September 2003
Environmental Technology
Verification Report
PS Analytical, Ltd.
Sir Galahad II Mercury
Continuous Emission Monitor
Prepared by
Battelle
Battelle
. . . Putting Technology To Work
Under a cooperative agreement with
£EPA U.S. Environmental Protection Agency
ETV EI VetV
-------
September 2003
Environmental Technology Verification
Report
ETV Advanced Monitoring Systems Center
PS Analytical, Ltd.
Sir Galahad II Mercury
Continuous Emission Monitor
by
Thomas Kelly
Zachary Willenberg
Karen Riggs
Battelle
Columbus, Ohio 43201
and
(under the support of the U.S. Department of Energy)
James Dunn
Shaw Environmental, Inc.
Oak Ridge, Tennessee
Kristopher Kinder
Shaw Environmental, Inc.
Knoxville, Tennessee
James Calcagno
University of Tennessee
Knoxville, Tennessee
-------
Notice
The U.S. Environmental Protection Agency (EPA), through its Office of Research and
Development, has financially supported and collaborated in the extramural program described
here. This document has been peer reviewed by the Agency and recommended for public release.
Mention of trade names or commercial products does not constitute endorsement or
recommendation by the EPA for use.
11
-------
Foreword
The U.S. Environmental Protection Agency (EPA) is charged by Congress with protecting the
nation's air, water, and land resources. Under a mandate of national environmental laws, the
Agency strives to formulate and implement actions leading to a compatible balance between
human activities and the ability of natural systems to support and nurture life. To meet this
mandate, the EPA's Office of Research and Development provides data and science support that
can be used to solve environmental problems and to build the scientific knowledge base needed
to manage our ecological resources wisely, to understand how pollutants affect our health, and to
prevent or reduce environmental risks.
The Environmental Technology Verification (ETV) Program has been established by the EPA to
verify the performance characteristics of innovative environmental technology across all media
and to report this objective information to permitters, buyers, and users of the technology, thus
substantially accelerating the entrance of new environmental technologies into the marketplace.
Verification organizations oversee and report verification activities based on testing and quality
assurance protocols developed with input from major stakeholders and customer groups
associated with the technology area. ETV consists of seven environmental technology centers.
Information about each of these centers can be found on the Internet at http://www.epa.gov/etv/.
Effective verifications of monitoring technologies are needed to assess environmental quality
and to supply cost and performance data to select the most appropriate technology for that
assessment. In 1997, through a competitive cooperative agreement, Battelle was awarded EPA
funding and support to plan, coordinate, and conduct such verification tests for "Advanced
Monitoring Systems for Air, Water, and Soil" and report the results to the community at large.
Information concerning this specific environmental technology area can be found on the Internet
at http://www.epa.gov/etv/centers/centerl .html.
111
-------
Acknowledgments
The authors wish to acknowledge the support of all those who helped plan and conduct the
verification test, analyze the data, and prepare this report. In particular we would like to thank
Wayne Davis of the University of Tennessee; Marshall Allen and Thomas Geisler of the
Hemispheric Center for Environmental Technology at Florida International University; and
Stephen Priebe of the Idaho National Engineering and Environmental Laboratory, U.S.
Department of Energy. We also acknowledge the assistance of ETV AMS Center stakeholders
Philip Galvin, New York State Department of Environmental Conservation; Will Ollison,
American Petroleum Institute; and Roy Owens, Owens Corning, and of Jeff Ryan of EPA's
National Risk Management Research Laboratory, as reviewers of the verification reports from
this test.
iv
-------
Contents
Page
Notice ii
Foreword iii
Acknowledgments iv
List of Abbreviations viii
1 Background 1
2 Technology Description 2
3 Test Design and Procedures 4
3.1 Introduction 4
3.2 Facility Description 5
3.3 Test Design 9
3.3.1 Equipment Setup 9
3.3.2 Test Schedule 9
3.3.3 Reference Method Sampling 11
3.3.4 Verification Procedures 14
3.4 Materials and Equipment 16
3.4.1 High Purity Gases 16
3.4.2 Mercury Standard Gases 16
3.4.3 Mercury Spiking Standard 19
3.4.4 Sampling Trains 19
3.4.5 Analysis Equipment 19
4 Quality Assurance/Quality Control 20
4.1 Facility Calibrations 20
4.2 Ontario Hydro Sampling and Analysis 20
4.2.1 Ontario Hydro Reproducibility 21
4.2.2 Ontario Hydro Blank and Spike Results 24
4.3 Audits 25
4.3.1 Technical Systems Audit 25
4.3.2 Performance Evaluation Audits 26
4.3.3 Data Quality Audit 28
5 Statistical Methods 29
5.1 Relative Accuracy 29
5.2 Correlation with Reference Method 29
5.3 Precision 30
5.4 Sampling System Bias 30
v
-------
5.5 Relative Calibration and Zero Drift 31
5.6 Response Time 31
5.7 Data Completeness 31
5.8 Operational Factors 32
6 Test Results 33
6.1 Relative Accuracy 33
6.2 Correlation with the Reference Method 34
6.3 Precision 36
6.4 Sampling System Bias 36
6.5 Relative Calibration and Zero Drift 37
6.6 Response Time 38
6.7 Data Completeness and Operational Factors 39
7 Performance Summary 48
8 References 49
Figures
Figure 2-1. PS Analytical, Ltd., SG-IICEM 2
Figure 3-1. Schematic of the TSCAI and Off-Gas Cleaning System 6
Figure 3-2. Overview of TSCAI Test Location 7
Figure 3-3. Side View of TSCAI Stack 7
Figure 4-1. Plot of Ontario Hydro Train B Results vs. Train A Results 24
Figure 6-la. Linear Regression Plot of SG-II Hg° Results Against OH Results 35
Figure 6-lb. Linear Regression Plot of SG-II Hgx Results Against OH Results 35
Tables
Table 3-1. TSCAI Stack Gas Characteristics 8
Table 3-2. Mercury CEM Verification Test Schedule 10
vi
-------
Table 3-3. Schedule of OH Method Sampling Runs in Initial Sampling Period
(August 8 - 11, 2002) 12
Table 3-4. Schedule of OH Method Sampling Runs in Final Sampling Period
(September 16- 19, 2002) 13
Table 3-5. Data Used for SG-II Performance Evaluation 14
Table 3-6. Results of Elemental Mercury Standard Analyses 17
Table 3-7. Precision of Elemental Mercury Standard Measurements 18
Table 4-1. Ontario Hydro Results from Initial Sampling Period (August 8- 11, 2002)
(|ig/dscm) 22
Table 4-2. Ontario Hydro Results from Final Sampling Period (September 16 - 19, 2002)
(|ig/dscm) 23
Table 4-3. Results of Linear Regression, Correlation, and Percent Relative Standard
Deviation of Paired Ontario Hydro Train Results (n = 18) 24
Table 4-4. Summary of PE Audits 26
Table 4-5. Results of PE Audit of OH Train Recovery and Analysis 27
Table 6-1. Summary of Results from OH Reference Method and SG-II (|ig/dscm) 34
Table 6-2. Relative Accuracy Results for the SG-II 34
Table 6-3. Coefficients of Determination (r2) for Correlation of
SG-II Results with OH Results 36
Table 6-4. Precision of the SG-II During OH Runs 9 and 12 36
Table 6-5. Sampling System Bias Results 37
Table 6-6. Calibration and Zero Drift Results 38
Table 6-7. Summary of Data Used to Estimate Response Time 39
Table 6-8. Extent of Down Time and Service Time 42
vii
-------
List of Abbreviations
AMS Advanced Monitoring Systems
CO carbon monoxide
C02 carbon dioxide
CEM continuous emission monitor
cm centimeter
CVAA cold vapor atomic absorption
dscf dry standard cubic foot
dscm dry standard cubic meter
DOE U.S. Department of Energy
EPA U.S. Environmental Protection Agency
ETTP East Tennessee Technology Park
ETV Environmental Technology Verification
FIU-HCET Florida International University, Hemispheric Center for Environmental
Technology
Hg° elemental mercury
Hgox oxidized mercury
Hgx total vapor-phase mercury
hr hour
|ig microgram
mL milliliter
m3 cubic meter
mg milligram
min minute
NIST National Institute of Standards and Technology
02 oxygen
OH Ontario Hydro
ORD Office of Research and Development
PE performance evaluation
QA quality assurance
QA/QC quality assurance/quality control
QMP Quality Management Plan
RA relative accuracy
RCRA Resource Conservation and Recovery Act
RPD relative percent difference
RSD relative standard deviation
SEI Shaw Environmental, Inc.
STL Severn Trent Laboratories
TSA technical systems audit
TSCAI Toxic Substances Control Act Incinerator
viii
-------
Chapter 1
Background
The U.S. Environmental Protection Agency (EPA) supports the Environmental Technology
Verification (ETV) Program to facilitate the deployment of innovative environmental tech-
nologies through performance verification and dissemination of information. The goal of the
ETV Program is to further environmental protection by accelerating the acceptance and use of
improved and cost-effective technologies. ETV seeks to achieve this goal by providing high-
quality, peer-reviewed data on technology performance to those involved in the design,
distribution, financing, permitting, purchase, and use of environmental technologies.
ETV works in partnership with recognized testing organizations; with stakeholder groups
consisting of buyers, vendor organizations, and permitters; and with the full participation of
individual technology developers. The program evaluates the performance of innovative tech-
nologies by developing test plans that are responsive to the needs of stakeholders, conducting
field or laboratory tests (as appropriate), collecting and analyzing data, and preparing peer-
reviewed reports. All evaluations are conducted in accordance with rigorous quality assurance
(QA) protocols to ensure that data of known and adequate quality are generated and that the
results are defensible.
The EPA's National Exposure Research Laboratory and its verification organization partner,
Battelle, operate the Advanced Monitoring Systems (AMS) Center under ETV. The AMS Center
recently evaluated the performance of continuous emission monitors (CEMs) for mercury,
including the PS Analytical, Ltd., Sir Galahad II (SG-II) mercury CEM.
1
-------
Chapter 2
Technology Description
The objective of the ETV AMS Center is to verify the performance characteristics of
environmental monitoring technologies for air, water, and soil. This verification report provides
results for the verification testing of the SG-II mercury CEM. Following is a description of the
SG-II mercury CEM, based on information provided by the vendor. The information provided
below was not subjected to verification in this test.
The SG-II is an automated continuous emission monitor for elemental mercury (ITg°) and total
vapor-phase mercury (Hgx) in combustion flue gases and other gas streams. The SG-II consi sts
of a Model S235C400 mercury speciation module; an enclosed cabinet housing the SG-II
amalgamation atomic fluorescence mercury detector (PSA 10.525); a stream selector module
(PSA S235S100); a personal computer, monitor, and keyboard; and a mercury calibration source
(PSA 10.533). The speciation module converts oxidized mercury in the sample gas to Hg° by
means of a proprietary aqueous reagent, allowing separate detection of Hg° and Hgx. The
speciation module is approximately 75 centimeters (cm) wide x 45 cm deep x 90 cm high (30
inches wide x 18 inches deep x 36 inches high), and can be mounted on the stack being
sampled, or on a wall or supporting frame. The cabinet enclosing the other modules is approxi-
mately 75 cm wide x 75 cm deep x 180 cm high (30 inches wide x 30 inches deep x 72 inches
high) and is mounted on wheels. The speciation module and detector cabinet of the SG-II are
shown in Figure 2-1.
A heated Teflon diaphragm pump draws a filtered
sample flow of approximately five liters per
minute from the gas source into the speciation
module, which contacts the gas stream with the
aqueous reagents in two bubblers. Two separate
gas streams are thus produced, one of which has
been scrubbed of oxidized mercury and therefore
contains only Hg°. In the other gas stream,
oxidized mercury is reduced to FIg°, producing an
Hg° concentration equivalent to the original sum
of oxidized mercury and Hg°. These two gas
streams flow to the stream selector module.
Mercury in the selected gas stream is collected by
passage through a preconcentration trap and
subsequently thermally desorbed into the SG-II
2
Figure 2-1. PS Analytical, Ltd., SG-II
CEM
-------
detector, which has a detection limit of as little as 0.1 picograms of mercury. In this verification
test, a sample flow of 0.5 liter/minute was passed through the preconcentration trap for 1 to 2
minutes. The resulting detection limit for vapor-phase mercury is approximately
0.001 microgram per cubic meter (|ig/m3), with a linear dynamic range of up to 2,500 |ig/m3.
