United States
Environmental Protection
Agency
Office of Air Quality
Planning and Standards
Research Triangle Park NC 27711
EMB Report 83-IBR-21
January 1984
Air
Industrial Boilers
SO2 Continuous
Monitoring
Emission Test Report
Argonne National
Laboratories
Argonne, Illinois
Volume 1: Summary
-------
CORPORATION
DCN 83-222-018-10-11
Industrial Boiler Continuous
Emission Monitoring
at the
Argonne National Laboratories
Argonne, Illinois
Revised Draft Final Report
Volume 1
Prepared for:
Ms. Nancy McLaughlin
and
Mr. Peter Westlin
Task Managers
Emission Measurement Branch
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
EPA Contract No. 68-02-3542
Work Assignment 10
ESED 83/24
Prepared by:
L. A. Rohlack
D. L. Lewis
C. S. Galloway
Radian Corporation
9 January 1984
8501 Mo-Pac Blvd. / P.O. Box 9948 / Austin, Texas 78766 / (512)454-4797
-------
CONTENTS
Section Page
1 INTRODUCTION 1
2 PRESENTATION AND DISCUSSION OF RESULTS 3
2.1 Presentation of Results 3
2.2 Discussion of Results 17
2.2.1 EPA Method 6B Precision and Reliability 17
2.2.2 Unit Availability—Effect on S02 Removal 17
2.2.3 S02 Removal Efficiency During "Trouble Free"
Operation 19
. 2.2.4 ANL Process Data Quality 20
3 DATA QUALITY. 25
3.1 Summary and Conclusion 26
3.2 QA/QC Program Objectives 28
3.3 Methods of Quarititating Data Quality 29
3.3.1 Definitions of Precision, Accuracy, and Bias 30
3.3.2 Assessment of Accuracy and Bias 32
3.3.3 Assessment of Precision 34
3.4 Quality Assurance Audits _.. . 37
3.4.1 Performance Audits 37
3.4.1.1 Method 6B Measurements 38
3.4.1.2 S02 Analyses 42
3.4.1.3 Dry Gas Meters 44
3.4.1.4 Balance 44
3.4.1.5 Proximate/Ultimate Fuel Analyses 44
3.4.2 Systems Audit 47
3.5 Quality Control Data 52
3.5.1 Control Sample Analyses 52
3.5.2 Duplicate Analyses 54
3.5.3 Duplicate Samples 54
3.6 Method 6B Reliability and Data Capture 58
iii
-------
CONTENTS (Continued)
Section Page
4 PROCESS DESCRIPTION . 62
4.1 Plant Configuration 62
4.2 Description of Sampling Points 67
5 SAMPLING AND ANALYSIS 71
5.1 Sampling 71
5.1.1 Flue Gas Sampling 71
5.1.2 Coal Sampling Collection 75
5.2 Sample Analysis 75
5.2.1 Flue Gas Analysis 75
5.2.2 Coal Analysis 76
6 PROCESS MONITORING PROCEDURES 78
7 EXAMPLE CALCULATIONS 80
iv
-------
FIGURES
Number Page
3-1 Systems Audit Checklist ............................... ....... 49
3-2 SOa Control Sample Data ...................................... 53
4-1 Schematic Diagram of the Argonne National Laboratory
Unit 5 Boiler Spray Dryer/Baghouse Flue Gas
Desulf urization System ..................................... 64
4-2 Schematic Diagram Illustrating the Location of the Inlet
and Outlet Flue Gas Sampling Ports at the Argonne
National Laboratory Unit 5 Spray Dryer/Baghouse Flue
Gas Desulf urization System ................................. 68
4-3 Diagram Illustrating the Relative Location of the Two
Sampling Probes Used in Collecting the Duplicate EPA
Method 6B Samples at the Inlet to the Argonne National
Laboratory Unit .'5 FGD System ............................... 69
4-4 Diagram Illustrating the Location of the Two Sampling
Probes Used in Collecting the Duplicate EPA Method 6B
Samples at the Outlet of the Argonne National Laboratory
Unit 5 FGD System .......................................... 70
5-1 EPA Method 6B SOa and COa Sampling Train ..................... 72
5-2 Method 6B Field Sampling Data Sheet .......................... 74
5-3 Method 6B Calculations Worksheet ............................. 77
-------
TABLES
Number page
2-1 Summary of EPA Method 6B Data Collected at ANL Unit 5
FGD System from August 1 through August 29, 1983 6
2-2 Summary of ANL Unit 5 Boiler and FGD 24-Hour Average Process
Data Collected from August 2 through August 29, 1983 12
2-3 Summary of Ultimate and Proximate Analyses Performed on
the Coal Used at the ANL Unit 5 Boiler During
August 1 through August 26, 1983 16
2-4 ANL Unit 5 Spray Dryer-Baghouse System Upset Summary
for the Period of August 1 through August 29, 1983 22
2-5 Summary of EPA Method 6B 24-Hour Average SOz Removal
and Emission Rate Data Collected at ANL from August 2
through August 26, 1983 During Periods of "Trouble
Free" Operation 23
2-6 Comparison of Process Data Collected Using the EPA Method 6B
and the Contraves/Goerz CEMS on the Stack at ANL from
August 2 through August 25, 1983 24
3-1 Summary of Estimated vs. Measured Data Quality 27
3-2 Measures of Precision. 31
3-3 Summary of Performance Audit Results 39
3-4 Method 6B Audit Results 41
3-5 S02 Analytical Audit Results % 43
3-6 Dry Gas Meter Audit Results 45
3-7 Balance Audit Results 46
3-8 Performance Audit Results for Coal Analyses 48
3-9 Summary of Significance Test Data for Paired Results 57
3-10 Summary of Analysis of Variance Results 59
3-11 Coefficients of Variation for Repeat and Duplicate
Measurements 60
4-1 State of Illinois Emission Limits for ANL Boiler No. 5 62
4-2 Spray Dryer Parameters 66
4-3 Fabric Filter Parameters 66
6-1 Summary of ANL Unit 5 Boiler and FGD Process Parameters
of Interest (Including DART Acronyms When Applicable) 79
vi
-------
SECTION 1
INTRODUCTION
The United States Environmental Protection Agency (EPA) is currently
engaged in developing an industrial boiler, sulfur reduction standard. In
order to support the New Source Performance Standard for industrial boilers,
sufficient performance and financial background data must first be developed
at a representative facility currently using best available control technol-
ogy (BACT) for sulfur dioxide control. To help develop the necessary support
data, Radian Corporation was contracted by the Emission Measurement Branch
(EMB) of EPA to conduct: a 90-day S02 and C02 monitoring program around the
flue gas desulfurization (FGD) system located at the Argonne National
Laboratory (ANL) near Chicago, Illinois.
The ANL-FGD system was selected for use during this program for two
reasons. First, ANL utilizes a spray dryer/fabric filter system. This
control technology is presently considered among the best available for
SOa control. Second, under an existing EPA-funded program (EPA Contract
No. 89-F2-A041), ANL was contracted to thoroughly characterize the FGD
system using high sulfur coal (^3.5%). The objective of the EPA/ANL pro-
gram was to evaluate the system's effectiveness under different operating
conditions. Data obtained during both the EPA/ANL program and EPA/Radian
program could be used to complement each other.
The original scope of work for this test program called for using the
existing Contraves-Goerz continuous gas monitors at ANL, if possible, for
collecting the desired FGD inlet and outlet SOa and COa emission data.
During initial phases of the program, an evaluation was performed by Radian
to determine if the existing Contraves-Goerz monitors could be used to
collect these data. Based on the results of the evaluation, a decision was
-------
made to not rely upon the Contraves-Goerz monitors. Instead, the EPA
Method 6B sampling and analysis procedure (described in Section 4) was used
throughout the program to collect the necessary S02 emission rate data.
S02 emission monitoring tests and process data collection were
initiated by Radian on August 1, 1983 at the ANL Unit 5 boiler spray dryer/
baghouse FGD system. Duplicate EPA Method 6B samples were simultaneously
collected each day at the FGD inlet and outlet. Pertinent boiler and FGD
operational data were also collected daily using a Radian DART data acqui-
sition system and operator process log data sheets. The test program was
halted on August 29, 1983 because the FGD system was not operating at its
maximum performance level of SOz removal.
