that  more  stringent methods for filling  in
      missing data periods may be required when data capture rates fall below the
      minimum threshold.

*     Most Subcommittee members agreed that  the source would be given more
      flexibility in the approaches used to fill in missing data above the minimum
      threshold value.
                                                            Preliminary Draft for ARAC
                                                Discussion Does Not Represent Ihc Position
                                              of the U.S. Environmental Protection Agency

-------
                                                                               £14
 •     A few Subcommittee members do not think a threshold is needed.

 Methods for Filling in Missing Data
 The Subcommittee discussed which techniques for filling in missing data are appropriate
 and under what conditions they would be used. The Subcommittee members considered
 statistical, parametric, and hardware methods for filling in missing data. A summary of
 the Subcommittee discussions follows:
 •     The Subcommittee members agreed that statistical methods look promising for
      filling in short-term data gaps.
 •     A suggestion was made to encourage the use of backup CEMs instead of statistical
      or parametric methods.
 •     The Subcommittee agreed that maximum uncontrolled emission rates (MUERs)
      must be addressed in the rule making, but should rarely be used in practice because
      of the large number of available methods for filling in missing data.
 *     The Subcommittee generally agreed that a number of methods could be appropriate
      depending upon the  length of the data gap and plant operating conditions.
 •     The Subcommittee agreed to further evaluate which methods could be appropriate
      tinder various operating conditions and data gap lengths.

 3.  UTILITY OF SUBPART Da REGULATIONS FOR TITLE IV
The topics discussed regarding the utility of Subpart Da regulations for Title IV emissions
monitoring included (1) procedures for the Relative Accuracy Test Audit (RATA) and
Relative Accuracy Audit  (RAA), in particular the use of instrumental versus- wet -•
chemistry measurement techniques for comparison with CEM data; (2) quality of the test
gases used in the Cylinder Gas Audit (CGA) and daily Calibration Drift Check (CDC);
(3) a linearity test as part of the CGA; (4) criteria defining "out-of-control" for the daily
CDC; and (5) Relative Accuracy (RA) criteria for the RATA and RAA.
                                                           Preliminary Draft for ABAC
                                               Discussion Docs Nol Represent the Position
                                             of the U.S. Environmental Protection Agency

-------
                                                                               E14
                                                                                  7
   TA Instrumental versus Wet Chemistry

Under Subpart Da, for the RATA and RAA, CEM measurements of emission
concentrations are compared to values measured using either instrumental or wet

chemistry techniques. It was suggested that the U.S. EPA consider limiting these test

methods to instrumental techniques only for Title IV.


Mr. Ralph Roberson (Science Applications International) presented a summary of findings

from the Utility Air Regulatory Group (UARG) on RATA results using both techniques
(wet chemistry and instrumental) versus only instrumental techniques (Document E1S).

Mr. Roberson stated that no distinct improvement (i.e., smaller values for RA) was

evident  for tests using only instrumental techniques. Mr. Roger Shigehera (Entropy

Environmentalists, Inc.)  also presented data on the consistency of CEM and instrumental

measurements in RATAs (Document E16); similar comparisons for wet chemistry

analyses were not available.


Several  potential benefits of limiting the  RATA and RAA tests to instrumental

techniques were identified:

     greater consistency among emissions units, simplifying data reporting requirements
     and inter-unit comparisons;

•    faster turnaround time, allowing for more rapid corrective action — results from
     instrumental analyses are  immediately available, while  results from wet chemistry
     analyses are typically delayed several days;

•    QA on both sampling and analysis, while wet chemistry  methods provide QA on only
     the analytical process; and

•    lower costs — costs for wet chemistry analyses are approximately three times
     those for instrumental methods, although it was observed that cost, by itself, would
     not be sufficient justification for restricting the RATA or RAA to only
     instrumental techniques.

Reasons cited for maintaining the current Subpart Da specifications, which allow either
instrumental or wet chemistry techniques, included the following:

*    greater flexibility, and

      possibility that instrumental techniques and CEMs have similar inherent biases,
      although no data were presented to support or refute this hypothesis.
                                                           Preliminary Draft f»r AKAC
                                                Discussion Does Not Represent the Position
                                              of the U.S. Environmental Protection Agency

-------
                                                                                £14
In general, it was decided that more analyses were needed to (1) compare CEM,
instrumental, and wet chemistry results over a range of conditions (e.g., high and low
concentrations), (2) distinguish between measurement accuracy and precision, and (3)
examine the available round-robin test data collected in earlier studies on the accuracy
and precision of wet chemistry and instrumental analyses.
Quality of CCA and CDC Test Ca
General agreement was reached on the following:
•     The specifications for commercial gases under Protocol 1 should be tightened to
      provide reliable reference gases for the CGA.
•     The U.S. EPA should implement auditing of gas manufacturers. Results from these
      audits should be reported regularly but not used for formal vendor certification.
*     Protocol 1 gases should be used for both the CGA and the daily CDCs.  By
      upgrading  the test gases used for the CDC, CEM data will be certified on a daily
      basis.

Prior to implementing the requirement for Protocol 1 gases for the daily CDC, the U.S.
EPA should confirm that the required daily use of Protocol 1 gases will not result in
excessive added costs relative to the benefits gained.  Abo, it was suggested that the
U.S. EPA consider allowing triplicate measurements of cylinder gas with certified
methods on a regular basis, to confirm gas concentrations, rather  than an absolute
requirement to use Protocol 1 gases  for both the CGA and CDC.

Linearity Test in the CGA                                  _      -            -
Emission measurements must be accurate over the full concentration range to achieve
the objectives of Title IV. In addition, some units may experience relatively large
variations in emission concentrations (e.g., units with scrubbers).  At present, the CGA
within Subpart Da requires testing with only two reference gas concentrations. It was
suggested that the U.S.  EPA consider increasing  this requirement  to three or four gas
concentrations in order  to confirm the linearity of the CEM response over the full
concentration range of interest.

The Subcommittee agreed that the concept of a linearity check within the CGA was
sound and appropriate, although further work was needed to define the exact

                                         8
                                                            Preliminary Draft for ARAC
                                                Discussion Docs Nut Represent the Position
                                              oflhu U.S. (Environmental Protect ion Agency

-------
                                                                                £14
specifications and protocols for such a test.  The major conclusions from these
discussions were as follows:
•    The reference gas concentrations used for the CGA must cover the full
     concentration range of interest.
*    A 3- or 4-point test  would be appropriate for units that may experience  large
     variations in emission concentrations (e.g.* many units with scrubbers).
     A 2-point test may be adequate, however, for units with smaller emission
     concentration ranges.
»    The Subcommittee should further investigate the utility of the 3-point
     calibration/error test proposed by the U.S. EPA during earlier regulatory actions
     for Subpart Da.

Out-of-Control Criteria for Daily CDC
It was noted that the term 'out of control" (OOC) used in Subpart Da was potentially
misleading, and perhaps should be changed to "outside the allowable range.*

Under Subpart Da, a CEM is considered OOC if the drift detected during the daily CDC
exceeds
•    4 times the system specification (2.5% of span for SO2 and NOX) on any given day,
or
•    Z times the system specification over 4 consecutive days.

In the first instance, the CEM data  are considered invalid (i.e., treated as missing data)
retrospectively from the time of the previous valid CDC (generally 24 hours earlier), and
prospectively, up until the problem  is corrected and the CEM is demonstrated to be back
in compliance.  In the second instance, where the CEM exceeds 2  times the
specifications over 4 consecutive days, the CEM data are considered only prospectively
invalid, from the time that the OOC is detected until it is demonstrated that  the
problem has been corrected.

It was proposed that the U.S. EPA consider tightening the criteria for OOC to make
them similar to the current OOC specifications used by the State of Pennsylvania. In
Pennsylvania, a CEM is considered OOC if the daily drift exceeds 2 times the system
specifications on any given day. Data are considered invalid retrospectively (since the
                                          9
                                                            Preliminary Draft for ARAC
                                                Discussion Docs Not Represent lh« Position
                                               of the U.S. Environmental Prelection Agency

-------
                                                                                £14
                                                                               JO
preceding test) and prospectively (until the problem is corrected).  No separate OOC
criterion is defined for drift over several consecutive days.

Data on CEM availability in Pennsylvania, presented by Mr. Joe Nazzaro (PaDER), are
similar to those for the United States as a whole, as presented at the January
Subcommittee meeting by the Utility Air Regulatory Group (UARG) and the U.S. EPA.
It was suggested, therefore, that the tighter OOC specifications in Pennsylvania result
primarily in greater attention to CEM maintenance and QC, rather than a measurable
increase in the quantity of invalid (missing) data. Specific analyses in support of this
conclusion were not available, however.

The Subcommittee was divided on the need for tighter regulations and the degree to
which the OOC criteria should be tightened for the daily CDC. Additional information
was requested on the following items:
•     the frequency with which OOC conditions result from (a) lack of adequate CEM
      maintenance or QC versus (b) fundamental equipment limitations and variability;
*     how many additional missing data periods would result if the criteria for OOC
      were tightened;
*     how often and under what  circumstances basic equipment or facility
      characteristics would limit the achievable criteria for valid data; and
•     the ultimate influence of tighter OOC regulations on the accuracy of emission
      estimates  for Title IV.

The Subcommittee agreed, in principle, that the OOC criteria and regulations should be
written to reflect the achievable system accuracy and precision in most units. Units or
systems which routinely cannot achieve these tighter limits, because of their basic
equipment or facility characteristics, should be treated as exemptions or separate
classes of units. The burden of proof for such exemptions would be on the facility
operator, although the U.S. EPA would identify general conditions under which these
exemptions apply.  Examples of  possible exemptions proposed include (1) units with low
emissions, (2) units that must be retrofit with CEMs, and (3) units with wet scrubbers. In
addition, the regulations should recognize natural constraints, such as inclement
weather, that may on occasion hinder the completion of the daily CDC.
                                         10
                                                            Preliminary Draft for AHAC
                                                Discussion Docs Not Represent the I'usilion
                                               of Ihe U.S. Environmental Protection Agency

-------
  RA Criteria for RATA and RAA
  Given the proposed requirements to use Protocol 1 gases for the daily CDC for Title TV,
  failure of the quarterly audits and RATA was considered unlikely, occurring rarely if at
  all. Most Subcommittee members proposed that CEM data be considered invalid
  prospectively only, following the failure of a quarterly audit (i.e., exceedance of the
  specified RA criteria for the RATA and RAA or accuracy  criteria for the CGA). Some
  members, however, also proposed retrospective data invalidation, deleting that portion
  of the previous quarter's data for which data errors are suspected based on an analysis of
  the daily CDC results (using protocols for defining missing data similar to those
  currently used in Pennsylvania).

  Further discussions on specific RA pass/fail criteria were  postponed until the March
  meeting.  Additional data are needed on (1) the frequency  with which CEMs would fail if
  the RA criteria were tightened (see Document E15), (2)  the potential effects of
  tightening the OOC criteria for the daily CDCs regarding  the frequency of failure during
  quarterly audits, and (3) the degree to which RATA or RAA failures have resulted from
^•^
"" roblems with data accuracy versus precision.

  4. COz MONITORING
  The Subcommittee agreed that the basic monitoring requirements for CO£ should be
  based on combustion  equations. Further work may be warranted, however, to modify the
  equations to account for the effects of  some combustion technologies and control
  devices on CO? emissions.  It is expected  that these modifications would be relatively
  minor.  It was suggested that the U.S. EPA conduct additional analyses comparing CO?
  estimates from combustion equations and available CEM data.

  In addition, it  was generally agreed that the regulations should be written to encourage
  the use of CO? CEMs, where possible, both for diluent monitoring and to provide CO?
  emission estimates as required by Title vm. Concerns were expressed  by some
  Subcommittee members that (1) some CEMs use only O; as the diluent, (2) O? monitoring
  may be more accurate and/or reliable than CO<> monitoring in at least some types of
  systems, and (c) COz levels can be directly computed from CEM data on O?.  The
                                          11
                                                             Preliminary Draft Air AitAC
                                                 Discussion Docs Not Represent I he Position
                                                of the U.S. Environmental Prelection Agency

-------
                                                                              £14
Subcommittee requested that members compile additional information on the relative

merits of Oz versus C0£ for diluent monitoring.


5.   SUBCOMMITTEE ASSIGNMENTS FOR THE MARCH MEETING

Requests were made for the following papers/information:


U.S. EPA

•    Strawman paper on recordkeeping and reporting of CEM data

*    Overview of the available demonstrated alternative(s) to CEMs

•    Specific proposal on procedures for 3- or 4-point linearity test for CGA


UARG
•    Paper on alternative methods, concentrating on alternatives to CEM measurements
     of emission concentrations for the March meeting, with a follow-up paper on
     alternatives to CEM  flow measurements planned for the April meeting.

     Summary of results from continuing work on statistical methods for estimating
     missing data


Pennsylvania OER
•    Summary paper on Pennsylvania's approach to alternative methods


Ohio Edison
*    Proposal for exemptions to the OOC criteria for the CDC for low-emitting units,
     including background data and analyses in support of the proposed special
     exemption criteria


Research-Cot trell Companies
*    Background information on CEMs:  (a) Oz versus COj diluent monitoring and
     (b) monitor accuracy and precision capabilities for SOj and NOX

The next meeting is scheduled for March 20-22, 1991. Thus, the above papers must be
distributed no later than March 13. Richard Poirot has already distributed his written

summary of proposed uses for short-term CEM data, as requested at the prior

Subcommittee meeting (Document El 8).
                                        12
                                                           Preliminary Draft for AllAC
                                               Discussion Docs Not Represent the Position
                                              of the U.S. Environmental Protection Agency

-------
                                                                               EH
  e Subcommittee meeting adjourned at 12:15 on February Zl, 1991. Nancy Wrona,

acting Subcommittee Chair, presented a summary of Subcommittee discussions to ,the

full ARAC Committee on February 22 (Document El7).
                                         13
                                                           Preliminary Draft for AKAC
                                                Discussion Docs Not Represent the Position
                                              of the U.S. Environmental Protection Agency

-------
M«mo—ARAC Continuous
      Monitoring Subcommittee
February 26* 1991
f age 2


all tests) and omit those 11 results  (6 — SO, and 5 — NO,)  for which
relative accuracies cannot be r•computed.  However,  we do not believe that
th* distribution of relative accuracy results would  change significantly
if this were don*.

Sine* intermediate value* are net reported for  eix SO, and five NO, testa,
we could not compute percent mean difference for these 11 tests.  This
explains why we plot 98 SO, relative accuracies  (based on all tests)  and
only 92 SO,  percent differences (based on all tests).  Likewise, we have
54 NO, relative  accuracies (based en all tests)  and only 49 HO, percent
differences (based on all tests).

If you have any questions concerning  this aemorandua,  please do not
hesitate to call.
RLR/ehw

Enclosures

-------
 Systems  Applications
       International
                                                   4600 Marriott Dr.. Suita 42a Raleigh. NC 27112
                                                   919-782-1033  Facsimile 919-782-1716  ____
                                                   A Division of Qamem International Cocpoiaiion
                                                                Health
                                HgMORAMPPK
FROM i

DAZE:

SUBJECI:
            ARAC Continuous Monitoring Subcommittee

            Ralph L. Roberson, P. I.

            February 26, 1991

            Tra»smittal at Presentation Materials
At your request, we are submitting copies of the  graphs presented at the
February 20, 1991 ABAC Subcommittee meeting.  The graph* are based on data
presented in Appendix B of the Entropy Report.1

The first four graphs examine relative accuracy for SOj  and NO, monitors.
Two graphs represent all relative accuracy tests,  and two represent those
based on instrumental test methods (i.e., EPA Methods 6C and 7E).
Relative accuracy Lm computed by adding the absolute value of the reported
mean difference (|d|) to the confidence coefficient (CC) and dividing the
sum by the average reference method (RM) value.

The second four graphs examine only the mean difference (i.e., confidence
coefficient is not included) for SO, and NO. monitors.   Again, two graphs
are for all tests, and two are for instrumental tests.  Mean difference is
computed by dividing the absolute value of the reported mean difference
(I "31) by the average reference method (RM) value.

We provide one additional note regarding our handling of the data
presented in Appendix B of the Entropy Report.  Wherever possible, we plot
relative accuracies based on "recalculated results" instead of "reported
results".  That is, we recalculate relative accuracies by adding each
reported mean difference to its respective confidence coefficient and
dividing the sum by the reported reference method value.  However,
Appendix B lists six SO? and five NO,  relative accuracy test results for
which neither the test method nor intermediate values (e.g.,  RM, CC, etc.)
are provided.  Since intermediate values are not  available,  we could not
recalculate relative accuracy result* for 11 of the tests.  Thus, for
these 11 results (6 — SO, and 5 — NO,) we plot reported results instead
of recalculated results.  If the Subcommittee believes it would be
helpful, we can regraph the SO, and NO, relative accuracy results (based on
1   Entropy Environmentalists, Inc., Accuracy and Reliability of CEMS at
    Subtaart Da tEleetrie Otilitvj Facilities, prepared for EPA under
    Contract No.  68-02-4550, Work Assignment 106A, March 1990.

-------

-------
 X
O
 o
 O
 O
    CO
 P w
 oS<
 QC
 o
 c
 CD
 ZJ
 cr
 CD
                            Aouanbejj

-------

-------
 X
O
 o
 o  Si
   .5
   "c
    
-------
O
CO


 O
 O
 c
 0

5f

 S<
 Q)
 Q

-------
CJ
O
CO

 O
 O  co
   _
-^   CO

5  §
 0 -N-

   £


 ^c,
 CD
 O
                           Aouanbejj

-------
CM

O

CO
 O
 O
 O
     CO


 P  OT
 o<
DC
 o
 c
 0

 O"
               c
               o
               o
             o
             o
              £
              "3
            00

            O)
            0)

            l£>

            O)
            GO
             i
             vo
            to

            «
            CM
                                                             cs
                                                                 o
                                                                 a
                                                                 •o
                                                                 o
                                                                 (0
                                                                 o
                                                                 o

                                                             iq <
                                                             CNJ  O
                                                             1-  >
                                                                 Q>
                                                                cc
                                                             to
                                                             c\
                                   O
                                   CM
                                Aouanbajj

-------
1
r



-U
iw
>*
^
W
o
r:
*--
«r
^
CO
'/>
8
i
0
TD
X
D
05
O)
2
*->
z






























*




i





i







i

• + *



i
- •*
•H
K • V

• 4-^B-
f*


IX
r

•
•




1 1



m

i + ^!
hX *

/A ••
+ *«
-^
/K
• +
K.
.



1^ M/
^r ^R


'

^w
/m|i\j/





If ^y •
1 7n
X

f 3K


* „






• -f *
•

I-HK




1










•


•X






II \|/
^r ^C











in
d
CO
"d

.in
d
i ^^*
^s J
M* *
O

in
d
^t
"d

m
-CO
d
CO
V *
o
in
-CM
d

CM
q

d
                                                  CO
                                                 T3

                                                  O
                                                  CD
                                                  
-------

 x
o
 0
 o
 c
o
 0)
 o

-------
 x

O
 o
 O en

 SI
Q §

 s
 CD

 O
                   Aouanbejj

-------
o
CD

O

"55
 CO

CO

LU
O

 02

*x
 o
CO
   CD
     *
   O
                    •+*
                                *
                                  • nix
                                  • I/TS
                                 ^ 1\1/
                                 • 1 A
                             n
                                                *
                    CVJ
                    o
C\j

9
o
CD

CD
                                                                 -CO
                                                                 -1C
                                                                     CD
                                                                     "O
                                                                     O
                                                                 CO g


                                                                     CD
                                                                -CM
CO
CD
                       nigujuu/qi

-------
o
CO

•D
O


"5
 C/D
o
 I


s
"x
 o

5

j5


CO















1
1








• 4





1







*





TK







1







1





1 +




:


1 i


I
I

\^
/TS»







• *


•
•H

,


•

,
:


I

1
X
(



•
•+*
| y ^/
tlK






I





HK
(
*


















H

f

^









•










i




X








1






\i/









5K
























^


























N-
O
CO
"d

in
"d
_^r
0
CO
•• •
0
CM
"d
^ *
o
-<->
co N, co in Tf co CM
o o q q q q q q
d d d d d d d d
                             OT-CVJCO  ^r in co N co en
                                OOOOOOOOO
                                                               CD

                                                              •E
                                                                  "
                                        ••

                                OOOOOO
                                 I   I   I   I   I   I
         O
O O O  '
                        n}giuiu/q| '

-------

-------
 CONTINUOUS EMISSIONS MONITORING
           SUBCOMMITTEE

 UTILITY OF SUBPART Da REGULATIONS
             FOR TITLE IV
MAJOR TOPICS:


   1. Subpart Da QA/QC Methods

   2. Definition of Invalid Data —*• Missing Data

-------
GENERAL CONSENSUS TO MAINTAIN
    4 BASIC QA/QC COMPONENTS
          IN SUBPART Da
         1. Calibration Drift Tests

         2. Cylinder Gas Audit

         3. Relative Accuracy Audit

         4. Relative Accuracy Test Audit

-------
     CALIBRATION  DRIFT (CD) TEST
Frequency:
Purpose:
Approach:
Daily
Are CEM readings "wavering" over time?
Challenge CEM with low and high
concentration test gases

-------
       CYLINDER  GAS AUDIT (CGA)
Frequency:
Purpose:
Approach:
1 / quarter, 3 quarters per year
Are CEM measurements accurate?
Challenge CEM with known
concentrations of reference gases

-------
    RELATIVE ACCURACY AUDIT (RAA)
Frequency:
Purpose:
Approach:
1 / quarter, 3 quarters per year
Alternative to CGA

Are CEM measurements accurate
relative to other alternative measurement
methods?