The PSA 10.533 mercury source provides a calibration gas and blank stream that can be
substituted for the sample stream on a scheduled or as-needed basis. This allows system bias
checking for the entire sampling system.
The SG-II uses Windows®-based operating software for calibrating and operating the instrument
and recording and displaying data. The duration, flow rate, and sequencing of the Hg° and Hgx
measurements are controlled by the software, as are the operation of the SG-II detector, the
graphical display of data, and the scheduling of internal calibration checks. The software checks
for alarm outputs from the various modules and can warn of any malfunctions of the system.
3
-------
Chapter 3
Test Design and Procedures
3.1 Introduction
This verification test was conducted according to procedures specified in the Test/QA Plan for
Field Demonstration of Mercury Continuous Emission Monitors at the TSCA Incinerator {l) The
purpose of the verification test was to evaluate the performance of mercury CEMs at a full-scale
field location, over a substantial period of continuous operation. The mercury CEMs were
challenged by stack gases generated from the thermal treatment of a variety of actual wastes in
the Toxic Substances Control Act Incinerator (TSCAI) at the East Tennessee Technology Park
(ETTP) in Oak Ridge, Tennessee. CEM responses were compared with reference mercury
measurements of total (Hgx), oxidized (Hgox), and elemental (Hg°) mercury. Mercury standard
gases were used to challenge the CEMs to assess stability in long-term operation, and the
instruments were operated for several weeks by TSCAI staff to assess operational aspects of their
use.
The performance of the SG-II was verified while monitoring emissions from the TSCAI that
were generated from treating actual waste. The reference method for establishing the
quantitative performance of the tested technologies was the Ontario Hydro (OH) method.(2)
The SG-II performance parameters addressed included
¦ Relative accuracy (RA) with respect to reference method results
¦ Correlation with reference method results
¦ Precision
¦ Sampling system bias
¦ Relative calibration and zero drift
¦ Response time
¦ Data completeness
¦ Operational factors.
Relative accuracy, correlation with the reference method, and precision (i.e., repeatability at
stable test conditions) were assessed for total and elemental mercury in the stack gas emissions.
Sampling system bias, calibration and zero drift, and response time were assessed for Hg° only,
using commercial compressed gas standards of Hg°. The data completeness, reliability, and
maintainability of the CEMs over the course of the verification test were assessed during several
weeks of continuous operation.
4
-------
This verification test was conducted jointly by the ETV AMS Center and the DOE. Under DOE
funding, Shaw Environmental, Inc. (SEI), under subcontract to Bechtel Jacobs Company LLC,
and the Hemispheric Center for Environmental Technology at Florida International University
(FIU-HCET) directed the field test. Reference method analyses were conducted by Severn Trent
Laboratories, and data analysis was conducted by the University of Tennessee. Funding for these
activities was provided by DOE's Transuranic and Mixed Waste Focus Area; the Characteriza-
tion, Monitoring and Sensor Technology Crosscutting Program; and FIU-HCET.
3.2 Facility Description
The TSCAI is designed and permitted for receiving, sorting, storing, preparing, and thermally
destroying low-level radioactive and Resource Conservation and Recovery Act (RCRA) mixed
waste contaminated with polychlorinated biphenyls. This waste is treated in a rotary kiln
incinerator with a secondary combustion chamber and off-gas treatment system for cleaning
combustion effluent gases. The TSCAI includes various support buildings, an unloading and
storage area, a tank farm, an incinerator area, concrete collection sumps, and carbon adsorbers.
A schematic of the TSCAI is shown in Figure 3-1, and photographs of the facility are shown in
Figures 3-2 and 3-3.
The TSCAI treats a wide range of waste categories, including oils, solvents and chemicals,
aqueous liquids, solids, and sludges. Solid and non-pumpable sludge material is typically
received and stored in metal containers and repackaged into combustible containers prior to
feeding. A hydraulic ram feeds containerized solids and sludges to the rotary kiln. Aqueous
waste is injected into the kiln through a lance. High heat-of-combustion liquids are burned in
either the rotary kiln or a secondary combustion chamber with gas burners. Both solids and
waste liquids are permitted for treatment in the primary combustion chamber, but only organic
liquids may be treated in the secondary combustion chamber. The typical temperature in the
primary combustion chamber is approximately 870°C (1,600°F), and in the secondary
combustion chamber is greater than 1,200°C (2,200°F).
Ash residue from the wet ash removal system is collected and handled through hazardous and
radioactive waste storage facilities. Selected residues are sent to a commercial landfill. Kiln
off-gas flows to the secondary combustion chamber. The off-gas from the secondary combustion
chamber then passes through a four-stage treatment system that includes a quench chamber and
scrubber treatment system for cooling, removing particulate matter, and neutralizing acidic
by-products. An induced-draft fan forces flue gases through the stack. Liquid waste generated by
the scrubber systems is treated by the Central Neutralization Facility, an adjacent on-site waste
water treatment plant. Solid waste, such as scrubber sludge, is collected in drums for off-site
disposal.
5
-------
N £
rO LO
Q LD
* Od
cr Q en
< I— UI
O (/) £D
ZDS
O CD <
O 2 X
UJ O O
(/) U
Q (/)
Om^
uj o
CD
D/j
#G
'S
ce
4)
ce
o
£
o
¦a
=
«
HH
<
u
in
H
a>
o
w
w
(Z3
fO
4>
•~
s
0/j
Q
3
o
6
-------
Figure 3-2. Overview of TSCAI Test Location. The incinerator stack is at
left, with waste feed area behind the stack. The trailers housing the mercury
CEMs for this test were located in the foreground at the base of the stack.
Figure 3-3. Side View of TSCAI Stack. Sampling platforms are at the left
and CEM trailers are at the lower right.
7
-------
The off-gas treatment system of the TSCAI produces a scrubbed, wet gas flow. The TSCAI stack
receives this water-saturated flue gas and vents it to the atmosphere. The stack is 100 feet high
and its inside diameter is 54 inches, with a gas velocity of approximately 20 feet per second. The
stack is equipped with several sample ports for flue gas sampling; a continuous emission
monitoring system for measuring carbon monoxide (CO), carbon dioxide (C02), and oxygen
(02); continuous sampling systems for radionuclides and metals; and two access platforms that
surround the full circumference of the stack at about 30 feet and 50 feet above ground level. The
combustion gas velocity is also monitored by means of the induced-draft fan current and
pressure drop across the fan.
The combustion process and off-gas cleaning systems are monitored by instrumentation for
process control and data collection. Operational parameters are automatically monitored and
logged by the incinerator Supervisory Control and Data Acquisition system.
Stack gas characteristics at the CEM sampling locations used in this test are summarized in
Table 3-1. Additional detail on the TSCAI configuration and operations are available in the
test/QA plan,(1) and in recent publications describing this test.(3"5)
Table 3-1. TSCAI Stack Gas Characteristics^
Parameter
Range
Units
Temperature
Static Pressure
Flow Rate
Velocity
02
C02
CO
Moisture
Particulate Matter
Loading
83.7-86.0 (182.6
186.8)
-0.25
6,065 -9,100
14,920-23,450
15.78 - 19.73
8.4-11.6
4.3-7.0
0-10.3
47.1 - 52.2
0.0012-0.0079
2.68-18.2
°C (°F)
inches H20
dry standard cubic feet (dscf) per
minute (min)
actual cubic feet per minute
feet per second
parts per million by volume
grain/dscf @ 7% 02
mg/dry standard cubic meters
(dscm) @ 7% 02
Values shown are actual conditions during OH reference method periods.
8
-------
3.3 Test Design
3.3.1 Equipment Setup
The SG-II was housed in the TSCAI Test Bed Mobile Laboratory Trailer located near the base of
the TSCAI stack. A dedicated data acquisition system was placed inside the trailer for logging
signals from the SG-II and other CEMs undergoing verification. The data logger was also
connected to the facility Supervisory Control and Data Acquisition system through an Ethernet
link to collect and log process parameters on the SG-II data logger.
At the lower of the two platforms on the TSCAI stack (i.e., about 30 feet above ground level),
one sampling port was dedicated to a probe that extracts stack gas to be analyzed for CO, C02,
and 02 by the facility CEMs. Other ports at this level were used for the SG-II and other CEMs
being tested.
For the SG-II, the vendor-supplied extractive sampling probe was connected to the CEM by
means of a 1/4-inch outside diameter (0.156-inch inside diameter), heated PFA Teflon sample
line. A vendor representative oversaw installation of the SG-II. The source sample was with-
drawn from the TSCAI stack through a Teflon-lined probe and then passed through a heated
fiberglass filter located outside the stack in a heater box maintained at 200°C (390°F). This filter
was periodically cleaned with a reverse pulse of air. A pump drew the sample gas through the
130-foot PFA Teflon sample line maintained at 200°C (390°F) to a sample splitter that provides
separate flows for determination of Hg° and total vapor-phase mercury (Hgx). The total sample
flow through the probe, filter, and Teflon line was approximately five liters per minute. Like all
CEMs in this verification test, the SG-II sampled at a single (fixed) point in the stack. This CEM
provided alternating batch measurements of Hg° and Hgx at approximately five-minute intervals.
Oxidized mercury (Hgox) can be determined by the difference between successive readings of
Hgx and Hg°. The SG-II does not determine particle-phase mercury. Verification of the per-
formance of the SG-II was based on comparison with the corresponding results from the OH
reference method.
3.3.2 Test Schedule
In this verification test, the CEMs undergoing testing sampled the TSCAI stack gas continuously
for nearly two months in the fall of 2002, while the TSCAI operated normally in destroying a
variety of waste materials. Stack sampling with the OH reference method was conducted in the
first week and the last week of the test, and between those two periods the CEMs operated
continuously for approximately five weeks. Table 3-2 summarizes the schedule of verification
testing at the TSCAI facility. Shown in this table are the activities conducted during various
periods, and the performance parameters addressed by those activities.
9
-------
Table 3-2. Mercury CEM Verification Test Schedule
Time Period (2002)
Activity
Performance Parameters
August 5-7
Installation and shakedown
—
August 8-11
OH method sampling; daily
challenge with mercury
standard gases
RA, correlation, precision;
sampling system bias,
calibration drift, zero drift,
response time
August 12 - September 15
Routine monitoring, with
scheduled challenges with
mercury standard gases
Calibration drift, zero drift
September 16-19
OH method sampling; daily
challenge with mercury
standard gases
RA, correlation, precision;
sampling system bias,
calibration drift, zero drift,
response time
The TSCAI was operated continuously during the first and last weeks of the test and was not
shut down overnight. Such continuous round-the-clock operation is the standard mode of
operation for the TSCAI. During the OH reference method sampling runs, the TSCAI burned
aqueous, solid, or a combination of aqueous and solid waste. The waste was characterized by
chemical analysis before the test began, and some measure of control of the stack mercury
concentration was achieved by varying the feed rate of aqueous waste and/or mixing solid and
aqueous waste materials.
After installation at the TSCAI in early August 2002, the CEMs went through a shakedown
period in which all CEMs sampled the facility stack gas. Sampling of the stack gas then
continued for the duration of the verification test, including during the performance of 10 OH
reference method sampling runs with dual OH trains on August 8 through 11. During this
period, the CEMs also were challenged with zero gas and with commercially prepared
compressed gas standards of Hg°. Vendor representatives oversaw installation and shakedown of
the CEMs and operated the CEMs through the first week of testing. Following this first OH
sampling period, vendor representatives trained site personnel on routine operation,
maintenance, and calibration checks of each of the mercury CEMs. The CEMs then operated for
five weeks with only routine attention and maintenance from TSCAI staff. During this period,
the staff recorded the maintenance and repair needs of each CEM and made observations on the
ease of use of each CEM. Finally, a second four-day period of OH method sampling with dual
trains was conducted on September 16 through 19, in which eight OH sampling runs were
conducted. The zero gas and mercury standard challenges were carried out by vendor
representatives through this period as well.
10
-------
The OH reference method results are presented in Section 4.2, along with evaluations of the
quality of these reference results. The commercial mercury gas standards are described in
Section 3.4.2, and the CEM results on those standards are reported in Section 6.