The primary purpose of this report is to present the results of the
monitoring program conducted at ANL, by Radian, from August 1 through August
29, 1983. Section 2 includes a summary and discussion of the S02 emission
rate data and process data collected during this program. The quality
assurance/quality control (QA/QC) results are presented in Section 3. An
evaluation of the EPA Method 6B precision and reliability, as it pertains to
this program, is included with the QA/QC results. A brief description of
the process configuration and location of the sampling points is presented
in Section 4. A description of the sampling and analysis procedures are
presented in Section 5. Section 6 includes a description of the process
monitoring procedures and Section 7 includes example calculations.
A copy of the actual EPA Method 6B sampling data sheets, the operator
logsheets, and DART process data sheets are included as appendices in a
separate data volume (Volume II). A copy of the EPA Method 6, 6A, 6B, and
Subpart D of the Federal Register are included in the appendices.
-------
Section 2
PRESENTATION AND DISCUSSION OF RESULTS
This section presents a summary of the EPA Method 6B test results, the
24-hour average process data, and the coal ultimate analysis and proximate
analysis results. A discussion of the EPA Method 6B, SOa emission rate data
and 24-hour average proc«>ss data collected during this program is also pre-
sented in this section.
2.1 PRESENTATION OF RESULTS
Table 2-1 contains a summary of results of the EPA Method' 6B tests.
For convenience, the S02 concentration was reported in parts per million
(ppm) and the average SOJ; emission rate was reported in both nanograms per
Joule (ng/J) and pounds per million Btu (lbs/106 Btu). Table 2-1 includes
a comment section to help explain certain discrepancies in the data and
identify process upsets. EPA Method 6B, SOa emission rate tests were not
conducted on August 26, 2.7, 28, and 29 because the FGD system was not operat-
ing because of an electrical short, and the sampling system was undergoing a
quality assurance audit. All of the supporting EPA Method 6B data sheets
are included in Appendix A of Volume II.
Table 2-2 contains a summary of Unit 5 boiler and FGD system 24-hour
average process data. The 24-hour average process data presented in Table
2-2 originated from one of three sources. These include the Radian DART
data sheets and the "Dry Panel" and "Wet Panel" operator log process data
sheets. The "Dry Panel" refers to the main instrument panel located near
the boiler control panel. The "Wet Panel" refers to the instrument panel
located in the slaking house. All three sources were used because the DART
was not in operation during the entire reporting period and because certain
-------
process data from both instrument panels could not be monitored by the DART.
The ANL Unit 5 boiler and FGD Operator Logsheets and the DART 60-minute and
24-hour average process data printouts are included in Appendices B and C,
respectively, of Volume II.
To help identify where the process data in Table 2-2 originated, numbers
in parentheses are used to denote "Dry Panel" observations and numbers in
brackets are used to denote "Wet Panel" observations. The remaining process
data were collected using the DART. Blank spaces in Table 6-2 indicate that
the instrument monitoring the parameter of interest was not operable or that
a portion of the boiler or FGD system was not operating.
Table 2-3 contains a summary of the ultimate and proximate coal analyses
performed on the two coals used by the ANL Unit 5 boiler from August 1 through
August 26, 1983. The results are reported on both a wet and dry basis. The
F -factor (dry basis) calculated from the analysis of each coal is also in-
cluded in Table 2-3.
-------
TABLE 2-1. SUMMARY OF EPA METHOD 6B DATA COLLECTED AT ANL UNIT 5
FGD SYSTEM FROM AUGUST 1 THROUGH AUGUST 29, 1983
Datea
(MMDD)
0801
0801
0802
0802
OB03
0803
0804
0804
0805
0805
0806
0806
0807
0807
0808
0808
0809
0809
Sampling
Location
Inlet
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Sample
S02
(ppm)
724
1550
135
1610
208
593*
116
1510
189
1640
289
1590
266
1130
487
1500
115*
Train A
C02
OS)
6.0
9.7
8.9
9.1
8.7
6.5*
9.8
9.7
10.0
9.6
9.8
9.8
9.6
9.3
9.6
9.2
7.9*
Results
Emission
Rate
(ng/J)
1530
2030
192
2240
303
1160*
150
1970
240
2170
374
2060
351
1540
643
2070
185*
Sample
S02
(ppm)
1310
1550
224
1660
1540
1590
235
1600
284
1360
276
1130
543
1490
796
Train
C02
9.4
11.4
9.5
9.0
___*
10.3
10.1
10.2
10.1
10.0
9.7
8.4
10.0
9.5
9.7
9.8
9.3
B Results
Emission
Rate
(ng/J)
1770
1720
299
2250
1900
1980
295
2030
371
2050
350
1510
710
1930
1090
Average
S02
Removal
Efficiency
86.9
86.5
92.1
86.5
82.2
83.0
55.6
45.7
"The duplicate EPA Method 6B sampling trains (.Train A and Train B) operated from 1000 hours
on the day indicated to 0900 hours on the next day to constitute a 24-hour sample.
-------
TABLE 2-1 (Continued)
Average S02
Emission Rate
ng/J lbs/106 BTU
Comments
Sampling Problems
Boiler/FGD Upsets
1650 3.83 Outlet sample not collected-
preliminary check-out of
. sampling system
1880
245
4.35
.570
2240 5.21 *0utlet B train impingers
improperly recovered. Data
303 .704 not used in average.
1900
150
1980
267
4.42
.348
4.59
.622
*Lealc indicated across Inlet
A train. Outlet B train
impingers improperly recovered.
Data not used in average.
Outlet Contraves inoperable-
poor system control.
2100
373
4.87
.866
Outlet Contraves inoperable-
poor system control.
2060
350
4.79
.814
Outlet Contraves inoperable
poor system control.
1520
681
3.54
1.57
Outlet Contraves inoperable—
poor system control.
2000 4.64 *High pressure drop across
drirlte impinger. Data in-
1090 2.52 dicate leak in system. Data
not used in average.
High absorber pressure—flue gas
bypasses FGD system—no scrubbing.
-------
TABLE 2-1 (Continued)
Date
(MMDD)
0810
0810
0811
0811
0812
0812
0813
0813
0814
0814
0815
0815
0816
0816
0817
0817
0818
0818
Sampling
Location
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Sample
S02
(ppm)
1530
736
1440
579*
1320
366
1400
182
1400
222
1470
242
1470
130
1660
165
1610
236
Train A
C02
(%)
9.3
9.6
9.6
7.9*
8.6
9.2
8.9
9.0
8.6
9.1
9.1
9.0
9.3
9.8
10.4
10.2
10.0
10.3
Results
Emission
Rate
(ng/J)
2090
972
1900
931*
1950
505
2000
257
2060
308
2050
341
2000
168
2020
205
2040
291
Sample
S02
(ppm)
1590
669
1490
1340
1360
412
1330
133
1430
249
1430
229
1530
120
1600
210
1660
254
Train
C02
(%)
10.0
9.6
10.0
9.8
8.8
9.2
8.5
9.1
8.9
8.7
8.9
9.0
9.6
9.5
9.9
10.0
10.2
10.1
B Results
Emission
Rate
(ng/J)
2020
884
1890
1730
1960
568
1980
185
2040
363
2040
•
323
2020
160
2050
266
2060
319
Average
S02
Removal
Efficiency
54.7
8.2
72.6
88.9
83.6
83.8
91.8
88.4
85.2
aThe duplicate EPA Method 6B sampling trains (Train A and Train B) operated from 1000 hours
on the day indicated to 0900 hours on the next day to constitute a 24-hour sample.
-------
TABLE 2-1 (Continued)
Average S02
Emission Race
ng/J lbs/106 BTO
2060 4.77
928 2.15
CommenCs
Sampling Problems Boiler/FGD Upsets
High absorber pressure — flue
gas bypasses FGD system — no
scrubbing
1900
1730
1960
536
4.41
4.01
4.55
1.25
^Obstruction in sampling system
resulted in low sample volume
and possible leak. Data not
used in average.
