United States
Environmental Protection
Agency
Office of Research and
Development
Washington DC 20460
EPA/625/R-97/001
August 1997
&EPA Handbook
Continuous Emission
Monitoring Systems for
Non-criteria Pollutants
-------
EPA/625/R-97/OQ1
August 1997
Handbook
Continuous Emission Monitoring Systems
for Non-criteria Pollutants
Center for Environmental Research Information
National Risk Management Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, OH 45268
Printed on Recycled Paper
-------
Notice
The information in this document has been compiled wholly, or in part, by the U.S. Environmental Protection Agency (EPA)
under Contract No. 68-C3-0315, Work Assignment No. 3-39, issued to Eastern Research Group, Inc. (ERG). The work
was performed by Pacific Environmental Services, Inc. (PES) and its consultants, Source Technology Associates, and
Emission Monitoring, Inc. under subcontract to ERG. This document has been subjected to EPA's peer and administrative
review and has been approved for publication as an EPA document. Mention of trade names or commercial products does
not constitute endorsement or recommendation for use.
-------
Contents (continued)
4,3 Parameter Monitoring Used as an Indicator of Equipment Operation and Maintenance . 58
4.4 Parameter Monitoring Used as a Surrogate for Emissions 58
4.4,1 Parameter Monitoring as a Surrogate for Reporting Excess Emissions 58
4.4.2 Parameter Monitoring for Direct Compliance 59
4.5 Emission Modeling - Predictive Emissions Monitoring Systems 60
4.5.1 First Principle Models , 61
4.5.2 Empirical Modeling 61
4.5.3 Model Development 64
4.5.4 Model Quality Assurance 64
4.5.5 Model Limitations , 65
4.6 Issues in Parameter Monitoring 65
Chapter 5 System Control and Data Recording - Data Acquisition and Handling Systems 70
5.1 Option 1: Simple Data Recording Device , 70
5.2 Option 2: Plant Mainframe Computer System 71
5.3 Option 3: Commercial DAHS 71
5,3.1 Emission Data Recording , 71
5.3.2 Emission Data Display 72
5,3,3 Sampling System Control 73
5.3.4 Calibration Control and Recording 73
5.3.5 Alarms 74
5.3.6 Plant Computer Interface 74
5.3.7 Report Generation 75
5.3.8 Multitasking . , 76
5.3.9 Expansion 76
5.3.10 Hardware 76
5.4 Summary 78
Chapter 6 CEM System Procurement, installation, and Start-up 80
6.1 Defining the Project Scope 80
6.2 Reviewing the Regulations and the Process 83
6.3 Assessing the Site 83
6.4 Reviewing Monitoring Options 84
6.4.1 Extractive Systems 84
6.4.2 In-situ Systems 84
6.4.3 Parameter Monitoring Systems 89
6.4.4 New Market Products and Systems under Research and Development ...... 91
IV
-------
Contents
Notice , , y
Acknowledgments , viii
Chapter 1 Introduction , 1
1.1 Background and Objectives of the Handbook 1
1.2 Evolution of Continuous Emission Monitoring - Regulatory Programs 1
1,3 Current Status of Continuous Emission Monitoring - Monitoring Technology 2
1.4 Organization of the Handbook , 4
Chapter 2 Implementing Rules - Requirements for Installation of OEM Systems 6
2.1 Clean Air Act, RCRA, and Other Federal Monitoring Requirements 6
2.1.1 NSPS Requirements 6
2.1.2 Acid Rain Program 13
2.1.3 NESHAP and Title ill - MACT Standards 14
2.1.4 Regulations for Sources Burning Hazardous Waste 16
2.1.5 Part 503 Sewage Sludge Incinerators 18
2.2 State and Local Agency Programs 18
2.2.1 State and Regional Initiatives 19
2.2.2 Compliance Assurance Monitoring (CAM) Program 23
2.2.3 Open Market Trading 24
Chapter 3 Monitoring Technology - Instrumentation ,, 26
3.1 Monitoring Systems for Non-criteria Gases 26
3.1.1 Sampling Problems for Reactive and Condensable Gases 26
3.1.2 Solutions to Sampling Problems 29
3.1.3 Analytical Techniques 34
3,2 Monitoring Systems for Particulate Matter 43
3.2.1 Sampling Problems for Particulate Matter , 43
3.2.2 Continuous Particulate Monitoring 45
3.3 Monitoring for Metals 49
3.3.1 Sampling Problems for Metals 49
3.3.2 Mercury Monitoring Methods 49
3.3.3 Multi-Metal Methods 51
Chapter 4 Alternatives to Monitoring Instrumentation 56
4.1 Parameter Monitoring 56
4.2 Parameters and Sensors 57
in
-------
Contents (continued)
6.5 Evaluating Potential Bidders 91
6.6 Writing the Bid Specifications 93
6.7 Reviewing the Proposals and Making a Decision 94
6.7.1 Areas for Review 95
6.7.2 A Matrix Evaluation Technique 99
6.7.3 Normalizing the Issues 99
6.8 Installing the System 100
6.9 Approving/Certifying the System 101
6,10 Implementing the QA/QC Plan and Operating the System 101
Chapter 7 Certification and Approval Mechanisms for Non-criteria Pollutant Monitoring Systems;
Calibration and Demonstration of Performance 104
7.1 Introduction 104
7.2 Instrument Calibration , 105
7.2.1 Instrument Function 105
7.2.2 Calibration Materials 106
7.2.3 Direct and System Calibration Procedures 109
7.3 Demonstration of Performance; Certification and Quality Assurance 110
7.3.1 EPA Performance Specifications for Non-criteria Pollutant Monitoring ..... 111
7.3.2 Performance Specification Testing 111
7.3.3 Ongoing Performance Checks and Quality Assurance 116
7.3.4 CEM System Approval Mechanisms in Germany .,.,.... 118
7.3.5 International Standards Organization 120
7.4 Suggested Approval Mechanisms and Approaches for Non-criteria Pollutant
and Application Testing 121
7.4.1 EPA Method 301 121
7.4.2 Dynamic Analyte Spiking , 122
Appendix A Acronyms 126
Appendix B Glossary 128
Appendix C Bibliography and Additional Reading 133
Appendix D Hazardous Air Pollutants - 1990 CAAA Title III Listing 137
Appendix E Units of the Standard 144
Appendix f Guide for Evaluating CEM System Costs 149
Appendix G Effects of Sample Matrix on Analyte Spike , 158
-------
Figures
1-1 Continuous Monitoring Methods 3
3-1 Hot/Wet Sampling Systems 30
3-2 In-stack Dilution Probe , 30
3-3 Out-of-Stack Dilution System 31
3-4 Close-coupled System 32
3-5 Close-coupled Laser Monitoring System 33
3-6 Double-Pass Transmissometer 33
3-7 Gas Chromatogram 35
3-8 Flame lonization Detector .,...,. 36
3-9 Mass Spectrum of Meta-xylene 37
3-10 Linear Quadrupole Mass Analyzer 38
3-11 Total Ion Current Chromatogram 38
3-12 Transmission Spectrum , , . , 39
3-13 Differential Optical Absorption Techniques 40
3-14 Ion-mobility Spectrometer 42
3-15 Angular Dependence 47
3-16 Side-scattering Continuous Mass Emission Monitor 48
3-17 Typical Beta Gauge Paper Tape Monitor 48
3-18 Selection Process for Particulate Monitors 50
3-19 Inductively Coupled Plasma Technique 51
3-20 Laser Spark Spectrometry 52
4-1 Uses of Parameter Monitoring in Regulatory Programs 56
4-2 Operational Parameters Correlated to Emissions. . 59
4-3 Test Data 62
4-4 Linear Regression of Test Data 62
4-5 Emissions Calculations (PEM system predictions) 67
5-1 Odessa Paperless Stripchart 70
5-2 Typical DAHS System 72
5-3 Example Historical Trend Screen 73
5-4 Example Calibration Cycle Configuration Screen 74
VI
-------
5-5 Competing Demands for Processor Time , 76
5-6 Custom GEM System Datalogger , , . , , , 77
6-1 Example Flow Diagram for CEM System Selection and Implementation . 81
6-2 Selection Considerations for Extractive Systems 85
6-3 Selection Considerations for In-situ Systems 88
6-4 Considerations for Parameter Monitoring Systems 90
6-5 Alternative Approach to Monitoring System Decision-making , 92
6-6 Areas for Proposal Review 95
6-7 Cost Estimation Method for PEM System , 97
6-8 Analysis of PEM System Lifetime Cost for Two Example Cases . 97
6-9 Simplified Example Matrix Evaluation 99
7-1 Certification and Approval Mechanisms 105
7-2 Instrument Function of Two Analyzers 106
7-3 Generalized Schematic of a Permeation/Diffusion-Based Calibration Gas Generator 107
7-4 Use of Calibration Transfer Standards (CTS) 108
7-5 Direct and System Calibrations 109
7-6 Continuous Particulate Monitoring Calibration Line Specifications 121
7-7 Analyte Spiking 124
Tables
2-1 NSPS Criteria Pollutant Monitoring Requirements 8
2-2 NSPS Non-criteria Pollutant Monitoring Requirements 10
2-3 Example NSPS Parameter Monitoring Requirements 11
2-4 NSPS Appendix B Performance Specifications 13
2-5 HON Rule Example Continuous Monitoring Requirements 16
2-6 Pennsylvania Specifications for Hydrogen Chloride Continuous Emission Monitoring Systems 20
2-7 Pennsylvania General Parameter Source Monitoring Specifications 21
3-1 Automated Measurement Methods for Particulate Matter 48
7-1 Existing Performance Specifications for Selected Pollutant Monitors 112
7-2 Principal Performance Specifications for TUV Suitability Tests of Emission Monitoring
Instruments 119
VII
-------
Acknowledgments
This document was developed and prepared under EPA Contract No. 68-C3-0315, Work Assignment 3-39, by Pacific
Environmental Services, Inc. (PES) and its consultants, James A. Jahnke, Ph.D., Source Technology Associates,
and James W. Peeler, Emission Monitoring, Inc. (EMI), under contract to Eastern Research Group, Inc. Dr. Jahnke
and Mr. Peeler were the principal authors. Phillip J. Juneau and Laura L. Kinner, EMI, assisted with the preparation of the
handbook with Mr. Juneau writing Chapter 5 and Dr. Kinner preparing the initial version of Chapter 7 and assisting with
Chapter 3. The EPA project officer was Justice A. Manning, Center for Environmental Research Information, National Risk
Management Research Laboratory, Office of Research and Development. The current contact is Scott R. Hedges, also
of the Center for Environmental Research Information, National Risk Management Research Laboratory, Office of
Research and Development. Special thanks are extended to Dan Bivins, EPA's Emissions Measurement Center, for this
guidance and technical support during the preparation of the handbook. Further technical and editorial support was
provided by PES staff through John T. Chehaske, project manager. Mr. Bivins and Tom Logan, Office of Air Quality
Planning and Standards, and Robert G. Fuerst, National Exposure Research Laboratory, ORD, provided peer review for
the handbook. Lastly, Scott Evans, Clean Air Engineering; Paul Reinermann, 111, Pavilion Technologies, and Gus Eghneim,
Ph.D., P.E., Texas Natural Resource Conservation Commission, were kind enough to provide comments on Chapter 4
of the document in a very timely manner. To each person responsible for the final product sincere thanks are extended.
VIII
-------
Chapter 1
Introduction
1.1 Background and Objectives of the
Handbook
This Handbook provides a description of the methods
used to continuously monitor non-criteria pollutants
emitted from stationary sources. The Handbook
contains a review of current regulatory programs, the
state-of-the-art sampling system design, analytical
techniques, and the use of computer systems for data
acquisition and predictive monitoring. The Handbook
is intended for those in industry or in government who
are charged with implementing a continuous emission
monitoring (CEIVI) program for this wider range of air
pollutants.
In 1979, a Technology Transfer handbook was
prepared on the topic of CEM systems to provide the
detailed information necessary for developing continu-
ous monitoring programs at stationary sources (EPA,
1979). This information was updated in subsequent
publications (EPA, 1991 and Jahnke, 1993); however,
the original Handbook and these later materials
focused on the monitoring of the criteria pollutants
such as sulfur dioxide (S02) and the oxides of nitro-
gen (NOX). Due to the successes of earlier CEM
programs, requirements for CEM have been and are
being extended to cover a wider range of pollutant
categories such as volatile organic compounds (VOCs)
and hazardous air pollutants (HAPsj.
Although the installation of monitoring systems is
most frequently initiated through regulation, the use
of these systems for optimizing and improving process
operations has proven to be an important benefit for
many companies. When accurate emissions data are
available, plant operators are provided with the
baseline information necessary to control operations
or to change operational practices. The net benefit
for both industry and the public is to improve opera-
tional efficiencies with a consequent reduction of
emissions to the atmosphere.
1.2 Evolution of Continuous Emission
Monitoring - Regulatory Programs
Federal CEM requirements were originally established
in the U.S. for tracking the performance of air pollu-
tion control equipment under the mandates of the EPA
New Source Performance Standards (NSPS). Data
obtained under this program are reported when
emission standards are exceeded. These "excess
emission reports" are then used to determine if
control equipment performance warrants conducting
an inspection or a reference method test (40 CFR 60
Appendix A, USEPA, 1996a) to determine whether
the source is in compliance with its emission stan-
dards. This program began in the middle 1970's
and after a number of years of experience, enough
confidence was achieved in CEM systems to use the
resultant data directly for the enforcement of emission
standards (McCoy, 1986).
To withstand the rigors of litigation, CEM quality
assurance programs became necessary to establish
the continuing validity of the data. Subsequently,
quality assurance requirements for CEM systems (40
CFR 60 Appendix F, USEPA, 1996b) were promul-
gated in 1986 for sources where the systems are
used for monitoring compliance with emissions
standards. By the late 1980's, federal experience
with electric utilities constructed after 1978 (the
Subpart Da sources) and the stringent CEM-based
enforcement program of Pennsylvania, indicated that
CEM systems could achieve both high levels of
accuracy and availability. For S02 and NOx monitoring
systems, accuracies of 10%, relative to the EPA
Reference Test Methods of 40 CFR 60 Appendix A,
were common. System availabilities of 95% were
attained for gas measurements and availabilities of
98% for opacity monitors had become achievable in
practice.
The encouraging experience of these earlier (and
continuing! programs helped to establish CEM as the
foundation for the acid rain program of the 1990
Clean Air Act Amendments. In this emissions trading
program, emission allowances (one allowance grant-
ing the right to emit 1 ton of SO2 per year) were
originally allocated and reserved so that a net reduc-
tion of 10,000,000 tons of S02 would be realized by
the year 2010. Allowances can be viewed as finan-
cial instruments that can be bought, sold, traded, etc.
CEM systems enter this picture by providing the
means of determining who has how many allowances
— how many are used up, and how many are unused
-------
and available for trading. The CEM systems are
essentially, the "gold standard" on which the allow-
ance trading program is based (EPA 1991b),
Stringent performance requirements were established
(40 CFR 75) for CEM systems used in the acid rain
program. As a result of these requirements and the
importance of CEM systems for tracking allowances,
the systems received greater attention from upper
management than in previous programs. Resources
were allocated for both purchase and maintenance
and responsibility for the systems was given to higher
levels of management. After a year of operation of
the first phase of the program, relative accuracies of
5-6% were common, with corresponding availa-
bilities of 95% or better.
The success of CEM systems at the electric utilities
has encouraged their application for monitoring a
wider range of pollutants at other types of emissions
sources. Monitoring programs are well established at
pulp and paper mills, petroleum refineries, municipal
waste combustors, hazardous waste incinerators, and
cement kilns. At hazardous waste incinerators,
stringent operating permits go so far as to require that
waste feed be shut off when the monitors indicate
that carbon monoxide emission levels are too high or
oxygen percentages are too low. Particularly in the
case of hazardous waste incineration, continuous
monitoring data provide assurances to the public that
plant emissions are being controlled continually within
safe levels.
The Clean Air Act Amendments of 1990 address two
other issues. One is the public concern for the
emission of hazardous air pollutants, the so-called "air
toxics" (CAAA 1990 - Title III). Title III contains a list
of 189 pollutants, principally organic compounds and
metals (see Appendix D). Depending upon rules
specifying "Maximum Achievable Control Technology"
(the so-called "MACT" standards), certain operations
likely will be required to monitor the emission of these
materials on a continuous basis.
Title VII of the Clean Air Act Amendments refers to
the term "enhanced monitoring" (§702(b)(3H, where
sources are required to certify that they are in compli-
ance with their emissions limitations. "Enhanced
Monitoring" has evolved into the "Compliance Assur-
ance Monitoring" (CAM) rule, where CEM systems or
other techniques may be used by the source to track
its compliance status.
The air toxics and CAM rules affect a wide variety of
emissions sources. Many of these sources are small,
relative to the larger industrial sources that have been
required to Install CEM systems through the NSPS
Part 60 or acid rain Part 75 rules. For these smaller
sources, arguments often arise that CEM systems are
too costly, that the technology is too new, or that
extensive evaluation programs need to be conducted
before CEM systems can be used for regulatory
purposes. However, much is known about CEM
systems. With over 25 years of experience both in
the United States and Europe, continuous emission
monitoring for many pollutants is a mature technol-
ogy. With the one time upsurge of Acid Rain CEM
system sales in the early 1990's, subsequent market
pressures have forced CEM system vendors and
integrators to reduce costs to remain competitive. In
addition to the present knowledge base and market
pressures, new technologies are being introduced into
this field at a rapid pace. This combination of factors
has created an array of options for meeting monitor-
ing requirements for hazardous air pollutants.
1.3 Current Status of Continuous Emission
Monitoring - Monitoring Technology
Monitoring pollutant emissions from stationary
sources involves two principle functions: 1) extracting
or locating a representative sample and 2} analyzing
that sample. Monitoring methods in use today are
illustrated in Figure 1-1.
Extractive and in-situ methods are used to monitor
gas concentrations directly. In the case of extractive
systems, gas is withdrawn from the stack, condi-
tioned, and then analyzed. In in-situ systems, gas is
not extracted, but is monitored directly in the stack
by the analyzer. In the indirect parameter monitoring
methods, plant operational or control equipment
parameters are correlated to emissions determined by
manual or instrument reference methods. To the
extent that the parameter method's validity depends
on correlated data, representative sampling and
analysis are equally as important as in the direct
extractive and in-situ measurement methods.
The choice and design of a CEM system depends on
both the regulatory requirements and the types of
pollutants and/or parameters that are to be monitored.
For example, when emissions are reported in kg/hr or
tons/year, a pollutant monitor plus a volumetric flow
determination are usually necessary. If concentration
corrected values are required (such as ppm corrected
to 3% 02 or 12% 02), then a pollutant monitor plus a
diluent monitor (O2 or CO2) are required.
Different gases have different properties and some are
more amenable to analysis than others. Many well-
proven techniques are available for monitoring sulfur
-------
TYPES OF CONTINUOUS MONITORING METHODS
Parameter I
Surrogates
Predictive
—| Theory-based
1 Empirical
Figure 1-1, Continuous monitoring methods,
dioxide and oxygen. However, concentration
measurements of metal vapors and complex organic
compounds can be very difficult to perform in field
installations. With the advent of modern micropro-
cessor systems, sophisticated laboratory analytical
techniques have become available for field use at
relatively low cost. Fourier transform infrared
spectroscopy (FTIR) and gas chromatographic mass
spectroscopy (GCMS) are two such methods that
can be used for measurement of a wide range of
compounds. Other techniques, such as ion mobility
spectroscopy, the use of diode lasers in optical
differential absorption systems, and the advent of
low-cost catalytic or semi-conductor sensors provide
options for monitoring acid and organic gases that
were previously very difficult and expensive to
monitor on a continuous basis.
Technologies for the measurement of flue gas metal
concentrations are developing. The partition of
metals between the gas phase and particles provides
a challenge in obtaining a representative sample and
in making the actual measurement. Some metals,
such as mercury, are relatively easy to monitor in
the vapor state. When metals are bound in the
paniculate matter, the continuous measurement of
particulate mass concentration can monitor an upper
bound of metal emissions. Continuous particulate
mass measurement techniques are well established,
having evolved in Europe1 over the past 20 years
(Peeler, 1996).
In some cases, sensors used to monitor plant or
control equipment performance parameters also can
provide continuous data for emissions tracking.
Here, data obtained from the sensors are used in
place of actually determining emission concentra-
tions. A sensor-determined value is assumed to
stand in place of, or correlate to, emissions levels
expressed in ppm or mg/m3. This approach has
been used in NSPS requirements since 1975 and
may see greater application for process units regu-
lated under the proposed CAM rule. Rather than
just serving as emissions surrogates, operational
parameter data can also be incorporated into com-
puter models to predict emissions. These models
generally are developed by correlating parameter
data to actual source test data obtained over a
range of operating conditions. However, to develop
a robust model, one that remains valid under a wide
range of operating conditions, the costs associ-
ated with source testing may become comparable
with the actual cost of CEM instrumentation.
One of the central problems associated with moni-
toring hazardous air pollutants is the lack of estab-
lished or validated reference methods. Traditionally,
CEM systems have been certified for use at a facility
by performing a series of wet chemical or instrumen-
tal reference method tests. The tests are used to
determine the "relative accuracy," the accuracy of
the CEM system relative to the reference method
values. In the past, the effect of other gases on the
reference test method results was generally well
understood for specific industrial processes. Collab-
orative tests and years of field data have given
confidence that the reference methods indeed give
true emission reference values under most condi-
tions.
To similarly validate reference methods for the
hazardous, non-criteria pollutants has been difficult.
-------
The reactive nature of many of these gases, their
presence in gaseous and solid forms, and their low
concentration in flue gases, provide challenges for
wet chemical reference methods and instrumental
monitoring techniques alike. Because of these
problems, few reference methods for the hazardous,
non-criteria pollutants have been established.
Instead, several new approaches using reference
spectra, internal standards, and dynamic spiking
offer alternative approval mechanisms.
The engineer choosing between the various monitor-
ing options is of course, looking for the "best"
system. However, no generic "best system" exists.
GEM systems are application dependent; depending
upon regulatory requirements, pollutants to be
monitored, location restrictions, flue gas conditions,
ambient conditions, and manpower and management
considerations. Cost is a factor in choosing a
system, but low cost should never be the single
deciding factor. In the end, the best system will be
one that can analyze a sample representative of flue
gas conditions, one that meets all regulatory require-
ments for accuracy and precision, and one that has
low capital costs and low maintenance require-
ments.
When choosing between parameter surrogates,
predictive systems, or CEM systems, cost can be
viewed from a different perspective. When applied
to market trading programs, the value of an emis-
sions "credit" or "allowance" may be related to how
the credit is determined. Credits determined with
some uncertainty may be worth less than those
obtained with more certainty. In the end, the cost
associated with the monitoring method may be
counterbalanced by the consequent value of the
data.
The decision-making process for choosing a CEM
system can become very complex (White, 1995).
All of the decision factors for the "best" system are,
of course, relative. For example, a 10% relative
accuracy may be important for a trading program;
however, a 20% relative accuracy may be adequate
if normal emissions are at a 50 ppm level when the
emissions standard has been set at 200 ppm.
Monitoring costs may not be a significant factor
when facing $25,OQO/day penalties for noncompli-
ance, but a small manufacturing facility may not
have the capital or manpower necessary to maintain
a complex electro-optical monitoring system. This
Handbook has been prepared to provide information
that can assist the decision-maker in choosing
between the variety of available monitoring options.
1.4 Organization of the Handbook
The Handbook is organized in a manner that allows
the reader to proceed from a logical progression
from implementing rules, technical evaluation,
system purchase, and certification. The objective of
the Handbook is to provide the reader with the
conceptual tools and detailed information necessary
to make informed decisions with regard to the
monitoring options available.
* The implementing rules through which CEM
systems are required to be installed are ad-
dressed in Chapter 2. This chapter contains
a discussion of various regulations and moni-
toring programs to assist in evaluating non-
criteria pollutant monitoring programs in the
context of established regulatory concepts.
The chapter provides a review of existing
requirements, emphasizing those where non-
criteria pollutant monitoring is required.
» Current monitoring instrumentation is dis-
cussed in Chapter 3. The authors focus on
the dependence of the sampling system and
instrumentation on the pollutants to be mea-
sured. Reactive and condensable gases,
particulate matter, and metals present spe-
cific challenges. Due to the special sampling
requirements for non-criteria pollutants,
sampling systems are addressed separately
for gases and particulate matter. The analyti-
cal techniques used in analyzers designed for
monitoring HAPs and particulate matter are
presented.
* Non-traditional methods of monitoring source
emissions, focusing on the parameter (surro-
gate) monitoring and predictive monitoring
(modeling) methods, are addressed in Chapter
4. Advantages and disadvantages of the
methods and the concept that the combina-
tion of CEM instrumentation with modeling
methods makes for a powerful tool for both
monitoring and process control are advanced
in this chapter.
» An important subsystem associated with any
monitoring program, the data acquisition and
handling system, is discussed in Chapter 5.
The state-of-the art of CEM system data
acquisition, control, and reporting systems is
reviewed in this chapter. It provides a review
of the various options available and how they
fit into the total system package.
-------
Common sense approaches to choosing a
monitoring system, emphasizing the relation-
ship of the type of CEM system or monitoring
technique to the application, are discussed in
Chapter 6. This chapter describes what
information is needed prior to writing the
technical specifications, how to write a
technical specification, and how to evaluate
bids received.
Certification and approval mechanisms that
can be used for validating non-criteria pollut-
ant monitoring systems are addressed in
Chapter 7, Mechanisms are necessary to
provide assurances that the installed system
will provide data that provide a degree of
representativeness, accuracy, and precision
consistent with regulatory specifications.
Approval procedures specific to advanced
monitoring techniques are presented. Issues
involved with multi-component systems (such
as FTIR and GCMS), where many compounds
may be measured concurrently or sequen-
tially, are discussed with respect to proce-
dures that may be required for demonstrating
system and analytical performance both on
an initial and continuing basis.
References
Jahnke, J. A. 1993, Continuous Emission Monitor-
ing, Van Nostrand Reinhold, New York, NY.
McCoy, P. 1996. "The Use of CEM Data in Subpart
D Enforcement" in Continuous Emission Monitoring -
Advances and Issues, Air Poll. Control Assoc.,
Pittsburgh, PA.
Peeler, J.W., Jahnke, J.A., and Wisker, S.M. 1996.
Continuous Particulate Monitoring in Germany and
Europe Using Optical Techniques. Continuous
Compliance Monitoring Under the Clean Air Act
Amendments, Air & Waste Management Associa-
tion, Pittsburgh, PA. pp 208-220.
U.S. Environmental Protection Agency, 1979. Con-
tinuous Air Pollution Source Monitoring Systems,
EPA 625/6-79-005.
U.S. Environmental Protection Agency. 1991 a.
APTl Course 474 - Continuous Emissions Monitoring
Systems, EPA 450/2-91-006A.
U.S. Environmental Protection Agency, 1991b.
Acid Rain Program: Permits, Allowance System,
Continuous Emissions Monitoring, and Excess
Emissions; Proposed Rule, 56 FR 63002.
U.S. Environmental Protection Agency. 1996a.
Code of Federal Regulations - Standards of Perfor-
mance for New Stationary Source - Appendix A -
Test Methods, 40 CFR 60 Appendix A, Washington
D,C.
U.S. Environmental Protection Agency. 1996b.
Code of Federal Regulations - Standards of Perfor-
mance for New Stationary Sources - Appendix F -
Quality Assurance Procedures, 40 CFR 60 Appendix
F, Washington, D.C.
White, J, R. 1995. "Survey Your Options: Continu-
ous Emissions Monitoring," Environmental Engineer-
ing World, July-August 1995. pp 6 - 10.
-------
Chapter 2
Implementing Rules - Requirements
for Installation of CEM Systems
A discussion of various regulations is presented in this
chapter to help the user evaluate non-criteria pollutant
monitoring programs in the context of established
regulatory concepts. A review of current require-
ments, emphasizing those where non-criteria pollutant
monitoring is required, is also presented. Monitoring
regulations now being developed are discussed
briefly.
The information contained in this chapter is intended
for those that are new to the field of continuous
monitoring. Understanding the evolution of existing
monitoring requirements and the interrelationship of
different regulatory programs is especially important
in the development of new monitoring programs. The
following sections illustrate a range of monitoring
approaches, quality assurance requirements, and
regulatory uses of continuous monitoring data. This
background information will be helpfu! in establishing
and negotiating monitoring requirements for non-
criteria pollutant monitoring programs.
2.1 Clean Air Act, RCRA, and Other Federal
Monitoring Requirements
Federal pollutant monitoring requirements for air emis-
sions are contained in regulations developed under
several different statutes. Most notably, monitoring
requirements are included in several different regula-
tory programs developed over the last twenty-five
years under the Clean Air Act fCAA). These have
included new source performance standards, national
emission standards for hazardous pollutants, and the
acid rain program. Also under the Clean Air Act au-
thority, the EPA has established federal requirements
for states to adopt and implement pollutant monitor-
ing programs in State Implementation Plans designed
to achieve conformance with ambient air quality stan-
dards. These requirements are contained in Part 51.
Air pollutant emissions from sources that burn hazard-
ous waste are regulated under federal regulations
derived from the Resource Conservation and Recovery
Act (RCRA). Proposed revisions to the hazardous
waste combustor regulations combine the CAA air
program requirements and RCRA requirements into a
single set of regulations for these sources. Air pollut-
ant emissions from sewage sludge incinerators are
regulated under Section 503 of the Clean Water Act.
Emissions of radioactive materials are not described
here.
2.7.1 NSPS Requirements
EPA regulations contained in 40 CFR 60, "Standards
of Performance for New Stationary Sources," (com-
monly referred to as NSPS} apply to selected catego-
ries of stationary sources of air pollution for which
EPA has developed specific regulations to limit emis-
sions of criteria pollutants and certain designated
pollutants. Over the past 25 years, the EPA has de-
veloped NSPS regulations for more than 68 source
categories. In most cases, these regulations only
apply to sources above specific size thresholds and
for which construction began after the date of the
proposed regulation. NSPS regulations limit emissions
of criteria air pollutants including particulate, S02,
NOX, and CO, Emissions of volatile organic com-
pounds (VOCs) also are limited because of their role
in photochemical reactions resulting in the formation
of ozone in the ambient air. Other non-criteria pollut-
ants that are regulated directly or indirectly under
NSPS include total reduced sulfur compounds (TRS),
hydrogen sulfide (H2S), and hydrogen chloride (HCI).
NSPS regulations are designed to require the installa-
tion and proper operation of "best demonstrated
control* better known as "Best Available Control
Technology" (BACT) to minimize emissions of pollut-
ants. In general, a performance-based approach has
been adopted whereby an affected facility can choose
any method of pollutant reduction provided that the
source operator can demonstrate compliance with the
applicable emission limits by conducting performance
tests. In most cases, specific test methods or moni-
toring requirements for conducting the demonstration
of compliance are detailed within the applicable NSPS
regulations. Where possible, the same methods that
were used during the standard setting process to
-------
determine the level of pollutant emissions correspond-
ing to BACT are also used to demonstrate compliance.
In this way, the uncertainty (bias and imprecision) of
the measurement method and the averaging period of
the applicable standard are taken into account. Be-
cause of this approach, the exact test methods and
monitoring procedures used for compliance demon-
strations are of great importance. Test methods are
found in Appendix A and Performance Specifications
for continuous emission monitors are included in Ap-
pendix B of 40 CFR 60.
Many sources subject to NSPS particuiate emission
standards are also subject to an opacity standard. For
the most part, compliance with the opacity standard
is determined by a trained human observer in accor-
dance with EPA Method 9. Continuous monitoring of
opacity of emissions is required for certain sources.
NSPS regulations adopted since the 1977 Clean Air
Act Amendments have prescribed "percent removal"
requirements in addition to emission limitations for
certain sources. In some cases, continuous emissions
monitoring is required to demonstrate compliance with
percent removal requirements.
2.1.1.1 "Proper O&M Monitoring" versus
"Compliance Monitoring"
Initial NSPS regulations required the use of specific
manual test methods for demonstrating compliance
with particuiate, S02, and NOX emission limits. Me-
thod 9, Visible Emission Observations, was specified
for determining compliance with opacity standards.
CEM requirements for large boilers (heat input & 250
million BTU/hr), sulfuric and nitric acid plants, and
non-ferrous smelters were included in NSPS regula-
tions promulgated in December of 1971. The data
provided by these monitors could not be used to de-
termine compliance with emission limits but were
used instead to determine if a source was "properly
operating and maintaining process and control equip-
ment in a manner consistent with good air pollution
control practices" as is required by 60.11(d). This
regulatory application of continuous monitors has
become known as "proper O&M monitoring." Moni-
toring results have been used to enforce the require-
ment of 60.11 (d). These monitoring results have also
been used as a trigger for other activities such as
agency source inspections, visible emission observa-
tions, or requiring additional compliance demonstra-
tion tests. The EPA expanded this regulatory applica-
tion of a CEM system to other source categories in
subsequent regulations. Today, opacity CEM systems
are by far the most widely used monitoring technique
at NSPS sources. They remain as tools for enforce-
ment of the general 60.11(d) proper O&M monitoring
requirement rather than for direct determination of
compliance with applicable opacity standards.
Subsequent to the 1977 CAAA, NSPS Subpart Da
was proposed and later promulgated after litigation.
Subpart Da applies to large electric utility steam
generators constructed after Sept. 1978. It requires
the use of a CEM system for the continuous demon-
stration of compliance with S02 and NOK emission
standards and S02 percent removal requirements.
Compliance with these requirements is determined on
a thirty-day rolling average basis. Subpart Da require-
ments include explicit minimum data availability
requirements and also impose additional quality assur-
ance requirements for S02 and NO x CEM systems
contained in Appendix F, Procedure 1 of Part 60.
Since the promulgation of Subpart Da, continuous
compliance monitoring applications of S02 and NOX
CEM systems have been promulgated for other large
boilers, municipal waste combustors, and several
other source categories.
2.1.1.2 Summary of NSPS Criteria Pollutant
Monitoring Requirements
A summary of the NSPS criteria pollutant continuous
emission monitoring requirements is presented in
Table 2-1. Diluent (O2 or CO2) monitoring that is
required to convert pollutant concentrations to units
of the applicable standard is also indicated for
different source categories. The compliance averag-
ing period for CEM data is shown for those source
categories and pollutants where monitoring data are
used to demonstrate compliance with emission
standards. The time period for determining and
reporting excess emissions is shown for other moni-
toring applications. The reader is cautioned that
exemptions and exceptions affecting many sources
are detailed within the regulations for some source
categories. The actual Part 60 regulations should be
consulted for additional information, specific exemp-
tions, averaging periods, reporting requirements, and
other information.
2.1.1.3 Summary of NSPS Non-criteria Pollutant
Monitoring Requirements
A summary of the NSPS non-criteria pollutant moni-
toring requirements Is presented in Table 2-2. Exist-
ing non-criteria pollutant monitoring requirements
address emissions of volatile organic compounds
(VOCsJ, total reduced sulfur compounds (TRS), and
hydrogen sulfide (H2S) monitoring in petroleum refin-
ery fuel gas. All of the NSPS non-criteria pollutant
monitoring applications are used to ensure proper
operation and maintenance of source process and
control equipment. Emissions exceeding specific
-------
Table 2-1. NSPS Criteria Pollutant Monitoring Requirements
Subpart -
Effective Date
D- 8/1 7/71
Da -9/1 8/78
Db- 11/25/86
NOX 6/1 9/84
DC - 6/9/89
Ea - 12/20/89
F- 8/1 7/71
G- R/17/71
H- 8/1 7/71
J- 6/1 1/73
J- 671 1/73
Source Category and Type
Fossil Fuel-Fired Steam Generators
> 250 x 10* Btu/hour heat input
Electric Utility Steam Generating
Units > 73 MW (250 x 1 08 Btu/hour)
heat input
Industrial-Commercral-lnstitutional
Steam Generators >29 MW
(100 x 10* Btu/hour) heat input
Small Industrial-Commercial-Institu-
tional Steam Generators >2,9 MW <
29 MW (10 to 100 x 106 Btu/hour)
heat Input
Municipal Waste Combustors >250
tons/day
Portland Cement Plants
Sulfuric Acid Plants
Petroleum Refineries
Fluid Catalytic Cracking Unit
Regenerators
Fuel Gas Combustion Devices
Pollutant and
Diluent Monitors
S02 and NOX (02 or
C02 as diluent)
Opacity
SO2 and NO* emis-
sions, S02 percent
removal (02 or C02
as diluent)
Opacity
SO2 and NO^ emis-
sions, SO2 percent
removal (02 or CO2
as diluent)
Opacity
SO2 emissions,
SO2 percent remov-
al (certain sources)
(O2 or COZ as dilu-
ent)
SO2 with O2 as dilu-
ent, SO2 percent
removal (certain
sources)
Oz as diluent
CO with O2 as dilu-
ent
Opacity
Opacity
SO2 emissions
(O2orCO2 moni-
tors or other pro-
cedures to calcu-
late emissions)
fc^&$Jj£^'85sS£;:?3;£ffi£
S'ji'»i:S»*;:;m :» if fitfVftti&iSSsf.
CO
SO2 emissions SO2
removal, sulfur
oxides (certain
sources)
Opacity
S02with Oj as dilu-
ent, (H2S in fuel
gas as alternative)
Compliance Aver-
aging Period For
CEMS DATA
M ilxtxmXxXxXtZjjfKxtt $3.
30-day rolling av-
erage (boiler oper-
ating days)
:; n
30-day rolling av-
erage (boiler oper-
ating days)
;|JiS{i5giiSSjA{|||g5«j5|:j:;:j<
30-day rolling av-
erage (boiler oper-
ating days)
24 hours
4 hour and 24 hour
periods
?!:;:!;!:::!:!x!:::*S;:::::::-^^^
:lPllfiii?ipi??p'|il
.IKWK Vtymi »:»•*:: Uf,t t¥ti X
•mmmmmvmi %% mt
;S:::::':::::'5:<'Svr:-,::::::::-x<->,5-;-,v::x-Xv:':"^^^
lixllsllASsl;IIIPSllsii§i!
*iSISISs*sw||lsS:?*^^^
7-day rolling aver-
age
!>'X-v*&^
Excess
Emissions
Reporting
Period
3 hours
6 minutes
5:«S::SiSV*j;s::;™<*»KH
6 minutes
i:r:>. s ' :• ";;';-"-;vw;'; v K-X-X v, v ,;, v
^x^^:>^v^^::^!•:;^•."l•^•:"^^x-^'-^:-•-^•"
6 minutes
mmmmmm
.;.;,..; x< :•>; x-C'Xv'.-.vXvX'X'X'.'.':
<<:V<^^:f-^''-y^^-y.^f^^
|$|:|x¥x^x;f *ttx$£g$fSf
6 minutes
6 minutes
3 hours
i;illlii.'ilil?«:
1-hour
liilillfiSi
6 minutes
3-hour rolling
average
8
-------
Table 2-1. Continued.
Sub part -
Effective Date
J - 10/4/76
J - 10/4/76
- 10/15/74
R- 10/16/74
Z- 10/21/74
AA- 10/21/74
to 8/17/83
Aaa - 8/17/83
BB - 9/24/76
CC- 6/1 5/79
HH - 5/3/77
NN- 9/21/79
LLL- 1/20/84
UUU - 4S3/86
Source Category and Type
Claus Sulfur recovery Plant > 20
LTD* with oxidation control system
Claus Sulfur recovery Plant > 20
LTD with reduction control system
Primary Copper Smelter
Dryer
Roaster, smelting furnace, and cop-
per converter
Primary Lead Smelter
Blast furnace, dross reverberatory
furnace, or sintering machine dis-
charge
Sintering machine, electric smelting
t A **
furnace, and converter
Ferroalloy Product Facilities, sub-
merged electric arc furnaces
Steel Plants-Electric Arc Furnaces
and Argon Decarburization Vessels
(exceptions for certain controls)
Steel Plants - Electric Arc Furnaces
and Argon Decarburization vessels
(exceptions for certain controls)
Kraft Pulp Mills
Recovery Furnaces
Glass Manufacturing Plants
Lime Manufacturing Plants, rotary
lime kilns
Phosphate Rock Plants, dryers, cal-
ciners, and grinders
Onshore Natural Gas Processing,
sweetening units
Calciners and Dryers in Mineral In-
dustries (with dry control devices)
Pollutant and
Diluent Monitors
S02with 02 as dilu-
ent
TRS with Oz as
diluent or dilution
sampling system
with oxidation and
SO, with 0, as dilu
errt
Opacity
SO,
Opacity
S02
Opacity
Opacity
Opacity
Opacity
Opacity
Opacity
Opacity
Velocity (also SO2
if oxidation control
system or reduc-
tion control system
followed by inciner-
ator is used)
Opacity
Compliance Aver-
aging Period For
CEMS DATA
;:::::;::::-::'"::::'::'::'::::::v;::::^,::-::,:: :; :::::"::^.::^.::-:: V; -
mm^Mm^mmffifgimi:
;.,.-v,.;.-. .•.-.,.-.;.,.;v,.-v .- -. .-.-. .-.-.,.;.-.,.- -. .- -. .- -.,.-.-.;.
:x:>v£:;:;^
6-hour period
m»xl)MmMXv,v?;ww^^^^^
il M
yV:: ":''< -:•? ^y^-yVs^xoM-x^x^x^fe^:^
Sx?:?:Wx^v":x;::.^;>vf;>:;;;';<^\
Excess
Emissions
Reporting
Period
1 2 hours
12 hours
6 minutes
Hiss illllHli
liii
6 minutes
'!SS£!?*?-JSJ8£§^$x£i;
:';, .: -./_ •-.:••:. ..v.v,:.y,v.:,v.:.v.-,y<
: :• -:-x-x-X'X-;-x.:.:.x.:.: •:•:••-• .x-:.x-:
6 minutes
6 minutes
6 minutes
mymmmmsm
6 minutes
6 minutes
6 minutes
6 minutes
24 hours
6 minutes
'LTD = Long Tons per Day
-------
Table 2-2. NSPS Non-criteria Pollutant Monitoring Requirements
Subpart -
Effective
Date
J
J -6/11/73
J - 10/4/76
BB-
9/24/76
DDD-
9/30/87
FFF-
1/18/83
III-
10/21/83
LLL-
1/20/84
QQQ-
5/4/87
sss -
1/22/86
VVV-
4/30/87
Source Category and Type
Petroleum Refineries
Fuel Gas Combustion Devices
Glaus Sulfur recovery Plant > 20 LTD with
reduction control system
Kraft Pulp Mills - Emissions from recov-
ery boilers, lime kilns, digester system,
brown stock washer system, evapora-
tor, and condensate stripper systems
VOC Emissions from Polymer Industry
Carbon absorbers
Condensers
Flexible Vinyl and Urethane Coating
and Printing - Rotogravure printing lines
with recovery units
Synthetic Organic Chemical Manufac-
turing Industry - Units with air oxida-
tion reactors
Onshore Natural Gas Processing,
sweetening units
Petroleum Refinery Waste Water Sys-
tem - Units with carbon absorbers
Magnetic Tape Coating Facilities - Units
with carbon absorbers
Polymeric Coating of Supporting Sub-
strates - Units with carbon absorbers
Pollutant and
Diluent Monitors
H2S in fuel gas (as alternative
to SO2with 02 in emissions)
TRS with O, as diluent or dilu-
tion sampling system with oxi-
dation and SO2with O2 as
diluent
TRS with O2 as diluent
yiimMmimm &'$ aslllrs "M iimimm
i timm ||si;ii m W. w«il
VOC
Temperature or VOC
VOC
Specified parameters or
VOC CEMS
velocity, (TRS and/or S02 as
alternative for certain
sources)
VOC
VOC inlet and outlet streams
for certain sources
VOC inlet and outlet streams
for certain sources
Excess
Emissions
Reporting
Period
:SS:l:j|:S:pg;g;g|il:
3-hour roll-
ing average
12 hours
1 2 hours
85SS; '; si jwS*®:;:'!; j ; ; J
3 hours
3 hours
3 hours
3 hours
24 hours
3 hours
3 hours
3 hours
levels are reported as "excess emissions," The mon-
itoring results from these applications are not used
to determine compliance with emission standards.
Other designated pollutants such as HCI are regu-
lated in Part 60 because of potential welfare impacts
resulting from emissions. During the development of
the Subpart Ea for municipal waste combustors, the
EPA considered including an HCI continuous emis-
sion monitoring requirement. However, the final rule
relies on annual HCI emissions tests in conjunction
with the continuous demonstration of S02 removal
efficiency to control HCI emissions. In this case, the
continuous monitoring of S02 emission levels up-
stream and downstream of the acid gas control de-
vice serves as a surrogate for the direct monitoring
of HCI emissions.
2.1.1.4 NSPS Parameter Monitoring
Requirements
Many parameter monitoring requirements are in-
cluded in NSPS regulations with measurement fre-
quencies ranging from monthly to continuous. Se-
lected NSPS parameter monitoring requirements are
presented in Table 2-3. Similar requirements are
contained in the applicable regulations for many
other source categories. These and other NSPS
parameter monitoring requirements are necessary for
a variety of purposes.
10
-------
Table 2-3. Example NSPS Parameter Monitoring Requirements
Subpart
0
E,F,G,S
GG
DDD.III.NNN
N,O,Y,HH, LL
and many oth-
ers
Source Category and Type
Sewage Treatment Plants
Incinerators, Portland Cement
Plants, Nitric Acid Plants, Primary
Aluminum Reduction Plants
Stationary Gas Turbines
VOC Emissions from; Polymer Man-
ufacturing Industry, SOCMI Air Oxi-
dation Unit Processes, and SOCM!
Distillation Operations
Secondary Emissions from Basic
Oxygen Steelrnaking Facilities, Sew-
age Treatment Plants, Coal Prepara-
tion Plants, Lime Manufacturing
Plants, Metallic Mineral Processing
Plants
Parameter
feed rate
fue! flow rate to
incinerator
feed rate or pro-
duction rate
natural gas flow
rate
water to fuel ratio
absorber scrubber
liquid temperature
and specific gravity
boiler or process
heater combustion
temperature
flare (on/off)
incinerator com-
bustion tempera-
ture
wet scrubber pres-
sure drop
Measurement
Frequency
continuously
continuously
daily
continuously
continuously
continuously
continuously or
every 1 5 minutes
continuously
continuously
continuously
Accuracy
±5%
±5%
±2%
±5%
±5%
±1% or
±0,5°C and
±0.02 s.g
±1%or
±0.5°C
±1%or
±Q,5°C
±1 inch H20
Parameter monitoring may be required where pollut-
ant monitoring is impractical or infeasible. For ex-
ample, NSPS regulations provide an exemption from
continuous opacity monitoring requirements if con-
densed water (droplets) exist at the monitoring loca-
tion and interfere with the measurement. Monitor-
ing certain parameters is required for determining
process or production rates in conjunction with pol-
lutant monitoring data to determine emissions in
units of the standard. For example, sulfuric and
nitric acid plants monitor certain process parame-
ters to calculate the mass of SO2 or NO x emitted per
ton of acid produced. Steam generators that com-
bust varying fuel mixtures must monitor the heat
input rate (fuel usage) for each fuel to select appro-
priate F-factors and properly calculate emissions in
units of mass of pollutant per unit of heat input.
Other sources are required to monitor production
rate or operating hours to demonstrate that they are
exempt from a particular monitoring requirement.
A number of "demonstrated compliance parameter
level" approaches occur within NSPS regulations. In
these applications, a specific parameter level is pre-
scribed within the regulation or determined by empir-
ical tests that will ensure compliance with the emis-
sion limitation. Thus, the parameter value becomes
a surrogate for the emission limit. For example, the
outlet temperature for catalytic VOC incinerators
must be monitored and maintained above a minimum
to ensure that VOCs are combusted properly. As
another example, the water injection rates neces-
sary to achieve compliance with NOX emission limits
for an NSPS Subpart GG gas turbine are demon-
strated at each of four operating loads during the
initial compliance test. Subsequently, the operating
load and water injection rate are monitored continu-
ously to ensure compliance with HOX emission limita-
tions.
Other NSPS parameter monitoring requirements are
quite diverse. They include monitoring of sulfur
content of coal to calculate SO2 emissions for cer-
tain sources where S02 monitoring is not required
and for other sources where fuel sulfur pretreatment
credits are applied towards SO2 removal require-
ments. Certain source categories involved in coating
operations must monitor the VOC content of coat-
ings to demonstrate compliance with applicable
limits. Other sources must monitor compliance with
various "work practice standards" ranging from sim-
ply covering solvent containers or closing ventilation
11
-------
hood inspection doors, to wetting of unpaved roads
to reduce fugitive paniculate emissions at stone and
mineral processing facilities. These types of require-
ments are specific to particular source categories.
They are established during the rule making process
in consideration of many factors including technol-
ogy limitations and cost.
2.1.1.5 NSPS Monitoring Regulations and
Performance Specifications
Important aspects of NSPS CEM programs are pre-
scribed in several sections of 40 CFR 60. Require-
ments applicable to all sources are found in "Subpart
A. General Provisions." Of these, "§60.13 Monitor-
ing Requirements" includes requirements for source
operators to conduct daily zero and upscale calibra-
tion checks, to perform an initial test of each CEM
system in accordance with applicable Appendix B
Performance Specifications, and other essential
elements of a monitoring program. Other require-
ments applicable to continuous monitoring programs
are found in §60.7 Notification and Recordkeeping,
and §60.11 Compliance with Standards and Mainte-
nance Requirements.
Generally, each source category subpart with a con-
tinuous monitoring requirement includes (or refer-
ences) additional specific monitoring requirements
and information. This typically includes require-
ments for the selection of the monitor span value,
identification of Appendix A test methods that may
be used to conduct relative accuracy tests, methods
for converting emissions to units of the standard,
and other technical requirements. For proper O&M
monitoring applications, reporting requirements and
specifications used to identify periods of excess
emissions are included. For compliance monitoring
applications, emissions averaging periods and de-
tailed compliance reporting requirements are includ-
ed.
Appendix B of Part 60 includes CEM performance
specifications for a number of different compounds
and applications. These are listed in Table 2-4.
These specifications and procedures are used to
determine whether a particular CEM system is ac-
ceptable at the time of, or soon after, installation at
a particular source. The performance specifications
are the minimum procedures that are required to
determine if a CEM system is capable of providing
reliable measurements. They are not sufficient to
assure the quality of the data obtained on an ongo-
ing basis. The first three specifications were origi-
nally promulgated in October 1975 and have been
revised substantially several times.
Performance Specification 1 (PS 1) applies to opac-
ity monitors and includes a detailed list of design
specifications to prescribe how the optical transmis-
sion measurement is to be made. In addition, per-
formance specifications are included in PS 1 to de-
termine 1) the stability of the monitor response
relative to its simulated zero and upscale calibra-
tion checks, 2} the monitor's calibration error rela-
tive to a set of external optical density filters, and 3)
the instrument's capability to operate for a period of
two weeks without unscheduled maintenance or
repairs. PS 1 does not include a relative accuracy
specification because no independent method is
available to measure the in-stack opacity. (Substan-
tial technical and administrative revisions to PS 1
were proposed on November 25, 1994 (59 FR
60585). EPA's and industry's consideration of these
revisions is ongoing at the time of this writing.)
All other performance specifications reflect a differ-
ent approach than PS 1. No substantial design re-
quirements are included. In essence, any sampling
system configuration and/or any analytical approach
may be used provided that the measurement system
can be shown to meet two basic performance speci-
fications. First, a drift test is conducted over a one-
week period to evaluate the stability of the monitor
response relative to the calibration materials and the
procedure used for the daily zero and upscale cali-
bration checks. Second, a relative accuracy test is
performed that involves the comparison of the CEM
system measurement results with concurrent inde-
pendent pollutant measurements obtained through
the use of specified test methods found in Appendix
A of Part 60. The relative accuracy test includes a
minimum of nine runs although twelve are more
common since the tester is free to reject up to three
runs on an arbitrary basis. The computation of the
relative accuracy test result includes both mean
difference and confidence coefficient terms based
on the paired CEM system and reference test re-
sults. Thus, both the accuracy and precision of the
paired measurements are evaluated. Failure of a
relative accuracy test may be due to problems with
the CEM system, problems with the reference test
methods, problems with the representativeness of
the sampling location, or other factors. A failed test
requires careful investigation to determine the cause
and then it must be repeated. A successful relative
accuracy test is considered to be an adequate dem-
onstration of the monitor's capability to provide
reliable data.
12
-------
Table 2-4. NSPS Appendix B Performance Specifications
Performance
Specifications
PS 1 j
PS 2
PS 3
PS 4
PS4A
PS 5
PS 6
PS 7
PS 8
PS 9
CEM
Systems
Opacity
S02 and NOX
02 and C02
CO
CO (applicable for municipal
waste combustors}
TRS
Continuous Flow
H2S (H2S in fuel gas)
VOC
Gas Chromatography
Performance Specifications 2 (SO2. and NO,) and 3
(O2 and CO2) were promulgated originally in October
of 1975 and included relative accuracy, calibration
error, response time, 2- and 24-hour zero and cali-
bration drift specifications, and continuous opera-
tional requirements for two separate one-week peri-
ods. Numerous corrections and more detailed tech-
nical test procedures were proposed as revisions to
PS 2 and 3 during 1983 in response to severe indus-
try criticism of the existing performance specifica-
tions. (This criticism arose as a result of the pro-
posal, and subsequent promulgation, of Subpart Da
which included the first use of a CEM system for
compliance monitoring.) The 1983 proposed revi-
sions to PS 2 were met with more objections from
the utility industry. The EPA subsequently repro-
posed and then promulgated revisions to PS 2 which
eliminated ail of the performance specifications ex-
cept for relative accuracy and the seven-week drift
test. In addition, specific CEM measurement loca-
tion requirements were reduced with general guid-
ance. These changes reduced the prescriptiveness
of the regulations and placed the responsibility on
industry for determining the acceptability of monitor-
ing systems. This philosophy has been maintained
in revisions to PS 2 and 3 and has been used to
develop PS 4 through PS 8. A different approach
has been used for PS 9 which eliminates the relative
accuracy test and relies solely on the use of a multi-
ple calibration gases to assess the accuracy of the
monitoring data.
Appendix F, Procedure 1 of Part 60 prescribes qual-
ity assurance requirements for CEM systems that are
used to demonstrate compliance with emission limi-
tations or percent removal requirements. They in-
clude: 1} requirements to develop quality control
procedures for five specific activities, 2) "out-of-
control" limits on daily zero and calibration drift
check results for determining when data can not be
used to satisfy minimum data requirements, and 3)
quarterly accuracy assessment procedures and "out-
of-control" criteria for such audits. Procedure 1
requires that a relative accuracy audit be performed
each year as one of the quarterly accuracy audits.
Three-run relative accuracy audits or cylinder gas
audits may be performed for the other three quar-
ters.
2.1.2 Acid Rain Program
EPA has developed the Acid Rain Program in re-
sponse to the 1990 Clean Air Act Amendments.
The program seeks to reduce SO2 emissions from
the electric utility industry by 10 million tons per
year (relative to 1985 emissions) by the year 2010
utilizing a market-based trading approach. Under
this program, each electric utility generating unit
received the right to emit a certain quantity of S02
each year. This right is expressed as a number of
"allowances" to emit one ton of S02. An allowance
trading program has been established to provide for
the sale or exchange of S02 allowances between
electric utility units, companies, or other parties.
Market forces within the utility industry are expec-
ted to determine the most effective means of
13
-------
achieving the overall SO2 emission reduction. Addi-
tional regulations are being developed to limit emis-
sions of NOX from the utility industry as part of the
Acid Rain Program.
Extensive continuous emission monitoring programs
serve as the basis for the allowance trading pro-
gram. These monitoring regulations are contained in
40 CFR 75 and require continuous monitoring of S02
and stack gas flow rate to determine the mass
emission rate of S02 (i.e., Ib. S02 per hour, tons S02
per year). Monitoring of NOX and diluent (O2 or C02)
concentrations also is required to determine emis-
sions in units of !b. NOK per million Btu of heat input.
{These sources must also report mass emissions of
C02.) The allowance trading program is based on
the premise that all emissions must be accounted
for. Thus, great emphasis has been placed on the
accuracy of the monitoring data at all operating
conditions (rather than only at the emission standard
level) in the adoption of Part 75 performance specifi-
cations and quality assurance procedures. In addi-
tion, special procedures have been developed to
account for missing data. Within the data substitu-
tion procedures required for missing data, incentives
are provided for affected sources to achieve high
levels of CEM availability. Other key aspects of the
Acid Rain Program are the development of electronic
reporting mechanisms and comprehensive efforts to
implement the program on a consistent basis in all
states and jurisdictions.
Essentially, the largest 263 S02 emission sources
were identified within the 1990 CAAA as Phase 1
sources. All other affected utility sources (approxi-
mately 2,000 units) are designated as Phase 2 units.
Phase 1 units were required to have a CEM system
installed and certified before November 1994. All
Phase 2 units were required to have a CEM system
installed and certified by Dec, 31, 1995 with the
exception of certain gas-fired units and peaking
units. The vast majority of all of the affected units
were able to install and certify the required CEM
system by the applicable deadlines (EPA, 1995a).