3.3.3 Reference Method Sampling
OH method sampling at the TSCAI was conducted at the upper platform on the stack (50 feet
above ground) by staff of SEI, who prepared the trains, conducted sampling at the TSCAI stack
using dual OH trains, and then recovered the resulting samples in a laboratory facility near the
TSCAI site. The dual OH sampling trains sampled isokinetically at separate ports located 90°
apart on the stack circumference and traversed the stack at points determined by EPA Method 1.
The two trains were interchanged from port to port at the halfway point in the OH sampling
period, so that the trains completed full and identical traverses of the stack during each OH run.
Severn Trent Laboratories supplied the chemical reagents used in the OH sampling train
impingers and performed the mercury analyses on the OH method samples. Containers for
collecting and storing samples were labeled for tracking by Severn Trent Laboratories and
subsequently supplied to the SEI field sampling team. Request for Analysis/Chain of Custody
forms accompanied the samples from the time of collection by the field sampling team through
analysis by the laboratory. Modified QA procedures for the OH method were followed, as
described in Section 4.3.1. In addition, two blank OH trains (one in each week for OH method
sampling) were spiked with known quantities of mercury to assess recovery in sample analysis.
The results of those mercury spikes are reported in Section 4.3.2.
Tables 3-3 and 3-4 summarize the schedule of OH sampling in the initial and final weeks of the
verification test, respectively, indicating the run number, date, and start and stop times of each
OH run. These tables also show the type of waste burned in each OH run. In most runs, the total
sampling period was made up of two separate periods of time, as necessitated by the port change
procedure noted above. In a few OH runs, other factors such as disturbances in the waste feed
required a stoppage in OH sampling; for those runs the total OH sampling period consists of
three or more segments, rather than two. A few OH runs of one hour duration were conducted in
the initial week of OH sampling (Table 3-3). However, it was recognized that this sample
duration allowed only a few measurements to be made within the OH sample period, by those
CEMs that provided sequential batch analyses, as opposed to continuous analysis.
Consequently, all OH periods in the final week (Table 3-4) were of two hours duration.
Note that the first 10 OH sampling runs (Table 3-3) were numbered 7 through 16, and the last
eight (Table 3-4) were numbered 18 through 25. The numbers 1 through 6 were assigned to OH
trains used in pre-test trial runs, and other numbers were assigned to trains used as field blanks
or as field spike trains. Each OH run number applies to two trains, designated A and B, which
were used in parallel sampling, as described above, or used for separate QA purposes. For
example, OH train 17A was spiked with known amounts of mercury, as described in Section
4.3.2, and train 17B was used as a blank. Similarly, train 28A was spiked and train 28B was a
blank.
11
-------
Table 3-3. Schedule of OH Method Sampling Runs in Initial Sampling Period
(August 8 - 11, 2002)
Run Number Date
7 8/8/02
8 8/8/02
9 8/9/02
10 8/9/02
11 8/10/02
12 8/10/02
13 8/10/02
14 8/11/02
15 8/11/02
16 8/11/02
Start Time
Stop Time
Waste Feed Type
09:10
09:28
09:43
10:25
Solids
10:55
11:55
14:40
15:40
Solids
16:10
17:10
10:50
11:50
12:15
13:15
Aqueous
14:35
15:35
16:10
17:10
Aqueous
9:35
10:05
10:25
10:55
Aqueous
12:15
12:45
13:10
13:40
Aqueous
15:00
15:30
15:50
16:20
Aqueous
08:20
08:50
09:10
09:40
Aqueous and Solids
10:40
10:52
11:05
11:23
Aqueous and Solids
11:45
12:15
13:45
14:15
Solids
15:00
15:30
12
-------
Table 3-4. Schedule of OH Method Sampling Runs in Final Sampling Period
(September 16 - 19, 2002)
Run Number Date
Start Time
Stop Time
Waste Feed Type
11:10
12:10
13:05
14:05
Aqueous
15:20
16:20
16:50
17:50
Aqueous
9:25
10:25
11:10
12:10
Aqueous and Solids
13:15
14:15
14:35
15:35
Aqueous and Solids
8:35
9:35
9:55
10:37
Aqueous
12:35
12:53
14:36
15:36
16:36
17:36
Aqueous
8:25
9:20
10:56
11:01
11:22
11:44
Aqueous and Solids
11:59
12:37
13:34
14:34
15:46
16:46
Aqueous and Solids
18
9/16/02
19 9/16/02
20 9/17/02
21
22
23
24
25
9/17/02
9/18/02
9/18/02
9/19/02
9/19/02
To ensure that the OH reference method and CEM data sets were indeed parallel and
comparable for each sampling period, the CEM vendors were notified of the start and stop times
of each OH period so that average analyte concentrations corresponding directly to the reference
method sampling period could be reported. The CEM vendors were given at least 15 minutes
notice prior to initiation of each OH method sampling run.
All OH trains were prepared, recovered, and analyzed in the same manner, with one exception.
The particulate filters from trains designated "A" and used for sampling at the TSCAI stack were
weighed before and after sampling to determine particulate matter loading in the flue gas,
whereas those from the trains designated "B" were not. The particulate loadings determined
from the A trains ranged from 0.0012 to 0.0079 grain/dscf (2.68 to 18.2 mg/dscm). Particulate
mercury was determined from the filter catch and probe rinse of both the A and B trains in all
samples, but was never found at significant levels (i.e., maximum values of particulate Hg were
less than 0.003 |ig/dscm). Given this negligible amount of particulate mercury, the total
vapor-phase mercury (Hgx) determined by the OH method can be considered as the total
mercury content of the stack gas.
13
-------
3.3.4 Verification Procedures
This section describes the test procedures that were used to verify mercury CEM performance on
each of the performance parameters listed in Section 3.1. Table 3-5 lists the quantitative
performance parameters and summarizes the types of data that were used to verify each of those
parameters.
Table 3-5. Data Used for SG-II Performance Evaluation
Performance Parameter
Objective
Comparison Based On
Relative Accuracy
Correlation with Reference
Method
Precision
Sampling System Bias
Relative Calibration/Zero
Drift
Response Time
Determine degree of
quantitative agreement with
reference method
Determine degree of correlation
with reference method
Determine repeatability of
successive measurements at
relatively stable mercury levels
Determine effect of the CEM's
sample interface on response to
zero gas and Hg° standard
Determine relative response to
zero gas and span gas over
successive days
Estimate rise and fall times of
the CEMs
Reference method results
Reference method results
Repetitive measurements
under stable facility
conditions
Response to zero gas and
Hg° standards at analyzer vs.
through sample interface
Zero gas and Hg° standards
CEM results at start/stop of
Hg addition
3.3.4.1 Relative Accuracy
The RA of the SG-II was verified using the OH reference method data. The Hgx and Hg°
readings of the SG-II during each OH sampling interval were averaged and compared with the
average of the results from the paired OH trains (see Section 4.2.1). The RA equation stated in
Section 5.1 was applied to the averaged CEM data, using the OH data as the reference values.
To optimize the comparability of the CEM and OH data, the OH sampling was coordinated with
the CEM operations as noted in Section 3.3.3.
3.3.4.2 Correlation with Reference Method
The correlation of SG-II results with the OH results was based on the same data used to assess
RA. No additional test procedures were needed to verify the correlation.
14
-------
3.3.4.3 Precision
Precision is the degree of variability of successive CEM readings under conditions of stable
mercury concentration. In this test, the TSCAI stack gas mercury concentrations resulted entirely
from the waste feed material being burned (i.e., no mercury was spiked into the flue gas).
Consequently, mercury concentrations in the TSCAI stack would be most stable when a waste
material of uniform mercury content was being fed into the incinerator at a uniform rate. For this
verification test, an aqueous waste was stockpiled in quantities sufficient for all the testing and
was characterized to document its mercury content. The aqueous feed rate data from the TSCAI
were then reviewed for the periods of each OH run in which only aqueous waste was burned (see
Tables 3-3 and 3-4). On the basis of the feed rate data, two OH runs (Runs 9 and 12, Table 3-3)
were selected as having relatively uniform feed rates. The variability of the responses of each
CEM during these two OH runs was then calculated to assess the variability of the CEM
response.
As described in Section 5.3, the assessment of precision is based on comparing the variability of
CEM readings to that of the aqueous feed rate, with variability expressed as a percent relative
standard deviation (RSD). This approach does not assume that the waste feed rate is the sole
factor affecting the variability of stack mercury concentrations, nor that the waste feed is
perfectly uniform in mercury content. This approach does provide a consistent basis for
reporting CEM variability in measuring mercury in the TSCAI stack gas.
3.3.4.4 Sampling System Bias
Sampling system bias was assessed using the commercial Hg°gas standards described in Section
3.4.2. To assess sampling system bias, a mercury gas standard was supplied at the analyzer
portion of the CEM, and separately at the stack gas sampling point of the CEM. Any difference
in the CEM responses in the two cases was attributed to the effect on the mercury level of the
sampling system components, i.e., the probe, filter, mercury conversion system, and transport
lines.
3.3.4.5 Relative Calibration and Zero Drift
Zero drift and calibration drift also were assessed using zero gas and the commercial Hg° gas
standards described in Section 3.4.2, respectively. Although the mercury standards were not
suitable for use as absolute standards, they did exhibit stable concentrations and so were useful
for assessing CEM relative calibration drift (see Section 3.4.2). These gases were supplied to the
CEMs on numerous occasions throughout the study; and the range, mean, and standard
deviation of the CEM readings were calculated as indicators of the drift of the instruments over
the course of the test. Both low (approximately 8 |ig/m3) and high (40 to 60 |ig/m3) mercury
standards were used for this evaluation. Zero gas (nitrogen) was used for a similar assessment of
the drift in CEM zero readings. The Hg° standards and zero gas were supplied to the analyzer
portion of each CEM for this assessment, with the exception of one CEM which was designed to
accept standard and zero gases only at its stack gas inlet.
15
-------
3.3.4.6 Response Time
Mercury CEM response time also was verified using zero gas and the commercial Hg° standards.
Response time was determined as the time required for the CEM to reach 95% of its final value,
after switching between zero gas and a mercury gas standard, or between different standards.
This procedure was used to assess both rise time and fall time. Because the SG-II is a batch (i.e.,
noncontinuous) analyzer, time response is reported as the percentage response to a step change
that is achieved in each measurement cycle.
3.3.4.7 Data Completeness
Data completeness was determined as the percentage of data that each CEM produced, relative
to the total possible data return. This parameter was evaluated both in terms of the percentage of
OH sampling runs for which each CEM produced data and in terms of the overall fraction of the
two-month test period in which the CEM was operating and producing data.
3.3.4.8 Operational Factors
Throughout the field period of testing the mercury CEMs at the TSCAI (August 8 -
September 19, 2002), the CEM vendors and TSCAI staff operating the CEMs recorded the
repair, routine maintenance, and expendable needs of each CEM and noted operational issues
such as the ease of use and calibration of the instruments. These observations are summarized
for the SG-II in Section 6.7.
3.4 Materials and Equipment
3.4.1 High Purity Gases
The high purity gas used for zeroing the CEMs during testing was commercial, ultra-high purity
(i.e., minimum 99.999% purity) nitrogen. Argon of ultra-high or industrial grade purity also was
obtained for those CEMs requiring it.
3.4.2 Mercury Standard Gases
Ten compressed gas standards of Hg° in nitrogen were obtained from Spectra Gases (Alpha,
New Jersey) for use in assessing drift and sampling system bias of the CEMs. These cylinders
were received in March 2002 and stored outdoors at the TSCAI site until the start of the verifica-
tion test. When used during the verification test, each mercury standard was placed inside the
instrument trailer near the CEMs for ease of access and to maintain the cylinders at room
temperature.
To assess their stability, the mercury gas standards were analyzed using various methods at
intervals before, during, and after the verification test. The 10 mercury standards were analyzed
by Spectra Gases in March, before shipment to the TSCAI site. In addition, a cold vapor atomic
absorption mercury analyzer (Seefelder Messtechnik) on loan from the EPA Office of Research
16
-------
and Development (EPA-ORD) was used to analyze the mercury gas standards at the TSCAI field
site. Analysis of all 10 cylinders was conducted with the Seefelder analyzer on August 8 and on
nine of the cylinders on October 17, after the field test had been completed. The contents of one
cylinder (CC133537) were unintentionally depleted during the verification test, and post-test
analysis was not possible. Eight cylinders, including the depleted one, were returned to Spectra
Gases, where the seven cylinders with remaining gas were analyzed on November 13.