High absorber pressure—flue
gas bypasses FGD system—no
scrubbing.
High absorber pressure—flue
gas bypasses FGD system—no
scrubbing.
1990
221
4.62
.51
2050
334
4.76
.779
2040
332
4.73
.771
2010
164
4.66
.380
2040
236
4.73
.548
2050
305
4.76
.708
(Continued)
-------
TABLE 2-1 (Continued)
Sample Train A Results
Date3
(MMDD)
OH19
0819
0820
0820
0821
0821
0822
0822
0823
0823
0824
0824
0825
0825
0826
0827
0828
0829
Sampling
Location
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
S02
(ppm)
1710
215
1580
169
1670
204
1580
195
1610
249
1630
207
1690
588
C02
«)
10
10
9
9
10
10
9
10
9
9
9
10
10
10
.4
.4
.7
.8
.2
.1
.9
.1
.9
.9
.9
.1
.4
.2
Sample Train B Results
1 •• Aveidgu
Emission Emission SO 2
Rate SO 2 CO 2 Rate Removal
(ng/J) (ppm) (%) (ng/J) Efficiency
2090
262
2070
219
2080
256
2020
245
2060
319
2090
260
2060
731
1600
228
1600
179
1590
209
1610
205
1530
252
1620
201
1610
585
9.9
10.2
10.0
9.9
9.9
10.0
9.9
9.8
9.6
9.7
9.8
10.1
10.0
10.2
2050
86.9
283
2030
89.1
229
2040
87.3
265
2060
87.5
265
2020
84.2
329
2100
87.7
252
2040
64.4
727
EPA Method 6B samples are not collected because an electrical short circuit caused a damper
to become maligned resulting in flue gas bypassing the FGD system.
On-site
On-site
On-site
Method 6B
Method 6B
Method 6B
audit
audit
audit
sample
sample
sample
no. 1
no. 2
no. 3
collected.
collected.
collected.
Damper still
Damper still
Damper still
misaligned.
misaligned .
misaligned.
aThe duplicate EPA Method 6B sampling trains (.Train A and Train B) operated from 1000 hours
on the day indicated to 0900 hours on the next day to constitute a 24-hour sample.
10
-------
TABLE 2-1 (Continued)
Average SO2
Emission Rate
ng/J lbs/106 BTU
Comments
Sampling Problems
Boiler/FGD Upsets
2070
273
4.80
.633
2050
224
4.77
.521
2060
261
4.79
.607
2040
255
4.75
.594
2040
324
4.73
.752
2100
256
4.86
.594
2050
729
4.76
1.69
Low slurry feed to spray
dryer—reduced 502 removal.
11
-------
TABLE 2-2. SUMMARY OF ANL UNIT 5 BOILER AND FGD 24-HOUR
FROM AUGUST 2 THROUGH AUGUST 29, 1983
AVERAGE PROCESS DATA COLLECTED
Parameter Units
Boiler Load 10* Ibs/hr
Boiler Exit Gas 0; Concen-
tration *
Baghouse Outlet Temperature *F
Spray Dryer lnl»-l Tenporature *F
Spray Dryer Outlet Temperature *F
Spray Dryer Slurry Feed gpm
Baghouse AP Inches HjO
Lime Milk Flow gpm
Central Gas Disperser AP Inches HZO
Baghouae Outlet Deupolnt *F
Atomizer Motor Amps asps
Line Milk Tank Level X
Slurry Density grans/cc
Slaking Dilution Hater Flow gpm
Line Milk Flow gpm
Line Milk Density graBS/cc
Slurry Dilution Water Flow gpm
Recycle Feed Bate Ibs/hr
Contraves Outlet COj X
Contraves Outlet SOi ppm
SO] Ealaaion Rate lbs/10* BTU
Slurry Mix Tank Levnl I
Date: 0802
24
Hour
Average
107
7.9
158
324
155
10.0
-.09
3.3
6.0
0.2
126
50
76
1.12
4.2
6.3
1.28
—
1.26
76
Number
of
Obser-
vation
(21)
(21)
(21)
(21)
(21)
(21)
(21)
(21)
(21)
(13)
(20)
(21)
(21)
(21)
[12]
[12]
[12]
—
(21)
(21)
Date t 0803
24
Hour
Average
113
7.6
158
325
155
10.6
-.08
3.9
7.6
0.5
128
52
78
1.12
5.1
7.2
1.29
—
1.25
76
Number
of
Obser-
vation
(21)
(21)
(21)
(21)
(21)
(21)
(21)
(21)
(21)
(21)
(20)
(21)
(21)
(21)
[12]
[121
[12]
—
(20)
(21)
Date: 0804
24
Hour
Average
112
7.5
159
325
155
10.4
-.09
3.6
6.6
J1.5
126
52
75
1.13
4.5
7.3
1.28
—
1.24
75
Dumber
of
Obser-
vation
(21)
(21)
(21)
(21)
(21)
(21)
(21)
(21)
'(21)
(21)
(20)
(21)
(21)
(21)
[11]
fill
[11]
—
(20)
(21)
Date: 0805
24
Hour
Average
110
7.7
160
324
155
10.5
-.08
3.8
6.7
0.4
128
54
74
1.13
4.6
7.5
1.27
—
1.10
75
Nuaber
of
Obser-
vation
(22)
(22)
(22)
(22)
(22)
(22)
(22)
(22)
(22)
(22)
(22)
(22)
(22)
(22)
[10]
[101
[10]
—
5
(22)
Date: 0806
24
Hour
Average
100
7.8
164
313
155
9.2
-.09
3.6
7.0
0.2
138
50
72
1.10
4.5
6.4
1.28
2.8
5
—
73
Number
of
Obser-
vation
24
(23)
24
24
24
24
(23)
(23)
24
(23)
(23)
(23)
m)
24
[12]
f!21
[12]
24
23
—
24
Date: 0807
24
Hour
Average
103
7.5
165
316
156
9.4
-.08
3.6
6.4
0.3
133
52
7S
1.11
4.3
7.1
1.29
2.9
6
—
73
Number
of
Obser-
vation
24
(23)
24
24
24
24
(23)
(23)
24
(16)
(23)
(23)
(21)
24
[11]
fill
.111]
24
23
~
24
Date: 0808
24
Hour
Average
100
6.9
167
309
159
8.1
-.09
3.0
6.1
0.1
130
50
7S
1.12
3.9
6.1
1.29
2.9
—
5.4
506
1.40;
75
Number
of
Obser-
vation
20
(20)
20
20
20
20
(20)
20
20
(16)
(20)
(20)
(70)
20
[12]
f!21
[12]
20
—
20
20
20
20
a The 24-hour average represents the average of
following day.
( ) dry panel log
[ ] wet panel log
— Data not available because the monitor was
that time period.
data collected from 1000 hours on the day Indicated to 1000 hours the
not functioning properly or the process was not operating during
(Continued)
-------
TABLE 2-2 (Continued)
Faruwlcr Unit*
Boiler Load 10 ' Iba/hr
Boiler Exit Gas O, Concen-
tration x
Baghouse Outlet Tesperature *F
Spray Dryer lnl«-l Tcmpuruturu *F
Spray Dryer Out let Tcnpi'riitm'i' *F
Spray Dryer Slurry Feed gpa
Baghouse Of Inche* UiO
Ll«e Milk Flow gp.
Central Gaa Dlsperser oP Inches UiO
Baghouse Outlet Dewpolnt *F
Atomizer Motor Amps aatps
Lime Milk Tank Level X
Slurry Density grau/cc
Slaking Dilution Water Flow gpm
Line Milk Flow gp.