Compare CEM emissions measurements
to values obtained using alternative
measurement techniques (3 replicates)

-------
RELATIVE ACCURACY TEST AUDIT (RATA)
Frequency:
Purpose:
Approach:
1 / year
Same as RAA, but more extensive
Same as RAA
9 replicates, rather than 3

-------

REASONS FOR MODIFYING SUBPART Da
o  Do the objectives of Subpart Da match those
   for Title IV?

       No flow specifications

       Interested in wider range of emissions
o  Is Subpart Da contemporary given today's
   CEM technology?

-------
        REFERENCE  GAS  QUALITY
ISSUE:   Is test gas quality sufficient to ensure accurate
         CEM data?
o   Quality of test gas critical for accurate data
o   Many test gases on the market not sufficiently reliable
o   Consensus to tighten specifications for Protocol 1
    gases, including EPA manufacturer audits
o   Recommendation to use Protocol 1 gases for daily
    CD test as well  as quarterly CGA
o   Accuracy of CEM data will be certified daily.

-------
   REFERENCE  GAS  CONCENTRATIONS
ISSUE:   Because OEMs-must accurately measure
         emissions over a wide range under Title IV,
         should the number of points checked in the CGA
         be increased from 2 (Subpart Da) to 3 or 4?
    General agreement that this is a legitimate issue
    for Title IV regulation

    Of particular concern for units with scrubbers which
    could experience large variations in emissions

    Further work needed by EPA to define specific
    protocols for test and pass/fail criteria.

-------
                TEST  CRITERIA
ISSUE:  Should the test criteria be tightened to encourage
         better maintenance and higner data accuracy?
o    Discussion of possible tightening of relative accuracy
     (RA) specifications postponed until next month
o    Discussion focused on daily CD test and definition of
     out-of-control (OOC) periods (i.e.,  data invalid).
    .   Under Da, CEM is OOC if drift exceeds:
         4 X spec for a given day
         2 X spec over 4 consecutive days
o    Proposal to tighten regulations similar to
     Pennsylvania standard
       Pennsylvania data availability similar to nationwide
p    Subcommittee split on degree to which regulations
     need to be tightened.
o    EPA regulations should  consider special categories
     of sources, such as
         Units with low emissions
         CEM-retrofit units
         Units with wet scrubbers

-------
             C02  MONITORING
ISSUE:  Should continuous emissions monitoring for
        COa be required?
o   Minimum requirement for C02 monitoring:
    combustion equations
o   Rulemaking could encourage use of COa CEMs.
o   Questions remain regarding
        Market availability of COa CEMs
        Need for some modification of combustion
        equations

-------
 CONTINUOUS EMISSIONS  MONITORING
            SUBCOMMITTEE
        MISSING DATA PERIODS
MAJOR TOPICS:
    1.  Appropriate Time Period for Accounting
    2.  Definition of a Valid Hour
    3.  Development of a Threshold
    4.  Methods for Filling In Missing Data

-------
PERSPECTIVES  ON ROLE OF MONITORING
Perspective 1:  Emphasis on calculation of annual tons
             of S02 emitted
Perspective II: Emphasis on maintaining continuous
             record of emissions
Implications:

 o   Selection of methods for filling in missing data

 o   Quality assurance/quality control (QA/QC)
     procedures

-------
  APPROPRIATE TIME PERIOD
        FOR ACCOUNTING
o  For CEMs an hour is appropriate.
o  Other time periods may be appropriate for
   evaluating alternatives to CtMs.
   Subcommittee did not explicitly address
   this issue.

-------
       DEFINITION OF  A VALID HOUR
ISSUE:   What constitutes a valid hour?
         More stringent requirements may result in
         more missing data periods.


 0  Subpart Da design specification is 4 data points
    per hour. Capturing 2 data points per hour is
    considered an exception.

 o  Subcommittee split on whether 2 or 4 data points
    should be required for a valid hour.

 o  Allowing 2 data points per hour may be
    advantageous, especially for calibration periods.

 o  Variation of measurements within an hour is an
    important consideration. However, Subcommittee
    members not aware of  any studies on this issue.

-------
                  THRESHOLDS
ISSUE:   Shquid thresholds be established for
         distinguishing the types of methods used
         to estimate missing data?
 o  Several members thought that the use of an
    incentive-based approach would be appropriate to
    encourage a smaller amount of missing data (more
    complete record of emissions monitoring).

 o  Most Subcommittee members agreed that some
    threshold is required.  Examples included minimum
    of 90% or 95% capture (total number of plant
    operating hours per year). Agreement was not
    reached on the exact threshold value.

 o  More stringent methods may be required to fill in
    missing data when capture rates fail below the
    minimum threshold.

 o  Unit would be given more flexibility in approaches
    used to fill in missing data above minimum value.

 o  A few Subcommittee members did not think a
    threshold is needed.

-------
METHODS FOR FILLING  IN MISSING  DATA
ISSUE:   Which techniques for filling in missing data
         are appropriate, and under what conditions
         would they be used.
    Subcommittee considered statistical, parametric,
    and hardware methods for filling in missing data.

    A suggestion was made to enc9urage the use of
    backup OEMs instead of statistical or parametric
    methods.

    Subcommittee agreed that MUER must be addressed
    in rulemaking, but should rarely be used in practice.

    Subcommittee Generally agreed that a number of
    methods could Be appropriate depending upon length
    of data gap and plant operating conditions.

    Statistical methods look promising for short-term
    data gaps.

    Subcommittee agreed to further evaluate which
    methods could be appropriate.

-------

-------
      State of Vermont
                                                  AGENCY OF NATURAL RESOURCES
                                                         103 South Main Street
                                                       Waterbury, Vermont 05676

                                                 Department of Environmental Conservation
                             MEMORANDUM
TO:        Members  ARAC   CEM Subcommittee

FROM:      Richard L. Poirot

SUBJECT:   Time Resolution for Reporting CEM Data

DATE:   •   February 19, 1991
      CEM reporting and recordkeeping are scheduled for discussion at our March
meeting. The appropriate time resolution for data reporting is an important aspect of
this issue, which could potentially affect other ongoing discussions regarding missing
data and alternative methods.

      The range between possible CEM output frequency, and the annual basis on
which  SO? totals (and  NOX rates)  are  assessed,  spans some 5 to 6 orders  of
magnitude*.  A  reporting requirement  at the shortest end of this time continuum
(seconds or minutes) would pose an unreasonable burden on affected sources and EPA
record keepers, would quickly result in an unwieldy data base, and would be difficult
to justify in terms of any potential use of the data.

      On the other hand, comparison with annual SO? allowances or NOX rates does
not constitute the exclusive purpose of the CEM requirement. If this were the case.
there are more cost-efficient means of determining annual emissions totals and rates.
The  CO2 monitoring requirement illustrates this point, as only an annual emissions
total is required, and the Conference Committee Explanatory Statement makes it quite
clear that CEMs are not required for the purpose of deriving an annual total.  Also,
the  CEM opacity  monitoring requirement has no direct  relevance to any annual
allowance consumption; though  it does not  seem logical that opacity should be
monitored, but  not reported.
              Regional Offices • Barre/Essex Jct./Ptttsford/N. Spnngfield/St. Johnsbury

-------
      My personal preference would be for reporting of CEM (or alternative) data as
hourly averages, submitted on a quarterly or monthly basis.  In the long run,  I
believe  this will minimize the overall  effort of affected sources and EPA record
keepers, and maximize the usefulness of the resulting database.  It would also
EPA to conduct an effective and timely external quality assurance program,
though not specified in detail in the statute . would nonetheless be highly desirable to
assure the integrity of the emissions reductions and allowance trading program.

      I believe  that  hourly emissions reporting  is thoroughly justified by the
statutory language in Title IV alone , and is further supported by the additional needs
for such data in implementing the (state-run) Title  V permitting program, and
particularly in fulfilling the many tracking or assessment responsibilities scattered
elsewhere throughout the Act .


Title IV  Justification
      The intended purpose of CEM equipment is unambiguously defined in section
402(7) as "to sample, analyze,  measure, and  provide on a continuous  basis a
permanent record of emissions and flow. . . " Given the well-known capabilities of CEM
technology, it is not reasonable to assume that a continuous . permanent record be
composed of a series  of annual or seasonal averages.  Nor is it reasonable to assume
that this continuous, permanent record was intended to reside in  a piecemeal and
inaccessible form in the private books of individual, affected sources.

      The Title IV definition goes on to specify several exemplary units, including
"pounds per hour", leaving no question that Congress expected EPA to maintain an
emissions record of relatively short-term time  resolution.   In any  event,  this
definition leaves no doubt that if hourly emissions reporting was not precisely what*
Congress had in mind, it  is certainly well within the range of EPA's discretionary
authority.


EPA Assessment Responsibilities (External to the Title IV Allowance Program)

      If we accept that comparison with annual SO 3 allowances does not constitute the
exclusive purpose of the CEM  requirement, and /"or that EPA at least maintains the
justifiable option of requiring  short -term  (hourly) emissions data, then it becomes
important to view Title IV in the context of other Clean Air Act requirements which
require consideration of emission quantities.  If EPA fails to exercise its option of
requiring short-term data, will it be able to adequately fulfill these other important
assessment responsibilities?

      It is instructive to consider here that the stated purpose of Title IV is not
simply to achieve a 10 million ton SO2 reduction, but  rather "to reduce the adverse
effects of acid deposition. . . " The 10 million ton goal is not a scientifically determined
figure,  designed to precisely eliminate all effects of sulfur and nitrogen emissions.
In truth, we don't know exactly what the benefits (or costs) of Title IV will be.

-------
Conceivably, we may get part way through the control program, and find a scientific
and  political consensus that  the last few  million  tons  are not worth  the cost.
Conversely, it may become apparent that 10 million tons are not quite enough, or that
the benefits could be substantially improved by  slight alterations in the seasonal
timing or locations of emissions reductions.

      By comparison to previous Clean Air Act amendments,  the 1990 Act includes a
proportionally large number of tracking and assessment responsibilities (research.
monitoring, inventorying, modeling,  source  attribution,  strategy development,
evaluation, reporting, etc.), many intended specifically to determine the changes in
air quality and effects which result from required emissions reductions.  The number
and specificity of these  required assessment activities is particularly overwhelming
in contrast to EPA's projected budget resources, and will require careful coordination
and efficient utilization of all available data resources.

      Attachment A contains a listing of various required Clean Air Act assessment
responsibilities which would benefit from access to hourly emissions data. Many of
these require  a direct assessment of the  effects  of emissions reductions achieved
under Title IV and other Clean Air Act control programs. Others may not refer back
to Title IV directly, but would nonetheless require (or benefit from) estimates (highly
uncertain)  or measurements of hourly emissions of 862. NOX and/or opacity.

      As a general rule, virtually every evaluation involving oxidant photochemistry
requires consideration of short-term emission rates.  This would include  any
assessment of ozone non-attainment (itself a 1 hour standard), and also evaluations
of sulfur transformation. transport and deposition (which are strongly influenced by
oxidant chemistry and other short-term variables).  Short-term  emissions data are
also obviously relevant to any effects which tend to occur on a short-term basis, like
episodic acidification of surface waters, visibility impairment and acute health or
ecological effects.

      The list in Attachment A was developed using the "EPA Shoe" Test. That is.
I tried to put myself in EPA's shoes, assumed  that it would be my responsibility to
conduct the required assessment, and asked if I would wish to have access to hourly
emissions data. This list includes those assessments for which the answer was yes.


Title V  Permit Requirements


      As a practical matter, it appears likely that the (state-run) Title V permit
program will also inevitably require emissions data expressed for short averaging
times, and that the necessary coordination between the Title IV  and Title V permit
processes would provide additional justification for hourly data reporting under Title
IV. Many currently operative state permit programs require hourly CEM emissions
reporting and are likely to continue or expand upon these requirements in the future.
The affected sources under Title IV and Title V are the same, and Title IV permits will
need to  be consistent with Title V requirements.

-------
      I would recommend, however; that the necessary and important Title IV/Title
V coordination be treated'as a separate issue in our discussions, for several reasons.
First,  this is a large, complex  issue for which the data reporting requirements
represent a  relatively small part.  Second,  I believe that an hourly  reporting
requirement is fully justified without consideration of additional Title V information
needs.  Finally, such discussion is likely to lead quickly to  the more controversial
questions of whether emissions data collected pursuant to Title IV should or could be
used for determining compliance under Title V.  This is an important question, but not
one for which the ARAC Committee is likely to reach a consensus of any use to EPA.
In any event,  the possibility that short term data, if available,  might be used
"inappropriately"  for  compliance  determination,  would not provide a defensible
justification for disguising the data by reporting only long-term averages.


RLP:ljl

-------
                                  ATTACHMENT A
             8PA ASSESSMENT RESPONSIBILITIES WHICH WOULD REQUIRE OR
            BENEFIT  FROM  ACCESS  TO SHORT-TERM (HOURLY)  EMISSIONS  DATA
             Attainment  and Maintenance
o   182(c}(l) Enhanced  monitoring in  "Serious"  (.160-.ISC  ppm)  or worse  ozone
    nor.- attainment  areas;  requires  states  (following EPA  rules)  to  "improve
    monitoring of emissions  of  oxides of  nitrogen."

o   182(1}(1)(B)  Each  state in a "multi-State ozone non-attainment area"  shall
    "use  photochemical  grid modeling"  or  other  equally  effective  analytical
    method.

c   184(d)    For  interstate  "Ozone  Transport  Regions",  EPA  must  promulgate
    criteria for determining interstate source/receptor contributions,  including
 •   "best availacle air quality monitoring  and modeling."

c   18S(B)   E?A/N'AS  study on  "role  of  ozone precursors  in tropospr.eric  ozor.e
    formation ar.d control."   Study shall  examine roles of nox  and  voc emissions
    redaction,   extent   to   which  NOX   reductions   rr.ay '  contribute    to  achievement  of  attainment in different non-attainment
    areas  .  .  .  role of biogenic VOC  emissions, and basic  information  required
    for AQ models.  .Report shall  use "all available  information."

c   190  EPA to issue guidance  on  RAC.M and  3ACM  en various categories  of sources
    contributing  to   PM-1C ' non-attainment   "taking   into  account   emissions
    reductions achieved or expected to be achieved under Title  IV .  .  . ••
             Hazardous Air  Pollutants
o  301(k)  EPA Research/Monitoring  (in representative numcer of urban areas) en
   Hazardous  air  pollutant effects  from  area  sources,  including  acute and
   chronic  effects,  consideration of  atmospheric  transformations  which can
   elevate  public health  risk,  and  role of  such emissions  as  precursors to
   ozone and  acid aerosols.  Report  in  three years;  National Strategy in  five
   years   (strategy   shall  provide  for  ambient  monitoring  and  emissions
   modeling).

o  301(a)    EPA/NOAA assessment  of  atmospheric  deposition  of  hazardous air
   pollutants  to  Great  Lakes,  Lake Champlain,  Chesapeake  Bay  and  Coastal
   Waters,   including  assessment  of   "sources  and  deposition  rates  of air
   pollutants  (and  their atmospheric transformation  precursors)."   EPA report
   in three  years,  then biennially,  on above including sources of atmospheric
   deposition  to  above  waters,   and  description of   any  revisions  of the
   requirements,  standards  and  limitations  pursuant  to  this  Act
   necessary to assure protection of  human health  and environment."

-------

o  301(n)
-------
o  812   (Amending  312(a))    Formation  of  Council  on  Clean  Air  Compliance
   Analysis.   EPA/Council shall  conduce  "comprehensive analysis  of the  impact
   of  this Act  on the  public  health, economy  and environment  of  the  Unites
   States";  including limitations  on  SOX and  NOX; considering  costs  and all
   economic,  public health  and environmental benefits.   EPA  "to  assure that
   damage  to  human  health  and  environment  is  more  accurately  measured".
   EPA/Comnterce/Labor/Council  report  in  2 years  and every  2  years including
   future  projections.    Council  to review data  for use  in  such analysis and
   make  recommendations  for  use and methods to analyze  such data.

o  812   (Amending   312(b))     GAO  Report  (in  2  years,  then  annually)  on
   "incremental  human health and  environmental  benefits and incremental  costs
   beyond  current clean  air requirements  of the  new  control  strategies and
   technologies  required  by   this  Act"   including  effects  en   jobs,   energy
   security, competitiveness in international markets,  etc.

o  81S(a)  EPA  and affected states negotiate with Mexico  to  develop program to
   monitor and  improve  air.  quality in  U.S./Mexico  border regions.   Monitoring
   to  include  met data,  air   quality,  emissions,   "sufficient, to  the  extent
   necessary to  support  use   of state-of-the-art  mathematical  air  modeling
   analysis".   Project  emissions reductions  to attain  primary  and secondary
   NAAQS and  other air  quality goals.   "Include to  the extent possible, data
   from monitoring programs  undertaken  by other parties."

c  81S(b)  Negotiate to develop joint  remediation measures ir. border region.

o  815(c)  EPA annual reports on  progress toward riAAQS  in  border region.

c  816  Amending Subpart 2.  part  C, title  :.  169 3  (aj  through  
-------
Title IX  Research                                                  . •

o  901(b)  Amending 103(O through (f)

o  (c)  EPA program of research,  testing and methods development for sampling,
   measurement, monitoring, analysis and modeling including:

   1}   Considerations   of  individual   pollutants,   complex   mixtures,   and
        transformations.

   2)   National  monitoring  network  to  collect  data -(with—quar.tif.ication of..
        certainty) in  status and  trends  of  emissions,  deposition,  air quality,
        water quality, forest condition and visibility.

   3)   Develop  improved  methods  of  monitoring,   analysis  and   modeling to
        increase  understanding  of  sources  of  ozone  precursors,  formation,
        transport,  regional  influences,  trends  and  interactions  with other
        pollutants

             Improve ability to inventory emissions cf VOC and NOX

             Improve understanding of reaction mechanisms through which  natural
             and anthropogenic emissions react to form ozone and other oxidants

   4)   Submit periodic  reports  (5  years)  to assess effectiveness cf pollution
        control regulations and programs.

c  (e)   E?A/HOAA/FWS/Ag  research  to understand  short  and  Icr.g-terrn  caus
   effects  and  trends  in  ecosystem  damage  from air  pollution includi
   characterization of causes and effects sf  chrsnis and episodic exposures at:
   determination   cf   reversibility;   development   cf   improved    atmospheric
   dispersion  modeling and monitoring;  assessments  of short-term and long-c.erni
   ecological  effects  cf acid deposition  and other air  pollutants on  surface
   waters...

o  901(c)    Amending lCj(g) through  (k)                '  -

c  (g)     Improvements  in  non-regulatory  strategies  and   technologies  for
   preventing  or  reducing multiple air pollutants including, SOX,  NOX,  metals,
   PM-10, CO and CO2 from  stationary  sources  including fossil  fuel  power plants
   (shall be considered  for new and existing  facilities).

o  (i)   EPA report on coordination of 901  research with  other programs (in 2
   years, then every 4 years)

o  (j)   Continuation  of  NAPAP to  review research status,  identify  information
   needs,  assure coordination among  federal agencies  to "ensure  availability
   and  quality of  data   and  methodologies  needed  to  evaluate  the status and
   effectiveness  of  the  acid deposition  control  program."   Including research
   and monitoring of

         (i)  Continuous  emissions monitoring of acid deposition  precursors.

-------
        (ii) Maintain,  upgrade  and  apply  models,  such  as  RADM,  describing
             interactions of emissions/ atmosphere and dose/response.
                                       /
        (iii)  Costs,   benefits and  effectiveness  of  acid deposition  control
               program

        Report every 2 years to include:

        (1)  Actual and projected emissions and acid deposition trends

        (ii) Ambient   concentrations   of   acid   deposition   precursors   and
             transformation products

        (iii) status of effected ecosystems, including visibility

        (iv) Causes and effects of such deposition

        (v)  Occurrence and effects of episodic acidification

          i) Confidence level associated wish each conclusion
o  901 (g)    EPA  research,  monitoring  and  annual  reports  on  occurrence  and
   effects of  acid deposition west of  Mississippi  River,  including occurrence
   and effects on high elevations,  and utilizing predictive modeling techniques.

-------

-------
                                                            JO
                                                            E 19
Central and South West Services, Inc.
             OlM*: Ttiu 7S26W16*
   February  13,  1991
                                             Fax *
   To:       Larry Kertcher             (202)  252-0892
             David Hawkins              (202)  783-5917
             Henry Seal                 (201)  685-4120
             Jerry Golden               (615)  751-3561
             Robert J.  McWhorter        (216)  384-5791
             Daniel R.  Plunley          (518)  873-6675
             Richard L. Poirot          (802)  244-5141
             Nancy Wrona                (602)  257-6874
             Robert Bergstroa           (515)  432-7096


   As agreed to  at the  last ARAC Eaissions Monitoring sub-committee
   meeting in Washington, January 28,  1991,  I am presenting for your
   review and discussion an issue paper that merits consideration in
   the CEMS rule making process.  The  paper represents the  needs of
   a large population of phase II affected gas units.

   I would be happy to  speak to any guest ion? next week at  the
   February ARAC meetings,
   Yours truly,
   N.N. Dharmarajan
   (214) 754-1373
   (214) 754-1380  (fax)

   NND/»h
                      A Mamtoaf of tfta1 CMWV! Mid SouOt WMI Syvtwn
       Cfrml Pom* and Ugnt Compwy • Pu»c S«vc* Compwiy * Owanoma • Sou0M*mm Etaetne Powir Company
                      •  Tranio*. Inc. • VWwi Tnii UWittM Company

-------

-------
                  ISSUE PAPER

CONTINUOUS EMISSION MONITORING EXEMPTION
            FOR GAS-FIRED UTELTTIES
                   Presented to:
          Hie Acid Rain Advisory Committee

                   Prepared by:

                Radian Corporation
                 8501 Mo-Pac Blvd.
                 P. O. Box 201088
              Austin, Texas 78720-1088

                       for

         Central and South West Services, Inc.
           1616 Woodall Rodgers Freeway
               Dallas, Texas  75266
                 12 February 1991

-------

-------
                                     ISSUE PAPER
                  CONTINUOUS EMISSION MONITORING EXEMPTIONS
                               FOR GAS-FIRED UTILITIES
      ISSUE
                 Gas-fired utility stations that fire fuel oil less than 10 percent of the time
      should be exempted from the requirements of Section 412 of Title IV of the Clean Air
      Act Amendments of 1990 (CAAA) to install continuous emission monitors (CEMs) for
      the measurement of sulfur dioxide (SO}), volumetric flow rate, and opacity. Further-
      more, the requirement of Section 821 of Title Vm of the CAAA to monitor carbon
      dioxide (CCK) emissions should not be interpreted to mean that a GEM for CO2 is
      required.  Acceptable alternate methods for the measurement of these emissions are
*az   available and should be allowed.