SEI staff also analyzed the remaining two cylinders (CC133359 and CC133367) using a
modified version of EPA Method 101 A(6), with sampling performed on November 5 and 6,
respectively, for the two cylinders. Finally, the contents of these two cylinders were determined
on November 6 using the EPA-ORD Seefelder analyzer. Upon return to Spectra Gases, the gas
in these two cylinders was analyzed on November 21 by the vendor. The results of these diverse
measurements on each of the cylinders are summarized in Table 3-6. This table lists the cylinder
numbers, the various analytical results obtained on each cylinder (Hg° results in |ig/m3), and the
percent difference between the initial and final concentrations determined by the gas vendor.
Table 3-6. Results of Elemental Mercury Standard Analyses(a)
Post-Test
March 1
August 8
October 17
November
5 & 6
November
6
November
13 & 21
Difference
Between
Cylinder
Number
Initial Gas
Vendor
Certified
Analysis
(Hg/m3)
EPA-ORD
Seefelder
Analysis
(Hg/m3)
EPA-ORD
Seefelder
Analysis
(Hg/m3)
Method
101A Mini-
Train
Analysis
(Hg/m3)
EPA-ORD
Seefelder
Analysis
(Hg/m3)
Final Gas
Vendor
Certified
Analysis
(Hg/m3)
Final Gas
Vendor
Certified
Analyses
(%)
CC133146
14.0
11.3
11.4
NA
NA
12.1
-13.3
CC133172
64.3
44.7
42.4
NA
NA
44.7
-30.4
CC133174
59.6
46.0
45.2
NA
NA
47.5
-20.3
CC133345
11.2
7.9
6.8
NA
NA
5.6
-50.0
CC133357
53.1
37.6
37.1
NA
NA
40.1
-24.6
CC133359
60.6
37.2
34.5
30.6
35.4
44.7
-26.2
CC133367
10.2
6.3
5.4
4.6
5.6
5.6
-45.4
CC133537
15.8
14.9
NA
NA
NA
NA
NA
CC133612
57.8
36.9
34.4
NA
NA
35.4
-38.7
CC133619
59.6
39.9
37.8
NA
NA
40.1
-32.8
(a) All measurements corrected to 1 atmosphere and 20°C.
NA: Not available, analysis not performed.
It is apparent from the last column of Table 3-6 that there was a substantial decrease in all the
concentrations determined after the test by Spectra Gases, relative to those determined before the
17
-------
test by Spectra Gases. This finding suggests a decay in the mercury content of all the standards
between these March and November analyses by the gas vendor. However, Table 3-6 also shows
that all analyses subsequent to the initial analysis by Spectra Gases show better agreement. This
observation suggests that any such decay in concentration must have occurred primarily before
the August 8 analyses. Unfortunately, no measurements were made between the original March
1, 2002, Spectra Gases analyses and the August 8 analyses made during the first week of CEM
testing. Thus, there is no way to determine whether the decrease occurred as a sudden, step-wise
drop or a gradual decay over time. However, the important point regarding Table 3-6 is that the
data indicate stable mercury concentrations in all cylinders throughout the period of the
verification test.
This point is supported by Table 3-7, which shows the mean, standard deviation, and percent
RSD of all analyses of each mercury standard from August 8 on. Table 3-7 indicates that the
RSD values for six of the standard cylinders were about 4% or less, and the RSD values for the
other three cylinders having multiple analyses were less than 17%. These results indicate that the
contents of the mercury standard cylinders were stable over the course of the verification test
and, consequently, were suitable for assessing the stability of the CEMs themselves.
Table 3-7. Precision of Elemental Mercury Standard Measurements
Cylinder
Number
August 8 and Later
Mean
(lig/m3)
Standard Deviation
(Hg/m3)
RSD
(%)
CC133146
11.6
0.5
4.0
CC133172
43.9
1.3
3.0
CC133174
46.2
1.2
2.6
CC133345
6.8
1.1
16.8
CC133357
38.3
1.6
4.1
CC133359
36.5
5.2
14.2
CC133367
5.5
0.6
11.4
CC133537
14.9
NA(a)
NA
CC133612
35.6
1.3
3.5
CC133619
39.3
1.2
3.2
Not applicable for one data point.
18
-------
Spectra Gases conducted a quality review of its production and analytical records to determine
the cause of the concentration decay observed.(7) The preliminary conclusion from the review
was that an important step had been omitted from the manufacturing process. Spectra Gases
tested this hypothesis by manufacturing two separate cylinder batches of three cylinders each.
The first batch was made according to procedure, and the second batch was made with the
suspect step omitted from the manufacturing process. After the cylinders were prepared, each
cylinder was analyzed every seven days over a 49-day period. After 49 days, the concentration of
the first batch was stable, but the second batch (with the manufacturing step omitted) exhibited
a sharp decay in concentration. This test seemed to validate the theory that an important step had
been omitted from the manufacturing process, which led to a decrease in concentration from the
initial certified analysis of the gases used in the TSCAICEM test.
3.4.3 Mercury Spiking Standard
A National Institute of Standards and Technology (NIST)-traceable aqueous mercury standard,
with a concentration of 1,000 mg/L of Hg as HgO in dilute nitric acid, was obtained from VWR
Scientific (Catalog No. VW4217-1). This solution was Lot No. B2015064 and had an expiration
date of August 2003. Dilution of this standard in American Society for Testing and Materials
Type II water with added nitric acid was used to prepare the 10 |ig/mL and 30 |ig/mL spiking
solutions for the performance evaluation (PE) audit of the reference method (Section 4.3.2).
3.4.4 Sampling Trains
The SEI field sampling team supplied the glassware, probes, heater boxes, meter boxes, and
other associated equipment for the OH method sampling. Severn Trent Laboratories supplied the
chemical reagents and materials used in the OH sampling train impingers. Multiple trains were
prepared each day so that as many as six trains (i.e., three sampling runs with two trains each)
could be sampled in a single day, in addition to at least one blank train. The SEI field sampling
team recovered samples from OH method trains in a laboratory facility near the TSCAI site.
Containers for collecting and storing samples were purchased and labeled for tracking by Severn
Trent Laboratories. Samples were packaged and delivered by the field sampling team to Severn
Trent Laboratories.
3.4.5 Analysis Equipment
Laboratory equipment for sample recovery and analysis was provided by Severn Trent
Laboratories. This included all chemicals and solutions for rinsing train components and
recovering impinger samples, as well as cold vapor atomic absorption (CVAA) spectroscopy
equipment for mercury determination.
19
-------
Chapter 4
Quality Assurance/Quality Control
Quality assurance/quality control (QA/QC) procedures were performed in accordance with the
quality management plan (QMP) for the AMS Center(7) and the test/QA plan for this verification
test.(1)
4.1 Facility Calibrations
During this verification test, the TSCAI facility was operated normally to carry out its function
of destroying hazardous waste. Consequently, calibration procedures and schedules for the
TSCAI monitoring equipment were followed throughout the verification test, as required to
maintain RCRA certification of the TSCAI. These procedures, which included both weekly and
monthly calibrations, took precedence over the conduct of the verification test. Included in these
activities were calibrations of the 02 and C02 CEMs on the incinerator stack. Records of all
such calibrations are maintained in the operation files of the TSCAI.
Measurements that factored into the verification test results were also the subject of PE audits,
as described in Section 4.3.2. Those audits included checks of the facility 02 and C02 CEMs.
4.2 Ontario Hydro Sampling and Analysis
The preparation, sampling, and recovery of samples from the OH trains adhered to all aspects of
the OH method,(2) with minor modifications as described in Section 4.3.1. The preparation and
recovery of trains was carried out by SEI staff in a laboratory on the ETTP site; trains were
sealed for transport between the preparation/recovery laboratory and the TSCAI. Blank trains
were prepared in both the initial and final weeks of OH sampling, taken to the sampling location
on the TSCAI stack, and recovered along with the sampled trains. Reagent blanks were collected
as specified in the OH method. OH trains and resulting samples were numbered uniquely, and
samples were transferred to the analysis laboratory (STL) within about 24 hours of collection,
using chain-of-custody forms prepared before the field period. As described in Section 4.3.1,
trial OH sampling by SEI and OH sample analysis by STL were both subjected to a pre-test
evaluation before the field verification took place.
Because of the importance of the OH data in this verification, the following sections present key
data quality results from the OH data.
20
-------
4.2.1 Ontario Hydro Reproducibility
The results of the OH flue gas sampling are shown in Tables 4-1 and 4-2, for the initial
(August 8-11) and final (September 16 - 19) weeks of OH method sampling, respectively.
Each table indicates the OH run number, and lists the Hg°, Hgox, and Hgx results from the paired
OH trains (designated A and B) in each run. Also shown are the mean values of the paired train
results, and the relative percent difference (RPD) of each pair of results (RPD = difference
between A and B results divided by sum of A and B results expressed as a percentage). All
mercury results are in |ig/dscm, adjusted to 20°C (68°F) at 7% flue gas 02 content. Particulate
mercury is not shown in Tables 4-1 and 4-2. Particulate mercury was determined from the
particulate filters in both OH trains for each run, but was always less than 0.003 |ig/dscm. Thus,
particulate mercury was a negligible fraction of the total mercury in the TSCAI stack.
Inspection of Tables 4-1 and 4-2 shows that Hg° composed most of the total mercury value,
consistent with the extent of scrubbing of the TSCAI flue gas. The total mercury level was
controlled to some extent by the choice of waste feed material and the waste feed rate entering
the TSCAI. Total mercury was less than 1.7 |ig/dscm in the first two OH runs and then was
progressively increased throughout the rest of the first 10 OH runs (Table 4-1), peaking at about
200 |ig/dscm in OH Run 16. In the eight OH runs conducted during the final week of the test
(Table 4-2), total mercury ranged from 22 to 85.4 |ig/dscm. All the CEMs tested produced
readings of Hgx that generally paralleled this progression of mercury levels during the two weeks
of OH method sampling. Hgox was typically about 1% of the total mercury, and in 17 of the 18
OH runs, the Hgox results from both OH trains were less than 2 |ig/dscm. The one exception was
the Hgox level °f about 15 |ig/dscm observed with the peak mercury levels in OH Run 16, when
Hgox was about 7% of Hgx.
Tables 4-1 and 4-2 show generally close agreement between the A and B train results for all
three mercury fractions. The reproducibility of OH results is an important indicator of the
quality of the OH reference data for this verification test. Consequently, that reproducibility was
quantified by the RPD values for each A and B pair, by linear regression of the A and B train
results, including the correlation of the A and B results, and by calculation of the mean RSD of
the paired OH results for Hg°, Hgox, and Hgx. Considering the RPD values in Tables 4-1 and 4-
2, only one of the 18 RPD values for Hg° exceeds 7%, and the same is true for Hgx. The RPD
values for Hgox range from 0.5 to 39.4%, with a median of 9.7%. These results indicate close
agreement at the low Hgox concentrations found. Figure 4-1 shows the linear regression of B
train results versus A train results, for all three mercury fractions. The data for all three mercury
fractions lie closely along the 1-to-l line shown in this figure. Table 4-3 summarizes the results
of the linear regression, correlation, and %RSD analyses for the duplicate OH trains for Hg°,
Hgox, and Hgx. The correlation between paired trains is shown in terms of the coefficient of
determination (r2). Table 4-3 shows that the slopes of the paired OH regressions are all close to
1.0, the intercepts are near zero, and the r2 values are all approximately 0.99. Mean RSD values
of about 5.5%) were found for the paired results for Hg° and total mercury. The mean %>RSD for
Hgox was higher, due undoubtedly to the low Hgox levels in the TSCAI flue gas.
Based on the close agreement of the duplicate OH results for all mercury fractions in all sample
runs, the mean OH results in each run were used for comparison to the CEM results.
21
-------
S
w
VI
"m
i
cT
o
o
fS
90
VI
S
0/j
s
¦a
'•Z
a>
a.