Lime Milk Density graos/cc
Slurry Dilution Water Flow gpm
Cootravea Outlet CO2 I
Contravea Outlet SOi ppn
SO: Enlssloo Rate lbs/10* BID
Slurry Mix Tank Level I
Date:
24
Hour
Average
95
7.8
247
300
227
2.1
-.03
2.9
1.9
0.2
130
52
77
1.28
4.2
6.9
1.29
—
8.2
903
1.749
85
0809
Number
of
Obser-
vation
24
(9)
24
24
24
24
(9)
24
24
(9)
(9)
(9)
(9)
24
[5]
[5]
[5]
—
23
23
23
24
Datel
24
Hour
Average
—
~
~
—
—
—
—
—
-_
—
—
—
—
—
—
—
~
—
—
—
--
0810
Humbel
of
Obser-
vation
—
—
—
—
— •
—
—
—
—
~
—
—
—
—
—
—
—
—
--
—
—
Date:
24
Hour
Average
92
—
203
313
180
—
~
—
—
—
—
71
—
—
—
—
—
8.6
1234
2.016
86
0811
Number
of
Obser-
vation
24
24
24
24
—
—
' —
—
—
—
24 .
—
—
—
—
--
23
23
23
24
Date:
24
Hour
Average
90
8.7
160
310
154
6.7
-.09
2.2
1.8
.01
123
49
68
1.14
4.4
1.8
1.25
1.3
8.1
463
1.323
73
0812
Number
of
Obser-
vation
24
(18)
24
24
24
24
24
24
24
(14)
(18)
(18)
24
24
[9]
[9]
[9]
24
23
23
23
24
Date :
24
Hour
Average
90
7.9
163
311
156
9.0
-.10
3.7
4.0
0.2
127
52
68
1.14
3.0
4.0
1.27
3.0
7.9
310
1.113
68
0813
Number
of
Obser-
vation
24
(21)
24
24
24
24
24
24
24
(21)
(21)
(21)
24
24
[12]
[12]
[12]
24
23
23
23
24
Date:
24
Hour
Average
95
9.1
163
312
156
9.1
-.10
3.7
6.0
0.3
127
50
70
1.14
3.0
5.7
1.27
3.0
8.3
349
1.205
70
0814
Number
of
Obser-
vation
24
(22)
24
24
24
24
24
24
24
(22)
(22)
(22)
24
24
[11]
[11]
[11]
24
23
23
23
24
Date:
24
Hour
Average
98
8.7
163
313
156
9.4
-.10
4.4
4.8
0.4
128
51
69
1.16
2.5
5.5
1.28
2.5
8.5
352
1.212
69
0815
NuBber
of
Obser-
vation
24
(22)
24
24
24
24
24
24
24
(22)
(22)
(22)
24
24
[12]
[12]
[12]
24
23
23
23
24
( ) dry panel log
[ ] wet panel log
— Data not available
that time period.
I
I
z
because the monitor was not functioning properly or the process wns no: operating during
(Continued)
-------
TABLE 2-2 (Continued)
Parameter Units
Boiler Load 10* Ibs/hr
Boiler Eilt Gas Oj Concen-
tration X
Baghouse Outlet Temperature *F
Spray Dryer luli-l 'IVBperaturt: *F
Spray Dryer Oul let Tunperatiui* *F
Spray Dryer Slurry Feed gpm
Spray Dryer Inlet Pressure laches UaO
Baghouae tf inches HiO
Line Hi Ik Flow gpm
Central Gas Disperser AP Inches HiO
Baghouae Outlet Dewpolnt *F
Atonlzer Motor Amps Mips
Lime Milk Tank Level X
Slurry Density grsas/cc
Slaking Dilution Water Flow gp.
Line HI Ik Flow gpm
Lime Milk Density grau/cc
Slurry Dilution Water Flow gpm
Becycle Feed Bate Iba/hr
Contravea Outlet COi X
Contraves Outlet SO? ppm
SO, Emission Rate lbs/10' BTU
Slurry Mix Tank Level X
Date: 0816
24
Hour
104
8.7
164
310
157
9.9
-1.0
4.6
7.0
0.45
129
52
67
1.19
4.4
5.1
1.27
1.6
8.9
319
1.017
68
Number
of
Obaer-
24
(22)
24
24
24
24
24
24
24
(22)
(22)
(22)
24
24
[12]
[12]
[12]
24
23
23
23
24
Date: 0817
24
Hour
109
8.1
160
321
156
10.3
-.09
4.8
7.7
0.5
130
53
70
1.21
4.9
7.8
1.29
3.6
9.5
319
1.02
72
Muster
of
Obaer-
(22)
(22)
(22)
(22)
(22)
(22)
(22)
(22)
(22)
(22)
(22)
(22)
(22)
(22)
[11]
fill
(111
7
6
6
(22)
(22)
Date: 0818
24
Hour
115
6.6
151
313
157
11.1
-0.9
3.7
8.5
0.6
129
57
69
1.15
5.0
8.6
1.14
3.6
9.5
400
1.201
71
Dumber
of
Obaer-
21
(23)
21
21
21
21
21
21
' 21
(23)
(23)
(23)
21
21
[10]
noi
[10]
21
21
21
21
21
Date: 0819
24
Hour
121
6.8
147
326
155
12.6
-.09
4.0
8.7
0.7
132
57
73
1.17
4.8
7.7
1.15
3.5
8.2
297
1.09
72
Number
of
Obser-
(23)
(23)
(23)
(23)
(23)
(23)
(23)
(23)
(23)
(23)
(23)
(23)
(23)
(23)
[12]
[12]
[12]
1
1
1
(23)
(23)
Date: 0820
24
Hour
106
7.7
146
322
155
11.4
-.10
3.8
7.5
0.3
129
52
73
1.18
4.2
7.8
1.15
3.8
8.8
375
1.10
72
Number
of
Obser-
(23)
(23)
(23)
(23)
(23)
(23)
(23)
(23)
(23)
(23)
(23)
(23)
(23)
(23)
[12]
[121
[12]
19
19
19
(23)
(23)
Date: 0821
24
Hour
115
7.4
146
322
155
11.1
-.10
3.8
7.7
0.5
131
55
74
1.19
3.9
6.6
1.16
4.8
9.6
372
1.12
71
Number
of
Obser-
(23)
(23)
(23)
(23)
(23)
(23)
(23)
(23)
(23)
(23)
(23)
(23)
(23)
(23)
[12]
[12]
[12]
4
3
3
(23)
(23)
Date: 0822
24
Hour
106
7.1
147
322
155
10.0
-.10
3.7
8.2
0.3
128
52
74
1.19
3.9
7.4
1.14
3.5
9.3
374
1.11
72
Number
of
Obser-
(23)
(23)
(23)
(23)
(23)
(23)
(23)
(23)
(23)
(23)
(23)
(23)
(23)
(16)
[12]
[12]
[12]
21
21
21
(23)
(23)
( ) dry panel log
[ ] wet panel log
— Data not available
that time period.
i
because the monitor was not functioning properly or the process was not operating during
(Continued)
-------
TABLE 2-2 (Continued)
Parameter Units
Boiler Load • 10 ' Ibs/hr
Boiler Exit Gas Oi Concen-
tration X
Baghouse Outlet Temperature 'I
Spray Dryer lnli-1 IV»|>orutui..- 't
Spray Dryer Out let Tempera tun- *?
Spray Dryer Slurry Feed gpa
Spray Dryer Inlet Pressure inches HiO
Baghouse AP Inches H,0
Lime Milk Flow gp.
Central Caa Dlsperser AP Indies UiO
Baghouae Outlet Dewpotnt *F
Atomizer Motor A0pa saps
Lime Milk Tank Level I
Slurry Density grama/cc
Slaking Dilution Water Flow gpm
Line Milk Flow gp.