2.1.2.1 Flow Rate Monitoring
Before the development of the Acid Rain program,
there was little experience with stack gas flow rate
monitors installed as part of emission monitoring
systems. The performance and reliability of these
devices was a very controversial subject during the
development of the Part 75 regulations. However,
experience has shown that flow rate monitors are
capable of meeting the applicable performance spec-
ifications in Appendix A of Part 75 including the
relative accuracy tests conducted at three different
operating loads (Bensink, 1995). Stack gas flow
rate monitors are considered reliable monitoring
devices even though differences between the stack
gas volumetric flow rate determined by heat rate
calculations and by flow rate monitors continue to
be the subject of investigation by both the EPA and
the utility industry.
2.1.2,2 Part 75 Performance Specifications and
QA Requirements
These requirements are similar, though somewhat
more restrictive, than the Part 60 requirements. The
Part 75 performance specifications require use of
calibration gases to check the performance of all gas
monitoring systems. The specifications include
response time tests and a three-point calibration
error test. A bias test is also included that is based
on comparison of the mean difference and confi-
dence coefficient terms determined during the rela-
tive accuracy test. The regulations require applica-
tion of a bias adjustment factor to emission values
if the CEM data are biased low by an amount greater
than the confidence coefficient relative to the refer-
ence method data.
Part 75 Appendix B quality assurance requirements
are also similar to Part 60 requirements. Accuracy
audits are required twice each year except for those
units which achieve an "incentive" specification in
the relative accuracy test. Units that achieve a
relative accuracy of less than 7.5% are allowed to
perform a single accuracy audit each year.
2.1,3 NESHAP and Title III - MACT Standards
2.1.3.1 Part 61 Existing NESHAP
National emission standards for hazardous air pollut-
ants (NESHAP) are contained in 40 CFR 61. Prior to
1992, EPA promulgated NESHAP for 22 source
categories. These regulations limit emissions of
arsenic, asbestos, benzene, beryllium, mercury, vinyl
chloride, radon, radionuclides, and HAP fugitive
emissions from equipment ieaks.
Several of the NESHAP regulations include CEM
requirements. Subpart A General Provisions, §61.14
includes general monitoring requirements for sources
required to monitor continuously. Opacity monitors
are required for sources affected by Subparts N, 0,
and P regulating inorganic arsenic emissions from
glass manufacturing plants, copper smelters, and
from arsenic trioxide and metal arsenic production
plants, respectively. These regulations refer to Per-
formance Specification 1 in Part 60 for evaluations
of opacity CEM systems. In addition, parameter
monitoring requirements similar to those contained
in Part 60 are included for certain sources. Subpart
14
-------
F requires installation of vinyl chloride CEM systems
at certain emission points in ethylene dichloride,
vinyl chloride, and polyvinyl chloride plants. Sub-
part F, §61.68 includes specific technical require-
ments for monitoring systems that use gas chroma-
tography and flame ionization detectors for analysis
of sequential samples. Daily span checks with a 10
ppm calibration gas are required. Procedures con-
tained in Appendix B, Method 106 are referenced for
certification of vinyl chloride cylinder standards and
for the preparation of calibration curves. A sum-
mary of NESHAP monitoring requirements is in-
cluded in the "Enhanced Monitoring Reference
Document" (EPA, 1993).
2.1.3.2 WIACT Standards
Title III, Section 112 of the 1990 CAAA identified a
list of 189 hazardous air pollutants (HAPs) and re-
quires EPA to establish new NESHAPs for all major
sources of HAPs (Appendix A) in accordance with a
prescribed regulatory schedule. Major sources are
defined as those that emit more than 10 tons per
year of one HAP or more than 25 tons per year of a
combination of HAPs. The schedule for promulga-
tion of these new standards for affected source
categories was originally published in the July 16,
1992 Federal Register (57 FR 31576). The schedule
organized sources into "bins." The two-year bin
included 40 source categories for which standards
were to be promulgated by November 1992. The
four-year bin included 25 percent of the listed
source categories and the seven-year bin included
50 percent of the listed source categories. All listed
categories are to have standards set no later than
November 2000.
These regulations are being developed for inclusion
in 40 CFR 63 and specify the maximum achievable
control technology (MACT) for each source cate-
gory. The CAAA require sources to obtain case-by-
case MACT determinations if EPA has not promul-
gated a standard within 18 months of the scheduled
date.
The MACT standards will include monitoring pro-
visions that will satisfy the requirements of the Act
to ensure that source owners are able to certify as
to the compliance status of affected emission units.
These monitoring requirements will be at least as
rigorous (i.e., direct emissions measurement and
monitoring of enforceable operational limits) as re-
quirements outlined under Part 64 or Part 70 peri-
odic monitoring requirements. EPA has developed
general provisions that are applicable to all MACT
standards. The Part 63 General Provisions were
promulgated March 16, 1994 and include perfor-
mance testing, monitoring, and recordkeeping and
reporting procedures for sources subject to MACT
standards. A current listing of the MACT standards
status is published periodically in the Federal Regis-
ter by the EPA in accordance with the requirements
of the CAA. (See April 17, 1996 notice on the EPA
Technology Transfer Network (TTN) bulletin board,
telephone no. 919-541-5742). Many of the MACT
standards require inspection or monitoring of pro-
cess or control device parameters on a quarterly,
monthly, weekly, or daily basis. These requirements
are not addressed in this handbook because of the
frequency of the measurements. Few of the MACT
standards include continuous monitoring require-
ments.
2.1,3.2.1 HON Rule - Example MACT Standard. As
an example, the Hazardous Organic NESHAP (known
as the "HON rule") covers manufacturing processes
in the synthetic organic chemical manufacturing
industry (SOCMI) and regulates emissions from
about 370 facilities including approximately 111 of
the 189 HAPS (EPA, 1994a). Subpart G of the HON
rule contains regulations for emission points at
SOCMI sources. Continuous monitoring require-
ments are summarized in Table 2-5.
2.1.3.2.2 Magnetic Tape - Example MACT Stan-
dard. Subpart EE of Part 63 applies to the produc-
tion of magnetic tapes. Monitoring requirements
similar to the HON rule apply for sources using com-
bustion devices for control of volatile emissions.
Those source using carbon adsorbers are required to
install a CEM system for volatile organic hazardous
emissions. Depending on the type of facility, either
a total hydrocarbon monitor or a gas ehromatograph
with an appropriate detector can be used. Perfor-
mance Specifications 8 and 9 in Appendix A of Part
60 are used to evaluate the performance of these
types of CEM systems, respectively. Very similar
requirements are found in NSPS Subpart SSS except
that the requirements are intended for the control of
VOCs.
2.1.3.2.3 Secondary Lead Smelters - Example
MACT Standard. A final NESHAP (Part 63, Subpart
X) for new and existing secondary lead smelters was
published on June 23, 1995 (60 FR 32587). These
facilities recover lead metal from scrap lead, primar-
ily from used lead-acid automotive batteries and
have been identified as significant emitters of lead
and arsenic compounds, and 1,3-butadiene.
15
-------
Table 2-5. HON Rule Example Continuous Monitoring Requirements
Emission Source
Process Vents
Storage Vessels
Control System
thermal incinerator
catalytic incinerator
boiler or process heater
flare
scrubber
bypass lines
fixed roof
external floating roof
closed vent and control
Monitoring
firebox temp.
inlet and outlet temp.
firebox temp.
presence of flame
pH of effluent streams
vent stream flow
visual inspection
visual inspection
presence of ffame or flare
Frequency
continuous
continuous
continuous
continuous
continuous
continuous
annual
when filled
continuous
The MACT standard regulates emissions of lead
compounds as surrogates for all metal HAPs and
total hydrocarbons (THC) as surrogates for all or-
ganic HAPs, respectively. Continuous monitoring
requirements are included for baghouse operation
and THC emissions.
The proposed regulation would have required the
installation of a continuous opacity monitor and the
development of a site-specific opacity standard to
ensure adequate collection of metal HAPs. How-
ever, in view of the many comments opposed to this
approach for technical and administrative reasons,
the EPA modified this regulation. The final rules in
§63.548, require a "standard operating procedure"
(SOP) for baghouse inspection and maintenance, and
a bag leak detection system with an alarm, and a
corrective action procedure for responding to alarms.
The bag leak detection system: 1} must be capable
of detecting paniculate matter concentrations at 1.0
mg/m3, 2) must provide an output of relative or ab-
solute particulate emissions, 3) must include an
alarm system that activates upon detection of an
increase in particulate emissions, 4) must be in-
stalled downstream of any wet acid gas scrubber or
on each compartment of a positive pressure bag-
house, and 5) must be installed and operated in a
manner consistent with any available guidance from
the EPA.
The secondary lead THC monitoring requirements
apply to emissions from process sources (i.e., blast
and reverberatory furnaces). The THC monitor is
used to continuously monitor compliance (3-hour
average) with the applicable emission limit ranging
from 20 to 360 ppm (as propane) depending on the
type of furnace used. The THC monitor must com-
ply with the all of the CEM requirements in the Sub-
part A, General Provisions.
2.1.4 Regulations for Sources Burning Hazardous
Waste
2.1.4,1 Existing Regulations for Sources Burning
Hazardous Waste
2.1.4.1.1 Hazardous Waste Incinerators, The
treatment of hazardous waste is regulated under the
Resource Conservation and Recovery Act (RCRA).
Hazardous waste is defined in Part 264. Air pollu-
tion emissions from hazardous waste incinerators are
regulated in Part 265. Facilities subject to these
regulations must demonstrate compliance with pre-
scribed destruction and removal efficiencies (DREs!
for metals and organic emissions. The DREs are
determined during the Part B permit trial bum by
comparing the quantity of materials fed into the
incinerator with measured emission rates. Trial
burns are extensive testing programs typically in-
volving the measurement of emissions of speciated
dioxin/furans, volatile organic compounds, non-di-
oxin/furan semi-volatile organic compounds, particu-
late, metals, HCI/C12 and other components. Be-
cause of the extreme public concern regarding these
facilities, source-specific trial burn plans are devel-
oped and negotiated with federal, state, and local
agency representatives. Trial burn test results are
subject to control agency and public review and are
the basis for many of the requirements included in
the incinerator operating permit. CEM requirements
for CO and THC as well as incinerator and control
device operation also are included in hazardous
waste incinerator operating permits.
2.1.4,1.2 BIF Regulations, On February 21, 1991
(56 FR 7134), the EPA published a final rule control-
ling hazardous waste burning by boilers and indus-
trial furnaces (known as the BIF Rule). Currently,
the BIF Rule regulates emissions of HCI/CI2, CO,
particulate matter, metals and organics in essentially
the same manner as RCRA hazardous waste inciner-
ators. BIF sources also are regulated under RCRA
Standards for treatment, storage, and disposal of
16
-------
hazardous waste. The BIF Rule organizes sources
using a three-tiered approach for each target pollut-
ant. Current provisions under the BIF Rule mandate
continuous emission monitoring for CO, 02, and
hydrocarbons based on this approach (EPA, 1992).
Tier I of the BIF Rule limits CO emissions to 100
ppmv (dry) based on an hourly rolling average cor-
rected to 7% 02. Monitoring hydrocarbon emissions
is not required if the source can meet this criteria.
If the source cannot meet this 100 ppmv Tier I CO
limit, then hydrocarbon monitoring is required in
addition to CO and O2 monitoring. The source is
also regulated under Tier II controls. Under Tier II,
hydrocarbon emissions are limited to 20 ppmv (dry,
corrected to 7% 02) and CO emissions are limited
based upon levels demonstrated during the compli-
ance test. Concentrations of CO and hydrocarbons
must be continuously monitored and corrected to
7% 02 on a dry basis. CEM systems for CO and
hydrocarbons must complete a minimum of one
cycle of sampling and analysis every 15 seconds,
and must record one data point each successive
minute. The 60 most recent 1-minute averages
must be used to calculate the hourly rolling average.
The current performance specifications for CEM
systems at BIF sources are included in 40 CFR, Part
266, Appendix IX, Section 2, Section 2,1 outlines
the performance specifications for CO and O2 ana-
lyzers. Included are procedures for conducting cali-
bration drift, relative accuracy, calibration error, and
response time tests to assess the conformance of
the CEM system with the specifications. The refer-
ence methods used for the relative accuracy deter-
minations are Methods 3 or 3A (for 02), and Meth-
ods 10, 10A, or 10B (for CO) in 40 CFR 60, Appen-
dix A. Performance specifications for hydrocarbon
analyzers are found in Section 2.2 of the BIF Rule.
They specify the use of a heated flame ionization
analyzer and sampling system maintained between
150-175°C, and include procedures for conducting
calibration error, calibration drift, and response time
tests. (Provisions are included also for the interim
use of sample conditioning systems that cool and
dry the stack gas sample prior to the analyzer.)
Relative accuracy test requirements are not included
in Section 2.2. Instead, procedures to challenge the
analyzer and system with calibration gases are used
to determine the conformance of the CEM system
with the specifications.
2.1.4.2 Revised Standards For Hazardous Waste
Combustors
Proposed revisions were published in the Federal
Register on April 19, 1996 to 40 CFR Parts 60, 63,
260, 261, 263, 266, 270, and 271 with respect to
the regulations for sources burning hazardous waste
(61 FR 17358). The rule was proposed under the
joint authority of the CAA and RCRA. The proposed
rulemaking action was taken for two main reasons;
1) to meet scheduled MACT standards requirements,
and 2) because of settlement requirements of a
lawsuit between the agency and several other par-
ties. The proposed rule revises standards for hazard-
ous waste combustors, boilers and industrial fur-
naces, and lightweight aggregate kilns burning haz-
ardous waste as supplemental fuels. MACT stan-
dards are proposed for dioxin/furans, mercury, semi-
volatile metals (Cd, Pb), low volatile metals (Sb, As,
Be, Cr), particulate matter, HCI, CI2, hydrocarbons,
and CO.
The proposed rules reflect a multifaceted approach
that establishes emission limits for dioxin/furans on
a "toxic equivalent basis", uses hydrocarbons and
CO as surrogates for volatile organic HAPs, and uses
particulate matter as a surrogate for 1) non-dioxin/
furin semi-volatile organics and 2) both low and
semi-volatile metals. No surrogates are proposed for
mercury and HCI/CI2. The proposed rule contains
monitoring requirements for CO and 02 (all data
must be corrected to 7% 02), hydrocarbons, particu-
late matter, and mercury. Continuous monitoring of
HCI/CI2 and other metals is optional.
New performance specifications to be included in
Part 60 have been proposed as follows:
PS 4B for CO and O2 for incinerators, boilers,
and industrial furnaces burning hazardous
waste (previously published in 56 FR 32688
July 17, 1991, with BIF regulations, but not
previously designated as PS 4B.)
PS 8A for hydrocarbons THC for incinerators,
boilers, and industrial furnaces burning hazard-
ous waste (previously published in 56 FR
32688 July 17, 1991, with BIF regulations, but
not previously designated as PS 8A.)
PS 10A mercury, semivolatile metals, and low
volatile metals for incinerators, boilers, and
industrial furnaces burning hazardous waste
(new).
PS 11A continuous monitoring of particulate
matter for incinerators, boilers, and industrial
furnaces burning hazardous waste {new but
similar to ISO 10155).
17
-------
PS 12A mercury emissions for incinerators, boilers,
and industrial furnaces burning hazardous waste
(new),
PS 13A HCi emissions for incinerators, boilers,
and industrial furnaces burning hazardous waste
(new).
PS 14A CI2 emissions for incinerators, boilers,
and industrial furnaces burning hazardous waste
{new}.
Extensive discussions of EPA's determinations re-
garding the feasibility, availability, and performance
of continuous monitors for organic compounds,
metals, and particulate matter are included in the
Technical Support Document 4A accompanying the
proposed regulations. Many references to the expe-
rience with monitoring particulate and other compo-
nents in Germany and Europe are included in the
technical support documents and the preamble of
the proposed regulations. EPA is conducting addi-
tional field evaluations for particulate and Hg moni-
tors as the rulemaking proceeds.
2.1,5 Part 503 Sewage Sludge Incinerators
EPA promulgated CEM requirements for sewage
sludge incinerators on February 19, 1993 in 40 CFR
503 under authority of the Clean Water Act, Sub-
part E of that regulation, §503.45, requires the in-
stallation of a total hydrocarbon (THC) monitor and
an oxygen monitor on each incinerator. The THC
monitor must use a sampling system maintained at
a temperature above 150"C and a flame ionization
detector. Monitoring results are corrected to 7% O2,
dry basis, and affected sources are required to com-
ply with an emission limit of 100 ppm. Detailed
guidance with respect to these regulations is found
in "THC Continuous Emission Monitoring Guidance
for Part 503 Sewage Sludge Incinerators" (EPA,
1994b). Performance specifications and quality
assurance procedures are modified from those in
Part 60. No relative accuracy test involving inde-
pendent THC measurements is required. Instead,
these regulations rely on the use of calibration gases
to assess the performance of the measurement sys-
tem.
In response to certain petitioners, EPA published a
modification to the sewage sludge incinerator moni-
toring requirements on February 25, 1994 {FR
9097). In this action, EPA agreed that a 100 ppm
CO standard imposed by the state of New Jersey
was more restrictive than the 100 ppm THC stan-
dard. EPA's amendment removed the THC require-
ment for those sources that install a CO monitor and
can demonstrate continuous compliance with a 100
ppm CO standard {monthly average).
2.2 State and Local Agency Programs
Most state and local air pollution control agencies
have broad authority to specify emission monitoring
and test methods. Monitoring requirements may be
adopted through applicable rulemaking procedures
for certain source categories or they may be in-
cluded in operating permits for individual sources on
a case-by-case basis. Source-specific monitoring
requirements are included in compliance orders or
consent decrees as a result of enforcement activi-
ties.
On October 6, 1975 (FR 40 48247), EPA estab-
lished requirements for states to adopt and imple-
ment continuous emission monitoring programs in
state implementation plans designed to achieve
conformance with ambient air quality standards.
These requirements are contained in Appendix P to
Part 51 - "Minimum Emission Monitoring Require-
ments." These minimum requirements identify af-
fected source categories; prescribe monitoring, re-
cording and reporting procedures for those sources;
and detail performance specifications and proce-
dures for converting monitoring data to units of the
state emission standard. Appendix P states,
"Such data must be reported to the
State as an indication of whether pro-
per maintenance and operating proce-
dures are being utilized by source op-
erators to maintain emission levels at
or below emission standards. Such
data may be used directly or indirectly
for compliance determination or any
other purpose deemed appropriate by
the State."
Appendix P addresses: opacity, SO2 and NO x moni-
tors (and diluent oxygen or carbon dioxide monitors)
for certain fossil fuel-fired steam generators, opacity
CEM systems for fluid bed catalytic cracking unit
regenerators, S02 monitors for sulfuric acid plants,
and NOX monitors for nitric acid plants.
The minimum requirements included in Appendix P
apply to opacity and criteria pollutant monitors at
fossil fuel-fired steam generators, sulfuric acid
plants, and nitric acid plants. Some states and local
agencies have required CEM systems for additional
sources and additional pollutants. In many cases,
these regulations are similar to the requirements
outlined in Appendix P.
18
-------
2.2,1 State and Regional Initiatives
State and local air pollution control agencies have
included continuous emission monitoring require-
ments in regulations and in permits for specific facili-
ties. With respect to non-criteria pollutant monitors,
many of these requirements are on a case-by-case
basis. The following are some examples. New Jer-
sey, Virginia, and Rhode Island have required Instal-
lation of ammonia monitors at certain cogeneration
or combined cycle turbine installations where NOX
control is required. A municipal waste combustor
facility in Connecticut is conducting an evaluation of
an installed HCI monitor to demonstrate that SO2
removal efficiency across a spray dryer is an ade-
quate surrogate for HCI emissions; the state agency
is determining the viability of HCI monitors based on
the same study {Anderson, 1996), Another HCI
CEM system is installed at a resource recovery facil-
ity in New Jersey in response to a permit require-
ment (Ballay, 1996), FTIR monitoring systems are
installed and reporting data at a hazardous waste
incinerator in New Jersey, Additionally, a prototype
gas chromatograph continuous monitoring system
measuring multiple organic compounds is installed at
B printing facility in North Carolina (Davis, 1996>.
Some of these installations, and^many others, are
installed on a trial basis; future requirements may
depend on the experience that is gained in these
efforts.
Three example state/regional initiatives are described
below which represent a range of CEM applications
and programs. These examples differ from the pre-
viously described federal regulatory programs and
illustrate alternative approaches, performance speci-
fications, problems, and solutions that may be useful
in other applications,
2.2,1.1 Pennsylvania Non-Criteria Pollutant
Monitoring
The Commonwealth of Pennsylvania has taken a
somewhat unique approach to the development and
implementation of CEM requirements. Pennsylva-
nia's requirements reflect differences in the technical
specifications, performance test procedures, and
reference test methods for criteria and non-criteria
pollutant monitoring. The requirements are con-
tained in, "Continuous Source Monitoring Manual"
Revision No. 6 (Commonwealth of Pennsylvania,
1996).
The Continuous Source Monitoring Manual contains
CEM requirements for 1) submittal and approval, 2)
recordkeeping and reporting, and 3} quality assur-
ance. The submittal and approval process includes
Phase I - Initial application, Phase II - Performance
testing, and Phase III - Final approval. Detailed re-
quirements for submission of the initial application
are provided; the initial application must be approved
prior to initial startup of new source and within six
months of promulgation of monitoring requirements
for existing sources. Performance Specifications for
opacity, S02, NOK, 02, and C02, CO, TRS and H2S,
and HCI monitors are provided. In addition, perfor-
mance specifications are provided for coal sampling,
stack gas flow monitors, temperature rate monitors,
"pollutants not listed elsewhere" and "parameters
not listed elsewhere." For gas pollutant and diluent
monitors, performance specifications are included
for 1) relative accuracy, 2) calibration error (three
points), 3) 24-hour zero and calibration drift, 4) 2-
hour zero and calibration drift, 5) response time, 6)
operational test period, and 7} data system accu-
racy. These are more comprehensive specifications
than are included in EPA performance specifications.
Unique to Pennsylvania is a requirement to verify the
performance of opacity monitoring systems by com-
parison to visual opacity readings. Also unique to
Pennsylvania is the requirement to evaluate data
system accuracy by comparing "manual calcula-
tions" based on monitoring values with data acquisi-
tion system output. Installation specifications estab-
lishing span value, range, data recorder resolution
(% of span and time), measurement cycle time,
frequency of zero and calibration checks, and other
requirements are also included. Pennsylvania re-
quirements for HCI monitors and "parameters not
listed elsewhere" are reproduced herein as Tables 2-
6 and 2-7, respectively.
As with performance specifications, CEM quality
assurance requirements in Pennsylvania are some-
what more extensive than federal requirements. For
example, three-point calibration error tests are re-
quired to be performed each calendar quarter. De-
tailed procedures are included for establishing the
values of the calibration standards used for daily
checks and for periodic audits. An annual relative
accuracy test is required. Also, annual review of the
quality assurance plan and quality assurance results
by the source operator is required for every facility.
Pennsylvania requirements detail criteria to deter-
mine when monitoring data are invalid and addi-
tional criteria to identify valid periods of data.
Non-criteria pollutant monitoring required in Pennsyl-
vania includes HCI monitors on municipal waste
combustors, TRS monitors at pulp and paper facili-
ties, H2S monitoring in petroleum refinery fuel gas,
and various parameter monitoring applications. As
with criteria pollutant monitoring, the data are used
to assess monetary penalties for excess emissions
19
-------
Table 2-6. Pennsylvania Specifications for Hydrogen Chloride Continuous Emission Monitoring Systems
Type
Install
Perform
Parameter
Span Value (nearest ppm equivalent)
Range (ppm)
Data recorder resolution !% of lowest std.)
Data recorder resolution (minutes)
Number of cycles per hour Smeas. and record!
Schedule for zero and calibration checks
Procedures for zero and calibration checks
Calibration gas ports
Relative accuracy in terms of standard
either {% of reference method)
or (% of standard!
or (abs ppm for ppm stds)
or (abs % for % reduction stds)
Calibration error (% of actual concentration)
or (abs ppm)
Zero drift - 2 hour (% of span)
Zero drift - 24 hour (% of span)
Calibration drift - 2 hour (% of span)
Calibration drift - 24 hour (% of span!
Response time (minutes to 95% response!
Operational test period (hours without corrective
maintenance)
Data acquisition system accuracy, 1-hour avgs
(%of lowest std)
Specification
2.0 times lowest std or
as specified in federal
regulations
0 to > = max. expected
and (> = 1.25 x highest std,)
1 .0 maximum **
5 maximum **
12 minimum **
daily minimum
ail system components checked
close to sample point
20 maximum *
1 0 maximum *
5 maximum
2.0 maximum
5 maximum *
1 maximum
4 maximum *
5 maximum *
4 maximum *
5 maximum *
5 maximum
168 minimum
1 maximum ***
* Expressed as the sum of the absolute value of the mean and the absolute value of the 95% confidence
coefficient.
** Must meet most stringent requirements of other analyzers in CEM system {except temperature).
*** If data recording is digital, expressed as the absolute value of the mean. If data recording is analog,
expressed as the absolute value of the mean and the absolute value of the 95% confidence coefficient.
20
-------
Table 2-7. Pennsylvania General Parameter Source Monitoring Specifications
Type
install
Perform
Parameter
Span Value {terms of measurement)
Range (terms of measurement)
Schedule for zero and calibration checks
Procedures for zero and calibration checks
Calibration point
Data recorder resolution (% of lowest std)
Data recorder resolution (minutes)
Number of cycles per hour {meas. and
record)
Calibration error (% of actual measurement
or simulated signal)
Zero drift - 24 hour {% of span)
Calibration drift - 24 hour (% of span)
Response time {minutes to 95% response)
Operational test period {hours without cor-
rective maintenance)
Data acquisition system accuracy, 1-hour
avgs (% of lowest std)
Specification
2,0 times lowest std or as speci-
fied in federal regulations
0 to > = max expected &
(> = 1 .25 x highest std)
daily minimum *
measurement simulation if possi-
ble, otherwise signal simulation *
close to measurement point *
1.0 maximum **
1 maximum **
60 minimum **
5 maximum ***
2.5 maximum ***
2.5 maximum ***
equal to recorder resolution
168 minimum
1 maximum ****
Specifications for parameters not listed elsewhere, based on basic measurements of length, mass, time,
temperature, current, luminous intensity or events, or derived from such basic measurements (for instance,
volume rate, mass rate, velocity, force, pressure, torque, rpm, voltage, resistance, spark rate, etc.). For use only
when specified or allowed by an applicable monitoring requirement, or when necessary to convert data to terms
of the applicable standard or operational criterion.
* This requirement may be waived if quarterly recalibration of the measurement device/readout device
combination is conducted by National Institute of Standards and Technology (NIST) or by a lab using NIST
procedures each calendar quarter.
** Must meet most stringent requirements of other analyzers in CEM systems (except temperature)
*** Expressed as the sum of the absolute value of the mean and absolute value of the 95% confidence
coefficient,
**** If data recording is digital, expressed as the absolute value of the mean, if data recording is analog,
expressed as the absolute value of the mean and the absolute value of the 95% confidence coefficient.
21
-------
and for poor GEM performance (i.e., monitor down-
time). Penalties for excess emissions depend on
both the magnitude and duration of periods when
the applicable emission standard is exceeded. Quar-
terly monitoring reports must be certified by the
source operator. Electronic data reporting formats
and telemetry protocols are specified also.
2.2.1.2 NESGAU1W CEM Guidelines for Municipal
Waste Combustors
The Northeast States for Coordinated Air Use Man-
agement, (NESCAUM} is an organization supported
by a group of eight states: Maine, New Hampshire,
Connecticut, Massachusetts, Rhode Island, Ver-
mont, New York, and New Jersey. NESCAUM
facilitates projects where the participating states
pool expertise and resources to address specific air
pollution problems. NESCAUM formed a workgroup
to develop CEM guidelines for municipal waste com-
bustors because of the number and impact of
these facilities being constructed in the northeast-
ern United States prior to the EPA's promulgation of
Part 60 Subpart Ea regulations for these sources. In
1990, NESCAUM published "CEM System Perfor-
mance Specifications and Quality Assurance Re-
quirements for Municipal Waste Combustors"
(Peeler, 1990), This guideline document is signifi-
cant because: 1) more extensive procedural and
technical requirements are recommended than are
included in the federal regulations for S02, NOX, CO,
and opacity CEM systems, and 2} specific perfor-
mance specifications and quality assurance require-
ments are recommended for HCI CEM systems.
NESCAUM CEM recommendations were specifically
developed to address technical monitoring problems
that are encountered at municipal waste Combustors
and the needs of the participating states. Technical
monitoring problems include low emission levels at
the control device outlet and widely fluctuating
emission levels (intermittent spikes) in CO and S02
concentrations at some sources; Major differences
in the NESCAUM performance specifications relative
to the existing Part 60 regulations were: 1} require-
ments that gas CEM systems use calibration gases
for drift checks and daily checks, 2) requirements
for quantitative determination of the calibration gas
values, 3) four-point linearity tests for all gas moni-
tors, 4) an additional minimum absolute accuracy
specification was included to reflect limitations of
the monitoring equipment, reference methods, and
relative accuracy test in certain cases, and 5) cycle
time/ response time specifications were added for
all monitors. The absolute accuracy specification
is a mean difference of 5 ppm (10 ppm for CO mon-
itors) during the relative accuracy test and reflects
the collective limitations of monitoring equipment
and reference methods. The limitations of the rela-
tive accuracy test at low pollutant concentrations
also are contained in this specification. This specifi-
cation is included in addition to the PS 2 relative
accuracy specifications of ^20 percent of the refer-
ence value or ^10 percent of the emission standard
for S02 and NOX monitors, and similar limits for CO
monitors. EPA subsequently included a 10 ppm
mean difference accuracy specification in PS 4A for
CO monitors at municipal waste combustors. How-
ever, similar revisions to PS 2 or PS 3 were not
made.
Along with additional and modified specifications,
the NESCAUM CEM guidelines include substantive
procedural changes to the performance specification
test procedures that address specific problems
which have occurred during field performance tests.
For example, controversy had arisen regarding the
validity of performance test results in cases where
monitor vendors performed numerous adjustments
during the test period and where "normal operating
procedures" had not been established for newly
installed CEM systems. The NESCAUM guideline
document includes recommendations to resolve
these issues by requiring source operators to estab-
lish prior to the test; 1) criteria for adjusting monitor
calibration, 2} criteria and schedules for routine
maintenance, and 3) the frequency and criteria for
additional checks of monitoring equipment.
Major differences in the NESCAUM QA require-
ments relative to the Part 60 regulations are: 1)
source operators must develop and submit a prelimi-
nary monitoring plan, 2} QA plans are required for all
sources and detailed guidance on QA plan content
is provided, 3} an annual review of QA plans by the
source operator is required, 4) a four-point linearity
test is required to be performed each calendar quar-
ter, 5) a relative accuracy test is required once per
year and must be performed immediately before or
after the quarterly linearity test, and 6) minimum
data availability specification <90 percent of source
operating time) is included.
2.2.1.3 SCAQMD RECLAIM Program
Since the passage of the 1990 Clean Air Act
Amendments, the South Coast Air Quality Manage-
ment District has been involved in the development
of Regional Clean Air Incentives Market, or "RE-
CLAIM" program. This innovative market-based
program was developed as an alternative to the
traditional "command and control" approach in an
22
-------
effort to achieve air pollution emissions reductions
in a more cost effective and efficient manner. The
evolution of RECLAIM and its objectives are dis-
cussed by Lents (Lents, 1996), RECLAIM is an
example of what has become known as a "cap and
trade" program.
A number of fundamental issues addressed in RE-
CLAIM are applicable to other cap and trade pro-
grams. These include: 1) determining the exact
population of sources included in the program (pol-
lutants, source categories, size thresholds, exemp-
tions, etc.), 2) determining the "cap" or total emis-
sions allowed for pollutants and determining how
the cap is to decrease over time to achieve neces-
sary emission reductions, 3) determining how per-
mitted emissions are to be allocated amongst the
population of sources initially, and over time, and
establishing baselines for historic emissions, 4) de-
termining how emissions are to be measured by the
affected sources and monitored by the agency to
provide an accurate, reliable, and systematic basis
for trades, 5) developing trading mechanisms that
would encourage rather than inhibit the process, 6)
reconciling the trading program with other applicable
regulations for the populations of sources, and 7)
changing the procedures used to issue permits and
enforce regulations and permit conditions.
RECLAIM originally was intended to apply to emis-
sions of NOX, S02, and VOCs. During the develop-
ment of the program, SCACLMD decided to postpone
the VOC program because: 1) technical difficulties
are encountered in attempting to quantify these
emissions (few historical quantitative measurements,
many small sources and many fugitive emission
points), 2) some of the VOC compounds also are
classified as hazardous pollutants and thus, different
regulatory considerations apply, 3} different VOCs
participate to different extents in reactions to form
ozone, and 4) the workload associated with the
program was too great, even after eliminating con-
sideration of the VOCs. SCAQMD is now attempt-
ing to implement RECLAIM for NOX emissions from
370 facilities and SO2 emissions from 40 facilities.
All of these facilities are required to install CEM
systems to quantify their SO2 and NQ< emissions.
Measurement data are averaged and 15-minute val-
ues are calculated by remote terminal units at each
facility. Data are transmitted electronically to a
central AQMD computer. The computer is pro-
grammed to deploy an inspector when a problem is
indicated at a particular facility.
RECLAIM represents a comprehensive program that
deals with many complex issues even for monitoring
of criteria pollutants. Even greater technical chal-
lenges must be overcome to apply such an approach
to non-criteria pollutants.
2.2.2 Compliance Assurance Monitoring
(CAM) Program
Before the 1990 Clean Air Act Amendments, EPA
and some state and local agencies had concerns that
some air pollution sources were not in compliance
with applicable regulations resulting in adverse air
quality impacts. The 1990 CAAA requires EPA to
develop regulations for permitted sources to en-
hance air pollution monitoring and certify compliance
with air pollution regulations. Permit regulations in
Part 70 require certain sources to perform periodic
monitoring and to submit annual certifications of
compliance. In October of 1993, the EPA proposed
the Enhanced Monitoring Program (58 FR 54648-
54699). This proposed program was to require all
major sources subject to federally enforceable re-
quirements to develop procedures and methods that
continuously demonstrate compliance with emission
standards. Data from enhanced monitoring were to
be viewed as "presumptively credible evidence" for
use in the enforcement of regulations. Thus, various
performance specifications and quality assurance
requirements were also proposed. The proposed
enhanced monitoring program fundamentally
changed the compliance methods for many sources
and was expected to impose great financial burden
on both regulated sources and air pollution control
agencies.
The enhanced monitoring rule was withdrawn in
April of 1995 and the EPA subsequently drafted the
compliance assurance monitoring program (CAM).
The draft CAM rule was released in September of
1995 and a second draft was released on August 2,
1996. (See the EPA TTN BBS, phone 919-541-
5742.) According to the accompanying announce-
ment, the CAM program attempts to build on
existing regulatory monitoring approaches by
focusing on "providing a reasonable assurance of
compliance with emission limits by monitoring that
ensures control measures are operated and
maintained in a manner consistent with good air
pollution practices." The CAM rule is intended to
satisfy the periodic monitoring requirements in Part
70 and the enhanced monitoring requirements in the
1990 CAAA. According to EPA, the proposed CAM
rule will cover about 60 percent of the emission
units with control equipment and 20 percent of alt
other emission units.
23
-------
Where continuous compliance monitoring is re-
quired, the draft CAM rule exempts the source from
additional monitoring. For affected units with con-
trol equipment, the source must develop and comply
with a CAM plan. The CAM plan is to include oper-
ating indicator ranges for control equipment that
represent good air pollution control practices that
minimize emissions. Excursions beyond these indi-
cator ranges trigger prompt corrective action. An
excessive duration of excursions requires more in-
tensive evaluation, corrective action, and requires
notification to the permitting authorities of potential
compliance problems.
The CAM rule does not require sources to install
CEM systems, so few sources are expected to do so
as a result of this rule. Depending upon the specific
requirements included in the final rule, and the out-
come of other rule making efforts (such as the
"credible evidence" rule), some sources may find
opacity or other pollutant monitors advantageous.
EPA has indicated that explicit requirements to sat-
isfy enhanced and periodic monitoring of hazardous
air pollutants will be included in future Part 63
NESHAP standards.
2.2.3 Open Market Trading
The 1990 CAAA encourages the use of market-
based approaches, including emission trading to
assist in achieving ambient air quality standards.
Market-based trading programs are intended to pro-
vide incentives for sources to reduce emissions be-
yond applicable requirements and encourage early
emission reductions and technological innovations to
reduce and measure emissions.
Emissions trading systems may be categorized as
being either "open" or "closed." Examples of "clos-
ed market" programs are the EPA Acid Rain Allow-
ance Trading Program and the SCAQIV1D RECLAIM
program. In closed markets, emission trading is
restricted to a defined population of sources, total
emissions are limited, or "capped" (which may de-
crease with time to achieve overall reductions!, and
portions of the total allowed emissions are allocated
among the affected sources.
In contrast, open market trading programs involve
voluntary participation, may include diverse types of
sources, and are designed to be compatible with
existing regulations. These trading programs typic-
ally involve the exchange of discreet, quantifiable
emission reduction credits between sources with
some portion of the reduction "retired" to provide
for improved air quality. Open market trading pro-
grams also may involve banking of emission reduc-
tions for use at a future date. The fact that open
market trading allows for the exchange of emissions
both over time and between sources distinguishes
this approach from emissions averaging between
sources. Open market trading programs may be
very flexible and avoid many of the problems associ-
ated with establishing baseline emissions and allo-
cating emission allowances among a specific popula-
tion of sources.
On July 26, 1995, the EPA administration signed
"Open Market Trading Rule for Ozone Smog Precur-
sors: Proposed Policy Statement and Model Rule
(USEPA, 1995b}." The preamble discusses many
aspects of open market trading approaches as they
apply to NOX and VOC monitoring. An important
aspect of these programs are the measurement pro-
tocols used to quantify the discreet emission reduc-
tions (DER) that are bought and sold. The DER must
be "real, surplus, and verifiably quantified" according
to the preamble of the model rule. Measurement
protocols may include a wide range of inputs includ-
ing emission factors, engineering calculations, peri-
odic source testing, predictive emissions models,
and CEM systems. The model rule allows for states
to adopt these programs and facilitates their rapid
approval by EPA.
Michigan, Texas, New Jersey, New York, and Vir-
ginia have developed, or are developing, open mar-
ket trading programs. NESCAUM/MARAMA, North-
east States for Coordinated Air Use Management/
Mid-Atlantic Regional Air Management Association)
have undertaken a project to encourage interstate
open-market trading of NOX and VOC emissions.
NESCAUM/IVIARAMA have developed a series of
measurement protocols that involve determining
baselines, applying emission factors, and using CEM
systems.
The Michigan Department of Environmental Quality
(DEQ) has specified some characteristic elements of
DERs (called Emission Reduction Credits (ERCs) in
Michigan}. "ERCs must be:
1) surplus, in that reductions are not required
by any applicable requirement;
2} real, in that ail emission reductions have
actually occurred;
3) quantifiable, in that all reductions can be
measured and are replicable;
4) enforceable, in that they can be enforced
by both DEQ and EPA; and
24
-------
5) permanent, in that the reductions were
continuous during the time the ERCs were
generated."
The viability of open market trading programs for
VOCs that rely on continuous monitoring of emis-
sions present several challenges. Where a monitor-
ing system providing a "total response" (as opposed
to speciating particular organic compounds) is used,
assumptions regarding the composition of the VOC
emissions may be necessary to account for varying
instrument response factors of the instrument for
different compounds. In addition, assumptions are
also necessary to estimate the molecular weight
used in calculating emissions on a mass basis. Addi-
tional considerations apply because many VOCs are
also hazardous air pollutants and because emission
reduction credits for hazardous compounds can not
be applied to achieve compliance with NESHAP
(MACT standards). Finally, an equitable basis for
the exchange of VOC emission reductions may need
to consider that different compounds participate to
different extents in reactions leading to the forma-
tion of ozone. The decision-maker considering par-
ticipation in an emission trading program for VOCs
or other non-criteria pollutant must seek resolution
of these issues with the applicable control agency.
References
Anderson, S. 1996, Connecticut Department of
Environmental Protection personal communication.
Ballay, F. 1996. New Jersey Department of En-
vironmental Protection & Energy personal communi-
cation.
Bensink, J., Beachler, D., Joseph, J. 1995. "Certifi-
cation of Flow Monitors for Utility Boi!ers, Acid Rain
and Electric Utilities: Permits, Allowances, Monitor-
ing, and Meteorology," Proceedings International
Specialty Conference, Air & Waste Management
Association, Tempe, AZ.
Commonwealth of Pennsylvania. January 1996.
"Continuous Source Monitoring Manual," Revision
No. 6, published by the Department of Environmen-
tal Protection, Bureau of Air Quality, Division of
Source Testing and Monitoring, Continuous Emission
Monitoring Section, P.O. Box 8468, Harrisburg, PA
17105-8468. The document can be obtained elec-
tronically over the Internet at;
http://www.dep.state.pa.us/dep/deputate/
airwaste/aq/cemspage/cemshome.htm
Davis, K. 1996. North Carolina Dept. of Environ-
ment, Health, and Natural Resources personal com-
munication,
Lents, J.M., Leyden, P. 1996. "RECLAIM: Los
Angeles' New Market-Based Smog Cleanup Pro-
gram," Journal of the Air & Waste Management
Association, vol. 46, No. 3, March 1996, pp 195-
206.
Michigan Dept. Of Environmental Quality. 1996.
"Michigan Air Quality Emission Trading Fact Sheet."
The document can be obtained electronically over
the Internet at:
http://www.deq.state.mi.us/aqd
Air Quality Division, P.O. Box 30260, Lansing, MI
48909-7760.
Peeler, J. W. September 1990, "CEM System Perfor-
mance Specifications and Quality Assurance Re-
quirements for Municipal Waste Combustors,"
NESCAUM Guideline, available on EMTIC BBS.
U.S. Environmental Protection Agency. 1992. Tech-
nical Implementation Document for EPA's Boiler and
Industrial Furnace Regulations, Office of Solid
Waste, 401 M Street, S.W. Washington, D.C.
U.S. Environmental Protection Agency. 1993. En-
hanced Monitoring Reference Document, Office of
Air Quality and Standards, Emission Measurement
Branch.
U.S. Environmental Protection Agency. July 1994a.
Final HQN Rule, Reference Volume APTI Course # 1-
050-95 APTB 050-95-B.
U.S. Environmental Protection Agency. 1994b. The
Continuous Emission Monitoring Guidance for Part
503 Sewage Sludge Incinerators, Office of Water,
EPA 833-B-94-003.
U.S. Environmental Protection Agency. 1995a. Acid
Rain Program Emission Scorecard 1994, EPA 430/R-
95-012.
U.S. Environmental Protection Agency. 1995b.
Open Market Trading Rule for Ozone and Smog Pre-
cursors: Proposed Policy and Model Rule, 60 FR
39668-39694, corrected 60 FR 44290-44296.
25
-------
Chapter 3
Monitoring Technology - Instrumentation
Continuous monitoring of non-criteria gases may require
specialized sampling and analysis procedures since the
measurement of reactive and condensable gases can be
particularly difficult. An understanding of the composi-
tion of the stack gas stream, the behavior of the com-
ponents of interest, and the potential physical and
chemical reactions that may occur in the stack or within
the sampling system is necessary to understand and
gauge the sampling problems that may be encountered.
Sampling approaches used in the measurement of crite-
ria pollutants can be applied to the measurement of
hazardous air pollutants if sufficient care is taken. Once
a representative sample can be delivered to an analyzer,
a number of options are available for measuring the
concentrations of organic compounds, paniculate mat-
ter, and metals. Approaches to both sampling and
analysis are discussed in this chapter.
3.1 Monitoring Systems for Non-criteria Gases
Solutions to sampling problems for non-criteria gases
are offered to varying degrees by different system
configurations including hot/wet extractive systems,
dilution systems, close-coupled systems and in-situ
monitoring systems. Each configuration has its
strengths and weaknesses. The sampling system cho-
sen must be compatible with the analytical instrumenta-
tion used to measure the gas concentrations. Selection
of both the sampling and analytical systems will depend
greatly on the chemical characteristics of the pollutant.
3.1.1 Sampling Problems For Reactive and
Condensable Gases
Reactive and condensable gases such as HCI, NH3, and
formaldehyde present great measurement challenges.
Such gases may react with other components within
the stack gas stream; they may condense or be ab-
sorbed by liquid condensate within an extractive sam-
pling system, they may adsorb onto surfaces, or they
may polymerize before reaching the analyzer. Where
these and other related phenomena occur, measurement
results will be affected. The extent of these effects
range from introducing bias into the data to completely
invalidating all measurements. Continuous monitoring
of reactive and condensable gases is much more diffi-
cult than monitoring of criteria pollutants. Depending
upon the components making up the flue gas stream,
special sampling equipment may be needed and spe-
cial operation and maintenance procedures may be
required to achieve reliable results.
3.1.1.1 Surface Adsorption
Different compounds may adsorb onto the surface of
various materials within the sampling system compo-
nents and therefore be removed from the sample
stream before reaching the analyzer. The extent of
the adsorption depends on many factors, including:
the physical properties of the compound of interest,
the gas concentration of the analytes and interfer-
ences, the type of material comprising the adsorption
surface, the amount of exposed surface area, the
surface condition, the gas and surface temperatures,
and the time needed for the adsorption process to
reach equilibrium. The effects of many of these fac-
tors are interdependent and vary with age and previ-
ous use (or abuse) of the sampling system.
The selection of appropriate sampling system materi-
als is important in minimizing the adsorption of many
compounds. In general, selection of inert materials
minimizes adsorption but several other factors must
be considered. For example, measurement of ppm
levels of criteria pollutants such as S02 and NO „ usu-
ally can be accomplished with sampling system com-
ponents fabricated of high quality stainless steel,
Teflon, or glass. In many applications, Teflon is view-
ed as a completely inert material. However, less ad-
sorptive materials, such as poly ether ether ketone
(PEEK) may be required for tubing used in measure-
ment applications of ppb level of organic compounds.
Similarly, the use of Teflon is not recommended for
measurement of "organofluoro" compounds because
they are chemically similar to the Teflon polymer and
may result in a positive measurement bias. Also,
studies have been performed that demonstrate that
certain materials can permeate Teflon {Dunder, 1995),
In general, increasing the surface temperature of a
solid will reduce gas adsorption on the solid. Heat
transferred from the surface increases the internal
energy of the adsorbed molecule helping it to over-
come the weak molecular attraction and escape from
26
-------
the surface. For highly adsorptive gases, sample lines
may be operated near the physical temperature limits.
However, for some compounds such as CO, permeation
through the walls of Teflon tubing occurs to a greater
extent at higher temperatures.
The presence of "active sites", e.g., irregularities in a
material surface at the molecular level, provides loca-
tions for chemisorption or the formation of weak chemi-
cal bonds with gas molecules that significantly affect
the adsorption of gases. In the measurement of low
concentrations of organics, highly polished stainless
steel vessels with thermally deposited nickel are used to
minimize the effects of adsorption. The virgin surface
is virtually free of active sites and good sample recovery
efficiencies have been obtained for many compounds.
When small amounts of moisture are present in the
samples less adsorption will occur. Speculation is that
the water molecules preferentially occupy the reactive
sites and thus minimize adsorption of other analytes.
Similarly, improved recovery efficiencies have been ob-
served in the presence of 2 percent moisture by volume
(Peeler, 1996).
Surface corrosion, due to the deposition of acids or
other factors, creates many active sites in metal sur-
faces and greatly changes the adsorptive effects of the
surface for many compounds. The potential for degra-
dation of a sampling system over time is great because
stack gases often contain significant concentrations of
sulfuric, nitric, hydrochloric, or other acids. A sampling
system that initially performs very well may be rendered
completely incapable of transporting HCI to the analyzer
due to corrosion caused by the accumulation of conden-
sate within the system. In addition, deposits of particu-
late matter on filters or other surfaces within the sam-
pling system may increase adsorption greatly. For these
reasons, the age and history of use of a sampling sys-
tem affects its performance. Therefore, periodic checks
of the sampling system are required.
Variations in the gas concentration, or changes in other
parameters, affect the adsorption equilibrium resulting
in subsequent increased adsorption or desorption of the
compound of interest. Adsorption affects the response
time of an extractive sampling system; the greater the
adsorption, the longer the time required for a measure-
ment system to display a stable and fully equilibrated
response to a step change in gas concentration. The
effects of adsorption may sometimes be determined
through response time tests using dry calibration gases.
However, as discussed above, the adsorption also may
be affected by the presence of moisture or other com-
ponents in the stack gas samples. Small unswept vol-
umes within the sampling system (e.g., calibration injec-
tion lines) can mimic the effects of adsorption by allow-
ing analytes to diffuse during sampling, thus eon-
founding attempts to quantify adsorption. Because
sampling systems often are fabricated of many com-
ponents and different materials, isolating adsorption
problems can be quite difficult.
3.1.1.2 Solubility and Condensation
The stack gas streams at stationary sources contain
compounds that will condense if the sample tempera-
ture is reduced. Depending on the type of sampling
system that is used, water vapor may condense as
the sample temperature is lowered. Other com-
pounds, such as sulfuric acid, may condense along
with water vapor to form acid condensate, or sulfuric
acid may condense even when the sample tempera-
ture is maintained above the moisture dew point.
Condensate formed by cooling stack gas samples can
be a complex mixture of substances. This condensate
may be detrimental to the sampling system materials
and cause corrosion or other problems.
Some compounds, such as HCI and NH3, are highly
soluble in water. The presence of condensate within
the sampling system will scrub water soluble com-
pounds such as HCI. Obviously, if HCI is the com-
pound of interest, the presence of condensed mois-
ture in the sampling system will invalidate measure-
ments. In other cases, where measurements of insol-
uble compounds are made in a sample stream that
contains HCI, the presence of condensed moisture in
the sampling system may protect the analyzer from
damage. Thus, a decision to use a condenser system
to remove moisture depends on the solubility of the
compound of interest and other materials present in
the stack gas matrix.
Industrial process emission streams will reflect a wide
range of moisture contents, depending on the nature
of the process and the type of control equipment that
is installed. The moisture content that will be encoun-
tered in a particular application must be known; 1) to
size condensers or dryers used for moisture removal,
2) to select an appropriate dilution factor to maintain
the sample above its dew point, or 3) simply to gauge
the significance of the problem. The moisture content
of combustion source exhaust streams typically
ranges from 8 to 20 percent by volume depending on
the fuel combusted. Hazardous waste incinerators
with quench towers and spray dryers may have emis-
sion streams containing as much as 50 percent water.
Portland cement kiln exhaust streams are likely to
contain moisture ranging from 10 to 35 percent by
volume depending on the type of process (i.e., dry
process, precalciner or wet process kiln). On the
other end of the spectrum, sulfuric acid plants will
have no moisture at all in the stack gas. The moisture
27
-------
content may be estimated based on knowledge of the
process or it may be measured directly over the range
of process/control equipment operating conditions.
The absorption of a somewhat water soluble component
will reach an equilibrium between the liquid and gas
phases given sufficient time and a constant concen-
tration of the component in the gas phase. (This is not
the case where chemical reactions with other compo-
nents occur in solution,! However, most sampling sys-
tems form new condensate continuously, and the
concentration in the gas sample may also change with
time. Therefore, the application of equilibrium solubility
constants may be inappropriate, or at best, an indica-
tion of a one-sided limit for estimating the extent of this
phenomena. The design of condensers in extractive
systems should minimize contact of the condensate and
the sample stream to minimize the absorption of slightly
water soluble compounds. That the condensate not be
allowed to accumulate is also important; it should be
continuously removed to minimize absorption of soluble
compounds and opportunities for reactions with other
stack gas constituents.
Solubility losses of many pollutants are understood
poorly. VOC or organic hazardous air pollutant monitor-
ing applications may contain mixtures of soluble,
slightly soluble, and insoluble components. Little quan-
titative information is available in the literature. A few
studies at hazardous waste incinerators and sewage
sludge incinerators have indicated that heated sampling
systems with total hydrocarbon (THC) analyzers mea-
sure higher concentrations of volatile organics than
systems that include refrigerant moisture removal sys-
tems (Cone, 1989). Thus, some regulatory applications
specify the use of a heated system even though opera-
tional problems are reduced extensively with a "cold"
system. Unfortunately, the identity of the specific
compounds that are removed in a water condenser and
the degree to which they are removed are largely un-
known because THC systems provide no information
about the individual organic species that are present.
A few studies have been performed using condensers
with either FTIR or GCMS analyzers where a variety of
organic compounds were dynamically spiked into sam-
ple streams (EPA, 1993; Peeler, 1996). The investiga-
tors of these studies have demonstrated acceptable
sample recovery efficiencies for certain water soluble
compounds in the presence of 8 to 35 percent water
vapor. (See Chapter 7 discussions regarding dynamic
spiking procedures.)
3.1.1,3 Chemical Reactions
Chemical reactions can occur between various stack
gas components resulting in the formation of new
chemical species. Because of the complex nature of
some stationary source emissions, it is difficult to
determine the extent of these chemical reactions.
Polymerization, neutralization, and sublimation/con-
densation reactions are suggested frequently by scien-
tists, engineers and technicians as causing sampling
and analytical problems. Many undocumented myths
exist about the various chemical reactions both within
the stack and within the sampling system. Determin-
ing whether chemical reactions occur in the stack or
in the sampling system when performing extractive
testing is very difficult, because the analyzer only
detects what the sampling system can deliver. Detec-
tion of chemical reactions that occur within the
source virtually are impossible to determine because
some species are short lived, white others reach a
state where the products of the chemical reaction are
in equilibrium with the reactants.
Polymerization Reactions. Polymerization reactions
are those that occur when a compound reacts with
itself (or a similar compound) to form a large mole-
cule. An example of this phenomena is the reaction
of formaldehyde with itself to form paraformaldehyde.
Paraformaldehyde is a solid that may be formed in
sampling systems where cold spots exist. Individual
formaldehyde molecules react with each other at the
surface of these cold spots to form a polymer layer.
This deposition results in a negative measurement bias
for formaldehyde. The results of polymerization reac-
tions manifest themselves in a manner similar to sur-
face adsorption.
Neutralization Reactions. Neutralization reactions are
those reactions that have a net effect in reducing an
acidic or basic component in the stack gas. In the
pure sense, neutralization reactions usually occur in
solution where hydrogen ions (H+) and hydroxide ions
(OH1 react to form water. In the gas phase, neutral-
ization may occur by solubilization first and neutraliza-
tion second, or by direct adsorption of components
onto particulate matter. The apparent removal of HC1
by CaO (lime) in baghouses following spray dryers on
the particulate filter cake of a sampling system are
examples of this type of reaction. Reaction of HF
with silica in glass surfaces to form SiF4 is another
neutralization reaction.
Salt-forming Reactions. Salts can form when two or
more gaseous compounds react. An important exam-
ple is the equilibrium reaction between gaseous hydro-
chloric acid (HCI) and ammonia (NH3) to form solid
ammonium chloride {NH4CI). Both HCL and NH3 are
volatile, non-condensable gases while ammonium
chloride is a water soluble solid compound having an
exceptionally low vapor pressure. This reaction is
28
-------
known to occur under atmospheric conditions {Seinfeld,
1986), and may cause either positive or negative mea-
surement biases that are dependent upon the stack gas
and sample delivery temperatures.
Another example of a salt-forming reaction is the combi-
nation of S02/ NH3, and water to form ammonium bisui-
fate. This reaction can occur in the atmosphere down-
stream of a source, producing a detached plume and in
some cases participate fall-out. This reaction can also
occur within the condenser of a CEM system creating
low biases for either S02 or NH3 measurements.
3,1.2 Solutions to Sampling Problems
Extractive sampling systems must be designed and
operated in a manner that provide consistently represen-
tative samples to the analyzer. The design of the sam-
pling system must eliminate, or at least minimize, any
reactions or loss of the analytes of interest before they
reach the analyzer. The operation of the sample acqui-
sition and sample handling components must ensure
that the necessary conditions are maintained over the
complete range of source operating conditions to afford
representative measurements of reactive and conden-
sable gases.
3.1,2.1 Hot/wet Systems
"Hot/wet systems" are extractive CEM systems that
maintain the sample temperature above its dew point
throughout the sampling system and within the ana-
lyzer. These systems may be used for the measurement
of water soluble compounds and commonly are used for
monitoring non-criteria pollutants such as HCI, NH3, and
VOCs. Other compounds can also be measured pro-
vided that a suitable heated analyzer is available. Hot/
wet systems have been used for many years to monitor
criteria pollutant emissions such as SQ2, NOX, and CO.
Important aspects of hot/wet sampling systems for
reactive and condensable gases are illustrated in Figure
3-1.
Hot/wet sampling systems must not only maintain the
sample temperature above the dew point to avoid con-
densation, they must also minimize adsorption and
avoid the potential for chemical reactions to occur.