0/j
=
«
in
©
.•-
s
K
0)
o
•-
¦a
w
o
%z
ce
=
o
Tf
4>
3
C3
H
RPD
*n
ON
CO
O
vq
t-»
r-
^t
00
(N
*r>
vd
o
vd
o
Mean
in
sq
co
^t
n
(N
'xi-
On
CO
^t
X
0
OX
ffi
§
o
o
o
o
o
o
O
o
CO
>r;
On
n
\l"
CO
Os
o
O
o
o
o
o
o
o
IT)
r;
^t
o
1—1
vd
Mean
On
00
00
CO
^t
i>
in
^t
o
00
^d
CO
(N
\l"
o'
\l"
r-'
CO
o
\i-
r-'
00
W)
ffi
co
On
CO
o
o
^t
G\
co
CO
^t
CQ
O
00
Os
CO
\6
CO
o'
\l"
r-'
CO
CO
'i-
\6
00
\o
t-»
On
r-»
*n
00
On
l
vq
in
%
O
t-»*
CO
00
\l-
r-'
\i-
^d
CO
00
CO
00
r-'
00
OH Run
Number
t-»
00
On
o
-
-------
s
w
0/j
i
cT
o
o
fS
cC
I
VO
•-
4>
A
s
0)
a
0)
in
¦a
'•Z
a>
a.
01)
"a
S
«
m
"«
£
S
©
.•-
s
VI
4>
©
¦-
¦a
w
o
%z
C8
=
o
ri
Tf
4>
3
ce
H
RPD
no
o
Os
Os
^t
o
co
N6
O
NO
Mean
no
^t
O
t-»
r-'
o
00
(N
in
cn
r;
On
On
cn
no
NO
co
n
O
O
o
o
O
O
O
o
«!
>r;
\|-
o
\l"
On
in
o
cn
r;
co
CO
r-
in
^t
CO
r-'
-------
0 50 100 150 200
OH Train A Hg, ug/dscm
Figure 4-1. Plot of Ontario Hydro Train B Results vs. Train A Results
Table 4-3. Results of Linear Regression, Correlation, and Percent Relative Standard
Deviation of Paired Ontario Hydro Train Results (n = 18)
Analyte Slope (CI)(a) Intercept (CI) |i.g/m3 r2 %RSD
Hg° 0.959 (0.027) 2.19 (1.73) 0.988 5.55
Hgox 1.104 (0.025) 0.053 (0.082) 0.992 20.9
Hgj 0.969 (0.025) 1.93 (1.65) 0.990 5.36
(a) (CI) = 98% confidence interval shown in parentheses.
4.2.2 Ontario Hydro Blank and Spike Results
None of the OH reagent blanks showed any detectable mercury. Also, OH sampling trains were
prepared and taken to the sampling location at the TSCAI stack on two occasions, and then
returned for sample recovery without exposure to stack gas. These blank OH trains provide
additional assurance of the quality of the train preparation and recovery steps. Four sample
fractions were analyzed from these blank trains: the particulate filter and probe rinse; impingers
1-3 (KC1); impinger 4 (H202); and impingers 5-7 (KMn04). Mercury was not detected in any of
the blank train samples. The detection limits for analysis of these fractions (in terms of mass of
mercury detectable) were 0.019 |ig, 0.005 |ig, 0.021 |ig, and 0.031 |ig, respectively, which
correspond to stack gas concentrations of less than 0.001 |ig/dscm under all sampling conditions
in this verification. Thus, the blank OH train results confirm the cleanliness of the OH train
24
-------
preparation and analysis procedures. The recovery of mercury spiked into blank train samples as
part of the PE audit also met the prescribed criteria, as described in Section 4.3.2.
Mercury spike recovery was also evaluated using sample fractions from selected trains used for
the 18 OH method runs in the TSCA stack. Those spike recoveries ranged from 85 to 101%, and
the results for duplicate spikes never differed by more than 4%, well within the 10% duplicate
tolerance required by the OH method.
4.3 Audits
4.3.1 Technical Systems Audit
Battelle's Quality Manager performed a pre-test evaluation and an internal TSA of the verifica-
tion test at the TSCAI. The TSA ensures that the verification test is conducted according to the
test/QA plan(1) and that all activities in the test are in compliance with the AMS Center QMP.(8)
The pre-test evaluation consisted of a visit on May 14, 2002, by a representative of the Battelle
Quality Manager to observe trial OH method sampling and to audit the laboratory conducting
the OH method analyses. Trial sampling was observed at the facilities of SEI, and analytical
procedures were observed at STL, both in Knoxville, Tennessee. The Battelle representative was
a staff member highly familiar with the sampling and analysis requirements of the OH method.
He used detailed checklists to document the performance of OH method train preparation,
sampling, sample recovery, chain of custody, and sample analysis. All observations were docu-
mented in an evaluation report, which indicated no adverse findings that could affect data
quality. An amendment to the test/QA plan(1) was prepared as a result of this evaluation,
documenting several minor procedural changes implemented in the OH sample recovery by
STL. These procedural changes were based on the experience of STL personnel in conducting
OH mercury analyses and other metals analyses, as well as on the numbers and types of analyses
needed for this verification. The most significant such changes were
¦ The analysis of one matrix spike duplicate for each type of sample received (i.e., filter catch
and probe rinse, KC1 impingers, H202 impingers, etc.), rather than the duplicate and
triplicate analyses stated in section 13.4.2.3 of the OH method.
¦ The analysis of one spiked sample for each type of sample received, rather than a spike after
every 10 samples as stated in section 13.4.2.4 of the OH method.
¦ The use of a 25% tolerance on spike recovery values based on the requirements of EPA
Method 7460 for metals analysis, rather than the 10% tolerance stated in section 13.4.2.4 of
the OH method.
The Battelle Quality Manager conducted the TSA in a visit to the TSCAI test location on August
8, 2002, which was the first day of OH sampling in the first intensive period. In that visit he
toured the incinerator and CEM locations; observed the OH method sampling; observed OH
sample recovery and documentation in the on-site laboratory at the ETTP; reviewed Battelle
25
-------
notebooks, gas standard certifications, and the CEM data acquisition procedures; and conferred
with the CEM vendors and facility personnel. The TSA report from this audit found no issues
that could adversely affect data quality. All records from both the pre-test evaluation and the
TSA are permanently in the custody of the Battelle Quality Manager.
4.3.2 Performance Evaluation Audits
A series of PE audits was conducted on several measurement devices at the TSCAI facility to
assess the quality of the measurements made in the verification test. These audits were
performed by Battelle staff and were carried out with the cooperation of SEI staff. These audits
addressed only measurements that factored directly into the data used for verification, i.e., the
CEMs undergoing testing were not the subject of the PE audit. Each PE audit was performed by
analyzing a standard or comparing to a reference that was independent of standards used during
the testing. Each PE audit procedure was performed once during the verification test, with the
exception that blank OH sampling trains were spiked with a mercury standard during both the
first and last intensive OH sampling periods, approximately six weeks apart. Table 4-4
summarizes the PE audit results on several measurement devices at the TSCAI facility.
Table 4-4. Summary of PE Audits
Measurement
Observed
Acceptable
Audited
Date
Audit Method
Difference
Difference
Flue gas 02
8/9/02
Comparison to independent
02 measurement
0.16% 02(a)
0.24% 02
1% 02
Flue gas C02
8/9/02
Comparison to independent
0.0%) of reading(b)
10% of
C02 measurement
3.3%o of reading
reading
OH gas flow rate
8/7/02
Comparison to independent
flow measurement
1.3%(c)
3.2%
5%
Flue gas
8/7/02
Comparison to independent
0.33%(c)
2% absolute
temperature
temperature measurement
0.07%
temperature
Barometric
8/7/02
Comparison to independent
0.5" H20
0.5" H20
pressure
barometric pressure
measurement
Impinger weights
8/7/02
Weighing certified weights
0.37%
greater of 1%>
(electronic
(1.7 gat 454 g)
or 0.5 g
balance)
The two results shown are for the two Siemens Oxymate 5E units (Serial Nos. D1-447 and D3-491, respectively)
used at the TSCAI facility.
(b) The two results shown are for the two Siemens Ultramat 22P units (Serial Nos. U01-483 and A03-277,
respectively) used at the TSCAI facility.
(c:i The two results shown are for the two NuTech meter boxes designated Unit A (Serial No. 80563) and Unit B
(Serial No. 008068), respectively.
Table 4-4 shows the type of measurement audited, the date the PE audit was conducted, the
basis for the audit comparison, the difference between the measurement and the PE audit value,
26
-------
and the acceptable difference stated in the test/QA plan. As Table 4-4 shows, all the PE audits
met the required tolerances stated in the test/QA plan.(1) The PE audits for 02 and C02 were
conducted by sampling the same cooled and dried flue gas analyzed by the facility's CEMs for
these gases. The 02 and C02 content of the flue gas were about 9.5% 02 and 6% C02. The
independent audit monitor and the facility CEMs sampled this gas simultaneously for the PE
audit. As noted in the footnotes to Table 4-4, both of the dual 02 monitors and dual C02
monitors installed at the TSCAI facility were audited. The gas flow rate measurements of the
two OH trains were audited using a certified mass flow meter. The temperature measurements
were audited at ambient temperature (approximately 27°C), rather than in the flue gas, because
of the limited access to the TSCAI stack. The PE audit of the electronic balance used certified
weights of approximately 200 and 500 grams; the observed agreement shown in Table 4-4 is for
the 500-gram weight, which showed the greater percentage deviation. A planned audit of the
flue gas static pressure® was not conducted, because the minimal differential relative to
atmospheric pressure (approximately -0.25 inches of H20) makes this measurement both
difficult to audit and relatively unimportant in calculating the reference mercury results. An
amendment to the test/QA plan was prepared and approved to document this change.
The PE audit of the OH train mercury recovery and analysis was performed by spiking blank
OH trains with NIST-traceable mercury solutions. In each case, impingers 1 (KC1), 4
(H202/HN03), and 5 (KMn04/H2S04) of a blank OH train were spiked. In the first week of OH
sampling, each impinger was spiked with 1 mL of a 10-|ig/mL mercury solution; and in the final
week of OH sampling, each impinger was spiked with 1 mL of a 30-|ig/mL mercury solution.
Table 4-5 identifies the OH trains that were spiked, the date of the spike, the amount of the
spike, and the analytical results for each spiked impinger in the train (i.e., impingers 1, 4, and 5
of each OH train).
Table 4-5. Results of PE Audit of OH Train Recovery and Analysis
Impinger
Hg Spiked
Hg Found
Observed
Target
Train
Date
Number
(M-g)
(M-g)
Agreement
Agreement
17A
8/8/02
1
10
9.7
3%
25%
4
10
7.8
22%
25%
5
10
8.3
17%
25%
28A
9/16/02
1
30
32.5
8.3%
25%
4
30
26.7
11.0%
25%
5
30
30.6
2.0%
25%
Table 4-5 shows that all of the six spike recoveries were well within the target of 25% agreement
with the spiked values that was stated in the amended test/QA plan (see Section 4.3.1). Further-
more, four of the six results were near or within the 10% tolerance stated in the OH method.(2)
These results support the validity of the OH reference method results used in this verification.
27
-------
4.3.3 Data Quality Audit
An audit was conducted to trace the test data from initial acquisition, through reduction and
statistical comparisons, to final reporting. All calculations performed on data leading to
verification results were checked. The Battelle Quality Manager reviewed the procedures and
results of this audit, and conducted his own independent review of a small portion of the data.
28
-------
Chapter 5
Statistical Methods
This chapter presents the statistical procedures that were used in calculations for verifying the
performance factors listed in Section 3.1.
5.1 Relative Accuracy
RA was verified by comparing the SG-II results against the reference results for each parameter
that the SG-II measured. The average of the paired OH train results was used as the reference
value for each OH run. The SG-II readings in each OH run were averaged for comparison to the
reference data.
The RA of the SG-II with respect to the reference method was calculated using
Where
R + 0)
RA = - v v 1
X
RM
M = the absolute value of the arithmetic mean of the differences, d, of the paired
SG-II and reference method results
Xrm = arithmetic mean of the reference method results
n = number of data points
t0975 = the lvalue at the 97.5% confidence with n-1 degrees of freedom
SD = standard deviation of the differences between the paired SG-II and reference
method results.
RA was calculated separately for the first and last weeks of OH sampling (n = 10 and n = 8,
respectively) and for all reference data combined (n = 18).
5.2 Correlation with Reference Method
Correlation of the SG-II with the OH method was calculated using the same data used to assess
RA. Correlation was calculated for each parameter measured by the SG-II. The coefficient of
determination (r2) was calculated to determine the degree of correlation of the SG-II results with
29
-------
the reference method results. This calculation was conducted using data from the first week, the
last week, and both weeks of OH reference method samplings.