Lime Milk Density grau/cc
Slurry Dilution Water Flow gpm
Becycle Feed Bate Ibs/hr
Contraves Outlet COj X
Contraves Outlet SOj ppm
SO] Emission Rate lbs/10* BTU
Slurry Mix Tank Uvel X
Date: 0823
24
llour
99
7.1
148
309
156
9.4
-.10
3.0
8.2
0.3
128
52
73
1.15
3.9
7.4
1.14
3.0
—
9.1
398
1.253
70
lumber
of
Obaer-
24
(23)
24
24
24
24
24
24
(23)
(23)
(23)
(23)
24
24
[12]
[12]
[12]
24
—
23
23
23
24
Date; 0824
24
Hour
108
7.5
150
311
156
10.3
-.10
3.2
8.3
0.2
130
51
73
1.14
4.1
6.5
1.13
3.4
—
9.5
370
1.103
70
Number
of
Obser-
24
(23)
24
24
24
24
24
24
(23)
(23)
(23)
(23)
24
24
[11]
[11]
[11]
24
—
23
23
23
24
Datet 0825
24
Hour
113
7.7
182
312
183
7.2
-.09
3.0
8.8
0.5
130
53
67
1.13
4.4
7.4
1.14
3.0
—
9.2
588
1.329
73
Number
of
Obser-
24
(23)
24
24
24
24
24
24
'(23)
(23)
(23)
(23)
24
24
[12]
[12]
[12]
24
—
23
23
23
24
Datei 0826*
24
Hour
i_—
Number
of
Obser-
Datei 0827*
24
Hour
Number
of
Obser-
Date: 0828*
24
Hour
Number
of
Obser-
Date: 0829*
2*
llour
Number
of
Obser-
( ) dry panel log
[ ] vet panel log
— Data not available
that time period.
*FGD system not operating/engineering process data sheets not available.
because the monitor was not functioning properly or the process was not operating during
-------
TABLE 2-3. SUMMARY OF ULTIMATE AND PROXIMATE ANALYSES PERFORMED ON THE COAL USED AT THE ANL UNIT 5
BOILER DURING AUGUST 1 THROUGH AUGUST 26, 1983.
8-1-83 to 8-23-83
Kentucky
Type of Analysis Parameter
Proximate % H^O
% Ash
% Volatile
% Fixed Carbon
Btu/Pound
% Sulfur
Ultimate % H20
% Carbon
% Hydrogen
% Nitrogen
% Chlorine
% Sulfur
% Ash
% Oxygen
8-24-83 to 8-26-83
Illinois
Concentration
(Wet Basis)
7.91
7.34
37.5
47.2
12,320
3.37
7.91
68.1
4.81
1.41
0.05
3.37
7.34
6.99
(Dry Basis)
7.96
40.7
51.3
13,380
3.66
74.0
5.22
1.53
0.05
3.66
7.97
7.60
(Wet Basis)
1.43
10.33
34.93
53.31
13,113
3.36
1.43
73.05
4.76
1.45
0.25
3.36
10.33
5.37
(Dry Basis)
10.48
35.44
54.08
13,303
3.41
74.11
4.83
1.47
0.25
3.41
10.48
5.45
F-factor (dry
basis)
1775 SCF/106 Btu
4.768 x 10~8 SCM/J
1788 SCF/106 Btu
4.804 x 10~8 SCM/J
-------
2.2 DISCUSSION OF RESULTS
This section includes a brief discussion of the results presented in
Section 2.1. The overall precision and reliability of the EPA Method 6B
sampling and analytical procedure, used during this program, are discussed
briefly in Section 2.2.1. The effect of boiler and FGD system upsets on
S02 emission rates are presented in Section 2.2.2. The efficiency of the
ANL Unit 5 FGD system during periods of "Trouble Free" operation are included
in Section 2.2.3. Finally, the quality of the ANL Unit 5 boiler and FGD
process data is briefly discussed in Section 2.2.4.
2.2.1 EPA Method 6B Precision and Reliability
During this program, the EPA Method 6B sampling and analytical procedure
proved to be a precise and reliable means of collecting SOa emission rate
data. The measured precision (replicability), based on observed variability
in results for duplicate samples collected using collocated sampling trains
was 6.2% overall during this program.. Of the total of 96 sampling train
days (i.e., four trains per day for 24 days, excluding the trial run at the
inlet on the first day), SO2 emission rate data were lost or invalidated in
only five instances, resulting in a sampling/analytical reliability of 94.8%.
A more detailed assessment of the precision and reliability of the EPA
Method 6B sampling and analytical procedure, used during this program, is
provided in Section 3.
2.2.2 Unit Availability—Effect on S02 Removal
From August 2a through August 29, 1983 the ANL Unit 5 boiler did not
experience any major upsets in operation. During this same time period, the
ANL Unit 5 FGD system experienced system upsets, of one form or another, on
13 of 28 days (46 percent). Table 2-4 presents a summary of the ANL Unit 5
FGD system upsets that occurred during this program. The upsets that had
a
Data collected on August: 1 is not included in the data assessment because
no outlet S02 emission rate data were collected on this day.
17
-------
an effect on the performance of the FGD system can be classified in four
categories. These include:
monitor failure,
• High central gas disperser pressure drop,
• Low slurry feed rate to the spray dryer, and
• Electric short-circuits within the control panel.
On August 5, 1983 (Friday night), the outlet Contraves-Goerz SC>2 monitor
suffered a power failure. The ANL instrument technician was off-duty and
could not be contacted. Therefore, the outlet Contraves monitor was inoper-
able from late Friday afternoon until Monday morning (August 8, 1983).
During this time period, the scrubber was placed in the manual mode and the
scrubber operator (s) controlled the scrubber by using a chart that relates
the required milk of lime flow needed to achieve sufficient S02 removal at
a given boiler load. This approach to scrubber control assumes that the
sulfur content of the coal and the reactivity of the lime do not change
appreciably. The average SOa removal efficiency for August 5, 6, and 7,
during manual operation of the scrubber, was approximately 84%. This com-
pares -to an average SOa removal efficiency of approximately 88% during
August 2, 3, and 4 when the scrubber system operated in the automatic mode.
Although the SOa removal efficiency decreased slightly (from 88 to 84%)
during manual operation of the scrubber, the average scrubber SOa emission
rate ("330 ng/J). did not rise above the maximum allowable Illinois state S02
emission limit (520 ng/J) during this period.
On Tuesday morning (August 9, 1983), a high pressure drop was detected
across the FGD system and the operators bypassed the FGD system for a period
of time. The high FGD pressure drop remained a problem until the morning of
August 12, 1983 when an 18-inch plug of solids was removed from the central
gas disperser. The maximum allowable SOa emission rate (520 ng/J) was ex-
ceeded on each of the four days that pluggage was a problem. During this
period, the 24-hour average SOa emission rate ranged from 536 ng/J to
18
-------
1730 ng/J. The exact cause for pluggage developing in the central gas
disperser is not known.
On Thursday, August 25, 1983 low slurry feed rate CV7.2 gpm) to the
spray dryer (compared to a normal 9-12 gpm rate) resulted in an SOa emission
rate CW29 ng/J) higher than the maximum (520 ng/J) allowed by the State of
Illinois. The exact ca.use for the low slurry feed rate to the spray dryer is
not known.
On Friday morning, August 26, 1983, an electrical short developed in
the "dry" control panel resulting in a damper becoming misaligned. The
FGD system was bypassed resulting in almost no SOa control during this
period. The electrical short was not corrected until after August 29, 1983.
2.2.3 S02 Removal Efficiency During "Trouble Free" Operation
From August 2 through August 29, 1983 the ANL Unit 5 FGD system operated
in a "trouble free" mode during 18 of the 28 days (64 percent). The term
"trouble free" is used here to denote periods of time when the FGD system
did not encounter documented process upsets.
Table 2-5 presents a summary of the EPA Method 6B 24-hour average S02
removal and emission rate data collected during periods of "trouble free"
operation. As shown in the table, the average SOa removal efficiency
ranged from 82.2% to 92.1%, with an overall average of 86.8%. The average
SOa emission rate ranged from 150 ng/J (0.348 lbs/106 Btu) to 373 ng/J
(0.866 lbs/106 Btu), with an overall average of 271 ng/J (0.629 lbs/106 Btu).