Consider for example, monitoring of HCI which is 1)
very water soluble, 2) adsorbs onto common sampling
system materials, and 3) participates in chemical reac-
tions with other stack gas constituents at certain
sources. Extractive HCI sampling systems have been
used that minimize adsorption effects by operating
Teflon sampling lines at temperatures between 350°F
and 375°F (the maximum operating temperature for
Teflon). High sampling flow rates are maintained (20
liters per minute) and short sampling lines are used in
these systems to minimize the residence time and to
minimize the surface area for adsorption. Heated
head pumps fabricated of 316 stainless steel or other
special alloys and with Teflon diaphragms are used to
avoid condensation or adsorption. All cold spots
within the sampling system such as connections be-
tween heated line segments or connections to pumps
or manifolds must be eliminated. Nothing less than
meticulous attention to ensuring that the entire sam-
ple path is heated will prove adequate. The perfor-
mance of "system calibrations" where calibration
gases are introduced at the outlet of the sampling
probe are very important for these types of systems
to identify and/or account for the effects of adsorp-
tion (see Figure 7-5), System calibrations are some-
times avoided in practice because of the high con-
sumption rate of calibration gases (and corresponding
cost) due to the high sampling rates and the longer
time required to achieve an equilibrated instrument
response.
HCI also serves as an example of a chemically reac-
tive component. HCI may participate in chemical
reactions with lime or similar materials collected with-
in the sampling system. Special provisions are re-
quired to minimize the collection of the particulate
matter where a stack gas stream contains solids or
liquids that are chemically reactive with the compo-
nent of interest. Since the reaction of HCI with par-
ticulate material is most likely to occur on filters, the
equipment must be designed to accommodate system
calibrations where calibration gases are introduced
upstream of the filters. This may help to detect
whether reactions with the particulate matter are
occurring. (Procedures described in Chapter 7, Dy-
namic Analyte Spiking, provide a method to detect
such chemical reactions.)
As previously described, HCI may participate in reac-
tions with ammonia to form ammonium chloride salts.
As discussed before, this reaction is sensitive to the
sample temperature. Where the sample temperature
within the analyzer is substantially below the stack
gas temperature, the reaction may consume HCI and
thus introduce a negative (low) bias in the HCI mea-
surement results. Conversely, where the measure-
ment system is maintained at a higher temperature
than the stack temperature, particulate ammonium
chloride may volatilize to form HCI and NH3 thereby
creating a positive bias in HCI monitoring results. The
potential for such biases to occur at particular
sources, and the effects of temperature changes be-
tween the stack and the analyzer, must be examined
for hot/wet measurement systems.
29
-------
L/ Primary particutate filter
r,-v.- , c (maintained above
.- ... •
• "" , ., .Heated'
"*/• '..-•' probe-,
.."";'., ./sack" "". •
Calibration
gases /
preheated x
and
injected
before R
filter met
U ^4
\
W ;
er I
f
!
L_
"
f
sample dew point)
350* F
heated
sample line
(connections
insulated)
1
\
All tubing connections
and components fn
heated compartments
r HiD
l^brat
0
w
$ <3 saes
. L__, „„.]
hea
Ventf
L
Flow Flowpf
controls metergi
rzsl?
/ Second
h volume particul,
ted-head fliter
pump
ary
ate
350° F
heated
sample
line
,^|Myze$4B
Figure 3-1. Hot/wet sampling systems.
3.1.2.2 Dilution Systems
Dilution systems quantitatively dilute stack gas sam-
ples with clean dry air to reduce the relative moisture
content so that the sample is maintained above the
dew point with little or no heating. Because moisture
condensation Is eliminated as a potential problem,
heated sample lines and manifolds can be replaced
with simpler, less expensive components. Sample
gas condensers or permeation dryers are eliminated
also. Usually an aspirator or eductor is used to move
both sample and dilution gas thus, eliminating the
need for a sample pump. On the other hand, clean-
up systems to remove moisture, CO2, oil, hydrocar-
bons, or other components from the dilution gas may
be necessary.
In-stack dilution probes use critical orifices to control
sample flow rate and aspirators to both draw stack
gas through the critical orifice and supply dilution air
(Figure 3-2).
The critical orifice in the dilution probe ensures that
the sample extraction rate is independent of the aspi-
rator vacuum thus providing a constant sample flow
rate and consistent dilution of the sample gas. Cali-
bration gases are introduced upstream of the critical
orifice and are diluted in exactly the same manner as
stack gas samples. In-stack dilution probes are avail-
able from many manufacturers and have been used
widely in monitoring criteria pollutants, particularly in
the acid rain program. Dilution ratios ranging from
Diluted
sample gas
Venturt throat
Ejector pump
Dilution air in
Critical orifice
Calibration gas
Glass wool
Figure 3-2. In-stack dilution probe.
30
-------
20/1 to 1,000/1 are used in practice. Analyzers that
were developed originally to monitor S02 and NO x
concentrations in ambient air are used for emission
monitoring in the acid rain program.
Variations in the sample dilution ratio, and resulting
biases in monitoring results, may occur where in-
stack dilution probes are subject to varying stack
temperature, pressure, or molecular weight. These
effects have been characterized by Jahnke (Jahnke,
1994a), Newer designs of dilution systems have
located the critical orifice and aspirator outside of the
stack in a temperature controlled region to minimize
the effects of stack gas temperature variations on the
dilution ratio (Figure 3-3).
sample out
Flow Control Ortffce Dilution Ecfuctar
SimpTirtg
P.pbf . -".-•'
ample,
gash" '
Sample Conditioning Enclosure
Figure 3-3. Out-of-stack dilution system.
A dilution sampling system is not appropriate where
an analyzer with the requisite sensitivity is not avail-
able to reliably measure the diluted samples. Usually,
the minimum dilution ratio is determined based on the
maximum expected moisture content of the stack gas
and the minimum temperature in the sampling system
or analyzer. Applying this dilution factor to the ex-
pected stack gas concentration provides an estimate
of the required measurement range. An analyzer with
sufficient sensitivity, resolution, and signal-to-noise
capability must be available for this operating range.
The analyzer must be designed for the specific
measurement range; incorporating measurement cells
with sufficient optical path length and appropriate
filters, detectors, and other devices necessary for the
measurement level. Simply increasing the electronic
gain of an analyzer does not always change the mea-
surement range.
Problems with adsorption may be encountered with
dilution systems even though condensation of mois-
ture is avoided. EPA studies to evaluate HCL dilution
system performance at municipal waste combustors
and at hazardous waste incinerators demonstrated
very slow response times that were attributed to
adsorption of the HCL on Teflon sample lines
(Shanklin, 1989). The use of heated sample lines can
improve the system response times and can provide
additional protection against condensation in applica-
tions where extremely cold ambient temperatures are
encountered. However, one of the major advantages
of using a dilution system is lost if heated sample
lines are required.
The accumulation of organic material within the aspir-
ator of an in-stack dilution probe has been observed
during monitor evaluation tests at a power plant in
Virginia. This was attributed to localized condensa-
tion occurring because of cooling by the dilution air.
Pre-heating of the dilution air eliminated this problem.
In addition, condensation of acids may occur in dilu-
tion sampling systems even though condensation of
moisture is avoided. Accumulation of sulfuric acid
has been observed in unheated dilution sampling lines
at coal-fired electric utility boilers.
One of the major advantages of dilution sampling
systems is that they minimize the volume of sample
gas extracted from the stack and thereby minimize
the contamination of the system by particulate mat-
ter. The frequency of replacing filters and other
maintenance activities is reduced for dilution systems
because less particulate matter is introduced to the
system. Nevertheless, the relative amounts of
particulate matter and gases extracted is the same
for dilution and conventional extractive systems.
Therefore, the potential for chemical reactions or
adsorption between gases of interest and the
particulate matter is not reduced.
3.1.2.3 Close-Coupled Systems
Close-coupled systems minimize extractive sampling
components by effectively placing the measurement
sensor in close proximity to the sampling point.
Many of the problems observed in other extractive
sampling systems are eliminated. Close-coupled
systems have been developed for the measurement
of criteria pollutants (Mandel, 1995). Close-coupled
systems also have applications in monitoring non-
criteria air pollutants particularly reactive and
condensable gases. Several fundamentally different
configurations have been developed.
Close-coupled systems have been developed that use
FID detectors in "total hydrocarbon" monitoring
systems as shown in Figure 3-4. In this example, the
heated FID is located in a thermally controlled
enclosure just outside the stack wall. The sample is
conveyed only a very short distance, thus the surface
area for adsorption/desorption reactions and the time
allowed for reactions to occur before the sample
31
-------
reaches the detector are greatly minimized. This
design allows for rapid responses of the measurement
system and minimizes sampling system maintenance.
Figure 3-4. Close coupled system.
Close-coupled systems have been developed that use
"reactive gas analyzers" for the measurement of NH3,
SO3, and H2SO4. These measurement systems incor-
porate a means of contacting the sample gas stream
with an absorbing solution at the outlet of the sam-
pling probe. The absorbing solution may simply ab-
sorb the compound of interest or it may facilitate
chemical reactions which convert this component to
a more stable form. The chemical solution is con-
tinuously renewed and after exposure to the gas
sample it is conveyed some distance to an analyzer
which provides for the determination of concentra-
tion. This approach provides for the immediate reac-
tion of the component of interest, and possibly the
selective chemical removal of interfering species, to
make the measurement. Many of the precautions
evident in other systems that are necessary to avoid
condensation, minimize adsorption, or avoid chemical
reactions are not needed in this approach.
Another close-coupled system configuration has been
introduced which uses a solid state tunable diode
laser analytical method (Frish, 1996). This technique
is capable of measuring many components in the
infrared spectrum and can be used for monitoring
NH3, HF, H2S and other toxic gases. In these sys-
tems, the laser source and photodelectors necessary
for the measurement are contained in a control mod-
ule at a remote convenient location as are analyzers
in extractive monitoring systems. A fiber optic cable
is used to connect the control module to a sample
probe. The probe provides for the continuous flow of
stack gas through an optical cell mounted in a ther-
mally controlled chamber immediately outside the
stack as shown in Figure 3-5.
Because this measurement system can employ fiber
optic cables as long as 1 km, the advantages of a
close-couple optical measurement eel! and the conve-
nience of a remote analyzer are provided. In addition,
a number of sample probes can be connected to the
same control module thereby offering additional cost
savings for applications where measurements are
required at several locations.
3.1.2.4 In-Situ Measurement Systems
In-situ systems for the measurement of gases sense
the concentration of the gas of interest within the
stack by either placing a detector within the stack or
by projecting a light beam through a portion of the
stack gas stream and analyzing various spectral phe-
nomena. Point in-situ systems measure the concen-
tration at a specific point or over a relatively short
path length through the stack gas. Cross-stack sys-
tems project a light beam across the stack. These
systems may be either single pass or double-pass
systems depending on whether the light source and
detector are on the same or opposite ends of the light
path. A double pass system is illustrated in Figure 3-
6.
in-situ systems for measurement of criteria pollutants
are described by Jahnke (Jahnke, 1993). Regardless
of the configuration, all in-situ systems must be de-
signed to determine the concentration at stack condi-
tions, which typically involves varying temperature,
pressure, and moisture content. Other factors, such
as particulate loading or the concentration of interfer-
ing gases, also vary and may affect the measurement
process or accuracy of results.
An advantage of in-situ systems is that many of the
sampling problems associated with extractive sys-
tems are eliminated. Assuming that the stack gas
temperature is above the dew point, condensation is
not an issue. Adsorption of gases is irrelevant. Re-
actions between gases and particulate matter can be
ignored for all in-situ systems except for those that
use a thimble or filter to protect an in-stack detector.
In-situ analyzers are particularly appropriate, and in
some cases, the only option for the measurement of
reactive and condensable gases because the influ-
ences of the extractive sampling system are elimi-
nated.
On the other hand, in-situ analyzers may need to
compensate for variations in gas density due to tem-
perature or pressure variations, variations in spectral
absorption due to temperature shifts, as well as varia-
tions in particulate matter loading and the presence
of other interfering species. The method used to
32
-------
Fiber Optic
Cable Optical
(up to 1 km Song) Measurement
Cell
Remote Microprocessor and
Laser Emitter Subsystem
.' Filtered Probe
in Stack
Figure 3-5, Close-coupled laser monitoring system.
Calibration Retro-
Half-silvered filter mirror
mirror
Lamp
Detector
Figure 3-6. Double-pass transmissometer.
compensate for these types of factors is inherent to
the analytical technique. Consider for example, an
optical infrared dispersive device. Variations in the
effluent temperature change the gas density and thus
the number of molecules present in the adsorption
path at a particular concentration. Sensing of the
effluent temperature is necessary to distinguish
between changes in infrared adsorption due to
concentration variations or due to temperature
variations. Also, the actual spectra! adsorption of
infrared radiation varies as a function of temperature
for different compounds. Thus, effluent spectra and
reference spectra obtained at the same temperature
must be "matched" in order to analyze the data
accurately.
Temperature compensation for spectral adsorption
can be performed by obtaining a series of spectral
libraries at different temperatures on a controlled
calibration facility. This approach has been used for
infrared gas filter correlation devices and UV differen-
tial adsorption instruments. A different approach has
33
-------
been developed for dispersive infrared measurements
that relies on HITRAN reference spectra and mathe-
matical transformations to adjust the reference
spectra to correspond to the effluent temperature
(Lang, 1991).
A major challenge for in-situ analyzers is the ability to
verify proper calibration while the instrument is
installed. Because the effluent is present, determin-
ing if the zero concentration value is correct is
difficult. Various schemes have been used to at-
tempt to overcome this problem. Calibration at
saturation (very high concentration) rather than at
zero, incremental calibrations superimposing gas
filters and the effluent gas, and using concentric
slotted pipes or other mechanical means to tempo-
rarily provide a zero calibration have all been used
with varying degrees of success. An in-situ monitor
with a slotted probe for effluent measurements and
gas audit cell to facilitate quality assurance checks
using external calibration gases has been developed.
The probe design ensures that calibration gases are
at the same temperature as the effluent gases for
reasons previously discussed. Another single-pass
cross-stack in situ analyzer uses a zero pipe to
provide a reference optical path, free of adsorption,
and a flow-through gas cell to facilitate the introduc-
tion of calibration gases. This instrument is an
ultraviolet differential absorption instrument that can
measure many gases including S02, NOX, H20, NH^
volatile organic compounds, and Hg (vapor).
In-situ devices typically isolate optical components
from the effluent stream by using optical windows
and an air-purge system that provides a flow of
filtered ambient air across the optical surfaces and
then into the stack. The analytical technique must be
insensitive to any dust accumulation on the optical
surfaces; otherwise, the decrease in light transmit-
tance might be interpreted as an increase in pollutant
concentration. The sample interface system must be
adequate to ensure that dust accumulation is held to
acceptable levels between maintenance intervals.
Otherwise, the intensity of the optical beam may be
diminished to the point where deterioration in signal
to noise levels reduces the accuracy of the measure-
ment results. Optical windows must be fabricated of
materials that transmit the measurement wavelengths
and are resistant to chemical reactions and mechani-
cal deterioration.
3.1.3 Analytical Techniques
The analysis of hazardous air pollutants (HAPs) is not
as straightforward as the measurement of the inor-
ganic, criteria pollutants such as S02 and NOX.
Because of the wide variation of properties associ-
ated with the different classes of HAPs (organic
compounds, metals, particulate matter), numerous
methods are used to analyze the flue gases after they
are sampled by the extractive or in-situ systems.
These include gas chromatographic methods used for
analyzing organic and inorganic compounds, light
absorption and scattering methods used for particu-
late monitoring, and atomic emission spectroscopic
methods used for the analysis of metals.
Due to the difficulties of analyzing multiple hazardous
air toxic materials in a flue gas matrix, chromato-
graphic separation techniques often are employed to
separate compounds in a gas mixture. Compounds
can then be measured individually by some type of
detector, such as a flame ionization detector (FID),
thermal conductivity detector (TCD), photoionization
detector (P1D), or electron capture detector (ECD).
Other techniques where separation of the gaseous
compounds is not performed, such as Fourier trans-
form infrared (FTIR) spectroscopy, ideally identify
and quantify all of the compounds in the sample at
the same time. A combination of separation and
analytical methods such as gas chromatography and
mass spectrometry (GCMS) can also provide for a
versatile analytical system. This section contains a
review of hazardous air pollutant monitoring methods
that are commercially available, describing principles
of measurement.
3.1,3.1 Gas Chromatography
Gas chromatography typically is used to isolate the
individual components of a mixture of organic and
inorganic compounds from each other for subsequent
identification and quantitative analysis. Chromato-
graphic separation principles are used in EPA refer-
ence methods (EPA, 1996b). For example, a detailed
gas chromatographic procedure is specified in EPA
Method 106 for vinyl chloride, EPA Method 16
provides for the chromatographic separation of four
total reduced sulfur (TRS) compounds, and Method
18 gives general sampling and analytical criteria for
gas chromatographic testing. The use of gas chro-
matography for CEM regulatory applications has been
limited; however, many installations are found in the
process industries where the equipment is used to
monitor production operations (Villalobos, 1975;
Coleman, 1996), Particularly in the refining and
chemical industries, resources necessary to provide
for continuous, accurate data are made available
because of the importance of that data for determin-
ing process efficiencies.
3.1.3.1.1 Basic Principles of Gas Chromatography.
Gas chromatography is based on the selective distri-
bution of compounds between a stationary phase and
34
-------
a mobile phase {carrier gas). In this process, the
moving gas phase passes over a stationary material
that is chosen to either absorb or adsorb the organic
molecules contained in the gas. In gas chromatogra-
phy, the stationary material or phase can be either a
liquid or a solid and is contained in a long, thin tube
referred to as a "column." Columns are made of
fused silica, glass, or stainless steel and vary in
diameter depending upon the type of column packing
(the stationary phase) used.
In an ideal column operated under ideal conditions,
each molecular species will exit the column at a
different time. The equilibration between the individ-
ual compounds and the column stationary phase is a
function of 1} the compound's affinity for the station-
ary phase relative to the mobile phase, 2) the temper-
ature of the column, and 3) the flow rate of the
mobile phase carrier gas. Individual molecules are
separated in the column by undergoing a series of
equilibrations between the stationary and mobile
phases (Giddings, 1965). Selecting the appropriate
column and optimizing the column temperature and
carrier gas flow rate should enable separation of the
gas sample into its individual components.
3.1.3.1.2 Gas Chromatograph Components, A
simple chromatographic system is composed of a
device for injecting the sample into the column, a
carrier gas to sweep the sample gas through the
column, a column oven, and a detector. The carrier
gas, such as helium or nitrogen, sweeps the sample
from the injection area into the heart of the system,
the column. The detector at the end of the column
ultimately produces an electrical signal that is propor-
tional to the quantity of molecules present (Figure 3-
7).
Separated peaks are identified and quantified by
comparison of their peak areas to a calibration with
known gas standards. However, for an unknown
sample mixture, one will not know what standards to
include in the calibration unless some prior knowledge
of the flue gas composition is available. In that case,
one must use techniques such as infrared, ultraviolet,
or mass spectrometry to first identify the compounds
and then select the column and appropriate stan-
dards.
3.1.3.1.3 Detectors. The separation performed in a
chromatographic column is sensed by a detector and
recorded. Any detector designed for use in a gas
chromatograph system must have a high sensitivity
for low concentrations of organic molecules, and a
rapid response time. Many detectors are available
that meet these requirements; the most common in
source monitoring applications are the flame ioniza-
tion detector (FID) and the photoionization detector
-------
Flame lonizatlorL Detectors. The flame ionization
detector is capable of sensing most organic com-
pounds and, because of its relatively high sensitivity,
has become widely used in environmental applica-
tions. In an FID assembly, the column effluent enters
the base of the burner, is mixed with hydrogen, and
the mixture burned in a jet with oxygen at a tempera-
ture of about 2,100°C. Ions and free electrons are
produced by the flame, which increases the current
sensed by an electrometer. The current is approxi-
mately proportional to the number of carbon atoms
entering the flame. However, the response of the
detector is slightly different for different types of
organic compounds. As a result, the detector must
be calibrated for the compounds being studied to
achieve the best accuracy (Figure 3-8).
Exhaust
Measuring
Circuit
Air./ V Sample and hydrogen
Figure 3-8. Flame ionization detector.
The FID is convenient to use in source sampling
situations since it does not respond appreciably to
gases such as 02, N2» H20, CO, S02, and NO.
However, organic compounds that contain nitrogen,
oxygen, or halogen atoms may give a response
reduced from that seen from hydrocarbons. In a
photoionization detector, organic molecules are
ionized by ultraviolet light;
R + hv
+ + e"
where R+ is the Ionized organic compound and hv
represents the energy of the light having frequency v
(h = Planck's constant}.
Photoionization Detectors. A typical PID incorporates
a UV lamp suitable for ionizing the analytes of inter-
est and a pair of electrodes to measure a current
proportional to the concentration. Again, the major
components of the flue gas such as O2/ CO, N 2, CO2
and H2O are not ionized.
PIDs are used in conjunction with gas ehromato-
graphs or alone as portable analyzers used in EPA
Reference Method 21 for detecting leaks in petroleum
refineries (Hellwig, 1986). They offer advantages
over FID detectors in that hydrogen is not required
and methane (a gas that is not required to be moni-
tored) is not ionized and therefore not detected by a
PID (Hewitt, 1981). Depending upon the instrument,
compounds detected by PIDs include; aliphatic and
aromatic hydrocarbons, halogenated organics, alco-
hols, ketones, aldehydes, ethylene oxide, vinyl
chloride, and inorganic compounds such as arsine,
phosphine, and hydrogen sulfide. In general, com-
pounds that have ionizable electrons can be detected.
Electron Capture Detectors. The electron capture
detector (ECD) is selective for certain groups of
organic compounds such as those containing halogen
atoms or nitro groups. The electron capture detector
works by using a radioisotope treated electrode that
emits high energy electrons as it decays (P emission).
The P electrons react with the carrier gas to produce
secondary, free electrons which move to a positively
charged anode to generate a current through the
system. When the nitrogen carrier gas contains
electron-absorbing molecules such as the halocar-
bons, the electric current will be reduced because the
flow of free electrons will be reduced by the absorp-
tion.
The ECD is more sensitive for specific groups of
compounds than is the FID, but the response can
again vary from compound to compound. Issues
associated with the transport, storage, and disposal
of the radioactive material have also been of some
concern.
Other Detectors. Other types of detectors are used
in ehromatographic systems applied to environmental
monitoring. Many of these are species specific like
the ECD. Among these are the Hall electrolytic
conductivity detector (HECD), used for halogen,
sulfur, or nitrogen compounds; the flame photometric
detector (FPD) used for sulfur or phosphorus com-
pounds; and the alkali flame detector (AFD), used for
nitrogen and phosphorous compounds.
Mass Spectrometers used as Detectors. Increasing
demands for the analysis of trace levels of toxic
36
-------
materials require new approaches to increase the
resolving power and sensitivity of chromatographic
systems. Multiple detector combinations have been
used, but combining the gas chromatograph with a
mass spectrometer used as a detector offers one of
the most powerful combinations for both identifica-
tion and quantification. This GCMS combination has
been the basis for most low concentration testing for
many years in air pollution monitoring and is seeing
increasing application (Peeler, 1996).
In the technique of mass spectrometry, molecules are
ionized by high energy electrons, or by other means
(such as chemical or photolonization process), and
the resultant molecular ions and ion fragments are
separated according to their mass to charge ratio
(m/e). This separation produces a "mass spectrum"
of the different ions generated from the fragmen-
tation caused by the high energy electron-molecule
collisions (Figure 3-9), The mass spectrum is unique
to the original molecule, as is an infrared or UV
spectrum.
Mass spectrometers are distinguished by the type of
mass separator used. Magnetic deflectors, time of
flight separators, quadrupole mass analyzers, and
ion-traps have all been used. For environmental
applications, the quadrupole mass analyzer is used
most commonly. In this technique, an oscillating field
of electromagnetic energy filters ions having a spe-
cific mass to charge ratio (Figure 3-10).
The linear quadrupole mass analyzer operates by
oscillating the ions in a radio-frequency field super-
imposed on the charged cylindrical rods. Most of the
ions will oscillate with increasing amplitude and strike
the rods, but one set of rod voltage and radio-fre-
quencies will exist where the ions of a specific m/e
ratio will be able to pass through to the detector.
The radio-frequency or rod voltage is therefore
scanned to obtain the mass spectrum.
O!
s
8. so-
!0 84 87°,
50 60 70
Mass Units
Figure 3-9. Wlass spectrum of meta-xylene.
37
-------
, Non-resonant
Ion
voltages
Figure 3-10. Linear quadrupole mass analyzer.
The mass analyzers used in mass spectrometer
systems are operated under high vacuum (1x10~5 or
1x10"6 mm Hg) to minimize scattering by collision
with other ions and gas molecules. This requires
some type of vacuum pump and a system design that
can maintain a proper vacuum. The cleanliness of
the system is important also, and adsorbed materials
on the walls of the analyzer can lead to the introduc-
tion of interfering ions. These requirements can lead
to problems in source monitoring applications and
must be overcome to achieve consistent results.
The GCMS combination provides both qualitative and
quantitative information since it looks at each differ-
ent type of molecule separately as it comes off of the
column (Figure 3-11). The GCMS output gives a
three-dimensional plot over time, giving information
both on the type of compounds in the sample and the
amount present.
-t
o c
if
Total Ion
chromatogram
Mass spectra
| iL/ of individual
| If GC peaks
IK
Retention time
Selected ion
current profile
Figure 3-11. Total ion current chromatogram.
Mass spectrometers and GC mass spectrometers
have been applied to monitor criteria pollutants on a
continuous basis (Bartman, 1990; Harlow, 1990) and
are being used increasingly in the process industries
to monitor a wide range of HAPs. Their use in
environmental monitoring has focused primarily on
the ambient monitoring of hazardous waste sites or
leak monitoring in the chemical and petroleum indus-
tries. However, GCMS systems are being used
increasingly in short term source tests to determine
baseline HAP emissions (Campbell, 1991; Peeler,
1996). These tests often point out areas where
changes in process operational efficiencies can lead
to emissions reductions. The design of continuous
sampling and operating strategies has been a chal-
lenge in many GCMS applications, but these chal-
lenges are being met (Kinner, 1993; Haile, 1995).
3.1.3.2 Total Hydrocarbon Analyzers
Depending upon the process and the mixture of
compounds present, a preliminary characterization of
the emissions frequently is necessary to optimize
ehromatographic and other instrumentation. The
speciatfon and quantification of volatile organic
compounds can be very expensive. Although such
knowledge may be useful for process control and
optimization, such information is not needed for some
facilities, such as incineration sources. In these or
other cases, a total hydrocarbon analyzer may be
sufficient for monitoring the sum of individual VOC
emissions.
Total hydrocarbon analyzers direct the sample to the
detector without column separation. The sampling
system may be either cold or hot as discussed
previously. The hot FID systems provide a more
accurate measure of the THC content; however, they
are more difficult to operate continuously (see Cone,
1990 for a discussion of this issue). Early hot FID
systems frequently incorporated design flaws where
organic compounds and water could condense at cold
spots in the analyzer plumbing and obstruct the flow
of gas or interfere with the control of the sample
pressure (Cone, 1990). Such problems are avoided
in newer systems.
The FID is the industry standard for total hydrocarbon
(THC) analyzers, and is in fact specified as the
required detector for use in THC monitoring systems
installed in boilers and industrial furnaces that burn
hazardous waste and for those installed in sewage
sludge incinerators (40 CFR 266). Certification
criteria for approving some THC analyzer installations
have been simplified by eliminating relative accuracy
test criteria and using audit gases to check the
system performance (40 CFR 266 Appendix IX and
40CFR 503 Subpart E). These certification criteria
are discussed in Chapter 7 of this Handbook.
3.1.3.3 Light Absorption Techniques
Light absorption techniques have been used tradition-
ally to monitor criteria pollutants such as CO, NO,
S02. The techniques can also be used for the envi-
ronmental analysis of organic compounds and have
been applied extensively both in research laboratories
and in process industries. The light absorption
38
-------
methods are based upon the phenomenon that mole-
cules will absorb light energy to rotate, vibrate, or
change their electronic patterns in characteristic ways.
This absorption occurs only for wavelengths of light
that are in tune with the properties of the molecule {see
for example, Willard, 1987).
Light absorption techniques are categorized as being
dispersive or nondispersive. In the dispersive methods,
the spectral absorption of a molecule is measured over
a limited region of the electromagnetic spectrum. A
spectral absorption pattern, or spectrum, characteristic
of the molecule is obtained that can be used to both
identify the molecule and determine the concentration
of the molecule in the sample. Scanning spectrometers
and Fourier transform infrared (FTIR) spectrometers
generate such spectra.
In the nondispersive methods, the so-called nondis-
persive infrared (NDIR) and nondispersive ultraviolet
(NDUV) techniques, the spectrum is not scanned. Here,
a wavelength where light energy is absorbed is used as
the basis for the instrument design. Such instruments
are constructed quite simply.
Pifjerejtiaj Absorption Spectroscopy. A typical non-
dispersive method measures light absorption at two
wavelengths, one where the molecule absorbs energy
and one where it does not (Figure 3-12).
This particular technique has been called differential
absorption Spectroscopy or differential optical absorp-
tion spectroscopy (DOAS). The ratio of the intensi-
ties, !/!„, at the two wavelengths is known as the trans-
mittance and is related exponentially to the concentra-
tion of the gas that absorbs light energy at the wave-
length, A.
In the nondispersive, differential absorption technique,
I0 is obtained from the detector when it responds at
the wavelength h0. In many criteria pollutant
monitors, the value of I0 is obtained instead by using
a reference gas or reference gas cell that does not
absorb light energy at the measurement wavelength,
i.e., c = 0, which gives I = I0 when the light passes
through the reference cell.
The differential absorption spectroscopic technique
has been used for many years. Early instruments
used filters to select the light wavelengths. Current
methods applied to discriminate between wavelengths
include;
Optical filters
Diffraction gratings and photo-diode arrays
1
2. LUIII cit_.uui i yiaimys aiiu (jnuiu-uiuui
3. Diffraction grating and moving slits
4. Diode lasers
These different techniques are illustrated in Figure 3-
13.
Numerous optical filtering instruments operating in the
infrared region of the spectrum have been developed
for the measurement of gases such as CO, CO2, and
the criteria pollutants, using "Luft" type or "micro-
flow" detectors. These detectors monitor pressure
changes due to differential absorption of light by mol-
ecules contained in the detector cell (Jahnke, 1993).
An optical filtering instrument using a photoacoustie
detector, developed for ambient and industrial appli-
cations, has been used in conjunction with dilution
systems to monitor organic compounds (Sollid, 1996).
This detector monitors acoustic waves resulting from
absorption of light by molecules directly in the sample
ceil.
100-4-
1/3
Molecule absorbs 1 o light
energy at wavelength XQ
- No Absorption,',
F
Molecule absorbs light
energy at wavelength X
Wavelength
Figure 3-12. Transmission spectrum - example illustrating the differential absorption technique.
39
-------
Lamp
Optical
Filters
Detector
Gas Cell
Optical Filters
Diode Array
Detector ^
Lamp
Diffraction
Grating
Diode Array Detector
Lamp
-------
than one gas species at the same time and is suffi-
ciently general that it can be used at a wide variety
of sources. For these reasons, the technique has
generated interest for a number of non-criteria pollut-
ant monitoring applications.
With the advent of compact and powerful micropro-
cessors, the computation demands of such methods
as mass spectrometry and Fourier transform infrared
spectroscopy can be handled easily in lower-cost field
instrumentation. Significant advances have been
made in such instrumentation since the early 1990's
to the point where the continuous measurement of
hazardous air pollutants has become practical.
In the method, infrared radiation typically is directed
through a sample cell as in the simplest of non-
dispersive infrared spectrometers. In contrast to us-
ing an optical filter or a laser to transmit light at a
specific wavelength through the cell, the FTIR source
transmits light over a broad range of wavelengths.
The IR radiation is modulated with an interferometer
before the light enters the sample, that is, the light
energy at each wavelength is varied from zero to
some maximum value by using a mirror that moves
over a small distance, x (in FTIR systems, a laser is
used to monitor the mirror position). This all results
in generating an "interferogram"
-------
merits have been developed to meet this objective.
A competing design goal relates to the distance that
the mirror moves. The greater the distance, the
greater the special resolution of the instrument.
Other differences between instrumentation rest in the
mathematical techniques used to extract the concen-
tration data from the interferogram Fourier transform
generated spectrum.
One of the more important decision-making factors
associated with the FTIR technique is whether it has
been applied successfully at a similar source for mea-
suring similar compounds. With its increasing popu-
larity, increasing numbers of reports are being pub-
fished of successful applications. Examples have
been given for monitoring criteria pollutants at coal-
fired electric utilities (Dunder, 1994), solvent emis-
sions at industrial coating facilities (Ayer, 1996,
Bartak, 1996, Stock, 1996), for monitoring formalde-
hyde, methanol, phenol, and carbonyl sulfide (Kinner,
1995, Geyer, 1996), cyclohexane, ammonia, formal-
dehyde, methanol, carbon monoxide, methane at a
high temperature, high moisture source (Reagen,
1996), hydrochloric, hydrobromic, and hydrofluoric
acids at municipal waste incinerators and HCI at in-
dustrial process plants (Vidrine, 1993). Numerous
other examples are given in the literature.
3.1.3.4 Ion-mobility Spectrometry
Ion-mobility spectrometry is a technique similar to the
time-of-flight technique used in mass spectroscopy,
except that the analyzer operates at ambient pressure
and the ions drift to the detector in an electric field.
Commercial ion-mobility instruments are single com-
ponent instruments and have been optimized for the
measurement of HF, HCI, NH3f hydrogen peroxide
|H202), hydrocyanic acid (HCN), CI2, and chlorine
dioxide (CI02). A wide range of other gases, such as
aldehydes, ketones, amines, polyaromatics, etc., are
said to be measurable by this technique (Bacon,
1993). However, specific instruments for these gases
have not been advertised cornmereially for source
monitoring applications.
In this method, a radioactive source ionizes the
molecules, which then drift through an electric field
to an electrometer (Figure 3-14). The mobility (drift
velocity/electric field strength) of the ions through
the field is dependent upon the charge, mass, and
shape of the molecule. An electronic shutter grid is
used as a "gate" to periodically allow ions to enter
the drift space. Smaller ions have a greater drift
velocity than the larger ions and reach the
electrometer earlier. The resultant instrument signal,
showing the different ion signals as a function of
time, appears much like a chromatogram and could
be called an "ionogram." Concentrations are
determined from the peak heights.
60 40 30 20 10 0
Time - milliseconds
Figure 3-14. Ion-mobility spectrometer.
42
-------
The ion mobility spectrometer does have some prob-
lems in discriminating between different compounds.
These interference problems are resolved by several
stratagems: 1) introducing chemicals ("dopants")
into the gas stream to inhibit the ionization of in-
terfering compounds, 2) using a permeation mem-
brane to exclude or retard interfering compounds, 3}
changing the electric field polarity to select positive
or negative ions at the detector, or 4) scrubbing out
interfering compounds before they enter the spec-
trometer. The research necessary to optimize the
technique for specific compounds has limited the
application of the method.
A related method to ion-mobility spectrometry is field-
ion spectrometry. In this method, molecules are ion-
ized by a radioactive source or are photoionized using
a UV lamp. The method uses an oscillating electric
field to neutralize interfering ions and allow ions of
interest to pass through the drift tube to the elec-
trometer. Field-ion spectrometry is said to be able to
measure a wide range of organic compounds at part
per billion levels; however, the method has been in-
troduced only recently and is just becoming commer-
cialized (MSA, 1996).
3.2 Monitoring Systems for Particulate
Matter
3.2.1 Sampling Problems for Particulate Matter
Continuous monitoring of paniculate matter emis-
sions presents special sampling problems that are not
encountered in the measurement of gaseous pollut-
ants. These include resolving problems associated
with 1) particulate stratification, 2) wet gas streams,
and 3) particle deposition in extractive sampling sys-
tems.
3.2.1.1 Particulate Stratification
Particulate matter stratification across a stack or duct
cross-section, including both variation in mass con-
centration and variation in particle size distribution,
occurs in many instances and must be considered in
the development of a particulate monitoring program.
The extent of these effects are dependent upon the
source application and the type of monitoring equip-
ment that is used to sense the particulate concentra-
tion. In addition, effluent flow rate stratification is
also likely to occur at most sampling locations and
this may complicate the measurement of particulate
matter mass emission rates.
Particulate matter stratification is due to the influence
of inertial and viscous forces acting on the particles
as they move with the gases through the effluent
pathway. The significance of these forces is depend-
ent on the flow stream characteristics and the spe-
cific bends, turns, flow obstructions, and length of
the effluent ducts between flow obstructions. In the
simplest terms, the particles within the effluent
stream have much greater mass than the gas
molecules and are thus subject to much larger inertial
forces where changes in direction of the flow stream
occur. Obviously, the extent of these forces depend
on the size and density of the particles.
Where inertial forces cause particles to move in a
direction different than the gas velocity, viscous
forces are also exerted on the particles. These forces
are dependent on the relative velocity between the
particle and the gas stream and the aerodynamic size
and mass of the particles. As a rough rule of thumb
for industrial emission air flows, particles having a
diameter smaller than 1 fjm can be assumed to
behave much like a gas and particles having a
diameter 10 fjm or larger are expected to exhibit
substantial inertial behavior. Stratification across the
flow stream is expected for 1 to 10 fjm particles.
The particle size distribution depends on the type of
industrial process and the type and efficiency of
control equipment that is installed.
Taken together, the above factors create particulate
matter stratification at the majority of monitoring lo-
cations. To estimate or forecast the specific impact
of these factors is infeasible based on theoretical
models. To characterize the stratification profile
would require obtaining particulate concentration data
at many points in the duct. This would be very diffi-
cult and cost prohibitive. Furthermore, stratification
profiles are likely to change 1) with varying flow rates
corresponding to process rate changes, and 2) over
time as a result of variations in control equipment
performance, process operation, and fuel or raw
materials. Decisions regarding monitor location can
not be based on exact knowledge of the particulate
matter stratification profile but instead must be based
on an understanding of basic principles and the use
of limited measurement data to make an informed
choice. Ultimately, the acceptability of a particulate
matter monitoring location is determined when the
monitoring measurements are correlated with the
results of manual gravimetric measurements. A non-
representative monitoring location will not correlate
well with manual test results that are obtained by
traversing the entire duct cross section (see Chapter
7).
The effects of particulate matter stratification on duct
concentration measurements cannot be completely
eliminated. However, the effects can be minimized
by selection of an appropriate measurement tech-
43
-------
nique and selection of a measurement location and
measurement point or path that will maximize the
opportunity to acquire representative samples.
For sampling locations downstream of high efficiency
control devices, very low particulate concentrations
are expected except during malfunctions or certain
periods of start-up or shut down when the control
device must be bypassed. In most cases where high
efficiency control devices are used, or where a strin-
gent particulate matter standard must be met, control
of large particles is reasonably assured and the re-
maining particle size distribution contains mostly
small particles. In such cases, the effects of stratifi-
cation are minimized since the small particles behave
much like gas molecules and a representative sample
can be obtained following simple principles described
below.
Where higher particulate matter concentrations and
larger particles are encountered, greater attention to
selecting the measurement location and measurement
points is warranted. Selection of a measurement lo-
cation in a long straight run of duct or as far from
flow disturbances as possible is a good first step.
Pitot tube traverses should be conducted to construct
velocity profiles at various operating loads or process
rates. If no basis for estimating the particle size dis-
tribution exists, measurements should be made at
several points across the duct cross-section using
cascade impactors or other in-situ particle sizing de-
vices. The resulting particle size information may be
useful both in selecting 1) the type of monitoring
equipment and 2) the measurement point or path for
the monitoring device.
Another approach to selecting a particulate sampling
location relies on the use of the monitoring device.
"Portable" transmissometers with slotted probes
have been manufactured and can be used (within the
physical limits of the probe length, duct wall thick-
ness, and duct dimensions) to perform measurement
at a number of locations across the duct. Some in-
situ light scattering devices can vary the area or
volume in which particulate matter is sensed by
changing the angle between the light source and the
detector. For single point extractive systems or point
monitors, varying the insertion depth of the probe or
making measurements in several sampling ports may
be possible. When attempting to use any monitoring
device to detect stratification, the effects of temporal
variations that occur during the experiment must be
minimized. Ideally, two instruments should be used,
one remaining at a fixed reference point and the other
moving to various traverse points. Where the use of
two instruments is impractical, care must be taken to
maintain steady source operating conditions and
sampling must be performed for a sufficient period at
each point to characterize the normal fluctuations in
emissions over time at each process rate of interest.
Good practice in all cases would generally require
that point monitoring systems be located within the
central portion of the flow stream (i.e., away from
duct walls to avoid boundary layer effects), and at a
point of average velocity in the flow stream with the
flow direction parallel to the duct or stack walls.
Cross-stack optical path measurement devices should
be oriented so that the beam passes through the
stratification gradient rather than parallel to the gradi-
ent. Such an orientation is much less susceptible to
non-representative sampling due to the influence of
stratification. Orienting the monitor path so that the
light beam is in the plane defined by an upstream
bend is an example of this approach. Also, locating
a cross-stack monitor so that the light beam passes
through the central area of the stack or duct mini-
mizes the effects of boundary layers and eddy flows
near the walls. Requirements for locating opacity
monitors and examples for commonly encountered
duct configurations are found in Part 60, Appendix B,
Performance Specification 1 (PS1) {USEPA, 1996)
which represent a consistent approach based on "en-
gineering judgment." These principles and the
requirements of PS1 generally should be followed for
path sampling devices in the absence of other appli-
cable criteria !or information) for a particular applica-
tion. However, the PS1 location criteria should be
considered as only a guide. Actual measurement
data or other specific information should be given
greater credence.
3.2.1.2 Particulate Monitoring In Wet Stacks
Particulate monitoring downstream of wet scrubbers
or process streams where water droplets are present
at the monitoring location limits the selection of mon-
itoring equipment and may require that additional
steps be taken. Some particulate monitoring devices
can not be used where liquid water droplets are pres-
ent. For example, charge-contact (triboftow) devices
cannot be used. Transmissometers or other optical
devices can not distinguish between water droplets
and other particles and therefore can not be used
without additional modifications to the source or ef-
fluent pathway. In the United States, new source
performance standards {40 CFR 60.13(0(1) (USEPA,
1996b) and state regulations provide for opacity
monitoring exemptions where liquid water are present
downstream of control devices because of this inter-
ference. However, in Germany, France and other
European countries, particulate monitoring in wet
stacks is required (Peeler, 1996).
44
-------
Several options are available for participate monitor-
ing in wet stacks. Single point extractive systems
with heated probes are available from several manu-
facturers. An extractive light scattering device has
been developed with sufficient heating capability to
vaporize liquid droplets before they reach the heated
measurement cell. This device has been evaluated
and tested in Germany (VDI, 1989), Extractive beta
gauge systems that also vaporize liquid droplets have
been used widely in France for wet gas monitoring.
Alternative approaches that convey a slipstream of
the effluent through a heated device to vaporize liq-
uid and which then employ either transmissometers
or light scattering analyzers have also been used in
Germany and other countries. A heated bypass sys-
tem has been used with transmissometers for particu-
late monitoring on refuse incinerators and power
plants in Germany. These systems have been
evaluated and approved by TUV Rheinland (TUV
Rheinland, 1985).
A similar approach has been developed for use with
light scattering instruments normally used as in-situ
devices. In this case, the slipstream is extracted
from the stack, heated to vaporize droplets in an
electrically heated cyclone, and then passed through
a small heated duct with an installed light scattering
analyzer to facilitate the measurement. This system
has been used in Germany for several years and has
been shown to perform successfully in the United
States at a hazardous waste incinerator having a sat-
urated exhaust stream and low particulate matter
concentrations (Joklik, 1995).
The above solutions to monitoring particulate matter
in wet stacks still require that comparisons to manual
gravimetric tests be performed to correlate the output
of the instrument to mass concentration units. Spe-
cial care is required in performing the manual test
methods because most in-stack particulate filtration
methods (e.g., Part 60, Method 17) (USEPA, 1996b)
can not be used in the presence of water droplets
without special precautions and heating to ensure
that the filters are not exposed to droplets. When
out-of stack filtration methods are used (e.g., Part
60, Method 5) the temperature of the filter should be
maintained only slightly above the sample stream
dew point.
3.2.1.3 Probe/Sample Line Deposition Problems
Deposition of particulate matter in the sample probe
and sample lines is a concern for extractive particu-
late monitoring devices because particulate matter
that is deposited in the probe or sample lines repre-
sents a low bias in the measurement results. As with
stratification, large particles are the most likely to be
affected by probe deposition.
The primary method used to minimize probe and sam-
ple line deposition is to maintain high transport
velocities through the tubing. Because isokinetic
sampling must be performed at the point of sample
extraction, the transport velocities are somewhat
limited. The internal tubing diameters may be
minimized to increase the transport velocity provided
that the vacuum is not too high to be overcome by
the pump or aspirator and thus create non-isokinetic
sampling conditions. Another method of minimizing
deposition is to keep the probe and sample lines as
short as possible by using a close-coupled system.
A supplier of beta gauge devices has included a probe
closure valve at the sample nozzle to protect the
measurement system by excluding effluent gases and
particulate matter when the monitor is not in the
sampling mode and to minimize the effects of
particulate deposition in the sample probe. The
measurement system is operated for a discreet period
to obtain a suitable amount of material on the paper
tape. At the end of each of these batch sampling
periods, the probe closure valve is shut briefly,
creating a vacuum in the sample probe. Then it is
opened quickly resulting in a pressure pulse of gas
traveling through the probe to re-entrain particles that
may have been deposited in the probe and transport
lines.
The extent of particulate deposition problems for a
specific system can be determined by periodically
cleaning the probe and sample lines upstream of the
analyzer. If a significant amount of material is found,
then the maintenance interval should be shortened
and greater cause for concern is warranted.
3.2.2 Continuous Particulate Monitoring
3.2.2.1 Perspectives for Continuous Particulate
Mass Monitoring
No U.S. Federal requirements exist for the continuous
measurement of particulate mass, although one is
being proposed for sources that incinerate hazardous
wastes (EPA, 1996). A few states have required
continuous mass measurement systems through
operating permits or through negotiated agreements;
however, these applications have not been extensive.
Continuous mass measurement requirements are
common in Europe, particularly in the Federal
Republic of Germany where both the regulatory and
technical sophistication of continuous mass
measurement has become quite advanced (Peeler,
1996).
45
-------
missometer measures the ability of a flue gas to
transmit light. A light scattering instrument measures
the light intensity scattered by the flue gas particles. A
beta gauge measures the transmission of electrons
through a spot of collected paniculate matter. All of
these commercially available continuous mass
monitoring instruments produce an instrument output
that is something other than "grams per cubic meter"
(or Ibs/ft3). This output, however, can be correlated to
the particulate concentration, To continuously measure
particulate mass, one first chooses an instrument that
measures some property of the particles in the flue gas.
The instrument readings are correlated with manual
particulate source test method data from a manual
reference method. Source and control equipment
operating conditions are varied to obtain a range of
particulate concentrations. A graph, or other
correlation, is then made between the instrument
response and the manually determined particulate
concentrations.
This correlation results in an "analytic function" that
relates the two techniques. The method defines how a
statistical correlation is to be made and defines the
acceptance criteria for the correlation. The principal
concern in obtaining a valid instrument-manual method
correlation is to make sure that the procedures are
conducted in a representative manner.
1. Comparative measurements should be made at
several source operating conditions to obtain a
data spread suitable for establishing the
correlation.
2. The automated system should measure a sample
that is representative of emissions to the
atmosphere.
3. The manual sampling method should extract a
sample representative of that measured by the
instrument system.
4. Measurements should be representative in time.
Instrumental and manual measurements should
be concurrent. Source operating conditions
should remain stable during these measurement
periods.
The correlation technique is valid only so long as the
conditions under which a correlation was developed are
representative of the source operation. Changes In
operation that lead to significant changes in particle
characteristics or the particle size distribution may
affect the slope of the correlation line greatly. Changes
in fuel, changes in control equipment, or changes in
process operation may contribute to this problem. A
new correlation should be developed in such situa-
tions.
Standards for continuous mass monitoring systems
are used today in the Federal Republic of Germany
(VDI, 1980 and FRG, 1992). A more general set of
standards has also been prepared by the International
Standards Organization (ISO) (ISO, 1995). A variation
of the ISO 10155 method has been proposed as
Performance Specification 11A for continuous particu-
late monitoring in the proposed hazardous waste
combustor rule (USEPA, 1996). These methods are
discussed further in Chapter 7 of this manual.
3.2.2.2 Measurement Techniques
Measurement techniques used in continuous par-
ticulate monitor systems are given in Table 3-1.
Of these methods, the light attenuation technique
using transmissometers has been the most extensively
studied. Extinction-mass correlation methods are
used routinely in Germany and occasionally in the
U.S. The light scattering and beta gauge techniques
are being applied increasingly due to the good correla-
tions that can be obtained.
3.2.2.2.1 Optical - Light A ttenuation (Transmissome-
ters}. In a transmissometer, the light attenuation or
transmittance through the flue gas is determined by
passing a light beam across the stack interior. The
intensity of the light returning, I, is compared with a
previously determined reference signal, I0, to give the
transmittance, T = 1/I0.
A transmissometer may be constructed in two ways,
using either a single pass system or a double pass
system. In a single-pass system, the light crosses the
stack directly to a detector. In a double-pass system
(Figure 3-6), the light crosses the stack twice. The
transceiver assembly on the left houses both the light
source and light detector. By reflecting the projected
light from a mirror located outside of the transceiver
window, systems can be designed easily to check all
of the electronic circuitry, including the lamp and
photodetector, as part of the operating procedure.
Most transmissometer systems include some type of
air purging system or blower to keep the optical
windows clean.
The way in which a transmissometer is used can
affect its design. If the transmissometer data are to
be correlated with particuiate mass, red or infrared
light may be more appropriate than using visible light
as in opacity monitors. The smaller particles {<5 /^m
in diameter) contribute greatly to the opacity but not
46
-------
Table 3-1. Automated Measurement Methods for Particulate Matter
Physical Basis
Optical
Nuclear
Electrical
Electromechanical (Loaded Oscillator)
Technique
Light Attentuation (Transmissiometers)
Light Scattering
Beta Ray Attenuation
Contact Charge Transfer
Piezoelectricity
Cantiievered Beam
to the paniculate mass loading of the flue gas. Red
light is not as sensitive to the small particles as it is
to the larger particles, and thus gives a better correla-
tion to particulate mass (Uthe, 1980). As discussed
above, this correlation is done with respect to manual
methods such as EPA Method 5 or 17 test data.
Since the light transmittance, T, is reduced exponen-
tially by an increase in mass concentration, a semilog
plot of light transmittance versus concentration
should yield a straight line, or linear correlation.
Another method of developing the correlation is to
first convert the transmittance data to optical den-
sity, where optical density, D, is defined as:
D = log(1/T)
Equation 3-1
Another expression that is used frequently for such
correlations is "extinction," a parameter that normal-
izes the path length:
b = 2.303D/I
Equation 3-2
where b
D
I
= extinction
= optical density
= light path length
A graphical plot of either optical density or extinction
against the manually determined particulate concen-
tration should give a straight line correlation. The
measurement sensitivity of this technique is approxi-
mately 10 mg/m3 for a one meter optical path.
Extinction-mass correlations have been developed
successfully for many types of emission sources.
However, as implied above, correlations may be
sensitive to changes in the particle-size distribution in
the flue gas. In a practical sense, transmissometers
used to provide mass measurements in Germany
with retesting every 3 to 5 years have been found to
maintain the original correlation (Peeler, 1996).
3.2.2.2.2 Optical - Light Scattering. When light is
directed toward a particle, the particle may both
absorb and scatter the light. If the wavelength of the
light is large with respect to the size of the particle,
a type of scattering called "Rayleigh" scattering
occurs. If the wavelength of the light is approxi-
mately the same as the radius of the particle, a type
of scattering called "Mie" scattering will occur
(originally described by Gustav Mie in 1908). This
form of scattering is shown in Figure 3-15.
Figure 3-15. Angular dependence of the intensity of
light scattered by a spherical particle with index of
refraction 1.20. The intensity is arbitrarily normal-
ized in each case, (Source: Ashley, L.E., 1958)
Note from the figure that for values of r/X less than
0.5 (where r is the particle radius and X the wave-
length of the light), the particle will scatter the light
in many directions - forward, backwards, up, down,
etc. For values of r/A > 1.0, the scattering will occur
principally in the forward direction.
Baghouses and electrostatic precipitators used to
control the emission of particulate matter will collect
particles that are greater than 1 /^m (1000 nm) in
diameter effectively. However, collecting particles
in the submicron range «1 Mm) is more difficult.
These are the particles that will have a higher proba-
bility of escaping into the atmosphere. Visible light
47
-------
(range 400 nm to 700 nnn) scattering from these
particles is, therefore, within the region of applicabil-
ity of Mie theory for visible and infrared light.
Analyzers have been developed to take advantage of
scattering effects. They can be designed to measure
either back-scattered light, forward scattered light, or
light scattered toward the side, at a specified angle,
A side-scattering instrument is illustrated in Figure 3-
16.
. Jight trap
.*>'"•'
Figure 3-16. A Side-scattering continuous mass
emission monitor.
In this side-scattering device, infrared light is focused
on a sample volume. Instead of measuring the back-
scattered radiation, the device locates a sensor above
the lamp such that side-scattered light is detected.
A reference measurement is made by monitoring the
lamp intensity through a tube passing from the lamp
to the detector,
3.2.2.2,3 Nuclear - Beta Ray Attenuation. When
beta rays pass through a material, they can be
absorbed or reflected by that material. The transmis-
sion of the beta rays is therefore attenuated and the
reduction in beam intensity can be correlated to the
amount of material present. By using a radioisotope
for the beta source (e.g. Kr8S, C14)> "beta gauges"
have been developed that can monitor particulate
mass continuously (Figure 3-17) (Nader, 1975). In
this device, the flue gas is drawn isokinetically
through a probe. The sample may then be diluted to
reduce the dew point to levels where condensation of
flue gas moisture will not occur in the instrument.
The gas is filtered through a glass fiber filter to
produce a spot of collected particulate matter, which
is moved between the beta source and detector for
a determination of the beta ray attenuation. In
practice, a moving filter tape allows the intermittent
collection and measurement of one data point to
produce a semi-continuous measurement.
Isokinetic
sampling probe
Figure 3-17. Typical beta gauge paper tape monitor.
The reduction of the beta ray beam intensity through
the spot depends upon the electron density of the
collected material and the amount of material pres-
ent. To produce consistent measurements, a con-
stant relationship must exist between the number of
electrons per molecule and the molecular weight.
This ratio is essentially the same for most particulate
matter found in coal and oil combustion sources. In
this method, the sample gas volume is controlled to
provide a value for the particulate matter concentra-
tion. The range of the instrument is typically from 2
to 4,000 mg/m3.
The method does require that the sample be collected
isokinetically. Problems may occur with particulate
deposition in the sample probe and sampling lines.
As discussed earlier, strategies have been devised to
minimize such deposition. Spot collection efficiency,
particle composition, and gas volumes and dilution
ratios are all factors that may produce error.
These problems may be minimized in some applica-
tions by first diluting the sample, using high transport
velocities, or pulsating flow (Farthing, 1996).
48
-------
3.2,2.2.4 Electrical - Contact Charge Transfer. When
two dissimilar materials make contact, a net transfer of
electrons from one material to the other can occur.
This is not an effect based on the accumulation or
transfer of static charges, but an effect based upon the
intrinsic electronic properties of the materials them-
selves. The amount of charge transferred depends on
the particle's work function, resistivity, dielectric
constant, and the physical conditions of contact (parti-
cle deformation, duration of contact, area of contact,
etc.) (Wang, 1988). The operating mechanism has
been advertised as the "tribo-electric effect," a term
which is not commonly found in the scientific literature.
This has tended to confuse the evaluation of the
technique.
The instrument is simple, consisting of a metal surface
probe inserted into the stack. It has been qualitatively
successful as a bag-house particulate alarm monitor.
The instrument, however, lacks a method of probe
calibration and has shown problems for monitoring after
electrostatic precipitators because of static electrical
charges on the particles. Also, small particles may
follow the gas streamlines around the probe and never
make contact for the measurement.
3.2.2.2.5 Electromechanical - Piezoelectricity, Loaded
Beam, Electromechanical devices have been developed
on the principle that the frequency of a vibrating
oscillator will change if the mass of the vibrating
element changes. A piezoelectric crystal, a cantilevered
beam, or oscillating metal band may be used to provide
the mechanical vibration. When particulate matter
comes into contact with the vibrating element, it
adheres to it and changes its total mass, and conse-
quently, its vibration frequency. This mass-dependent
vibration frequency is then measured as the correlation
parameter.
When applied to flue gas measurements, the sample
must be withdrawn isokinetically from the flue and it
must be diluted to avoid condensation of the flue gas
moisture. When the particles don't adhere to the
vibrating element, the data are not representative.