      BACKGROUND
                  Section 412 of Tide IV of the CAAA requires that all sources subject to
      the title install and operate CEMs on each affected unit for the measurement of SO^
      volumetric flow rate, and opacity. Also, Section 821 of Title Vm of the CAAA requires
      that each affected source shall monitor CO2 emissions, although no specifications are
      made as •Ifct method of  monitoring.  Regulations for the CEMs must be issued not
      later th^Hfcl+en months after enactment of the CAAA (by 15 May 1992). Phase I
      affected m*,** defined by Section 404 of Title IV of the CAAA, must install and
      operate CEMs not later than 36 months after enactment of the CAAA (by 15 November
      1993). Phase H affected units, as defined by Section 405 of Title IV of the CAAA, must
      install and operate CEMs by 1 January 1995. CEMs are defined as the equipment used
      to sample, analyze, measure, and provide  (on a continuous basis) a permanent record of
      emissions and flow, expressed in pounds per million British thermal units (MM Btu),
                                           1

-------
 pounds per hour, or such other form as the EPA Administrator may prescribe by
 regulations. Section 412(a) allows for the provision of "any alternative monitoring system
 that is demonstrated as providing information with the same precision, reliability,
 accessibility, and timeliness as that provided by CEMs...".

 DISCUSSION

            Phase I affected units are primarily coal-fired units, some of which will
 benefit from the use of SO2 and volumetric flow rate monitors through the establishment
 of the SO2 Allowance Transfer Program as provided by Section 403 of Tide IV of the
 CAAA.  Exchange of  emission credits is expected to take place upon enactment of the
 Phase I regulations, 1  January 1995.

            All gas-fired units are included in the Phase n list These units, for the
                                                                                .>

 provide a reliable electrical supply, a majority of these units have expensive oil as a
 backup fuel. Oil firing is typically less than one percent in any one year and occurs only
 when gas supply is curtailed due to weather conditions.  The contribution of such gas-
 fired units to SO2 emission rates and opacity is  an insignificant factor compared to units
which are primarily coal- or oil-fired (1).

 SOT CQT tmd Volumetric Flow Rate
            Gas-fired units should be excluded from the requirement of installing
CEMs for continuous monitoring of SO» and the requirement to monitor CO2 emissions
should not be interpreted to mean that a CEM is required for this purpose. Analyses
providing the sulfur and carbon content of the fuel, whether the fuel is fuel gas or fuel
oil, will provide data of equal or better quality than that provided by CEMs.  NSPS
Subpan D already allows use  of fuel analysis to monitor SO2 emissions for gas- and
fired units. Volumetric flow rate will be needed to quantify SO: and CO2 emissions.

-------
 Fuel firtaf rate or unit load measurements can be substituted for a GEM for determining
 volumetric^** rate.  These analyses and measurements are generally already practiced
 at most gas-fired utility stations and would seem to satisfy the requirements of precision,
 reliability,  accessibility, and timeliness for alternative monitoring systems.

            The estimated cost for installing CEMs for SO» COj, and volumetric flow
 rate, measurement is $ 173,000 per unit This cost includes the CEMs, sample delivery
 and conditioning system, instrument housing, data logging or recording system, access
 ladders and platforms, and certification.  It does not include continuing costs for
 calibration, maintenance, and reporting.  EPA's National Utility Reference File  lists 622
 gas-fired units as of 1985 (2). An additional 161 units are planned according to the
 Utility Data Institute (3).  Therefore, installation of CEMs represent a capital expendi-
 ture of approximately $135,000,000 for these units.  These figures dearly demonstrate
 that any benefits associated with the requirement to install CEMs to monitor SO* CO*
 and volumetric flow rate on gas-fired units as opposed to using  alternative monitoring
 systems are far exceeded by the costs.

 Opacity

            Gas-fired units should be excluded from the  requirement of installing
 CEMs for continuous monitoring of opacity. There are no discernible visible emissions
 from units  fixing 100 percent natural gas.  When a unit is firing fuel oil, several  factors
 have aa jlhfMB opacity,  including sulfur and vanadium content of the fuel, upset
 coadit&olHMifc at sun-up), and nozzle plugging. Nozzle plugging can be addressed by
 a infiiliTiiaairi linn ( schedule and operator training.  Many gas-fired utilities  use the
 practice of test fixing fuel oil for several minutes on a monthly or bimonthly basis to
 ensure that the fuel ofl delivery, control, and firing system is operating property. This
practice keeps operators trained in oil firing procedures and oil fixing equipment
maintenance.  Visible emissions during start-up with fuel oil represent, at this time, an
unsolvable  problem.  However, this is classified as an "upset* situation which occurs very

-------
 infrequently and typically lasts less than 30 minutes. Opacity studies have indicated
 opacity diptds directly on the sulfur and vanadium content of the fuel oil (4).  Gen-
 erally, units that fire low sulfur (less than 0.7 percent), low vanadium (less than 50 ppm
 by weight) oil have opacities less than 15 percent Furthermore, the highest fraction of
 the visible emissions from firing fuel oil results from the condensation of sulfuric add
 (HjSO4), which generally occurs after the stack exit as the flue gas is mixed with ambient
 air.  A CEM measuring opacity in the stack will not detect this event

             An alternative monitoring system for the measurement of opacity could
 involve the services of a certified opacity reader during periods when fuel oil is  being
 fired, thereby providing the same level of precision, reliability, and accessibility  as a
 CEM. Frequency of measurement can be established by regulation to satisfy the level of
 timeliness.

             The estimated cost for installing CEMs for opacity measurement is
 5105,000 per unit if installed independently of other CEMs. This cost includes the
 CEMs, data logging or recording system, access ladders and platforms,  and certification.
 The total cost for instillation of* opacity monitors on 622 existing and 161 planned gas-
 fired units  is  approximately $85,000,000. These figures dearly demonstrate  that any
                                                               %.
 benefits associated with the requirement to install CEMs to monitor opacity on gas-fired
units as opposed.to using an alternative monitoring system are far exceeded by  the costs.
Therefore, gas-fired units should not be required to use CEMs to monitor opacity, since
data eta tginitably obtained by an alternative monitoring systems.

         **
REFERENCES

 1.          AP-42, Section 1.4, Table  1.4-1.  Uncontrolled Emission Factors for Natural
             Gas Combustion.
2.           National Utility Reference File,  EPA, November 1989.
3.           Utility Data Institute, Capital Expenditure File, July 1990.

-------
    Sulfate Formation in Oil-Fired Power Plant Plumes, Volume 1, Final
    Report, 1983.
T. -

-------

-------
                                                                   E 20
                               February 12, 1991
 TO:   Members of Emission* Monitoring Subcommittee,  ABAC
             Re:  Continuous  Emission Monitors  for Gas/Oil  Units  <10% Oil
Attached is a  paper prepared  to address  the rationale  and alternatives to
requiring continuous emission monitors on Sec. 403(h) units, i.e.  that  class of
gas/oil  units which  fire  less than 10 percent oil.

As was discussed  in  the last meeting of  the Emissions Monitoring  Subcommittee,
there are a number of reasons to  consider alternatives to  continuous emissions
monitors for this specific class of  units.   While the time constraints  in the
preparation of  this  document have kept us from providing all  the  details which
might be necessary in rulemaklng,  ve feel that the framework  and  concepts upon
which the Subcommittee can make a decision  are included.

If there are any  questions which  require further  information  or clarification,
we will be pleased to respond and provide that information  to  the  Subcommittee.
Thank you for your consideration  of  this issue.
J.BL. Smith
Houston Lighting & Power Co.
P. 0. Box 1700
Houston, Texas  77001
(713) 922-2190
                           Wade Stansell
                           Texas Utilities Electric Co.
                           2001 Bryan Tower, Room 2070
                           Dallas, Texas  7S201
                           (214) 812-4814
cc:
Lertcher • EPA
    tz - EPA
 Clauasen - EPA

-------

-------
                            EMISSION MONITORING APPROACHES
         Oil and Gas-fired Units Less Than 10  Percent Oil  Consumed
i.

       1.0 INTRODUCTION

           The requirement for Continuous Emissions Monitoring Systems
       (GEMS)  on all affected units is designed to allow determination.
       of the actual  tonnage of SO2 and  average NOx  emission  rate
       emitted during  any given year.  On some affected units the CEMS
       are  also used for  compliance  with  emission  limitations*
       Section 412  of the Clean Air Act Amendments of  1990  provides
       for monitoring  of opacity, SO* and NOx emissions and  volumetric
       flow from the stack to make the tonnage determination for SOg.
       Section 412  also makes provisions  for alternative  methods  of
       monitoring the  emissions from these units.  The purpose of this
       paper is  to  delineate alternative methods  that,  under certain
       conditions,  will provide the same tonnage  information for SO2
       and NOX emission rate and opacity as that provided by  CEMS for
       Gas/oil-fired  units firing less  than  10 percent oil.  These
       alternatives are appropriate since the Section 405 (h) units are
       a class of unit which has relatively small contributions of S02
       and NOx  due to the  fuel  use,  basic  design  and  operating
       capacity.  The  purpose of this paper is  not to provide all  of
       the details of the methods but only a rudimentary understanding
       of  principles behind these methods.
                 alternate monitoring methods  for SO2 are proposed
      which  depend upon the measurement of the fuel  sulfur  content
      and the fuel flow to individual units or groups of units.  The
      three  alternative methods are:

          1) INPUT .METHOD
             Periodic measurement of the fuel and sulfur input to
             the plant  or unit,

-------
    2) THROUGHPUT METHOD
       Continuous measurement of the fuel  flow to each unit
       and gross measurement of plant fuel sulfur content,
    3} CONTINUOUS METHOD
       Continuous measurement of fuel flow to each unit and
       batch or  continuous  measurement of the  sulfur con-
       tent of the fuel to each unit.

Each  of  these  methods has its  usefulness depending upon the
particular configuration of the oil supply and delivery system
and the method of fuel oil circulation within the system.

    Two  alternate  NOx monitoring  methods are  proposed which
depend upon 1)  accurate characterization  of the NOx emissions
over the load range and 2) characterization of the influence of
excess  oxygen  on  NOx emissions.    One  alternate  method  of
monitoring the  opacity during  periods when oil is  fired  is
proposed that  relies on the  currently accepted EPA standard
visual observation method.

    The following Sections provide insight into the rationale
for the potential use of  these alternative monitoring methods.
Section  2.0  provides a discussion  of the potential relative
accuracy requirements of the  alternate SO2 monitoring systems
in terms of quantifying the national SOg emission tonnage on a
yearly  basis.    The  rudiments  of  the  three   alternate  SO2
monitor^sjg methods are described and some of the configurations
       36f
for vbirfflfctfeftse may be used are listed.   Section 3.0 provides
similar^tiscussions  for alternate  NOx  monitoring methods.
Section  4.0  provides a  brief discussion  on   the  alternate
opacity monitoring method.   Section 5.0 provides conclusions
related to the  utilization of alternate monitoring methods for
gas/oil units firing  less than  10  percent oil.

-------
2.0 SULFUB OZZDS EMISSIONS

2.1 Accuracy Requirements

    Current CEMS requirements for Subpart Da units require that
each monitor be  certified according to specific criteria for
relative accuracy and  calibration  drift.   These requirements
have been  in effect  for more than  a decade.  The accuracy
criteria is  based upon  comparison  of the CEMS output  for a
specific emission specie  (NOX or S02) compared to a reference
method.  The allowable relative accuracy for certification of
a CEMS is permitted  to  be as high as ± 20 percent.  Consequent-
ly, a CEMS which was certified at a relative accuracy of + 20
percent could conceivably report S02 emissions which were at
least 20 percent above the actual emission level.

    It would be reasonable to assume that the CEMS certifica-
tion criteria for the  1990 Clean Air Act Amendments would be
nearly  identical to  those  for  units  required to  meet  the
Subpart  Da  New  Source  Performance  Standards  provisions.
Consequently, due to  the relative  accuracy specifications on
CEMS the potential will exist for inaccuracies (as high as ± 20
%) in the estimation of the total SO2 tonnage for the US boiler
population if all boilers utilized CEMS.   It is  not certain
what the exact inaccuracies would be since not all CEMS would
consistently have inaccuracies on the high or low side of the
reference method.  It is  certain, however,  that  the  average of
all reiafcjfcve accuracies for the US  boiler population would not
       *tr*
result VF^swro inaccuracies compared to the reference method.
For the sake of  the following discussion,  it is assumed that
the mean relative inaccuracy for the US boiler population is 10
percent.

    Based upon the EPRI  boiler database, it is estimated that
one third of the potential utility  fossil  fuel capacity  in the
                              3

-------
US* is comprised of gas-,  oil- or gas/oil-fired units.  Of this
one third potential generation, only a fraction of these burn
both gas and oil.  Furthermore, only a fraction of those that
burn both  gas and oil regularly  burn less than  10 percent.
Consequently, only a very small percentage of  the potential
total US generation  is represented by those units that burn
both gas and oil and burn less than 10 percent oil.

    Due to the price differential between  coal-fired generation
and gas and/or oil fired generation, coal-fired unit capacity
factors are  generally higher than for  gas/oil  fired units.
Most large coal-fired  units  are base loaded whereas gas/oil-
fired units are cycled and are generally used to  carry, the load
swings.  The net result of  this  is that gas/oil-fired units
would represent only a small  fraction (much less than 1/3) of
the  actual  total US  electric generation due to  their lower
capacity factor.

    Most gas/oil units are located in metropolitan  areas where
coal firing is less desirable.  In the large metropolitan areas
the  sulfur content of  fuel  oil is generally limited.  In the
Los Angeles area and some northeastern metropolitan areas the
allowable sulfur content  is much less than one percent  (approx.
0.5%).  In general,  the sulfur content  of  fuel oils used by
utilities is  less  than the  sulfur content for coals.  If one
assumes that the average sulfur content  for all units burning
oil is approximately 0.9  percent,  then the emission rate would
be 1.0 J|»JIOCBtu (18,000 Btu/lb oil). Using this assumed rate,
one can estimate the relative contribution of the gas/oil-fired
units  firing less than  10  percent  oil  to the total US SO2
tonnage under the most conservative conditions.

    The conservative estimate can be made by assuming capacity
factors, and average emission rates  for  all coal and gas/oil-
fired units.   In addition,  by assuming  that  all US gas/oil-

-------
fired unit* fir* oil at a  rate  less than 10 percent,  the SO2
contribution by these Section 405  (h) units  is maximized.  The
following Table provides assumptions for an extremely conserva-
tive estimate  of  the SOg  emissions  from gas/oil-fired units
firing less than 10 percent oil.
                          TABLE 2-1
                 ESTIMATE O7 SOj CONTRIBUTION
    ASSUMPTIONS
    FUEL TYPE                              COAL      GAS/OIL
    PERCENTAGE OF FUEL                     100       90/10
    CAPACITY, % of US Total                67        33
    AVERAGE ANNUAL CAPACITY. FACTOR, %      60        30
    SOj EMISSION RATE, Ib/MMBtU    .        0.6.       1.0

    802 CONTRIBUTION ESTIMATE

    PERCENTAGE CONTRIBUTION                96.0      4.0


The results presented in Table 2-1 are only for the purpose of

providing an upper  bound for the estimated SOg contribution.

The 4.0  percent contribution  by units burning less  than 10

percent  oil  is  extremely  conservative  for  the  following

reasons:


    1) Hot  all  of the non-coal  fired units  burn  both gas
       and oil, therefore, much  less  than 33 percent actu-
       ally fire both gas and  oil.

    2) OjUy a  nail percentage  of the unite that do fire
             gas and  oil fire  less  than 10  percent oil,
            •fore, the flection 405(h)  units represent only a
             small  fraction  of  the   actual  US  generation
           IB less than the 33  percent assumed above)*

    3) The emission rate is  low for coal and high for oil,
       therefore, the contribution for coal-fired units is
       actually much greater than indicated.


Neglecting the fact that the emission rates  are conservative.

-------
if it is assumed that  70  percent (high estimate), of the non-
coal fired units fire gas  and oil and 50 percent of these fire
less than 10 percent oil,  then the SQg contribution would be
less than 2 percent of  the US total.  It is believed that these
assumptions regarding the number of units burning less than 10
percent oil are very conservative as well and a more realistic
SO- contribution vould  be more in the order of 1 percent rather
than 2 percent.

    It is very unlikely that the accounting of all US SO2 emis-
sions for a particular  yearly period vould have an accuracy of
± 2 percent much less ± 1 percent.  This is particularly true
since the  allowed inaccuracies  for certified  CEMs is  ±  20
percent.    In  the end,  even  completely  neglecting the S02
contribution from the gas/oil fired boilers that burn less than
10  percent  oil would not  detract significantly  from  the
accounting  of  the total  US  SOg  tonnage.   As  a consequence,
relatively simple SO. accounting methods (other than CEMs) for
these boilers would  be sufficient to estimate  this  1 percent
SO2 contribution.

2.2 SOX Formation

    Alternate methods  for monitoring SOg emissions are depen-
dent upon  the fate  of  the  fuel sulfur.  If the  fuel  sulfur
eventually  leaves the boiler  by  paths  other than  through
gaseous emissions or if gaseous sulfur species (SO^) other than
SO,  ar^present   in large  quantities,  then  the  alternate
monitoMML methods described in Section 1 would not be appro-
     ^^»^ff^^
priate.   Since the fuel  sulfur content of natural  gas is
essentially zero,  there is no discussion of alternative methods
of monitoring sulfur dioxide emissions  from the combustion of
natural gas.  The  following paragraphs  illustrate the fate of
the fuel sulfur for oil combustion and demonstrate that  for oil
firing, the alternate  methods would be  appropriate.

-------
Oil BO  Teraation

    The  oil burned  in  utility plants  today  is a  refined
product.  In addition, the tolerance on certain constituents in
the oil can be held to reasonably close to specifications.   As
mentioned  previously,   most  oil  fired  utility  boilers  in
metropolitan areas fire relatively  low sulfur oil.  The sulfur
content is restricted by contract with the supplier and due to
the cost of desulfurization, the delivered oil is usually close
to the specified sulfur content (not significantly lower).  As
a consequence,  in most  cases (but not  necessarily all) the oil
delivered from one supplier has a  relatively  constant sulfur
content.   Even  in the  case where  the oil  is  delivered from
different suppliers to  the  same specification,  the variation
between suppliers would not necessarily be significant.

    The sulfur in a batch of oil is, for all practical purpos-
es, homogeneous.   Furthermore,  in most cases  oil from different
batches is circulated within tanks  and therefore becomes rela-
tively well mixed at some point in time.  As  a result of the
relatively constant oil sulfur content and the ability of the
oils to  mix, the sulfur content of fuel  oil does not change
significantly over short periods of time.

    The ash content in  fuel  oils is generally much less than 1
percent (appro*.  0.1%), consequently,  there  is little opportu-
nity fcor sulfur capture as sulfates in the ash as is the case
with aj^ftV  Fuel oil combustion, therefore, results entirely in
the
.tion of
                      and SOj.   Generally, SOj formation from
oil combustion in utility boilers  is in the order of 5 to 10
ppm or less than 0.00001 percent.  Since there are insignifi-
cant  amounts of  sulfates  and  SOj  formed  during  fuel  oil
combustion, essentially all of the sulfur in the oil converts
directly to SO..  Tha sulfur content of fuel  oil la. therefore.
a direct measure of the SO* emissions.

-------
    1} Measure the fuel consumed over the monitoring period
       by accounting for quantities on hand plus deliveries
       minus quantities remaining.

    2) Calculate the  average sulfur content by a weighted
       average of  the sulfur content of quantities at the
       beginning and  end  of the period vita those for the
       contents of the deliveries.

    3) Calculate the average fuel oil higher heating value
       (HHV) by a weighted average of the HHV of quantities
       at the  beginning and end of  the period with those
       for the contents of  the deliveries.