Also included in Table 2-5 is the corresponding 24-hour average spray
dryer approach to saturation temperature (ATAC) for each period of "trouble
A.O
free" operation. The AT is calculated based on the difference between the
Ao
spray dryer outlet process gas temperature and the baghouse outlet dew point
temperature. Except for August 6, 1983, AT ranged from 23 to 29°F with an
Ao
19
-------
average of 27°F during "trouble free" operation. On August 6, 1983, the
average AT g was 17 °F and the emission rate (0.866 lbs/106 Btu) was the
highest for any of the "trouble free" days. The spray dryer is designed
to maintain a ATAg of 22°F at the spray dryer outlet. It is not known if
the difference between the design ATAC value of 22°F and the average value
A.O
of 27°F is due to inaccurate gas temperature and/or gas dew point measure-
ments or if the system was not operating at the designated AT conditions.
A.O
2.2.4 ANL Process Data Quality
The quality (precision) of the EPA Method 6B data collected during this
program was carefully determined and the results are discussed in detail in
Section 3. Assessment of the quality of the ANL Unit 5 boiler and FGD process
data, summarized in Table 2-2, was beyond the scope of this program. There-
fore, the quality of these data is subject to conjecture. One possible ex-
ception to this is the Contraves/Goerz continuous emission monitoring system
(GEMS) data collected from the stack during this program. The relative accu-
racy of the Contraves/Goerz GEMS was not determined directly using EPA-accepted
methodology, during this program. However, by comparing EPA Method 6B results
to Contraves/Goerz GEMS results obtained during the same time period, the
relative accuracy of the Contraves/Goerz can be estimated.
Table 2-6 presents a summary of the EPA Method 6B and Contraves/Goerz
S02 and COz concentration data and the SOz emission rate data. The Contraves/
Goerz SOz and COz concentration data are reported on a wet basis. The EPA
Method 6B S02 and COz concentration data are reported on a dry basis because
the moisture content of the flue gas was not determined during this program.
Therefore, EPA Method 6B and Contraves/Goerz SOz and C02 concentration data
cannot be compared directly. However, if you assume a moisture content of
about 15 percent (dew point temperature ^130°F) in the stack, a basis of
comparison can be established. The percent difference between the EPA Method
6B and the Contraves/Goerz S02 and COz concentration data should approach the
fraction of moisture (^15%) in the flue gas if the two sources of SOz and COz
20
-------
data agree with each other. Based on the data in Table 2-6, the Contraves/
Goerz COa data compare within ±13% of the EPA Method 6B data on all days
except August 6, 7, and 9, 1983.
The Contraves/Goerz SOa concentration data and SOa emission rate data,
in Table 2-6, do not compare favorably with thecorresponding EPA Method 6B
SOa data collected during periods of S02 compliance (1.2 Ibs S02/106 Btu).
The Contraves/Goerz SOa monitor on the stack does not appear to be very
linear below about 400 ppra SOa (dry basis). In most cases, as the actual
SOa concentration (based on EPA Method 6B) decreases below 400 ppm, the
error in the Contraves/Goerz monitor increases substantially. For example,
on August 12, 1983 the Contraves/Goerz average SOa value (463 ppm) was
approximately 40% higher than the corresponding EPA Method 6B SOa concen-
tration (389 ppm) after adjusting the Contraves/Goerz SOa data for 15%
moisture. On August 26, 1983 the Contraves/Goerz average SOa value (319
ppm) was approximately 200% higher than the corresponding EPA Method 6B
SOa concentration (125 ppm) after adjusting the Contraves/Goerz SOa data
for 15% moisture. The high Contraves/Goerz SOa concentrations resulted
in the Contraves/Goerz SOa emission rates being proportionately higher
than SOa emission rates based on EPA Method 6B data.
21
-------
TABLE 2-4. ANL UNIT 5 SPRAY DRYER-BAGHOUSE SYSTEM UPSET SUMMARY
FOR THE PERIOD OF AUGUST 1 THROUGH AUGUST 29, 1983
Date
• nuns
ftfln?
0808
nftAQ
UolU
nfli i
0312
0825
0826
AQO7
nooo
Approximate Tine Period
1700 hra - 2400
0000 hra - 2400
0000 hra - 2400
0000 hra - 1100
0245 hra - 0600
1900 hra - 2400
0000 hra - 1400
2200 hra - 2400
0000 hra - 2400
0000 hra - 1300
Approximately 0500-2400
0000 hrs - 2400
0000 hra - 2400
0000 hrs - 2400
Type of Upset
Outlet Contravea
Inoperable
Outlet Contravea
Outlet Contravea
Inoperable
Outlet Contravea
Tnonerahle
High Abaorber Pressure
High Abaorber Pressure
Hlah Abaorber Preaaure
High Abaorber Preaaure
Hlah Abaorber Preaaure
Hlfth Absorber Preaaure
Low Slurry Feed
Electrical Short Circuit
Electrical Short Circuit
Electrical Short Circuit
Electrical Short Circuit
Effect on Performance
Pnnr Syat-pm Prtnrrnl
Poor Syntpm Control
Poor Ryatj»m Pnnrrnl
Flue Gaa Bypasses FGD
System — No Scrubbing
Flue Gaa Bypasses FGD
System — No Scrubbing
Flue Gas Bypasses FGD
System — No Scrubbing
Flue Gas Bypasses FGD
System — No Scrubbing
Flue Gaa Bypasses FGD
System — No Scrubbing
Flue Gas Bypasses FGD
System — No Scrubbing
Reduced SO 2 Removal and
Higher Spray Dryer Exit
Gas Temperature
Flue Gas Bypasses FGD
System—No Scrubbing
Flue Gas Bypasses FGD
System — No Scrubbing
Flue Gas Bypasses FGD
System — No Scrubbing
Flue Gas Bypasses FGD
System — No Scrubbing
22
-------
TABLE 2-5. SUMMARY OF EPA METHOD 6B 24-HOUR AVERAGE S02 REMOVAL AND
EMISSION RATE DATA COLLECTED AT ANL FROM AUGUST 2 THROUGH
AUGUST 26, 1983 DURING PERIODS OF "TROUBLE FREE" OPERATION
Average SOz
Date Removal
(MMDD) Efficiency
0802
0803
0804
0805
0806
0807
0813
0814
0815
0816
0817
0818
0819
0820
0821
0822
0823
0824
Overall
86.9
86.5
92.1
86.5
82.2
83.0
88.9
83.6
83.8
91.8
88.4
85.2
86.9
89.1
87.3
87.5
84.2
87.7
Average 86.8
aThe EPA Method 6B sampling
indicated to 0900 hours on
Average SOa
ng/J
245
303
150
267
373
350
221
334
332
164
236
305
273
224
261
255
324
256
271
trains were
the next day
Emission Rate
lbs/106 Btu
0.570
0.704
0.348
0.622
0.866
0.814
0.510
0.779
0.771
0.380
0.548
0.708
0.633
0.521
0.607
0.594
0.752
0.594
0.629
operated from
to constitute
Spray Dryer
Approach To
Saturation (°F)
29
27
29
27
17
23
29
29
28
28
26
28
23
26
24
27
28
26
27°
1000 hours on the day
a 24-hour sample.
ture and the baghouse outlet dew point temperature.
"Does not include value for August 6, 1983.
23
-------
TABLE 2-6. COMPARISON OF PROCESS DATA COLLECTED USING THE EPA
METHOD 6B AND THE CONTRAVES/GOERZ GEMS ON THE STACK
AT ANL FROM AUGUST 2 THROUGH AUGUST 25, 1983
Stack S02 Concentration
Stack CO 2 Concentration (%) (ppm)
Date Cont raves/
(MMDD) Goerz3
0802
0803
0804
0805
0806
0807
0808
0809
0810
0811
0812
0813
0814
0815
0816
0817
0818
0819
0820
0821
0822
0823
0824
0825
5
6
5.4
8.2
—
8.6
8.1
7.9
8.3
8.5
8.9
9.5
9.5
8.2
8.8
9.6
9.3
9.1
9.5
9.2
3Contraves/Goerz S02 and
b£PA Method 6B data are
C(
I1-
Contraves
EPA
Method
6Bb
9.2
8.7
10.0
10.0
9.8
9.8
9.6
9.3
9.6
9.8
9.2
9.0
8.9
9.0
9.6
10.1
10.2
10.3
9.8
10.0
10.0
9.8
10.1
10.2
Contraves/
AZC Goerza
-40
-28
-34
3.7
—
3.2
3.6
-3.3
9.7
11.1
9.1
10.7
9.6
-6.3
5.6
12.9
9.4
9.2
10.7
6.1
CO 2 values are on
on dry
^
15 Moisture Fraction/
EPA
Contraves - EPA
EPA
AZ
basis.