When the particles do adhere, the vibrating element
eventually will become overloaded and it must be
cleaned and recalibrated. Although some interest in this
method has been shown, it has not become commer-
cially available technology.
3.2.2.2,6 Other Methods. Several newer methods for
continuous particulate monitoring have appeared on the
commercial market. !n what is termed an "acoustic
energy" technique, particles impacting on a probe
produce acoustic waves in the probe transducer. The
oscillations produced are used to count the impact of
single particles on the probe and thus provide a relation-
ship to the flue gas particulate matter concentration.
In another technique, variation on transmissometry
utilizes fluctuations in the light transmission due to
the flue gas particles passing through the light beam.
This generates an instrument response that is directly
proportional to the particulate concentration.
3.2.2.3 Choosing a Continuous Particulate
Monitoring System
Numerous technical issues are involved in choosing a
continuous particulate monitoring system. First, the
flue gas stream must be well characterized to deter-
mine the presence or absence of water droplets,
particulate concentration levels, and the degree of
particulate stratification. A selection process that
considers these issues is illustrated in Figure 3-18.
3.3 Monitoring for Metals
3.3,1 Sampling Problems for Metals
Either extractive or in-situ sampling methods can be
used for monitoring metals. However, measurement
problems develop when a metal is present in the flue
gas in both the vapor and solid phases. Mercury is a
typical example, where most of the total mercury is
present as a vapor, although some may be bound in
the form of mercuric chloride or other compounds in
the particulate matter. Simple optical instrumentation
can detect the mercury vapor, but not the other
mercury compounds. To obtain a value for total
mercury, either an extractive technique must be em-
ployed to reduce these compounds for subsequent
measurement, or the particles and vapor alike can be
heated to extremely high temperatures
-------
Limit choices to:
Transmissometer with slip steam
Extractive light scattering
Beta guage
Are
water
droplets/condense
moisture present
at monitoring
location?
is
there also
an opacity
monitoring
require -
ment?
Dust
concentratio
above \ YES
Qmg/m3/meter?
Dust
concentration
above \ YES
Qmg/nf/meter?
Limit choices to:
Extractive light scattering, or
Beta giiage
if dust loading fs below 30mg/ms
Select appropriate instrumentation
based upon dust loading, particle
size distribution, and competitive
commercial factors.
Evaluate commercial
factors and other
selection criteria
Select
Extractive Scattering
,OQ5-20mgM
.05-10 jim
Select Transmissometer
>30 mg/m* for
1 meter path
.5-25 urn
Figure 3-18. Selection process for particulate monitors.
In most mercury monitoring methods, mercury vapor
is analyzed by light in the ultraviolet region of the
spectrum. The simplest method is to merely project
a UV light into the stack as in an in-situ instrument
and monitor the absorption of light by the elemental
mercury vapor. However, if mercury is present in
other molecular forms, such as mercuric chloride
(HgCI2), total mercury will not be measured. Al-
though in many incineration facilities elemental
mercury comprises greater than 90% of the total
mercury emitted, interest remains to measure total
mercury. As a result, various stratagems have been
devised to reduce the mercury compounds to elemen-
tal mercury, vaporize the elemental mercury, and
measure the vapor through UV light absorption
techniques.
In one method, an isokinetically extracted continuous
sample is heated in an infrared oven to volatilize
particulate bound mercury and a sodium hydroboron
50
-------
solution is used to reduce all mercury compounds to
elemental mercury. Elemental mercury then is
measured using a UV photometer. In another method,
mercury compounds are reduced chemically using
stannous chloride and the elemental mercury is
amalgamated with gold, which then is heated to
release the mercury as vapor to be measured using a
UV photometer.
In a new microsensor technology, mercury vapor
adsorbs on the surface of a thin noble-metal film.
The electrical properties of the film change quantita-
tively to give a measurement of the amount of
mercury present (Glaunsinger, 1995).
Although mercury monitoring methods are commer-
cially available, to measure total mercury adsorbed or
bound on paniculate matter, a continuous isokinetic
sample must be obtained and the sample reduced to
elemental mercury. However, chemical systems and
ovens used in the various system designs require
frequent periodic maintenance. These maintenance
demands are typically greater than for traditional
criteria pollutant monitoring systems.
3.3.3 Multi-Metal Methods
Title III of 1990 Clean Air Act Amendments has listed
ten other metals, in addition to mercury, that are to
be regulated as hazardous air pollutants (air toxics).
These metals are: antimony (Sb), arsenic (As),
beryllium (Be), cadmium (Cd), chromium (Cr), cobalt
(Co), lead (Pb), manganese (Mn), nickel (Mi), and
selenium (Se). Whether these metals will be required
to be monitored individually and continuously is
unclear. Nevertheless, considerable interest exists in
the development of instrumental methods for metals.
(See, for example, Durham, 1995). As in the initial
development of gas monitors, multi-metal methods
are essentially laboratory techniques adapted for field
use. The techniques of x-ray fluorescence spectros-
copy and atomic emission spectroscopy have seen
the most application. The atomic emission spectro-
scopic techniques differ by the manner in which the
metal atoms are excited - either through the use of a
laser, a radio-frequency plasma, or a microwave
generated plasma.
3.3.3,1 X-Ray Fluorescence Spectroscopy
X-ray fluorescence spectrometry is being applied in a
manneY similar to that of the beta gauge used to
continuously monitor particulate matter. To analyze
for metals, particulate bound and gaseous metals are
collected on an activated carbon impregnated filter.
The collected material is then exposed to x-rays,
which excite the atoms to higher electronic levels.
As the excited atoms de-excite, they fluoresce,
emitting light at wavelengths specific for each metal.
The amount of emitted light is measured and the flue
gas metal concentration calculated.
X-ray fluorescence spectroscopy is a well-established
technology; however, the eommereiaS continuous
monitoring systems applying the technique have not
been developed fully. Current systems involve
collecting and analyzing the filter samples in a batch
mode, not a continuous mode.
3.3.3.2 Atomic Emission Spectroscopic Systems
Atoms can be excited in many ways and when
excited sufficiently, will emit light energy. Two
excitation methods under development are that of
inductively couple plasma and laser spark spectrome-
try.
3.3.3.2.1 Inductively Coupled Plasma Spectrometry,
In this technique, a radio frequency generated plasma
is used to heat the flue gas sample to temperatures
greater than 10,000°C (Figure 3 -19).
The light emitted by the metal atoms excited at this
extreme temperature is measured using a diffraction
grating and photodetectors as shown in the figure.
A system is under development (Seltzer, 1994) that
extracts a sample continuously and isokinetically from
the stack, which is then subjected to the heat of a
plasma torch.
Plasma Torch
Photomultiplier
Ql
Diffraction
Grating
Figure 3-19. Inductively coupled plasma technique,
3.3.3.2.2 Laser Spark Spectrometry. In laser spark
spectrometry, a high energy laser is used to excite
metal atoms in the stack (Figure 3-20). Here, the
laser provides energy to excite the metal atoms, both
in the gaseous phase and those adhering to the
surface of the particulate matter. The light emitted
by the excited metal atoms is again collected and
analyzed.
Validating multi-metal monitoring methods has proven
difficult. Uncertainties associated with the manual
51
-------
wet chemical test methods (e.g., Method 29) have
also led to uncertainties in assessing the performance
of the instrument methods. As discussed above, the
partition of metals between gaseous and solid phases
introduces additional difficulties since problems of
both gaseous and particulate stratification and both
gas and particle sampling losses need to be over-
come. In addition to these sampling issues, conve-
nient calibration methods and traceable standards
have not been developed. At present, data quality
can be estimated only through extraordinary research
efforts. These challenges are formidable and will
take several years of research and development to
overcome.
-._. -Partible! '
a'nfl gases j
"~*
LASER
SPECTROMETER
Figure 3-20. Laser spark spectrometry.
References
Ashley, I.E. and Cobb, C.M. 1958, Single Particle
Scattering Functions for Latex Spheres in Water, J,
Optical Soc. of America. 48:261-268.
Ayer, J. and Williams, G.M. 1996. "FTIR Monitoring
for Real Time Organic Emissions Measurement and
Process Control at Military or Industrial coating
Facilities and Other Solvent Operations," Paper
presented at the Air & Waste Management Associa-
tion Meeting. Nashville, TN. Paper No. 96-WA65.01.
Bacon, T. 1993. Ion Mobility Spectrometry Tackles
Continuous Emission Monitoring Problems. Pollution
Equipment News. Vol. 26. No. 7. pp 4-5.
Bacon, A.T, and Reategui, J. 1993. Ion Mobility
Spectroscopy Applications for Continuous Emission
Monitoring, in Continuous Emission Monitoring - A
Technology for the 90s. Air & Waste Management
Association. Pittsburgh, PA. pp 288-300.
Bacon, T. and Webber. 1996. "Acid and Halogen
Gas Monitoring Utilizing In Mobility Spectroscopy
(IMS)." Paper presented at the Air & Waste Manage-
ment Association Meeting. Nashville, TN. Paper No,
96-TA30.02
Bartak, D.E., Wiley, R.M., Crome, C, and Rover, M.
1996. "Analysis of Volatile Organic Compounds
(VOCs! by Extractive FTIR Spectroscopy." Paper
presented at the Air & Waste Management Associa-
tion Meeting. Nashville, TN. Paper 96-RP65.Q6,
Bartman, C.D., Renfroe, J., Robards, H., and Con-
nolly, E. 1990, "A Mass Spectrometer-Based
Continuous Emissions Monitoring System for Hazard-
ous Material Stack Gas Measurements," Paper
presented at the International Joint Power Generation
Conference, 1990.
Biermann, H.S., and Winer, A.M. 1990. Recent
improvements in the design and operation of a
differential optical absorption spectrometer for in-situ
52
-------
measurement of gaseous air pollutants. Paper
presented at the Air Pollution Control Association
Meeting, Pittsburgh, PA. Paper No. 90-87.2.
Campbell, K. R., Hallett, D.J., and Resch, R.J. 1991.
Assessment of an on-Line CI-Mass Spectrometer as
a Continuous Emission Monitor for Sewage Sludge
Incinerators, in Municipal Waste Combustion. Air &
Waste Management Association. Pittsburgh, PA. pp
880-884.
Coleman, W.M., Dominguez, L. M., Gordon, R.J.
1996. A Gas Chromatographic Continuous Emissions
Monitoring System for the Determination of VOCs
and HAPs. J. of the Air & Waste Management
Association. Vol 46, No.1, pp 30-34.
Cone, L., Logan, T., and Rollins, R. 1990. "Carbon
Monoxide and Total Hydrocarbon Continuous Moni-
toring at Hazardous Waste Incineration Facilities" in
Continuous Emission Monitoring: Present and Future
Applications, Air & Waste Management Association.
pp. 338-349.
Dominguez, L.M., Gordon, B.M., and Coleman, W.M.
1995. "A Commercial Gas-Chromatographic Continu-
ous Emissions Monitoring System for The Determina-
tion of VOCs and HAPs." Paper presented at the Air
& Waste Management Association Meeting, San
Antonio, TX. Paper 95-TA16B.06.
Dunder, T.A. 1994. "Evaluation of Fourier Transform
Infrared {FTIR} Technology for Continuous Emissions
Monitoring." Paper presented at the Air & Waste
Management Association Meeting, Cincinnati, OH.
Paper No. 94-TP29B.02.
Dunder, T.A., Stone, C.D. 1995. "Permeation and
Adsorption of Carbon Monoxide in Polymeric Tubing
Used for Extractive Emissions Monitoring Applica-
tions." Paper Presented at the Air & Waste Manage-
ment Association Meeting, San Antonio, TX. Paper
No. 95-TA-16B.02.
Durham, M.D, and French, N.B. 1995. Status of
Emerging Technologies for the Continuous Monitoring
of Metal Emissions from Waste Combustion, in Solid
Waste Management: Thermal Treatment & Waste-to-
Energy Technologies. Air & Waste Management
Association, Pittsburgh, PA.
Eldridge, J.S., Stock, J.W., Reagen, W.K., and
Osborne, J.M. 1995. "Extractive FTIR: Manufactur-
ing Process Optimization Study." Paper presented at
the Air & Waste Management Association Meeting,
San Antonio, TX. Paper No. 95-TA32.01.
Farthing, W.E. and Williamson;, A.D. 1996. Charac-
terizations of Continuous Particulate Monitoring
Approaches for Stationary Sources. Continuous
Compliance Monitoring Under the Clean Air Act
Amendments. Air & Waste Management Association.
Pittsburgh, PA. pp 194-207.
Federal Republic of Germany fFRG), Federal Ministry
for the Environment, Natural Conservation, and
Nuclear Safety. 1992. Air Pollution Control Manual
of Continuous Emission Monitoring, Regulations and
Procedures for Emission Measurements, 2nd Revised
Edition.
Frish, M.B. 1996. "The SpectraScan® Family of
Truce Gas Monitors Based on Tunable Diode Laser
Spectroscopy," Paper presented at the Air & Waste
Management Association Meeting, Nashville, TN.
Paper 96-TP26B.05.
Geyer, T. 1996. "Performance Specification and
Evaluation of Fourier Transform Infrared {FTIR!
Continuous Emission Monitors for Measuring Hazard-
ous Air Pollutants." Paper presented at the Air &
Waste Management Association Meeting. Nashville,
TN. Paper 96-WA65.02.
Giddings, J.D. 1965. Dynamics of Chromatography.
Marcel Dekker. New York, NY,
Giel, T.V. and Douglas, J.R. 1995. "Measurement of
Particle Volume Concentrations with a New CEM for
Particulate Emissions." Paper presented at the Air &
Waste Management Association Meeting. San
Antonio, TX. Paper No. 95-MP17.03.
Glaunsinger, W. 1995. "New Chemical Micro-
sensors and Implications for Continuous Mercury
Monitoring." Paper presented at the Air & Waste
Management Association Meeting, San Antonio, TX.
Paper 95-MP-21.05.
Gnyp, A.W., Price, S.J.W., St. Pierre, C.C., Smith,
D.S. 1979. Long Term Field Evaluation of Continu-
ous Particulate Monitors, in Proceedings: Advances in
Particle Sampling and Measurement (Ashevi/le, NC,
May 1978), EPA-6QO/79-065, pp 122-168.
Haile, D.M., Dorris, E.H., and Finney, G.L. 1995.
"Field Analysis of Hazardous Air Pollutants with an
Ion Trap Mass Spectrometer." Paper presented at
the Air & Waste Management Association Meeting,
San Antonio, TX. Paper No. 95-TA32.05
Harlow, G., Bartman, C.D., and Renfroe, J.R. 1990.
"Design of a Continuous Emissions Monitoring Sys-
tem at a Manufacturing Facility Recycling Hazardous
53
-------
Waste," Paper presented at the Great Lakes 90
Hazardous Materials Control Research Institute.
Hellwig, G.V., Dunbar, D,, and Richards, J. 1986,
Portable Instruments User's Manual for Monitoring
VOC Sources. EPA-340/1-86-015,
Hewitt, G.F. and Driscoll, J.N. 1981, A New Concept
in Environmental Chromatography. Analytical Instru-
mentation. Vol. 19. pp 5-6.
International Standards Organization (ISO). 1995.
Automated Monitoring of Mass Concentration of
Particles in Stationary Source Emissions: Perfor-
mance Characteristics, Test Procedures and Specifi-
cations. ISO 10155.
Jahnke, J.A. 1984. Transmissometer Systems -
Operation and Maintenance, An Advanced Course.
APTI Course 476A. EPA-450-84-004.
Jahnke, J.A. 1993. Continuous Emission Monitoring,
Van Nostrand Reinhold. NY, NY.
Jahnke, J.A., Maybach, G.B., and Marshall, R.P.
1994a. "Pressure and Temperature Effects in Dilution
Extractive Continuous Emission Monitoring Systems,"
Electric Power Research Institute, Palo Alto, CA.
Jahnke J.A. 1994b. An Operators Guide to Eliminat-
ing Bias in CEM Systems, EPA/400/R-94/016, Wash-
ington, D.C., U.S. Environmental Protection Agency.
Joklik, R.G. 1995. "Technology-Based Approach to
Establishing Maximum Achievable Control Technology
for Hazardous Waste Combustion Devices, Draft
Report on Trip to Visit TUV Rheinland," EPA Con-
tract 68-D2-0164, February 1995.
Kinner, L.L., Geyer, T. J. and Lay, L.T. 1995. "Field
Validation Testing of Fourier Transform Infrared
(FTIR) Spectrometry Method for Measurement of
Formaldehyde, Phenol, and Methanol from Stationary
Sources." Paper presented at the Air & Waste Man-
agement Association Meeting. San Antonio, TX
Paper No. 95-TA32.01.
Kinner, L.L. and Plummer, G.M. 1993. An Investiga-
tion of Process Mass Spectroscopy as a Continuous
Emission Analyzer for Stationary Sources, in Mea-
surement of Toxic and Related Air Pollutants. Air &
Waste Management Association. Pittsburgh, PA. pp
414-422.
Lang, F.D., Shliftshteyn, A. 1991. "Emission Spectral
Radiometer/Fuel Flow Instrument for Determination of
Instantaneous Heat Rate." Proceedings of Heat Rate
Improvement Conference, Scottsdale, AZ., May 7-9,
1991.
Mandel, S.B. and Gottlieb, M.S. 1995. Development
of a Close-coupled Multigas Anaty?er for Continuous
Emissions Monitoring. Paper presented at the Air &
Waste Management Association Meeting, San Anto-
nio, TX. Paper 95-TA16B.04.
McDonald, W.C., Erickson, M.D., Abraham, B.M.,
and Robbat, A. 1994. Developments and Applica-
tions of Field Mass Spectrometers, Environ. Sci.
Technol. Vol. 28 No.77 pp 336A-343A.
Mine Safety Appliances (MSA). 1996.
literature. Pittsburgh, PA.
Commercial
Nader, J.S. 1975. Current Technology for Continuous
Monitoring of Particulate Emissions. J, Air Poll.
Control Assoc. Vol 25, No. 8 pp 814-821.
Pedersen, S.V. 1989. Measurement of Process
Emissions with a New Multigas Monitor.
Peeler, J.W., Jahnke, J.A., and Wisker, S.M. 1996.
Continuous Particulate Monitoring in Germany and
Europe Using Optical Techniques. Continuous Compli-
ance Monitoring Under the Clean Air Act Amend-
ments. Air & Waste Management Association.
Pittsburgh, PA. pp 208-220.
Peeler, J.W., Kinner, L.L., and DeLuca, S. 1996.
General Field Test Method Approval Process and
Specific Application for a Direct Interface GCMS
Source Test Method. Paper presented at the Air &
Waste Management Association Meeting, Nashville,
TN. Paper 96-WA 132.01.
Podlenski, J. Peduto, Mclnnes, R. Abell, F., and
Gronberg, S. 1984. Feasibility Study for Adapting
Present Combustion Source Continuous Monitoring
Systems to Hazardous Waste Incinerators; 1. Adapt-
ability Study and Guidelines Document. EPA 600/8-
84-Olla.
Reagen, W.K. and Wright, B.D. 1996. "Production
Facility Emissions Assessment Using Heated Extrac-
tive FTIR Spectroscopy and EPA Method 25A." Paper
presented at the Air & Waste Management Associa-
tion Meeting. Nashville, TN. Paper 96-WA65.05.
Seinfeld, J.H. 1986. Atmospheric Chemistry and
Physics of Air Pollution, Wiley intsrscience Publica-
tions, New York, NY. pp 374-375.
Seltzer, M.D. and Green, R.B. 1994. Instrumentation
for Continuous Emissions Monitoring of Airborne
54
-------
Metals. Process Control and Quality. No. 6 (1994)
pp 37-46.
Shanklin, S.A., Logan T.J. 1989. "An Evaluation of
Current Instrumentation for Continuous Monitoring of
Hydrogen Chloride Emissions from Waste Incinera-
tors," Proceedings Air & Waste Management Associa-
tion Specialty Conference on Continuous Emission
Monitoring, Pittsburgh, PA,
Sollid, J.E,, Trujiilo, V.L., Limback, S.P., and
Woloshun, K.A. 1996. "Comparison of Photo-
acoustic Radiometry to Gas Chromatography/IVIass
Spectrometry Methods for Monitoring Chlorinated
Hydrocarbons." Paper presented at the Air & Waste
Management Association Meeting, Nashville, TN.
Paper No. 96-WA65.08.
Stock, J,W. and Eldridge, J.S. 1996. "Process and
Emission Evaluation Using Extractive FTIR." Paper
presented at the Air & Waste Management
Association Meeting, Nashville, TN. Paper 96-
WA65.04.
TUV Rheinland. 1985, Institute for Energy
Technology and Protection of the Environment,
Report no.: 936/806001/GM21.
Thijssen, J.H.J, 1995. "Measurement of Organic
HAP Emissions with Laser Induced Fluorescence."
Paper presented at the Air & Waste Management
Association Meeting, San Antonio, TX. Paper 95-
TA32.03.
U.S. Environmental Protection Agency. 1993. "Field
Validation Testing at a Coal Fired Boiler," United
States Environmental Protection Agency Report, EPA
Contract No. 68D20163, WA No. 2. U.S. Environ-
mental Protection Agency. 1996. Revised Standards
for Hazardous Waste Combustors. 61 FR 17358.
U.S. Environmental Protection Agency. 1996b.
Standards of Performance for New Stationary
Sources - 40 CFR 60. Code of Federal Regulations,
Washington, D.C.
Uthe, E.E. 1980. Evaluation of an Infrared Trans-
missometer for Monitoring Particulate Mass
Concentrations of Emissions from Stationary Sources.
J. Air Poll. Control Assoc. 30:382-386.
Uthe, E.E. 1982. Particle Size Evaluations Using
Multiwavelength Extinction Measurements. Applied
Optics, 21:454-459.
Verein Deutscher Ingenieure (VDI). 1980 Staub-
messung in Stromenden Gasen Bestimmung der
Staubbeladung durch Kontinuterliches Messen der
Optischen Transmission. VDI 2066. VDI-Handbook
Reinhaltung der Luft, Band 4. Dusseldorf.
Verein Deutscher Ingenieure (VDI). 1989.
Measurement of Particulate Matter in Flowing Gases;
Determination of Dust Load by Continuous
Measurement of Scattered Light with the Photometer
KTN, VDI 2066 Band 6, Dusseldorf.
Vidrine, D.W. and Mclntosh, B. 1993. Hydrogen
Chloride Monitoring Using an FTIR Based CEM
System, in Continuous Emission Monitoring - A
Technology for the 90s, Air & Waste Management
Association. Pittsburgh, PA. pp 280-287.
Villalobos, R. 1975. Process Gas Chromatography,
Analytical Chemistry. Vol 47. pp 983A-992A.
Wang, H. and John, W. 1988. Dynamic Contact
Charge Transfer Considering Plastic Deformation. J.
Aerosol Set. Vol. 19 No,4, pp 399-411.
Willeke, K. and Baron, P.A. 1993. Aerosol
Measurement - Principles, Techniques and Applica-
tions. Van Nostrand, Reinhold, NY.
55
-------
Chapter 4
Alternatives to Monitoring Instrumentation
Alternatives to the installation of traditional CEM
systems have been, and can be, used to monitor
emissions. These alternatives are often applicable to
sources where independent monitoring instrumenta-
tion may not be required to meet the goals of a
regulatory program. To monitor process and control
equipment performance, certain NSPS sources employ
"parameters," such as pressure drop, temperature, or
water injection rates (see Chapter 2, NSPS Parameter
Monitoring Requirements and Table 2-3), instead of
installing CEM systems. Similar provisions are in-
cluded in other federal and state monitoring require-
ments. Sources regulated under the air toxics or the
"enhanced monitoring" provisions of the 1990 Clean
Air Act Amendments are likely to have the flexibility
to consider such options.
Alternative monitoring options include: 11 using
parameters as indicators of proper operation and
maintenance practices, 2) using parameter values
directly as surrogates for emissions determinations, 3)
using parameters in models that calculate emissions,
4) performing mass balance calculations, or 5) em-
ploying a CEM system to monitor a more easily
analyzed gas as a surrogate for one that is more
difficult to analyze. Deciding whether to use an
alternative method depends upon the application and
how the data will be used. To assist in such deci-
sions, the basic principles of these techniques as well
as their advantages and limitations will be discussed
in this chapter.
4.1 Parameter Monitoring
Parameter monitoring has been used in a variety of
ways in regulatory programs. These are summarized
in Figure 4-1.
Beginning with the original application of the monitor-
ing process and control equipment performance,
parameter monitoring has extended to providing a
basis for the calculation of source emissions. Emis-
sion calculations have been performed since the
original development of air pollution control equipment
in the form of engineering design calculations (for
example, see Mycock, 1995). Design engineers
typically will attempt to determine equipment perfor-
mance from operational parameters. In terms of air
pollution control equipment, the question facing the
engineer is, if I vary this parameter, or if 1 vary that
parameter, will the pollutant emissions increase or
decrease?
Regulatory Uses of Parameter Monitoring
O&MPractidte
Indicators
Theory-Based Models
Empirical Models
Least Squares
Neural Nets
Figure 4-1. Uses of parameter monitoring in regulatory programs.
56
-------
This question also has drawn the attention of
regulatory agencies, and due to relationships be-
tween parameter values and emissions, has led to
the reporting of parameter data to both federal and
state agencies. The original regulatory use of
parameter data and GEM data was to indicate
whether process and control equipment were being
operated properly. A continued record of unsatisfac-
tory parameter data could result in a notice and
finding of violation of operation and maintenance
requirements. A subsequent requirement to perform
a source test for determining compliance with an
emission standard might also result.
A more direct application of parameter data is to use
it as a surrogate for emissions. Rather than requir-
ing the determination of emissions by GEM systems
or a manual method, a surrogate parameter level is
established which is correlated to the emissions
standard. An exceedance of this parameter level
could then be used for enforcement and a source
test may not be necessary (depending upon the
applicable regulation or permit requirements).
An extension of the use of parameter data is to
"predict" the performance of the process or control
equipment from the parameter data. If the process
is well understood, first principle (i.e, theoretical or
phenomenological) calculations may be performed.
Another technique is to "correlate" parameter data
to emissions data. An initial study is performed by
varying and monitoring process and control equip-
ment parameters while monitoring emissions using
reference methods or GEM systems. One can then
correlate the data to develop a statistical model that
can "predict" emissions.
Both theoretical and statistical emission models
based on process and control parameter inputs are
used today to meet emission monitoring require-
ments. A number of states have accepted their use,
and the federal proposals addressing the 1990 Clean
Air Act enhanced monitoring requirements have
been receptive to their application (USEPA, 1993,
Bivins, 1996).
The different uses of parameters, illustrated in Figure
4-1, provide a means for agencies to track control
equipment performance and emissions without the
application of GEM systems or performing source
tests. Although in some cases the data may not be
as accurate as that obtained from independent test
methods, such levels of accuracy may not be
necessary. For stable, nonfluctuating sources
having low emissions relative to the compliance
limit, data that can assure compliance with the
standard may have a higher inaccuracy than that
generally acceptable (e.g., >20% of the reference
method or >10% of the standard), but may still be
defensible. For example, if a stable source has a
VOC emission limit of 50 ppm, but normally emits 5
ppm, even with 100% uncertainty {±5 ppm), one
could still assume that the source is in compliance.
However, the acceptability of such uncertainty
depends on the use of the data. Moreover, this
scenario may not be acceptable if the emission limit
is 10 ppm or the data were to be used in a market
trading program.
4.2 Parameters and Sensors
A parameter is a property whose value can charac-
terize or determine the performance of process or
control equipment. Such properties may be temper-
ature, pressure drop, liquid to gas ratios, percent
oxygen, or even the position of a damper.
The values of parameters are determined by "sen-
sors." In the broadest sense, a sensor is "a device
that receives and responds to a signal or stimulus
{Fraden, 1993]." A thermocouple or resistance
temperature device (RTD) may be used to measure
temperature, a pressure transducer to measure
pressure drop, flow monitors to measure liquid to
gas ratio, an oxygen monitor sensor to measure
percent oxygen, and a simple on-off switch to
monitor the damper position. Sensors normally are
assumed to be some mechanical, electrical, or
chemical device that generates an electrical re-
sponse, but human perception also can serve to
determine parameter values. For example, the
visible emissions observer performing EPA Method 9
is determining a value for the parameter, opacity,
which may characterize the performance of a bag
house or an electrostatic precipitator. Or, in an iron
and steel plant (NSPS subpart AA), each shift
operator may note the furnace static pressure, again
a parameter.
In process and control equipment, sensors tend to
be simple devices that generate an electrical re-
sponse. They do not, in general, incorporate sophis-
ticated linearization and calibration features such as
those used in GEM systems. They are basically the
stripped down version, the sensing elements incor-
porated in GEM systems. For example, the zirco-
nium oxide electrocatalytic cells used in many CEM
oxygen analyzers are the same type of cells used as
"sensors" in automobiles for engine and emissions
control.
57
-------
4.3 Parameter Monitoring Used as an
Indicator of Equipment Operation
and Maintenance
As discussed in Chapter 2, plant operational para-
meters or control equipment parameters have been
specified in the New Source Performance Standards
for certain source categories. In these applications,
the parameter data are intended to determine whe-
ther the process or control equipment is operating
properly. Proper operation, in turn, means that the
plant equipment is operating in a manner such that
the emissions limits (standards) are most likely being
met. In NSPS applications, parameter data may
indicate possible problems in plant performance, but
compliance with emission standards traditionally
has been determined by source testing (or in some
cases, CEM systems) under specified operating
conditions.
The use of the parameter values varies depending
upon the regulation. In the simplest case, a regula-
tion merely may require that the parameter values be
determined, such as in the NSPS requirements for
monitoring the pressure drop across wet scrubbers
used to control particulate emissions in the phos-
phate fertilizer industry, coal preparation plants,
ammonium sulfate manufacture, etc. (CFR Subparts
T, U, V, W, Y, and PP, respectively). No recording
or reporting requirements are given for these exam-
ples, but the required permanent record can be used
by the regulatory agency to assess plant operations
and target problem facilities for follow-up actions.
State programs also have used parameter monitoring
for determining the adequacy of plant operation and
maintenance practices. Many programs include
general provisions in state implementation plans re-
quiring good engineering practice in the operation
and maintenance of control equipment.
4.4 Parameter Monitoring Used as a
Surrogate for Emissions
An extension to requirements for monitoring and
recording operational parameters is to establish
some trigger value for the parameter. Here, a
parameter baseline or trigger value is established
during a source test, where the source test is
conducted to determine compliance with the emis-
sions standard (Ibs/hr, ppm, etc.). This trigger value
then can be used in one of two ways: 1) it can
trigger a reporting requirement, analogous to report-
ing excess emissions by sources with CEM systems,
or 2! it can be used directly as a surrogate for an
emission compliance limit.
4.4.1 Parameter Monitoring as a Surrogate for
Reporting Excess Emissions
Parameter values can be used as surrogates for
emission values to report problems in the operation
and maintenance of emission control equipment, A
typical regulatory statement for parameter monitor-
ing for particulate control equipment is:
"...report to the Administrator, on a semian-
nual basis, all measurements (pressure loss
and water supply pressure) over any 3-hour
period that average more than 10% below the
average levels maintained during the most
recent performance test in which the affected
facility demonstrated compliance with the
mass standards..." (40 CFR 60.143 Subpart
N).
Here, one assumes that the facility is operating at or
below its compliance value and that the parameter
values reflect the operating conditions of the control
device (usually a wet scrubber} at that level. If
parameter levels fall within a range of acceptable
values established during the compliance test, one
assumes that the scrubber is continuing to operate
in a manner where the facility is in compliance with
the underlying emission limit. In this sense, the
parameter value stands in as a surrogate for the
emission compliance value.
Note, however, in the example given, that the
trigger ieve! is not at the parameter leve! determined
during the compliance test, but at a level 10%
lower. In a typical venturi scrubber, an increase in
pressure drop increases the particulate removal
efficiency of the scrubber. A decrease in the pres-
sure drop means that less particuiate matter will be
removed. Since the sensor measurement accuracy
may be ±5% and since some inaccuracies may
have occurred in the source testing, the pressure
drop is allowed to decrease by 10% from the base-
line level before it is required to be reported. For
even greater flexibility, some regulations allow a
reduction of 30% before reporting is required (see
for example, 40 CFR 60, Subpart HH, lime manufac-
turing plants).
Many other examples could be given of this use of
parameters as surrogates for emissions. In Subpart
GG, gas turbines using water injection to control
NOX emissions must report any one hour period
where the water-to-fuel ratio falls below the value
determined to demonstrate compliance. Incinerator
temperatures in Subpart MM, RR, and other coatings
operations are to be reported when they change
from some specified compliance level.
58
-------
If parameter values fall below such trigger levels, the
agency may require the source to improve its opera-
tion and maintenance practices, or may require a
compliance test be performed to determine if the
source is meeting its emissions standards. Also, if
sensors installed to monitor equipment performance
are not operating properly, parameter data will no
longer be useful and a compliance test again may
become necessary,
4.4.2 Parameter Monitoring for Direct
Compliance
A new approach appeared in a 1993 proposal for
enhanced monitoring (EPA, 1993) where parameter
monitoring was proposed as a method for determin-
ing directly the compliance of a source with emis-
sions standards. In this approach, the source owner
or operator would be required to "justify that a
known and consistent relationship exists between
the emissions subject to an applicable limitation or
standard and the parameters being monitored." This
is not much different from using parameters as a
surrogate limit in excess emissions monitoring, as
prescribed in certain NSPS requirements. The
differences are that;
• The parameter level is used as a surrogate
for the emissions standard,
• The source owner would be required to
establish the parameter value or values that
would assure that the source is in compli-
ance with the emissions standard.
That established parameter value has been called the
"demonstrated compliance parameter limit," or
DCPL. Although the proposed enhanced monitoring
rule was withdrawn in April 1995, the concept of
the DCPL, the stand-in or surrogate emissions
standard, still remains.
The question as to how parameter limits actually
relate quantitatively to emission values depends on
the underlying relationship, the accuracy and preci-
sion of the emission measurement method, averag-
ing periods, and other factors. To answer this in
practice, one must obtain further information by
testing at different operational levels. One can
correlate the parameter values with emissions over
a narrow range of operating conditions to establish
the DCPL. A typical performance curve for a venturi
scrubber is given in Figure 4-2. Here, as the pres-
sure drop across the venturi throat increases, the
efficiency in removing paniculate matter increases.
At some point on the curve, a pressure drop will
correspond to the compliance value, the emissions
standard for the facility. Some variability will exist
in the testing procedures used to establish the
correlation and in the accuracy of the sensor used to
determine AP, This range of variability is shown by
the confidence intervals in the figure. The agency
may therefore not accept a 1:1 correlation between
the emissions standard and the correlation, but
establish the surrogate standard, PsW, at a higher
value to account for the variability. If the pressure
drop is maintained at the DCPL value or higher,
Confidence
Intervals
Standard (compliance value)
/
(—-DCPL
AP
Figure 4-2. Operational parameters correlated to emissions. Establishing the demonstrated compliance
parameter limit (DCPL),
59
-------
some assurance exists that the emissions standard
is being met, within the uncertainty of the correla-
tion.
Many examples of using parametric monitoring to
demonstrate compliance with emissions regulations
have been published. Bills (Bills, 1995] describes
applications in the pharmaceutical industry for
thermal oxidation systems, condensers, and adsorp-
tion systems used to control VOC emissions. In
these applications, direct measurements of multiple
gaseous components in the process gas streams
became overly complicated and resource intensive.
Instead of using sophisticated monitoring instrumen-
tation, simpler parameter monitoring approaches
were adopted in exchange for accepting more
stringent "worst case" DCPLs.
This use of parameter surrogates in place of using
CEM systems to monitor emissions may appear
straightforward. However, as in all emission moni-
toring techniques, the method is application depend-
ent. In determining the appropriateness of the
method, one should consider the following factors:
• The relationship between the parameter or
parameters should be straightforward.
• The relationship should hold for all operating
conditions.
* The correlation between emissions and the
parameter should have a high degree of
confidence with narrow confidence and
tolerance intervals.
Parameter surrogates are most appropriate when the
relationship is a simple one, such as in using inciner-
ator temperature for VOC control, or water-to-fuel
ratio for gas turbine NOX control. For the method to
be practical, no other operating variables should
affect the correlation significantly. If they do, they
should be included in the correlation, which will then
make the correlation more complicated and more
uncertain. These issues are discussed in detail
by Evans (Evans, 1994).
The DCPL approach is similar to a predictive method
since the DCPL value is established on past data or
performance data to predict present compliance with
emissions standards. It differs from the statistical
predictive emission monitoring (PEM) models in that
it is less refined (see Evans, 1994) and is established
over a narrow range of operating conditions. In a
DCPL approach, parameter values such as pressure
drop, temperature, or supply pressure are reported
to the agency. In a predictive method, emissions
values calculated by the model from parameter
values are reported to the agency.
4.5 Emission Modeling - Predictive
Emissions Monitoring Systems
Many applications occur where parameter relation-
ships to emissions are more complicated than
establishing a simple DCPL and sensor determined
parameters might be better incorporated into a
predictive emission monitoring (PEM) model, or
system. In the PEM model approach, emission
values are calculated by the model from the input
parameter values.
When developing a predictive model, parameter and
emissions data are accumulated under various
operating conditions. Then the data are used to
develop the model. Two approaches are possible:
1) If the source or control equipment operation is
well understood, first principles calculations can
be made to determine the emissions. Based
upon the physical and chemical effects of oper-
ating parameters on emissions, the actual emis-
sions may be determined. This approach is often
called the "first principles" or "phenomenologi-
cal" approach.
2) If the effects of operating parameters on emis-
sions are not well understood, or if theoretical
calculations become too complicated, statistical
methods may be applied. Linear and nonlinear
regression techniques have been used success-
fully in many situations. Neural net methods
also have become popular in these applications.
No one modeling approach can be said to be the
"best." As with CEM systems, one must consider
tradeoffs in each application. Both first principle and
statistical models have passed relative accuracy
tests in specific applications (Hung, 1994, Clap-
saddle, 1995, Clapsaddle, 1996).
Predictive systems are basically empirical models,
Even the first principle approaches use past data for
evaluation purposes. The theory establishes the
form and mathematical functions of the model, and
test data commonly are used to introduce empirical
constants to fine-tune the model. In building these
models a test program must be developed that can
provide representative emissions and parameter data
over the probable, expected range of operating
conditions of the emission source.
60
-------
4.5.1 First Principle Models
Models can be developed from the fundamental
theory associated with the operation of a process or
a pollution control device. Depending on the pro-
cess, principles of thermodynamics, chemical kinet-
ics, fluid flow, and so on, may be applied. First
principle models have been applied to calculate
nitrogen oxide emissions for can-type gas turbines
(Hung, 1993, 1994, 1995). Also, theoretical design
equations have been developed for most types of air
pollution control equipment; particulate scrubbers,
gas absorbers, condensers, electrostatic precipita-
tors, etc.
First principle models provide for an understanding
of a process and the relative importance of the input
parameters to its performance. A pure first princi-
ples model does not depend upon historical data and
can be used over the full range of process operation.
4.5.1.1 Semi-empirical Models
Theoretical design equations can give good qualita-
tive information, but quantitative information having
the accuracy necessary to be legally enforceable
usually is not obtainable without empirical correla-
tion. Not all of the effects of process variables may
be known and the values of the necessary input
parameters may not be of sufficient accuracy to give
correct results. The problem here is that most
devices are too complex. Calculating emissions
from first principles often requires too much informa-
tion to be practical. For example, in the "infinite
throat model" for Venturi particulate scrubbers a
knowledge of the particle size and droplet size is
necessary
-------
In practice, one desires a model that will be valid
under a range of values of x. Thus, a number of
tests would be conducted to obtain data that might
appear as that shown in Figure 4-3.
Parameter Value
Figure 4-3. Test data. Emissions as a function of
the parameter values, x.
Obviously scatter occurs in the emissions data. The
scatter in y is due to uncertainties and variability in
the emissions measurements because of problems in
measurement, difficulties in keeping process vari-
ables (parameters) other than x constant during the
testing, and to unknown variables that may be
affecting the process.
In the statistical methods, the goal is to find an
equation that best summarizes the test data and
predicts emissions to within an acceptable level of
confidence. Based upon the quality of the input
data and the choice of parameter or parameters or,
based on past data, how confident are we in our
equation predicting future values?
Statistics can answer these questions for us and
many statistical methods have been developed to fit
curves to data and qualify the results. The least
squares linear regression and multiple linear regres-
sion techniques are very useful in this regard. The
extended statistical method of neural nets has also
been useful in providing greater flexibility for nonlin-
ear expressions,
4.5.2.1 Least Squares Linear Regression
Techniques
The least squares technique is one of the simplest
methods used to fit a line to emissions/parameter
data. In the example above, if y is a linear function
of the parameter x, the line y = a + bx is such that
if one takes the deviation from each point to the line
and squares the deviations, the constants a and b in
the equation will be such that the sum of the
squares is the smallest possible value (Figure 4-4).
Parameter Value
Figure 4-4. Linear regression of test data.
In mathematical terms a and b are calculated to be
such that
- (a +
= a minimum value
One can show that the values for a and b which
define the line of minimum deviation (the regression
line), can be calculated from n data sets {xif y,) as
follows:
a =
b =
nix,2 - (£x,)2
n£xz
The calculations are tedious if performed manually,
however. Computer programs are available to do
such routine calculations easily and rapidly.
The determination of a and b is not a matter of
guess or iteration. In this statistical method, an
underlying assumption is that the value of x has
negligible error or is free from error. It assumes that
the deviation lies principally in the measurement of
y, in this case, the emissions.
Other assumptions in the method exist, but the main
point here is that the values of a and b are mathe-
matically determined and depend only on the original
test data. If the dependence of y on x should
change because of a change in operations, system
degeneration, or faulty determination of x, the best
line will no longer be valid.
62
-------
4.5.2.2 Multiple Linear Regression
More than one parameter may influence the value of
y (the emissions). The equation for y may then
appear as:
y = b0 + b,x, + bzx2 -i-
bkxk
for k parameters. This is the more likely case in
emissions applications and techniques of multiple
linear regression can be used. The approach is
similar to that given for simple linear regression, but
a set of k simultaneous linear equations are solved
to determine the values of b0 to bk , In multiple
regression, the effect of one parameter on the
emissions can be determined while the other param-
eters are kept constant,
Linear regression statistical techniques have been
used widely to develop PEM system models. Pub-
lished examples most often address monitoring NOX
emissions from industrial boilers. Evans (1995) has
developed a PEM NOX modal for a gas-fired boiler
using two parameters, % excess 02 and flue gas
temperature. Macak (Macak, 1988}, using multiple
linear regression techniques, developed a model for
a natural gas-fired boiler using three equations, each
used over different load ranges, and each using two
to three different parameters.
4.5.2.3 Higher Order Multiple Linear Regression
Higher order linear models may be used to provide
curve fits to data (Draper, 1981). In the case of
more than one parameter, the regression expression
may include higher orders of the input variables and
incorporate expressions such as:
y = a + bx 4- cxz
For multiple input parameters both polynomials and
polynomial cross-products may be included:
y =
b-,x,
These expressions are termed linear regression equa-
tions since they are linear in terms of the regression
coefficients. The first expression is a second order
linear equation. The second expression is a third
order linear equation.
4.5.2.4 Nonlinear Least Squares Regression
Nonlinear models express the model output (emis-
sions) as a nonlinear function of the regression coef-
ficients (a, b,...). The nonlinear function may be a
power function, or a logarithmic or exponential func-
tion of the regression coefficients. Examples of non-
linear models include:
y = a 4- x"
y = a + log(bx)
y = a 4- e"bx
The constants for nonlinear models are determined
by iteration techniques similar to those in linear re-
gression. Some non-linear forms present significant
mathematical complexity and require numerical
methods rather than analytic solutions.
Many more mathematical options are available in the
nonlinear and multiple order curve-fitting techniques.
One may be able to represent the initial input data
well with such models, but a danger exists in "over-
fitting" the data. One can "correlate" or fit a curve
between any two sets of numbers, but if no actual
relationship exists between them, future predictions
will not necessarily be valid.
Clapsaddle (Clapsaddle, 1995, 1996) has used poly-
nomial expressions to represent emissions in gas and
oil-fired boilers. Four to six parameters such as fuel
oil flow rate, air flow rate, excess 02, fuel gas flow
rate, air damper position, air heater outlet tempera-
ture, fuel bound nitrogen, were used in the various
models. Snyder, et. al. (Snyder 1996) used up to
five parameters in nonlinear models for predicting
NOX and CO emissions from stationary gas turbines.
4.5.2.5 Neural Network Models
Neural network models have been applied recently
to model source emissions. Although the neural net-
work methods are inherently mathematical, analo-
gies can be made to biological learning processes.
The regression methods discussed so far provide a
model that is calculated by using a set of equations.
In the neural net method, the model constants are
not calculated, but are determined by iteration. The
constants in the model are varied incrementally until
a set of constants is obtained that will reproduce the
actual emissions of the input data set. Least
squares regression methods are still used in neural
net models to minimize the residual differences
between the test emission data and the model pre-
dicted emissions.
In developing neural network models, extra sets of
constants are introduced into what are called
"hidden layers (or nodes)." These constants, in
conjunction with nonlinear functions, can weight the
importance or unimportance of different input
parameters in contributing to a given result. This
"weighting" also is done by an iterative process.
63
-------
This weighting process is similar to what neurons do
in the brain, hence the word "neural."
Neural network procedures are somewhat of a
"brute force" approach in statistical model
development. The procedures use the iterative
capabilities of the computer to choose optimum
constants that best represent how a process
operates. The method is powerful since most
process operations are complicated and the
interrelation between input parameters is not always
well understood. The approach is similar to using a
computer to numerically integrate an integral that is
too complex to solve analytically.
Neural network models do offer some greater
flexibility in optimizing system operations. Since the
contribution of the various parameters to the
operations is better understood, this information can
be fed back to improve system performance. Neural
network process optimization models have been said
to have been developed for the chemical, petro-
chemical, semiconductor, and mining industries
{Keeler, 1993).
In one study, 21 input parameters, selected from
120, were used to develop an NOX prediction model
for a gas fired boiler (Collins, 1994). Clements, et.
al. (Clements, 1996) applied both multiple linear
regression methods and neural network techniques
to develop PEM models to predict NOX, CO, O 2, and
stack gas flow rates for gas turbines and
reciprocating gas engines. In this study comparing
the two types of models, regression coefficient (R2}
values were found to be better for the neural
network analysis than for linear regression analysis.
Note: R2 values give a measure of the "fit" of the
model to the data, but are not necessarily an
indicator of which model will provide the best
prediction from new data (Evans, 1995).
4.5.3 Model Development
A typical approach to model development is to first
review all potential operational parameters that
affect emissions, determine the full operating range
for each parameter, examine potential cross para-
meter interaction, and then develop a test matrix to
evaluate the effect of parameter changes on the
emissions (Clapsaddle, 1996). The emissions are
then characterized according to the test plan and a
regression model developed from the test data.
Evans (Evans, 1994) proposes a similar approach
that emphasizes care in the selection of model
parameters. The test must concentrate on
parameters that have an effect on emissions.
Inclusion of parameters that have negligible effects
will complicate the test without commensurate
improvements in the model. The main problem in
developing models is in not recognizing parameters
that can affect emissions. This "lurking parameter",
being as simple as an open or closed damper, can
easily invalidate any model (Evans, 1994).
A model should be developed over a sufficient
period of time where a full range of operating
conditions can be correlated to the emissions.
During this time both accurate emissions data and
sensor data are required. If either inaccurate sensor
or emissions data are used to build the model, the
model itself will be inaccurate.
Usually more than one model is examined when
developing a PEM system. Different combinations
of parameters and both linear and nonlinear
regression equations may be used to examine those
having the best fit with the data. An examination of
residuals, confidence intervals, and regression
coefficients typically are examined in the evaluation
process (Evans, 1995).
A relative accuracy test audit (RATA) used to certify
GEM systems is not necessarily sufficient to validate
a PEM model (Eghneim, 1996). The RATA normally
is conducted at only one operating condition of the
source. For a PEM system, if that operating con-
dition were to be one under which the correlation
were developed, the system should obviously pass.
A truer evaluation would be to conduct a RATA, or
RATAs, at operating conditions different from those
used to develop the correlation. To be "predictive,"
the model must provide true emissions values from
parameter data not previously provided to the
model.
4,5.4 Model Quality Assurance
In NSPS requirements for operations monitoring,
sensor calibrations are required to be checked
annually. The manufacturer of a sensor is expected
to deliver a product to specified levels of precision
and accuracy. These levels are chosen to be within
an acceptable range for the intended application,
usually ±5% of reading. However, users often
assumed incorrectly that the performance of a
sensor will remain constant with time. The
calibration may drift over time, the sensor may
become fouled, or it might not work at all. Plant
maintenance routines provide for inspection of these
devices, but the sensors most critical to plant
operation usually receive the most attention.
If a parameter monitoring program is to be initiated,
the sensors used to provide parameter values for a
64
-------
model or correlated relation must provide values at
consistent levels of precision and accuracy. To do
otherwise leads to faulty data, just as would a
poorly calibrated or malfunctioning analyzer of a
CEM system. A quality assurance program institu-
ted to assure that the sensor data continues to be
valid is as essential in PEM systems as a preventive
maintenance program is in a CEM system (Macak,
1996, DeFriez, 1996). The "once yearly" check for
a sensor may no longer be sufficient and quarterly
or monthly preventive maintenance procedures may
be necessary to ensure the continued validity of the
model.
PEM systems do, however, have the capability of
performing self-diagnostics and making adjustments
for poor sensor data. If redundant sensors are
installed in a system, the PEM system might switch
to another operating sensor after one fails.
Algorithms can be instituted to perform cross-
checks, or reality checks for sensors. Historical data
might even be substituted for missing data until a
faulty sensor is replaced or the PEM system may
model the sensor data itself from other input
variables. However, the regulatory acceptability of
these substitution procedures has not been
addressed in most state or federal monitoring
programs, except for a few instances (for example,
see 40 CFR 75 Appendix C).
EPA historically has required daily calibration checks
for monitoring instrumentation so as to provide for
legally defensible data for each operating day. The
necessity of providing a means for assuring data
quality on a routine basis is as important in PEM
systems as it is in CEM systems
-------
happens under upset conditions, what are the
emissions when the process is not operating
properly (see for example, Macak, 1996). Since
most correlations are not developed under such
conditions, the model predictions will either be
incorrect during the period of malfunction or the
model may flag the data as "missing." in the first
case, the data will be misleading (Figure 4-5b); in
the second case, no data will exist {Figure 4-5c).
The scenarios of Figures 4-5a and 4-5b may be
convenient for the source required to report periods
of excess emissions, but not desirable for the
regulatory agency. A more complete model would
be able to determine that the upset occurred
through appropriate sensing devices, and would then
switch to an alternate model or algorithm to account
for the upset. The problem here is multifold. First,
one would have to be aware of all possible upset
conditions. One would then have to either initiate
or simulate the possible upset conditions and obtain
reference emission data to develop the alternate
models. This could be prohibitively expensive, or
impossible if one could not or did not wish to
produce the upset conditions during the
development effort. Another approach might be to
establish "control limits," or limiting conditions for
the model. When the process operates outside
these limits, the model will be out-of-control and
data cannot be used. Such limits would be anal-
ogous to the 40 CFR 60 Appendix F out-of-control
limits for a GEM system.
For example, in gas turbine models, NOX emission
predictions are based upon the inherent assumption
that the turbine is maintained and operated under
the conditions under which the model was
developed. Should water injection nozzles become
plugged, or water distribution in the turbine become
uneven, the model becomes invalid. The wear and
tear on a system, the normal degeneration of system
components due to continual operation will remove
the system from the baseline conditions from which
the model was developed. These issues are similar
to those associated with GEM systems, since a GEM
system must be maintained properly to operate
under the conditions at which it was certified. As
with GEM systems, a program of quality control, of
checking model performance and sensor perfor-
mance, must be instituted to ensure that the model
continues to represent current conditions.
PEM model maintenance issues are therefore very
important. Just as with GEM systems, when a
component is modified or replaced, the question of
recertification arises when the model is modified.
Agency guidance in this area is still developing, but
questions posed by Clappsaddle are pertinent (Clap-
saddle, 1995): "Can periodic adjustments be made
to the PEM model equation (without recertification)?
If a RATA on the PEM shows inaccuracies, can the
PEM model be adjusted and then accepted until the
next required quality assurance (QA) audit? How
many times can a PEM fail a QA audit before a
regulatory agency requires that the PEM be
replaced?" Other questions arise with regard to
techniques used in "sensor validation." If a sensor
is replaced, is the model still valid? Is recertification
required? If a sensor fails and sensor data are
reconstructed to substitute and maintain data
availability, should the model be certified under
those conditions and other possible sensor data
substitution scenarios? (See DeFriez, 1996 for a
discussion of sensor issues.) If a model contains
different algorithms for different operating
conditions, should the model be certified at each
operating condition?
The cost issue has become distorted considerably in
PEM system vs GEM system arguments. PEM
system costs are often compared inappropriately to
the costs associated with GEM systems installed to
meet the requirements of the Part 75 acid rain pro-
gram. In Part 75, the hardware precision and
accuracy requirements and detailed data acquisition
and handling systems (DAHS) specifications
necessary to legitimize data for allowance trading
justified higher costs. Systems installed to meet the
NOX or VOC monitoring requirements of a state Title
V permit program are not required to meet such
stringent specifications, and their costs are con-
sequently much less.
Many GEM system suppliers now offer GEM systems
at prices comparable to commercial PEM systems.
When one considers the correlation testing neces-
sary to develop a PEM system model and the
certification testing necessary to validate the model,
initial capital costs are often comparable. Also, if
the original correlation becomes invalid due to
process changes or system degeneration, the
correlation testing would have to be redone. The
ongoing quality assurance costs can be comparable
between the two methods. These cost issues are
further discussed in Chapter 6 of this manual.
Those required to monitor non-criteria pollutants
may have the flexibility to choose between a GEM
and PEM system to meet the requirements of future
rulernaking. Depending on the application, the two
techniques can be competitive both in terms of cost
and accuracy. However, one of the most powerful
66
-------
a. Actual Emissions
^IpfWllW^
b. PEM system not accounting for upset conditions
missing data
c. PEM system detecting upset having no input
data to correlate to upset conditions
d, PEM system detecting upset and switching to alternate
model to account for upset conditions
Figure 4-5. Emissions calculations (PEM system predictions) based upon upset conditions not
accounted for by the model.
67
-------
options is the combination of two methods
(DeFriez, 1996, Macak, 1996), With a combined
CEM/PEM system, the PEM system can serve as a
backup to the CEM system, as in the prescribed
techniques can be competitive both in terms of cost
and accuracy. However, one of the most powerful
options is the combination of the two methods C-
EM/PEM system, the PEM system can serve as a
backup to the CEM system, as in the prescribed
methods of the acid rain CEM program (40 CFR 75),
Or, a CEM system can serve as a backup to a PEM
system. When a sensor validation program causes
the PEM system to invalidate data and report miss-
ing data due to failed sensors or an unrecognized
operating condition, the CEM system can fill in the
missing data gap. The CEM system can provide the
data where it may be most needed, the period of
plant upset conditions not accounted for by the PEM
model.
The final decision in choosing a monitoring method,
whether a CEM or a PEM system, depends upon
how the data will be used and the constraints of the
application. Questions need to be asked about
whether a DCPL or predictive model will meet mon-
itoring requirements, will the method be sufficient
for demonstrating compliance, or will the data be
accurate enough for a trading program? A
consideration of these questions should involve an
evaluation of cost, regulatory, and technical issues;
none should be considered solely.
References
Bills, R.T. and Rae. J.T. 1995. "Parametric
Monitoring of Pollution Control Equipment for Non-
Continuous, Flexible Manufacturing." Paper
presented at the Air & Waste Management
Association Meeting, San Antonio, TX, Paper 95-
FA160.01.
Bivins, D. and Marinshaw, R. 1996. Update on
Compliance Assurance Program: Revising the
Enhanced Monitoring Rule, In Continuous
Compliance Monitoring Under the Clean Air Act
Amendments, Air & Waste Management
Association. Pittsburgh, PA. pp 243-259.
Calvert, S. Goldshmid, J., Leith, D., and Mehta, D.
1972. Wet Scrubber System Study, Volume I:
Scrubber Handbook. EPA-R2-72-118a. U.S.
Environmental Protection Agency. Research Triangle
Park, NC.
Cashin, M.G., Love, J.E., and MuIIer, J.D. 1996.
"Resolving the Difference — Mass Balance vs. CEM
Air Emissions Estimates. Paper" presented at the
Electric Power Research Institute CEM Users Group
Meeting, Kansas City, MO.
Clapsaddle, C.A., 1995. "Comparison of NOX
Emission Rates Estimated by 40 CFR Part 75
Appendix E Procedures and Parametric Emission
Models with Reference Method CEM Measure-
ments." Paper presented at the Air & Waste
Management Association Meeting, San Antonio, TX.
Paper 95-MP16A.05.