5.3 Precision
As described in Section 3.3.4.3, precision was assessed based on the individual readings
provided by the SG-II over the duration of OH method sampling Runs 9 and 12. Precision of the
SG-II was determined by calculating the percent relative standard deviation (RSD) of a series of
SG-II Hgx measurements made during stable operation of the TSCAI in these OH runs. The
%RSD is the ratio of the standard deviation of those readings to the mean of the readings,
expressed as a percentage:
SD = standard deviation of the SG-II readings
X = mean of the SG-II readings.
The calculated precision values from Equation 2 include the variability of the TSCAI stack gas
mercury concentration, as well as the variability of the SG-II itself. To estimate the precision of
the SG-II, it was assumed that the two sources of variability combine in root-mean-square
fashion, with the variability of the TSCAI mercury concentration represented by the variability
of the aqueous waste feed rate. Consequently, the CEM precision was estimated in terms of a
%RSD by means of Equation 3:
where %RSDr is the relative standard deviation of the CEM readings, %RSDWF is the relative
standard deviation of the aqueous waste feed readings, and %RSDCEM is the resulting relative
standard deviation attributable to the CEM variability. It must be noted that the total variability
of the TSCAI may not be fully represented by the variability of the waste feed rate. Conse-
quently, the CEM variability (%RSDCEM) calculated from Equation 3 must be considered as the
maximum variability that could be attributable to the CEM.
5.4 Sampling System Bias
Sampling system bias (B) reflects the difference in SG-II response when sampling Hg° standard
gas through the SG-II's entire sample interface, compared with that when sampling the same gas
directly at the SG-II's mercury analyzer, i.e.:
%RSD = ^xlOO
X
(2)
where
%RSDr = [(%RSDwf)2 + (%RSDcem)2]1/2
(3)
B =
XlOO
(4)
30
-------
where
R, = SG-II reading when the standard gas is supplied at the sampling inlet
Ra = SG-II reading when the standard is supplied directly to the analyzer.
Since the effect of the inlet is expected to be a negative bias on measured Hg levels, R, is
expected to be less than Ra. Equation 4 thus gives a positive percent bias value for what is
understood to be an inherently negative bias. In rare instances R, was found to exceed Ra slightly
due to normal instrument variation. In such instances, B was reported as 0.0%.
The purpose of this part of the verification was to assess the bias introduced by the sampling
probe, filter, and long (> 100-foot) sampling lines in sampling Hg°. It must be pointed out that
delivery of the standard gas to the sample inlet also required a Teflon line over 100 feet in length.
Thus, the observed bias may include a contribution from the standard gas delivery system, as
well as from the sampling system.
5.5 Relative Calibration and Zero Drift
Calibration and zero drift were reported in terms of the mean, relative standard deviation, and
range (maximum and minimum) of the readings obtained from the SG-II in the repeated
sampling of the same Hg° standard gas and of zero gas. The relative standard deviation of
standard gas or zero gas readings was calculated according to Equation 2 above. This calculation,
along with the range of the data, indicates the variation in zero and standard readings.
The SG-II was challenged with three Hg° gas standards in this test, cylinders CC133537,
CC133612, and CC133619, which had nominal average Hg° concentrations of 14.9, 35.6, and
39.3 |ig/m3, respectively. These nominal averages are based on all analyses of the gas standards
from August 8, 2002, through November 21, 2002, (Table 3-7), i.e., excluding the vendor's
initial pre-test analysis of the standards in March 2002.
5.6 Response Time
The response time refers to the time interval between the start of a step change in mercury input
and the time when the SG-II reading reached 95% of the final value. Both rise time and fall time
were determined. SG-II response times were obtained in conjunction with a calibration/zero drift
check or sampling system bias check by starting or stopping delivery of the mercury standard gas
to the SG-II or sampling interface, recording all readings until stable readings were obtained, and
estimating the 95% response time.
5.7 Data Completeness
Data completeness was assessed by comparing the data recovered from the SG-II with the
amount of data that would be recovered upon completion of all portions of these test procedures.
31
-------
5.8 Operational Factors
Maintenance and operational needs were documented qualitatively, both through observation and
through communication with the vendor during the test. Factors noted included the frequency of
scheduled maintenance activities, the down time of the SG-II, and staff time needed for main-
taining it during the verification test.
32
-------
Chapter 6
Test Results
The results of the verification test of the SG-II are presented below for each of the performance
parameters.
6.1 Relative Accuracy
Table 6-1 lists the OH reference method results and the corresponding SG-II results for Hg° and
Hgx in all 18 OH sampling runs. The OH results are the averages of the results from the paired A
and B trains in each run; the SG-II results are the averages of the SG-II readings over the period
of each OH run.
The SG-II operates by switching at five-minute intervals between Hg° and Hgx measurements,
which are conducted alternately rather than simultaneously. The SG-II also determines Hgox by
difference between successive Hg° and Hgx measurements. This difference approach is subject
to greater uncertainty when Hg° comprises the great majority of Hgx, as in this test. In fact, the
SG-II Hgox results were found to be unrealistic relative to the OH reference method results. For
example, Table 6-1 shows instances in which the SG-II Hg° value exceeded the Hgx value, most
notably in OH Run 16. In that run the waste feed was packets of solid waste of relatively high
mercury content, with no accompanying liquid waste. This waste is likely to have produced a
variable mercury content in the flue gas, which may have been difficult to determine accurately
with the SG-II's batch analysis process. After inspection of the data, it was concluded that the
conditions of this test were not appropriate for determining Hgox with the SG-II, and no relative
accuracy comparison for Hgox was conducted.
Table 6-1 shows that the SG-II readings were usually lower than the corresponding OH results
for Hg° and Hgx, sometimes by a factor of two or more, throughout both weeks of OH sampling.
Table 6-2 shows the resulting RA values for the SG-II, for Hg° and Hgx, based on the first week,
the last week, and both weeks of OH sampling. The RA for Hgx was substantially improved in
the last week of OH sampling, relative to the first week, but the RA for Hg° did not show the
same improvement. The overall RA results for Hg° and Hgx are 54.7% and 59.8%, respectively.
The RA results for Hgx improve to 42.8% if OH Run 16 is excluded, but no other substantial
change in RA for Hg° or Hgx would result from exclusion of any single OH run.
33
-------
Table 6-1. Summary of Results from OH Reference Method and SG-II (|i,g/dscm)
OH Run
Hg°
((ig/dscm)
HgT
((ig/dscm)
Date Number
OH
SG-II
OH
SG-II
8/8/2002 7
1.49
1.64
1.65
1.61
8
0.18
0.09
0.36
0.25
8/9/2002 9
18.1
8.04
18.4
11.6
10
36.8
11.8
37.2
26.8
8/10/2002 11
42.3
29.8
42.6
30.2
12
47.6
29.5
48.1
33.2
13
37.4
28.0
37.8
30.6
8/11/2002 14
40.7
23.4
41.1
22.5
15
67.5
40.2
68.9
40.4
16
184.4
134.6
198.8
91.9
9/16/2002 18
70.9
41.7
71.6
56.4
19
77.0
48.9
77.5
64.7
9/17/2002 20
83.5
42.2
84.1
60.5
21
52.2
27.0
52.5
29.9
9/18/2002 22
22.8
14.8
23.1
15.8
23
32.8
22.1
33.1
22.9
9/19/2002 24
23.2
7.92
23.4
13.3
25
59.2
20.1
60.0
29.2
Table 6-2. Relative Accuracy Results for the SG-II
Relative Accuracy (%)
Test Period
Hg°
HgT
First Week (n = 10)
57.7
87.2
Last Week (n = 8)
66.4
44.1
Overall (n = 18)
54.7
59.8
6.2 Correlation with the Reference Method
The correlations of the SG-II readings with the OH results for Hg° and Hgx were calculated
using the data shown in Table 6-1. To illustrate the correlations, Figures 6-la and 6-lb show
linear regression plots of the SG-II results against the corresponding OH results for Hg° and Hgx,
respectively. The linear regression equations and coefficients of determination (r2) are shown on
the graphs. Table 6-3 shows the r2 values for the first and last weeks of OH sampling, and for the
two periods combined, for Hg° and Hgx.
34
-------
160
140
120
y = 0.7048x-5.6155
r2 = 0.9478
" 100
80
60
40
20
0
20
40
60
80
100
120
140
160
180
200
Ontario Hydro Hg°, ug/dscm
Figure 6-la. Linear Regression Plot of SG-II Hg° Results Against OH
Results
120
100
y = 0.4973x +6.8904
r2 = 0.8753
E
o
«
5
u>
3
I-
O)
x
a
tn
<
tn
a.
~ *,
0
20
40
60
80
100
120
140
160
180
200
220
Ontario Hydro HgT, ug/dscm
Figure 6-lb. Linear Regression Plot of SG-II HgT Results Against OH
Results
35
-------
Table 6-3. Coefficients of Determination (r2) for Correlation of SG-II Results with OH
Results
r2
Test Period
Hg°
HgT
First Week (n = 10)
0.982
0.963
Last Week (n = 8)
0.827
0.894
Overall (n = 18)
0.948
0.875
Table 6-3 shows that the SG-II results were strongly correlated with the OH results for both Hg°
and Hgx, with all r2 values exceeding 0.82. The overall r2 value for Hgx was 0.875.
6.3 Precision
Table 6-4 summarizes the observed precision of the SG-II, in terms of the stability of its Hgx
readings during two periods of relatively stable introduction of mercury in aqueous waste into
the TSCAI. For OH Runs 9 and 12, Table 6-4 shows the %RSD of the aqueous waste feed rate
into the TSCAI, the corresponding %RSD of the SG-II Hgx readings, and the resulting estimate
of the variability attributable to the SG-II, calculated according to Equation 3 in Section 5.3.
(The integrated OH and average SG-II results in these two runs are shown in Table 6-1 above.)
Table 6-4. Precision of the SG-II During OH Runs 9 and 12
Aqueous Feed Rate
SG-II HgT
Maximum CEM
OH Run
Variability
Readings
Variability
Number
(%RSDwf)
(%RSDr)
(%RSDcem)
9
2.4
9.2
8.9
12
13.9
21.1
15.9
The results in Table 6-4 show that the SG-II readings of Hgx exhibited variability of about 9 to
21%RSD under conditions of relatively stable mercury feed into the TSCAI. The maximum
variability attributable to the SG-II was 8.9%RSD in OH Run 9 and 15.9%RSD in OH Run 12.
6.4 Sampling System Bias
On four days during the verification test, an elemental mercury gas standard was supplied directly
to the analyzer of the SG-II, and then to the inlet of the CEM's sampling system on the TSCAI
stack. Table 6-5 shows the date, the mercury standard, and the SG-II readings obtained for each of
these sampling system bias checks. The SG-II responses are the average of two or more successive
stable readings on the standard gas. When the standard gas was supplied at the inlet, it was
36
-------
analyzed through both the Hg° and Hgx channels of the SG-II. Since these are separate flow paths
within the SG-II, separate calculations of the sampling system bias were made using Equation 4 in
Section 5.4, and the results are shown in Table 6-5. In the sampling system bias checks conducted
on August 8 and 9 and on September 17, all analyses were completed within about one hour.
However, in the test on September 19, the analyses of gas supplied to the inlet occurred about
seven hours after those at the analyzer.
Table 6-5. Sampling System Bias Results
Response at Inlet (Rj) Bias(b)
Hg°
(ng/m3)
_ Response at Analyzer _
(%)
Date
Standard'3'
Hg°
HgT
(Ra) ((ig/m3)
Hg°
HgT
8/8/02
CC133537
16.86
17.35
18.10
6.9
4.1
8/9/02
CC133537
13.45
13.28
13.98
3.8
5.0
9/17/02
CC133537
14.74
15.29
15.17
2.8
0.0
9/17/02
CC133619
36.42
37.02
38.86
6.3
4.7
9/19/02(c)
CC133537
14.13
14.70
15.11
6.5
2.7
9/19/02(c)
CC133619
37.56
36.11
39.77
3.1
6.9
See Section 3.4.2 for information on mercury standard gases.
(b) Calculated according to Equation 4, Section 5.4.
(c) Measurements at analyzer and at inlet separated by about seven hours.