EPA
—
—
506
903
—
1230
463
310
349
352
319
319
400
297
375
372
374
398
370
588
wet basis.
AZ
EPA
Method
6B
180
208
116
212
286
271
515
796
702
1340
389
158
236
236
125
188
245
222
174
206
200
250
204
586
A%C
—
—
16
33'
—
8.0
40
131
74
75
200
100
92
57
154
112
120
87
113
18
SO 2 Stack Emission
(lbs/106 Btu)
Contraves/
Goerz
1.26
1.25
1.24
1.10
—
1.41
1.75
—
2.02
1.32
1.11
1.20
1.21
1.02
1.02
1.20
1.09
1.10
1.12
1.11
1.25
1.10
1.33
EPA
Method
6B
0.570
0.704
0.348
0.622
0.866
0.814
1.57
2.52
2.15
4.01
1.25
0.510
0.779
0.771
0.380
0.548
0.708
0.633
0.521
0.607
0.594
0.752
0.594
1.69
Rate
A%d
121
78
256
77
—
- 10
- 31
—
- 50
5.6
118
54
57
168
86
69
72
111
85
87
66
85
-21
24
-------
SECTION 3
DATA QUALITY
The test approach used during this project for FGD system characteriza-
tion incorporated a comprehensive quality assurance/quality control (QA/QC)
program as an integral part of the overall sampling and analytical efforts.
The QA/QC program was designed, in part, to ensure that the SOa emission
rate/removal efficiency data collected during the test program were complete,
representative, and comparable to other similar data. It was also designed
to control measurement data quality within prescribed limits of acceptability,
and to ensure that the resulting data were of known quality with respect to
precision and accuracy. The QA/QC efforts addressed only efforts associated
with Method 6B sampling and analysis. Control and assessment of process
data quality were not within the scope of work for this project.
This section presemts an assessment of the quality of measurement data
collected during this test program. This assessment is based upon QC data
and quality assurance audit results, and provides estimates of the uncertainty
associated with the measurement data. Section 3.1 presents conclusions and
a summary of QA/QC results. A discussion of the objectives of the QA/QC
efforts, and the general approach used in achieving these objectives, is
presented in Section 3.2. Methods used in quantitating data quality, along
with definitions and explanations of QA/QC and statistical terminology are
discussed in Section 3.3. Audit procedures and results are presented and
discussed in Section 3..4, while Section 3.5 addresses QC procedures and re-
sults used to assess precision of the Method 6B data. Section 3.6 contains
a brief discussion of reliability of the Method 6B sampling system and the
impact upon data capture (completeness).
25
-------
3.1 SUMMARY AND CONCLUSIONS
Quality control data collected throughout the course of the measurement
program, along with performance audit results, provide the basis for assess-
ing the quality of the SOz emission rate/removal efficiency data. These
qualifying data indicate that the measurement data are of adequate quality
to fulfill the program objectives. Table 3-1 presents measured values for
precision and accuracy (bias) of the SOa emission rate data and the measure-
ment parameters required to calculate emission rate. Precision and accuracy
objectives which were presented in the Quality Assurance Project Plan (4)
for this project are shown for comparison.
As indicated in Table 3-1, precision and accuracy of the emission rate
data were well within the objectives presented in the QA Project Plan. SO2
and C02 concentrations and sample volume are not separate measurement para-
meters as such, but rather are component parts of the SOa emission rate
determination. Data quality for each of these individual components was
also within the specified objectives. Based on measured precision and bias
for the various components of the measurement system, 95% of the removal
efficiency data would be expected to be in error by less than ±5% of the re-
ported value (i.e., for at least 95% of the data, the "true" value should be
within ±5% of the reported value).
The data quality measurements presented in Table 3-1 are average
values and, as such, provide only a cursory glimpse of the data quality
assessment performed for this project. The QA/QC program was designed to
provide detailed information pertaining to the limitations associated with
the measurement data. For instance, results for duplicate samples indicate
that precision of the emission rate data at the inlet of the FGD system
(2.8%) was considerably better than that at the outlet (8.3%). While the
performance audit results and QC data presented in the remainder of this
section provide the primary basis for evaluation of uncertainty in the
emission rate measurements, this evaluation requires careful interpretation
of the audit and QC data in the context of the measurement data and the
manner in which the individual measurement parameters are related.
26
-------
TABLE 3-1. SUMMARY OF ESTIMATED VS. MEASURED DATA QUALITY
N)
a
„_ _ Precision
Parameter Estimated0 Measured
S02 Emission Rate 8.0% 6.2%
S02 Concentration 5.0% 0.5%
C02 Concentration 5.0% 3.0%
Sample Volume 2.0%
Accuracy fe
c e £
Estimated Measured J
±7.7% 0.0% ± 1.6%
±5.0% -0.2% ± 0.1%
±5.0% 0.0% ± 0.2%
±5.0% -2.6% ± 1.7%
Coefficient of variation for replicate samples
Bias (systematic error), expressed as a percentage of the measured value (i.e., relative error)
Q
Data quality objective presented in the Quality Assurance Project Plan
Pleasured precision (replicability) based on observed variability in results for duplicate
samples collected using colocated sampling trains
£
95% confidence interval for mean relative error (bias) based on performance audit results
Precision and bias of analytical phase of the method (i.e., barium-thorin titrations)
-------
3.2 QA/QC PROGRAM OBJECTIVES
For any measurement effort, there always exists some degree of uncer-
tainty associated with the measurement data due to inherent limitations of
the measurement system. Usefulness of the measurement data is dependent
upon the degree to which the magnitude of this uncertainty is known and upon
its relative impact. The industrial boiler FGD system testing described in
this report included a comprehensive quality assurance/quality control (QA/QC)
program. The objectives of the QA/QC efforts were twofold. First, they
provided the mechanism for controlling data quality within acceptable limits.
Second, they form the basis for estimates of uncertainty by providing the
necessary information for defining error limits associated with the measure-
ment data.
The quality assurance function was organized to provide independent
review and assessment of project activities and their ability to achieve
the stated data quality objectives. The QA Coordinator for the project had
the responsibility of evaluating the adequacy and effectiveness of the QC
system and providing assurance that it was, in fact, responsive to the
specific needs of the program.
In addition to reviewing the test plan and providing input into design
of the QC efforts, the QA Coordinator conducted both performance and systems
audits during the test program. The performance audits were designed to
provide a direct, quantitative, point-in-time assessment of data quality
in terms of accuracy. This was achieved by using equipment and standards
which were independent of those used by the field personnel. The systems
audit was designed to provide a systematic, qualitative review and assessment
of the critical elements of the various measurement systems and associated
internal quality control (QC) systems, with emphasis upon procedures and
documentation.
28
-------
A quality control system is a system of routine internal procedures
for assuring that the data output of a measurement system meets prescribed
criteria for data quality. Inherent and implied in this control function
is a parallel function of measuring and defining the quality of the data
output. A well-designed internal QC program must be capable of controlling
and measuring the quality of the data in terms of precision and accuracy, as
well as ensuring that the data are complete, representative, and comparable.
Precision reflects the influence of the inherent variability in any measure-
ment system. Accuracy reflects the degree to which the measured value
represents the actual or "true" value for a given parameter, and includes
elements of both bias and precision. The precision and bias of the final
data are related to the precision and bias of the component parts of the
measurement system. While the QA activities served an evaluative function
which was independent of the testing efforts per se, the QC system was an
integral part of the daily technical effort. Together, the QC data and
the audit results may be used to qualify the measurement data, as discussed
in the remainder of this section.