Clapsaddle, C.A., and Cunningham. 1996.
Performance Evaluation of Parametric Emission
Monitoring Systems for Boilers. Continuous
Compliance Monitoring Under the Clean Air Act
Amendments. Air & Waste Management Associa-
tion. Pittsburgh, PA. pp 162-176.
Clements, B., Hayden, A.C.S., Zheng, L., and Dock-
rill, P. 1996. "Comparisons of Neural and
Statistical Methods for the Parametric Prediction of
NOX from Natural Gas-Fired Engines." Paper pre-
sented at the Air & Waste Management Association
Meeting, Nashville, TN. Paper 96-RA109.01.
Collins, N. and Terhune, K. 1994. A Model Solution
for Tracking Pollution, Chemical Engineering -
Environmental Engineering Supplement.
DeFriez, H., Seraji, H., and Schillinger, S. 1996.
"Neural Logic Solutions for Emission Monitoring."
Paper presented at the Air & Waste Management
Association Meeting, Nashville, TN. Paper 96-W-
P92.05.
Denmark, K., Farren, M., and Hammack, B. 1993.
Turning Production Data into SPC Gold. Intech.
Vol. 40. No. 12. pp 18-20.
Draper and Smith. 1981. Applied Regression
Analysis. Wiley Interscience. New York, NY.
Eghneim, G. 1996. "Thoughts on Predictive
Emissions Monitoring from a Regulatory
Perspective," Journal of the Air & Waste
Management Association, Volume 46, pp 1086-
1092.
Evans, S. 1994. "Predictive Monitoring: The Way
Things Ought to Be." Paper presented at the Air &
Waste Management Association Meeting, Cincinnati,
OH. Paper 94-WA73.06.
68
-------
Evans, S. 1995. "Developing Predictive Models for
Enhanced Monitoring." Paper presented at the Air
& Waste Management Association Meeting, San
Antonio, TX. Paper 95-TP64.01
Hung, W.S.Y., 1993. Predictive Emission Monitoring
System (PEMS): An Alternative to In-Stack
Continuous NOX Monitoring. Continuous Emission
Monitoring - A Technology for the 90s. Publication
SP-85. Air & Waste Management Association,
Pittsburgh, PA. pp 314-323,
Hung, W.S.Y., 1994. "Predictive Emission
Monitoring System (PEMS): A Proven Alternative to
In-Stack Continuous NOX Monitoring." Paper
presented at the Air & Waste Management
Association Meeting, Cincinnati, OH. Paper 94-
TA29A.03,
Hung, W.S.Y. 1995, "Predictive Emission Monitoring
System (PEMS); The Established NOX Monitoring
System for Industrial Gas Turbines." Paper pre-
sented at the Air & Waste Management Association
Meeting, San Antonio, TX. Paper 95-MP16A.01.
Keeler, J., Havener, J., Hartman, E., and Magnuson,
T. 1993. Achieving Compliance and Profits with a
Predictive Emissions Monitoring System: Pavilion's
"Software CEMtm".
Macak, J,J. 1988. Development of an Operations
Monitoring Plan for a Subpart Db Industrial Boiler.
Paper presented at the Air & Waste Management
Association Meeting, Dallas, TX. Paper 88-137,4.
Macak J,J., 1996. The Pros and Cons of Predictive,
Parametric, and Alternative Emissions Monitoring
Systems for Regulatory Compliance. Paper pre-
sented at the Air & Waste Management Association
Meeting, Nashville, TN. Paper 96-WP92.02.
Mycock, J.C., McKenna, JD., and Theodore, L.
1995. Handbook of Air Pollution Control
Engineering and Technology. CRC Press. Boca
Raton, FL.
Samdani, G.S. 1994. Software Takes on
Monitoring. Chemical Engineering, pp 30-33.
Air
Siebert, P.C., Aydil, M.L, and Jackson, R.W. 1995.
"Emission Estimating Procedures for Non-Traditional
and Small Emission Sources at DOD Facilities."
Paper presented at the Air & Waste Management
Association Meeting, San Antonio, TX, Paper 95-
WA85.05.
Snyder, R.B., Levine, P., and Bautista. 1996.
Continuous Parameter Monitoring System
Development for a Stationary Gas Turbine.
Continuous Compliance Monitoring Under the Clean
Air Act Amendments. Air & Waste Management
Association, Pittsburgh, PA. pp 73-84.
Steven, T. 1994. Software Cuts Clean Air Costs.
Industry Week. January 17, 1994, pp 45-48.
U.S. Environmental Protection Agency. 1993.
Enhanced Monitoring Program; Proposed Rule.
Federal Register, Vol. 58, No. 203, October 22,
1993, p 54648.
Yung, S., Calvert, S., and Barbarika, H.F. 1977.
Venturi Scrubber Performance Model. EPA 600/2-
77-172. U.S. EPA. Cincinnati, OH.
69
-------
Chapter 5
System Control and Data Recording - Data
Acquisition and Handling Systems
A data acquisition and handling system {DAHS) is a
combination of hardware and software that is used to
record data and generate reports for submission to the
regulatory agency. A properly designed DAHS also
can provide warnings of CEM system fault conditions
and excess emissions so that adjustments and repairs
can be made in a timely manner. The DAHS is con-
sidered an integral part of the CEM system and must
be included as part of the CEM certification and audit
procedures. Because CEM data are a key enforce-
ment tool used by EPA and local agencies, emissions
reports must be prepared in a manner that accurately
reflects the quality of the data and any anomalies that
occurred during the reporting period (McCoy, 1986).
Although not generally considered part of the DAHS,
control of CEM system functions such as calibration,
probe blowback, and probe switching (for time
sharing systems) is often performed by the same
hardware and software that performs some data
acquisition functions. For example, the same pro-
grammable logic controller (PLC) may be used to
calculate emission rates and to control calibration
cycles. Decisions about the CEM control hardware
and software must be made with a consideration of
how it will affect the ability of the user to change
DAHS vendors in the future {see Commercial PLCs
section). This chapter contains discussions of CEM
system control and data acquisition functions.
The complexity of the DAHS required for a given
application depends on the number of emission
parameters being monitored and the applicable report-
ing requirements. Three basic options for handling
emissions data are described in this chapter: a
simple data recording device (stripchart or data-
logger), a plant mainframe computer (often used to
implement a distributed control system {DCS}}, and a
commercially available DAHS. The design elements of
commercial DAHSs are described in detail
5.1 Option 1: Simple Data Recording
Device
In the case of very limited data reporting require-
ments, a simple stripchart recorder or data logger may
be adequate. Digital stripchart systems will average
data, calculate emission rates, and display the data
either graphically or in tabular form (Figure 5-1).
Often they have some data storage capability so that
data can be downloaded periodically to a personal
computer. The simple recording device may be the
preferred option if the facility is required to monitor
and certify compliance, but is not required to submit
summaries of emission data or CEM system quality
assurance data to the regulatory agency. A PLC or
other device may be necessary to control calibrations
and sampling, but in some cases these functions can
be controlled, by the gas analyzer itself.
Figure 5-1. Odessa paperless stripchart.
To quantify negative zero drift during calibrations, the
recorder scale must be capable of providing concen-
tration results well below zero. For example, 40 CFR,
Part 60, Appendix F CEM quality assurance proce-
dures require that a CEM system be declared "out of
control" if the calibration drift exceeds four times the
drift limit in the applicable performance specification.
70
-------
For example, if a CEM system with a range of 0-1000
ppm was subject to a drift limit of 2,5% of span
during the Performance Specification Test, the DAHS
would have to be capable of displaying a reading at
least as low as -100 ppm (-0.025 x 4x 1000 =
-100).
Some regulatory agencies no longer accept stripchart
data for reporting purposes (even as a backup system)
or they place restrictions on the resolution of the
recorder. Therefore, consultation with the regulatory
agency must occur before deciding to use this option.
Choosing a simple data recorder will reduce the initial
capital Investment in the CEM system, but it may be
more costly in the long run. If the CEM system does
not have a method of recording fault conditions or
process operating status, interpreting the emissions
data retroactively for CEM system or process upset
conditions may be difficult and time-consuming.
Thus, the validity of the data for extended periods of
time could be questionable. Without a DAHS, the
user will be required to keep up with the calibration
and other quality assurance data manually, making
implementation of a CEM system quality assurance
plan more difficult. Also, a backup recorder should be
used to prevent failures in the data recording device
from causing violations of CEM system downtime pro-
visions.
5.2 Option 2: Plant Mainframe Computer
System
A limited number of facilities have opted to develop
their own DAHS using the plant mainframe computer
system. Plant process control systems typically are
equipped with excess data recording and storage
capability. The plant Internal programming option is
attractive because the CEM data can be integrated
easily with the other plant data display and backup
systems. The programming also can be tailored to
meet the specific needs of the plant operators. The
downside to this approach is that the extra program-
ming required to keep up with CEM emission rate
calculations, alarm conditions, calibration data,
emission data validation, and report generation can be
underestimated. An internally programmed system
may not be as flexible as a commercial product and
may need periodic reprogramming to keep up with
changing regulations or operator needs. When
choosing between internal programming and a com-
mercial DAHS, the user should examine carefully all of
the functions offered by commercial systems and
make an honest appraisal of the cost and delivery
schedule issues associated with developing the
programming in-house. Commercial DAHS provided
either by the CEM vendor or by an independent DAHS
provider have been found to be more cost effective by
most CEM system users,
5.3 Option 3: Commercial DAHS
A DAHS is a stand-alone system that is used to
display data in units of the emission standard, provide
alarms, and prepare reports for submission to regula-
tory agencies. It can be used by the operator to
produce edited emission summaries, excess emission
reports, alarm reports, and CEM system downtime
and corrective action reports. A typical system
(Figure 5-2) consists of a personal computer, a data
interface (typically a PLC or datalogger), a backup
data storage device, and a printer. The DAHS com-
puter may be connected to a local area network (LAN)
and/or it may be equipped with a modem for remote
access.
Since the implementation of the new source perfor-
mance standards (40 CFR, Part 60) and the Acid Rain
Program (40 CFR, Part 75), CEM system and DAHS
vendors have made great strides in the reliability,
flexibility, and utility of personal computer-based
DAHSs. Recent additions to DAHS software pack-
ages include reporting requirements for 56 CFR 7134
(BIF Rule), 40 CFR 63 (MACT standards), 40 CFR 503
{sewage sludge incinerators), and the revised stan-
dards for hazardous waste incinerators. Early DAHS
software was written specifically for each installation,
but most vendors now have standard software
products that can be configured by the user for each
regulatory application.
If the user has special data reporting requirements
from an unusual permit condition or a state rule, the
software provider may need to perform custom
programming. Such custom programming can be
costly for the user and can lead to software
"bugs." If possible, the user may propose a more
standardized reporting format to the regulatory
agency to avoid custom programming.
DAHS functions and some typical hardware com-
ponents are described in the following sections.
5.3. / Emission Data Recording
Some gas analyzers and sensors send digital signals
directly to the DAHS and others send analog signals
(usually 4-20 mA) that must be converted to a digital
value and scaled to the appropriate units (concentra-
tion, flow rate, temperature, etc.). Concentration
and flow rate data are used to calculate emission
rates in units of the emission standard. The units of
the emission standard vary according to the type of
facility, the applicable regulation, or the permit
71
-------
Process
Data
Gas
Analyzers
GEM Fault
Sensors
COMPANY
LAN OR WAN
DATA INTERFACE
(PLC or Datalogger)
BACKUP
DATA STORAGE
Figure 5-2. Typical DAHS system.
conditions. Some common units for emission
standards are:
* ppmy of pollutant
• ppmv of pollutant corrected to a standard 02
dilution basis (e.g., ppmv at 7% 02)
» ppmtf of pollutant corrected to a standard
C02 dilution basis (e.g., ppmu at 12% CO2)
• Ib of pollutant/MMBtu of fuel combusted
» Ib of pollutant/hour
• Ib of pollutant/lb of product
A more comprehensive discussion of the units of the
standard are found in Appendix E.
Because analyzer measurement ranges and the
parameters used to calculate emission rates can
change over time, most DAHSs include a menu
system to allow the user to scale the analyzer
channels and edit the emission rate calculations.
Emission rate calculation errors are not uncommon,
therefore the emission rates should be verified with
hand calculations when the DAHS is first installed
and after each change of the calculation parameters.
The applicable monitoring regulation or permit
conditions should state the minimum frequency of
data collection and recording for each parameter; if
not, it should be discussed with the regulatory
agency before the DAHS is chosen. Minimum data
collection frequencies can be from once per 10
seconds to once per 15 minutes, depending on the
variability of the emission parameter and the re-
sponse time of the CEM system, If the sampling
frequency is high, use of a PLC or data logger may
be necessary to preserve processor time on the
DAHS PC. The minimum frequency of data collec-
tion may be different from the minimum frequency
of data recording. For example, BiF regulations for
monitoring of total hydrocarbons require that data
be collected at least once each 15 seconds, but
average data values need to be recorded as one-
minute averages. For sources subject to 40 CFR,
Part 60 monitoring requirements, the minimum data
collection frequencies can be found in section
60.13.
The averaging periods should be flexible to allow
easy data collection during quality assurance testing.
Having the ability to print real-time one-minute
averages simplifies conducting and documentation
of cylinder gas audits and relative accuracy tests,
5.3.2 Emission Data Display
The main viewing screen of the DAHS should have
a real-time data display that is easy for operators to
read. The screen should show emissions data, CEM
system status (on-line, off-line, calibrating, etc.) and
the status of alarms. If the DAHS computer is not
located near the CEM system hardware, a second
display terminal should be located near the gas
analyzers for quality assurance checks and trouble-
shooting. Often the reading on the front panel of an
analyzer does not match the reading on the DAHS
screen due to small signal conversion errors or
automated calibration corrections. Therefore, the
analyzers should be tuned to provide accurate
results using results from the DAHS. A PLC or data
72
-------
logger located near the analyzers often can be used
as a display device for troubleshooting.
Most DAHS vendors include software that allows
the user to look at historical trends in the data
(Figure 5-3) for user-selectable intervals and aver-
aging periods. Trending of different combinations of
emission and operating parameters can be used to
determine whether changes in emissions data are
linked to plant operation or to CEM system opera-
tion. Key plant operating parameters such as
production rate, temperatures, or control device
parameters can be trended with emissions data for
troubleshooting. The trending software also may be
helpful in determining the nature of emission exceed-
ances when preparing emission reports for the
regulatory agency. If the plant computer control
system has ample capacity, some users may prefer
to send CEM system data from the DAHS to the
plant computer so that plant operators can use
display and trending software that is familiar to them
and already available at their workstations.
., SWI»WM
.'.vZBRjs
. w$
'• :12B.O;
ysisf
';:;MA
';?•,$&<$,
;A/^8
'••'•fa»i'
Figure 5-3. Example historical trend screen,
5.3.3 Sampling System Control
The DAHS software and/or computer may be used
to control routine sampling system control functions
or it may simply monitor control functions that are
conducted by a PLC. If the CEM system uses an
extractive sampling system, periodic purging of the
sample probe with compressed air to remove partic-
ulate matter from the probe tip filter may be neces-
sary. The duration and frequency of air purges
depend on the particular type of emission source
and on the source operation. After the CEM system
installation, it may require several months of opera-
tion to determine the optimum purge duration and
frequency settings. These settings should be
configurable by the user on the DAHS computer,
PLC, or datalogger. Likewise, CEM systems that
perform time-sharing of sample probes (i.e., using
the same analyzers for more than one emission
point) should include a method for adjusting the
frequency of probe switching. For both time-sharing
and air purges, the emissions data will be invalidated
for short periods. Some DAHS systems use "sample
and hold" circuits to keep the analyzer response
constant during purges, while other DAHSs label the
data as "invalid" and do not include it in averages.
Also, the time required for the sampling system to
be completely purged can change over time due to
filter plugging or changing sample pump perfor-
mance. The CEM system user periodically should
verify that the purge times used by the DAHS are
adequate to ensure that only representative sample
data are being recorded,
5.3.4 Calibration Control and Recording
Automatic daily zero and span calibrations of the
CEM system may be controlled by the DAHS or the
DAHS may monitor the activity of a PLC. For
calibration using cylinder gases, the DAHS should
have a feature that easily allows the user to edit the
initiation time, concentration values, and the dura-
tion of each calibration gas injection. For each
calibration standard, the following information
should be recorded:
• time and date
» value of the calibration standard
• instrument response to the calibration
standard
* amount of calibration error (as a percent of
instrument span)
An example calibration cycle configuration screen is
shown in Figure 5-4.
The DAHS also should allow the user to initiate an
automatic calibration cycle at any time. This feature
is especially useful when CEM system data are in
doubt or after any recalibration or repair.
Rather than adjusting the zero and span of an
analyzer at the instrument itself, some users prefer
to make zero and span adjustments by automatically
applying correction factors using the DAHS after
each calibration check. When such adjustments are
made, the calibration report must include the values
used for the calibration adjustment. Automatic
calibration adjustments apply only to data collected
after a given calibration and before the next calibra-
tion adjustment. Retroactive calibration adjustments
generally are not allowed. The accumulated amount
of calibration adjustment made by the software
should be limited so that degradation in CEM system
performance can be detected. An excessive soft-
73
-------
ware correction is a sign that the CEM system
hardware should be inspected. Typically, a DAHS
will generate an alarm when the zero or span soft-
ware correction exceeds 5 or 10 percent of span.
f J-s,
;>y\' ••-jjf-*^--' 7.77'rt£r;?r^*lt:?||^
>',?yJ5e! !^'A!"iL'/ f.!*t!^- ^feiN*
lfei:^-~(.tiiyhM-
, .. _ ^
*!!« v7 "•••', jO< '-y
Figure 5-4. Example calibration cycle configuration
screen.
5.3.5 Alarms
All commercial DAHSs can be configured to provide
alarms for excess emissions and CEM system fault
conditions. Depending on the design of the DAHS,
the alarms may be triggered at the DAHS computer,
PLC, datalogger, or at the analyzer itself. The DAHS
is equipped with alarm acknowledgments that allow
the user to enter the reasons and corrective actions
for alarm conditions. The alarms, reasons, and
corrective actions can be reviewed periodically to
identify recurring problems with the process or CEM
system operation. Triggers for excess emission
alarms can be set at lower setpoints or at shorter
averaging periods than the emission standard so that
corrective action can be taken before an actual
emission exceedance occurs. Some excess emission
indications during conditions such as unit startup/
shutdown or equipment malfunctions may be ex-
empted by the applicable regulations. CEM system
operators have found that when data are examined
only during the.preparation of quarterly emission
reports, incomplete records often lead to the inability
to determine the reason and corrective action for
each exceedance. For this reason, commercial
DAHSs contain lists of numbered reason codes that
are entered by the operator as part of the alarm
acknowledgment procedure. The operator also may
opt to enter a reason code for "unknown" or manu-
ally enter a reason that is not on the list. The list of
reason codes for a given facility should be reviewed
periodically to make sure that the list is complete
and that the reasons are adequately descriptive.
Some regulatory agencies can offer guidance on
which reasons constitute "excused" exceedances.
Some example excess emission reasons are pro-
vided below:
• bag leak
« baghouse blowback
* air damper malfunction
• plugged spray nozzles
* start-up
• shut down
The use of CEM system fault alarms can reduce the
time required for diagnosis and repair as well as
reducing the CEM system downtime. Fault alarms
also can be used to invalidate questionable emis-
sions data automatically, simplifying report produc-
tion. Some common CEM system fault conditions
are listed below:
• analyzer flame-out
* sample flow low
• sample vacuum high
» calibration gas pressure low
* sample line temperature low
« condenser temperature high
• water in sample line - pump shut off
* dilution air pressure low
5.3.6 Plant Computer Interface
As discussed in Option 2: Plant Mainframe Computer
System, the plant computer control system can be
used to display CEM data in a manner that is useful
to the process operators. Likewise, plant opera-
tional data should become part of the DAHS. The
most useful application of plant data is for determin-
ing when the process is on-line, off-line, or in a
startup or shutdown condition. Threshold levels for
such parameters as steam flow, fan speed, or stack
temperature can be used by the DAHS to automati-
cally mark the data as valid or invalid for emission
reporting purposes. In addition, EPA guidelines for
open market trading programs require correlation of
emissions data with the activity level (percent
capacity) of the emission source. Having production
unit and control device operational data in the data-
base while preparing periodic emission reports also
can be useful in explaining anomalies in the emis-
sions data or "unknown" reason codes for excess
emissions.
74
-------
5.3.7 Report Generation
Commercial DAHSs are designed primarily for
sources subject to monitoring provisions of 40 CFR
Parts 60 and 75. These monitoring requirements
include detailed reporting requirements for emissions
summaries, excess emissions, CEM system down-
time, and CEM system data quality assurance test
data. Even if the applicable reporting requirements
do not specify detailed reporting, the reporting
software that already has been developed for other
applications can be very useful for internal quality
assurance programs. DAHS systems include editing
functions that allow the user to review questionable
data and mark it as "valid" or "invalid" based on
what was known about the CE1V1 system and plant
operation at the time the data were collected. In all
cases, the raw unedited data files are retained
separately for later reference.
5.3.7.1 Data Flags
During data collection, the DAHS labels each dis-
crete emission data average according to the quality
of the data or the status of the CEM system at the
time that the data were collected. When emission
reports are prepared, the flags allow the editor to
sort data by its designation. DAHS vendors use
different systems for labeling, but a typical list of
data flag designations might be;
• valid data
» invalid data due to an alarm condition
• questionable data due to alarm condition
« process unit off-line
• process unit in startup or shutdown condi-
tion
» calibration in progress
» analyzer logged off-line for maintenance or
other QA activity
• substituted backup reference method data
during extended CEM system downtime
The data flags allow the DAHS software to calculate
percent data availability, process unit on-line hours,
and emission averages. The data can be sorted to
compile summaries of excess emissions, CEM
system downtime, and process status summaries.
5,3.7.2 Emission Summary Reports
To prepare a summary of emissions data the user
must first review all of the data for the reporting
period and resolve all issues of questionable data
quality. Preparation of this summary may involve
reviewing the plant or CEM system operation and
maintenance logs. When all data have been sorted
correctly, the DAHS automatically recalculates
averages and produces data summaries based on the
data that have been through the review process.
The editing function is particularly useful when
calculating long-term or roiling averages such as 24-
hour block averages or 30-day rolling averages from
one-hour average values.
5.3.7.3 Excess Emission Reports
Most monitoring regulations require the CEM system
operator to report the time, date, magnitude, reason,
and corrective action for each exceedance of the
emission standard. A common feature of almost all
DAHSs is the ability to produce automatically reports
summarizing excess emissions for a given reporting
period. The reason and corrective action codes
entered by the operator are provided for each
exceedance. During the review process, the user
can edit the reasons and corrective actions based on
new information. EPA has produced a guidance
document which describes how enforcement agen-
cies use excess emission reports (Paley, 1984).
5.3.7.4 CEM System and Unit Downtime
Reports
For each reporting period, the CEM system operator
generally is required to report the total hours of unit
operation and the percent availability of CEM system
data. By monitoring process parameters, the DAHS
can easily keep track of when the source is on-line
or off-line and in most cases can be used to produce
a summary report showing the time, date, and
duration of each outage. Using CEM system alarms,
analyzer on-line/off-line indicators, and information
entered during the emission summary editing pro-
cess, the DAHS can calculate the percent data
availability for each emission parameter and also can
produce a report indicating the time, date, duration,
reason, and corrective action for each CEM system
downtime incident.
5.3.7.5 Alarm Reports
The DAHS can be used to produce a list of all alarms
that occurred during a given period, including CEM
system faults and excess emissions. This summary
list can be used during the emission data review
process to ensure that all periods of questionable
data are resolved. It also can be used to identify the
most common CEM system fault conditions that
may require an equipment or procedure modification.
5.3.7.6 Calibration Drift Reports
The DAHS can be used to generate a summary of all
zero and span calibration data for the reporting
period. This report documents the time of each
daily calibration and whether the results are within
the applicable performance specifications. Even if
the drift results are within specifications, the sum-
75
-------
mary may indicate a long-term degradation in instru-
ment performance.
5.3,7.7 Data Assessment Reports
For those sources which are subject to the ongoing
CEM quality assurance procedures in 40 CFR, Part
60, Appendix F, a requirement exists to prepare a
data assessment report (DAR) for each calendar
quarter. The DAR contains a summary of all quality
assurance activities, including daily calibrations,
cylinder gas audits, and relative accuracy tests. The
DAR may be reviewed by the enforcement agency
to evaluate the status of a CEM system (Von
Lehmden, 1989).
5.3.8 Multitasking
The DAHS often is required to perform several
functions simultaneously. Some of the competing
demands placed on the software are illustrated in
Figure 5-5. The DAHS must have a resilient multi-
tasking capability to resolve the competing demands
without generating errors or "locking up" (Baranow-
ski, 1995). A procedure called "kernelling" is used
by the operating system to prioritize tasks and
prevent conflicts. In recent years, many DAHS
vendors have moved their software to operating
systems such as UNIX or OS/2 which are designed
specifically for multitasking and multi-user applica-
tions. Other systems that use MS DOS or Windows
operating systems rely on PLCs or dataloggers to
handle most of the routine tasks such as calibra-
tions, data averaging, and emission rate calculations.
Multitasking conflicts often are inconsistent and
difficult to trace. Oniy by interviewing other DAHS
software users or by conducting extensive accep-
tance testing can the purchaser be assured that a
vendor's software is free from such conflicts.
5,3.9 Expansion
Early providers programmed DAHSs specifically for
each application and any changes in the number of
inputs being monitored often required an expensive
software modification. Most DAHS vendors now
provide software that will allow the user to add gas
analyzers, emission rate calculations, alarms, and
even additional production units using a series of
menus within the software. Each vendor offers a
different level of flexibility. Before purchasing a
DAHS the user must determine the degree to which
the system can be expanded without reprogramming
by the vendor. Purchasing extra hardware capacity
to accommodate additional analog and digital signals
is prudent. After operating the CEM system through
the break-in period, many users find adding more
input signals from the plant distributed control
system (DCS) or providing more fault alarms for the
CEM system is helpful in the data review process.
5.3.10 Hardware
5.3.10.1 DAHS Computer
Most DAHS vendors use a stand-alone IBM-com-
patible personal computer for data storage, display,
and the generation of reports. However, vendors
who use the UNIX operating system occasionally
will install the DAHS software on the plant main-
frame computer system. Most DAHS vendors will
provide the computer with the DAHS but may allow
the user to select or purchase their own computer if
the company has a preferred manufacturer. The
DAHS computer most often is located in the plant
control room so that plant operators can have easy
access to the data. This also provides a clean
environment, a well-regulated power source, and
backup for emergency power.
Alamis
Data
Backup
Maintenance
Requests
User
Configuration
Figure 5-5. Competing demands for processor time.
5.3.10.2 Signal I/O Boards
The least expensive hardware option for processing
signals from analyzers and alarms is to use multi-
function boards that fit directly into the expansion
slots of the DAHS PC or into a chassis that is
attached to the PC. Multifunction boards accept
multiple analog input signals (from analyzers or other
sensors) as well as digital input and output signals
(e.g., alarms). Processing the signals places a heavy
burden on the PC processor to perform such real-
time functions as analog-to-digital conversion, signal
scaling, emission rate calculations, and averaging.
In addition, the PC may be required to control such
functions as calibration and blowback. With the
increased burden on the processor, multitasking
becomes more difficult. Vendors using multifunction
boards may require the user to conduct data review
and report generation on a separate computer.
Another drawback to this system is that If the PC
stops running for any reason, emission data are not
collected or stored.
76
-------
5.3.10.3 Commercial PLCs
Many DAHS vendors use standard PLCs to gather
analyzer and alarm signals, calculate data averages
and emission rates, and control calibrations and
blow/back. CEM data are downloaded periodically to
the DAHS computer where the data are displayed
and stored for report generation. !n some cases the
PLCs are equipped with memory modules so that
data can be stored for days at a time in case of a
failure in the DAHS computer. When the computer
is restored, the stored data are down-loaded and
no data loss occurs. Often plant instrument techni-
cians are already familiar with programming PLCs
and may even have backup hardware available in
case of a failure. The DAHS vendor should be
familiar with the PLC that is chosen for a given
application. Often DAHS vendors offer a choice of
PLCs, based on the preference of the plant opera-
tors. Some common PLC manufacturers that are
supported are:
DAHS vendors can operate several dataloggers (and
CEM systems) using one DAHS computer.
General Electric
Allen Bradley
Siemens
Modicon
controllers using the
nications protocol
Modbus eommu-
Since PLCs use different communication protocols,
the DAHS vendor should be asked if they have
installed other systems using the PLC that is chosen.
A significant advantage to using a commercial PLC
is that it may be compatible with DAHS computers
from several different vendors. This flexibility makes
a change of software vendors possible without
purchasing additional hardware. Previous experience
by DAHS users in the acid rain program has shown
that keeping all options open can be valuable in case
the chosen software does not perform adequately
(Huberland, 1995).
B.3.10.4 Custom CEM System Dataloggers
Instead of PLCs, some DAHS vendors provide their
own custom data loggers that are designed specifi-
cally for DAHS systems {Figure 5-6). Some of these
systems were developed originally for use at ambi-
ent air monitoring stations. The dataloggers perform
all of the same functions of a PLC, but often have
added features such as real-time data display,
enhanced data storage, and battery backup. In the
event of a DAHS PC failure, the datalogger will
continue to collect data until the PC comes back on-
line. The datalogger can be programmed from the
front panel or from the DAHS PC using menu-driven
software. Because of the added functionality,
Figure 5-6. Custom CEM system datalogger.
5.3.10.5 Backup Data Storage
The DAHS computer should be equipped with a
backup data storage device to prevent data loss in
the event of a hard disk failure. Many hardware
options are available including magnetic tape,
removable hard disks, and optical disks. The backup
routine should be automatic and should include a
procedure for verifying the integrity of the backup
copy,
5.3.10.6 Component Failure Analysis
DAHS failure is among the most common causes of
CEM system downtime. Often when a component
fails, excessive time is spent finding a replacement
and then making the software function properly
again. As part of the development of the CEM
system quality assurance program, a set of proce-
dures should be established in the event of a DAHS
component failure. Determine how long obtaining
replacement hardware (I/O card, PLC, datalogger, or
computer) takes. If the lead time necessary for
shipping could lead to a violation of CEM system
downtime limits spare components should be kept
on-hand. If a failure of a programmable component
{e.g., the PLC, datalogger, or computer) occurs, the
software and programming instructions should be
kept up-to-date and in a secure location. The
operators should practice reprogramming the compo-
nents to make sure that the procedure is adequate.
5.3.10.7 Remote Access Within the Plant or
Company
Many DAHSs now are integrated into the plant local
area network (LAN) or company's wide area network
77
-------
(WAN). For example, the plant environmental
coordinator may find monitoring the CEM system
and preparing emission reports easier from his or her
office. Networking of the DAHS should be dis-
cussed with the DAHS vendor at the time of pur-
chase so that communication issues can be re-
solved. Some functions, such as initiating calibra-
tions, editing data summaries, and entering reason
codes, should be restricted with password protec-
tion. Often the DAHS provider will install a modem
so that diagnostics and software upgrades can be
accomplished remotely.
A company may find integration of CEM data from
all remote locations into a central location for review
and planning purposes advantageous. To integrate
results from several locations, the emissions data
must be stored in a format that can be converted
easily or directly downloaded into a central database
(Long, 1995).
5.3.10.8 Remote Access by the Regulatory Agency
Increasingly, state and local agencies are requiring
the installation of remote terminal units (RTUs) that
allow the agency to monitor CEM system status,
source status, and emission data remotely on a real-
time basis or to download emissions data via modem
periodically (Friedlander, 1992). Remote access
usually is accomplished using a separate datalogger
that receives concentration and emission rate data
from the CEM system. Remote reporting allows the
agency to have access to raw data for auditing and
monitoring, but the source still is required to submit
emissions data according to the regular reporting
requirements. Remote terminal units are being used
on a limited number of sources by regulatory agen-
cies in Pennsylvania and New Jersey as well as
several local districts in California.
5,4 Summary
The ideal DAHS for a given application depends on
the complexity of the applicable reporting require-
ments. A simple stripchart recorder or datalogger
may be sufficient for cases where the source is not
required to submit detailed emission monitoring
reports to regulatory agencies and the monitoring
requirements are simple and straightforward. A
computer-based DAHS can be an indispensable tool
when emission reporting requirements include
calculations of CEM data availability, complex data
averaging, and generation of reports for CEM down-
time and excess emission reasons/corrective actions.
A computer-based DAHS can be programmed using
the plant mainframe computer or it can be pur-
chased from the CEM vendor or a DAHS vendor.
The best DAHS is one that allows the user to be
sure of the status of the emission source, control
equipment, and CEM system for the entire reporting
period. The first part of preparing emission reports
is to answer basic questions such as:
» During what periods was the emission
source operating or in a startup/shutdown
condition?
» During what periods was the CEM system
off-line due to equipment failure or quality
assurance and maintenance activities?
• Do any of the reported emission averages
need to be recalculated based on new
CEM or plant operational data?
* What was the reason and corrective action
for each period of CEM downtime or ex-
cess emissions?
If the DAHS incorporates sufficient plant operational
data and CEM status indicators, these questions can
be answered more easily and many labor hours can
be saved. Also, the DAHS should include an editor
that allows the user to input new information and
recalculate emission averages while maintaining the
original unedited data in a separate location.
The selection of a particular DAHS can affect the
flexibility that the user will have in the future. Some
software changes may require reprogrammtng by the
vendor, but many vendors provide software that
allows the user to add analyzers, averaging periods,
alarms, and even additional emission sources with-
out reprogramming by the vendor. The user also
should be cautious about integrating CEM control
functions (calibration valve switching, probe blow-
back, etc.) with the software on the DAHS com-
puter. Combining the control and data acquisition
functions in the same software may make the ability
to switch DAHS vendors at a later time more diffi-
cult and costly. Often CEM system control func-
tions can be performed by a PLC or by the analyzers
themselves while the DAHS computer monitors the
status of the sampling system and records data.
Finally, determining whether a given DAHS software
package will work properly can often be difficult at
the time of purchase. Many DAHS users have been
forced to switch DAHS vendors shortly after pur-
chase due to poor software performance. The
purchaser should interview previous users of the
candidate software packages to evaluate the long-
term reliability and utility of the software.
78
-------
References
Baranowski, S.T., Gulp, L.R., Jonas, T.S., Williams,
J.L., and Menge, J.F, 1995. The Design, Develop-
ment, and Implementation of a Data Acquisition and
Handling System at Basin Electric Cooperative, "Acid
Rain & Electric Utilities: Permits, Allowances,
Monitoring & Meteorology," Air & Waste Manage-
ment Association Specialty Conference, Tempe, AZ,
pp. 741-747.
Friedlander, D., Sikorsky, D. 1992. The Santa
Barbara County APCD Data Acquisition System,
"Continuous Emission Monitoring: A Technology for
the 90s," Air &. Waste Management Association,
Pittsburgh, PA, pp. 186-194.
Haberland, J.E. 1995. CEM Data Acquisition and
Handling Systems: Updated Experience of the
Utility Industry, "Acid Rain & Electric Utilities:
Permits, Allowances, Monitoring & Meteorology,"
Air & Waste Management Association Specialty
Conference, Tempe, AZ, pp. 751-758.
Long, A., Patel, D. 1995. Opening CEM Vendor
Data-bases, "Acid Rain & Electric Utilities: Permits,
Allowances, Monitoring & Meteorology," Air &
Waste Management Association Specialty Confer-
ence, Tempe, AZ, pp. 748-750.
McCoy, P.G., Schulz, D.A. 1986. "The Use of
CEM Data in Subpart D Enforcement," Continuous
Emissions Monitoring - Advances and Issues, Air &
Waste Management Association Specialty Confer-
ence, Pittsburgh, PA, pp. 159-174.
Paley, L.R. 1984. Technical Guidance on the
Review and Use of Excess Emission Reports, EPA
340/1-84-015.
von Lehmden, D.J., Walsh, G.W. 1989. Appendix
F DARs for GEMS at Subpart D Facilities, "Continu-
ous Emission Monitoring: Present and Future
Applications," Air & Waste Management Association
Specialty Conference, Chicago, II, pp. 103-119,
79
-------
Chapter 6
GEM System Procurement, Installation, and Start-up
Purchasing a CEM system should be a step-wise,
methodical process. As we have discussed in this
handbook, many .options must be considered and
many decisions must be made before a system
actually comes on-line. The manager of the CEM
project must assimilate and understand the
information available on these options, review that
information objectively, and choose the options that
best meet the project objectives. Many pitfalls exist:
one can rely too heavily on the experience of others,
become deluded by aggressive advertising, be led
astray by low cost factors, or enter the process with
decisions already made. As stated earlier, the goal is
to choose the best system for the intended
application. Achieving this goal reflects on both the
technical and the management skills of the project
manager.
Several central factors should be considered in the
program developed to accomplish this goal (White,
1995a). They are:
1!
2)
3)
The design should meet regulatory requirements
and be consistent with plant operating
requirements.
The materials, components, and techniques
should be both reliable and durable under the
constraints of ambient and effluent gas
conditions, and operating conditions.
The system should be easy to use, serviceable,
and cost-effective in its long term operation.
The risks associated with existing and new
technologies applied to monitoring non-criteria
pollutions should be minimized.
These factors apply equally to extractive, in-situ, or
parametric systems. They apply to systems that are
either commercially available or to research systems
under development. Cost is always a constraint, but
in the field of continuous emission monitoring, cost is
not necessarily related to system quality or
performance. Evaluations should first be made on a
technical basis to determine whether the proposed
systems can meet both the regulatory and technical
criteria. Costs should then be normalized between
those acceptable options so that cost comparisons are
made between equivalently performing systems
(Brown, 1992). Nevertheless, some risk will be
associated with applying existing or innovative
techniques to monitoring non-criteria pollutants. If
the technique is inappropriate to the application, costs
associated with the initial resource investment and
noncompliance with monitoring requirements must be
considered. These issues will be discussed in more
detail later in the chapter.
The selection of a monitoring system should be
conducted in a systematic manner (Kopecky,
1979), following established project management
procedures in place at the company. Although these
procedures may differ from company to company,
typical CEM evaluation programs follow approximately
ten basic steps. These steps are:
1) Defining the project scope
2) Reviewing the regulations and process
requirements
3) Assessing the site
4) Reviewing monitoring options
5) Evaluating vendor options
6) Preparing and transmitting a request for
proposal
7) Reviewing bids and awarding contracts
8) Installing the system
9] Approving (certifying) the system
10) Implementing a QA/QC program and operating
the system
An example flow diagram that incorporates these
steps is given in Figure 6-1.
This process can be lengthy. One author estimated
that a typical CEM project will take approximately 40
weeks from inception to certification (Passmore,
1991), Each of these steps will be discussed in detail
in the following sections.
6.1 Defining the Project Scope
A CEM project most commonly begins with the
assignment of a project manager, usually a project
engineer, to develop a technical specification for the
80
-------
Assign GEM,
project rasponsiil
Individual or team
Deffne"pro)§et'SDdpe
• Goals
• Budget limitations
• Procurement criteria
• Establish schedule
Review
regulations,
permit - monitoring
requirements
Review potential
process/control
application for
monitoring data
Assess site specific eohlttons- *
I]
• Access & location
• Flue gas conditions
• Ambient conditions
• Personnel capabilities
Review options
Conduct supplier survey
Commercial
systems & analyzers
Commercial
systems & analyzers
£valuitet.potenlfe! bidders
1 Evaluate system/analyzer
performance parameters
• Contact users for opinions/experience
1 Conduct preliminary field test
(if possible)
Figure 6-1. Example flow diagram for CEM system selection and implementation.
81
-------
(continued)
i
Make tentative choice on
system type or keep
multiple options open
Write technical
specification
Packaged {turn-key) systems
or
Multiple subsystem specs
Send out request for
proposal to pre-selected vendors
• Adherence to specifications
• System/component quality
• Past performance
• In-house review
• External consultant
Risk evaluation
' Past performance
1 Potential performance
• Strength of guarantees
Costs
• Capital
• Operating/maintenance
1 Intangibles
Award contract(s)
Conduct
tests
Install,
• Access / platforms / parts
1 OEM hardware
•DAS
Debug system
Develop/finalize daily
calibration routine
Approve system
Performance
specification test
(where applicable)
Method 301
or alternative
approval criteria
I
Research study
evaluation
(where applicable)
Develop QA/QC plan
10
Provide operator training
Implement QA/QC plan
OPERATE I
Figure 6-1. Continued
82
-------
proposed system. Today, most project managers will
assemble a team to assist in reviewing regulations
and technology, preparing the technical specification,
and evaluating the bids received. Instrumentation
and control supervisors, technicians, and other plant
personnel can be invaluable contributors to this
process. The team should also include purchasing
personnel, environmental personnel, and, where
necessary, legal counsel.
The project scope should be defined at an initial
meeting. In this meeting, a decision should be made
whether the CEIV1 system will be used solely to meet
agency requirements or if it also will be used as input
for plant operations control. The implementing rules,
the required pollutant measurements, and the data
acquisition and reporting requirements should be
discussed so that the project team will have an
understanding of the purposes of the project. The
project manager should have gained enough initial
understanding of source monitoring options so that
the advantages and disadvantages of extractive, in-
situ, or parameter monitoring can be discussed. The
estimated budget and procurement criteria as well as
the project schedule also should be discussed at this
time.
6.2 Reviewing the Regulations and the
Process
After the planning stage, the monitoring regulations
should be reviewed in detail. This is important to
establish the system design criteria for the technical
specification. Particular points to focus on are:
• Monitors required, based on how the data
are to be reported (see Appendix E of this
manual).
» Are parameter surrogate or predictive sys-
tems allowed?
» Are time-shared systems allowed?
• Daily calibration techniques allowed (e.g.,
are Protocol 1 cylinder gases required or are
reference spectra or gas cells allowed?).
• Instrument span (range) requirements.
» Sampling, analyzing, and recording frequen-
cies.
» Drift, relative accuracy, and availability
requirements.
* Recording and reporting requirements
The requirement for a CEM system may have come
through a permit, a state rule, or a federal rule. In
some cases more than one rule may require a CEM
system. Conflicts between specifications may arise
from multiple monitoring requirements. Such con-
flicts should be resolved with the regulating agency
before preparing the request for proposal. Also,
many states have developed CEM guideline docu-
ments (Nazzaro, 1986, Seidman, 1990). These
incorporate essentially de-facto requirements that are
often referenced to in the permit. The state should
be contacted early in the project to determine if such
guidelines have been published and if they must be
met by the installed system.
The applicability of a CEM system to both monitor
process operations and to provide data for operations
control should also be investigated at this time. For
example, many companies are now using NOX moni-
toring data to allow operators to adjust load in
accordance with the plant NOX control program. VOC
monitors may be used to track the efficiency of a
catalytic converter or to monitor loss of product in
process operations. Predictive emission monitoring
systems have proven extremely useful to monitor;
and control many chemical process operations often
the regulatory emissions data are merely an extra
benefit from the model. However, one must be
careful in designing a system for both regulatory and
process application. The process monitoring require-
ments may be either more or less stringent than the
regulatory requirements and finding a system to
satisfy both at reasonable cost may be difficult.
6,3 Assessing the Site
The plant site and flue gas characteristics should be
reviewed next. A suitable location for the CEM
system already may have been identified; however,
its advantages and disadvantages should be exam-
ined carefully. The EPA has established siting criteria
for monitoring systems that are very flexible (USEPA,
1996). The intended location(s) should be evaluated
for the following:
• Can a representative sample or measure-
ment of the actual flue gas emissions be
obtained at the location?
* Is the site accessible?
* Are ambient and physical conditions at the
site suitable for monitoring instrumentation?
83
-------
Sample representativeness is often the biggest Issue
in monitor siting (EPR1, 1993, Jahnke, 1994). One
problem is that flue gas can be stratified, having high
and low pollutant concentrations over different areas
of the measurement cross-section. Another problem
is the presence of non-parallel or swirling (cyclonic)
flow, which affects the measurement of flue gas
velocity. These problems can be identified by con-
ducting a stratification study, measuring gas con-
centrations and flow angles over the cross-section at
the tentative monitoring location. If the gas is found
to be stratified and the location is still viewed as the
best available, sampling options will be affected. !n-
situ path monitors or multiple point sampling systems
may be necessary to obtain averaged measurements
over the cross-section (Jahnke, 1994).
Flue gas characteristics should be either measured or
estimated. This information may be obtained from
prior stack test reports and/or from process data.
Particularly important for the instrument vendor and
the system design are:
* Flue gas temperature and static pressure
* Pollutant gas concentrations (average and
range during upset, or other conditions)
• Moisture percentage/presence of water
droplets
* Particulate loading and paniculate or precipi-
tate carry-over at upset conditions
» Flue gas velocity
Hot metal stacks or flues, excessive vibration, weep-
ing brick stacks after wet scrubbers, should also be
noted.
Ambient conditions at the candidate locations should
be evaluated. The effects of cold or hot weather,
pressure variation at high elevations, sunlight, light-
ning, entrained dust, or duct gas leaks or stack
down-wash on exposed instrumentation and probes
should be examined. These issues are critical in the
continuing performance of in-situ systems, but can be
less of a problem for extractive system instrumenta-
tion housed in environmentally controlled shelters.
Another factor that should be considered in the
choice of systems is the manpower capability at the
facility. Personnel will be required to perform preven-
tive and corrective maintenance on the system after
it is certified. Manpower at appropriate skill levels
must be available to meet these maintenance de-
mands if high system availability is to be achieved.
6.4 Reviewing Monitoring Options
Sufficient information should be available by this part
of the process to begin an evaluation of the various
monitoring options. One of the first decision points
will be whether to focus on extractive, in-situ, or
parameter monitoring systems. Each has its own
advantages and disadvantages - again, one must
remember that the evaluation should concentrate on
determining the best system for the application. This
chapter contains guidance for that evaluation, provid-
ing flow charts and comments on the application
specific features of various systems designs. How-
ever, many exceptions also exist. Due to the practi-
cally infinite permutations between process units,
control devices, flue gas characteristics, and person-
nel resources, no fool-proof scheme can be developed
for monitor selection. In the end, the merits of each
option must be considered in terms of the application.
In general, extractive systems provide the most
options and are the most flexible in meeting sampling
challenges, particularly for HAPs monitoring. In-situ
systems can provide low maintenance options when
only a few target compounds are required to be
monitored. Parameter monitoring systems can offer
both monitoring and process control capabilities. For
these or other reasons, decision makers may have an
initial interest in one type of system over another.
Upon further examination, a first preference may not
lead to a system suitable in meeting the project
goals. The following sections contain flow diagrams
that can aid in determining the suitability of various
options.
6,4.1 Extractive Systems
Extractive monitoring is the most developed of the
three monitoring techniques. It can be a "brute-
force" approach to monitoring in the sense that
extractive system components can be modified or
replaced until some combination of hardware and
operating conditions is found that can deliver a viable
sample to an analyzer. This is where most HAPs
monitoring development programs start. As more
experience is gained in extracting and measuring
specific compounds, the systems may be simplified.
Figure 6-2 is an example flow diagram for selecting
between the various extractive system options.
6.4.1.1 Basic Issues
In the first gate for entry into extractive systems, one
should consider three basic issues. If an interest
exists in time-sharing one set of analyzers among
84
-------
EXTRACTIVi
, NO
p
Consider
in-situ
Consider
parameter
method
i
L
Time-share
analyzers?
Multiple Gases?
Maintenance
requirements
OK?
Will
reactions
be minimized
by keeping
gas hot?
Will
reactions
be minimized
by dilution?
Are
the target
compounds
eactive?
to non-reactive
Is the target
compound soluable?
&
Does the effluent
contain condensable
olsture levels?
Are
analyzers
available
in dilution
range?
Are
nalyzer
available in
the dilution
range?
Can
analyzer
be operated
hot?
Muracompound
analyzers available?
Equal accuracy
for all gases?
Multicompound
* No options
remaining
Gases
Metals
FTIR
GC-MS
GFC
Diode Array
DOAS
ICP
XRF
LASS
Single compound
analyzers
Gases
Particles
IR
UV
Ion Mobility
FID (THC)
Solid State
P- Gauge
Figure 6-2, Selection considerations for extractive systems.
85
-------
several units, extractive systems, not in-situ, are
appropriate. If the user requires or desires to monitor
several gas species, more multi-gas analyzers are
available commercially for extractive systems than for
in-situ systems. Lastly, extractive systems require
more maintenance than either in-situ or parametric
systems. These basic issues should be considered
when choosing the system. A summary of other
technical issues discussed in more detail in Chapter
3 follows.
6.4.1.2 Reactive Gases/Condensable Gases
The tendency of target gases to react or decompose
before analysis is a primary consideration when
selecting a monitoring approach. Two gases may
coexist in the flue gas, but when extracted and
cooled they may react. The resultant compounds
may not be representative of the target pollutants
and, in some cases, may form a precipitate that plugs
the system. An organic compound may react with
the oxygen in the dilution air of a dilution system, or
a compound may catalytically decompose on the
probe, tubing, or pump surfaces. Other compounds
may be unstable and decompose merely on standing.
Where the sampling system perturbs the gases or
particles being measured, a representative sample
may not be obtained and other monitoring techniques
may need to be considered.
The next consideration should be whether the gases
being measured are condensable or soluble. Gases
like hydrochloric acid, ammonia, formaldehyde, and
rnethanol will drop out along with water in any chiller
of a dry-extractive system. Other gases with varying
water solubilities may be partially lost to the conden-
sate. If the gases are not condensable, a dry-extrac-
tive system would offer a greater range of options in
selecting analytical techniques. Of course, dilution
systems can be used for non-condensable gases, but
the choice of analyzers may be limited to those that
are sensitive enough to measure at the lower, diluted
concentrations. If the gases are condensable, hot-
wet source level systems, permeation dryers, or dilu-
tion systems can be used. Dilution systems are
preferable here if analytical techniques are available
at low ranges. Hot-wet systems require greater
maintenance, since parts and materials degrade faster
at elevated temperatures. Nafion permeation driers
will lose some polar compounds such as ammonia,
alcohols, and organic acids, but are acceptable for
others, such as the halogenic acids, inorganic acids,
and aldehydes (1995, Permapure).
6.4.1.3 Multi-gas vs Dedicated Analyzers and
Analytical Methods
The next decision addresses the type of analytical
system to be used. If a number of gas species are to
be analyzed, multi-component analyzers such as gas-
chromatographs, gas chromatography coupled with
mass spectrometry (GCMS), or FTIRs may be consid-
ered. However, one should evaluate the sensitivity of
the analyzer to each of the species analyzed. The
analyzer may be able to measure one gas to a 1 ppm
level, but other gases only to a 10 ppm level. A
separate analyzer may be required to measure the
diluent gases
-------
The analytical techniques have been discussed in
greater detail in Chapter 3 of this handbook. At this
point, an extension of the flow chart is difficult since
many factors are involved in choosing the analytical
method. One must evaluate the instrument response
time, drift, and precision specifications, interference
rejection capability, stage of development, rugged-
ness, cost, and so on. Some familiarity must be
gained with the technologies used to monitor toxic
gases and Chapter 3 should be reviewed.
6.4,2 In-situ Systems
In-situ systems were developed as an alternative to
extractive systems where maintenance levels became
unacceptable. Plugging, leaks, and corrosion eventu-
ally will occur in even the best designed extractive
system, but often these problems can be avoided if
the gas is measured without extracting it in the first
place, !n-situ instruments that require only 50 to 60
hrs of preventive maintenance a year are available
commercially (Karpinsky, 1995). Unfortunately, at
this point, these instruments have been developed
primarily for the monitoring of criteria pollutants.
Therefore, the first question that must be asked is
whether an analyzer has been developed that can
measure the pollutants of interest. Figure 6-3 is an
example flow diagram that illustrates points to be
considered for decision-making for in-situ systems,
6.4.2.1 Siting Considerations
If a pollutant can be monitored by the in-situ tech-
nique, some site specific questions need to be ad-
dressed before proceeding. The first and most
obvious question is whether an interest exists in
timesharing the system between multiple ducts or
stacks. If so, go back to extractive systems. Time-
sharing a spectrometer might be feasible in path
systems using fiber-optic cables, but no known
applications of this technique exist. Other barriers to
the use of in-situ systems pertain to the capability of
the monitor to operate at the chosen duct or stack
location.
Siting for in-situ systems is important since limits to
environmental conditions are encountered that even
the best of systems cannot withstand. Exposure to
temperature extremes may limit an instrument's
application. Although the instrument may have an
internal heater to maintain temperature if it gets too
cold, the instrument may not have an internal cooler
to reduce the temperature if it gets too hot. Light-
ning can be a problem. An exposed in-situ monitor
mid-way up a stack makes an excellent lightning rod.
Excessive vibration also may be a problem. This
factor often is pointed out as a disadvantage in in-
situ systems, however, techniques have been devel-
oped to minimize both the effects of vibration and
lightning on in-situ instrument systems.
Flue gas characteristics certainly limit the application
of in-situ systems. High gas temperatures !>500 °F)
affect molecular infrared light absorption characteris-
tics. Hot-metal stacks will emit infrared radiation and
reduce the sensitivity of in-situ infrared analyzers. In
addition, cycling stack temperatures can distort
untempered probes, or cause misalignments in poorly
designed path systems. Electro-optical in-situ analyz-
ers require a certain amount of light to reach the
detector after it traverses its optical path in the
stack. If particuiate levels are high, or if high levels of
water droplets or other aerosols are present, the
intensity of the light may be too weak to afford good
signal to noise levels. Opacity levels common to
most flue gas exhausts «20% opacity) do not
present a problem for most in-situ monitoring applica-
tions.
The monitoring location for the in-situ system must
be accessible. Duct or stack locations accessible by
catwalks or stairs are ideal. Systems located in the
annulus between chimney and flue and accessible by
man-lift have traditionally been preferred but may
now require confined space entrance permits due to
OSHA regulations. Sites accessible only by ladder,
requiring a safety harness, are not ideal. The rule-of-
thumb is: if no one is willing to go up and service the
system in the middle of winter, find another location
or purchase some other system.
6.4,2.2 Flue Gas Stratification Issues
Flue gas stratification can be a major problem in
obtaining representative flue gas measurements.
Stratification is the uneven distribution of pollutant,
diluent, or particuiate concentrations over the area of
the sampling cross-section. Air in-leakage, combining
ductwork, mixing or combining effluent streams, duct
geometry, and flue gas physical properties can cause
pollutants in the gas to stratify. The flue gas velocity
can also vary over this cross-section; in the case of
non-parallel cyclonic (swirling flow), its direction can
vary also. Cyclonic flow presents one of the most
challenging sampling problems. Selecting an alternate
sampling site is usually more prudent than attempting
to measure flow under cyclonic conditions.
Three approaches can be taken to minimize stratifi-
cation effects. One is to install a path in-situ system,
where a measurement is made on a line across the
stack. The measurement represents a line average,
not an area average, but the average measurement
87
-------
1. Are monitors
developed to measure
target gases in-situ?
2, Is the monitoring location
accessible?
3, Are dedicated monitors
preferred over
lime-shared
systems?
Is gas
stratified?
Is
optical pa
long enough
for sensitivity
squired
rule requires
gas calibration,
does system
have flow-through
gas cell?
UV - Second Derivltlve
DOAS - UV, IR
Electrocatalylic
Electrochemical
DOAS - IR, UV
GFC (IR)
Transrmssometry
Back-scattering
contact charge-transfer
Differential absorption
Thermal
Figure 6-3. Selection considerations for in-situ systems.
88
-------
can be more representative than a single point
measurement under varying stratification conditions.
Another approach is to use multiple probes or moni-
tors to sample at several points on the cross-section.
This technique has been used in monitoring stratified
flow in ductwork by placing a grid of thermal sensors
at measurement points defined by EPA Method 2,
Sometimes, increasing the number of sampling points
from one to just two or three can greatly improve the
representativeness of the measurement.
In situations where the stratification pattern is stable
over varying process load or operating conditions, a
single point in-situ monitor or extractive system probe
may be suitable if a point representing the average
cross-sectional concentration or velocity can be
found. Detailed solutions to these problems are
treated elsewhere (Jahnke, 1994 Chapter 2 and
references therein).
6.4,2.3 Path in-situ Monitor Considerations
If the decision-making process has led this far to path
in-situ monitors, two other factors must be consid-
ered. First, for small diameter ducts or stacks, the
sensitivity of the instrument may be limited. At low
concentrations and short pathlengths, too few
pollutant molecules will be present to absorb much
light energy. The instrument detector may not be
sensitive enough to see very small changes in the
returning light energy and the sensitivity of the
instrument will suffer. Longer path-length options
might be considered, such as measuring across the
stack at an angle (Reuter-Stokes, 1995!, measuring
lengthwise down the duct or stack, or even pulling
the flue gas into a longer by-pass tube to make a
hybrid in-situ/extractive system.
The second important issue regarding path in-situ
systems is a regulatory one. in the United States,
many CEM regulations require the use of calibration
gases for daily calibration
-------
alto* use of
Parameter
systems?