    4) Calculate the  SO-  emission  rate (lb/HX8tu)  and the
       total SO. tonnage  during the period  using the fuel
       oil  quantity  consumed  during  the  period  and the
weighted  average  sulfur
heating value of the oil.
                                  content  and  the  higher
Throughput 8O« Monitoring Method
     T       Z
                                                    •v

    Under certain circumstances, the measurement of .the total

quantity of  fuel oil flow by the INPUT METHOD  may not be as

accurate as necessary.  This may be due to large tank capaci-

ties or the  method  of fuel oil supply to the unit.  In these

cases it may be necessary to measure the  fuel flow  to the unit

rather than  the  net amount of oil  in the tanJc(s).   The major

difference between  the IVPOT and THROUGHPUT  methods is that

the  actual  fuel  flow to  the  unit is measured by use  of a

continuous fuel oil  flow meter.  The calculational procedure is

identical between the two methods.
    The assumption is  made that the  potential inaccuracies
       iBk'^y the measurement of the total fuel oil quantities
do  not ntreduce  significant  inaccuracies  in  the  weighted
average sulfur content of the fuels burned.  It is assumed that
widely varying fuel sulfur  content fuels are not delivered to
the plant for the reasons stated in 2.2.
                              10

-------
    The proposed  alternate THROUGHPUT METHOD  so.  monitoring
approach im as follows:

    1) Measure the fuel  flov  consumed over the monitoring
       period by utiliration of an in-line continuous flov
       meter.
    2) calculate the average  sulfur  content by a weighted
       average of the  sulfur  content of  quantities at the
       beginning and end of the  period with those for the
       contents of the deliveries.
    3} calculate the average fuel oil higher heating value
       (HHV) by a weighted average of the HHV of quantities
       at the beginning  and end of  the  period with those
       for the contents of the deliveries.
    4) Calculate the SO. emission  rate (Ib/KKBtu)  and the
       total' BO. tonnage during  the  period using the fuel
       oil  quantity  consumed  during  the  period  and  the
       weighted  average  sulfur  content  and  the  higher
       heating value of the oil.
Continuous 802 Monitoring Method

    Under some situations, neither the measurement of the total
quantity of fuel oil flov from storage tanks nor the assurance
that the sulfur contents  of various sources of fuel can be made
with certainty.  In these instances, these measurements must be
made at the boiler inlet.  As with the THROUGHPUT METHOD, the
fuel flov  can be measured vita  a  continuous totalizing flov
meter.  The measurement  of  sulfur  content in the fuel can be
par f oraeil- -in a standard grab sample/ analysis method.
          lie grab  samples  in the fuel  line leading  to the
boiler can be made  as required.   The  frequency of sampling may
be dictated by the variability of the fuel sulfur supplied to
the storage  tanks,  the degree  to which the storage tank is
mixed and the flovrate of the  fuel oil.  Not all  of the samples
need be analyzed individually  since all that is necessary is to
know the  average sulfur content over a specified  period of
                              11

-------
time.  The individual samples can be accumulated and thoroughly
mixed prior to analysis at the reporting times.

    The proposed  alternate CONTINUOUS  METHOD so2 monitoring
approach is as follows:

    1) Measure the  fuel  flow consumed over the monitoring
       period  by  utilization  of  an  in-line  continuous
       totalizing flow meter.
    2) Accumulate  discrete  in-line  fuel  oil samples  at
       specified times during the reporting period and mix
       representative portions for all samples taken during
       the period  for  analysis  of the average sulfur con-
       tent and the higher heating value.
    3) calculate the SO-  emission rate (Ib/HMBtu) and the
       total 80. tonnage  during the period using the fuel
       oil  quantity consumed  during  the  period  and the
       weighted  average  sulfur  content  and the  higher
       heating value of the oil.

3.0 NITROGEN OXIDE EMISSIONS

    Section 407 of the 1990 Clean Air Act Amendments addresses
the  requirements  for the  Nitrogen Oxides Emission Reduction
Program.  This  section states that the applicability of this
section is for coal-fired units  only.  Emission limitations, are
specified in this section for tangential- and dry bottom wall-
fired units that burn coal.  No nitrogen oxide limitations are
specified  for  gas- or  gas/oil-fired units.    Section  412
indicate^ that  sources subject to the Title  must install and
operatsjjj^MB  for sulfur oxides, nitrogen oxides, opacity and
volumetric  flow on  each affected  unit and  that alternative
monitoring approaches  may be approved by the Administrator.

    By virtue of the fact that gas/oil fired units firing less
than  10 percent oil (Section 405  (h))  are affected units for
S02 emissions,  Section 412 is applicable for these units for

                              12

-------
SO- emission, however, since HOx  emissions  are not regulated
for gas/oil-fired units under  this  Title,  it is questionable
whether  HOx monitoring should be  required.    The  following
paragraphs briefly describe alternate NOx monitoring approaches
which  could provide  emissions information  at least at  the
accuracy level as that provided by CEMs.

3.1 Qas/OiT NOX Formation

    HOx  formation  in utility  boilers  is  a  function  of many
parameters.   Some  of the  parameters  are  peculiar  to  the
particular boiler and cause the NOx emissions characteristics
to be unique to that particular boiler.  These parameters are
related to  the design of  the  boiler (tangential, wall-fired,
etc.)   and   cannot  be changed  without major changes  in  the
design, consequently, for  a given unit,  the NOx characteristics
are relatively constant.   Other NOx  influencing parameters are
related the type of fuel combusted and still  others are related
to the mode of operation  which generally is dictated by the
type of fuel burned.

    In the end, the parameters that can change NOx characteris-
tics from a particular boiler are related to the type of fuel
burned.  There are two basic types of NOx formation mechanisms
- Thermal NOx and Fuel NOx.  Thermal NOx formation results from
the thermal fixation of atmospheric nitrogen.  Fuel NOx results
from oxidation of the fuel-bound nitrogen.  The contribution of
each ofTJfreie formation mechanisms  is shown  in Table 3-1.
                              13

-------
                          TABLE 3-1
            CONTRIBUTION OF FORMATION MECHANISMS
                  70R UTILITY BOILER FUELS

       FUEL                     FORMATION MECHANT8M
                              THERMAL NOX    FUEL NOX
      Natural Gas                100%           0
      Fuel Oil  "               70-80       20-30
      Coal                     20 - 50       50 - 80

For coal combustion, the  influence of  fuel-bound nitrogen is
the predominate source  of NOx.   For  most fuel oils it  is a
secondary source and for gas fuels it is not a factor.

    The fuel properties for natural gas are, for all practical
purposes, invariant and  contain virtually no sulfur or nitrogen
bearing  compounds.    Fuel oils,  which are  supplied with a
purchase  specification  for maximum sulfur content,  usually
result in a relatively narrow nitrogen content variation.  As
a consequence of  the  relatively constant fuel properties for
gas/oil-fired boilers, the NOX characteristics are relatively
easy to characterize for a given condition.  In addition, these
KOx characteristics are very repeatable.

    On gas/oil-fired boilers,  the units are generally operated
in, more or less,  fixed  configurations at each load point.
This is primarily due to the operational factors that influence
heat r*£» and boiler efficiency.   In these fixed configura-
tions, SpH major  parameters that influence NOx formation are
the load point and the operating excess oxygen.  Most gas/oil-
fired boilers are generally operated over the load range with
a  excess oxygen curve  set by the automatic  boiler fuel/air
ratio  controls.   This results in near constant excess oxygen
levels at steady load  conditions.   Some variation exists in the
operating excess oxygen level dependent upon the steam tempera-

                              14

-------
ture control method and during times when the unit is operating
in a transient load condition.  In most cases the excess oxygen
variation about  the  normal level varies no  more that  ±  0.5
percentage  points  in 02 and  at the  most by ±  1  percentage
point.

    With the repeatable nature of NOx emissions from gas/oil-
fired boilers,  it  is possible to establish  the  NOx emission
rate  characteristics by a periodic  measurement of  the  NOx
versus  load.   The  frequency of  this  NOx  characterization
depends upon the alternate monitoring accuracy requirements.
Depending upon the  degree to which 02 influences NOX emissions,
a characterization of this parameter may also be required for
a thorough emission rate characterization.  The assessment of
whether 02 characterization is necessary is dependent upon the
slope of the NOx versus 02 curve.   If the change in NOx over
the normal  0. excursion  at a particular load  causes  the NOx
estimation  to fall outside acceptable  accuracy  limits, then
this factor  may  need to be included.   If, on the other hand,
the excursion results in a  variation about the normal 02 point
of no more that i 10 percent, then inclusion of this factor may
not be necessary.

3.2 Alternate NOX Monitoring Approaches

    Two alternate  approaches are proposed which depend upon
adequate^ characterizations of the NOx emissions  for both gas-
and oi^b^ing conditions.  The frequency of these character-
izationsFVould  likely be once per year or at a maximum twice
per  year.    The  two  approaches require  differing  amounts of
characterization depending upon the sensitivity of NOx to 02
excursions.   They  differ only  in the  necessity for character-
izing excess oxygen.
                              15

-------
Load CB«^e.qterizatien NOx Monitoring Method


    The  proposed alternate LOAD  CHARACTERIZATION  METHOD NOX

monitoring approach is as follows:


    1) Perfora a thorough NOx versus load characterisation
       using  EPA approved methods  at a ainiaua  once per
     .  year.

    2) Measure the  gas fuel flov consumed  over the morii-
       toring period by utilization of an in-line continu-
       ous totalizing flow aeter.

    3) Measure the oil fuel flow by one of the three aeth-
       ods described  in the Alternate  SO.  Monitoring Ap-
       proaches  (section 2.3).

    4) Continuously measure and record the unit load.

    5) Calculate the. annual NOx emission rate by integrat-
       ing the NOx versus load over the operating tiae for
       gas and  oil firing  by weighting the  emissions by
       operating  tiae on  each  fuel  over  the monitoring
       period.


Load/O» Character!eation NOx Monitoring Method


    If it is determined during the initial testing of the unit

that excess oxygen significantly influences  NOx emissions at a

particular load/ this  factor  will need to be included in the

NOx characterization.
                 alternate LOAD/O, CHARACTERIZATION METHOD NOx
      ^B.  ~                     *
aonltozjng approach is as follows:


    1) Perform a thorough NOx versus load characterization
       using  SPA approved methods at a minimum  once per
     •  year.  At each  load point establish the o. character-
       istics for an excursion at least 1 percent above and
      .below the normal 0- set point.

-------
                                                           -77
    2) Measure the gas  fuel flow consumed over  the  moni-
       toring period by utilization of an in-line continu-
           totalieiog flow meter.
    3} Measure the oil fuel flov by one of the three meth-
       ods described  ia the Alternate SO. Monitoring Ap-
       proaches (section 2.3).
    4) Continuously measure and record the unit load and O2*
    S) calculate the annual NO* emission rate by integrat-
       ing the NOX versus load and 0- characteristics over
       the operating time for gas ana oil firing by weigh-
       ting the emissions  by operating time  on  each fuel
       over the monitoring period.
4.0 ALTERNATE OPACITY MONITORING METHOD

    Prior to the use of CEMs for monitoring opacity, EPA Method
9 was used to assess the  opacity  of  oil-fired units.   Due to
the relative small amount of time that oil is burned on these
Section  405(h)  units,  it would  seem  appropriate to  again
utilize this as an alternate primary opacity monitoring method
during periods  of  oil  firing.   The method  requires a trained
qualified  observer to periodically  visually determine  the
opacity from the utility boiler stack.  The Method 9 procedures
are delineated  in 40 CFR  Part 60, Appendix A.

    Units that fire less  than 10 percent oil are not likely to
constitute a  serious  opacity problem.  This  is particularly
true of  units that fire gas and  oil  simultaneously.   During
period *t time when these units are firing less than 50 percent
oil, ivfr extremely unlikely that opacity problems  would occur
due to •olfur oxide plumes for any of  the commonly used sulfur
content oils.   In addition, since most units that burn oil in
metropolitan areas fire relatively low eulfur  oil.  Even under
conditions where  100  percent oil was  fired it is  not likely
that sulfur oxide plume opacity difficulties would  occur.   As
a  consequence  of  these   factors  and  the  fact that  opacity

                              17

-------
limitations are not specified for units under Section 405 (h),
installation of  a  costly  CE«  for  monitoring opacity  would
appear to  be inappropriate.   Method 9  could be used  as  an
assurance  that  opacity levels  were maintained  within  local
required levels.  As  an alternative to the installation of a
GEM for  opacity monitoring on  units that fire  less  than  10
percent oil, Method 9 is proposed  as  the primary monitoring
method.

    The frequency and appropriate periods of time to apply the
visual monitoring  approach would be dependent  upon  the oil
firing scenario.  Certainly under conditions where oil and gas
were fired simultaneously at an oil  level less than 50 percent
monitoring would be  inappropriate.  In most instances where the
sulfur content was less than 1 percent  it may also be inappro-
priate.  The following is  a proposed method to ascertain the
need and frequency of visual opacity monitoring.

    1) Through controlled tests over the  load range estab-
       lish the opacity levels under typical gas/oil firing
       coaditioas.
    2) Determine the conditions under which opacity levels
       exceed federal or local standards or other appropri-
       ate criteria.                                 ...
    3) Bstablish the  conditions under which opacity moni-
       toring is appropriate.
    4) Bstablish the  frequency of  monitoring for the ap-
       ipopriate operating  conditions.
Under the title, there are no limitations for opacity.  In the
absence  of this limitation,  some criteria would  have to be
developed to determine when monitoring was appropriate. If it
was determined that regular opacity monitoring was not warrant-
ed based upon the above procedure, an  annual re-certification
of  the opacity  characteristics would  likely be  sufficient.

                              18

-------
Under other conditions visual monitoring might be required only
during periods which could result in exceedences of the opacity
criteria.

5.0 CONCLUSION

    As the Clean Air Act Amendments (CAAA)  of 1990 are pres-
ently written, installation of expensive CZMs is required for
all affected units.  Ostensibly the Amendments are designed to
reduce SO*  and NOx  emissions  from coal-fired  boilers.   The
Amendments recognize that units firing oil can contribute
emissions as well and therefore under Sections 405 (e) ,  (f) , (g)
and (h) allowances for S02 are specified.  Under Section 407 no
limitations on NOx emissions are specified for the aforemen-
tioned units under Section 405.   The  amendments recognize that
the emissions of SO£ and NOx from coal fired units  is the major
concern.   As indicated in Section 2.1  of this report,  units
under CAAA Section 405 (h) represent much less than 1 percent of
the  S0g  emissions  from  the total US  boiler  population.
Furthermore,  there is  no NOx  emission  limitation  on  these
units.   Nevertheless, the  requirements of  CAAA  Section 412
presently require expensive CEMs  to monitor both S02 and NOx on
these units even though NOx has no limitation and SO* emissions
from  these units  are  less  than 1  percent  of the  total  US
emissions  (much less than the measurement accuracy of CEMs) .

    Installation of a complete CEM on a gas fired unit could be
as  higJJ^a  $300,000  (1990  $) .    In  addition,  maintenance,
supplieir and administrative reporting  associated  with the
monitor would amount  to  $30,000  or more  per year.   For a 15
year  monitor life, this  would  amount  to a typical utility
annualized cost of approximately  $87,000.  on a dollars per ton
monitored basis for a 250 MWe gas/oil fired boiler firing less
than 10 percent oil, this is  in the order of $275/ton.  This is
approximately 15 times the cost for a similar coal-fired unit.
                              19

-------
Baaed  upon  this  significant  disparity  and  the  small  SO,
contribution  from  units under Section 405(h)  to the total us
S02  emissions,  it is  appropriate that other  less expensive
monitoring approaches be used on  these units.

     Based upon the small S02 contribution and the fact that NOx
emission and  opacity limitations  are not addressed for units
that fall  under  Section  405(h),   the   following  alternate
monitoring requirements would seem appropriate for these units:

     1) Monitor SO. emissions  based upon the sulfur content
       of the fuel oil and measurement of  the fuel usage
       for the monitoring period.
     2) Wave the  monitoring requirement for NOx emissions
       for this class of boiler.
     3) Establish opacity monitoring requirements based upon
       EPA Method  9.

Adoption of  these alternate  monitoring methods would supply   (H|
information as  accurate as that  supplied by  OEMs  at  a much
smaller  dollar/ton monitored cost.    In  addition,  it  would
eliminate the need for the use.of  volumetric flow measurements
which  may  introduce more  inaccuracies  than  the  alternate
monitoring methods.

    Use of the alternate monitoring methods proposed does not
detract from  the intent of  the Clean Air  Act Amendments of
1990.  ftAe* the coal-fired units are the major  contributors to
      ^f»
both  t^KMu and  NOx and  since  units under  Section  405(h)
       jfy   *
contribute 802 l€Vttls *ar lesfl than the accuracy of the present
monitoring methods, these proposed alternate monitoring methods
will not result  in estimates of  the  emission  rate or annual
tonnage that are measurably either over or under the true total
US levels.
                              20

-------
                          ISSUE PAPER —                   E21
                  PROVISIONS FOR ALTERNATIVES TO
               CONTINUOUS EMISSION MONITORS  (CEMS)

                          March 13,  1991
I S SOB PRESENTED

     This paper  discusses criteria and procedures  that could be
applied to  alternative  monitoring systems seeking approval under
Section 412.  The paper is being presented for discussion; none of
the options herein are  being  endorsed by  EPA at this time.  ARAC
members  are  encouraged  to •  present   other  options  they  think
appropriate.

INTRODUCTION

     Section 412 (a)   requires  each affected  unit to have  a full
complement of emission monitoring equipment.  A unit is an affected
unit  if  it is subject  to any  emission reduction requirement or
limitation under Title IV.  Table 1 lists the affected categories
of units.

     Each category listed in Table 1 would be required to install
a full complement of monitors as described in Section 412 (a)  (S02,
NOX,  opacity,  and  volumetric  flow) .    Depending  upon  the NOx
  >nitoring requirements of Title IV,  a diluent  (O2 or C02) monitor
    also be required.

     Section  412 (a)   directs   the  Administrator  to  establish
demonstration  criteria  for  alternative  systems  that  meet  the
criteria   specified   with   the  same  precision,   reliability,
accessibility, and timeliness.   An alternative monitoring system
is  a  CEMS in which  any or all of the named monitors  have been
replaced with equivalent emission measurement  and data reduction
techniques .

ALTERNATIVE DEMONSTRATIONS

      The objective  of the alternative regulations is to provide
guidance concerning  permissible alternatives and to minimize the
need for large scale case-by-case evaluations.

Who Makes the Demonstration?

     Section 412 (a)  directs  the  Administrator  to specify  the
requirements for CEMS and alternative systems  demonstrated to be
the  same  as  CEMS   in  the  four attributes  discussed  in  the


             Preliminary Draft for ARAC Discussion —
           Does Not Represent the Position of the U.S.
                 Environmental  Protection Agency

-------
Equivalency  Considerations  below.    It  is  assumed  that  the
demonstration is made to the Administrator.  Though the wording of
the Act  may be  interpreted  to  mean demonstrated by  an affected
source,  there is  no preclusion  as to  the Administrator's  own
demonstrations.  Who may make a demonstration  runs  from a single
affected  source,  industrial group,   or  the  Agency  (including
States).  Because the "who" is not defined, the legislation would
allow  the  Administrator to  specify avenues for  demonstrations.
Specifying how demonstrations could be  made will reduce the number
of demonstrations  and maintain  order  in the  allocation process
while achieving the national reduction goals.  The regulation may
specify  acceptable alternative  demonstration  programs  afforded
affected units.  Alternative demonstration programs may be made on
a:

         •  case-by-case unit specific basis.   This  has been the
         historical procedure of the Agency, e.g., NSPS Subpart Db
         for NOX monitoring;

         •  representative testing applied to a class or category
         of affected units.   A demonstration may be  made for an
         alternative monitoring system with  applicability to  a
         defined  category   of   like   units.     This  successful
         demonstration, by  industry or EPA, would be limited to
         the specified category.  The  EPA would have the latitude
         to define the class or category based on Table 1 or sub-
         sets of Table  1 by criteria such as pollutant, size, fuel,
         control equipment,  etc.  The validation protocol mentioned
         in a later section  addresses this demonstration.   Any
         source not included in  the category would be afforded the
         case-by-case option;

         •  administrative determination. The Administrator could
         specify  acceptable alternatives  based upon historical
         information on alternative demonstrations.

    The  regulations would specify who may make  a demonstration and
how the demonstrations can be made.

Approval by Whom"?

    Section 412(d) implies that the alternative monitoring systems
must be approved by the Administrator.   Historically this authority
has been delegated to afford orderly functioning of Agency actions.

    The  demonstrations must be  conducted prior to the date a GEMS
system is required to be installed and  certified.  For purposes of


            Preliminary Draft for ARAC Discussion —
           Does Mot Represent the Position of the U.S.
                 Environmental Protection Agency

-------
  e Act  this date is  November  15,  1993 for  Phase 1  sources and
  nuary  1,  1995  for  Phase  2.    Approximately  2,000  potential
alternatives could be applied for within the next  four years.

    The  approval process may consist of three phases:  the initial
application, performance testing, and final approval.  The initial
application must be submitted to the Agency.   The  latter two steps
would need to  be  coordinated with  the  Agency depending upon the
Agency's desire for  process efficiency and to have certification
observers  present.

    The  regulations should clearly identify  the  organization in
EPA responsible  for administering  and  receiving   demonstrations,
approval  authority  over  the  demonstrations,  and certification
authority  for all CEMS  and  alternative methods.  The regions and
States,  as proposed  in the  original overall  program  framework
paper, would act as the field observers and preliminary reviewers
of the demonstrations and certifications.  They would make findings
and recommendations  to the specified authority . on acceptance or
rejection.  The Region and State activity  is predicated on intra-
 and  interagency  agreements;  but  operated  in   this  manner  a
nationally consistent deliberation  and  approval process would be
maintained.

EQPIVALENCY CONSIDERATIONS

    Section 412(b)  and (c)  of the Clean Air Act (the Act) require
generators subject to the Phase  l and Phase 2  limitations of Title
4  to  install  and  operate  CEMS, quality  assure   data,   and  keep
records  and  reports in accordance  with  regulations issued under
(a).   Section  (a)  requires the   Administrator  to specify  the
requirements  of CEMS  by  regulation within  eighteen months of
enactment.  The Administrator must also specify requirements for
any alternative monitoring system that is demonstrated as providing
information with the same  precision, reliability, accessibility,
and timeliness as that provided  by the CEMS.   The requirements for
alternatives  may   also  specify   limitations  on the   use  of
alternatives as are necessary to preserve  the orderly functioning
of the allowance trading system and assure the reductions called
for in th« Act.

    The  Act's  definition of a CEMS system includes the pollutant
monitor, effluent flow  monitor,  and the data system to create on
a  continuous  basis, a  permanent record  of the  emissions.   For
alternative requirements developed by the Administrator, the system
includes pollutant and flow such that the  reported units are on a
mass per unit of time basis.