Table 6-5 shows that all sampling system bias results were less than 7% for both measurement
channels of the SG-II, and most results were less than 5%. In one of the bias checks, the average
SG-II Hgx response to standard gas at the inlet actually slightly exceeded the average response at
the analyzer, indicating no sampling system bias.
6.5 Relative Calibration and Zero Drift
Mercury gas standards and zero gas (high purity nitrogen) were analyzed by the SG-II periodically
throughout the verification test to assess the drift in calibration and zero response of the CEM. The
results of these analyses are shown in Table 6-6, which lists the date of each analysis and the SG-II
Hg° readings on zero gas and on the mercury standards. Four instances of multiple analyses
conducted on a single day are included in the table. Also shown in Table 6-6 are the mean,
standard deviation, %RSD, and range of the SG-II readings on zero gas and on the mercury
standard gases.
Table 6-6 shows that the zero gas readings of the SG-II averaged -0.05 |ig/m3 over the duration of
the verification test, with a standard deviation of 0.10 |ig/m3. These results indicate minimal drift
of the zero readings of the SG-II. Eighteen analyses of the lowest concentration standard
(CC133537) took place over a period of about six weeks and exhibited an RSD value of 11.9%.
The 12 analyses of the middle concentration standard (CC133612) over a five-week period showed
an RSD of 10.2%. It should be noted that the results for these two standards are greatly dependent
on the first set of analyses conducted on August 22, which produced by far the lowest
37
-------
Table 6-6. Calibration and Zero Drift Results
SG-II Hg° Readings (|ig/m3)
Mercury Standard Mercury Standard Mercury Standard
Date
Zero Gas(a)
CC133537
CC133612
CC133619
8/8/02
0.00
18.10
8/9/02
0.05
13.98
8/10/02
-0.06
14.94
36.96
8/11/02
-0.28
14.52
37.55
8/12/02
-0.11
14.50
8/12/02
15.17
39.03
8/22/02
9.56
25.54
8/22/02
13.91
36.29
8/29/02
16.50
36.28
8/29/02
12.74
31.20
9/4/02
13.65
34.00
9/5/02
15.38
36.75
9/11/02
15.29
34.86
9/12/02
14.62
35.61
9/16/02
0.02
13.52
34.10
32.47
9/17/02
-0.00
15.17
38.86
9/18/02
-0.06
15.79
39.20
9/19/02
0.02
15.11
38.77
Mean
-0.05
14.58
34.85
37.32
Std. Dev.
0.10
1.74
3.55
3.24
%RSD
—
11.9%
10.2%
8.7%
Range
-0.28-0.05
9.56-18.10
25.54- 39.03
32.47-39.20
(a) High purity nitrogen used for zero checks.
readings for both of those standards. Exclusion of those values would result in RSD values of
8.3% and 5.9% for standards CC133537 and CC133612, respectively. Finally, the four analyses
of the highest concentration standard (CC133 619) over a four-day period resulted in an RSD of
8.7%.
6.6 Response Time
On several occasions during the verification test, successive readings were recorded at times when
the SG-II switched from zero gas to a mercury standard gas, or vice versa, or from one standard gas
to another. These records were used to evaluate the response time (i.e., the rise and fall times) of
38
-------
the SG-II. The SG-II is a batch analyzer, which, for the analysis of response time, provided readings
at intervals of a few minutes. Consequently, the evaluation of response time, is reported in terms of
the extent of response to a step change in mercury concentration that was achieved in each
measurement cycle. Table 6-7 summarizes the response time data, showing the date and time of
each reading, the indicated mercury concentration, and the resulting percent rise or percent fall in
successive readings. Table 6-7 includes primarily data from tests in which the zero and standard
gases were supplied to the mercury analyzer of the SG-II, as well as a few data from tests in which
the gases were supplied directly to the inlet of the SG-II's sampling system.
Table 6-7 shows that the percentage response of the SG-II to rising mercury levels within one
measurement cycle ranged from 84.7 to 99.6%, and averaged 94.8%. Thus the 95% rise time of the
SG-II was essentially one measurement cycle. Table 6-7 also shows that the percentage response of
the SG-II to falling mercury levels within one measurement cycle was nearly 100%, i.e., response
dropped by 99.3% or more in all four cases in which fall time could be determined. Thus the 95%
fall time of the SG-II is also within one measurement cycle.
6.7 Data Completeness and Operational Factors
The operational factors associated with use of the SG-II were evaluated by SEI staff, who operated
the SG-II during the five-week period of routine monitoring. These operators recorded observations
on daily maintenance, repair, expendables use, waste generation and disposal, etc., in a separate
logbook for each CEM. The SG-II vendor also recorded activities in the first and last weeks of the
field period. Particular attention was paid to the cause and extent of any down time of the SG-II
during the field period. Table 6-8 lists the dates of significant down time of the SG-II during the
entire verification period, along with the duration of the down time, the duration of the service
time, and a description of the cause and resolution of each problem.
The operation and maintenance activities listed in Table 6-8 include only those that were not
required by the test/QA plan (e.g., time required to conduct zero and standard gas checks was not
considered down time) and that were responsible for either CEM down time or for operator
intervention. In the case of the SG-II, routine calibration checks using the CEM's internal mercury
standard (the "Cavkit") also were not included in Table 6-8 because these checks required minimal
operator intervention. The Cavkit checks are typically used as an automated QC feature of the
SG-II. On the other hand, the time needed for recalibration of the SG-II with an absolute standard
mercury vapor source is included in Table 6-8 because this is a manual procedure carried out
externally to the CEM. As Table 6-8 shows, the most common maintenance needed on the SG-II
included frequent calibration checks, preparation of the aqueous reagents, and changing of argon
cylinders. The latter two activities were performed every few days. The most common repair
problems had to do with the liquid flow system in the mercury speciation module. Problems
occurred in maintaining the liquid reagent flows and levels in the impingers; in mass flow control
of the sample gas flows through the impingers; in shutdowns due to moisture carryover from the
impingers; and in the buildup of a yellow precipitate that clogged the liquid flow. These problems
were the cause of four periods of down time ranging from 10 hours to more than 34 hours, on
August 15, 20, and 31 and September 8. The total down time
39
-------
Table 6-7. Summary of Data Used to Estimate Response Time
Date
Time
Zero/Span
Gas
Analyzer/
Inlet(a)
SG-II Response
(ng/m3)
Result
8/10/02
16:55
CC133537®
A
14.94
17:00
CC133612
A
36.62
98.5% rise in one cycle
17:05
CC133612
A
36.90
17:10
CC133612
A
36.96
8/11/02
17:26
CC133537
A
14.52
17:31
CC133612
A
35.58
91.4% rise in one cycle
17:35
CC133612
A
36.67
96.2% rise in two cycles
17:40
CC133612
A
37.55
17:45
Z
A
-0.01
99.3% fall in one cycle
17:50
z
A
-0.26
17:54
z
A
-0.28
9/4/02
9:50
CC133537
A
13.65
9:56
CC133612
A
33.44
97.2% rise in one cycle
10:01
CC133612
A
34.00
9/11/02
13:19
CC133537
A
15.29
13:24
CC133612
A
33.08
84.7% rise in one cycle
13:29
CC133612
A
36.21
99.6% rise in two cycles
13:34
CC133612
A
36.29
9/12/02
11:06
CC133537
A
14.62
11:11
CC133612
A
37.07
99.6% rise in one cycle
11:16
CC133612
A
37.17
9/16/02
19:07
Z
A
0.02
19:12
CC133619
A
31.83
98.5% rise in one cycle
19:16
CC133619
A
32.33
9/17/02
16:18
CC133537
A
15.17
16:23
CC133619
A
38.24
97.4% rise in one cycle
16:28
CC133619
A
38.24
16:33
CC133619
A
38.86
16:37
Z
A
0.07
99.8% fall in one cycle
16:42
z
A
0.00
40
-------
Table 6-7. Summary of Data Used to Estimate Response Time (continued)
Zero/Span
Analyzer/
SG-II Response
Date
Time
Gas
Inlet(a)
(ng/m3)
Result
9/19/02
10:16
CC133537
A
15.11
10:21
CC133619
A
37.89
90.8% rise in one cycle
10:26
CC133619
A
38.82
94.5% rise in two cycles
10:31
CC133619
A
40.21
10:36
CC133619
A
38.77
10:40
Z
A
0.16
99.6% fall in one cycle
10:45
z
A
0.06
10:50
z
A
0.05
10:55
z
A
0.02
9/19/02
18:32
CC133537
I
14.70
18:37
Z
I
0.16
99.5% fall in one cycle
18:41
z
I
0.08
(a) Indicates whether zero and standard gases were supplied to the CEM's mercury analyzer (A) or to the inlet (I) of
the CEM's sampling system.
(b) See Section 3.4.2 for information on mercury standard gases.
experienced during the six-week test period was 7,100 minutes (approximately 118 hours, or
about 5 days), and the required service time during the same period was 1,730 minutes
(28.8 hours). The total down time amounted to about 11.7% of the total duration of the field
period (August 8 through September 19), so that data completeness was 88.3%.
The cost of the SG-II also was considered as an operational factor. The approximate purchase
cost of the SG-II as tested was $70,000.
41
-------
Table 6-8. Extent of Down Time and Service Time
Date
Down Time (a)
Service Time(b)
Activity
8/9/02
1 hour (hr)
1 hr
Recalibrated SG-II using vapor-injection method.
8/9/02
20 min
20 min
Recalibrated SG-II using vapor-injection method.
8/10/02
10 min
10 min
Recalibrated SG-II using vapor-injection method.
8/10/02
5 min
5 min
Recalibrated SG-II using vapor-injection method.
8/11/02
15 min
15 min
Recalibrated SG-II using vapor-injection method.
8/11/02
25 min
25 min
Recalibrated SG-II using vapor-injection method.
8/12/02
10 min
10 min
Recalibrated SG-II using vapor-injection method.
8/12/02
10 min
10 min
Recalibrated SG-II using vapor-injection method.
8/12/02
25 min
25 min
Changed collection time from 2 to 1 minute and
changed calibration factor (CF) from 1 to 2.
8/12/02
NA(c)
10 min
Emptied liquid waste containers into sump.
8/13/02
3 hr
10 min
Changed argon cylinder after finding system in
standby mode due to empty argon cylinder.
8/13/02
30 min
30 min
Liquid level in the total mercury channel impinger
appeared to be high, and liquid was bubbling
excessively over to chiller. Heated Teflon valve for
total mercury channel in speciation unit was
slightly closed to reduce bubbling in the impinger.
Mass flow controller (MFC) was inadvertently shut
down. Logged out and logged in to restart
operating system and reset mass flow controller to
25%. Afterward, liquid level still appeared to be
high.
8/14/02
45 min
1 hr
Performed daily checks.
Total mercury impinger liquid level still appeared
to be too high and the Hg° impinger level was too
low. Noticed that no waste was flowing into the
Hg° waste container. The mass flow meter dialog
box indicated "MD" rather than a setpoint of 25%.
Notified the vendor of symptoms.
42
-------
Table 6-8. Extent of Down Time and Service Time (continued)
Date
Down Time (a)
Service Time(b)
Activity
8/15/02
18 hr
2 hr
Found SG-II in standby mode due to Conditioner 1
alarm (moisture on HgT channel sensor). Dried the
sensor and restarted the system, but the liquid level
in the HgT channel impinger was too high. Stopped
the instrument until receiving further instructions
from vendor.
Reinitialized the mass flow controller and reset to
25% per instructions from the vendor. Discovered
that the peristaltic pump tubing for the Hg° reagent
line was reversed so that the direction of flow was
from the impinger to the clean reagent container.
Adjusted the flow through the HgT channel
impinger to match the flow through the Hg°
channel impinger. Reversed the tubing and
restarted the system at 12:08.
8/16/02
15 min
15 min
Changed argon cylinders and switched from UHP
argon to industrial grade argon. (Obtained vendor
approval via telephone on 8/15/02).
Recalibrated SG-II using vapor-injection method.
8/19/02
5 min
5 min
Changed argon cylinders.
8/20/02
17 hr 35 min
5 min
Found system in standby mode due to Conditioner
1 alarm. Restarted system. Buildup of yellow
precipitate in HgT impinger appeared to worsen.
8/21/02
10 min
25 min
Prepared seven liters of Hg° reagent and four liters
of HgT reagent.