Required
tyNSPS
or permit
Can
tw2
pargmsters
be used as
surrogates?
tea
flret-prfnc!ples\YgS
model
available?
swpte-Bw
aJ mpiissent
ofcp»ratlons?
th,e
process
aeneral^ stable?
Will 3 corretaferi
hold for most
ordinary
conditions?
Figure 6-4. Considerations for parameter monitoring systems.
90
-------
monitoring is also acceptable in the proposed CAM
program.
The simplest approach to using operational parame-
ters in emission monitoring is as a surrogate for the
actual emissions. If a clear relationship can be
established between emissions and one or two
parameters such as pressure drop, temperature, etc.,
a parameter value (the DCPL - direct compliance para-
meter limit) may be acceptable as a de facto emis-
sions standard. If this is not acceptable in meeting
agency requirements, then a predictive model may
be developed.
As discussed in Chapter 4, two types of predictive
models are used; the first principles or phenomeno-
logical models and statistical models. Both models
may provide additional insights into process opera-
tions. Models should be tested and verified. If they
fail verification, one may always return to CEM
analytical hardware.
Simple statistical models can be developed in-plant,
by consultants, or companies that specialize in their
development. If the model is to be used for compli-
ance with short term emission exceedances, exces-
sive correlation testing may be required to model all
the conditions under which an exceedance might
occur. This testing may be very difficult to perform
or prohibitively expensive. If, however, longer term
averages are to be reported or the data will be used
to report adherence to emission limitations, the model
may not have to be as robust.
The statistical model should be developed under
varying operating conditions and also evaluated for
the effects of parameters or conditions not included
in the model. The final model should be robust,
having an accurate solution for each set of operating
conditions. Models can cross-check themselves, but
all models should be re-verified periodically.
6,4.4 New Market Products and Systems
under Research and Development
Figures 6-2 and 6-3 are flow diagrams for evaluating
monitoring approaches that are, in general, commer-
cially available, or where some previous similar
experience exists in their application. In evaluating
new market products or newly developed techniques,
following such a diagram may be difficult if little is
known about the system limitations. For such new
systems, an alternative flow diagram is suggested,
Figure 6-5.
In this approach, the pollutant properties are consid-
ered first, the analyzer capabilities are evaluated, and
then the sampling system options are reviewed. This
evaluation approach proceeds from the analyzer to
system selection rather than the approach given in
Figures 6-2 and 6-3, from the system type to the
available analyzers,
6.5 Evaluating Potential Bidders
After the site characteristics and monitoring options
have been evaluated, information should be obtained
on potential suppliers and bidders for the project.
Potential bidders may include CEM system integra-
tors, instrument manufacturers, data acquisition
system (DAS) vendors, computer software firms, and
environmental consulting/contracting companies.
Many of these companies will exhibit at trade shows
and advertise in trade journals. Lists of instrumenta-
tion companies and CEM system integrators are given
in trade journal "Buyer's Guides," but these lists
generally are not very discriminating.
Once contacted, vendors will send literature, call,
visit, and market. Although marketing calls may
seem onerous, they can provide a means for assess-
ing both the applicability of the vendor's approach
and the capabilities of the vendor. On the other
hand, the vendor who understands the requirements
and limitations early in the proposal process, can
deliver a better product in the end.
At this point in the monitoring system selection
process, the project manager and team should
concentrate on screening the assembled list of
potential vendors to those that meet project
selection criteria (Brown, 1992). These initial criteria
should include an evaluation of:
» System capability in meeting regulatory
requirements
» System capability in meeting plant
process/control needs
» Suitability of the system for operation at
the plant, at the intended location, and
under plant operating conditions.
Contractual requirements such as bid bonds, warran-
ties, guarantees, and other commercial aspects
should be discussed before a formal bid solicitation.
Preliminary cost estimates, maintenance costs, and
vendor service capabilities also may be important
screening issues.
To help in this part of the process, the team should
contact other users of monitoring systems, preferably
users in the same or similar industries. Here, the
91
-------
Determine gaseous compounds
required to be monitored
by regulation, permit, etc.
Identify potential proven analytic techniques
for each compound
Compound Technique
1. NDIR, UV, FTiR
2. GFCIR, FTIR, GCMS
3. IMS, UV, UVDOAS
If appropriate, consider analytic techniques
with multi-component analysis capabilities
(GFCIR, FTIR, DOAS, GCMS...)
Identify research level
analytic techniques
List commercially available sampling systems:
Extractive In situ
List research sampling systems:
Extractive In situ
Select sampling system compatible with analytic
technique and effluent matrix conditions:
- Hot-wet extractive
- Cold extractive
- Dilution extractive
- In situ- point or path
NO
NO
Consider target
compounds and effluent
'matrix. Can the system/analyzer handle:"'
- Reactive gases?
- Condensible components?^
- Soluble analytes?
Can the
system/technique
handle constraints imposed 1
- Measurement range?
- Gas density, temperature, pressure?
- Reference spectra, temperature
compensation?
^ Moisture, COz, other
interterents?
i YES
"Can the"
system meet:
• Response timefcycle time
•Requirements? (response times < 1 mirT
may necessitate an in situ analyzer)
- Regulatory calibration requirements and constraints^,
- Are gas ceils or filters allowed?
- Are calibration gases required?,
s^_Are calibration gases
available?
NO
NO
NO
Figure 6-5. Alternative approach to monitoring system decision-making.
92
-------
industry network becomes useful. The opinions and
experience of others can be extremely helpful in
maintaining objectivity when evaluating the claims of
the various suppliers. However, as a cautionary note,
be aware that the experience of others is based on
specific applications and may not be appropriate to the
current project. Industry networks can fool themselves
by reaching an overall consensus - everyone agreeing
that only one specific approach is the correct ap-
proach. A project manager conducting a more objec-
tive evaluation may end up in the position of deviating
from the "conventional wisdom," and find difficulty in
justifying a different, but more appropriate technology
to management.
When a corporate office is planning to purchase a
number of monitoring systems for company facilities,
management often desires to standardize, having one
vendor supply identical systems to all of the plants.
However, prudence should be shown by first purchas-
ing one system, evaluating the system technically, and
evaluating the performance of the vendor in terms of
response and service before committing to a corporate-
wide purchase. Another approach is to invite vendors
to demonstrate equipment during a short-term field
trial. A field trial will be particularly appropriate if the
source conditions at the trial are similar to those of the
intended application. Vendors are becoming hesitant
in participating in such programs since they are costly,
do not necessarily reflect actual equipment perfor-
mance, and may offer no return if the system is not
selected. However, if an instrument manufacturer has
developed a new analytical method or monitoring
technique, it must be field demonstrated before it can
hope to have any market acceptance. The manufac-
turer will be eager to obtain the cooperation of a plant
to help in proving the system. Many good system
applications have resulted from such developmental
work.
The screening process should identify at least three,
but preferably more, potential bidders. If company
policy is to accept the lowest bidder, then only ven-
dors that can provide a product that the plant can live
with should be put on the bidder's list. If the company
requires that all requesting or potential bidders be
included on the list and that no prior discrimination is
allowed {as is the case in some municipal facilities),
then this entire exercise is useless.
6.6 Writing the Bid Specifications
The purpose of performing an initial vendor screening
is to simplify the writing of the technical specification.
As a result of the evaluations conducted on the
available technologies (Step 4 of this strategy] the
project manager may have decided initially to focus on
extractive or in-situ systems. However, a vendor may
have later pointed out the merits of predictive monitor-
ing systems, Options still can be kept open in the
technical specification by incorporating a set of uni-
form criteria that either technique can satisfy.
After a consensus has been reached by the GEM
project team, a detailed technical specification can be
prepared. In most cases, the team will not have to
start from ground zero and develop an original docu-
ment since model and example specifications are often
available. The Electric Power Research Institute has
developed a model Request for Proposal package for
CEM systems required under 40 CFR 75, the acid rain
program (EPRI, 1993), The package includes an
example transmittal letter, suggested terms and
conditions, and a guideline technical specification. The
package can be modified easily for other monitoring
programs and serves as a good starting point. Techni-
cal specifications also tend to circulate within the
networks of a given industry and technical specifica-
tion packages frequently can be "borrowed" from
someone else. Vendors also offer bid packages to
"assist" potential clients in their monitoring programs.
However, these packages are usually self-serving,
being written in a manner to exclude competing
technology or competitors. The company itself may
have standard specification requirements and the
monitoring specification may have to be tailored to
meet the criteria of the contracts department. Which-
ever route is chosen, the specification should provide
a basis for the purchaser to obtain the system desired
and a basis for legal action if it is not satisfactory after
installation.
A typical CEM system specification incorporates the
following (from Jahnke, 1993):
1) Purpose. A brief statement of where the CEM
system will be installed, the number of units that
will be monitored, and statement of the regulatory
requirements applicable to the installation.
2) Scope of Work. An outline of hardware, soft-
ware, and services to be provided by the vendor.
This section may include a basic system configu-
ration, list of the number of analyzers required,
data acquisition/control requirements, and may
specify brand-names of analyzers or components
if desired. Vendor furnished services may include
complete system engineering, installation, and
start-up, If desired.
3) Equipment and Services Provided by Others. A
listing of equipment and services that the vendor
is not expected to supply. This may include
93
-------
equipment or supplies such as elevators, ports,
catwalks, platforms, electrical supplies, founda-
tions, or calibration gases. Services supplied by
plant personnel or others may include system
installation, wiring, or certification.
4) Description of Operating Conditions.. A descrip-
tion of environmental and stack gas conditions at
the sampling locations. Diagrams of sampling
ports and access conditions should be provided
here or referred from here to the appendix of the
specification. Flue gas characteristics such as
moisture content, velocity, and temperature, and
the expected composition and concentrations of
pollutants in the flue gas should be supplied.
This information is critical to the vendor for the
design of the system.
5} Design Criteria and Construction. A detailed
description of the system on which the bid is to
be prepared. The intent is not to provide al!
design data, but to provide the vendor with an
understanding of the system requirements from
both regulatory and operational aspects. Design
requirements include adherence to standards,
codes, and regulations. They also include
specifications for instrument range, drift, and
response time. They may include specifications
for sample conditioning, interfacing with other
plant systems, and data acquisition requirements
and reporting formats.
This section will constitute the bulk of the speci-
fication. However, care must be taken not to
"over-specify" the system. The vendor must be
allowed leeway in the design to use his own
experience in CEIV1 systems for the job. If the
requirements are too stringent, either no one will
bid on the system or they will be ignored in the
systems offered.
6) Vendor Furnished Services. A listing and descrip-
tion of services desired from the vendor. These
may include, total project management, installa-
tion, training, performance testing, or on-going
maintenance services.
7) Inspection and Testing. A listing of certification
guarantees and warranties expected from vendor.
These may include factory checkout and certifica-
tion provisions, performance specification test
guarantees, and system availability requirements.
8} Equipment Delivery Requirements. A statement
of progress report requirements, delivery dates,
and shipping requirements.
9} Engineering Data and Documentation. A listing
of required system documentation. This should
include accurate system schematics and wiring
diagrams, operating manuals, maintenance in-
structions, and DAS operating instructions and
documentation.
Separate technical specifications may be written for
various subsystems. Frequently, a separate specifica-
tion is written for the data acquisition and handling
system. Special precautions are necessary to ensure
that the two systems are fully compatible (see Chapter
5).
Services such as training, maintenance, and perfor-
mance testing should be requested to be bid separately
as options. Costs for these services should also be bid
separately since they tend to be quite variable between
bidders. Separating services from the system technical
specifications will assist in the evaluation process and
will offer more flexibility in subsequent negotiations.
A set of technical specifications normally will be
accompanied by a set of "Standard Terms and Condi-
tions" prepared by the company contracts department.
This document will include legal requirements for
insurance, limits of liability, remedies, disputes, etc.
After a draft bid package has been prepared, it should
be circulated to other departments within the company
before it is released (Brown, 1992). Legal depart-
ments, computer systems, purchasing, and various
engineering departments may wish to provide input or
comment on areas of their particular concern. The
package should then be revised appropriately, finalized
and issued. If not already having done so, bidders
should be allowed the opportunity to visit the site and
observe plant conditions and operations to better
design a proposed system. Approximately four weeks
should be allowed for bidders to respond to the Re-
quest for Proposaf.
6.7 Reviewing the Proposals and Making
a Decision
Reviewing proposals objectively is difficult since, by
the time proposals are received, members of the CEIV1
team will have some preconceived idea about what will
be the "best" system or who will be the "best" vendor.
However, the decision should be based on a system-
atic evaluation of the proposal following objective
criteria established before the RFP, keeping in mind
that the quality of the proposal will often be indicative
of the quality of the work.
94
-------
6,7.1 Areas for Review
Three principle areas need to be reviewed on a prelimi-
nary basis after receipt of the proposals (Figure 6-6).
These areas include a technical review, an evaluation
of risk (especially for new systems for new applica-
tions), and cost. This review is preliminary because,
despite all good intentions in drafting clear, concise,
and explicit technical specifications, one vendor will
offer oranges and another will offer apples. One
usually must obtain further information, clarify excep-
tions, and normalize equipment costs and services to
make valid comparisons,
6.7.1,1 Technical Evaluation
In the technical review, a determination should be
made whether the system will meet all regulatory
specifications and any additional specifications given
for the plant application. Most proposals will state
that all specifications are met; however, corroborating
information should be provided so that the review team
may verify such statements.
Technical Review
Adherence to Technical Specs.
System/Component Quality
Proposal Quality
Risk Evaluation
Past Performance
Potential Performance
Strength of Guarantees
Costs
Capital
Maintenance
Intangibles
Figure 6-6. Areas for proposal review,
The system and component quality should be evalu-
ated. A proper evaluation here may require prior
knowledge on the part of the team members, or a
reliance on the experience of others who have pur-
chased similar systems or instrumentation. System
quality may be difficult to evaluate for newly devel-
oped analyzers. The frequent lack of experience of
entrepreneurial instrumentation companies in stack
monitoring applications is a factor that must be consid-
ered in the review.
The best proposals are those that are written specifi-
cally for the plant application. The worst consist of a
price quote and a collection of vendor trade literature.
Most proposals will fall somewhere in between these
extremes, generally towards the lower end of quality.
Not that better proposals can't be written, but when a
vendor sales or project manager has to respond to
numerous proposals, few receive the attention that is
desired. However, if a vendor really wants the job,
enough care and time will be put into the proposal to
show that a good system can be provided for the
application. Ultimately, well-prepared proposals have
greater success in being awarded.
6.7.1.2 Risk Evaluation
The evaluation of risk in a CEM program is not simple.
Frequently, the results of important decisions are not
known until much later in the implementation of a
monitoring program. An untested assumption or a
faulty decision early in the development of a monitor-
ing program may affect many subsequent steps. The
project manager must recognize the risk associated
with critical decisions and understand the impact of
unfavorable outcomes on both the immediate task and
subsequent activities.
The primary risk associated with selection of a moni-
toring approach or specific equipment is that the
selected system will not perform acceptably. When a
genera! failure occurs, the investment of resources
(time for internal personnel, subcontracted services,
and capital expenses) produces neither benefit nor
compliance with applicable monitoring requirements.
An entire system may need to be replaced or new
regulatory requirements may need to be negotiated.
The project manager must focus on the specific issues
of the particular monitoring application. Past perfor-
mance of similar systems may provide a limited basis
for evaluation of risk. Contractual conditions, warran-
tees, and performance guarantees can draw attention
to specific issues and may also provide some recourse
for the purchaser if performance proves to be unac-
ceptable. However, to minimize risk of unacceptable
performance is always better than to rely on the threat
of litigation.
Because of the importance of CEM data to source
compliance programs, the CEM project manager must
consider how inaccurate, imprecise, or non-representa-
tive monitoring data affects source operation and
control equipment costs. The risks and costs associ-
95
-------
ated with inaccurate data must be weighed against the
cost for more reliable monitoring results. Achieving
the appropriate balance between these risks and costs
becomes more crucial as emission levels approach the
applicable limit. One may expect that a source operat-
ing well within the applicable limits will be willing to
accept a greater level of uncertainty in the monitoring
data. In this case, the source operator may be able to
choose more economical alternatives over more
accurate data. On the other hand, a source with a
minimal compliance margin will place greater value on
accurate monitoring data and may be able to justify the
greater expense associated with achieving higher levels
of performance. Costs associated with modification of
process or control equipment and operational costs due
to changes in source operation and maintenance
practices necessary to meet emission limitations are
considered "compliance costs" for the applicable
standard. Such costs should not be attributed to the
monitoring program. Similarly, the risks associated
with non-compliance with emission standards aiso
must be considered separately even though such
problems might be detected only by means of the
monitoring program.
The proper balance between risk and cost will depend
on the monitoring approach and many source-specific
factors. This balance must be determined on a case-
by-case basis. Two examples of risk evaluation are
provided below:
6.7.1.2.1 Example 1 - PEM System Risk Assessment.
Consider the selection of a PEM system relative to a
conventional CEM system for the measurement of
gaseous pollutants. In this approach, hardware ex-
penses are mostly eliminated and the initial capital
outlay for the PEM system may approximate the cost
of the CEM system DAHS. As compared to the CEM
system approach, additional comparative emissions
testing is required to develop either the "first principals
- phenomenological" PEM system or the "statistical -
inferential" PEM system. Thus, an investment occurs
in direct costs (i.e., planning and engineering time for
company personnel or similar subcontract services by
consultants or vendors). PEM system performance
tests involving comparisons with independent emis-
sions measurements may be required soon after the
model is established. A similar test may have to be
repeated on a periodic basis as a QA procedure.
To assess the risk associated with a PEM system
approach, estimates must be made of the initial model
development cost, initial confirmation performance test
cost, and subsequent periodic QA test cost. These are
the expected costs that will be incurred in all PEM
system applications. In addition, the potential costs
associated with unfavorable outcomes must be consid-
ered. The likelihood that the PEM system will fail
either the initial or subsequent tests and the potential
costs of retesting must be assessed. Also, the likeli-
hood that the initial model development effort will need
to be repeated either because of repeated failures of
performance tests or because of fundamental changes
in the fuel, raw materials, process, or control device
function must be considered.
The ultimate cost of the PEM system will be very
sensitive to the risks discussed in the previous para-
graph. The relationship as shown in Figure 6-7 is
suggested as a method of estimating the cost of the
PEM system.
The consistent application of arbitrary estimates or
ranges of the likelihood of various outcomes in the
above relationship will assist the decision maker in
assessing the overall cost for the PEM system. An
example of such an analysis is shown in Figure 6-8.
Prudence dictates assigning higher risk factors to
industries or applications where little previous experi-
ence has accumulated. Higher risk factors should be
assigned to applications within agency jurisdictions
where strict monitoring regulations or QA requirements
may be adopted.
6.7.1.2.2 Example 2 - CEM System Risk Assessment.
The likelihood that a CEM system will perform accept-
ably will depend to a great extent on the pollutants or
compounds to be measured. Opacity, SO2, and NOK
CEM systems are now considered to be very reliable
for many applications. However, the development
status of CEMS for non-criteria pollutants spans a very
wide range. For example, "total hydrocarbon moni-
tors" employing FIDs have been installed at certain
sources in the United States for a number of years.
CEM systems for Hg are available from several manu-
facturers, but issues related to calibration standards
and reference test method procedures are being
resolved. Monitors for other metals (e.g., "multi-
metals CEM systems") are beginning to become
available commercially. The project manager must
recognize that the likelihood of a general failure of the
monitoring system to perform acceptably is very
significant until commercially available systems are
installed and experience is gained at a substantial
number of facilities.
For commercially available CEM systems, risks can be
minimized by careful consideration of each step in the
CEM program implementation. The selection of a CEM
measurement location is a key aspect in the develop-
ment of a monitoring program and is an example of a
96
-------
$LC(N) = I$IM
Where;
P1)$TP][1
$LC!N)
SO&IV)
N
M
P,
P,
P4]
P2)$T
Q&
Equation 6-1
Lifetime cost for N years
Initial model development cost (including software, hardware,
installation, and development test costs)
Performance test cost
Periodic quality assurance test cost
Annual PEM system operation and maintenance costs (exclusive
of periodic quality assurance tests)
Years of expected operation
Number of periodic QA tests required per year
Likelihood of performance test failure (range of 0 to 1)
Likelihood of periodic QA test failure (range of 0 to 1)
Likelihood model must be redeveloped because of model
failure during expected useful life of PEM system (range of 0
to 1)
Likelihood that a new model must be developed because of
fuel, process, or control equipment changes during expected
useful life of PEM system (range of 0 to 1)
Figure 6-7. Cost estimation method for PEM system.
PEM System Lifetime Cost
o"
T"
o"
11,000,000 -
$900,000
$800,000 i
I
$700,000 -
$600,000 -
$500,000
$400,000 -
$300,000
$200,000
$100,000
$0
— f,
»•* — *^ — m*
-a a-o-o- '
>j-*_j
$LC(N) IS LSFETIIVE
P3 IS LIKELIHOOD
- P4 IS LIKELIHOOD
CONTROL EC
Case 1 Case 2
$50,000 $50,000
$30,000 $10,000
$10,000 $10,000
$25,000 $25,000
10 10
4 4
0.1 0.25
0.05 0.25
I I
P3 IS LIKELIHOOD MODEL MUST BE REDEVELOPED BECAUSE OF MODEL FAILURE
P4 IS LIKELIHOOD NEW MODEL IS REQUIRED BECAUSE OF FUEL, PROCESS, OR
$ft/l is Initial Model Development Cost
$Tp is Performance Test Cost
$Tqa is Periodic Quality Assurance Test Cost
$O&M Is Annual Operating & Maintenance Cost
N is Number of Years
M is Number of Periodic QA Tests per Year
P1 is Likelihood of Performance Test Failure
P2 Is Likelihood of Periodic QATest Failure
0.2
0.4
0.6
0.8 1 1.2
P3 + P4 Likelihood
1.4
1.6
1.8
Figure 6-8. Analysis of PEM system lifetime cost for two example cases.
97
-------
decision that affects many subsequent steps.
Selecting a tentative monitoring location early in the
planning phase of a monitoring program to specify
emission and environmental conditions is very
important for potential vendors. Furthermore, the
costs for access to the monitoring location (ladders,
elevators, platforms, sample ports, etc.) and provid-
ing support utilities may be greater than the capital
cost for the monitoring equipment. Flue gas (or
flow) stratification at a measurement location can
affect the representativeness of monitoring data
substantially. In evaluating a tentative monitoring
location, the following must be considered: 1) the
availability of alternate monitoring locations, 2) the
cost to relocate a monitor if the tentative location is
later found to be unacceptable, 3} the likelihood of
pollutant, diluent, or flow stratification at the loca-
tion and the reliability of indications that it may
exist, 4) the likelihood that a failed relative accuracy
test would have to be repeated and the costs for
retesting, 5) the cost associated with potential
delays in the approval process of the CEM system,
and 6) the practicality and cost associated with
performing a stratification test before finalizing the
location selection or prior to performing the relative
accuracy test. The risks and costs associated with
all of these factors must be balanced during the
planning effort.
Assessment of risk for the initial acceptance test
and subsequent QA checks and audits is also impor-
tant. The cost of the initial acceptance test is
usually substantial. A well planned and executed
test can be a thorough evaluation of CEM system
performance and the adequacy of the calibration
procedure. A poorly performed test may result in
failure to meet specifications, even for the best
designed and implemented monitoring program,
resulting in a costly retest and substantial delays. A
poorly performed test also can allow a flawed CEM
system design or improperly applied CEM system to
be accepted when it should not. This outcome
avoids immediate retesting but leaves the facility
with an unreliable monitor. Unfortunately, this
situation may continue for many months until the
underlying problems are identified by subsequent QA
audits or other activities.
In planning the initial and subsequent tests, the
following must be balanced: 1) the level of prepara-
tion and oversight for the test and the associated
costs, 2} the qualifications and capabilities of the
test team, 3) the likelihood that the CEM system will
fail a test and the cost for retesting, 4) the likelihood
that the test will not detect an important perfor-
mance problem affecting either regulatory require-
ments or contractual guarantees, and 5) the cost
associated with potential delays in the approval
process resulting from a failed test,
Other QA risk criteria to consider are: 1) the likeli-
hood or expected frequency of daily drift check
failures, 2) likelihood of periodic QA test failures and
the need for retests, 3) expected and/or guaranteed
CEM availability, 4) availability of replacement parts
and service from the vendor, and 5} the supplier's
response time to provide service or replacement
parts/monitors. These factors will affect the on-
going operating cost for the monitoring program and
they will affect the risk associated with satisfying
minimum data availability requirements of the regula-
tory agency.
Evaluating the risks associated with a CEM system
or PEM system monitoring program requires the
likelihood and impact of unfavorable outcomes of
various decisions and events. Considering the
downside risks can help to focus attention on the
most critical issues and may be helpful in preventing
negative possibilities from turning into actual
experiences.
6.7.1.3 Cost Evaluation
In addition to the initial capital expense for the
purchase of a monitoring system, other start-up
costs are associated with engineering, selection,
purchase, installation, and performance testing of
the monitoring system. Recurring or ongoing costs
for the operation, maintenance, quality assurance,
recordkeeping and reporting associated with the
monitoring program also must be considered. The
costs that will be incurred at a particular facility will
depend on many factors and will differ greatly
between facilities. The pollutants to be monitored,
the type of monitoring approach selected, the
monitor installation location and conditions, and
many other source-specific factors will affect the
cost of the monitoring program.
All of the costs that may be expected should be
taken into account in the cost evaluation of competi-
tive monitoring proposals. In most cases, the
majority of costs that will be encountered over the
life of the monitoring equipment will not be repre-
sented in a monitor equipment vendor's proposal.
Many other external and source-specific factors over
which the vendor or equipment supplier has little
control need to be considered. Nevertheless, all of
the expected costs should be projected and consid-
ered in conjunction with each vendor's cost proposal
and technical approach. This is necessary both 1) to
evaluate different proposals on a consistent basis,
98
-------
and 2) to determine the total cost of the monitoring
program for the affected facility. Much experience
has shown that the proposal offering the lowest
initial purchase price is not necessarily the least cost
approach over the CEM program lifetime,
Costs must be evaluated over the useful life of the
equipment or on another basis that accounts for
both initial and ongoing costs. The average life of a
CEM system is likely to vary between five and ten
years depending on the type of equipment used and
the monitoring application. The distribution of these
costs between capital outlays for initial equipment
purchases, facility modification costs, software pur-
chases, subcontracted support services, and internal
costs for engineering/planning, purchase, operation
and maintenance, reporting, recordkeeping, and
administrative support will vary for each particular
monitoring application. A checklist is provided in
Appendix F which may be helpful in evaluating some
of these costs. The checklist is offered as a guide
to encourage thinking about CEM program costs in
a comprehensive manner.
6.7.2 A Matrix Evaluation Technique
A common technique used in evaluating proposals is
to prepare a matrix of various evaluation factors and
to assign a maximum possible score to each factor.
For example, each listed item in Figure 6-6 could be
used as an evaluation criteria and assigned a maxi-
mum score of 10. The matrix would then be com-
posed of 9 criteria evaluated for each proposal
received (Figure 6-9h
Each returned proposal is scored and the total
scores for each category (technical, risk, cost) are
obtained. Each member of the evaluation team
should complete the matrices for each vendor
independently. Then, they should meet as a com-
mittee and reach a scoring consensus on each
evaluation criteria. Each member should justify his
or her score until agreement is reached. The use of
an outside consultant or CEM expert from another
company to provide an independent, objective
review of the proposals also may be helpful.
6.7.3 Normalizing the Issues
The proposal review may identify the best system
for the application, but frequently many questions
and issues are raised that require clarification. One
vendor may have drawn exception to numerous
specifications, whereas another may have agreed to
all of the specifications. Another may have pro-
posed an inferior conditioning system, but a superior
analyzer. A service contract of one may have
proposed 12 trips in a year including expenses.
TECHNICAL REVIEW
"Adherence to
Technical Specs.
* System/Component
Quality
« Proposal Quality
RISK EVALUATION
* Past Performance
« Potential
Performance
"Strength of
Guarantees
COSTS
• Capital
« Maintenance
» Intangibles
BID1
BID 2
BID 3
BID 4
Figure 6-9. Simplified example matrix evaluation.
99
-------
whereas another offered 8 trips plus expenses.
Another may have a superior system but misunder-
stood the data acquisition system criteria. Often,
another round of evaluation will be necessary to
bring the proposals to a stage where they can be
compared on a uniform basis. This can be achieved
by listing all exceptions and questions and requiring
a response from the vendor suitable for review
within one to two weeks. For large projects, bring-
ing the vendors to the plant to review their propos-
als and clarify outstanding issues may be necessary,
All costs should be normalized so that equivalently
performing systems would be purchased for the
stated quotations. Some costs may have to be
adjusted up or down if equivalent services are not
offered by the vendors. Intangible issues also need
to be considered and factored into the risk rankings.
These intangibles include an assessment of the
stability of the company (its ability to provide parts
and service five years or ten years from the equip-
ment purchase), its ability to honor guarantees, the
likelihood of the company being sold, the probability
of the only person who knows anything leaving, etc,
By this time, the project team should have deter-
mined which proposals are technically acceptable
and have ranked the proposal costs. Depending
upon company policy or pre-established guidelines,
a contract may be awarded to the lowest bidder of
those technically acceptable, or technical/cost
considerations may be weighted. For example, the
project team may agree initially that technical
factors will contribute 80% in the final evaluation
and costs 20%, or vice versa. On the other hand,
after all this analysis has been done, it may be
thrown out and the award decision based on subjec-
tive opinions that a favored vendor is the only
vendor that can develop a successful operating
system for the application. If the latter approach is
taken, company resources have been wasted.
An evaluation can be objective and systematic, or
subjective. An ostensibly systematic evaluation can
also be distorted and biased to achieve a desired
and. However the decision is made, the company
must understand that plant managers, plant engi-
neers, and technicians will have to work with the
system for many years into the future. The choice
of systems will be considered a good choice for all
concerned if it meets the three criteria discussed in
the introduction to this chapter:
1) The design meets regulatory requirements and
is consistent with plant operating requirements.
2} The materials, components, and techniques are
reliable and durable under the constraints of
ambient and flue gas conditions, and operating
conditions.
3} The system is easy to use, serviceable, and
cost-effective in its long term operation.
6,8 Installing the System
After the contract is awarded, construction will
begin on the system. This will take several months,
depending upon parts availability and the vendor's
backlog. Few vendors have systems that can be
provided "off-the-shelf," since most begin construc-
tion after receipt of an order. A typical time to
fabricate and deliver the CEM system can take from
two to four months (Passmore, 1991, Ferguson,
1991, Retis, 1992). Installation and construction
will take four to six weeks, as will start-up and
certification. Thus, at least 8 months lead time
should be planned.
Depending upon the size of the project and the
project budget, conducting a factory acceptance
test (FAT) before the system is shipped may be
desirable. This requires some member of the project
team to visit the vendor and evaluate the perfor-
mance of the system on the shop floor. An excel-
lent checklist has been developed by EPRI {EPRI,
1993) for conducting such an evaluation.
One of the most common problems noted during
factory acceptance tests is that the data acquisition
system software is not completed (Porter, 1990).
Although the system hardware may be fully func-
tional, an essential subsystem basically will be non-
operable. Caught between contract requirements
and regulatory deadlines, this scheduling failure is
often overlooked, with the expectation that the DAS
will be ready sometime during the on-site installa-
tion. Approving the FAT without the data acquisi-
tion system will be the beginning of trouble in the
CEM program. Nonadherence to contractual condi-
tions at this point can lead to further exceptions and
delays during installation.
Installing the system will require the cooperation of
plant personnel, the vendor, and either plant or
subcontracted construction engineers. Platforms
may have to be constructed, ports installed, and
electrical cables run to the sampling site. These
activities may be handled by the plant or the vendor,
although vendors often prefer not to be responsible
for these construction activities. Major problems
can develop at this stage if the construction engi-
neers are not fully informed about the purposes of
100
-------
the CEM system and the vendor's installation re-
quirements, A port installed at a 45° angle may not
align with one on the other side, the analyzer cabi-
net door may bump into the scaffolding hand rail
when opened, or no support is provided for the
sampling line or electrical cables. Adequate commu-
nication will prevent such problems from developing.
The installation of the system will be relatively
straightforward. Debugging the system will not be
straightforward, particularly for newer systems
developed for the measurement of HAPs. Unantici-
pated interferences, electrical stability, and calibra-
tion drift problems are common in the initial stages
of operation. Since the daily calibration check
routines for HAPs monitors may differ from the
traditional usage of calibration gases, the routines
themselves may have to be fine-tuned. A daily
calibration instability of greater than 2.5% of span
may be indicative of a site-specific installation
problem. The problem must be investigated and
resolved. Calibration stability should be achieved
before proceeding with system certification or other
approval procedures.
6.9 Approving/Certifying the System
Certification and approval methods for non-criteria
pollutant monitoring systems are discussed in detail
in Chapter 7 of this manual. The traditional "perfor-
mance specification test" certification procedures
incorporate the "relative accuracy test," which
compares the continuous monitoring measurements
to those obtained using an EPA reference method.
Performance specification test procedures have
proven to be an excellent means for certifying
monitoring systems for criteria pollutants. However,
validated methods exist for only about 40 of the
189 air toxics materials listed under Title 111 of the
Clean Air Act and these for only a very limited range
of sources. This lack of appropriate test methods
limits the application of similar testing procedures
for these monitoring systems.
Other approval mechanisms can be considered by
the environmental control agency. The Method 301
validation criteria (see Chapter 7) are applicable for
short-term test methods, but involve an extensive
validation program on a site-specific basis only,
Alternative methods have been proposed that
incorporate laboratory evaluations with field tests
conducted at representative "challenging" applica-
tions (Peeler, 1996). Proposed laboratory tests
include checks for calibration stability, response
time, interference rejection, and linearity. Field tests
include techniques such as analyte spiking, direct
calibration checks, and sampling system bias tests.
Similar methods have been used by Kinner {Kinner,
1996) for FTIR applications in cement kilns.
In other cases, such as for metals (as vapor or as
particulate matter combined or uncombined), refer-
ence methods may not produce data sufficiently
precise to compare to monitoring system data.
Calibration standards may not be available or may
have to be specially generated. Special spiking
procedures may also have to be developed. The
limited experience in these techniques with certain
materials can transform an intended validation
program into a research study. Attention then must
be paid to all facets of the validation program. The
experimental research design must determine the
representativeness of the sample measured as well
as the precision and accuracy to which it is mea-
sured.
Approval methods also are necessary for parametric
monitoring systems (parameter surrogate or predic-
tive). Where validated reference methods are not
available alternatives must be considered.
6.10 Implementing the QA/QC Plan
and Operating the System
Certification or approval of a monitoring system
provides no guarantees that the system will continue
generating accurate and precise emissions data. An
implemented quality assurance (QA)/quality control
-------
Utilities - Permits, Allowances, Monitoring & Meteo-
rology. Air & Waste Management Association.
Pittsburgh, PA, pp. 589-593.
Carman, T.A., Anadi, R.S. 1993. A Multi-compo-
nent, Across-Slack, NDIR Analyzer Meets On-Line
Cylinder Gas Audit Requirements of a CEMS.
Continuous Emission Monitoring - A Technology for
the 90's. Air & Waste Management Association,
Pittsburgh, PA. pp 373-382.
DeWees, W.G., Steinsberger, and Buynitzky, W.D,
1994. "EPA Method 301 Validation of a Source
Measurement Method for Title III Polar Volatile
Organic Compounds (PVOCs)." Paper presented at
the Air & Waste Management Association Meeting,
Cincinnati, OH. Paper 94-TA27.04.
Electric Power Research Institute. 1983. Con-
tinuous Emission Monitoring Guidelines. Report No.
EPRl CS-3723. Palo Alto, CA.
Electric Power Research Institute. 1988. Continuous
Emission Monitoring Guidelines - Update. Report No.
EPRl CS-5998. Palo Alto, CA.
Electric Power Research Institute. 1993. Continuous
Emission Monitoring Guidelines - 1993 Update,
Volume 2. Report No. EPRl TY-102386-V2. Palo
Alto, CA.
Ferguson, A.W. and Cochran, J.R. 1991.
Continuous Emission Monitoring — "Getting Your Act
Together to Ensure 1990 Clean Air Act
Compliance." Paper presented at the Power-Gen
'91 Conference, Tampa, FL.
Jahnke, J.A. 1993. Continuous Emission
Monitoring. Van Nostrand Reinhold. New York, NY.
Jahnke, J.A. 1994. An Operator's Guide to
Eliminating Bias in CEM Systems. EPA 430-94-016.
Kinner, L.L. and Peeler, J.W. 1996. Development of
Fourier Transform Infrared (FTIR) Spectrometry and
Gas Filter Correlation Infrared (GFC-IR) Instrumental
Test Method Protocols for the Portland Cement
Industry.
Kopecky, M.J. and Rodger, B. 1979, Quality
Assurance for Procurement of Air Analyzers. ASQC
Technical Conference Transactions- 1979. Houston,
TX; pp 35-40.
Krooswyk, E.D. and Saliga, J.J. 1994. Recent
CEMS Installations Provide Valuable Insights. Power
Engineering, pp 35-38.
Mariatt, R.D. and Teacher, G.F. 1995. Design
Criteria Manual Guides CEM Project. Project
Engineering. August 1995. pp 59-61.
Nazzaro, J.C. 1986. Continuous Emission Monitoring
System Approval, Auditing, and Data Processing in
the Commonwealth of Pennsylvania. Continuous
Emission Monitoring - Advances and Issues. Air &
Waste Management Association. Pittsburgh, PA. pp
175-186.
Parrish, C. 1993. A Comparison of CEM Method
25A and Method 18 Test Results for Determining
VOC and Air Toxic Emission Rates. Continuous
Emission Monitoring - A Technology for the 90's.
Air & Waste Management Association, Pittsburgh,
PA. pp 165-173.
Passmore, J.G. and Brodmerkle, S.L. 1991. "A
Systematic Approach to the Selection and
Installation of Continuous Emissions Monitoring
Systems." Paper presented at the Power-Gen '91
Conference, Tampa, FL.
Peeler, J.W., Kinner, L.L, and DeLuca, S. 1996.
"General Field Test Method Approval Process and
Specific Application for a Direct Interface GCMS
Source Test Method." Paper presented at the Air &
Waste Management Association Meeting, Nashville,
TN. Paper 96-RP132.01.
Permapure, Inc. 1995. Nation Gas Sample Dryers.
Trade Literature. Permapure, Inc. Toms River, NJ.
Porter, T. 1990. Experience in Design, Installation,
Certification and Operation of Continuous Emission
Monitors at Resource Recovery Facilities.
Continuous Emission Monitoring; Present and Future
Applications. Air & Waste Management Association.
Pittsburgh, PA. pp 148-161,
Retis, C.E. and Henry, R.E. 1992. Meeting CAA
demands on CEM Systems. Power Engineering, pp
46-49.
Reuter-Stokes. 1995. Gas Analyzer Mixes in Some
Solid Savings. World Cement.
102
-------
Scott, N, 1995. The Care and Feeding of Your
GEMS, Acid Rain & Electric Utilities - Permits,
Allowances, Monitoring & Meteorology. Air &
Waste Management Association, pp 589-593.
Seidman, N.L., Peeler, J.W., and Queries, P. 1990.
Developing State Guidelines for Continuous Emission
Monitors. Continuous Emission Monitoring - Present
and Future Applications. Air & Waste Management
Association. Pittsburgh, PA. pp 174-180.
Tatera, J.F. 1993. A Review and Discussion of
Several Design Requirements that will Impact CE
systems Used in Hazardous Process Plant Environ-
ments. Continuous Emission Monitoring - A Technol-
ogy for the SO's. Air & Waste Management Associ-
ation, Pittsburgh, PA. pp 99-107.
U.S. Environmental Protection Agency. 1995. EPA
Method 301 - Field Validation of Pollutant Measure-
ment Methods from Various Waste Media. U.S.
Code of Federal Regulations, 40 CFR 63, Appendix
A.
U.S. Environmental Protection Agency. 1996. Code
of Federal Regulations Standards of Performance for
New Stationary Sources - Appendix B - Performance
Specifications,.40 CFR 60, Appendix B. Washing-
ton, D.C.
Wagner, G.H. 1993. "Meeting Your Monitoring
Requirements." Paper presented at the Air & Waste
Management Association Meeting, Denver, CO.
Paper 93-FA-164.01.
White, J.R. 1995a. Survey Your Options: Continu-
ous Emissions Monitoring. Environmental Engineer-
ing World. July-August 1995. pp 6-10.
White, J.R. 1995b. Technologies for Enhanced
Monitoring. Pollution Engineering. June 1995. pp
47-50.
Weldon, J. 1990. Emission Monitoring System
Design and Cost Considerations for Resource Recov-
ery Facilities. Continuous Emission Monitoring:
Present and Future Applications, Air & Waste
Management Association, pp 162-173.
103
-------
Chapter 7
Certification and Approval Mechanisms for
Non-criteria Pollutant Monitoring Systems;
Calibration and Demonstration of Performance
7.1 Introduction
The purpose of this chapter is to aid in understanding
how CEM systems are evaluated and approved for use
in regulatory applications. Quality assurance proce-
dures that are used on an on-going basis to maintain
CEM data quality within acceptable limits are dis-
cussed. Understanding these issues is important
because, 1) specific technical requirements in this
area may constrain the selection of CEM equipment,
2} the complexity of a monitoring program at a
particular facility may be significantly affected, and 3)
the specification or choice of initial approval proce-
dures and quality assurance activities may affect the
cost and risk associated with a monitoring program.
Many non-criteria pollutant monitoring applications
present special and difficult challenges in this area.
The lack of appropriate calibration materials or proce-
dures, inadequate performance specifications and
quality assurance requirements, and the absence of
reference test procedures all constrain the availability
of non-criteria pollutant monitoring systems. Some of
these issues are being addressed by instrument
developers and control agencies and progress is being
made in resolving problems. Many approaches that
have been successful in the field of criteria pollutant
monitoring cannot be transferred to monitoring of
non-criteria pollutants. An overview of certification
and approval mechanisms is presented in Figure 7-1.
As indicated on the chart, initial monitor set-up
activities are very important considerations for non-
criteria pollutant monitors. For such monitors, the
type of procedures and approaches used by the equip-
ment supplier or facility personnel to establish proper
calibration and operation of a measurement system
are quite diverse and are highly dependent on the
specific sampling and analysis system under consider-
ation. For some systems, these initial set-up and
calibration procedures necessitate understanding the
intricacies of the analytical technique, the limitations
of calibration check procedures, and the constraints
of available calibration standards. The "black box"
empirical approach traditionally used for criteria
pollutant CEM systems is not applicable to many non-
criteria pollutants. For many of these monitors the
constraints which affect the initial set-up of the
monitor also limit the methods which may be used to
demonstrate the performance of the CEM system.
Many of the issues associated with the initial set-up
and calibration of non-criteria pollutant CEM systems
are discussed in Section 7.2.
Methods and procedures used to demonstrate the
performance of non-criteria pollutant CEM systems
both initially and on an ongoing basis are shown in
Figure 7-1, Monitoring system certification is based
on determining whether the analyzer is properly
calibrated, and if the sampling system can deliver the
gaseous compounds to be measured to the analyzer
within a prescribed tolerance. A number of ways to
assess analyzer calibration and sampling system bias
exist. Specifications for non-criteria pollutant moni-
toring systems generally require a performance
specification test that includes calibration drift and
accuracy tests. These tests are performed soon after
installation of the monitor. Daily calibration checks
are performed for virtually all CEM systems to assess
data accuracy on a day-to-day basis. For some
applications, development of a quality assurance (QA)
plan and performing periodic accuracy checks may be
required. Existing and proposed regulatory mecha-
nisms for performing the initial certification test,
conducting subsequent evaluations of monitor perfor-
mance, and alternate techniques to assess a CEM
system's effectiveness are discussed in Sections
7.3.4, 7.3.6, and 7.4.
Because the existing regulatory mechanisms do not
address all of the potential issues and applications for
non-criteria pollutant monitors, alternate instrumental
protocols and acceptance criteria may be necessary
for non-criteria pollutant CEM systems data validation.
In some cases, approval mechanisms for instrumental
test methods may be modified for this application.
Techniques such as dynamic analyte spiking may be
necessary to properly evaluate CEM system perfor-
104
-------
CERTIFICATION and APPROVAL MECHANISMS
Initial Se
1. Determine Instrument Function,
Measurement Level, Detection
Limit
2. Obtain Calibration Materials
3. Perform Direct and System
Calibrations
D6MonstratioiiifOf Performance
INITIAL
A. Performance Specification Testing
1, Drift Tests
2. Relative Accuracy
3. Calibration Error
4, Response Time
5. Other
Figure 7-1. Certification and approval mechanisms,
mance necessary to properly evaluate CEM system
performance, especially for reactive or condensable
compounds, or compounds that have many interfer-
ences in the analytical procedures used for quantifi-
cation. Alternative approaches such as those used
in Germany, or those included in standards devel-
oped by the International Standards Organization
may also be applicable to non-criteria pollutant
monitors. These alternative approaches are dis-
cussed in Section 7,4.
7.2 Instrument Calibration
The term "calibration" is used to represent many
concepts in the CEM field. These concepts include
simple daily QA checks, routine adjustments to
compensate for analyzer drift, benchtop adjustments
made by the manufacturer, comparisons with
manual test methods, and fundamental adjustments
of various system components. Here, "calibration"
is defined as the set of necessary activities that
establish the relationship between the analyzer
output (instrument response) and a series of stan-
dard materials that span the anticipated operating
range. In the traditional and most straightforward
case, a system measuring gaseous pollutants is
calibrated by introducing a series of calibration gases
at known concentrations to assess performance.
Appropriate adjustments are then made so that the
instrument provides a proper response over the
measurement range. In other cases, where appropri-
ate calibration materials, performance specifications
or quality assurance procedures have not been fully
developed, the project manager must evaluate the
specific instrument response, the effective measure-
ment range, and the procedures used to establish
and verify proper calibration.
ON-GOING
A. Daily Calibration Checks
B. Performance Audits
1, Relative Accuracy Test Audits
2. Calibration Gas Audits
3. Other
In this chapter, the term "analyte" will refer to
gaseous compounds that are to be measured. Using
this term, calibration can be said to establish the
proper measurement system operation for the target
analytes.
7-2.7 Instrument Function
"Instrument function" is a term that refers to the
instrument's response versus concentration for each
analyte. A graphical representation of response
versus concentration defines the "shape" of the
instrument function. For example, some instruments
generate a linear response (i.e., flame ionization
detectors, FIDs) through a given calibration range,
while other instruments deviate from linearity at low
or high concentration ranges (infrared). Some
instruments have complex instrument functions such
as a logarithmic or quadratic response with respect
to concentration.
The instrumental function must be known for each
analyte to select the appropriate calibration points
over the anticipated operating range. For linear
devices such as FIDs, zero and upscale calibration
points may be sufficient. However, for non-linear
instrument functions, additional calibration points
that encompass each analyte concentration are
necessary to ensure proper quantification. Figure 7-
2 is a simplified representation of two instrument
functions.
Multi-component analyzers frequently have different
instrumental responses (response factors) for each
gas (analyte) that is being measured. In all cases,
the instrument function depends upon the analytes'
physical properties and interferences from other flue
gas components. For example, an FTIR analyzer
105
-------
may exhibit a linear response for most compounds
over a certain measurement range but will exhibit a
non-linear response for compounds such as carbon
monoxide.
Flame lonization
Detectors
Confidence
Intervals
Infrared Based
Analyzers
Concentration
Figure 7-2, Instrument Function of Two Analyzers.
Inherent to the instrument function is the detection
level achievable for each analyte. The "detection
level," or detection limit, is defined as the lowest
concentration level for each analyte that the instru-
ment is able to measure. The detection limit often
is determined by the instrument's "signal to noise
(S/N) ratio." This is the ratio of the instrument
response to the lowest level calibration standard,
divided by the instrument response when no analy-
tes are present (noise) which gives a good approxi-
mation of the S/N ratio, and thus the instrument's
detection limit. An instrument S/N ratio of 10
should be adequate to detect most analytes; how-
ever, this should be verified by direct observation.
Another important characteristic of the instrument
function is the range of uncertainty associated with
the response. Many optical instruments are very
repeatable. On the other hand, mass spectrometers
exhibit greater imprecision in their quantitative
output. Imprecision should not be mistaken for bias.
The level of repeatability or precision associated
with a specific instrument function must be consid-
ered in determining the appropriate number of cali-
bration check repetitions to perform and in establish-
ing control limits for initiating corrective action.
7.2.2 Calibration Materials
Calibration standards may be comprised of certified
gaseous standards in gas cylinders, permeation or
diffusion devices that generate known concentra-
tions of gas, optical filters, gas filled cells, chemical
solutions, or reference spectra. The target analytes,
sampling system, and analytical techniques deter-
mine which type of calibration standard should be
used. Regulatory requirements, the instrument
function, CEM system design and the expected
target analyte concentration range must be consid-
ered when selecting calibration standards. In gen-
eral, a CEM system calibration technique should
provide for checks at a zero, a mid-point, and a
high-range concentration for each target analyte.
When certifying and calibrating multi-component
analyzers, all, or as many of the analytes as possi-
ble, should be in the same gas cylinder or set of
permeation/diffusion vials. This will enable simul-
taneous calibration and calibration checks to be
conducted providing for both cost and time savings.
The calibration standard manufacturer should be
contacted to ensure that calibration standards are
available as stable blends at the required concentra-
tions.
7.2,2.1 Gaseous Standards
Virtually all extractive systems can analyze calibra-
tion gas directly. Many point in-situ analyzers can
analyze calibration gases also. Use of standard
reference materials (SRMs! for routine calibrations is
cost prohibitive. However, secondary calibration gas
standards, called Protocol 1 gases, are referenced to
the SRM, and are lower in cost. For some gases,
SRMs are available from the National Institute of
Standards and Technology. Protocol 1 (USEPA,
1996a) gases are certified by the gas manufacturer
under a protocol accepted by EPA. Analyzers that
accept calibration gases directly can use Protocol 1
gases (±1-2%), certified reference standards
(usually ±2 to 5% accuracy), or permeation and
diffusion devices that generate a known concentra-
tion of gas. Also, Part 51, Method 205 provides for
the dilution of calibration gases for multi-point
calibrations to reduce the number of cylinders and
amount of gas that is required.
Protocol 1 gases have a period for which they are
certified, whereas merely certified reference materi-
als usually do not. Calibration standards for non-
criteria pollutants usually are not available as Proto-
col 1 gases, and therefore, the best available stan-
dards are those that have reference values certified
by the manufacturer. All calibration standards
should be checked periodically to ensure that the
compounds have not degraded from their certified
values. Suspect calibration gases should be reana-
lyzed and recertified. Compounds that are particu-
larly problematic in compressed gas form are those
that are reactive or have the tendency to polymerize.
Examples of these gases are halogenated acids {i.e.,
HCI and HFS and formaldehyde, respectively.
106
-------
Some compounds are not available in gaseous form
because their vapor pressures are too low, or they
are not stable at elevated pressure. Gas standards
for these compounds can be generated from the
liquid or solid material using either permeation tubes
(for liquids) or diffusion vials (for solids). Permeation
tubes are sealed devices that contain the analyte
and usually are constructed from semipermeable
Teflon™ tubing. Diffusion vials are constructed from
glass or quartz and have a small orifice through
which they are filled with the solid material. Gas
generators and permeation/diffusion apparatus are
available with emission rates certified within ± 2 to
5 percent.
The generation of gaseous standards from liquids or
solids is fairly simple. A sample of the liquid or solid
contained in the vial is placed in a temperature
controlled oven through which a carrier gas is
swept. The concentration of the generated gas is
directly proportional to the permeation/diffusion rate
of the tube or vial. Permeation and diffusion rates
are based upon the vapor pressure of the compound,
the temperature and pressure of the oven, and the
flow rate of the carrier gas. Figure 7-3 is a genera-
lized schematic of a permeation/diffusion based
gaseous standards generator.
Figure 7-3, Generalized schematic of a permeation/
diffusion-based calibration gas generator.
Another means of generating gaseous standards is
termed the "hanging drop" approach. In this proce-
dure, a drop of the standard solution is introduced
into a gas stream via a syringe. The syringe assem-
bly is placed into a heated compartment so that the
drop is continuously volatilized by carrier gas. Use
of the hanging drop technique involves many param-
eters that introduce errors that could invalidate test
results. Factors such as preparing the standard
solution and determining the exact volatilization rate
affect the resultant concentration of analyte in the
calibration gas, and thus the accuracy of the calibra-
tion. This procedure should be used only when
certified gaseous standards and permeation/diffusion
standards are not available commercially.
7.2.2.2 Reference Spectra
Some analyzers employ reference spectra standards
for calibration purposes. Reference spectra may be
used repeatedly without having to recalibrate the
analyzer, and thus provide savings in cost and time.
Fourier transform infrared (FTIR) and tunable diode
laser analyzers are examples where reference spec-
tra standards are used to quantify data. A reference
spectrum is composed of digitally stored data
obtained from an instrument's spectral response to
a known gas standard, acquired at a known con-
centration, temperature, pressure, and measurement
pathlength. These spectra represent graphically the
instrument's response (relative absorbance of elec-
tromagnetic radiation) with respect to wavelength.
To employ reference spectra acquired at one point
in time to quantify data acquired at a later date, the
instrument response obtained during reference
spectral acquisition is compared with the instrument
response obtained during the time of data collection.
A comparison of the instrument response at differ-
ent points in time is performed by analyzing a
calibration transfer standard (CTS). Similarly,
reference spectra acquired on one instrument may
be used to quantify data collected on another
instrument provided that the two instrument re-
sponses are related through the CTS.
The CTS gas is used as the link between the instru-
ment response at the time of calibration and the
instrument response at the time of actual sample
acquisition. A CTS spectrum generated during
instrument calibration is compared to a CTS spec-
trum acquired during sampling. A mathematical
transformation (correction factor! is then applied to
match the reference spectrum CTS to that acquired
during sampling. The exact same process is used to
adjust the reference spectra so that they represent
the instrumental response at the time of sampling,
and therefore may be used to quantify data.
Figure 7-4 is a simplified diagram of the use of
calibration transfer standards. In this figure, the
CTS gas collected during field sampling is 0.85
times the response of the CTS collected with the
reference spectra. Therefore, to employ the refer-
ence spectra, a mathematical transformation must
be applied to the CTS reference spectrum to match
the CTS collected with the sample spectrum, and
thus adjust for the difference in instrument re-
sponse. This same transformation is then applied to
the analyte reference spectrum to quantify the
analyte in the sample spectrum.
107
-------
Analyte
Reference Spectrum
Wavenumber
CTS Reference Spectrum
Abs = 1.0
Gas
Sample Spectrum
Analyte
i^-j Interference
Bands
Wavenumber
CTS Spectrum during sampling
Abs = 0.85
Wavenumber
Wavenumber
Figure 7-4. Use of calibration transfer standards (CTS).
The concept of calibration transfer is simple in
theory; however, in practice it may be quite com-
plex. In theory, any set of reference spectra may be
used provided that the calibration transfer is valid.
In practice, reference spectra are most often gener-
ated on the actual field instrument to minimize errors
associated with the necessary CTS mathematical
transformations and added noise. Detailed proce-
dures for acquisition of FTIR reference spectra
standards and their transference are beyond the
scope of this text. These may be found in an EPA
FTIR Protocol (USEPA, 1996b).
7.2.2.3 Other Calibration Techniques
Some analyzers are not capable of directly accepting
calibration gases. Many cross-stack in-situ analyzers
are not designed to accept calibration gases while
the design of others allows the use of a flow-
through gas cell. The approach used to check the
calibration for these monitors also may be
constrained because the flue gas cannot be removed
from the measurement path and thus the zero
concentration is difficult to verify. Some in-situ
monitors have been designed or modified to allow
the introduction of calibration gases {Jahnke, 1993,
1994), For many of these instruments, the
calibration gas cell must be maintained at the same
temperature as the flue gas sample to obtain valid
calibration measurements.
For in-situ gas analyzers that cannot analyze calibra-
tion gases, a zero reflector or other device may be
used to simulate the zero concentration condition,
Calibration standards may be comprised of a gas-
filled cell or an optical filter that is inserted into the
measurement path to generate an upscale instru-
mental response. Where gas cells are mounted
within a temperature controlled region of an
analyzer, temperature compensation features that
are used in the sampling mode may need to be
disabled to adjust for optical absorption and/or gas
density variations in the flue gas. In this situation,
calibration checks with gas cells may not represent
fully the performance of the entire measurement
system. Problems in the temperature compensation
feature cannot be detected if it is disabled during
the calibration checks. When a significant
component of the monitoring equipment is excluded
from the routine calibration check procedure, a
separate means of verifying the performance of the
excluded component must be used.
The primary calibration for most in-situ analyzers
that cannot accept calibration gases is established
typically during laboratory bench tests at the
manufacturing facility. Based on the design of the
instrument, various internal functions are monitored
to maintain performance within acceptable limits or
to provide an indication to the operator that the
initial calibration is no longer valid. The parameters
and functions that are monitored depend on the
analytical principal and the design of the instrument.