             Preliminary Draft for ARAC Discussion —
           Does Not Represent the Position of the  U.S.
                 Environmental Protection  Agency

-------
                                                          It*
     The Act specif les that an alternative monitoring system is only
approvable if  it is  demonstrated to meet  the four  criteria of
equivalency established by CEMS.   Those criteria  in terms of the
GEMS  must  be  established  in  order  to  judge  an  acceptable
alternative.    Those  criteria  and  "Webster's Ninth  New College
Dictionary" definitions are:

         Precision -       The degree of refinement with which an
                           application   is    performed    or   a
                           measurement stated;

         Reliability -The extent  to which an  experiment, test, or
                       measuring procedure yields the same results
                       on  repeated trials;

         Accessibility -   Capable of being used or seen or ability
                           to obtain or make use of;

         Timeliness -      Appropriate or adapted to the times or
                           the occasion,  coming early or  at the
                           right time.

     These  definitions must  be   augmented   to  conform  to  the
conventions  of CEMS  as  practiced over  the  last two  decades.
Specifically, reliability has also come to include the availability
of  the  monitoring  system.    This  should  be retained  in  our
definition   of  reliability.      In  addition,   timeliness  and
accessibility may not be mutually exclusive.

     The Act  states,  "with the same,"  preceding  these criteria.
This should mean that  the alternative system is capable of being
placed in a one-to-one correspondence (equivalent) to the CEMS.

  ;   The Act does not prevent the Administrator from imposing other
requirements for possible alternatives.  Substitute data systems,
quality assuring data,  and recordkeeping and reporting  data are
applicable at  all  times  and  equally  apply  to  CEMS and their
alternatives.

     In th« context of paragraph (a) , successful demonstrations of
alternatives should be required to meet  the capability of the CEMS
in fulfilling these  requirements under the general criteria above.

     The alternative  requirements  to  be proposed  should  also
incorporate the Agency policy that BO  exemptions to the monitoring
requirements of Title 4 will be allowed.  A successful
             Preliminary Draft for ABAC Discussion —
           Does Not Represent the Position of the U.S.
                 Environmental Protection Agency

-------
                                                                (J -J
 . . a
•;
   onstration should not be interpreted as an exemption to the CEMS
 equirements, e.g., fuel monitoring'of natural gas for SO2 emission
monitoring.

Precision. Reliability*  Accessibility*and Timalineaa as Defined
bv the GEMS and Plov Monitoring

Pollutant Monitors

     The proposed requirements for Acid Rain contemplate that the
pollutant monitors should first be capable  of meeting the accuracy
requirement as detailed  in the  regulations.   In  addition,  the
monitors must be linear across the range of  potential measurements.
The  pollutant  system  should  probably have an  hourly continuous
measurement capability.    The reporting requirements contemplated
require quarterly submissions of daily accumulations based on the
1-hour records.  The data,  and reliability of the monitor, would
be  required  to  be demonstrated  over a  16 8-hour test  where no
adjustments to the system are allowed.  In addition, QA is required
to assure continued monitor reliability, quarterly and  annually.

     The results  of pollutant  monitor  certifications,  relative
accuracy  audits,  relative accuracy  test  audits,  cylinder  gas
audits, and retrievals from AIRS/AFS on monitor availability have
shown that-the current pollutant compliance monitoring  systems for
^ SPS Subpart Da, over long periods of  time  and repeated tests, are
 eliable.  The pollutant monitors routinely  pass the short-term
relative accuracy requirements and have availabilities greater than
95 percent.    The contemplated requirements would also require that
the pollutant monitors not exceed the design drift limitations on
a 24-hour basis.

Flow ..Monitors

     The probable availability is expected  to be  similar to that
achieved by the pollutant monitors above.  The  accuracy of flow
monitoring  systems  appears   to  be   approximately  5  percent.
Preliminary information indicates that some monitors currently meet
this requirement.  Additional data is being obtained.

     The -requirements   contemplate  initial  certifications  for
accurate measurements  over  the range  of   flows  and  reliability
assurances on a periodic basis.

     Accessibility  and timeliness  appear to  be  the same  as  for
pollutant CEMS.
               Preliminary Draft for ARAC Discussion —
             Does Not Represent  the  Position of  the  U.S.
                   Environmental Protection Agency

-------
Opacity^ ftpnitors

     Opacity monitors are used widely  for  NSPS (proper operation
and maintenance) and SIPs for compliance with opacity limitations.
These systems provide 1-minute averages based on six or more data
points equally spaced over the minute.  Their availability has been
documented  to  be greater than  98 percent.   Design requirements
specify that they be  accurate to within  3  percent over the range
of emission measurements (calibration  error).   The resolution of
the measurements are to  be to the nearest 0.5 percent opacity.

     The  opacity monitoring  systems are limited  in application.
They are  not applicable  in  effluent streams  that  contain water
droplets and do not measure opacity forming downstream of the stack
exit.

     The  contemplated requirements for Acid Rain will require the
installation  of opacity monitors  where  no  current requirement
applies.      The .  systems  must   meet  the  current  performance
specifications  required  of  NSPS  or the  SIP,  which  ever  is more
stringent.  Opacity  limitations currently required of the unit will
continue.  Recordkeeping and reporting will be  consistent with the
current   NSPS   or   SIP   requirements   for  existing  and  new
installations.

Data Acquisition^Systems

     The  data management system requirements  of  the regulations
discussed in another paper would:

     •  require the maintenance of all emission-related measurements
     on site;

     •  require quarterly reports  of summarize quality assured data;

     •  require reporting of  the data  in  some type of electronic
     media.

     The  paper further states that,  the state of  the art process
control  and accompanying computer  equipment  allow  operators of
combustion  and other  process equipment  to sample, record,  and
analyze  a  large  number of  process parameters.    This  greatly
enhances the timeliness  and accessibility attributes of the GEMS.
             Preliminary Draft for ARAC Discussion —
           Does Not Represent the Position, of the U.S.
                 Environmental Protection Agency

-------
            An alternative is something that can be approved in place of
       the GEMS.  The alternative must provide information with the same
       attributes as described  for  the combined pollutant CEMS and flow
       monitoring system required of the unit.  This must be demonstrated.

       precision

            A current-emission.measurement  validation protocol provides
       two  statistical  tests which must  be satisfied in order  to show
       equivalency on a  system basis.  The document is  entitled, "Protocol
       for the Field Validation of Emission Concentrations  from Stationary
       Sources"  (protocol).  .   Briefly the proposed alternative  must
       demonstrate  a  precision equivalent  to  the validated  method (F-
       test) and an accuracy  (t-test) test showing there is no bias at the
       80-percent  confidence  level.     If   the  bias is  shown  to  be
       statistically significant a  correction factor is  evaluated, but
       cannot be outside the range of 0.90 to 1.10.

            This field demonstration program would  address the equivalent
       accuracy and precision of the alternative monitoring system under
       field  conditions to the  CEMS.   In  the protocol,  18  samples (9
       pairs) are obtained and compared.  The F-test  and t-test analyses
       are  independent  of the  time between  sampling and  the emission
1        eveIs at which  the sampling is conducted.   The field test could
        e conducted over  the  range  of  emissions that  occur  in some time
       frame, e.g., 48 hours.

            Other protocols may also be  referenced by the rules for certain
       applications.  Test Method  19,  Determination of Sulfur Reduction
       Removal  Efficiency and  Particulate  Matter, Sulfur  Dioxide, and
       Nitrogen Oxides Emission Rates, could be referenced appropriately
       for alternative systems that would be incorporating fuel sampling
       and  analysis.   These  requirements would  be  in addition  to the
       precision and accuracy test mentioned above.

            This fuel sampling and analysis  protocol only covers coal and
       oil,  and has the  limitation of  not being an as  fired  system.
       Sampling and analysis results do not  necessarily reflect what coal
       is actually being burned. This is due to the practice of bunkering
       fuel  for combustion.   The regulations  should  specify  "as fired"
       sampling.  An additional protocol  for natural  and fuel (refinery
       process) gas fired generators would need to be provided.
                    Preliminary Draft for ARAC Discussion ~
                  Does Not Represent the Position of the U.S.
                        Environmental Protection Agency
L

-------
                                8

Reliability

     The CEMS  and  flow monitor  system  requirements  specify  a
continuous demonstration of reliability,  at a minimum,  on a 24-
hour  basis  (accuracy  within 2.5  percent  of  a  known reference
value).    Zn   addition,   the  reliability must  be  initially
demonstrated over a 168-hour test in which no adjustments are made
to  the  system.    A  similar  test  could  be  conducted   on  the
alternative system.  A time period could be  established over which
no adjustments to the alternative system could be made.  The output
of the system would be checked against the  CEMS/flow system  output.

     Continual  reliability of alternative systems  (QA) must  provide
the. same assurances as established by the  CEMS.   In general the
rules should specify  that  alternative  systems must  provide data
validation procedures,  data reduction procedures and recalibration
procedures.  The rules  would specify the data validation criteria,
the  . minimum  data   reduction   procedures  and  recalibra.tion
requirements.

     There  are  no specific  Federal  criteria that could be cited.
In general the regulation may be framed around the following:

     • For  alternative  systems employing parameter monitoring;

            Data Validation.  Data  must be  considered invalid if
          any of the following conditions occur:

             a.  The equipment being used  to monitor the parameter
             is  not operated in accordance  with the manufactures
             requirements,

             b.  The equipment being used  to monitor the parameter
             is  not  being  maintained in  accordance with  the
             manufacturers  specifications,

             c.  The parameter monitoring equipment is inoperative,

             d.  The monitored  process  is not operating.

         -  Data Reduction Criteria:

             a.  All averages must  be calculated using valid data
             only,

             b.  A one-hour average will  be considered valid only
             if  it contains 100  percent  of the  readings  of all
             parameters used in the calculation,

            Preliminary Draft for ARAC  Discussion —
           Doea Not Represent the Position of the 0.8.
                 Environmental Protection Agency

-------
           -  Maintenance Requirements:
               a.   Minor maintenance.   Any maintenance done  that
               does  not  effect  the  integrity  of  the  parameter
               monitor.   Requirement  for a  calibration check prior
               to and after maintenance,
               b.   Major maintenance.   Any maintenance  that would
               effect  the  integrity  of  the  parameter  monitor.
               Recalibration of the equipment upon completion of the
               maintenance or repair.
               Periodic Recalibration.   The  recalibration  of  the
           parameter monitor shall be conducted on a quarterly basis.

      •   For systems using fuel  sampling and analysis  the  QA may
     . include the analytical QA published by ASTM in addition to the
      following;
              Data Validation Criteria
               a.   Criteria for daily composite unit samples.
j|t            b.   Criteria for daily composite system samples.
               c.   Criteria for laboratory  sample analysis
           -  Data Reduction Procedures.  These would be the same a
  for CEMS and parameter monitoring.
                Maintenance  Requirements.    Besides  manufacturer
           requirements the rules may have:
               a.   Requirements for the sulfur analyzer.
               b.   Requirements for the calorimeter.
               c.   Requirements for the sample acquisition  system.
               d.   Requirements for the sample preparation  system.
               Periodic  Recalibration.    Quarterly  recalibration
           through performance specification testing.
               Preliminary Draft for ARAC Discussion ~
             Does Not Represent the Position of the U.S.
                   Environmental Protection Agency

-------
                                10

Accessibility

     In the case  of the GEMS  and flow system,  the generator is
expected to obtain and use quality assured data on a 24-hour basis.
The information is available on site.  Systems that would require
longer periods are not equivalent.


Timeliness

     The  GEMS  system includes the data acquisition system.  Every
hour on the hour an emission rate  is permanently  recorded for Acid
Rain.  On a 24-hour period the data is quality assured.  The data
then becomes available.  Adjustments are automatically injected to
the permanent  record, e.g., substitute  data, as  flagged  by the
system as it monitors the performance of the GEMS.  At the end of
each  24-hour  period  and  quarter,  the  data  is  summarized  and
available within  the  CPU  time of  the data  acquisition system.
These  timeliness  criteria must  be  matched, with the alternative
systems established to monitor  the data  inputs to the system and
quality assurance checks of the system.
            Preliminary Draft  for ARAC Discussion ~
           Does Not Represent the Position of the U.S.
                 Environmental Protection Agency

-------
                     11
    Tabl* 1.   Affected  Source  Categories
                  category
       All units named in Phase I

       All units with a nameplate capacity equal to or
       above 75 Mve and annual baseline emission rates
       greater than 1.20 Ibs/mmbtu

       All  coal or  oil  fired  units  with  nameplate
       capacity less  than  75 Mwe and  annual baseline
       emission rates greater than 1.20 Ibs/mmbtu

       All  coal  fired  units  with  annual  baseline
       emission rates .less than 1.20 Ibs/mmbtu .

       All oil and  gas fired units with annual baseline
       emission rates equal to  or  greater  than 0.60
       Ibs/mmbtu

       All oil and  gas fired units with annual baseline
       emission rates less than 0.60 Ibs/mmbtu

       All oil and gas fired units consuming less than
       10 percent oil per annum

       All  new units commencing  commercial  operation
       between 1986 and December 31, 1995

       Any process  source or combustion unit that elects
       into Phase I or II
 Preliminary  Draft  for ABAC Discussion  —
Does Mot Represent the Position of the U.S.
      Environmental Protection Agency

-------

-------
           Issue Paper - Reporting and Recordfceeping of        £22
           Continuous Emissions Monitoring (GEM) Data
 SSUE
     What are the appropriate content,  frequency, and form for the
reporting  and recordkeeping  of continuous  emissions  monitoring
(CEM)  data as required  by the  acid deposition control program
provisions of the 1990 Clean Air Act Amendments (CAAA)?

INTRODUCTION

     Section  412  of  the  1990 CAAA  mandates  EPA  to  specify
requirements  for  the  reporting  and  recordkeeping by  affected
sources  of  various types  of  CEM  (or  equivalent  alternative
monitoring system) data for each of their affected units.
These CEM data include:

     • Sulfur dioxide  (SO2) emissions,

     • Nitrogen oxides (NOJ  emissions,

     • Opacity, and

     • Volumetric exhaust  gas flow.

 For  ease  of discussion, the  term  "CEM  data"   will  be  used
  roughout this paper,  although it is intended to include  data from
 Iternative monitoring systems judged to have the "same precision,
reliability, accessibility, and timeliness as CEMs.")

     EPA  needs  accurate  and timely  CEM  data  to fulfill  its
Congressionally mandated responsibilities to:

   .  • Determine  whether  an  affected  unit's emissions  during a
       given year  exceed its SO2 allowances and/or NOX  emissions
       limitation.

     • Implement  the Act's  excess  emissions penalty  and offset
       (Sec. 411) and enforcement provisions (Sec. 414).

     • Ensure  achievement  of  the  annual  10  million  ton   SO,
       reduction and the mandated  reduction of NO  (Sec. 401 and
       Sec. 404-407).

     • Develop numerous  reports to  Congress on  assessments  and
       program  evaluations   of  the  changes  in  air  quality,
       visibility, and acidic deposition effects resulting from the
       S02 and NOX emissions reductions  (Titles IV, VIII,  and IX).

                    PRELIMINARY DRAFT FOR  ARAC
          DISCUSSION DOES  NOT REPRESENT THE POSITION OF
               U.S.  ENVIRONMENTAL PROTECTION AGENCY

-------
     While the statute delineates  the  applicability and scope of
the  CEM  data reporting and  recordkeeping  requirements  quite
clearly,  it'  appears  to leave  much  of  the  specific  content
(including the appropriate  time  resolution for data reports and
records)  as well as the frequency and form of reporting to EPA's
discretion.  Our objective  in  this issue  paper,  therefore,  is to
set forth options for each of these parameters  for consideration by
the Subcommittee for Emissions Monitoring  (Subcommittee)  of the
Acid Rain Advisory  Committee (ARAC).   To  facilitate focused and
fruitful discussion, we will attempt to relate these reporting and
recordkeeping parameters both to the Act's purposes and to EPA's
mandated responsibilities.

DISCUSSION OF REPORTING AND RECORDKEEPING OPTIONS

     Preliminary discussion  during the ARAC meetings  indicates some
confusion may exist about the  relationship between recordkeeping
and reporting in EPA's traditional CEM data collection require-
ments.   To the extent possible,  this paper  attempts  to clearly
delineate  one from the  other.    Under  Subpart  Da New  Source
Performance Standards (NSPS) and subsequent federal and state CEM
regulations, EPA has required affected  sources to maintain records
of pollutant  concentration  and opacity measurements (or approved
surrogates  for  these)  necessary to  assure   compliance  with
applicable emissions standards. EPA has required these sources to
report  only a  small subset of  the   actual  data  generated  and
recorded by CEMs—only those data  necessary to  make a compliance
determination,  assist in  regulatory  development,   or  otherwise
needed  for  implementation  of the  air  pollution  control program.
EPA can require the  sources to submit additional information from
their CEM data records,  however,  when reports indicate potential
noncompliance or when supplemental information is needed.

     The same philosophy  will apply to  reporting and recordkeeping
under the acid deposition control  program provisions of the 1990
CAAA:  EPA will require only those data elements needed to ensure
compliance  with   the   CEM  regulation   and  to   perform  its
Congressionally  mandated   responsibilities  listed  previously.
Emissions-related data not  reported to EPA should be retained by
the sources for a limited period  of time  (e.g.,  three years) for
on-site review by federal and state inspectors and for assessment
and program evaluation purposes.   Records  will be  available for
collection and review through Section  114  authority  under the Act,
as well  as  to the  public through  the  Freedom of  Information Act
(FOIA) process.

     Our  discussion of  considerations for CEM  data  report and
record  contents  (i.e.,  data elements  and the appropriate time
resolution) is organized into three sections around the monitored
pollutant emissions  and conditions:   (1) S02 emissions, (2) flow,

                    PRELIMINARY DRAFT  FOR  ARAC
          DISCUSSION DOES NOT REPRESENT THE POSITION OF
              U.S.  ENVIRONMENTAL  PROTECTION AGENCY

-------
and  (3)   NO   emissions.   Opacity  has  been omitted  simply  to
streamline the articulation of the  key acid deposition precursor
emission  reporting  and  recordkeeping  issues.   Opacity will  be
 onitored and  probably reported to EPA, although  no  compliance
 etermination will be made under the acid rain rule.  (Title IV's
requirement  to monitor opacity  might  possibly be  related  to
visibility for  sources located near national parks or  in Clean
States.)

     In addition to discussing the rationale for various proposed
reporting and recordkeeping requirements, each section presents our
initial thoughts on "strawman" templates for the reporting of CEM
data.  Our  goal is to develop standardized  templates  that will
interface with many (or at least the most advanced)  CEM automated
data acquisition and handling  systems (DAHSs)  and will accommodate
all  the  allowed combinations  of  primary/approved  alternative
system/substitute data methods that  may be employed  to account for
a unit's  SO2 and NO, emissions throughout the year.  Admittedly,
this is an ambitious undertaking as there  are numerous DAHSs and
many combinations of potentially acceptable methods for emissions
data capture and estimation.     .               .

     Our strawman templates focus on the allowed data combinations
for  SO,  and   NOX  emissions,   as   defined  by   the  following
possibilities:

     • CEM/equivalent alternative monitoring systems for measuring
       S02 and NOX pollutant concentrations.

     • Flow  monitor/alternatives  to flow monitor  for converting
       pollutant  concentration measures into  measures of mass
       emissions per unit time.

     • Substitute data methods for  filling  in missing CEM-and/or
       flow data

          - Statistical estimation
          - Parametric methods
          - Other backup monitoring methods/systems
          - Application of minimum data availability threshold (s).

     EPA  is  in  the  process of analyzing whether  CEM DAHS  report
generation  software in use  today  can  be  characterized  into  a
manageable set  of generic classes.   If it  can,  perhaps tailored
versions  of the standard templates could be created for automatic
interfacing with CEM DAHSs.  (EPA is considering the  possibility of
offering  user-friendly, ready-to-install,  PC-based  software  to
facilitate the preparation of accurate  CEM data reports by  sources
who elect to use electronic transfer.)


                    PRELIMINARY DRAFT FOR ARAC
          DISCUSSION DOES NOT REPRESENT THE POSITION OF
               U.S.  ENVIRONMENTAL PROTECTION AGENCY

-------
     A related  issue, involves  the  computation and  reporting of
pollutant emission rates by sources operating continuous emissions
rate monitoring systems  (CERMSs).  A CERMS is defined  to consist of
pollutant concentration  (SO, and/or NO ) monitors  (PCMs),  a flow
rate monitor  (FRM) ,  and a  data recorder to measure and provide
permanent records of SO2 and/or NOX emissions.   (We have identified
at least seven coal-fired and four other utilities that have been
operating CERMSs for a year or  longer.)   The  primary question is
whether and how the DAHS combines FRM data  (in  scfh)  with PCM data
(in ppm)  to calculate hourly pollutant emissions rates (in Ib/hr).
Do the FRM and  the  PCM(s)  have  separate DAHSs or does integrated
system software calculate pollutant  emissions rates directly?  The
answers to these questions could have important ramifications for
the CEM data templates.

     The  last  two  sections address the  frequency  and  form for
reporting under  the acid rain rule which,  most  likely,  would be
consistent across the various types of CEM data.

SO, Emissions

     The reporting and recordkeeping of S02 emissions are pivotal
to the stated purpose of Title IV, "to reduce the adverse effects
of acid deposition  through the  reductions in  annual emissions of
sulfur dioxide  of  ten  million tons ... and,  in combination with
other  provisions of this  Act,   of  nitrogen oxides  emissions of
approximately two million tons ....  It  is  the  intent  of this title
to effectuate such reductions by requiring compliance by affected
sources  with   prescribed   emissions   limitations  by  specified
deadlines,  which  limitations  may  be  met  ...   by  an  emission
allocation and transfer system."

     Implicit in this goal and crucial  to  the success of a market-
based  allowance  trading   program  for   S02  emissions  are  two
principles  that have been deliberated in  previous Subcommittee
discussions and papers:                    ....