8/22/02
30 min
30 min
Replaced HgT impinger with a clean impinger.
Replaced peristaltic pump tubing for Hg° and HgT
reagent lines. Emptied condensate container for
excess flue gas line. Instrument recalibrated using
vapor injection method.
8/23/02
5 hr
1 hr
Found system in standby mode with two alarms:
empty argon cylinder and Conditioner 1 alarm.
Liquid in the HgT Peltier cooler impinger had
reached the moisture sensor causing the
conditioner alarm. Changed argon cylinders.
Replaced peristaltic pump tubing for HgT and Hg°
waste lines. Restarted system.
43
-------
Table 6-8. Extent of Down Time and Service Time (continued)
Date Down Time(a) Service Time(b)
8/24/02 10 min 10 min
8/24/02 20 min 1 hr
8/26/02 10 min 10 min
8/26/02 10 min 10 min
8/26/02 NA 1 hr
8/29/02 1 hr 1 hr
8/29/02 30 min 30 min
8/31/02 10 hr 10 min 25 min
9/2/02 10 min 10 min
9/3/02 10 min 10 min
9/3/02 NA 1 hr 15 min
9/4/02 20 min 20 min
Activity
Discovered peristaltic tubing for HgT waste line
disconnected from waste container drain line.
Paused the SG-II, cleaned up spill on the floor,
reconnected the lines, and restarted the system at
8:30.
Recalibrated SG-II using vapor-injection method.
Prepared four liters of HgT reagent and three liters
of Hg° reagent.
Recalibrated SG-II again using vapor-injection
method.
Recalibrated SG-II using vapor-injection method.
Changed argon cylinders.
Prepared five liters of Hg° reagent and four liters
of HgT reagent.
Replaced HgT impinger and peristaltic pump lines.
Recalibrated SG-II using vapor-injection method.
Found system in standby mode due to Conditioner
1 alarm as a result of liquid in the HgT Peltier
cooler impinger. The peristaltic pump tubing for
the HgT waste line was pinched. Drained impinger
and dried moisture sensor. Moved to a new
position on the tubing and swapped tubing holders
on the peristaltic pump. Restarted the system.
Found SG-II in alarm. Argon cylinder was empty.
Changed argon cylinders.
Emptied condensate container for flue gas bypass
line.
Prepared seven liters of Hg° reagent and four liters
of HgT reagent.
Observed yellow precipitate in the HgT waste line
downstream of the Peltier cooler impinger and the
peristaltic pump.
Replaced the peristaltic pump tubing for the
reagent lines. Replaced the HgT impinger.
Recalibrated SG-II using vapor-injection method.
44
-------
Table 6-8. Extent of Down Time and Service Time (continued)
Date Down Time(a) Service Time(b)
9/5/02 35 min 35 min
9/6/02 NA 2 hr
9/7/02 15 min 15 min
9/8/02 34 hr 20 min 1 hr 35 min
9/9/02 NA 10 min
9/10/02 NA 10 min
9/10/02 20 min 20 min
9/11/02 NA 40 min
9/11/02 10 min 45 min
Activity
Found the drain elbow from the HgT Peltier cooler
impinger plugged with yellow precipitate.
Removed, cleaned, and reinstalled the elbow and
waste line.
Prepared 10 liters of HgT reagent and eight liters of
Hg° reagent.
Replaced the HgT impinger due to low instrument
response.
Found SG-II in standby mode due to Conditioner 1
alarm resulting from liquid in the HgT Peltier
cooler impinger. Elbow beneath impinger was
plugged with yellow precipitate. Removed elbow,
cleaned, and reinstalled. Discharge line that mates
to elbow had swollen and had difficulty making up
connection.
Replaced peristaltic pump tubing for HgT waste
line.
Changed argon cylinders.
Heard pressurized air leak from polyflow tubing
airline. Identified source of the leak and wrapped
in duct tape temporarily.
Permanently repaired polyflow tubing airline with
Swagelok connector.
Replaced elbow at base of HgT Peltier cooler
impinger with parts sent by vendor. Nipple on
impinger broke off while replacing the elbow, thus
replaced the impinger also. The original Teflon
union that connects the impinger to the moisture
sensor was loose, so replaced this union.
Adjusted excess flow on Channel 1 from 600 to
350 ml/min. Replaced peristaltic pump tubing for
both reagent lines and for Hg° waste line.
Noticed that liquid levels in the HgT and Hg°
impingers were low. Further noticed bubbling in
the reagent tanks. Moved reagent line tubing to
two different bridges and the bubbling stopped.
Bridges were either not seated properly or just
needed to be swapped out.
45
-------
Table 6-8. Extent of Down Time and Service Time (continued)
Date Down Time (a) Service Time (b) Activity
9/11/02
10 min
10 min
Recalibrated SG-II using vapor-injection method.
9/12/02
5 min
5 min
Changed argon cylinders.
9/13/02
NA
1 hr 15 min
Prepared six liters of Hg° reagent and six liters of
HgT reagent.
Observed small amount of liquid on Teflon
Swagelok fitting at base of HgT Peltier cooler
impinger. Slightly tightened connection and moved
bridge for HgT waste line to new position on
peristaltic pump.
9/13/02
35 min
35 min
Replaced the HgT impinger.
9/13/02
10 min
10 min
Recalibrated SG-II using vapor-injection method.
9/15/02
20 min
20 min
Recalibrated SG-II using vapor-injection method.
Liquid level in Hg° impinger appeared to be low.
Moved the bridge to a different position on the Hg°
reagent tubing. Emptied the liquid waste
containers.
9/15/02
5 min
1 hr
Changed argon cylinders.
Prepared one liter of Hg° reagent and one liter of
HgT reagent.
9/16/02
NA
NA
Vendor representative came to site to check SG-II
before starting OH reference method testing.
Routine maintenance and operation of SG-II
returned to vendor.
9/16/02
16 hr
NA
Found system in standby mode due to nearly
closed valve on argon cylinder.
9/16/02
50 min
50 min
Restarted argon flow to instrument. Changed HgT
reagent chemistry to overcome yellow precipitate
problem. Recalibrated SG-II using vapor-injection
method.
9/17/02
25 min
25 min
Recalibrated SG-II using vapor-injection method.
9/18/02
20 min
20 min
Recalibrated SG-II using vapor-injection method.
46
-------
Table 6-8. Extent of Down Time and Service Time (continued)
Date Down Time (a) Service Time (b)
9/18/02 20 min 20 min
9/18/02 1 hr 5 min lhr5min
9/19/02 15 min 15 min
TOTAL 7,100 min 1,730 min
(a)
(b)
(c)
(d)
Activity
During the second half of Run 22, noticed that
MFC had fallen to 20% of full-scale deflection
(FSD). To overcome this issue, decreased MFC to
12.5% FSD and set CF to 4.
At the end of Run 23, noticed that MFC had
fallen to 10% FSD. This could cause a slight
negative bias to the results. The unit was
inspected and the sample block valve was causing
a restriction. The block valve was bypassed to
resolve the problem.
Changed argon cylinders.
Recalibrated SG-II using vapor-injection method.
Observed that the system ran without problem
during the night. The MFC was set to 25% FSD
and the CF to 2.
Recalibrated SG-II using vapor-injection method.
88.3% availability and 28.8 service man-hours (d)
Down Time = time that the CEM was not operating, or was operating but not reporting reliable measurements. The
period over which down time was evaluated begins at the start of OH method testing on 8/8/02 and ends at the
conclusion of testing on 9/19/02. The amount of time was rounded to the nearest 5 minutes.
Service Time = time spent to perform daily checks, conduct routine operation and maintenance activities, and
troubleshoot problems. The period over which service time was evaluated begins at the start of OH method testing
on 8/8/02 and ends at the conclusion of testing on 9/19/02. The amount of time was rounded to the nearest 5
minutes.
NA = not applicable.
Availability = the ratio of time that the CEM was not experiencing down time to the total time available for
monitoring mercury emissions from the start of OH reference method testing on 8/8/02 to the end of testing on
9/19/02. The total time that was available for monitoring was 60,936 minutes or 1,015.6 hours.
47
-------
Chapter 7
Performance Summary
The RA of the SG-II for measuring Hgx and Hg° was verified by comparison with the results of
18 sampling runs using dual trains of the OH reference method at Hgx levels of <1 to
200 |ig/dscm. The overall RA results were 54.7% and 59.8% for Hg° and Hgx, respectively. The
RA for Hgx is 42.8% if OH Run 16 is excluded from the calculation.
Correlation of the SG-II Hg° and Hgx results with the OH results showed r2 values of 0.948 and
0.875, respectively, when all 18 OH results were included.
Precision of the SG-II was estimated using two OH sampling periods having relatively stable
introduction of mercury in aqueous waste into the TSCAI. The estimated maximum variability
attributable to the SG-II was 8.9% and 15.9% for these two periods.
The bias introduced by the SG-II sampling system was evaluated by introducing Hg° standard
gas both at the SG-II and at the inlet to the sampling system. Sampling system bias was 2.8% to
6.9% for Hg° and 0.0% to 6.9% for Hgx in the two measurement channels of the SG-II.
Zero gas and mercury gas standards were used to assess the calibration drift of the SG-II
throughout the verification test. Zero gas readings over the six-week field period averaged
-0.05 (±0.10) |ig/m3, indicating minimal drift of the SG-II zero readings. Eighteen analyses of an
approximately 14.9 |ig/m3 Hg° standard over six weeks resulted in an RSD of 11.9% (8.3% with
one outlier excluded). Twelve analyses of an approximately 35.6 |ig/m3 standard over five weeks
resulted in an RSD of 10.2% (5.9% with one outlier excluded). Four analyses of an approx-
imately 39.3 |ig/m3 Hg° standard over four days resulted in an RSD of 8.7%.
Rise and fall times of the SG-II response were determined at times of switching between zero
and mercury standard gases. The SG-II achieved 95% rise and fall times in approximately one
measurement cycle.
The SG-II data completeness was 88.3%. The most common maintenance needed was replace-
ment of chemical reagent solutions and argon cylinders, which was done every few days. The
most common operational problems were in the liquid flow system of the mercury speciation
module.
48
-------
Chapter 8
References
1. Test/QA Plan for Field Demonstration of Mercury Continuous Emission Monitors at the
TSCA Incinerator, Revision 3, Battelle, Columbus, Ohio, June 2002.
2. Standard Test Methodfor Elemental, Oxidized, Particle-Bound, and Total Mercury in
Flue Gas Generated From Coal-Fired Stationary Sources (Ontario Hydro Method),
American Society for Testing and Materials, Draft Method, September 3, 2001.
3. Dunn, J. E., Jr., Kelly, T. J., and Allen, M. W., "Field Demonstration of Mercury
Continuous Emission Monitors at the TSCA Incinerator," in Proceedings of the
International Conference on Air Quality III: Mercury, Trace Elements, and Particulate
Matter, Arlington, Virginia, September 1-12, 2002.
4. Kelly, T. J., Dunn, J. E., Jr., and Allen, M. W., "Evaluation of Mercury Continuous
Emission Monitors at the Oak Ridge TSCA Incinerator: ETV Phase II Verification Test,"
presented at the EPRICEM User's Group Meeting, San Diego, California, May 13-16,
2003.
5. Dunn, J. E., Jr., Kinder, K. K., Calcagno, J. A. m, Davis, W. T., Geisler, T. J.,
Allen, M. W., Kelly, T. J., and Priebe, S. J., "Evaluation of Mercury Continuous
Emission Monitors at the U.S. DOE TSCA Incinerator," paper #70249, presented at the
Air and Waste Management Association 96th National Conference and Exposition, San
Diego, California, June 22-26, 2003.
6. Method 101A - Determination of Particulate and Gaseous Mercury Emissions from
Sewage Sludge Incinerators, 40 Code of Federal Regulations 61, Appendix B, pp. 1,731-
1,754.
7. Dunn, J., Klamm, S., and Mandel, S., "Status of Mercury Calibration Gas for Mercury
CEMs," presented at the EPRI CEM User's Group Meeting, San Diego, California,
May 13-16, 2003.
8. Quality Management Plan (QMP) for the ETV Advanced Monitoring Systems Pilot,
Version 3.0, U.S. EPA Environmental Technology Verification Program, Battelle,
Columbus, Ohio, December 2001.
49
------- |