Understanding the relationship among these
parameters and the acceptable tolerances is usually
more complex than simply evaluating a monitor
response to an external calibration standard.
108
-------
Some analyzers employ a correlation factor derived
from comparing the instrumental response to
another sampling method. The initial instrument
function is determined, and the initial calibration is
conducted by setting and monitoring internal
instrumental parameters. Then the instrument
response is correlated to concentration by
comparison with reference test methods.
Polynuclear aromatic hydrocarbon (PAH) analyzers
are an example of such instruments. Calibration is
factory set, and correlation to concentration is per-
formed by comparing the instrument response with
EPA Office of Solid Waste Method SW846-
0010/8270 {SW846 Manual, Nov-ember 1986).
This instrumental correlation to reference method
determined concentration must be performed on a
source-by-source basis.
Continuous particuiate monitors are another example
of analyzers calibrated by correlation to manual
sampling methods. This can be accomplished in
accordance with the provisions of the International
Standards Organization 10155 (ISO, 1995, see also
7.3.4 CEIV1 System Approval Mechanisms in
Germany, or the proposed Part 60, Performance
Specification 11 A). This standard establishes
performance specifications and procedures for
performing the calibration of continuous particuiate
matter measurement techniques relative to the
results of a series of
manual tests. Recalibration is required when
conditions change at the facility that may affect the
calibration.
For particuiate monitors, the specific parameters
that are monitored on a day-to-day basis (or in some
cases on a continuous basis) depend on the mea-
surement technique. For example, transmissometers
used as particuiate monitors incorporate a simulated
zero mechanism (often a reflector to simulate zero
opacity) and an internal optical filter (as an upscale
check) to verify proper performance. Similarly, in-
situ light scattering devices used as particuiate
monitors include a two point calibration check by
temporarily not illuminating the effluent and by using
the light source and a filter to check the response of
the detector. These and other internal checks are
sufficient to determine that the instrument operates
within the limits established by the manufacturer.
7,2.3 Direct and System Calibration Procedures
Once the instrument function and necessary
calibration materials are determined, the monitor
may be calibrated. There are two conventional
calibration methods for CEM systems that accept
calibration gases. These methods are referred to as
"direct" and "system" calibrations, respectively.
Figure 7-5 is a generalized schematic of direct and
system calibrations.
From
Probe
Flue Gas Flow=QT
•vr ..__»-;• .
• Vf Ir
f"
S: ;:,;«
Primary
1 — Particuiate
Filter
. .Ca
w /
/
librati
Pump
Analyzer
Calibration Gas
Introduction Assembly
SystemCalibration
Qa>QT
Direct Calibration
Q2 = QT
Direct Calibration
Calibration Gases
Figure 7-5. Direct and system calibrations.
109
-------
A direct calibration consists of injecting the stan-
dards directly at the analyzer and recording the
response. Direct calibration checks assess the
stability of the analyzer and the extent to which the
instrument has drifted from its zero and span calibra-
tion points. Direct instrument calibrations test the
instrument function only.
A system calibration injects the calibration stan-
dards at a point in the sample delivery system as
close to the probe tip as possible. Calibration
standards must be introduced at a point upstream of
the primary particulate matter filter to account for
any filter/particulate cake reactions with the anal-
ytes. System calibrations provide an indication of
the sample handling system's effectiveness to
deliver the analytes to the analyzer, as well as
testing the instrument function.
Optimally, direct and system calibrations should
agree. Large differences between these two calibra-
tion techniques for the same analytes could indicate
that a leak is present or that losses due to chemical
or physical reactivity are occurring. Losses due to
adsorption and condensation are the most typical for
non-criteria pollutants. System calibrations are also
necessary to determine if compounds may have
adsorbed on the filter. Adsorbed or condensed
compounds that volatilize later can cause a positive
measurement bias.
For those analyzers that measure multiple compo-
nents, daily calibrations of the instrument may not
be economically feasible for all compounds that are
measurable. In this instance, surrogate compounds
must be chosen for calibration checks. Surrogates
are target analytes that can also represent a range
of other target analytes through similar instrumental
response, sampling system bias response, and other
important characteristics. Selection of surrogates
depends on their ability to represent more than one
of the target analytes, based on similar physical
properties. An example of surrogate selection is the
use of one water soluble compound to represent
many. If a CEM system is designed to measure HCI,
formaldehyde, acetaldehyde, ammonia, methanol,
and benzene, toluene and xylenes (BTX), the use of
HCI and benzene may be possible as surrogates for
the entire population of compounds. HC! will repre-
sent the water soluble and reactive compounds,
while the benzene can represent the aromatic
compounds. Protocols using surrogate recovery
check compounds have been prepared by the
Portland Cement Association based upon using
surrogate compounds (Kinner, 1995). In these
protocols, three recovery check compounds were
used to represent fourteen volatile organic com-
pounds to reduce the cost and complexity of the
testing program.
Optical analyzers that employ a reference spectral
library for quantification also need to perform cali-
bration status checks on a periodic basis. This
ensures that a valid calibration transfer exists be-
tween the time of reference spectra acquisition
{reference library) and the time of sample collection
and analysis. A gaseous calibration transfer stan-
dard {CTS> should be analyzed and the response
compared to the response at the time the reference
spectra were obtained. Direct and system calibra-
tions using the CIS should be performed so that the
effect of sampling system bias on instrumental
response may be measured. Ethylene and carbon
monoxide have been successfully used as calibration
transfer standards for extractive FTIR measurement
systems {USEPA, 1993; Kinner, 1994).
Software audits also can be used as a tool to evalu-
ate optical CEM system quantitative algorithms and
their ability to accurately quantify data. Software
audits may consist of synthetic spectra (a number of
reference spectra superimposed upon one another),
or actual measurement data that challenges the
ability of the analysis technique to distinguish
between the target analytes and their interferences.
Such audits can be useful where a computer is used
to analyze broad-band spectral measurements such
as those employed by an FTIR. In such cases, the
selection of particular spectral regions over which to
perform an analysis, or the particular procedure used
to remove the affects of water or other interfer-
ences, may significantly affect the results.
An audit using artificially generated spectra, or
spectral data files from prior measurements obtained
at another source, can be employed to gauge the
effectiveness of the analytical procedure. However,
audits are likely to generate one-sided conclusions.
For example, an analysis routine that cannot prop-
erly identify target analytes in an audit spectrum
most likely will fail to identify these analytes in
actual samples. The converse is not necessarily
true. Obviously, a software audit cannot evaluate
the physical performance of a sampling system or
the performance of an analyzer.
7.3 Demonstration of Performance;
Certification and Quality Assurance
CEM systems installed to comply with regulatory
requirements must demonstrate acceptable perfor-
mance both initially and on a continuing basis. The
110
-------
specific requirements vary for different regulatory
applications (see Chapter 2) or according to specific
conditions or requirements included in the facility's
permit. Nevertheless, most CEM systems are
subject to a performance specification test, or
"certification test," soon after installation. Source
operators are required to perform daily checks of
CEM systems calibration status to ensure the validity
of data on a day-to-day basis. Some sources are
required to develop quality assurance plans and
conduct periodic audits of CEM systems perfor-
mance. In other cases, the regulatory agency relies
on a general requirement for the proper operation
and maintenance of the CEM system and can require
performance specification tests to be conducted as
necessary.
7.3.1 EPA Performance Specifications for
Non-criteria Pollutant Monitoring
Performance specifications are a set of procedures
and criteria that are used to evaluate the acceptabil-
ity of the CEM system at the time of installation and
whenever specified in the applicable regulations.
Existing EPA performance specifications (PS) for
non-criteria pollutant analyzers address monitoring of
VOCs and TRS only. Performance specifications for
H2S in refinery fuel gas also have been developed to
provide an alternative to measurement of SO2
emissions at many emission points. Existing EPA
performance specifications for non-criteria pollutants
are summarized in Table 7-1 (requirements from PS
2 are shown for comparison).
Performance specifications for VOC CEM systems
are included in PS 8 and PS 9 of Part 60 for sources
subject to New Source Performance Standards. PS
8 is applicable to analyzers that provide a total
hydrocarbon response and PS 9 applies to analyzers
that speciate hydrocarbons using a gas chromato-
graph. Neither of these performance specifications
require the use of a specific type of detector.
Performance specifications for hydrocarbon monitors
also are included in 40 CFR, Part 266, Appendix IX
for sources regulated by the boilers and industrial
furnaces (BIF) rule, and in 40 CFR, Part 503 (subpart
E) for sewage sludge incinerators. These specifica-
tions require a heated sampling system and heated
flame ionization detector (FID) analyzer.
Performance specifications 8A for hydrocarbons,
10A for multi-metals, 11A for particutate matter
(PM), 12A for mercury (Hg), 13A for hydrochloric
acid (HCI!, and 14A for chlorine (CI2) monitors are
proposed in the hazardous waste combustor rule.
(FR 4/19/96) This proposed rule covers both haz-
ardous waste incinerators and BIF sources. These
performance specifications will be included in Part
60 if this proposed rule is adopted. All of the
performance-based specifications listed above are
derived from PS 2, ISO 10155 (for paniculate
monitoring), and various instrumental test methods.
7.3,2 Performance Specification Testing
A performance specification test (PST) is performed
soon after installation of the CEM system and peri-
odically as specified in the applicable regulations.
The test is performed in accordance with procedures
detailed in the applicable performance specifications
previously described. In some regulatory applica-
tions (and in common usage), this initial test is
referred to as a "certification test." The PST usually
includes a calibration drift test and an accuracy test,
(For Part 75 applications, the term "calibration error
test" is used to describe the drift test and daily zero
and upscale calibration checks. This terminology is
avoided here to minimize confusion.) The accuracy
test is often a "relative accuracy test" involving
comparison with results of an independent reference
test method (RM). For some monitors, the accuracy
test requires analysis of several calibration standards
to determine calibration error only. Depending on
the specific regulatory requirements, the PST also
may include response time or cycle time tests, short-
term drift tests, and demonstrations that the mea-
surement system can operate continuously for
prescribed periods without unscheduled maintenance
or repairs. (Part 75 performance specifications are
described in Chapter 2, Acid Rain Program.)
7.3.2.1 Calibration Drift Test
A typical PST includes a calibration drift test per-
formed over 7 consecutive days. The calibration
drift test evaluates the stability of the monitor
response to the procedure used for daily checks of
the monitoring system (see also Daily Calibration
Checks). The calibration drift test is conducted by
introducing the zero and span gases, gas cells,
optical filters, or electric signals to the system and
recording the instrument response. The deviation of
the instrument response at the zero and upscale
calibration point must meet the applicable criteria.
For example, the allowable calibration drift in PS 8A
is 2.5% of the span. Therefore, a hazardous waste
incinerator required to monitor hydrocarbons subject
to PS 8A has an allowable analyzer span value of
100 ppm, and may have a maximum of ±2.5 ppm
drift at the zero and span level for each 24-hour
period over the seven consecutive days during the
performance specification test. The tester is al-
lowed to adjust the calibration accordingly after
each daily check during the drift test.
111
-------
Table 7-1. Existing Performance Specifications for Selected Pollutant Monitors
PS#
2*
S02
NOX
5
TRS
7
H2S in
fuel gas
8
Total
VOCs
9
GC
CEMS
VOCs
BIF1
THC
ss*"1
THC
40CFR
Part 60
Appendix A
Part 60
Appendix A
Part 60
Appendix A
Part 60
Appendix A
Part 60
Appendix A
Part 268
Appendix IX
Part 503
Subpart E
PERFORMANCE SPECIFICATION TEST
Calibration
Drift
7 day test
2.5% of span
7 day test
5% of span
<6/7 days)
7 day test
5% of span
{6/7 days)
Same as PS2
N/A
7 day test
3% of span
Annual PST
7 day test zero and
span within 6%
Calibration
Error
N/A
N/A
N/A
N/A
7 day test
3 injections
at three
levels
Precision
and Linear-
ity
r2 = 0.995
CE *10%
for all levels
5% of span
for all cali-
bration
points
zero within
5 ppm
mid- and
span within
10 ppm
Relative
Accuracy
±20% of Refer-
ence Method or
1 0% of stan-
dard
Same as PS2
Same as PS2
Same as PS2
Performance
Audit during CE
testing
±10% criteria
N/A
N/A
Response
Time
N/A
N/A
N/A
N/A
< 5 min-
utes or as
specified
1 20 sec-
onds for
95% step
change
200 sec-
onds for
90% step
change
DAILY
CALIBRATION
REQUIREMENTS
zero and span drift
within 2X's PST
before required
adjustment
Same as PS2
Same as PS2
Same as PS2
Triplicate mid-level
calibration checks
must be within
1 0% of certified
value or repeat ini-
tial 3-point calibra-
tion
zero and span drift
<3 ppm
Same as PS2
* Included for comparison purposes
** Sewage Sludge Incinerators
1 No number assigned
112
-------
7.3.2.2 Relative Accuracy Test
Relative accuracy (RA) tests are required for almost
all criteria pollutant CEM systems except for opacity
monitors. A relative accuracy test is also required
for certain non-criteria CEM systems. The RA may
be conducted concurrently with the 7 day calibration
drift test. The relative accuracy test involves
comparing the average CEM system response with
the integrated average pollutant concentration or
emission rate for each corresponding reference
method test run. A minimum of nine RA compara-
tive test runs are required; however, EPA specifica-
tions allow the rejection of up to three test runs.
Therefore, 12 runs usually are performed in practice
with the best nine results used in the RA calcula-
tions. The RA specification for a CEM system
subject to PS 8 is <:20% of the average reference
method (RM) result or <:10% of the applicable
emission standard, whichever is greater.
Relative accuracy calculations are based upon the
results of at least nine comparative runs. The
difference between the run averaged CEM system
results and the reference method is calculated and
averaged. The absolute value of the summed
differences for each run plus a confidence coeffi-
cient (CC) is divided by the average of the reference
method values (RMavs):
RA = [|d| + CC]/RMave *100 Equation 7-1
The confidence coefficient is expressed In terms of
a statistical t-value according to the following
equation for a t-value at a = 0,025 for n-1 degrees
of freedom."
CC = t
Where
0.97B
5fSri/n1/2
Equation 7-2
Corresponds to the probability that a
measured value will be biased 2.5% at
the 95% confidence level.
The standard deviation of the differ-
ences of the 9 data pairs.
Number of test runs.
As shown above, the relative accuracy result in-
cludes both mean difference and confidence coeffi-
cient terms based on the paired CEM system and
reference test results. Both the accuracy and
precision of the paired measurements are evaluated.
A user must recognize that the relative accuracy Is
affected by the accuracy and precision of both the
monitoring system and the reference test method.
Failure of a relative accuracy test may be due to
problems with the CEM system, problems with the
reference test methods, problems with the represen-
tativeness of the sampling location, or other factors.
A failed test requires careful investigation to deter-
mine the cause, after which the test must be re-
peated.
Relative accuracy is based on comparing concurrent
monitoring data and reference method test results.
Therefore, the response time of the CEM system
must be taken into account to directly compare the
monitoring with the reference method results.
Where instrumental reference test methods are
used, the response time and sampling frequency of
the reference method must be considered also. For
relative accuracy tests with these systems, sampling
runs must be of sufficient duration to allow compari-
sons of valid averages obtained from both the CEM
system and the reference system.
Where CEM systems are compared to reference
methods that provide integrated results (e.g.,
impinger trains and most other manual methods), the
exact start time and stop time for manual sampling
runs land time associated with interruptions for port
changes or other problems) must be included. When
substantial changes in emission concentrations or
emission rates occur over the course of the run, or
when short duration spikes occur during runs,
comparisons between integrated reference methods
and CEM systems are often invalid.
The relative accuracy comparison of a CEM system
to a reference method assumes that the RM gener-
ates accurate and precise data, and that the source
testers have adequate experience in correctly
executing the reference method. Interferences
affecting the test method results can be mistaken
for bias in the CEM system measurements. Use of
an imprecise test method increases the magnitude of
the confidence coefficient in the relative accuracy
calculation. Similarly, the imprecision due to sloppy
or inexperienced testers can affect the results
dramatically. These issues were initially of great
concern in criteria pollutant monitoring applications
but eventually were minimized through experience
and development of more precise instrumental test
methods for use in place of wet-chemistry methods
for most applications. For non-criteria pollutant
monitoring applications, these factors are still of
great concern because: 1) interferences affect some
reference methods when extended to new applica-
tions, 2! sample recovery documentation and other
QA criteria are not included in all methods, 3) lower
and potentially variable emission concentrations are
encountered, 4) the precision of some methods is
113
-------
poor and for some other methods is unknown, 5)
some methods are not demonstrated fully where
they applied before promulgation of standards at the
specific source category, and 6) source testers have
much less experience in using these methods for
performing relative accuracy tests. In summary, an
imprecise, inaccurate, or improperly conducted
reference method test will increase the probability
that the GEM system will fail the PST even if it is
functioning and calibrated properly.
Further disagreements may result in relative accu-
racy testing due to actual technical limitations. For
example, PS 8 includes a relative accuracy specifi-
cation, however, the only reference method that
generates a total hydrocarbon response is EPA
Method 25A {Part 60, Appendix A), which specifies
the use of a flame lonization detector (FID). Method
25A is the only reference method directly compara-
ble to a CEM system installed to satisfy PS 8.
However, PS 8 does not specify the use of a partic-
ular detector; it simply specifies instrumentation that
generates a total hydrocarbon response.
Presuming that the CEM system uses an FID, then
the relative accuracy test would involve comparing
two FID measurement systems. Because the two
systems use the same analytical technique, interfer-
ences within the effluent stream could go unde-
tected. Additionally, differences in the compound-
specific response factors between two FIDs could
affect the outcome of the relative accuracy test
even though both systems are properly calibrated
using the same standard (Cone, 1989).
Also, consider the example where a photoionization
detector (PID) is chosen for the CEM system in-
stalled to comply with PS 8. A relative accuracy
comparison with a Method 25A FID may not be
appropriate because PID and FID responses differ
substantially for different classes and functional
groups of hydrocarbons. Because neither the CEM
system nor the reference method provides any
information regarding the speciation of individual
compounds within the flue gas, one is unable to
decide whether apparent agreement or disagreement
between the CEM system and the reference method
was due to differences in response factors or other
calibration/accuracy issues. In this case, the out-
come of the relative accuracy test is likely to be
indeterminate and may generate more controversy
than answers.
The BIF and sewage sludge incinerator performance
specifications for hydrocarbon monitors, and the
proposed PS 8A, specify using a heated FID and
heated sampling system. PS 8A requires a three-
point calibration error test rather than a relative
accuracy test comparison using Method 25A. This
approach avoids the relative accuracy test issues
described above for PS 8. However, the three-point
calibration error test, using calibration gases, pro-
vides only a limited evaluation of the monitoring
system. (See also Section 7,3.2.3.)
Proposed performance specifications 10A (multi-
metals CEM systems) and 12A (mercury CEM
systems) include relative accuracy specifications
(USEPA, 1996c). Although multi-metals CEM
systems are not commercially available at this time,
prototypes have been developed using a variety of
approaches. Limited field tests have been con-
ducted. Calibration standards and techniques,
measurement sensitivity, and accuracy are critical
technical issues for these developing measurement
systems. The accuracy and precision of proposed
Method 29 has not been demonstrated to allow for
relative accuracy testing of multi-metals CEM
systems over the expected range of pollutant
concentrations (Brown 1996, Laudal 1996). Modifi-
cations to proposed Method 29 are being developed.
Mercury may exist as elemental mercury (Hg°) and
as other oxidized Hg (Hg+, Hg*2) species in effluent
streams. Mercury CEM systems generally sense
elemental mercury and employ sample conditioning
elements that convert all mercury in the sample
stream to this form (Richards, 1996, Roberson,
1996). (See also Chapter 3.) Proposed EPA Method
29 attempts to speciate the different oxidation
states of mercury. Studies have been conducted on
coal-fired facilities to evaluate the performance of
Method 29 by spiking elemental mercury and HgCI2
(Laudal, 1996). These investigators have shown
that the presence of significant quantities of SO2
(> 1,500 ppm) and the presence of 10 ppm chlorine
(Cl2) affects substantially the collection of elemental
mercury and the determination of mercury species
present. Other mercury test methods have been
shown to be affected significantly by the presence
of NOX in the sample stream. Until the performance
of the manual test methods are understood and the
effluent conditions that affect the results are well
known, an appropriate and reliable relative accuracy
test procedure cannot be devised for mercury CEM
systems.
Proposed PS 13A and PS 14A for HCI and CI2
specify calibration gas audits instead of comparison
to reference methods. The reference method for
HCI and CI2 is EPA Method 26A, which collects the
gaseous components in an impinger train and
114
-------
measures the dissociated chloride ions by ion
chromatography. Proposed PS 13A and 14A specify
that HCI and CI2 must be measured in the gas phase
as opposed to the measurement of the dissociated
chloride ion in solution. A cylinder gas audit only is
required in these performance specifications, not a
relative accuracy test employing Reference Method
26. HCI continuous emission monitoring require-
ments in Pennsylvania; however, require a relative
accuracy test using Method 26. Pennsylvania
requirements do not mandate that HCI and CI2 be
measured in the gas phase.
Even where reference methods have been developed
for the pollutants of interest, the comparison of CEM
system and RIV1 data may not assess the ability of
the CEM systems to measure the compounds of
interest accurately. Consider for example measure-
ment of formaldehyde which is a very water soluble,
reactive compound and which may polymerize in
sample delivery systems. Several reference methods
exist for the collection and measurement of formal-
dehyde. These methods rely on sample collection in
high purity water, the Parosanaline Method (EPA
TTN EMTIC BBS, 919-541-5742); in dinitro-phenyl
hydraztne (DNPH) impregnated silica gel tubes,
NIOSH Method 3500, (NIOSH 1994); or in sampling
trains containing DNPH (SW846-0011). The results
provided by all three of these methods are biased by
the presence of other aldehydes and ketones (Serne,
1993). In addition, the reaction of formaldehyde
with DNPH is very pH specific. The two DNPH
methods are sensitive to the presence of acid gases
(e.g., HCI, SO2, H2SO4, etc.) and base gases (NH3).
Thus, the use of these methods at any source
having significant concentrations of these com-
pounds in the flue gas would affect adversely
relative accuracy test results for a formaldehyde
CEM system.
In summary, relative accuracy specifications and
test procedures have long served as the primary
evaluation standard for the acceptability of criteria
pollutant CEM systems. For some non-criteria
pollutants and monitoring applications, the relative
accuracy test continues to play an important role in
demonstrating performance and certifying CEM
systems. However, for other applications relative
accuracy tests are inappropriate or must await
future development of appropriate reference test
procedures.
7.3.2.3 Calibration Error Tests
As used here, a "calibration error test" consists of
multi-point comparison of monitor response to a
series of calibration materials, (The usage of
"calibration error test" varies. In the Part 75 Acid
Rain Program, this term is used to refer to calibration
drift tests and daily checks. In Part 75, the term
"linearity test" is used to refer to multi-point accu-
racy tests relative to calibration materials.) In
certain applications, as indicated below, the calibra-
tion error test is required in place of the relative
accuracy test. However, in other regulatory applica-
tions, both a calibration error test and a relative
accuracy test are included in performance specifica-
tion tests.
As previously mentioned, BIF and sewage sludge
incinerator performance specifications for total
hydrocarbon monitors, and proposed PS 8A, do not
include a relative accuracy test but instead require
a calibration error test to determine the accuracy of
the monitoring system. The test involves analysis of
a zero gas and two upscale calibration gases. The
calibration gases used for these tests are propane in
a balance gas of nitrogen. The instrument response
of these measurement systems to different com-
pounds (i.e., "response factors") is known to vary as
a function of: 1} the number of the carbon atoms,
2) the degree of carbon bond saturation (carbon
double and triple bonds) within the molecule, 3} the
presence of chlorine, and 4} other instrument-
specific factors. However, the determination of
these response factors is not included in the calibra-
tion error test. Furthermore, the effect of other
effluent stream constituents (e.g., moisture, C02,
CO, 02, NOX, etc.} on the accuracy of the monitoring
data can not be determined by the calibration error
test. The calibration error test in this case may
therefore not adequately validate the monitoring
system for actual flue gas hydrocarbon measure-
ments.
PS 9 applies to CEM systems that employ a GC to
separate organic compounds. As with PS 8, it also
does not specify a particular type of detector.
Multiple target analytes are speciated by an
instrument-specific calibration curve generated from
certified calibration standards. A preliminary calibra-
tion error test of the analyzer is accomplished by
analyzing low-level (40-60%), mid-level {90-110%),
and high-level (140-160%) calibration standards that
represent the percent of the measured concentration
levels for all analytes. A screening test using the
appropriate instrumental test method may be re-
quired to identify the effluent analytes and deter-
mine their relative concentrations.
The PST required by PS 9 includes a performance
audit test and a seven-day calibration error (CE) test
period during which the initial instrument perfor-
115
-------
manee is evaluated. The CE is performed daily by
analyzing calibration standards at each level in
triplicate (3 analyses X 3 levels = 9 analyses/day).
The calibration error for each analyte is determined
by comparing the average instrument response for
each level of triplicate analyses to the certified
reference value of the calibration standards. The
calibration error may not exceed ±10% for each
analyte.
The performance audit test required by PS 9 con-
sists of analyzing an EPA audit gas during the seven-
day CE test. This test is intended to provide a
somewhat independent check because the audit gas
must be obtained from a different supplier than the
gases used for the CE test. The concentration of
the analytes in the audit gas generally are not
known to the user.
7.3.2.4 Response Time Tests and Cycle Time
The measurement system response time is important
for monitoring applications with short averaging or
reporting periods or where spikes in the effluent
concentration occur. The sampling system response
time is determined by monitoring the instrumental
response to a calibration standard while performing
a system calibration. Upscale and downscale
response times are determined by introducing high-
range and zero calibration gases respectively.
System response time is defined as the time it takes
the analyzer to reach 95% of the step-change in
concentration.
Determination of the measurement system response
time can be important even when it is not specified
in the applicable regulations. The system response
time must be considered when conducting the
relative accuracy test. For measurement systems
that accept calibration gases, the system response
time must be known so that calibration gases are
introduced for a sufficient period to provide a stable
measurement response.
Some regulations specify the frequency for obtaining
measurements. 40 CFR 60.13 specifies that all
measurement systems must complete a cycle of
sampling, analysis, and data recording at least once
during each 15 minute period. In contrast, B1F
regulations specify that a sample must be acquired
and analyzed every 15 seconds. The average
emission rate is computed and recorded at least
every 60 seconds, with" an hourly rolling average
calculated each minute. Usually, the cycle time is
immediately apparent from the design of the GEM
system and programming of the data acquisition and
handling equipment. In some cases, time-shared
extractive systems are used to allow a single set of
analyzers to monitor emissions from several loca-
tions. Other functions such as particulate filter or
sampling line blow-back or purges may also be
included for certain CEM systems. These functions
must be considered when determining the cycle time
for some systems.
7.3.2.5 Other Performance Test Criteria
Some regulations include additional performance
specifications and test procedures. For example,
transmissometers may be used as particulate mass
concentration monitors and as opacity monitors.
These instruments, if subject to the proposed
revisions to Performance Specification 1, would be
subject to a short-term drift test to detect diurnal
variations in monitor performance. This drift test
includes zero and upscale calibration drift checks at
one-hour intervals over a period of 24 hours. Similar
short-term drift tests were included in Part 60
performance specifications prior to 1983 and are
included in some state monitoring requirements
(e.g., Pennsylvania).
Many performance specification test requirements
include an "operational test period" during which
the CEM system must operate continuously without
unscheduled maintenance or repairs. This is typi-
cally a one week (168-hour) test. Such a test period
may seem insignificant for a measurement system
that is intended for continuous monitoring over a
period of many years. However, early experience
with criteria pollutant monitoring systems revealed
that this was a difficult specification for many CEM
systems to meet. Although much progress has been
made in improving the ruggedness and reliability of
conventional systems, the application of new
technologies or new measurement systems attempt-
ing to utilize laboratory instruments may require that
more attention be devoted to achieving reliable
operation.
7.3.3 Ongoing Performance Checks and
Quality Assurance
Additional performance checks and quality assur-
ance activities are required to be performed after
initial installation, calibration, and performance
testing (certification) of a CEM system. In general
these include whatever is necessary to properly
operate and maintain the monitoring system, daily
calibration checks to determine the ongoing validity
of the data and, in some cases, periodic quality
assurance audits.
116
-------
7.3.3.1 Daily Calibration Checks
Daily calibration checks are the primary basis for
evaluating the validity of data on a day-to-day basis.
This is of fundamental importance for many sources
because 1) reporting of emissions data is based on
daily averages, 2) monitor availability may be ex-
pressed as a percentage of valid source operating
days, and 3) monitor control limits and the corre-
sponding requirement to initiate corrective action are
based on the daily calibration check responses.
Also, enforcement of regulations under the Clean Air
Act is based on the specific days during which a
source is considered to be in violation of applicable
regulations. Thus, the daily calibration checks for
CEM systems provide the critical determination of
validity for each day that data are recorded.
As discussed in Chapter 2, the General Provisions of
the Part 60 new source performance standards
include requirements to conduct daily zero and
calibration drift checks for all CEM systems. Specifi-
cally, 60.13 states,
"(d) (1) Owners and operators of
all continuous emission monitoring
systems installed in accordance
with the provisions of this part
shall check the zero (or low-level
value between 0 and 20 percent
of span value) and span {50 to
100 percent of span value) calibra-
tion drifts at least once daily in
accordance with a written proce-
dure. The zero and span shall be
adjusted whenever the 24-hour
zero drift or 24-hour span drift
exceeds two times the limits of
the applicable performance specifi-
cations in Appendix B. The sys-
tem must allow the amount of
excess zero and span drift mea-
sured at 24-hour interval checks
to be recorded and quantified,
whenever specified. ..."
Similar requirements are found in the general provi-
sions for Part 61 and Part 63 NESHAP requirements.
Most state regulations also contain similar provi-
sions.
For monitors that accept calibration gases, the daily
calibration checks usually are performed by conduct-
ing system calibrations using zero and mid- or high-
range calibration gases. Using an upscale calibration
value that approximates either the pollutant con-
centrations or the level corresponding to the emis-
sion standard is preferred. Most, but not all, regula-
tions require that the calibration gases for extractive
systems are introduced at or "as near as practical"
to the probe outlet so that as much as possible of
the sampling system is checked. Some regulations
specify that only gases meeting EPA Protocol 1
(USEPA, 1996a) be used for daily calibrations.
Other regulations do not specify a traceability
protocol or other standard for calibration gases.
Such approaches emphasize the use of the daily
checks as a stability test and rely on initial and
periodic accuracy tests to ensure that monitoring
data are accurate,
For in-situ analyzers, daily calibration checks typi-
cally include zero checks and upscale checks using
the appropriate signals, filters, cells, or other calibra-
tion jig assemblies. The approaches that are used
are quite diverse depending on the type of instru-
ment to be calibrated. The approaches used for
conducting daily calibration checks may rest on a
number of important assumptions or conditions
which may not be known or well understood by the
user. In practice, these assumptions may become
invalid because conditions change that result in daily
calibration procedures that do not provide results
which are representative of monitor performance.
Therefore, a thorough understanding of the daily
calibration check procedure and its limitations is
important to interpret results correctly.
Additionally, many of the approaches that have been
used in the past for daily checks of in-situ analyzers
do not evaluate important functional parts of the
monitoring instrumentation. As previously de-
scribed, temperature compensation circuitry is often
disabled when checking in-situ monitors using cali-
bration gas cells within the analyzer. In this case,
errors arising within the temperature compensation
circuitry (which can significantly bias the CEM
system measurement results) are likely to go unde-
tected. A thorough understanding of the daily cali-
bration check procedure is needed to interpret
results correctly.
An important aspect of the §60.13 (d)(1) regulation
is that it requires a written procedure to be devel-
oped and followed for performing the daily calibra-
tion checks. (For certain sources, this procedure is
included in the quality assurance plan.) This written
procedure should take into account the important
details regarding the calibration of the specific
monitor. For extractive systems, it should indicate
where the gases are introduced, how the values of
check gases are established, provisions to ensure
that the gases are injected at the proper flow rate
117
-------
and pressure, etc. For in-situ monitors, the proce-
dure should specify exactly what is done and it
should contain the important assumptions or other
conditions that must be maintained for operation of
the instrument. Alternatively, the procedure should
prescribe the appropriate actions to take when
unacceptable performance is observed.
Most importantly, the written procedure provides a
standard operating practice to be followed during
the initial performance test and each day that
emissions data are recorded. In this way, initial and
periodic accuracy tests serve to verify the adequacy
of the calibration procedure. Implementing the daily
calibration procedure verifies the data each day until
the next accuracy test is performed. The daily
calibration check procedure must conform with
applicable regulatory requirements and must also
accommodate the technical and practical constraints
of the sampling and analysis system as described
previously in Section 7.2,
For both extractive and in-situ CEM systems, daily
calibration checks are performed and adjustments to
the CEM system are required if the indicated drift
exceeds specified control limits. When such adjust-
ments are made, or when other corrective action is
undertaken, a daily calibration check must be
performed after these activities are completed to
demonstrate that the monitor has been returned to
service properly. Records of adjustments, corrective
actions, and the results of daily drift tests are
required by virtually all regulatory monitoring require-
ments.
7.3.3.2 Quality Assurance/Quality Control
40 CFR 60, Appendix F, Procedure 1 contains QA
procedures for CEM systems used to demonstrate
compliance with emission standards. Although non-
criteria pollutant monitoring is not specified as the
compliance demonstration method in Part 60 regula-
tions, many permits and state regulations applicable
to such monitors adopt Appendix F, Procedure 1 or
very similar provisions. The quality assurance
procedures for CEM systems in the acid rain pro-
gram are contained in Part 75, Appendix B and are
similar in principal to the Part 60 requirements. Part
75 requirements require that all monitors be able to
analyze calibration gases.
Appendix F, Procedure 1 requires that sources
develop and implement a QC program with written
procedures that describe in detail the complete,
step-by-step procedures for calibration of CEM
systems, calibration drift determination and adjust-
ment procedures, preventative maintenance, data
recording calculations and reporting, accuracy audit
procedures, and corrective action procedures for
malfunctioning CEM systems. The results of the
daily drift determinations are used within Procedure
1 to determine if the CEM system is in control.
Procedure 1 specifies that a monitor system is out-
of-control if the zero or high-level response exceeds
either 1 > twice the Appendix B calibration drift
performance specification limit for five consecutive
days, or 2) four times the Appendix B calibration
drift performance specification limit on any one day.
Appendix F, Procedure 1 requires that sources
conduct accuracy audits once each calendar quarter.
For systems that can use calibration gases, the
accuracy audits may include cylinder gas audits
conducted during three of four calendar quarters and
a relative accuracy test conducted during the other
quarter. Alternatively, relative accuracy audits may
be performed instead of cylinder gas audits for three
quarters. A relative accuracy audit is simply a three-
run relative accuracy test evaluating only the mean
difference between the CEM system and the RM,
The relative accuracy test required by Appendix F,
Procedure 1 is identical to the test required during
the initial performance specification (i.e., "certifica-
tion test").
One can see from the above, which is confirmed by
experience, that the selection of a CEM system is
constrained by the applicable QA requirements. The
cost and inconvenience of conducting three cylinder
gas audits per year is much less than the cost and
inconvenience of conducting three relative accuracy
audits per year. The majority of criteria pollutant
CEM systems incapable of analyzing calibration
gases were excluded from compliance monitoring
applications based on this requirement.
7.5,4 CEM System Approval Mechanisms
in Germany
Other approval mechanisms have been used interna-
tionally that may serve as models or may contain
elements that are useful for certification of non-
criteria CEM systems. For example, Germany has
implemented a rigorous and comprehensive approval
process for many years for CEM systems installed
for regulatory purposes. Frequently when monitor-
ing equipment used in Europe is first offered for sale
in the United States, potential suppliers will claim
that the measurement system is "TUV certified,"
Various control agencies have indicated interest in
these certification procedures and they may be
useful in the evaluation of non-criteria pollutant
monitors in the absence of EPA approval procedures.
The German approach to monitor approval and TUV
118
-------
certification often are misunderstood and sometimes
misrepresented in the United States. For these
reasons, information regarding CEM system QA
requirements in Germany are described below.
Monitoring equipment must first successfully com-
plete certification testing by the TUV before it can
be sold for use in regulatory programs. This is a
rigorous technical evaluation of each equipment
model to ensure that it is suitable for particular
applications. In addition, testing and evaluation is
performed for each monitor at the time that it is
installed and this testing is repeated every three to
five years depending on the type and size of the
facility. This initial calibration program for monitors
in Germany is similar to performance specification
tests for monitors in the United States. Additional
tests are performed by an independent authority on
an annual basis to verify performance of the moni-
toring equipment. Finally, maintenance procedures
are performed by the facility's personnel on an
ongoing basis in accordance with the manufacturer's
instructions and results from the suitability test.
7.3.4.1 Initial TUV Certification (Suitability) Tests
Each instrument model must first pass a Technischer
Uberwachungs-Verein (TUV) certification before it
may be offered for sale in regulatory monitoring
applications. The TUV certification is a rigorous
evaluation requiring at least three to six months to
complete and includes both laboratory evaluations
and field suitability testing. The fee for a TUV
certification test is substantial (typically S 50,000 to
§100,000) and is paid for by the monitor manu-
facturer. Repeating portions or all of the certifica-
tion test increases the cost to the manufacturer.
Basic TUV monitor performance specifications are
listed in the Table 7-2, The reproducibility specifica-
tion demands that two randomly selected instru-
ments provide readings with a mean difference of
less than 3.3% over a three-month period. The
reproducibility specification also ensures that instru-
ments of the same model number are interchange-
able. Zero and calibration drift are limited to ±2%
over the maintenance interval. The instrument must
provide an alarm indicating the need for immediate
maintenance when zero or calibration drift limits are
reached. The time needed to reach these limits
defines the maintenance interval. The maintenance
interval is recommended by the manufacturer but is
determined by the TUV as part of the suitability
tests. The maintenance interval should be at least
one week; however, four weeks is preferable.
Determination of the maintenance interval based on
suitability tests and drift limits (relative to the
maintenance interval) provides incentives for manu-
facturers to build stable instruments.
Tests to determine the effects of line voltage varia-
tions, ambient temperature variations, and other
factors that may influence monitor performance are
also performed as part of the TUV certification.
The TUV may investigate other areas or issues
that are considered to be relevant to the perfor-
mance of a particular measurement system. Field
suitability tests are performed for each type of
application (e.g., gas-, oil-, coal-, refuse-fired boilers)
but in some cases, success at more difficult monitor-
ing applications is taken as a sufficient demonstra-
tion for less demanding applications.
7.3.4.2 Initial and Periodic Monitor Calibration
Tests
These tests are in many ways similar to CEM system
performance specification tests performed in the
Table 7-2. Principal Performance Specifications for TUV Suitability Tests of Emission Monitoring
Instruments
System Parameter
Analytical Function {ma vs. concentration)
Detection Limit
Reproducibility
Zero and Calibration Drift
Availability
Interference from other species
Maintenance interval
Criteria
By Reference Method (regression analysis)
2% of most sensitive range
R = measurement range ;> 30 mean difference
± 2% in maintenance interval
90% in three months minimum; 95% is the goal
± 4% of full scale
Determined by test program (limited to three months
maximum}
119
-------
United States. However, a number of important
differences exist in both the philosophy, nomen-
clature, and procedures used for these tests. The
term "calibration" as used in VDI 3950, Part 1,
"Calibration of Automatic Emission Measuring Instru-
ments" refers to the entire process of establishing and
verifying the correct performance of a monitoring
system (VDI 3950). This includes: 1) procedures for
the selection and preparation of measurement sites,
2) procedures to check for proper Installation of
equipment, 3) leak checks of extractive monitoring
systems, 4) a five-point verification of the instrument
characteristic (similar to calibration error in the United
States using zero and upscale calibration gases), 5)
monitor-specific procedures for evaluating measure-
ment interferences, 6} zero and reference (upscale
calibration) drift stability tests, 7) response time tests,
8) procedures for measuring and processing data, 9}
procedures for checking the representativeness of
sampling points, 10} procedures for determining the
analytical function for the complete measuring system
relative to independent flue gas measurements
(reference methods), and 11) procedures for analysis
and reporting of the calibration results.
In the United States, many of the calibration issues
that are specifically addressed by VDI 3950, Part 1
are either lumped together and evaluated as a group
during the relative accuracy test or are simply ignored.
For example, the German procedures evaluate sepa-
rately: 1) the representativeness of the sampling
points, 2} the influence of interferences, and 3) the
system performance relative to independent emission
measurements through a series of tests. In the United
States, we rely on a single relative accuracy test to
determine the acceptability of all of these factors
taken together. The German "calibration" procedure
also relates test procedures and results back to the
initial TUV certification test. For example, the calibra-
tion interference test procedures specifically address
the measurement interferences identified during the
initial certification because this is obviously dependent
on the specific monitor design and analytical principal.
Also, the zero and reference point drift determinations
are determined relative to the maintenance interval
established during the initial certification test rather
than an arbitrary time interval.
The German comparison with independent reference
methods relies on a minimum of fifteen paired mea-
surements (CEM and reference method} conducted at
different emission levels to facilitate a regression
analysis that is used to determine subsequent emis-
sion levels. Linear or quadratic regression analysis
may be used and calculations of the confidence
coefficient and tolerance intervals are included.
These statistical quantities also are employed in the
interpretation of the monitoring data.
The field calibration test is performed by an independ-
ent expert agency (the TUV) authorized by the
German government to perform these evaluations.
However, the TUV office that performs the calibration
test at a particular industrial facility is not necessarily
the TUV office that performed the initial certification.
Again, the field calibration procedure is performed
initially and then is repeated every three to five years
depending on the type and size of the facility.
7.3.4.3 Annual Calibration Tests and Maintenance
Section 8, "Periodical Functional Test" of VDI 3950,
Part 1 describes the annual evaluation requirements.
These include; 1) checks of the operational status
(leak checks, optical contamination, etc.), 2) records
review of zero and reference point checks, 3) zero
and reference drift tests, 4) checks for interferences,
5) multi-point verification of the instrument character-
istic (calibration error test) using zero and upscale
calibration gases or other calibration materials, 6)
certain monitor functional tests, 7) a minimum 4-run
comparison with independent reference methods, and
8) inspection of the data transmission to the chart
recorder, integration device, or data logging system.
These are more extensive evaluation procedures than
are required for monitors subject to EPA regulations.
These procedures also are performed by an independ-
ent licensed agency (TUV) rather than by source
personnel. Maintenance is performed by the industrial
facility's trained personnel. Minimum procedures are
specified for in-situ devices and for extractive sys-
tems.
7.3.5 International Standards Organization
International Standards Organization (ISO) has devel-
oped standards for certain monitoring applications.
ISO Standard 7935 is for S02 CEM systems and the
standard for NOX CEM systems is currently in draft
form (ISO, 1991). A discussion of these standards
for criteria pollutant monitoring and a comparison with
United States regulations is presented by Jahnke
(Jahnke, 1993). Other ISO standards for emissions
test methods are under development.
The ISO committee TC146/SC1/WG1 has prepared
ISO standard 10155 "Stationary Source Emissions,
Automated Monitoring of Mass Concentration of
Particles - Performance Characteristics, Test Proce-
dures, and Specification". The standard, prepared
over the last 10 years, was published in its final form
on April 1, 1995 (ISO, 1995). It does not prescribe
a particular method or analytical technique but instead
reflects a general approach and provides performance
120
-------
specifications to evaluate any specific method that is
offered. Fundamental to the ISO standard is the
requirement to perform a site-specific calibration or
correlation to ISO manual method 9096 (which is
similar to EPA isokinetic participate sampling methods)
(ISO, 1991). As should be expected, ISO 10155
specifies that the calibration procedure must be
repeated when changes in emission controls, fuel
type, or other factors occur that may be expected to
influence the calibration. Although this is a general
requirement, it reflects the fact that the need for
recalibration is best based on practical judgment.
ISO 10155 prescribes several performance speci-
fications including the following:
* Response time of less than 1/10 of the manual
sampling time
» Zero and span drift less than ±2% of working
range per month
* Accumulated automatic zero and span adjust-
ments less than 6% of working range
* Calibration (correlation) line specification
1, Correlation coefficient 2:0,95
2. 95% confidence interval shall be s10% of
emission standard
3. Tolerance interval; 95% confidence that
all possible values are within ±25% of
emission standard
The ISO 10155 standard requires that sample runs be
performed at three different emission levels to estab-
lish the calibration. A minimum of 9 sampling runs
must be performed but 12 or more runs typically are
expected. The ISO standard prescribes that the
process operating conditions should be varied if
possible to create the different emission levels.
Where this is not possible, variation of the control
equipment operating parameters to create a range of
emission levels is accepted.
The Central European Normalization Committee {CEN
committee TC264/WG55 is developing requirements
applicable to continuous paniculate monitoring
{Peeler, 1996, b). Unlike ISO, which is a voluntary
organization of participating countries, CEN mandates
requirements for the 14 countries comprising the
European Community. The CEN committee will
establish: 1) emission standards and limits, 2) manual
test methods, and 3) automated monitoring methods,
Each participating country must adhere to the CEN
requirements or adopt more restrictive/more rigorous
requirements. CEN typically adopts ISO standards/
methods where available. The CEN committee has
adopted the ISO 10155 continuous particulate moni-
toring standard and applied it as a requirement for
hazardous waste incinerators.
The calibration line specifications are illustrated in
Figure 7-6,
80
60-
0>
tf!
I40
0,95)
Tolerance Interval
(9i% confidence all values
within * 25% of Std.)
.02 ,04
Extinction
Figure 7-6. Continuous particulate monitoring calibra-
tion line specifications.
7.4 Suggested Approval Mechanisms and
Approaches for Non-criteria Pollutant
and Application Testing
The requirements and procedures discussed in Sec-
tions 7.3.1-7.3.3 for conducting initial performance
specification tests and audits are consistent with
existing regulatory programs and can be used for
many non-criteria pollutant monitoring applications.
However, some of these procedures do not address
problems fully that may be encountered in evaluating
non-criteria CEM systems. New evaluation proce-
dures likely will be needed as technology continues to
evolve and as monitoring applications for additional
specific non-criteria pollutants expand through regula-
tions, permit requirements, or market based trading
programs. Several alternative approval mechanisms
that may be applicable in these situations are de-
scribed in this section.
7.4.1 EPA Method 301
Method 301 of Part 63
-------
initially was developed to allow a specific source to
demonstrate the validity of a hazardous air pollutant
test method for that source to qualify for an "early
reduction" option. Method 301 was intended to
encourage development of methods where methods
did not already exist. The practical application of
Method 301 has been expanded greatly beyond its
original purpose; it now serves as a model for evaluat-
ing many measurement methods for both criteria and
non-criteria air pollutants.
Method 301 incorporates procedures to determine if
the precision and bias of a candidate method are
acceptable based on comparisons with a validated
method, use of isotopic spiking, or analyte spiking.
Method 301 is directed primarily at traditional source
testing approaches that include discrete sampling and
analysis phases. The implementation of the procedure
requires the use of quad-trains or paired sampling
trains to evaluate precision and bias. Alternative
procedures have been developed and accepted by the
EPA to allow the application of Method 301 to direct
interface methods (methods where the effluent is
directly injected into the analyzer) such as instrumen-
tal methods using FTIR and GCMS analytical tech-
niques (USEPA, 1994). These alternative Method 301
procedures include collecting a series of spiked and
unspiked effluent data for calculation of the precision
and bias of a method. These analyte spiking proce-
dures have been used to demonstrate the acceptabil-
ity of various test methods used by industry in MACT
standard development programs (Kinner, 1996,
LaCoss, 1995).
Note: Method 301 is only a source specific test
method determination. Method 301, Section 12
includes a general discussion regarding "conditional
approval" of a method which might allow the transfer
of results to additional sources. Conditional approval
waiver requirements include: method documentation,
ruggedness tests, sample stability, and practical limits
of quantification. EPA has granted test method
approval to groups of sources in several circum-
stances. A voluntary administrative procedure
documenting the applicant's and EPA responsibilities
for instrumental field test methods has been proposed
by an instrument vendor, and is being evaluated for a
direct interface gas chromatograph mass spectrometer
method (Peeler, 1996).
As noted above, Method 301 can be adapted to
evaluate direct interface or other instrumental test
methods. Such methods are similar to CEM systems
in that successive samples are acquired and analyzed.
However, instrumental test methods generally involve
a much more detailed protocol and many more
specific procedures than are associated with a CEM
system. Also, an instrumental test method usually
requires the full time attendance of an operator as
compared to a CEM system, which is designed to
function automatically with very little human interven-
tion. Because of these differences, many modifica-
tions to Method 301 are necessary to apply it to the
evaluation of CEM systems.
7,4,2 Dynamic Analyte Spiking
Dynamic analyte spiking involves the quantitative
introduction of a calibration gas of known concentra-
tion containing one or more analytes, to an effluent
sample stream. The spike gas represents only a small
fraction of the combined stream. Therefore, the
spiked sample stream contains essentially the same
constituents at the same concentrations as the
unspiked samples. A comparison of spiked and
unspiked samples provides an evaluation of both the
effectiveness of the sampling system and the perfor-
mance of an analytical system.
Because of the similarity of the spiked and unspiked
samples, "matrix effects" (interference effects due to
the presence of other sample constituents such as
moisture, oxygen, carbon dioxide, etc.) can be
assessed by this procedure. The analyte spiking
approach is particularly useful in the evaluation of
non-criteria pollutants that are reactive, condensable,
water soluble, or that have the tendency to polymer-
ize in the sampling system. These types of com-
pounds represent the most difficult measurement
challenges and can invalidate traditional approaches
for assessing monitor performance.
Selection of the spike analytes depends upon two
items: the number of target analytes, and the ability
to obtain these compounds in gaseous form. For
multi-component CEM systems, anaiyte spiking with
all of the target analytes may not be economically
feasible. In this case, some surrogates that represent
the analytes must be chosen. Selection of surrogates
depends on their ability to represent more than one of
the target analytes based on similar physical proper-
ties.
Dynamic analyte spiking is a more rigorous evaluation
procedure than the traditional system calibration.
Simply introducing dry calibration standards in a
system calibration is useful to check for leaks, adsorp-
tion/desorption in the sampling system and to check
the analyzer calibration. However, even the straight-
forward absorption of analytes by condensate formed
from sample moisture cannot be detected in the
system calibration because dry calibration gases are
used. Similarly, the system calibration cannot detect
122
-------
analytical interferences due to anaiyte interaction with
other components present in the effluent samples
because the calibration gas does not usually contain
these components.
By performing dynamic anaiyte spiking, the effective-
ness of the sample delivery system and the analytical
components are assessed more thoroughly for the
compounds of interest in the presence of actual
source matrix interferences and moisture. This
procedure should be performed in conjunction with
both direct and system calibrations. Together these
procedures provide a check of the CEM system both
on a dry basis and under actual sampling conditions.
7.4,2.1 Anaiyte Spike Procedure
Any CEM system that has the capability to accept
calibration gases at a point upstream of the particu-
late filter is a candidate for anaiyte spiking. The spike
gas is delivered into the sampling system, typically in
a ratio of 1 part spike to 9 parts sample gas, resulting
in a ten-fold dilution of the spike gas. The spike
should not exceed this 1:10 ratio to avoid excessive
dilution of the analytes of interest or substantially
changing the sample matrix. Figure 7-7 is a general-
ized schematic of the anaiyte spiking technique.
The spike gas should be preheated to prevent local-
ized condensation at the point of injection. The spike
must be delivered to the sampling system at a point
upstream of the paniculate matter filter to detect
possible gas reactions with the accumulated particu-
late material on the filter.
Calibrated mass flow meters or controllers are neces-
sary to deliver the spike at a precise, measured flow
rate. A calibrated rotometer for measurement of total
flow is also necessary. The rotometer should be
installed in the sample delivery system at a point that
provides an accurate measurement of total sample
flow. Experimental errors associated with calculating
the dilution factor of the anaiyte spike arise from the
inaccuracy of the spike and total sample flow mea-
surement devices (see Appendix G, Equation G-2). In
addition, calibration of the total flow measurement
device with wet stack gas can be difficult or impracti-
cal. In addition, the effects of errors associated with
the anaiyte spike calibration standard values are also
magnified.
As an alternative to the above, the measurement of
the spike dilution factor may be determined directly
through measurements of a tracer compound con-
tained in the anaiyte spike gas. A tracer must be
chemically inert and not present in the source efflu-
ent. Sulfur hexafluoride (SF6| has been used success-
fully as a tracer in extractive FTIR test methods, and
is included in draft Methods 318 and 320. (Method
318 may be proposed as an FTIR test method for
inclusion in 40 CFR Part 63.) The amount of spike
gas introduced during a spiking experiment can be
determined from the concentrations of the anaiyte
and tracer components of the calibration standard.
(See Appendix G). Errors associated with use of the
tracer technique are associated with the analyzer's
ability to measure the tracer gas concentration
accurately and errors associated with the anaiyte and
tracer concentrations in the spike gas standard.
From
Probe
Flue Gas Flow=Qi
'• • !
•" :|t
• "•••
]
: Primary
Particulate
Filter
I r,a
w
lihrat
Total System Flow
QT = Qi + Qa
Analyzer
Pump
Calibration Gas Flow = Qs
Calibration Gas
Introduction Assembly
Calibration gas flow = Qz
ANALYTE SPIKE
Flue Gas Flow Qi = 0.9 QT
Spike Flow Qa = 0.1 QT
txj-
Direct Calibration
Calibration Gases
Figure 7-7. Anaiyte spiking.
123
-------
Acceptable measurement system performance
should be based upon the intended use of the data.
However, analyte spike recovery values of ±30% of
the expected value are typical. These limits are
consistent with the Method 301 bias limits and have
been applied in determining the acceptability of
direct interface instrumental test methods (Powell,
1996). Analyte spike recoveries and their calcula-
tions are discussed in further detail in Appendix G.
REFERENCES
Brown, Thomas D. 1996. "ND Energy and Environ-
mental Research Center for Air Toxic Metals,"
Newsletter, Volume 2, Issue 2.
Cone, L.A. 1989. "Carbon Monoxide and Total
Hydrocarbon Continuous Monitoring at Hazardous
Waste Incineration Facilities," Proceedings Air &
Waste Management Association CEM Specialty
Conference, Chicago, IL, pp. 338-349.
International Standards Organization. 1991. Sta-
tionary Source Emissions Determination of Mass
Concentration of Sulfur Oxides - Performance Char-
acteristics of Automated Measurement Systems.
ISO 7935, {Available from American National
Standards Institute, New York, NY.)
International Standards Organization. 1995. Auto-
mated Monitoring of Mass Concentration of Particles
in Stationary Source Emissions: Performance
Characteristics, Test Procedures and Specifications.
ISO 10155. (Available from American National
Standards Institute, New York, NY.)
Jahnke J.A. 1994. An Operators Guide to Eliminat-
ing Bias in CEM Systems, EPA/400/R94/016,
Washington, D.C., U.S. Environmental Protection
Agency.
Jahnke J.A. 1993. "In-situ Monitoring Systems for
the Measurement of Gas Concentrations and Flue
Gas Velocities," in Continuous Emission Monitoring,
Van Nostrand Reinhold, NY, pp 156-199.
Kinner, L.L., Peeler, J.W. 1996. "Development of
Fourier Transform Infrared Spectrometry and Gas
Filter Correlation Infrared Spectrometry Test Method
Protocols for the Portland Cement Industry," Pro-
ceedings Air & Waste Management Association,
Nashville, TN, 96-TA37.03.
Kinner, L.L., Peeler, J.W. 1995. "Protocols for
Measurement of Gaseous Volatile Organics and HCl
from Cement Kilns by FTIR and GFCIR," Available
from the Portland Cement Association, Skokie, IL.
Kinner, L.L., Geyer, T.J., Plurnrner, G.M., Dunder,
T.A. 1994. "Application of FTIR as a Continuous
Emission Monitoring System," Proceedings Air &
Waste Management Association Specialty Confer-
ence on International Incineration, Houston, TX, pp.
178-196.
LaCoss, J., McCarthy, J. 1995. "Measurement of
Air Toxics Using Extractive FTIR Spectroscopy,"
Proceedings Emerging Technologies for Hazardous
Waste Management, Special Symposium of Indus-
trial and Engineering Chemistry, Division of Ameri-
can Chemical Society, Atlanta, GA, paper no.
194.3.
Laudal et. al. 1996. "Mercury Speciation: A
Comparison between EPA Method 29 and Other
Sampling Methods," Proceedings Air & Waste
Management Association, Nashville, TN, paper no.
96-WA64A.04.
NIOSH. 1994. NIOSH Manual of Analytical Meth-
ods, 4th Edition. U.S. Department of Health and
Human Services, Public Health Service, Centers for
Disease Control and Prevention, National Institute
for Occupational Safety and Health, Division of
Physical Sciences and Engineering, Cincinnati, Ohio,
August, 1994.
Peeler, J.W., Kinner, L.L., DeLuca, S. 1996. "Gen-
eral Field Test Method Approval Process and Spe-
cific Application for a Direct Interface GCMS Source
Test Method," Proceedings Air & Waste Manage-
ment Association, Nashville, TN, paper no. 96-
RP132.01.