     • Need for 100% Accounting  of SO2  Emissions
                                                      (
     • Need for Quality Assured  CEM Data.

In addition, EPA needs accurate and  timely reports on CEM data, as
well as ready  access to supporting  emissions-related records, to
measure progress towards  achievement  of  the  mandated  annual 10
million ton SO2 reduction and to assure  compliance with the Title's
innovative and flexible  SO, emissions source control program. These
requirements underscore  the importance of another principle  that
has been alluded to, but not discussed, within the Subcommittee:

     • Minimization of Errors in CEM Reports.

                    PRELIMINARY DRAFT FOR ARAC
          DISCUSSION DOES NOT REPRESENT THE POSITION OF
               U.S.  ENVIRONMENTAL PROTECTION AGENCY

-------
In the  paragraphs which  follow,  we discuss the  implications of
these  principles  for  the  reporting  and  recordkeeping  of  S02
  issions.
Need for 100% Accounting of S0: Emissions.    EPA needs a complete
(100%) accounting of a unit's S02 mass emissions in tons to develop
a cumulative  annual total for comparison with the  allowances it
holds.  If the tons of SO2 emitted exceeds the allowances and the
unit is not using an alternate  compliance method such as a Phase Z
extension technology or repowering,  the  Act's excess  emissions
penalty  and  offset would  be  applied.    Further,  as  discussed
previously in the Missing Data Periods Issue Paper,  we expect the
accounting of a  unit's annual  S02  emissions  to include estimates
and/or other substitute data during periods when CEM measures are
not available—at least  for units  that do not maintain redundant
CEMs.

     Several  of  the concepts discussed (and,  in  some instances,
agreed upon)  by the  Subcommittee  within  the  context  of missing
data  appear  pertinent  to  the  consideration  of   reporting  and
recordkeeping for SO2 emissions.   These concepts include:*

     • An hour is the appropriate time period for the accounting of
       CEM data.

     • Use  of  an  incentive-based  approach  with  data  capture
       thresholds would encourage a smaller amount of missing CEM
       data and,  thus,  provide a more  complete record  of actual
       measured emissions.  Examples of threshold values included
       90* and 95% of total unit operating hours per year.

     • An  estimate  of  "probable  actual  emissions"  based on  a
       statistical evaluation of available CEM measurements would
       be used for filling  in missing data above the threshold.

     e Alternative  approaches  being considered  for filling in
       missing data  would depend on the length of  the  data gap.
       Statistical   methods , for  estimating   emissions  appear
       satisfactory for short data gaps of a  few hours1 duration.

     • The  frequency  of  data  gaps  could  also   influence  the
       determination  of acceptable methods for  filling in missing
       data.

These  principles   for  S02 emissions accounting   suggest  the
following:  an hour would be an appropriate time increment for CEM
data recording; and reports should contain sufficient information
on CEM data gaps and the  methods used to fill them for EPA to check
the validity of substitute  data when appropriate.


                    PRELIMINARY DRAFT FOR  ARAC
          DISCUSSION  DOES NOT REPRESENT THE POSITION OF
               U.S.  ENVIRONMENTAL PROTECTION AGENCY

-------
Need, for Quality, Assured CEMData.    The Subcommittee  reached
general consensus on the  importance  of accurate CEM; data for SO2
emissions accounting  and  agreed  to maintain  the basic  quality
assurance (QA)/quality control (QC) procedures of NSPS subpart Da
for the acid rain rule.  Two of the concepts discussed within the
context of quality assuring CEM data appear germane  to reporting
and recordkeeping:

     • Accuracy  of  CEM data  should be  certified daily.   Using
       certified or Protocol  1 gases for daily calibration drift
       (CD)  tests would accomplish this.

     • CEM data recorded during periods when the monitor is deemed
      . out-of-control  (OOC) by daily CD tests are not valid for
       emissions accounting.   Thus, CEM OOC  periods become missing
       data periods.

These QA/QC principles suggest the following:  a day would be the
appropriate time resolution for CEM data reporting; reports should
contain sufficient information on  daily CD tests and OOC periods
for EPA to confirm the validity of  CEM  data; and reports should be
filed periodically, say, on a monthly or quarterly basis, so that
any apparent discrepancies noted in  EPA's checks can be reviewed
with the source and resolved in a timely manner.  Periodic reports
would also facilitate the orderly administration of the Acid Rain
Allowances and  Emissions  Data Systems and  the  sharing of timely
information  with  the  regulated  community  and other potential
stakeholders in allowance trading market.

Minimization  of  Errors  in  CEM  Reports.    Another  pertinent
observation  for reporting  and recordkeeping  that  has surfaced
during previous Subcommittee discussions is the surprisingly high
incidence and severity of computational errors plaguing calculated
data  in  source-generated  CEM  reports.    This is  one  of the major
conclusions  of a  study by Entropy  Environmentalists,  Inc.,  of
quarterly relative accuracy test audit  (RATA) reports submitted by
Subpart Da sources  during 1988 and 1989.   This study, which was
reviewed by both EPA and the Utility Air Regulatory Group (UARG),
indicates that EPA should require the reporting of actual measures
(e.g., ppm SO,, ppm KOx, scfh exhaust flow rate, etc.) comprising
the calculated units of the emissions standards  (i.e., tons/yr SO2/
and  Ib/MMBtu NOJ—at  least  initially until the  accuracy of the
sources1 computational  procedures  can  be confirmed.   The Entropy
analysis shows that 24% of the reported CEM  relative accuracy (RA)
values contain significant computational error  and 8% deviate from
the correct value by one-third to over 300%.

     Moreover,  the  risk of computational  error is substantially
higher in the algorithms for combining  CEM pollutant concentration
measures with volumetric exhaust gas flow measures (or equivalent

                    PRELIMINARY DRAFT FOR ARAC
          DISCUSSION DOES NOT REPRESENT  THE POSITION OF
               U.S.  ENVIRONMENTAL  PROTECTION AGENCY

-------
                                                          70
alternative measures) to yield estimates of SO2 mass emissions per
unit time.  The computation of CEM RA involves the addition of two
 eras, divided by a third term; whereas  the computation of tons of
 O2 emitted per year involves  the use of two different equations
(depending on whether the pollutant concentration and flow measures
are taken on the same or different moisture bases) with as many as
five multiplicative terms.

      A related objective  is  to  reduce, and hopefully eliminate,
computational and other software-generated  errors  in reports and
records  from DAHS1  computers integral to  CEM systems  (GEMS).
Experience shows that CEM  data produced by the most accurate and
reliable  pollutant  monitors   are  meaningless  if  the  emissions
information  is  processed  incorrectly.   Typical  computational
blunders  and  software  misspecifications which EPA  and  state
agencies have observed include:  use of "dry-basis" algorithms to
calculate emissions in terms of the standard  (i.e., Ib/MMBtu) from
in-situ or "wet-basis" monitors; and substitution of the average of
30 discrete  24-hour averages  for  a  true 30-day rolling average
(i.e.,  the average  of  all valid  hourly data  over the  last 30
operating days).           '

     Developing   report   template   software    that   interfaces
automatically with common CEM  DAHSs should reduce the incidence of
such errors.   Another  approach, which  EPA is  evaluating,  is to
develop  DAHS certification test and  audit procedures  designed
specifically to uncover incorrect algorithms and other programming
 rrors.   Some states have reported  finding major software errors
 uring audits (See Document E23).

Strawman  Template  for  SO^ETnisslons-.JtepQrting.   Table 1 presents
our  initial thoughts  on  how the principles   for  SO,  emissions
reporting  discussed  above might be combined  into  a standardized
template.  One of the notes to this table refers to a sample Header
Report, which is displayed as  Figure 1.  Insofar as possible, EPA
hopes
-------

-------
                                                                                                                                       '   i
i
a.

w


£
o
&

S
CO
to
w>

S
ui


tf
• *
!J
? C
S uj

if
*
PI
I™
SO2 1
Emisslont 1
|
1


Substitute M<


•

e
e
1
q
e
e
e
M
n

zt
M>
i
i
e

q
A
£
* *
1

§


I

 S
                                                                                                                c

                                                                                                                3
                                                                                                                o
           i

           i
            a

           I    *

           tf   I
                                                                                                                                 •

                                                                                                                                H
                                                                                                                              *
                                                                                                                                . it  *
                                                                                                                                 s  o
                                                                                                                           ~ «*<•»
I

-8


-------
 IU
 i
i

-------
          (3)  Daily calibration checks,  daily  zero and span drift
               checks, adjustments, and  maintenance  performed on
               CEMS                          \

          (4)  Periods when CEMS is inoperative

          (5)  Hourly:   pollutant  concentration  (ppm),  diluent
               .concentration  (t 02 or  C02),   exhaust  flow  rate
               (dscf/hr), calculated Ib/hr  (SO2).

          (6)  Daily:   fuel characteristics  (average sulfur and
               heat content) and usage (tons or gallons per day);
               to provide periodic  independent  check against CEMS.

It  has  been  suggested  by  some  that  EPA consider developing
standardized templates for  CEM  data records (as well as CEM data
reports) so as to ensure the availability of consistent and useable
short-term  emissions  data for the Congressionally mandated acid
deposition  control program  evaluations and assessments which may
require or benefit from such data.   (Attachment A of.a Memorandum
to  the  Subcommittee  by  Richard Poirot,  Vermont Department  of
Environmental  Conservation, on  Time Resolution for Reporting CEM
Data, dated February 19, 1991, lists these  studies.)

Flow

     Traditionally,  CEM  data  has  been  reported in  pounds  of
 ollutant per unit of heat input (Ib/MMBtu), ppm, or micrograms per
   ic meter.   Under Title IV,  however,  volumetric  flow data are
required  to compute each  unit's S02  emissions  in tons/year for
comparison with the allowances it holds.  Such data are also needed
for levying the  Title's excess emissions  fee  and offset against
units who do not have "balanced books"  at the  end  of the year.
Flow data would  also be required  to  convert  a unit's  annual NOX
emissions  rate into  tons,  should the  Act's  penalty provision
(expressed  as  $2000/excess  ton)   apply.    In  addition,  units
participating  in the annual NO, emissions averaging pools allowed
under Sec.  407 (e)  may  also need to develop estimates of NOX mass
emissions in tons.

     As  in  the accounting  of a unit's  annual SO2  emissions,  we
expect  the  accounting  of a unit's  annual  volumetric exhaust gas
flow to  include  estimates and/or  other substitute  data  during
periods when  flow  measures are  not available.   The Subcommittee,
however, has not yet addressed the issue of missing flow data.

     The  principles  enunciated  under  SO, emissions  for  100%
accounting, accurate and quality assured data,.and minimization of
errors   would   also  appear  to  apply   to  the  reporting  and
recordkeeping of flow (or alternative  methods to convert pollutant

                    PRELIMINARY  DRAFT  FOR ARAC
          DISCUSSION DOES NOT REPRESENT THE POSITION  OF
               U.S. ENVIRONMENTAL PROTECTION AGENCY
                               10

-------
concentration  measures  Into  mass  emissions  measures).    The
alternative methods  to flow monitors   that have  been mentioned
informally within  the Subcommittee  (e.g., heat  input, feedwater
flow rates, steam flow rates, etc.) would add greater complexity,
however, to standardized reporting templates.  These alternatives,
although  not  shown  as yet  to  be  equivalent  to  flow monitors,
generally employ different 'units of measure than  flow monitors and
involve different computational procedures for the determination of
SO2 mass emissions  per unit time.  Heat input,  for example, "which
is  typically  measured  in MMBtu/hr, would  be combined with SO2
pollutant concentration measures on a Ib/MMBtu, not ppm, basis to
yield S02 mass emissions per hour.     ;    ';'

     Table 2 is a strawman template for the reporting of flow (or
equivalent) data.  This template and the strawman template for SO2
emissions, shown previously as Table  1,  illustrate  one approach
that would accommodate the reporting of the heat  input alternative
method, although it has not^ been demonstrated to be equivalent to
CEM and flow monitor systems for estimating SO, mass emissions per
unit time.

     Supporting records to  be maintained by the sources for flow
reports might include:

           (1)   Stack inside diameter (ft)

           (2)   Calibration  checks, adjustments, and maintenance
               performed  on flow monitor

           (3)   Periods when flow monitor is inoperative

           (4)   Hourly:  average  stack  temperature  (°F), average
               gas   exit  velocity   (ft/sec),   and   stack  gas
               volumetric flow  (scfh).

           (5)   Daily:  average stack gas moisture  content  (%}.

Some of  these data elements  (i.e.,  Items 2, 3, and  parts of 4)
would not be germane  for certain alternative methods that have been
mentioned.

NO  Emissions

     The reporting and recordkeeping of NOX emissions are obviously
important  to  the Act's  stated  purpose, quoted  previously.  The
accounting  of KOX emissions  may  not  need  to be as  stringent,
however,  as  that described for S02 emissions  for the  following
reasons:
                    PRELIMINARY DRAFT FOR ARAC
          DISCUSSION DOES NOT REPRESENT THE POSITION OF
               U.S. ENVIRONMENTAL PROTECTION AGENCY
                                 11

-------




(

tt



*






UJ
Q.
UJ

.
§
UJ
C

o
Zj
^»
i (
e
«
I
M

o
h-








(
1



1
b


«
5

5
J^
2
0
2


z
*








•1
V







$
?
0
|


fm
3
i
.
P;
1




•u

•5
w

3*

7
3
CO




i
:
^B

^
3
e


fcS
5 •
J5 <

2;
Z j

t
M •
U m
S J
WJ j

4
Si
!1
»-t
fi
sl
D
0


flg

**<
S
^
5


•b
i


GL
3
i
a
CO
CO
<
ut

te

§


9



-"S
£2

V
3
P


D
•
i ^
• u


fi

>o
5
E


e
e

e
e




0



0
^



Z


o
"





z
1



o
8



5


p



i.
e

e
e

e
o




Ci



o
d



Z


O
M





z
i



o
8



5


q



1
e

e
o

e
e




d



e
o


<
Z


o
CM





Z
1



e
8



5


e



I
a

o
e

o
d




n



a
a



Z


o
«





Z
t



e
8



5


e
5



1
O

O
e

e
e








0
o



Z


o
d





Z
1



O
8



0


S



m
p
O
















































e
0

e




m
.*''


' 0
*



«s»


e
A




^.
z*
i



o
§
^


o
8
*

g
l»

e
t
* *
H









'








-
























•




e
d

e
d




d



a
cu


'
Z


O






Z
J



O
8



|
^

e
en



i.
s

o
e

e
d




.



e
o


;<
Z


O
(O





Z
i



a
8



|
^

e



I
0
















































e
6

q




5



o
S



3E*


e
U)
S




^.

1



e
o
o



g
o

e
«


^«
**
i

o
I
 S
la
=
 S
i
I
 =
s R
i
7
                                                  a
                                                  x

-------
     • There  is  no explicit requirement  for  a cumulative annual
       total of NOX emissions from each affected unit.  ,

     • The  emissions  allowances and  transfer  system does  not
       presently extend to NOX.

     • The NOX emissions source control program adheres generally
       to  EPA's  traditional  method  of  specifying  applicable
       emissions standards.

     The  treatment of  missing  data  for NOX  emissions may also
differ from that prescribed  for  S02 emissions.  (The Subcommittee
has  not  formally discussed  the  issue of missing  NOX  CEM data.)
Since the statutory limitation is expressed as an annual average,
it appears that some  consideration  may be given to the impact of
discrete (small) data "pieces" on the overall average.  It may be
argued that a small amount of missing data, say, 5% or less could
be tolerated for NOX emissions for units not participating in the
annual emissions averaging pools  allowed under Sec.  407(e). On the
other  hand,  some  estimating rules  will be  needed for  all NO
emissions and to gauge progress towards achievement of the annual
NOX reduction.

     Table  3  is  a strawman template for  the reporting of NO,
emissions.  While  it  closely resembles the strawman template for
SO2 emissions,  shown previously as Table 1, there are two notable
exceptions:

       •  Average  daily NOX emissions  must  be calculated on  a
          Ib/MMBtu  basis,  corresponding  to   the  units  of  the
          standard, as well  as reported on a ppm basis.

       •  No requirement exists for the calculation of total daily
          NOX  emissions  in  Ibs  or tons.   (This  exception will
          probably not  apply to units averaging  their emissions
          with others.)

Table  3  contains  a  column for  measures   from  an  equivalent
alternative  monitoring   system   (EAMS)   although   no  candidate
alternative  systems  or  methods have  been  identified  for  NOX
emissions.

     Supporting records  to be maintained by  the sources for NOX
emissions reports would probably  be  similar to those listed for S02
emissions.

Frequency of Reporting

     As discussed under SO2 emissions, periodic reports—probably
on a monthly or  quarterly .basis—are  essential to  the  orderly

                    PRELIMINARY DRAFT FOR ARAC
          DISCUSSION DOES NOT REPRESENT THE POSITION OF
              U.S.  ENVIRONMENTAL PROTECTION AGENCY
                             13

-------
§
a

I
QC
O
a.
I


*

f!
-
5
I


i

i
s.

•5
|
to








»
? §
is
5


•






1


eo
'•o
8*
i
i
s

<3
§
E
£


3

£
S
to
CO
i

M
s
UJ
O
3
5
S
S

cL
u

4
I
E
e
o
s
t
e
e
o
d

o
o



*
o
^




z


o
co
CM
d


o
S

o
CM
i.
a
o
o
5
CM
e
d
o
d

o
o




e
CM




z"


o
CM
CM
in
d


o
S
o
e

1
o
o
o
CM
pi
o
o
e
d

o
o




e
^




Z


e
n
CM
5
d


o
S
«
0
CM
|
O
a
*
CM
O
O
d
op

e
o




o
CM




Z


CM
^
""
d


e
S

o
5
|
o
o
CM
5
o
d
p
CM

O
O




o
o




z


in
^
"
v
d


o
2

to

I
o
































•

•9
a>
1
0
wi

o
o




«
f*.




z*


CA
0
CA
in
b


e
CM
r>
e
S
e
i
K


































'~0
O
CM
n
o
d
o
d

o
d




e
e







o
^
CM
(A
d


o
3
o
0

i
o
o
d
?
d
e
o
d

o
0




e
^




z*


a
n
CM
^
d


o
S

o
CM
1
e

































•

e
6
1
e
e
o
6

o
o




in
n
^



z"


in
CO
(0
tn
. e


e
CM

0
CM
u.
1
H
                                            Q

                                            J

                                            2
                                            T3
                                            I
                                            £
                                            M
?
1

i
1
                                                 *H
                                                 418
                                                 ill
                                                 £ii
                                                    to
                                                 fill


                                                 I«I<
                                                 •: N *"2
                      14
                                              I
                                                          e>
                                                          j

-------
 administration  of the  Acid Rain  Allowances  and Emissions  Data
 Systems and  to the  timely resolution of  apparent  discrepancies
 (noted in EPA's data quality checks) with the  sources.   Monthly
 reporting  would  lead   to  more  efficient   and  cost/effective
 administrative  processing  at EPA,  primarily by avoiding periodic
 (albeit predictable)  "crunches"  of processing  relatively  large
 batches of CEM data.   Also,  monthly reporting  would  enable the
 sources'  CEM  operators  to  keep  current  with  the  reporting
 requirements and minimize time spent "re-learning" the process.  On
 the  other  hand, quarterly  reporting is used  in most  state CEM
 programs and,  as  such,  represents the de facto  standard.   A few
 programs  (i.e., Connecticut,  Washington,  and  California - Bay
 Area),  however, already have  monthly  reports  and others  have
 indicated that monthly reporting would be preferable  to the current
 practice (See Document E23).

 Form of Reporting

      EPA wants  to encourage and facilitate the use  of electronic
 reporting, via either a PC-floppy diskette or a modem with a toll-
 free "1-800" telephone number, since this form has proven to be a
 feasible and efficient method for ensuring timely and accurate CEM
 data in several state programs.   EPA is considering offering user-
 friendly,   ready-to-install,  PC-based  software  containing  the
 standardized report templates (with built-in data edit checks and
 HELP screens)  to assist the sources in preparing accurate automated
 CEM data reports.  States could use this software as  the "core" for
 their  CEM  reporting  requirements,  adding  and/or modifying  data
 elements  as  needed.   User Manuals, Easy  Reference Cards,  and
 program documentation would be included with the software, possibly
 to be packaged as "self-installation" CEM data report kits.

      We have discussed this approach with several states, and they
 enthusiastically  endorse  the  concept.    Further,  we intend  to
 solicit recommendations from  different kinds  of  sources   (with
 varying levels of CEM experience)  on the types of user interfaces,
 data edit  checks,  software  interfacing protocols, etc., that would
 be  most  helpful.    Any  suggestions  made  by  members  of  the
, Subcommittee or other ARAC participants would  be most welcome-
 particularly if they come within the next few months as we develop
 functional specifications  for the  EPA  Acid Rain Allowances and
 Emissions Data Systems.

      Figure 2,  Acid Rain Bulletin Board, depicts one  method for the
 electronic transfer of automated CEM data reports from the sources
 to EPA.  Although EPA's Acid Rain Emissions Data System, like the
 Aerometric Information and Retrieval System (AIRS),  will probably
 reside in  a mainframe  computer,  PCs would  be  used  to "buffer"
 emissions data  submitted via modem and  to  process data on floppy
 diskettes.   The PCs  could employ  an  electronic bulletin  board

                    PRELIMINARY  DRAFT FOR ARAC
           DISCUSSION DOES NOT REPRESENT THE POSITION OF
                U.S. ENVIRONMENTAL PROTECTION AGENCY


                            15

-------
                                    2_-;-.
                                    Tf. • •*.-*--
                                 :-Tv£U
5  E  g
 (0  O

-------
system to log receipt* of data  and perform minimum QA checks before
uploading to the,mainframe system database.