Peeler, J.W., Jahnke, J.A., and Wisker, S.M. 1996.
Continuous Particulate Monitoring in Germany and
Europe Using Optical Techniques. Continuous
Compliance Monitoring Under the Clean Air Act
Amendments, Air & Waste Management Associa-
tion. Pittsburgh, PA, pp. 208-220.
Powell, J.H., Dithrich, E.G. 1996. "Hot-Wet
Instrumental Hydrogen Chloride Emissions Quan-
tification Using GFCIR Method Validation and
Comparison," Proceedings Air & Waste Management
Association Specialty Conference on Boilers and
Industrial Furnaces, Kansas City, MO.
Richards, Marta K., et, al. 1996. "Performance
Tests of Mercury Continuous Emissions Monitors at
the U.S. EPA Incineration Research Facility,"
124
-------
Proceedings Air & Waste Management Association,
Nashville, TN, paper no. 96-WA64A.02.
Roberson, R. 1996. "Status of CEM Systems for
HAP Emissions," Proceedings Electric Power Re-
search Institute CEM Users Group, Kansas City, MO.
Seme, J.C., White, M.O., Burdette, J.W. 1993.
"Recent Experience Measuring Emissions of Alde-
hydes and Ketones Using EPA Method 001," Pro-
ceedings Air & Waste Management Association
Waste Combustion in Boilers and Industrial Furnaces
Specialty Conference, Clearwater, FL.
U.S. Environmental Protection Agency. 1977.
Traceability Protocol for Establishing True Concen-
trations of Gases Used for Calibration and Audits of
Continuous Source Emissions Monitors (Protocol
Number 1], June 1978. Section 3.0.4 of the
Quality Assurance Handbook for Air Pollution
Measurement Systems. Volume III. Stationary
Source Specific Methods. EPA 600/4-77-027b.
U.S. EPA, Office of Research and Development
Publications, 26 M.L. King Dr., Cincinnati, OH
45268.
U.S. Environmental Protection Agency. 1981. A
Procedure for Establishing Traceability of Gas
Mixtures to Certain National Bureau of Standards
Reference Materials. Joint publication NBS/EPA-
600/7/81-010. Available from U.S. EPA, Quality
Assurance Division (MD-77), Research Triangle Park,
NC.
U.S. Environmental Protection Agency. 1986. Test
Methods for Evaluating Solid Waste, Physical/Chemi-
cal Methods, SW-846 Manual, 3rd, ed. Document
no. 955-001-0000001. Available from Superinten-
dent of Documents, U.S. Government Printing
Office, Washington, D.C.
U.S. Environmental Protection Agency. 1993. Field
Validation Testing at a Coal Fired Boiler, U.S. EPA
Report, EPA Contract No. 68D20163, WA No. 2.,
Research Triangle Park, NC.
U.S. Environmental Protection Agency. 1994. Field
^Validation Test Using FTIR Spectrometry to Measure
Formaldehyde, Phenol, and Methanol at a Wool
Fiberglass Production Facility, United States Environ-
mental Protection Agency Report, EPA Contract No.
68D20163, WA No. 32.
U.S. Environmental Protection Agency. 1995.
Protocol for the Use of Extractive Fourier Transform
Infrared (FTfRl Spectrometry in Analysis of Gaseous
Emissions from Stationary Industrial Sources, United
States Environmental Protection Agency, Research
Triangle Park, NC.
U.S. Environmental Protection Agency. 1996a.
Code of Federal Regulations. Continuous Emission
Monitoring, Appendix H to Part 75 - Revised Trace-
ability Protocol No. L 40 CFR 75 Appendix H.
Washington, D.C.
U.S. Environmental Protection Agency. 1996b.
Code of Federal Regulations. Method 318, Extrac-
tive FTIR Method for the Measurement of Emissions
from the Mineral Wool and Wool Fiberglass Indus-
tries, 40 CFR 63 Appendix A., Washington, D.C.
U.S. Environmental Protection Agency. 1996c.
Revised Standards for Hazardous Waste Combus-
tors, Proposed Rule, 61 Federal Register 17358.
U.S. Environmental Protection Agency. 1996d.
Code of Federal Regulations. Method 301, Field
Validation of Pollutant Measurement Methods from
Various Waste Media, 40 CFR, Part 63 Appendix A.,
Washington, D.C.
VDI Guideline 3950. 1994. "Calibration of Auto-
matic Emission Measuring Instruments," Part 1.
Available from Beuth Verlag Gmbh, 10772, Berlin.
125
-------
Appendices
Appendix A
Acronyms
A&WMA Air & Waste Management Association
BACT Best Available Control Technology
BBS Bulletin Board System
BIF Boiler and Industrial Furnace
BTU British Thermal Unit
BTX Benzene, Toluene, Xyienes
CAA Clean Air Act
CAAA Clean Air Act Amendments
CAM Compliance Assurance Monitoring
CC Confidence Coefficient
CEM Continuous Emission Monitoring
CFR Code of Federal Regulations
CTS Calibration Transfer Standard
DAHS Data Acquisition and Handling System
DAR Data Assessment Report
DCPL Demonstrated Compliance Parameter Limit
DCS Distributive Control System
DEQ Department of Environmental Quality
DER Discrete Emission Reductions
DNPH Dinitro phenol hydrazine
DOAS Differential Optical Absorption Spectroscopy
DOS Disk Operating System
DRE Destruction and Removal Efficiencies
ECD Electron Capture Detector
EMC Emission Measurement Center
EMTiC Emission Measured Technology Information Center
EPA Environmental Protection Agency
EPR1 Electric Power Research Institute
ERC Emission Reduction Credit
FID Flame lonization Detector
FR Federal Register
FTIR Fourier Transform Infrared Spectroscopy
GC Gas Chromatography
GCMS Gas Chromatography Mass Spectrometry
HAP Hazardous Air Pollutant
HON Hazardous Organic NESHAP
I/O Input/Output
1R Infrared
ISO international Standards Organization
LAN Local Area Network
126
-------
LTD Long Tons Per Day
MACT Maximum Achievable Control Technology
MARAMA Mid-Atlantic Regional Air Management Association
NDIR Nondispersive infrared
NDUV Nondispersive ultraviolet
NESCAUM North East States Consortium for Air Use Management
NESHAP National Emission Standards for Hazardous Air Pollutants
NIOSH National Institute of Occupational Safety and Health
NIST National Institute of Standards and Technology
NSPS New Source Performance Standards
OAQPS Office of Air Quality Planning and Standards
O&M Operation and Maintenance
OMT Open Market Trading
OMTG Open Market Trading Guidance
OMTR Open Market Trading Rule
PEEK Polyether ether ketone
PEM Predictive Emission Monitoring
PID Photoionization Detector
PLC Programmable Logic Controller
PS Performance Specification
PST Performance Specification Test
QA Quality Assurance
QC Quality Control
RA Relative Accuracy
RATA Relative Accuracy Test Audit
R&D Research and Development
RCRA Resource Conservation and Recovery Act
RECLAIM Regional Clean Air Incentives Market
RTD Resistance Temperature Device
RTU Remote Terminal Unit
SCAQMD South Coast Air Quality Management District
S/N Signal to Noise
TC Technical Committee
TCD Thermal Conductivity Detector
THC Total Hydrocarbon
TUV Technische Uberwachung Verein
TRS Total Reduced Sulfur
TTN Technology Transfer Network
USEPA United States Environmental Protection Agency
UV Ultra-violet
VD1 Verein Deutscher Ingenieur
VOC Volatile Organic Compound
WAN Wide Area Network
WG Working Group
WTE Waste To Energy
127
-------
Appendix B
Glossary
Accuracy - The closeness of the measurement made by a continuous emission monitoring system, a
pollutant concentration monitor or a flow monitor, to the true value of the emissions or volumetric flow. It
is expressed as the difference between the measurement and a reference method value, which is assumed
to be equivalent to the true value. Variation among these differences represents the variation in accuracy
that could be caused by random and/or systematic error.
Analyte - A compound or set of compounds that are to be measured by an analytical method.
Analyze - To conduct a measurement and arrive at a specific result or set of results.
Analyzer - An instrument that is capable of performing the measurement of the compounds of interest and
generating an output proportioned to the concentration of the analyte.
Acid Gas - A gas comprised of compounds such as S02, HC1, or H2SO4 that is corrosive in nature and that
can be difficult to sample.
Bias - Systematic error. The result of bias is that measurements will be either consistently low or high,
relative to the true value.
Elowback - A procedure conducted periodically by some extractive sampling systems during which
compressed air is blown out of the sample probe to remove accumulated paniculate matter from the probe
tip.
Calibration - The procedure for adjusting the output of a device to bring it to a desired value (within a
specified tolerance} for a particular value of input (typically the value of the reference standard).
Calibration check - The procedure of testing a device against a known reference standard.
Calibration Drift - The difference between 1! the response of a gaseous monitor to a calibration gas or
standard and the known concentration of the gas or standard, or 2) the response of a flow monitor to a
reference signal and the expected value of the reference signal, or 3) the response of a continuous opacity
monitoring system to an attenuation filter and the previously determined value of the filter after a stated
period of operation during which no unscheduled maintenance, repair, or adjustment took place.
Calibration Gas - A gas of known concentration that is traceable to either a standard reference material gas
or a National Institute of Standards and Technology or whose concentration is established by an analytical
method, on a manufacturer's certification.
Calibration Gas Cell or a Filter - a device that, when inserted between the transmitter and detector of the
analyzer, produces a desired output level on the data recorder.
Chemiluminescence - Loss of energy by a chemically excited molecule that results in emission of electro-
magnetic radiation at a particular wavelength. The energy of the radiation is indicative of the amount and
type of species that luminesce.
128
-------
Chillers - See condenser systems.
Close-coupled - An extractive CEM system that is installed at the sampling location.
Condensate - The resultant water and water soluble compounds that are removed from a flue gas sample by
condenser systems.
Condenser Systems - A system designed to physically remove the moisture from the flue gas sample stream
by cooling before introduction into an analyzer.
Condensible Gas - A gas that has chemical and physical properties that allow it to change from a gas to a
liquid in sampling systems before analysis.
Continuous Emission Monitoring System - The equipment used to analyze, measure, and provide, on a
continuous basis, a record of flue gas emissions.
Corrective Action Codes - Entered by the data acquisitions system operator to describe actions taken to
correct CEM fault conditions or emission exceedances,
Data Acquisition and Handling System (DAHS) - A system of hardware and software that is used to collect
and store emissions data from gas analzyers and to produce summary reports. For regulatory purposes, the
DAHS is considered to be an integral part of the CEM system.
Data Acquisition System - One or more devices used to receive, compute, store, and report CEM system
measurement data from single or multiple measurement devices.
Data Assessment Report (DARJ - For a CEM system subject to 40 CFR 60, Appendix F, quality assurance
requirements, the DAR is a quarterly report which includes all accuracy audit results, reasons for downtime,
and corrective actions.
Data Availability - A statistic used to indicate the percentage of plant operating time for which valid CEM
system data are available. Some monitoring regulations or permit conditions have minimum data availability
requirements for each quarter of the year.
Data Flags - A code that the data acquisition system associates with each data point to identify whether the
data point is valid, invalid, or questionable. Data flags are used to determine which data points are included
in emission averages.
Data Logger - A simple digital data recorder that can be used to convert gas analyzer output signals to
concentrations and emission rates and to calculate averages. Emission results may be printed real-time or
stored in memory for later retrieval by computer.
Data Recorder - A device capable of providing a permanent record of data.
Desulfurization - Removal of sulfurous compounds in stack emissions. Often accomplished at utility boilers
by use of lime/limestone scrubbers.
DeNox - Removal of oxides of nitrogen from flue gas using selective catalytic reduction (SCR) or non-
selection catalytic reduction (non-SCR).
Detector - The device used to sense an analyte in a monitoring system,
Detection Limit - The lower level of quantification achievable by a particular measurement.
129
-------
Diffusion - The transport of a liquid or gaseous substance through a solid material.
Diluent Gas - A major gaseous constituent in a gaseous pollutant mixture. For combustion sources, carbon
dioxide, nitrogen, and oxygen are the major diluent gases.
Direct Calibration - Introduction of the calibration gas directly to an analyzer without passing it through the
sampling system,
Distributed Control System (DCS) - A type of computer system used in plant environments for process
control. It relies on a system of independent processors that are linked to a central computer.
Drift - Change in analyzer output, over a period of time, that is unrelated to input or equipment adjustments.
Dual Range System - A pollutant concentration monitor that has two distinct ranges of values over which
measurements are made.
Emission Standard - The maximum emission level, in specified units, and averaging period allowed by an
environmental regulation.
Extractive System - A monitoring system that withdraws a gas sample from the stack and transports the
sample to the analyzer.
Flow Monitor - A stand alone monitor, or a component of the continuous emission monitoring system that
generates an output proportional to the volumetric flow of exhaust gas.
Hazardous Air Pollutants {HAPs} - Specific pollutants and groups of pollutants listed in Title III of the 1990
Clean Air Act Amendments. (See Appendix D of this handbook.)
Hydrocarbons - Compounds composed of carbon and hydrogen.
In-situ Monitor - A monitor that senses the gas concentration, particulate concentration, opacity, or
velocity in the flue gas and does not extract a sample for analysis.
Interference Rejection - The ability of a CEM system to measure a gaseous species, within specified limits,
without responding to other gases or substances present in the flue gas.
Invalid Data - Data that were generated while the measurement device(s) was out-of-control.
Linearity - The degree to which a CEM system exhibits a straight line (first order) response to changes in
concentration (or other monitored value), over the range of the system.
Lower Detection Limit - The minimum value that a device can measure,
Mass Flow Meter/Controller - A device that is used to measure precisely known volumes for flow rates of
gas.
Measurement Cell - The chamber where a gas sample is subject to analysis.
Opacity - The degree to which a flue gas stream reduces the transmission of visible light or obscures the
visibility of an object in the background.
Path Continuous Emission Monitoring System - A continuous emission monitoring system that measures the
pollutant concentration along a path greater than 10% of the equivalent diameter of the stack or duct cross
section.
130
-------
Point Continuous Emission Monitoring System - A continuous emission monitoring system that measures the
pollutant concentration either at a single point or along a path equal to or less than 10% of the equivalent
diameter of the stack or duct cross section.
Parts Per Million (ppml - One part in one million total parts (1x10"6).
Parts Per Billion (ppbj - One part in one billion total parts (1x10"9}.
Path Length - The distance a light beam travels through the a sample gas before reaching the detector.
PEEK™ - A tubing (poly ether ether ketone) substance manufactured to be inert relative to chemical
adsorption.
Permeation - Diffusion of a gaseous substance through a solid material.
Precision - The closeness of a measurement to the actual measured value expressed as the uncertainty
associated with repeated measurements of the same sample or of different samples from the same process
(e.g., the random error associated with simultaneous measurements of a process made by more than one
instrument), A measurement technique is determined to have increasing precision as the variation among
the repeated measurements decreases.
Programmable Logic Controller (PLC} - An electronic device that can be used to automatically control
sampling and calibration cycles and alarm condition responses in a CEM system.
Reason Codes - Entered by the data acquisition system operator at the time of a CEM malfunction or
process upset, they are used to describe the reasons for invalid data or emission limit exceedances.
Reference Spectra - Spectra that have been acquired under specific conditions and stored for later use.
Reference value - The known concentration of a verification standard or calibration gas or the known value
of a reference thermometer or output value of a temperature, pressure, current or voltage calibrator.
Relative Accuracy - The absolute mean difference between the gas concentration or emission rate
determined by a CEM system and the value determined by an appropriate reference method plus the 2.5
percent error confidence coefficient of a series of tests, divided by the mean of the reference method tests.
The relative accuracy provides a measure of the systematic and random errors associated with data from a
continuous emission monitoring systems.
Remote Terminal Unit (RTU) - An electronic device that is used to periodically report emissions data to the
control agency by telephone modem.
Response Time - The amount of time required for the continuous emission monitoring system to display on
the data recorder, 95% of a step change in pollutant concentration. This period includes the time from
when the sample is first extracted from the stack (if any extractive system is used) to when the concentra-
tion is recorded.
Span Gas - A high range calibration gas.
Span - The algebraic difference between the upper and lower range values of the monitoring system or
analyzer.
Standard Reference Material - A reference material distributed and certified by the National Institute of
Standards and Technology (NIST).
131
-------
System Calibration - A calibration procedure that directs gas through the entire sampling system.
Stripchart Recorder - A device for recording data in graphical form on a continuous chart.
Test Method - Any method of sampling and analyzing for a substance or determining the flow rate as
specified in the applicable regulations.
VEO (Visible Emission Observation) - A measurement of the opacity of a plume by a trained human observer
in accordance with EPA Method 9.
Volatility - The degree to which any compound is a gas at specific temperature and pressure conditions.
132
-------
Appendix C
Bibliography and Additional Reading
BIBLIOGRAPHY
The following publications provide background documentation on CEM regulation, CEM monitoring
techniques, and air toxics.
Air Pollution Control Association. 1987. Continuous Emission Monitoring: Advances and Issues.
Pittsburgh, PA.
Air & Waste Management Association. 1990. Continuous Emission Monitoring: Present and Future
Applications. Pittsburgh, PA.
Air & Waste Management Association. 1993. Continuous Emission Monitoring - A Technology for
the 90's. Pittsburgh, PA.
Air & Waste Management Association. 1996. Acid Rain & Electric Utilities - Permits, Allowances,
Monitoring & Meteorology. Pittsburgh, PA.
Air & Waste Management Association. 1996. Continuous Compliance Monitoring Under the Clean
Air Act Amendments. Pittsburgh, PA.
Bundesministerium fur Umvelt. 1988. Air Pollution Control Manual of Continuous Emission
Monitoring, Regulations and Procedures for Emissions Measurements, 2nd Edition. Bundes-
ministerium fur Umvelt, Naturschutz und Reactorsicherheit. PO Box 120692, D 5300, Bonn 1,
Federal Republic of Germany,
CRC Press. 1996. Handbook of Chemistry and Physics. CRC Press, Boca Raton, FL.
Electric Power Research Institute. 1993. Continuous Emission Monitoring Guidelines - 1993 Update,
Volumes 1 & 2. Report No. EPRI TY-1Q2386-V1 & V2. Palo Alto, CA.
Jahnke, J.A. 1993. Continuous Emission Monitoring. Van Nostrand Reinhold. New York, NY,
Jahnke, J.A. 1994. An Operator's Guide to Eliminating Bias in CEM Systems. EPA 430-94-016.
Jahnke, J.A. 1984. Transmissometer Systems - Operation and Maintenance, An Advanced
Course. APT! Course 476A. EPA-45Q-84-004,
Keith, L.H, and Walker, M.1V1. Handbook of Air Toxics: Sampling, Analysis, and Properties. CRC
Press. Boca Raton, FL.
Manahan, S.E. Environmental Chemistry. Fifth Ed. Lewis Publishers. Boca Raton, FL.
Patrick, D.R. 1994. Toxic Air Pollution Handbook. Van Nostrand Reinhold. New York, NY.
Peeler, J.W. 1990. Guidelines for CEMS Performance Specifications and Quality Assurance
Requirements for Municipal Waste Combustion Facilities. Northeast States for Coordinated Air Use
Management.
133
-------
Podlenski, J. Peduto, Mclnnes, R. Abell, F,, and Gronberg, S, 1984. Feasibility Study for Adapting
Present Combustion Source Continuous Monitoring Systems to Hazardous Waste Incinerators:
J. Adaptability Study and Guidelines Document. EPA 600/8-84-011 a.
Seinfeld, J.H. 1986. Atmospheric Chemistry and Physics of Air Pollution, John Wiley and Sons.
New York, NY.
U. S. Environmental Protection Agency. 1996, Code of Federal Regulations: Acid Rain Program -
Continuous Emissions Monitoring. 40 CFR 75. Washington, D,C.
U. S. Environmental Protection Agency. 1996. Code of Federal Regulations: Standards of
Performance for New Stationary Sources - Appendix A - Test Methods, 40 CFR 60. Washington,
D.C.
U. S. Environmental Protection Agency. 1996. Code of Federal Regulations: Standards of
Performance for New Stationary Sources - Appendix B - Performance Specifications. Washington,
D.C.
U. S. Environmental Protection Agency. 1996. Code of Federal Regulations: Standards of
Performance for New Stationary Sources - Appendix F - Quality Assurance Procedures. 40 CFR 60.
Washington, D.C.
U.S. Environmental Protection Agency. 1966. Code of Federal Regulations: Standards for the
Management of Specific Hazardous Wastes and Specific Types of Hazardous Waste Management
Facilities - Appendix IX - Methods Manual for Compliance with BIF Regulations. 40 CFR 266.
Washington, D.C.
ADDITIONAL READING FOR CHAPTER 5
Jahnke, J.A, 1993. Continuous Emission Monitoring, Van Nostrand Reinhold, New York, NY, pp.
188-203.
ADDITIONAL READING FOR CHAPTER 6
Acklin, M.W., McCullough, M., and Tolk, J.D. 1995. Utility Engineering's Experience in the Design,
Equipment Selection, and Operation of CEMS for Utilities. Acid Rain & Electric Utilities - Permits,
Allowances, Monitoring & Meteorology, Air & Waste Management Association. Pittsburgh, PA. pp.
190-198.
Bolstad, J. 1990. Design of Two Systems for VOC Emission Monitoring from Solvent Recovery/VOC
Control Devices. Continuous Emission Monitoring: Present and Future Applications. Air & Waste
Management Association. Pittsburgh, PA. pp 314-326.
Farber, P.S. 1992. Advanced CEMs Test Air Emissions. Environmental Protection, pp 12-19.
Gaffron, G., McCall, E., Myers, R.L., Fletcher, and Miller, M. 1995. Experiences in the Installation,
Certification, and Maintenance of Transit-Time Ultrasonic Flow-Monitors on Utility Stacks with High
Temperature and Velocity. Acid Rain & Electric Utilities - Permits, Allowances, Monitoring &
Meteorology. Air & Waste Management Association. Pittsburgh, PA. pp. 171-181.
Groce, P.J. 1995. Thermal Flow Monitor Design and Performance in Acid Rain Stacks - 1991-1994.
Acid Rain & Electric Utilities - Permits, Allowances, Monitoring & Meteorology. Air & Waste
Management Association. Pittsburgh, PA. pp. 199-209.
Johnson, D, 1991. Continuous Emissions Monitoring Devices. Plant Engineering, pp 76-80.
134
-------
Johnson, L.D. 1995. "Research and Evaluation of Organic Hazardous Air Pollutant Source Emission
Test Methods." Paper presented at the Air & Waste Management Association Meeting. San Antonio,
TX, Paper 95-TP62.02.
Klompstra, T.A. 1990. A Comparison of Extractive and In Situ Technology. Continuous Emission
Monitoring: Present and Future Applications, Air & Waste Management Association. Pittsburgh, PA.
pp 84-92.
Knoll, J.E. 1991. "Protocol for the Field Validation of Stationary Source Emission Measurements."
Paper presented at the Air & Waste Management Association Meeting. Vancouver. Paper 91-58.5.
Patton, D.M. 1990. Techniques for Optimizing Continuous Gas Emission Monitors. Continuous
Emission Monitoring: Present and Future Applications. Air & Waste Management Association.
Pittsburgh, PA. pp 360-369.
Radigan, M.J, 1994. How to Select a Continuous Emission Monitoring System. Hydrocarbon
Processing, pp 73-75.
Rihs, P.W. 1990. CEM Ultrasonic Flow Monitoring Design Installation and Certification Results at the
Salt River Project. Acid Rain & Electric Utilities - Permits, Allowances, Monitoring & Meteorology,
Pittsburgh, PA. Air & Waste Management Association, pp. 182-189.
Wagner, G.H. 1994. Air Compliance Falls Short Without CEMs. Environmental Protection, pp 37-40.
ADDITIONAL READING FOR CHAPTER 7
Bundesministerium fur Umvelt. 1988. "Air Pollution Control Manual of Continuous Emission
Monitoring," Regulations and Procedures for Emissions Measurements, 2nd ed. Naturschutz und
Reactorsicherheit, PO Box 120692, D 5300 Bonn 1, Federal Republic of Germany.
Allendorf, S.W., Ottensen, O.K. 1994. "Tunable Diode Lasers as Continuous Emission Monitors for
Thermal Waste Treatment Processes," Proceedings International Incineration Conference, Houston, TX.
Bacon, T., Webber, K. 1996. "Acid and Halogen Gas Monitoring Using IMS," Proceedings Air & Waste
Management Association, Nashville, TN, paper no. 96-TA30.01.
Coleman, W.M., Dominguez, L.M. 1996. "A Gas Chromatographie Continuous Emission Monitoring
System for VOCs and HAPs," Journal of Air & Waste Management Association, Volume 46.
Carroll, G.J., Thurnau, R.C. 1996. "Mercury Emissions from a Hazardous Waste Incinerator Equipped
with a State-of-the-Art Wet Scrubber,"Journal of Air & Waste Management Association, Volume 45.
Eldridge, J.S., Stock, J.W. 1995. "Extractive FTIR: Manufacturing Process Optimization Study,"
Proceedings Air & Waste Management Association, San Antonio, TX.
Flower, W.L., Peng, L.W. 1994. "A Laser-Based Technique for Continuously Monitoring Metal
Emissions from a Thermal Waste Treatment Units," Proceedings International Incineration Conference,
Houston, TX,
Haile, D.M., Dorris, E.H. 1995. "Field Analysis of Hazardous Air Pollutants with an Ion Trap Mass
Spectrometer," Proceedings Air & Waste Management Association, San Antonio, TX.
Kagaan, R.H., Garbis, S.D. 1996. "Generating Reference Spectra for Calibrating Open-Path FTIR
Measurements," Proceedings Air & Waste Management Association Specialty Conference Measurement
of Toxic and Related Air Pollutants, Research Triangle Park, NC.
135
-------
Jockel, W. 1996, "Present Situation of Mercury CEMS in Germany," Proceedings Air & Waste
Management Association, Nashville, TN,
Kleindienst, T.E., Blanchard, F.T. 1996. "Measurement of C1-C4 Carbonyls on DNPH-Coated Silica
Gel and C18 Cartridges in the Presence of Ozone," Proceedings Air & Waste Management Association
Specialty Conference Measurement of Toxic and Related Air Pollutants, Research Triangle Park, NC.
Lemieux, P.M., Ryan, J.V. 1996. "Emissions of Trace Products of Incomplete Combustion from a
Pilot-Scale Incinerator Secondary Combustion Chamber," Journal of Air & Waste Management
Association, Volume 46.
Ogle, L.D., LaCoss, J.P. 1996. "Validation of Extractive FT1R Method for the Analysis of Aldehydes
in Natural Gas Fired Stationary Engine Exhaust," Proceedings Air & Waste Management Association
Specialty Conference Measurement of Toxic and Related Air Pollutants, Research Triangle Park, NC,
Parmar, S.S., Ugarova, L. 1996. "A Study of Ozone Interferences in Carbonyl Monitoring Using DNPH-
Coated C18 and Silica Cartridges," Proceedings Air & Waste Management Association Specialty
Conference Measurement of Toxic and Related Air Pollutants, Research Triangle Park, NC.
Reagen, W.K., DePuydt, M.M. 1995. "Comprehensive VOC Source Emissions Assessment, A
Combined Approach of EPA Method TO14, EPA Method TO-11, and Extractive FTIR," Proceedings Air
& Waste Management Association, San Antonio, TX.
Russwurm, G.M., Childers, J.W. 1996. "Compendium Method TO-16 Long path. Open path FTIR
Method for Monitoring Ambient Air," Proceedings Air & Waste Management Association, Nashville, TN.
Schlager, R.J., Durham, M.D. 1996. "Continuously Monitoring Mercury from Thermal Treatment
Processes," Proceedings International Incineration Conference, Houston, TX.
Yang, P., Chikhiliwahla, E.D. 1993. "Real-Time Polycyclic Aromatic Hydrocarbon Monitor - A critical
Evaluation," Proceedings Air <5 Waste Management Association Specialty Conference Measurement
of Toxic and Related Air Pollutants, Durham, NC.
136
-------
Appendix D
Hazardous Air Pollutants 1990 CAAA Title HI Listing
CAS NUMBER
75070
60355
75058
98862
53963
107028
79061
79107
107131
107051
92671
62533
90040
1332214
71432
92875
98077
1 00447
92524
117817
542881
75252
106990
156627
105602
133062
CHEMICAL NAME
Acetaldehyde
Acetamide
Acetonitrile
Aeetophenone
2-AcetyIaminofluorene
Acrolein
Acrylamide
Acrylic Acid
Acrylonitrile
Alyl Chloride
4-Aminobipheny!
Aniline
o-Anisidine
Asbestos
Benzene (including benzene from gasoline)
Benzldine
Benzotrichloride
Benzyl chloride
Biphenyl
BIs(2-ethy!hexy!)phthalate(DEHP)
Bis(chlorometbyl)ether
Bromoform
1,3-Butadiene
Calcium cyanamide
Caprolactam
Captan
137
-------
CAS NUMBER
63252
75150
56235
463581
1 20809
133904
57749
7782505
79118
532274
108907
510156
67663
107302
126998
1319773
95487
108394
106445
98828
94757
3547044
334883
132649
96128
84742
106467
91941
111444
542756
CHEMICAL NAME
Carbaryl
Carbon disulfide
Carbon tetrachloride
Carbonyl sulfide
Catechol
Chloramben
Chlordane
Chlorine
Chloroacetic acid
2-ChIoroacetophenone
Chlorobenzene
Chlorobenzilate
Chloroform
Chloromethyl methyl ether
Chloroprene
Cresols/Cresylie acid {isomers and mixture)
o-Cresol
tn-CresoI
p-Cresol
Cumene
2,4-D, salts and esters
DDE
Diazomethane
Dibenzofurans
1 ,2-Dibromo-3-chloropropane
Dibutyfphthalate
1 ,4-Dichlorobenzene{p)
3,3-Dichlorobenzidene
Dichloroethy! ether (Bis{2-chloroethyl)ether
1 ,3-Dichloropropene
138
-------
CAS NUMBER
62737
111422
121697
64675
1 1 9904
60117
119937
79447
68122
57147
131113
77781
534521
51285
121142
123911
122667
106898
106887
140885
100414
51796
75003
106934
107062
107211
151564
75218
96457
75343
CHEMICAL NAME
Dichlorvos
Diethanolamine
N,N-Diethyl aniline {NrN-DimethyIaniline)
Diethyt sulfate
3,3-Dimethoxybenzidine
Dimethyl aminoazobenzene
3,3'-Dimethoxybenzidine
Dimethyl carbamoyl chloride
Dimethyl formamide
1,1 -Dimethyl hydrazine
Dimethyl phthalate
Dimethyl sulfate
4,6-Dinitro-o-cresol, and salts
2,4-Dinitrophenol
2,4-Dinitrotoluene
1 ,4-DIoxane (1 ,4-Diethyleneoxide)
1 ,2-Diphenylhydrazine
Epichlorohydrin f 1 -Chloro-2,3-epoxypropane)
1,2-Epoxybutane
Ethyl acrylate
Ethyl benzene
Ethyl carbamate (Urethane)
Ethyl chloride (Chloroethane)
Ethylene dibromide (Dibromoethane)
Ethylene dichloride (1,2-Dichloroethane)
Ethylene glycol
Ethylene imine (Aziridine)
Ethylene oxide
Ethylene thiourea
Ethylidene dichloride (1,1-Dichloroethane)
139
-------
CAS NUMBER
5000O
76448
118741
87683
77474
67721
822060
680319
110543
302012
7647010
7664393
123319
78591
58899
108316
67561
72435
74839
74873
71558
78933
60344
74884
108101
624839
80626
1 634044
101144
75092
CHEMICAL NAME
Formaldehyde
Heptachlor
Hexachlorobenzene
Hexachlorobutadiene
Hexachlorocyclopentadiene
Hexachloroethane
Hexamethylene-1 ,6-diisocyanate
Hexamethylphosphoramide
Hexane
Hydrazine
Hydrochloric acid
Hydrogen fluoride (Hydrofluoric acid!
Hydroquinone
Isophorone
Lindane {all iosmers)
Maleie anhydride
Methanol
Methoxychlor
Methyl bromide {Bromomethane!
Methyl chloride (Chloromethane)
Methyl chloroform {1,1,1-Trichloroethane)
Methyl ethyl ketone (2-Butanone)
Methyl hydrazine
Methyl Iodide (lodomethane)
Methy! isobutyl ketone (Hexone)
Methyl isocyanate
Methyl methacrylate
Methyl tert butyl ether
4,4-Methylene bis (2-ch!oroaniline)
Methylene chloride (Diehloromethane)
140
-------
CAS NUMBER
101688
101779
91203
98953
92933
100027
79469
684935
62759
59892
56382
82688
87865
108952
106503
75445
7803512
7723140
85449
1336363
1120714
57578
123386
114261
78875
75569
75558
91225
106514
100425
CHEMICAL NAME
Methylene diphenyl diisocyanate (MDIS
4,4-MethylenedianiIine
Naphthalene
Nitrobenzene
4-Nitrobiphenyl
4-Nitrophenol
2-Nitropropane
N-Nitroso-N-methylurea
N-Nitrosodimethylamine
N-Nitrosomorphloline
Parathion
Pentachloronitrobenzene (Quintobenzene)
Pentachlorophenol
Phenol
p-Phenylenediamine
Phosgene
Phosphine
Phosphorus
Phthalic anhydride
Polychlorinated biphenyls (Aroelors)
1,3-Propane sultone
beta-Propiolactone
Propionaldehyde
Propoxur (Baygon)
Propylene dichloride (1,2-DichIoropropane)
Propylene oxide
1,2-Propylenimine (2-Methyl aziridine)
Quinoline
Quinone
Styrene
141
-------
CAS NUMBER
96093
1746016
79345
127184
7550450
108883
95807
584849
95534
8001352
120821
79005
79016
95954
88062
121448
1582098
540841
108054
593602
75014
75354
1 330207
95476
108383
106423
0
0
0
0
CHEMICAL NAME
Styrene oxide
2,3,7, 8-Tetrachlorodibenzo-p-dioxin
1 ,1 ,2,2-Tetrachloroethane
Tetrachloroethylene (Perchloroethylene)
Titanium tetrachloride
Toluene
2,4-Toluene diamine
2,4-Toluene diisocyanate
o-Toluidine
Toxaphene (chlorinated camphene)
1 ,2,4-Trichlorobenzene
1 ,1 ,2-Trichloroethane
Trichloroethylene
2,4,5-Trichlorophenol
2,4,6-Trichlorophenol
Triethylamine
Trifluralin
2,2,4-Trimethylpentane
Vinyl acetate
Vinyl bromide
Vinyl chloride
Vinylidene chloride (1,1-Dichloroethylene)
Xylenes (isomers and mixture)
o-Xylenes
m-Xylenes
p-Xylenes
Antimony Compounds
Arsenic Compounds (inorganic including arsine)
Beryllium Compounds
Cadmium Compounds
142
-------
CAS NUMBER
0
0
0
0
0
0
0
0
0
0
0
0
0
CHEMICAL NAME
Chromium Compounds
Cobalt Compounds
Coke Oven Emissions
Cyanide Compounds1
Glycol ethers2
Lead Compounds
Manganese Compounds
Mercury Compounds
Fine mineral fibers3
Nickel Compounds
Polycylic Organic Matter4
Radionuclides (including radon)5
Selenium Compounds
NOTE: For all listings above which contain the word "compounds" and for glycol ethers, the following applies;
Unless otherwise specified, these listings are defined as including any unique chemical substance that contains
the named chemical (i.e., antimony, arsenic, etc.) as part of that chemical's infrastructure.
1 X'CN where X = H' or any other group where a formal disoeciation may occur. For
example KCN or CA(CN}2
2 includes mono- and di-ethers of ethylene glycol, diethylene glycol, and triethylene glycol
R-(OCH2CH2)n-OR. Polymers are excluded from the glycol category.
3 includes mineral fiber emissions from facilities manufacturing or processing glass, rock, or
slag fibers (or other mineral derived fibers) of average diameter 1 micrometer or less.
4 •
includes organic compounds with more than one benzene ring, and which have a boiling
point greater than or equal to 100°C.
a type of atom which spontaneously undergoes radioactive decay.
143
-------
Appendix E
Units of the Standard
Regulations can impose design constraints on a GEM system through the form in which an emissions
standard is expressed. Different types of standards have been used by environmental control agencies
for a variety of purposes. In these regulations, emissions have been required to be expressed in the
following forms:
A. Concentration when expressed in mg/m3, corrected to standard conditions (20°C, 101,325
kpa)
B. Wet-basis concentration (ppm, mg/rn3) corrected to dry conditions
C. Concentration (ppm, mg/m3) corrected to 12% (or other percentage) C02
D, Concentration (ppm, mg/m3) corrected to 6% (or other percentage) 02
E, Mass emission rate (kg/hr, tons/yr)
F. Mass emission rate (thermal! (ng/Joule)
G. Process weight rate (kg/ton of product produced)
H. Control device efficiency (%)
Calculating emissions in any of these forms implies that specific types of instruments or parameters are
to be monitored. A control device efficiency requirement implies that measurements must be made both
upstream and downstream from the control device. The basic calculations and their implications follow;
A. Concentration when expressed in mg/m3 corrected to standard conditions (20°C, 101.325 kpa)
P T
rs ' std
This calculation is necessary when reporting particulate, metal, or other concentrations in units of mg/m3
corrected to standard conditions. It implies that both temperature and pressure measurements be made.
Note that by applying this correction only, emissions will be expressed on a wet basis. A wet basis
calculation can be used to calculate a pollutant mass rate.
144
-------
B. Concentration (ppm, mg/m3) corrected to dry conditions
- Bws>
When a concentration standard is to be used alone, without subsequently calculating a pollutant mass
rate, results typically are required to be expressed on a dry basis. In this case, the flue gas temperature,
pressure, and moisture content must be known. For wet-basis monitoring systems such as in-situ
systems, in-stack dilution systems, out-of-stack/non-dry dilution systems, and hot-wet source level
extractive systems, the moisture content then must be measured, or in the case of stable processes,
often is assumed to have a constant value, based on stack test data. For dry-basis source level
extractive systems, pollutant gases are measured on a dry basis and a moisture determination is not
necessary,
C. Concentration (ppm, mg/m3! corrected to 6% lor other percentage) 02
cg (20.9 - 6.0)
Cs,
•0*0, 20.9 - %02
The expressions given in A and B do not correct for the effects of dilution air. Dilution air corrections are
almost always required so that emission's requirements can be normalized between sources. Since
dilution air is usually ambient air, the percent oxygen concentration or percent C02 concentration is
measured to perform the correction.
D. Concentration
-------
F. Mass emission rate (thermal) {ng/Jouie)
_ - _ 20.9 _ - _ 100
E = C.F., _____ £ = c.F,.
8 d 20.9 - %O2 s ° %CO2
-c F 2°'9
~ Cswrw
20,9(1 - Bwa) - %0
2w
20.9
ws d 20.9(1 - BWJ - %O.
Electric utilities and industrial boilers combusting carbonaceous fuels usually are required to express
emissions in terms of mass per unit heat input. These calculations are performed by using the proce-
dures of EPA Method 19, the F-factor methods, which require the measurement of either oxygen or CO2
to correct for dilution. A number of monitoring options are available using the F factors. If the Fc factor
is used, either a wet or dry system can be used provided that both the pollutant and C02 are measured
on the same basis. If the Fd factor is used, the pollutant and oxygen concentration measurements must
be made on a dry basis. If the pollutant and oxygen are measured on a wet basis, the "wet" F factor
can be used if the control system does not employ a wet scrubber. After wet scrubbers, the moisture
content must be measured.
G. Process weight rate (kg/ton of product produced)
PWR = ML
Process weight rate standards are based upon the amount of product produced from the process
industries. The first of these types of standards were established in the NSPS nitric acid and sulfuric
acid plant and petroleum refinery requirements. The method requires accounting for the mass of the
pollutant emitted and accounting for the amount of product produced that is associated with the
emissions. This, in general, requires the measurement of the mass emission rate as in E, above, and a
determination of the production rate, p. For some industries, such as certain types of suifuric acid
plants, alternative calculation methods based on process parameter measurements have been developed.
H. Control device efficiency (%)
% Efficiency = _ " 100
c.
's(in)
earr
146
-------
Removal efficiency is required to be reported in some standards. This requires the use of two CEM
systems or one time-shared system measuring or extracting sample gas both upstream and downstream
from the control device. If a time-shared system is used, with one bank of analyzers, multi-ranging
analyzers may be required because of the difference in concentration between the uncontrolled and
controlled flue gas stream. Concentration measurements are corrected for variations in diluent gas
concentrations for expressed as thermal mass emission rates) prior to calculating the control device
efficiency.
The monitoring requirements of A-H above are summarized in Table E-1,
TABLE E-1. Monitoring Requirements Based upon Conditions of the Units of the Emissions Standard
Eq.
Gases Monitored
Parameters
Monitored
Pollutant
P, T
B
Pollutant
H20 or none
Pollutant
O,
Pollutant
CO,
Pollutant
None or H20
Flow
Pollutant
O2 or CO2, possibiy H2O
Pollutant
Flow, Production
Rate, Parameters
H
Pollutant
llnlet & Outlet)
02 or CO2 (Inlet & Outlets
147
-------
Symbols:
rstd
Ts!d
CsS%02
20.9
pmr
Qsw
PWR
P
r
v's(in)corr
average stack concentration in mg/m3 corrected to standard conditions
uncorrected concentration
standard pressure (101.325 kpa)
standard temperature (2Q°C, 493°K)
stack temperature (°K, °R)
stack pressure (kpa)
average, dry concentration
average, wet concentration
moisture fraction
average stack concentration corrected to 6% O2
percentage of O2 in ambient air
average stack concentration corrected to 12% CO2
pollutant mass rate
wet, flue gas volumetric flow rate
dry F factor
F factor for CO2
wet F factor
moisture fraction in ambient air
process weight rate (Kg, tbs/tonne, ton product produced, for example)
production rate (units/ton, tonnes/hr, for example)
average concentration, corrected for dilution, in gas entering
control device
•'s(out)corr
average concentration, corrected for dilution, in gas leaving
control device
thermal mass emission rate (ng/J, Ibs/IVIlvlBtu)
148
-------
APPENDIX F Guide for Evaluating CEM System Costs
Table F-1. CEM System Cost Elements Checklist
1. Design/planning of the CEM system program - This element includes the identification of all applicable regulations (e.g., standards, monitoring
requirements, location requirements, testing and QA requirements, reporting and recordkeeping requirements) and source-specific constraints (i.e.,
tentative sampling locations, physical installation constraints, effluent conditions, environmental conditions, personnel constraints, etc.). It may
also include some amount of training, formal or otherwise, for the person(s) responsible for the design of the system.
ACTIVITY
A. Familiarization with regulations (read,
research...)
B. Resolution of regulatory questions (calls, letters,
meetings, etc.)
C. Source-specific constraints
1 . Review drawings/plans
2. Inspect source (inspect similar facility if source
not yet built)
3. Identify physical installation constraints,
evaluate existing utilities (electricity, air )
4. Estimate effluent conditions and parameters
(based on stack test report reviews, measure-
ments, and/or engineering judgment, etc.)
5. Estimate environmental conditions
6. Evaluate personnel constraints (interview
process operators, instrument tech., supervisory
personnel, and corporate representatives to
determine availability, expertise, previous
experience, and opinions/bias)
7. Summarize results of 1 .-6. in written form
D. Training for person(s) responsible for
desiqn/Dlanninq of the CEM system
LABOR HOURS
(PM, TECH.
MGT, ADM)
LABOR
COST
SUB-
CONTRACTS
OTHER
DIRECT
COSTS
TOTAL
COSTS
-------
Table F-1. Continued
2. Selection of CEM system equipment - This element includes deciding between component purchases versus system purchases and deciding
about support alternatives (i.e., vendor maintenance agreements, emergency repair services, need for instrument training, availability of spare
parts, etc.) It also includes the effort and cost associated with a) investigating monitoring technologies and available equipment options,
b) developing equipment specifications, c) identifying and selecting potential vendors, d) developing an RFP and performance guarantees,
e) evaluating proposals, and f) negotiating and executing a contract with the successful bidder(s).
ACTIVITY
A. Decide on basic approach (general approach,
type of monitoring system, components vs. system
purchase, level of vendor support, etc.)
B. Develop written equipment specifications
C. Identify potential bidders (call other sources, go
to trade shows, read journal articles and trade
magazines, call consultants, rely on previous
experience)
D. Develop RFP and performance guarantees
(Define the monitoring program in terms of the
applicable regulations, source specific constraints,
as well as the performance specifications and
guarantees that are needed. This must be done in
accordance with the purchasing practices of the
buyer and may also include legal, insurance,
performance bonds, and other terms and
conditions)
E. Compile and send out RFP
F. Conduct bidders meeting and response to
questions raised by the bidders
G. Bidder presentations (some companies allow or
require that all or selected bidders make an on-site
presentation of their proposal)
LABOR HOURS
(PM, TECH.
MGT, ADM)
LABOR
COST
SUB-
CONTRACTS
OTHER
DIRECT
COSTS
TOTAL
COSTS
CJl
o
-------
Table F-1. Continued
ACTIVITY
H. Evaluate proposals and select a winner
I. Negotiate details of the contract, draft and
execute the agreement
J. Administrative costs associated with
imDlementinq the contract
LABOR HOURS
(PM, TECH.
MGT, ADM)
LABOR
COST
SUB-
CONTRACTS
OTHER
DIRECT
COSTS
TOTAL
COSTS
3. Purchase of Capital Equipment - This element should reflect the actual cost associated with the CEM system and would include all components,
calibration materials, and support utilities. Technical and administrative time associated with the purchase activity should also be included.
ACTIVITY
A. Pollutant and diluent analyzers
B. Monitor remote control units and junction boxes
C. Sample acquisition, sample conditioning, and
sample transport equipment for extractive CEM
systems
D. Sample interface equipment (i.e., air purge
blowers, filters, etc.) for in situ CEM systems
E. Signal cables, communications, alarms, etc.
F. DAHS computers, software, chart recorders,
remote readouts, etc.
G. Support utilities, electrical power conditioners,
isolation transformers, lightening protection,
compressed air supply, instrument air, air clean-up
systems, etc.
H. Evaluate proposals and select a winner
LABOR HOURS
(PM, TECH.
MGT, ADM)
LABOR
COST
SUB-
CONTRACTS
OTHER
DIRECT
COSTS
TOTAL
COSTS
-------
Table F-1. Continued
ACTIVITY
I. Negotiate details of the contract, draft and
execute the agreement
J. Administrative costs associated with
implementinq the contract
LABOR HOURS
(PM, TECH.
MGT, ADM)
LABOR
COST
SUB-
CONTRACTS
OTHER
DIRECT
COSTS
TOTAL
COSTS
4. CEM System Installation - This element should include costs for a) stratification tests, if necessary (see text), b) agency approval of the
monitoring location, if required, c) installation of sample acquisition/conditioning equipment, sample transport lines, analyzers, monitor control
units, calibration gases and related equipment, and data handling/recording equipment, d) installation of sampling ports and utilities as well as
ladders, platforms and other access to both the monitoring location and the manual testing location, and e) construction of protective shelters
and safety equipment.
ACTIVITY
A. Submit proposed monitor locations to the
control agency
B. Discuss agency response
C. Conduct stratification test if necessary
1. Assemble equipment or hire testing firm
2. Travel to source and return
3. Perform test and reduce data
4. Write report and letter to the agency
5. Determine agency's response
D. Determine alternate monitoring location, if
necessary
E. Install monitor and manual testing ports
LABOR HOURS
(PM, TECH.
MGT, ADM)
LABOR
COST
SUB-
CONTRACTS
OTHER
DIRECT
COSTS
TOTAL
COSTS
Ol
-------
Table F-1. Continued
ACTIVITY
F. Install necessary support utilities (electricity,
compressed air, communications, etc.)
G. Design and erect protective shelters for monitor
components and/or manual sampling
H. Design and install necessary scaffolding and
access (i.e., ladders, elevators etc.) for both
monitor and manual sampling locations
I . Receive and check out monitoring equipment
from vendor(s)
J. Supervise and inspect vendor, or install sample
probes, sample acquisition equipment, sample
lines, analyzers, monitor control units, calibration
aases/manifolds, and DAHS
LABOR HOURS
(PM, TECH.
MGT, ADM)
LABOR
COST
SUB-
CONTRACTS
OTHER
DIRECT
COSTS
TOTAL
COSTS
Ol
to
5. Start-up and Performance Testing - This element should include the costs for start-up and debugging of the system, on-site training, preliminary
testing, arrangements for the performance specification test (PST) (e.g., selecting a contractor, pretest meeting, development and submission of
a protocol to the agency, safety requirements, etc.) and the actual PST cost including preparation of a test report.
ACTIVITY
A. Observe start-up
B. On-site training
C. Debugging and problem resolution during first
few weeks
D. Select contractor for RA test or entire
performance specification test
LABOR HOURS
(PM, TECH.
MGT, ADM)
LABOR
COST
SUB-
CONTRACTS
OTHER
DIRECT
COSTS
TOTAL
COSTS
-------
Table F-1. Continued
ACTIVITY
E. Travel and attend pretest meeting with testing
contractor and agency, if necessary
F. Develop and submit test protocol to agency, if
necessary
G. Notify agency of test dates
H. Travel and contractor travel for PST
I. Preliminary testing as necessary
J. Conduct RA test (source and contractor
expenses)
K. Source personnel conduct drift test and forward
data to contractor (assumption)
L. Contractor prepares report, source reviews
report, and submits to agency (assumption)
M. Allowance to represent likelihood of test
postponement
N. Allowance to represent likelihood of PST failure
and subsequent repeat tests
LABOR HOURS
(PM, TECH.
MGT, ADM)
LABOR
COST
SUB-
CONTRACTS
OTHER
DIRECT
COSTS
TOTAL
COSTS
Ul
6. Operation and Maintenance - At a minimum, this element should include the costs associated with performing the manufacturer's recommended
maintenance activities. It should also include the costs associated with expected use of spare parts, maintenance of a parts inventory, emergency
repair service, etc. Furthermore, this element would include the cost of performing daily zero and upscale calibration checks of the CEM system
(including the costs of calibration gases or materials).
-------
Table F-1. Continued
ACTIVITY
A. Spare parts inventory (capital cost plus interest)
B. Administrative effort to maintain parts and
supply inventory
C. Effort to perform manufacturer's recommended
maintenance activities
D. Daily calibration checks (time to review
calibration data and adjust the CEM system if
necessary; calibration gas costs)
E. Corrective action and emergency repair service,
if not included in QA costs
LABOR HOURS
(PM, TECH.
MGT, ADM)
LABOR
COST
SUB-
CONTRACTS
OTHER
DIRECT
COSTS
TOTAL
COSTS
en
ui
7. Quality Assurance - The costs for this element depend on the applicable requirements. For this example, we assume that requirements similar
to 40 CFR 60, Appendix F, Procedure 1 would apply. To estimate these costs accurately, some frequency of failure to meet QA criteria must be
assumed. It may be appropriate to estimate costs for participation in, or observation of, agency inspections or audits of the monitoring program.
ACTIVITY
A. Development of QA/QC Plan
B. Annual and/or quarterly accuracy audits
C. Daily precision checks
D. Corrective action for malfunctioning monitors
E. Aqencv Inspections or Audits
LABOR HOURS
(PM, TECH.
MGT, ADM)
LABOR
COST
SUB-
CONTRACTS
OTHER
DIRECT
COSTS
TOTAL
COSTS
-------
Table F-1. Continued
8. Recordkeeping and Reporting Costs - This element should include the costs associated with the preparation of the required periodic reports
describing both emissions problems and monitoring problems/events and data quality reports. It should also include the costs associated with
maintenance of the records necessary to generate these reports and other records which are required to be maintained on-site for agency
inspection. Most companies will have to maintain additional records in accordance with their own internal procedures to substantiate labor
and capital expenditures as well as contracts with other involved parties.
ACTIVITY
A. Compilation of CEM system QA Data - (daily
and periodic checks, work requests, corrective
action records, preventive maintenance records,
monitor logs, audit/test reports, etc.)
B. Back-up procedure for computer files and
magnetic media
C. Implementation of emissions and monitor record
keeping and retrieval system
D. Preparation of draft periodic reports
1 . Confirm unit and CEM system on-line/off-line
periods
2. Review DAHS alarms for effect on data quality
3. Confirm or resolve issues or reason/corrective
action codes
4. Review emissions data for consistency with
known performance
5. Prepared edited emission summaries
6. Prepared Data Assessment Report
E. Internal review of draft report
F. Revision and submission of report to aqencv
LABOR HOURS
(PM, TECH.
MGT, ADM)
LABOR
COST
SUB-
CONTRACTS
OTHER
DIRECT
COSTS
TOTAL
COSTS
-------
Table F-1. Continued
9. Response Plan and Other Action - Sources will need to develop a plan for responding to emissions problems and CEM malfunctions. The costs
for developing such a plan and responding to problems obviously will depend on the specific requirements and how the regulations are written and
enforced. Costs may also be incurred in responding to agency questions or other follow-up actions. Guidance for estimating these costs is not
provided here.
Table F-2. Summary of CEM System Cost Elements
ACTIVITY
1 . Design/planning of the CEM System Program
2. Selection of CEM System Equipment
3. Purchase of Capital Equipment
4. CEM System Installation
5. Start-up and Performance Testing
Subtotal Initial Costs
6. Operation and Maintenance
7. Quality Assurance
8. Recordkeeping and Reporting Costs
9. Response Plan and Other Action
Subtotal Recurring Costs
TOTAL Lifetime Costs
ESTIMATED COST
-------
Appendix G
Effects of Sample Matrix on Analyte Spike
Spiking the target analytes/surrogates into a flue gas matrix and calculating the recoveries can be challenging and
can be influenced by experimental noise. To calculate the recovery of the anaiyte spikes, one must know the
concentration of the target or surrogate compounds in the flue gas and whether these concentrations fluctuate
with time. Calculating the recovery of the spiked analytes requires accurate knowledge of: 1) the flow rate of
the spike gas, 2) the total sample flow rate (spike plus flue gas sample), 3) the concentration of the tar-
get/surrogate in the flue gas, and 4) the extent to which the concentrations vary with time. When flue gas
concentrations vary significantly with time, an accurate calculation of the spike recovery may not be possible. A
graphical representation of the effects of concentration variations on an anaiyte spiking experiment is given in
Figure G-1.
Spiked Flue Gas Sample
Spiked Flue Gas Sample
Time
Time
Figure G-1. Effects of concentration variations on anaiyte spiking experiments.
An ideal anaiyte spike concentration is one that approximates twice the flue gas concentration so that the spike
most closely approximates the flue gas concentration. This may be difficult to accomplish without prior knowledge
of the actual flue gas anaiyte concentrations. In practice, the spike concentration is constrained both by the avail-
able range of calibration standards and by the desire to limit the spike flow rate to less than 10 percent of the total
sample flow rate so that sample matrix effects are not obscured by excessive dilution of the sample gas. In some
cases, high level calibration standards may be diluted quantitatively with nitrogen to obtain spike gases that
approximate the ideal anaiyte spike concentration.
Anaiyte Spike Calculations
The percent recovery (%R! of the spiked analytes are calculated as:
%R = (SJCJxIOO
Equation G-1
Where:
Sra
c.
Mean concentration of the anaiyte spiked flue gas samples (observed)
Expected concentration of the
spiked samples (theoretical!
158
-------
The expected concentration (Ce) of the spiked samples are calculated as:
Ce = D, Cs + Su H - Df) Equation G-2
Where:
D, = Dilution Factor {Spike flow/Total flow)
total flow = spike flow plus flue gas sample flow
Cs = Cylinder concentration of spike gas
Su = Native concentration of analytes in unspiked samples
The spike dilution factor may be confirmed by measuring the total flow and the spike flow directly. Alternately,
the spike dilution can be verified by comparing the concentration of the tracer compound in the spiked samples
(diluted) to the tracer concentration in the direct (undiluted) measurement of the spike gas. If SF6 is the tracer
gas, then:
Df = [SF6]spike / [SF6Idireot Equation G-3
Where:
[SF8]sp)ke = The diluted SF6 concentration measured in a spiked sample
[SF6]direct = The SF6 concentration measured directly
The bias is determined by calculating the difference between the observed spike value and the expected response
(i.e., the equivalent concentration of the spiked material plus the analyte concentration adjusted for spike dilution).
Bias is defined by EPA Method 301 (Section 6,3.1) as:
B = Sm - CB Equation G-4
Where:
B = Bias at spike level
Sm = Mean concentration of the analyte spiked samples
Ca = Expected concentration of the analyte in spiked samples
The bias (accuracy) is defined in terms of concentration where the recovery is expressed as a percentage of the
expected concentration. For example, If a measurement technique gives a 2 ppm positive bias when attempting
to measure 10 ppm, then the percent recovery would correspond to 120%.
•U.S. Government Printing Office: 1997 — 549-001/60149 159
------- |