     Under this scenario, the mainframe software could produce an
"echo" report of the  emissions data  and header  information as
received from each source, QA'd, and tabulated for entry into the
EPA Acid Rain Emissions and Allowances Data Systems.  The reports
would  then  be sent  back to  the  sources,  either  electronically
through the bulletin  board or transmitted in hard copy form through
the mail, for review  and verification.  Such an "echoing" procedure
would not only eliminate electronic transcription errors, but also
assure the sources' concurrence with data in EPA's databases.
                    PRELIMINARY DRAFT FOR ARAC
          DISCUSSION DOES NOT REPRESENT THE POSITION OF
               U.S.  ENVIRONMENTAL PROTECTION AGENCY
                            17

-------
                                                                              £23
            State Agency Experience in Data Recordkeeping and Reporting
The following draft chart summarizes information from state agencies on various CEM
reporting and recordkeeping issues.  These issues include the time resolution of data in
reports, the frequency of reports, the use of hard copy or computerized data, the use of
automated data handling systems, the degree of errors in reporting, and the use of flow
monitor data.  The chart will be updated as further information becomes available.

-------

-------
' £

 2*


£ S
        1
       ?i
   ill
   I |i

   ill
    in

    s
    U
    •i
    M
    fl

    55

    US

    I
    V
     I
 H'lllfl'
 •Nil1 illl

 iJllSllflJ
   n
?!
i;
        If
         8
        - O
         I
             II
   i;
I
i||ti|


iliii!
                           «/ 6
                                        « •»
                     ^ • *
                     In
                       il
                          I
                  S||





                  fill I
                  x> . B -1
                                  l
                          |
                                        D
                                        i
             •E
             ili

            !l!
                il
                •
                                  I
             I]


             !
             JS-L.
                        Hi
                        X^J
                        3 . . I
I '



IBS
TS |fi

111

                                i
                                     I


                                     I
                          ill
                          I e i

                          i.
                  fi
                       I
                                  II
                                         1
                                 2 B frT5
!|1




III
                                  Hi
                           8

-------
-» g a.
  a
                         Ml
                          l 11 ill

                        l 1
« Q

II
                                 li
                                  a
ndcipit
An
lu
or by

-------
 "8
-i a
to

S
|
i
      !
u


i
fl
i
    I1
    -If
          I
            liili
               1

                     i
             III
                   I
                   jl
                   If

                        is
                        BJ(
                        3 9
                         ll
                        i

11 n
I
                         1
                          i
                          JB-
                            a  e

-------
       Baseline Tons of SO, (1985) for Selected Categories of Affected Unit* Under Tttte IV
Category of Units
Alt units with nameplate capacity
<7SMW • . .
Units with nameptate capacity
<75MW and baseline emission rate
>l.2lb/MMBtu
All oil- and gas-fired units
(<50% coal-fired)
Oil- and gas-fired units consuming
< 10% oil per year
Number of Units
639
194
859
525
Number of Tons SO,
665.569.72
575.943.94
592.258.00
7626.36
Source: National Mlowane* OaubaM
                            Preliminary Draft for ABAC Discussion
          Does Not Represent the Position of the U.S. Environmental Protection Agency

-------
                  Presented at  1983 Engineering Foundation Conference
                                Pocono Hershey Resort, PA
                              COAL SAMPLING AND ANALYSIS
                  AS AN ALTERNATIVE TO CONTINUOUS EMISSION MONITORING
                                    IN PENNSYLVANIA
                                   Joseph C. Nazzaro
                         Department of Environmental Resources
                                    Harrisburg, PA


   On August 1, 1979, the Pennsylvania Department of Environmental Resources
   promulgated regulations which implemented the EPA requirements of 10/6/75 (40 FR
   46240} concerning the preparation, adoption and submittal of implementation
   plans for emission monitoring of stationary sources.  The regulations included
   requirements for installation of SO. continuous emission monitoring systems
   (CEMS's) on solid-fossil-fuel-fired combustion units having greater than 250.mm
   Btu/hr heat input (83 sources, representing approximately 75% of all SO.
   emissions in the State).  Also included was a provision allowing the Department
   to use the data from the CEMS's to determine compliance with SO.
   emission-averaging standards for daily and running 30-day time periods.
•to
In addition to general regulatory requirements, three references were cited
  Oich provided the specific requirements.  Two of the references (40 CFR 51 and
   CFR 60) existed as parts of the Code of Federal Regulations.  The third
reference, the Continuous Source Monitoring Manual (Manual) was developed by the
Department and contained the administrative procedures for obtaining approval of
monitoring installations, additional performance testing requirements,
recordkeeping and reporting requirements and quality assurance requirements.

In response to "industry requests, the Department developed general requirements,
design and performance specifications, performance test procedures, submittal
requirements, recordkeeping and reporting requirements and quality assurance
requirements for coal sampling and analysis systems (CSAS's) to be used as
alternatives to SO- CEMS's.  On April 21. 1981, these items were added to the
Manual and the Department's regulations were amended to allow use of CSAS's to
meet regulatory requirements for SO. emission monitoring.

-------

-------
                                                                                110
           COAL SAMPLING AND ANALYSIS AS AN ALTERNATIVE
             TO CONTINUOUS EMISSION MONITORING SYSTEMS
                             IN PENNSYLVANIA

                               INTRODUCTION
           The request to allow use of CSAS's. in place of CEMS's raised three
 main concerns in DER--

      l.   The data would not be real-time and, thus, of little use in managing
           emissions to avoid violations of SO2 standards.

      2.   Such systems would be difficult to audit for continued accuracy and
           repeatability.

      3.   EPA, while indicating that CSAS's could be used, "reserved" the section
           of the Code which would provide the requirements for CSAS's. Although
           ASTM procedures for obtaining gross samples of coal existed, they
           were not directly applicable  to a CSAS required to produce data for  ,
           direct comparison with a stationary combustion source SO2 emission
           standard for enforcement purposes.

           Upon consideration of the first two concerns, it was decided that the
 increased response time and difficulty in auditing would be acceptable only if
 CSAS's could provide data that could be applied to the emission standards in a
.manner identical to CEMS's.

           With respect to the third concern, it soon became evident that some
 design and performance specifications would have to be developed and other,
 existing specifications modified in order to serve the Department's purpose.


                          DESIGN SPECIFICATIONS


           The decision to develop design specifications in addition to performance
 specifications was made in order to avoid a "black box" approach to CSAS's. Adherence
 to general design criteria, it  was hoped, would increase confidence in  the CSAS
 repeatability and would allow inspectors to detect system changes that could
 affect performance.  The design specifications cover the general areas of (1)
 sample acquisition point location, (2) sample collection technique, and (3) sample
 preparation and analysis.

           The sample acquisition point location criteria are directed at obtaining
 a sample representative of the coal actually burned  in the combustion unit, rather
 than an estimate of coal quality as received at the plant. This requires  that the
 sample acquisition points be  located as close as possible to the point at which the
 coal is burned.  Alternative locations are allowed only upon demonstration that
 the system results are still representative of the coal "as-fired."

           The sample collection technique criteria are designed to provide
         that can be related to specific time periods of SO2 emissions in order to
      a comparison of all CSAS results with reference method SO2 sampling. ASTM

-------
sample type, condition and spacing criteria, as well as general ASTM equipment
criteria, are specified.  Additionally, in order to make possible combining of
samples from individual units to represent combined emissions (as when several
units discharge to a common stack), proportional sampling is required. While it is
assumed that most users will prefer automatic sampling techniques, a proportional
sampling strategy for manual sampling is also specified.


            Sample preparation and analysis criteria are specified in accordance
with ASTM procedures with alternatives allowed upon demonstration of equivalent
results.
               design specification details, see the "Initial Application (Phase I)"
section of the Manual which is reproduced in the Appendix.

                       PERFORMANCE SPECIFICATIONS


           In addition to the design specifications, the following performance
specifications were developed in order to define the acceptable level cf CSAS
accuracy, repeatability and reliability.

      Sampling;

      1.    Number of subincrement point samples per hour per point of sample
           acquisition.  (See  Appendix for terminology specific to "as-fired"
           CSAS's)

           The number of sample increments was arrived at by application of the
           criteria in Table 2 of ASTM  D223A
iABlC, 2 ^•H
Top Sid

'» in. II* mmi

' 10. 1)0 mmi

• m. 1 1 M mmi
MovnuioilT Cleaned Corf
Mimrnum number of mer*
Muumvm *«f M of ineren
Minimvm »«if M of Mown
menu U
•«u.i» :
•ML If I
1)
•
J
1)
IS
1
IU» (tncfeiMd Coalr
Minimum number of incrv
Minimum xif hi of mcren
Minimum veifln of inertn
mnt% n
tenu.le> I
•em*. k| 1
15
M
T
            Under cnndiitom C  clumd. u may
         mil tho» fnn »anuiion brcaute uf beinf a Mend uf i»o di(Teren« puniont of one turn or a Mend of two different team*.
         to weft caaev the nwmocr of tncremcnu »ho»ild be aa iFectficd for ra» lunckanodl vo»l.

-------
 2.    Weight of hourly increment point sample.

      The criteria of the above-referenced Table 2 was applied, resulting in
      the specification of 2 pounds per increment.,

 3.    Variation of actual factor of  proportionality for daily composite unit
      samples (seven days, individual unit).

      In order to define acceptable CSAS repeatability and prevent significant
      bias in the 30-day average calculated SO2 emission rate, the ratio of
      coal sampled to coal burned for each daily  time period must be within
      20% of the average ratio as determined during a 7 .day test period.
      The 20% specification allows for a 10% actual variation with a 10%
      weighing error.

 ft.    Variation of actual factor of  proportionality for daily composite unit
      samples (daily, all units within system).

      In order that the daily composite system samples are not significantly
      biased by improper weighting of,certain daily  composite unit samples*
      the ratio of coal sampled to coal burned for each unit in a system
      (more than one unit discharging through a single stack) must be shown
      to be within 20% of the average ratio for all units in the system.
 Analysis;

 Specifications listed in the appropriate ASTM. standard methods ~

 5.    Precision of sample preparation.

 6.    Calibration error for percent sulfur analysis, dry basis.

 7.    Precision of analysis for percent sulfur, dry basis.

 8.    Standardization of calorimeter water equivalent.

 9.    Precision of analysis for Btu/lb., dry basis.

 System Operation:

10.    Response time of system.

      In order to allow identification of problems with coal quality and
      possible corrective action, the CSAS must yield final results within
      168 hours of completing the collection of the daily sample.

11.    Operational period of system.

      In order to encourage a high reliability of system components, the
      CSAS must operate for at least 16S hours without any corrective
      maintenance.

12.    Relative accuracy of system Ibs. SO2/106 Btu results.

-------
           In order to qualify as an acceptable alternative to CEMS's for direct
     i      comparison of data to SO2 emission standards, the CSAS must meet
           the same relative accuracy specification as is required for CEMS's.

                            TEST PROCEDURES

           While existing ASTM procedures were available for demonstration of
compliance with the performance specifications  for the analysis portion of the
CSAS, procedures appropriate for demonstration of compliance with the performance
specifications for the sampling and system operation portions had to be developed.
For details of these procedures, see the "Performance Testing (Phase II)" section
of "the Manual reproduced in tne~A"ppendix.

           The relative accuracy performance specification test for CSAS's is
identical to that used for CEMS's. A  minimum of nine  1-hour reference method
source tests must be conducted for comparison with CSAS data which must be
collected for corresponding hourly time periods.


           While the performance specification  tests are designed to verify the
operational parameters of the CSAS initially, periodic testing was deemed necessary
in order to evaluate the operation of  the CSAS and validity of the resultant data
over time. Appropriate performance testing is required whenever corrective
maintenance is performed on the sulfur analysis, heating value analysis, sample
acquisition or sample preparation portions of the CSAS. In addition, the Phase II
performance specification testing, with the exception of relative accuracy testing,
is required to be conducted once during every  calendar quarter.

                          ACTUAL  APPLICATIONS

           While several different approaches were proposed by companies in
attempts to satisfy the Department's CSAS requirements, only one type of CSAS
has been demonstrated to conform to all design and performance criteria or
equivalents, where permitted.  Several unsuccessful proposals were denied mainly
due to inability to represent SOj emissions for known, discreet, hourly time
periods.  To date, no company has proposed a manual sampling program as outlined
in the Department's Manual.

           The Pennsylvania Electric Company (Penelec) Automatic "As-Burned"
Coal Sampler System (PACSS), patented by Mr. Al Slowik of Penelec, has undergone
successful demonstration of compliance with design and performance criteria at
six different utility stations on a total of  16 generating units which have a total of
nine emission points (due to instances of several  units discharging through a single
stack).

           The sample acquisition portion of the PACSS consists of a stainless
steel probe located so as to convey pulverized coal from a coal pulverizing mill
exhauster to a cyclone collector which discharges to a coal sample can, as shown
in Figure 1 of the Appendix. The coal is collected periodically, at least two
samples per hour, by automatic, timed operation of a pneumatic pinch valve which
allows or prevents coal flow in the probe. When  coal is not being sampled, high
pressure instrument air is supplied to back-purge the probe, thus reducing plugging
problems.

-------
D.


E.



P.


G.
 18.
19.
20.
21.
      Analysis of each laboratory sample  for BTU/lb.  must be conducted  in
      duplicate using ASTM 02015-77(78) or methods which produce equivalent
      results.  Results must be recorded on  a dry basis.

      Analysis of each laboratory sample for percent sulfur must be conducted
      !n duplicate using ASTM 03177-75, Method B- Bomb Washing Method or
      methods which produce equivalent results.  Results must be recorded on
      a dry basis.

      Analysis of each  laboratory  sample  for percent  moisture must  be
      conducted in accordance with ASTM 03173-73(79).

      Results for each laboratory sample  must  be converted to tb. SO2/106
      BTU using  the average values of percent sulfur and BTU/lb. from the
      duplicate analysts as follows:

      S = (Sa * S2)/2
      H « (Hj * H->)/2
     Where
                Sj s first measured value of percent sulfur
                §2 = second measured value of percent sulfur
                S = average of Sj and 5*
                HI « first measured value of BTU/lb.
                Hj ~ second measured value of BTU/lb.
                H « average of HI and Hi
                      . SO2/106 BTU     . •    •
22.   The calibration error  with  respect  to  percent  sulfur analysis must  be
      checked at minimum every seven days using .either NBS SRM I632a or
      NBS SRM I631a-midrange.

23.   The value of  the  calorimeter  water  equivalent  must  be, checked  at
      minimum every seven days using ASTM D2Q15-77-{78), Section 7.

The claimed performance specifications as listed in Table V (will be verified as
part of Phase II).

Process and pollution control equipment  operating parameters which affect
the SO? emission level, along with an  explanation of the method to be used to
record these parameters.
Calibration,   operational,
recommended schedules.
                           and   maintenance   procedures,   along   with
An explanation of  procedures to  be  used .to  satisfy  the Department's
requirements as listed in the "Record keeping'and Reporting" section of this
manual.
                              a

-------
              = rated coal burning capacity of base unit (Ibs. coal burned/hr.)

         Nm s maximum  number of sample  acquisition points for any unit
               within the system.

         NOTE:  Record Fo retaining two significant digits.

    d.   Calculate  the  subincrement   point  sample  weight   for  each
         combustion unit within the system by the equation
    Where Wjs subincrement point sample weight for unit i (Ibs.)

          F0 - ideal factor of proportionality (Ibs. sampie/lb. coal burned).

          Cj0 = rated coal burning capacity of unit i (lbs./hr.).

          Nj s number of sample acquisition points for unit i.

          NOTE:  Record Wj retaining two significant digits.
e.
          At the beginning of each discrete hourly time period* determine
          the number  and spacing of subincrement  point  samples  to  be
          collected at each point of sample acquisition for each combustion
          unit within the system according to the following table.
                     Number of Subincrements          Spacing
     0.00 toO. SO                2                       30  minutes
     0.51 toO.75                3                       20  minutes
     0.76 lol.OO                4                       15  minutes

          Where Cja =  actual coal burning rate  for unit i anticipated for me
                       hour (Ibs. coal burned/hr.)

          Cj0 s rated coal burning capacity for unit i (Ibs. coal burned/hr.)

     f.    Collect the samples according to the specified weight, numbers,
          and spacing*.

16.   Each daily  composite unit sample must be weighed prior to combining, in
     accordance with  all  quality  assurance  criteria,  to  form  the daily
     composite  system sample.  All data  necessary to calculate the actual
     factors of  proportionality (Fja) for daily composite unit samples from
     each individual combustion unit  within the system (i.e.t  the  weight  of
     each daily  composite unit sample and the  weight of coal burned in the
     unit during the same daily time period) must be recorded.

17.   Preparation of a 50-gram laboratory  sample from each daily  composite
     system sample  must be  conducted in accordance with  ASTM 02013-72
     (78) as for Group B samples.

-------
 9,    An hourly increment system sample, which consists of hourly increment
      unit samples for all combustion units which discharge to a common flue
      during a  particular  discrete  hourly  time  period, must accurately
      represent the actual SO2 emissions from the flue for that time period.

 10.   A daily composite unit sample must consist  of all subincrement  point
      samples collected for a particular combustion unit during a discrete daily
      time period.

 11.   A daily composite system sample, which consists of daily composite unit
      samples for all  combustion units  which discharge to a common  flue
      during  a  particular daily  time period,  must accurately represent the
      actual SO2 emissions from the flue for that time period.  Combination of
      daily composite unit samples to form  daily .composite system samples
      must be conducted in accordance with all  applicable quality assurance
      criteria.

 12.   For sampling of coal streams other than fSuidized.  pulverized coal, each
      subir.crement  point sample  must consist of  a Type 1, Condition A or 8,
      Spacing 1 sample as specified in ASTM 02234-76.

 13.   For  sampling of  fluidized.  pulverized coal,  each subincrement point
      sample must consist of a Type I,  Condition A, B, or C, Spacing 1 sample
      as specified in ASTM 02234-76.

14.   Subincrement  point samples must  be  collected  in  proportion  to the
      weight of coal passing the  point of sample acquisition during the time
      period represented by the samples.  The factor of proportionality tlbs.
      sample/ID,   coal burned) must be as nearly  identical as possible for all
      sample acquisition points within a particular system.

15.   For sampling systems  that  do  not  inherently  sample on a  proportional
      basis,  :ne  following method shall be  used  to determine the  sampling
      strategy.

      a.   Determine the maximum rated  coal burning capacity in ib. per hour
          for each combustion unit within 'a system (ail units discharging to a
          common flue).

     b.   Select the unit with the lowest rated coal burring capacity as the
          base unit.

     c.   Calculate the ideal factor of proportionality for the system by the
          equation:
     Where Fo = ideal factor of proportionality (Ibs. sample/lb. coal burned}

-------
     *
II.   Coal Sampling/Analysts Systems

     A.    A general description of  the process(es). and pollution control equipment. All
           factors that may affect the operation or maintenance of the sampling/analysis
           system must be included.

     B.    The location of the sample acquisition point(s) in relation to:

           1.    The point at which the coal is burned
           2.    Any coal processing devices
           3.    Pollution control equipment
           4.    Emission point of pollutant gases to the atmosphere

           Provide a  flow diagram which  clearly shows the  location ^>f "The  sample
           acquisition point(s).  Include any  test data and an explanation as to the basis
           for the choice of the location.

     C.    A description of the equipment, methods, and procedures to be used to comply
           with each of the following system design specifications  or their equivalent,
           where  applicable (for explanation of  terms, see Table IV).  Equivalency must
           be demonstrated to the Department's satisfaction.

           1.    Points of sample acquistion must  be located as close as possible to the
                point at  which the coal is burned.

           2.    Points of sample acquisition  must be located downstream of any coal
                processing devices, including but not limited  to pulverizers,  unless  an
                alternate location will yield representative results.
           3.    A separate point of sample acquisition  must be located in each
                 coal  feed stream  to a  particular combustion unit  unless  it  can
                 demonstrated  that sampling  conducted  at  fewer  points  will  y
                 representative results.

           4.    Sample collection  must be by  means  which do not allow for operator
                 discretion with respect to portions of sample retained or rejected.

           5.    Sampling devices  must  comply  with  ASTM  D'2234-76,  Sections  6.4
                 through 6.10, unless alternate devices yield representative results.

           6.    A minimum  of two subincrement point samples must be collected from
                 each point of sample acquisition for each discrete hourly time period.

           7.    An hourly increment point sample must consist of all subincrement point
                 samples  collected at  a  particular sample acquisition  point  during a
                 discrete hourly time period.  Each hourly increment  point sample must
                 weigh at least two pounds* except for flutdized. pulverized  coal where
                 lower sample weights yield representative  results.

           8.    An hourly increment  unit sample  must consist of  all hourly increment
                 point samples  for a  particular combustion  unit  during a particular
                 discrete hourly time period.

-------
                          INITIAL APPLICATION (PHASE D

Upon promulgation of a > monitoring  requirement,  the  following information must be
submitted within six months  to the appropriate Regional Office.  This information must
indicate the probable capability of a system to meet all of the regulatory requirements.
Only information concerning  one specific proposed system should be submitted.  Multiple
proposals  will not  be evaluated.   The information must  be clearly identified in the
submittal.

-------

-------
APPENDIX

-------

-------
systems appear to be sufficient to result in data that can be directly compared
with existing combustion source SO2 emission standards. Coal sampling and analysis
systems designed to meet these requirements appear to be reliable and cost effective
alternatives to continuous emission monitoring systems.

-------
            During performance testing, Penelec demonstrated that the results-
 obtained by sampling from a single exhauster from a particular coal pulverizing
 mill was equivalent to results obtained by sampling all exhausters from that mill.
 Penelec also demonstrated that results  obtained by sampling quantities of coal
 weighing less than two Ibs. per hour were equivalent to results obtained by sampling
 at a rate of at least two Ibs. per hour. In accordance with Department criteria,
 these equivalent procedures were approved.

                         RELIABILITY COMPARISON

            Although the Department does not yet have sufficient data to conduct
 a rigorous evaluation of data availability of CSAS's versus CEMS's , such information
 is available on a limited scale. The table below indicates percent "valid days" (a
 valid day is one containing no more than six continuous hours of invalid data) for
 CSAS's and CEMS's reporting during the first quarter of  1983.

                                     CEMS's                          CSAS's
Percent Valid Davs                Number oi Units               Number of Units

       100.0       '                    10                           *  .
       98.9                              u
       97.8                              1
       94.ft                              4
       93.3                              <*
       92.2                              3
       90.0                              2
       88-9                              1
       84.4                              7                           1
       80.0                              1
       78.9                              1
       72.2                              I
       66.7                              3
       57.8                              I

            The average percent valid days for CSAS's during the first  quarter of
 1983 was 97.4% versus 89.9% for CEMS's.

                          ECONOMIC  COMPARISON

            According to Penelec, the cost of installing a PACSS (labor plus materials)
 is approximately $1,500 per sampler.  At one sampler per pulverizing mill, the
 largest PACSS installed to date has eight samplers for a total installation cost of
 approximately $12,000.

            Although CEMS equipment costs of $10,000 to $20,000 are  often cited
 by manufacturers, total installed costs of $50,000 to $100,000 are not  unusual.

            Annual operation and maintenance costs for either CSAS or CEMS may
 be estimated to be approximately equal to the installed equipment cost.

                               CONCLUSIONS

            The requirements established by the Department for coal sampling and
 analysis systems to be used as alternatives to continuous emission monitoring

-------
                                    TABLE IV                            	
 DEFINITION OF SAMPLE TERMINOLOGY FOR COAL SAMPLING/ANALYSIS SYSTEMS
Time Period

(less than one hour)


hourly
Term

subincrement


increment  _
daily
composite
Can Represent

point (1)


point


unit


system


unit


system
Definition

individual sample collected
at a single point

accumulation of (1) lor a
single point for one hour

accumulation of (I) for all
points in unit for one hour

accumulation of (1) for aJJ
units in system for one hour

accumulation of (1) for all
points in unit for 2
-------
                                     TABLZ V
                 COAL SAMPLING PERFORMANCE SPECIFICATIONS
1.


2.

3.



4'.



5.

6.
     Parameter
     Number of subincrement point samples per hour
     per point of sample acquisition

     Weight of hourly increment point sample

     Variation of actual factor of proportionality for
     daily composite unit samples (7 days, individual •
     unit)

     Variation of actual, factor of proportionality for
     daily composite unit samples (daily, all units
     within system)

     Precision of sample preparation

     Calibration error for percent  sulfur analysis.
     dry basis

7.   Precision of analysis for percent sulfur, dry basis
8.    Standardization of calorimeter water equivalent


9.    Precision of analysis for BTU/lb., dry basis

10.   Response time of system

11.   Operational period of system

12.   Relative accuracy of system ib. SCH/IO6 BTU
      results
                                                        Secification
>2 IbS.

Each daily value must be within
+20% of the average
Each unit value must be within
+20% of the average value


The ratio of variance must be £3.29

£10 percent of each NBS SRM
value (high, mid, and low ranges)

£0.05 percent sulfur if sample
contains <2.Q percent sulfur.  £0.1
percent sulfur if sample contains
>2.0 percent sulfur

Must comply with ASTM D2015-77(7e
Section 6

£50  BTU/lb.

£168 hours

M68 hours

£20  percent of mean value of
reference method  tests plus
95 percent confidence interval
                                          13

-------
                         PERFORMANCE TESTING (PHASE H)
     f
After approval of Phase I, the applicant should proceed with purchasing, installation, and
performance testing.   The Source Testing and  Monitoring  Section must be advised in
writing at least  10  days prior to Performance  Specification Testing  and  provided the
opportunity to  observe and participate in all testing.  The Section must  also be advised in
writing within  10 days after the completion of testing.  The Section reserves the right to
conduct testing during the Performance Specification Testing or at any time thereafter.
Phase II must be completed within 10 months  after  Phase I approval.   AH performance
specification testing must be conducted in accordance with the appropriate performance
specification test procedures in 40 CFR, Part 60, Appendix B, of the Code of Federal
Regulations and in this manual. Additional requirements are as follows:

-------

-------
in.   Coal Sampling/Analysis

      A.   Conditioning Period

           1.   Determine  the calorimeter water equivalent in  accordance with ASTM
                02105-77(78), Section 6.  Record all data and results for submission with
                performance test report.

           2.   Operate the  system for an  initial  168-hour  conditioning period in a
                normal operating manner.

      B.    Operational test period.  Operate the system for an additional 168-hour period
           in a normal operating manner during which time all performance testing must
           be completed (with the exception of Sections B.5. through  B.9. below, which
           must be completed within  168  hours  after the completion of the operational
           test period).

           I.   Test for number of subincrement point samples per hour.  Conduct this
                test in triplicate for each point of sample acquisition.

                a.    Collect and weigh a single subincrement point sample.

                b.    Collect  and weigh  the next  subsequent  hourly increment  point
                      sample.

           2.    Test for weight of hourly increment point samples. Conduct this test in
                triplicate for each point of sample acquisition.

                a.    Use the weights determined in B.l.b. above for this test.

           3.    Test for variation of actual factor of proportionality for daily composite
                jnc samples (7 days, individual unit).

                a.    For each unit monitored,  collect all data necessary  to  determine
                      the actual factor of proportionality (Fja)  for each  of  the seven
                      daily   time  periods  during the operational  test  period  (i.e., the
                     weight of each daily composite unit sample and the weight of coal
                     burned in the unit  during the same daily time period).

           fc.    Test for  variation of actual factor of proportionality for daily composite
                unit samples (daily, all units within system).

                a.   Use the data collected as in B.3.a. above for this test.

           5.    Test for precision of sample  preparation.  Conduct this test for each
                system monitored using any  five of the seven  daily composite system
                samples normally collected during the 168-hour operational  test period.

                a*   Divide  the  daily  composite   system  sample  Into two  equal
                     subsamples.
                                         17

-------
      b.    Prepare and analyze one  of the  subsamples according to normal
           procedures (in duplicate).

      c.    Prepare and analyze the remaining subsample for'dry ash content*
           using ASTM 02013-72(78),  Appendix A2, Section A2.2.

6.    Test for calibration error of percent sulfur analysis, dry basis.

      a.    For each analyzer  to be  used* conduct  five non-consecutive
           analyses on each of MBS SRM 163la low, middle, and high ranges
           for percent sulfur, dry basis at  any time  during  the 163-hour
           operational test period  (NOTE:  If NBS SRM  163 la is not available.
           conduct this test using NBS SRM 1635 and NBS SRM 1632a).

7.    Test for precision of analysis for percent sulfur, dry basis.

      a.    Conduct  the  normal duplicate analyses of  the daily  composite
           system samples for the  168-hour operational test period.

8.    Test for precision of analysis for BTU/lb., dry basis.

      a.    Conduct  the  normal duplicate analyses of  the daily  composite
           system samples for the  168-hour operational test period.

9.    Test for response time of monitoring system.

      a.    Record the date and time that  each calculation of Ibs.  SO2/106
           BTU is completed  for each daily  composite system sample during
        -  the 168-hour operational test period.

10.   Test for standardization of calorimeter water equivalent.

      a.    Use data and results as  obtained in A.I. above.

11.   Test for operational period of monitoring system.

      a.    Keep   records   indicating  compliance  with   all  performance
           specifications for the i68-nour operational test period.

12.   Test  for relative accuracy  of  monitoring system  Ibs.  SO2/106  BTU
      results.  This test must be conducted for each system monitored.

      a.    Conduct  a series  of  nine  source  tests  for SOj   emissions  in
           accordance  with  the   requirements of  Chapter  139  of  the
           Pennsylvania Department  of  Environmental Resources'  Rules and
           Regulations.  Each test must consist  of  the following determin-"
           ations:

           i.    Effluent  SO2  concentration   in   accordance   with   the
               procedures specified  in Chapter 139, Section 139.4(10).
                              18

-------
                ii.   Effluent volumetric flow rate according to Methods I, 2, 3
                     and » of 40 CFR, Part 60, Appendix A of the Code of Federal
                     Regulations.

                Hi.   Heat input to  the source(s)  being  monitored,  using  heat
                     balance or actual fuel feed data.
  i
                iv.   If it can be demonstrated that u and iii above are relatively
                     constant,  the Department may allow a single measurement of
                     each  to represent conditions  for up  to three measurements
                     of i.

           b.    Results of each source test must be expressed as Ibs. SO2/106 BTU
                for a known,  discrete hourly time period, using the data collected
                in i, ii and iii above.

           c.    Collect and  analyze,  in the  normal manner,  hourly  increment
                system  samples  for  the  system  monitored  for time  periods
                corresponding to each source test.

           d.    Results of  the monitoring  system  must  be  expressed  as  Jbs.
                SO2/I06 BTU for the  time  periods  corresponding  to  each source
                test.

C.   Calculations

     1.    Number of subincrement point samples per hour.

         .  a..   Using  the  data collected  as in B.l.a and  B.l.b. calculate  the
                average number of subincrement point samples as follows:
                       _
                     xsi

           Where N - average number of subincrement point sarrpies per hour

                Xi s average of the three weights determined as .n ?.l.b.

                Xsi * average of the three weights determined as ;n B.l.a.

     2.    Weight of hourly increment point samples

           a.    Using the data collected as in B.l.b., calculate the average weight
                of hourly increment point samples as follows:

                     3
                      EXi
               •«*   i»l
                Xi «       -


                where X~i a average weight of hourly increment point samples

                Xi * individual weights determined as in B.I. b.


                                    19

-------
3.    Variation  of actual factor  of proportionality  for daily  composite unit
      samples (seven day, individual unit).

      a.    Calculate  the actual  factor  of proportionality  for.  each.daily
           composite unit sample for each daily time period as  follows:

           F-  *£
           Fia s c'ia

           where F|a Actual factor of proportionality for daily composite unit
                      sample

           X! s weight of daily composite unit sample

         Cla  =  weight of coal burned in unit during the corresponding daily
                 time period

      b.    Calculate the average of the actual factors of proportionality for
           the 168-hour operational test period as follows:
     where F;a average of the actual factors of proportionality for the 163-
                hour operational test period
      * • •

           Fja s individual actual factors of proportional!iy  for  each daily
                composite unit sample

     c.     Calculate the variation as follows for  each of the 7 actual factors
           of proportionality:
                Vi-
                       ia
           where Vj ~  variation   of   an   individual   actual   factor   of
                      proportionality

                Fjm s  individual actual factor of proportionality

                Fja -  average of  the actual  factors  of proportionality  as
                      calculated in C.3.b.

     Variation of actual factors of proportionality for daily composite unit
     samples (daily, all units within system)
                              20

-------
            Calculate the average of the actual factors of proportionality for
            all units within a system, for each daily time period during the 168-
            hour operational test period as follows:
                       n
                       I Fs
      b.
where Fsa = average of the actual factors of proportionality for all
           units within a system  for a particular daily time period
           during the  168-hour operational test period.

      Fja = individual actual factor of proportionality

      n = number of units within system

Calculate the variation as follows for each of the units for each of
the seven daily time periods during the operational test period:
                        sa
           where V; a variation   of   an
                      proportionality
                                 individual   actual    factor   of
                 Fia = individual actual factor of proportionality

                 Fsa s average of the actual factors of proportionality for all
                      units within a system  for a particular daily time period
                      as calculated in C.4.a.

5.    Precision of sample preparation

      a.    Using the data collected in B.5.C., calculate the ratio of the largest
           variance of any set of four  subsarr.pies to trie  average variance cf
           the five sets of four suosampies according to ASTM  02013-72(73),
           Appendix A2.

      b.    The ratio calculated in C.S.a. must be <3.29  in order  to  comply
           with Performance Specification S.

6.    Calibration error of percent sulfur analysis, dry basis

      a.    Using the data collected in B.S.a., calculate each error as follows? "

                      |%Smi
                       %Sci
                              x 100% I- 100%
                               21

-------
7.
            where E; = error of an individual analysis
            % Srai » measured percent sulfur, dry basis
            % Sci = certified percent sulfur, dry basts
      Precision of analysis for percent sulfur, dry basis
      a.    Using the data collected as in B.7.a.,  calculate the precision of
            analysis for percent  sulfur,  dry basis for each pair of duplicate .
            analyses as follows:
            where P; s precision of analysis for an individual pair of duplicate
                 analyses
            % SI; s percent sulfur results for first analysis
            % S2j = percent sulfur results for second (duplicate) analysis
 8.    Precision of analysis for BTU/lb., dry basis
      a.    : Using the data collected in B.S.a.. calculate the precision of
            analysis for BTU/lb., dry basis for each pair of duplicate analyses
            as follows:
                 Pi* IHlj-H2il
            where P; s precision of analysis for an individual pair of duplicate
                 analyses
            HI-, - BTU/lb. results for first analysis
            H2-, s BTU/lb. results for second (duplicate) analysts
9.    Response time of monitoring system.
      a.     Using the data collected as in 8.9.a., calculate the  time between
            recording of Ibs. SOs/lO6 BTU results and the end  of the daily time
            period represented by the results.
10.   Standardization of calorimeter water equivalent
      a.     Using the data collected in  A.I.,  calculate the standard deviation
           of the test series in accordance with ASTM 02015*77(78), Appendix
           Al. This value must be <3.6 BTU/degree F in order to comply with
           Performance Specification 8
11.   Operational period of monitoring system.
                               22

-------
      a.    If  the monitoring system  fails  to  comply with  any performance
           specification during the 168-hour operational test period, the test
           period must be repeated.   During the repetition, compliance need
           be demonstrated only with the failed specifications).

12.   Relative accuracy of monitoring system Ibs. SOg/lQ6 BTU results.  Using
      the data collected as in 8.12.&.  through B.12.d., calculate  the relative
      accuracy plus 95 percent confidence interval as follows:
9
'£ r
5=1
                       EX$i
                                              x 100%
CIo.95 * O.Q<
 9               9
9U
-------

-------
                                                                                         I •"
•a'
                            FINAL APPROVAL (PHASE HQ

A report must be submitted  to the Bureau verifying the monitoring system's compliance
with all regulatory  requirements.  The report must be submitted within two months after
completion  of Phase II and  must include the data  as  specified in  Phase II and in the
following:

I.    Continuous Source Emission Monitoring

      A.    For  opacity   monitoring,  W  CFK,  Part  60,  Appendix  B, Performance
           Specification  1, Sections S and 9.

      B.    For sulfur dioxide or nitrogen oxides monitoring, 40 CFR, Part 60, Appendix B,
           Performance Specification 2, Sections 6 and 7.

      C.    For oxygen and carbon dioxide monitoring, 40  CFR,  Part  60,  Appendix B,
           Performance Specification 3, Sections 6 and 7.

II.    Coal Sampling/Analysis Systems

      A.    No additional information.

The method used to convert the monitoring data to the required reporting format must be
verified in the report using actual test data. The report must also include a description of
any changes,  additions,  or deletions made to the information submitted in the  initial
application (Phase I).

-------

-------
       II.    Coal Sampling/Analysis Systems

             A.   Record Keeping

                  1.    The company shall reduce all of the system results to daily averages in
                        LB SOW106 BTU in  accordance with the data validation and reduction
                        criteria in the Quality Assurance section of this manual.

                  2.    A chronological file shall be maintained by the company which includes:
a
                        a.    All laboratory samples identified by system and date represented.

                        b.    The results of each analysis for percent sulfur and BTU/lb.

                        c.    All valid averages as calculated  in I. above,  along with the date
                             and time the result was recorded.

                        d.    The cause, time periods,  and magnitude of all  calculated emissions
                             which exceed the applicable emission standard(s).

                        e.    Data and results for all performance tests and recalibrations.

                      •  f.    The data, necessary to show compliance with all data validation and
                             reduction criteria in the Quality Assurance section of this manual.

                        g.    The cause  and time periods for any invalid data averages.

                        h.    A record  of any  repairs,  adjustments,  or maintenance  to the
                             system.

                        i.     The process and pollution  control equipment operating data for all
                             parameters wnicr. affect the emission level of 5O->.

                  3.     All data must 5e maintained 5v t.K.e  company for a  ?er:oci of two years
                        and  ae  provided  :o  :re  Department  upon  request  u   any  time.
                        Laboratory  samples  .rust  3e   maintained  antil  tre  =rtd of  the  next
                       subsequent reporting period.

            8.    Reporting Requirements

                  I.    The  following information  shall be  reported  to  the  Department  on a
                       calendar quarter oasis:

                       a.    For each day, the  daily average emission rate  and  causes for any
                             daily averages which exceed the 30-day standard.

                       b.    For each day, the number of valid hours and causes  for any invalid
                            daily averages.

                       e.   The   results  from all  performance  tests  and   recalibrations
                            conducted  during the quarter.


                                                27

-------
The report shall be submitted in two copies to the central office by the
30th day following the close of the reporting period.

The report shall" be submitted in a format specified by the Department
and must be signed  by the person exercising  managerial responsibility
over the operation of  the sources) for which monitoring is required.
                         23

-------
II.
Coal Sampling/Analysis Systems

A.   Data Validation Criteria

     1.
           2.
                Daily composite unit samples.  A daily composite unit sample shall be
                considered invalid if any of the following conditions OCCUR

                a.   The sampling/analysis system  is not operated in accordance with
                     the performance specifications set forth in this manual.

                b.   The sampling/analysis system  is not operated in accordance with
                     the quality assurance criteria of this manual.

                c.   Any  combination  of  sampling/analysis  system  downtime  and
                     monitored unit downtime exceeds six consecutive hours.

                d.   The actual weignt of the daily composite unit sample is less than
                     0.75 Fja Cia Ibs.

                where Fja  =     average  of  the actual factors of proportionality  for
                                 unit i determined during the most  recent performance
                                 specification test (Ibs.  sampie/lb. fired).
                      C1m  =
                           weight of coal burned in unit i that day Obs. fired)
                e.
                The  actual factor of proportionality  for the daily composite unit
                sample is not witnin » 20% of the average of the actual factors of
                proportionality for all  valid  composite unit samples  within  tne
                system* unless all valid  daily composite unit samples are  analyzed
                individually and results weigr.ted according to tne actual amount of
                coal fired in eacn unit.
           Daily Composite System Samples

           a.    A daily composite system sample snai! 3e considered ir.valid if *.h
                total actual vaiid sample weignt is less than
O.T5
                             M
                             I
                                               |DS-
          Where M
                                number of units within the system
           3.    Laboratory sample analysis.  The results of  analysis of a  laboratory
                sample shall be considered invalid  if any of  the following  conditions
                occur:

                a«    The next subsequent calibration check indicates  non-compliance
                     with Performance Specification 6 or Performance Specification 3.
                                         31

-------

-------
                                          (U) Instrument Air,  5C psig
                                                                ( >) Electronic
                                                                    Timers (2)
,»
V
puiveritad coal
frmn miTla to boi-Tor' *
(1) Stainless
0 1 0 C 1 — — 	 » 	 -^
probe (1/2" 1.3.)
i»* 'X£rr£


•^1 1
•
^2> — '
Purge ^

••



C<
,4
MM
H> tf
•• ^

^- (3) Solenoid
- (2) Air Opera
/ r-r
3 1
A

•
*
                                                                (  ) Cyclone
                                                              Sa  pie Can
                        "Cycle of ...'.Operation*1
Let t - Start of Cycle
(1)  (t) to (t * 5 »«c.) — purge
(2)  (t * 5 »«c.) to (t *  13 tee.) — sample
(3)  (t •»• 13 sec.) to (t * 60 minutes) —-off
Repeat Cycle
Samples collected from noon to noon each day.
••Actual timing is site dependent (depends on  coal  flow,  pros BUT-
 in coal pipe, etc.)
                             Figure 1
                   Pennsylvania ElaetTlc C
                Automatic  "As Burned" Coal Sampler

-------

-------
           b.   The precision  of  analysis for percent sulfur, dry basis, is  not in
                compliance with Performance Specification 7.

           c.   The  precision  of analysis for  BTU/lb., dry  basis,  is  not  in
                compliance with Performance Specification 9.

B.    Data Reduction Procedure

      I.    All data averages must be calculated using valid data only.

      2.    A daily average  shall be considered valid if all of the data validation
           criteria above are complied with.

      3.    A running 30-day average shall be considered valid  if it contains at least
           23 valid daily averages.

C*    Maintenance Requirements

      1.    Sulfur analyzer maintenance

           a.   Conduct the test  for calibration error, as specified in item UI.B.6
                of the "Performance Testing"  section of  this  manual, immediately
                following any corrective maintenance to the sulfur analyzer.   •

      2.    Calorimeter Maintenance

           a.   Conduct standardization of the calorimeter  water equivalent, in
                accordance with  ASTM 02015-77(78),  Section 6,  immediately
                following any  corrective maintenance to  the calorimeter.

      3.    Sample Acquisition Maintenance

           a.   Conduct tests,  as specified  in  items III.B.l, III.B.2, III.B.3, and
                III.B.!* of  the  "Performance  Testing"  section of  this  manual,
                immediately following any corrective maintenance to the point of
                sample acquisition.

      ft.    Sample Preparation Maintenance

           a.   Conduct tests,  as specified in item I1I.B.3 of  the "Performance
                Testing"  section  of  this  manual,  immediately  following  any
                corrective maintenance on equipment used in sample preparation*

O.    Periodic Recalibration
                                                                             * .-
                                                                            **
      I.    Quarterly Recalibration

           a.   Performance  specification  tests,  as  specified  in  items  01.5.1
                through  ULB.11  of  the "Performance  Testing" section of  this
                manual, must be conducted quarterly.
                                    32

-------

-------