3.3 METHODS OF QUANTITATING DATA QUALITY
Internal quality control data associated with sampling/analytical
aspects of this project, along with performance audit results and the
measurement data themselves, provide the basis for a quantitative assess-
ment of measurement data quality. The two aspects of data quality which
are of primary concern are precision and accuracy. Accuracy reflects the
degree to which a measured value represents the actual or "true" value
for a given parameter, and includes elements of both bias and precision.
Precision is a measure of the variability associated with the measurement
data.
The quality control system for this measurement effort, in its broadest
sense, included all procedures which ensured that the resulting measurement
data were of adequate quality to fulfill the program objectives. Some pro-
cedures which fall into the category of QC were primarily intended to control
29
-------
data quality within acceptable limits (e.g., adherence to specified sampling
and analytical procedures, calibration of instrumentation, careful documen-
tation of field data and results, etc.). Other QC procedures were intended
to provide data pertaining to precision (variability) and accuracy of the
measurement data. In some cases, a single QC procedure might fulfill both
control and assessment functions.
The data necessary for assessment of precision and accuracy were
obtained in several ways. The performance audits for the various measure-
ments parameters were meant to address accuracy of the measurement systems
and consisted of challenging component parts of the system with audit samples
or standards. Variability (precision) associated with the measurement system
was measured and documented using QC procedures such as control sample
analyses, duplicate analyses, and collection of duplicate samples. These
procedures are discussed in more detail in Sections 3.4 and 3.5. This
section is devoted to discussion of the procedures and operating definitions
used in quantitating data quality.
3.3.1 Definitions of Precision, Accuracy, and Bias
Precision, by the definition presented in the EPA Quality Assurance
Handbook (3), is "a measure of mutual agreement among individual measurements
of the same property, usually under prescribed similar conditions."
Different measures of precision exist, depending upon these "prescribed
similar conditions." Radian typically uses the EPA definitions for replica-
bility, repeatability, and reproducibility, as summarized in Table 3-2.
Accuracy is a measure of the degree of agreement between a measured
value and the true value of the measured parameter. For single measurements,
accuracy includes components of both bias and precision, i.e., both systematic
and random error. Accuracy of the average of individual measurements equates
accuracy with bias and represents an attempt to quantitate systematic error
(bias) independent of random error (precision). The validity or significance
of the estimate of bias is directly related to the number of individual
30
-------
TABLE 3-2. MEASURES OF PRECISION
Source of
Variability
Replicability
Repeatability
Reproducibility
Specimen
(subsample)
Sample
Analyst
Apparatus
Day
Laboratory
Same or
different
Same
Same
Same
Same
Same
Same or
different
Same
At least one
of these must
be different
Same
Most likely
different
Same
Different
Different
Same or
different
Different
31
-------
measurements used to compute the average. It is based on the principle
that, as the number of individual measurements is increased indefinitely,
the sample mean, X, approaches a definite value, y. The difference between
y and the true value, T, represents the magnitude of the measurement bias,
or systematic error. The error in each individual measurement reflects this
systematic error plus random error due to imprecision.
3.3.2 Assessment of Accuracy and Bias
Performance audits represent the primary mechanism for assessing
accuracy of a measurement system and, by extrapolation, accuracy of the re-
sulting measurement data. When a measurement system is challenged with an
audit sample, the true value of which is known, the degree of agreement
between the measured value and the true value reflects the accuracy of the
measurement. The difference between the two is due to measurement error and
includes both random error (imprecision) and systematic error (bias). This
difference, expressed as "percentage of the true.value," is often referred
to as "relative accuracy," or simply "accuracy," although technically it
represents inaccuracy.
Typically, repeated measurements are made of the parameter of interest
for the same audit sample, or for additional samples at different levels,
and the average error is then calculated. As discussed in Section 3.3.1
above, this error value represents an estimate of measurement bias or
systematic error, although it is also often labeled "accuracy."
The significance of this bias estimate may be evaluated using confidence
intervals. An approximate 95% confidence interval for the mean error can be
calculated using:
—. . / Standard Deviation \
CI = Mean(X) ± t. --c , 1N
mean error v ' 0.025,(n-l) \ /— /
\ Vn /
32
-------
where n is the number of measurements used to compute the average and stan-
dard deviation and t is a table statistical value (0.025 confidence level,
n-1 degrees of freedom; where n is greater than 10, t approaches 2.0).
As an example, for a particular set of nine measurements, an overall
mean of 20 ppm is reported, and the standard deviation of these data is 10
ppm. Also, the true concentration if 30 ppm. for these measurements, the
95% confidence interval is:
95% CI = 20 ± 2.3 | 10
mean 1
or 20 ± 7.7
which is the interval ranging from 12 ppm to 28 ppm. Since this interval
does not include the true value, 30 ppm, a conclusion of bias is justified.
The magnitude of this bias is between 2 and 18 ppm. The uncertainty in the
bias estimate is due to variability arising from random error.
For the audit data presented in Section 5.4, results are presented in
terms of relative error where:
-. T . . „ Measured Value - True Value ,....
Relative Error = - —t x 100
True Value
Results for a given set of audit data are typically summarized in terms of
"mean relative error," which represents an estimate of the bias of the
measurement system. Variability among the individual error values used
to calculate mean relative error reflects one aspect of the overall preci-
sion associated with the measurement system. This variability is typically
quantitated in terms of the standard deviation of the relative error, which
is also presented.
The confidence interval approach may also be applied to audit results
expressed in terms of mean relative error. For example, consider a set of
33
-------
audit data for which a mean relative error of -5.0% is reported with a
standard deviation of 6.0%, based on five observations (i.e., five audit
sample analyses). For these measurements, the 95% confidence interval for
the mean is:
95% CI = -5.0% ± 2.8 2^*- or -5.0% ± 7.5%
5
Since the confidence interval (-12.5%, 2.5%) includes zero, conclusion of
bias is not justified and the audit data indicate that the measurements are
accurate within the limits of precision.
3.3.3 Assessment of Precision
As stated above, accuracy of measurement data is a function of both
bias (systematic error) and precision (random error). If a particular measure-
ment method is known or assumed to be unbiased, i.e., free of systematic
error, then accuracy of the results is limited only by random variability,
i.e., by the precision of the measurements. For most standard or accepted
source'test methods, random error is the major source of measurement error.
For the sampling/analytical procedures used in this program, the measure-
ment data precision (i.e., random error, exclusive of temporal variability)
is a function of the combined effects of analytical variability and sampling
variability. Each of these two sources of variability could be further sub-
divided into numerous specific components of variability such as that asso-
ciated with standardization of the barium chloride titrant, sample handling,
etc.
The precision estimates presented in Section 3.5 are based on observed
variability among replicate or repeat measurements made under various "pre-
scribed similar conditions," selected for specific purposes. This variability
was quantitated by first calculating the standard deviation for each set of
34
-------
measurements. The standard deviation is a measure of the average distance
of individual observations from the mean. It is usually denoted s and
defined as:
S (X.-X)2
s - - - ±=1
o
n-1
where: n is the sample size,
X. is the i observation in the sample, and
X is the sample mean.
In order to facilitate comparison of variability at different concen-
tration levels, measured variability is reported in terms of the coefficient
of variation (also known as relative standard deviation) which is defined as:
CV = Standard Deviation ^ 1Q()%
Mean
When individual measurements of variability (i.e., CV) were combined (pooled)
to obtain an overall measure of variability for a given set of conditions or
measurements, the following technique was used:
Pooled CV
where X. = CV of data set i (e.g., CV for one duplicate pair, i),
DF. = degrees of freedom for data set i (k.-l),
n = total number of data sets (e.g., total number of
duplicate pairs),
k. = number of data points in set i (e.g., k=2 for duplicates),
i = data set 1,2,3 ... n
35
-------
In Section 5.5, variability in the Method 6B data is evaluated in terms
of that arising from various components of the method. Magnitudes and
relative contributions of each component or source of variability are pre-
sented. This evaluation of the measurement data was performed using a
statistical technique known as analysis of variance (ANOVA). This technique
separates the variation present in a set of data into independent components
and then tests hypotheses about these components. A complete discussion of
the ANOVA technique is given by Cochran and Cox (5).
Also presented in Section 3.5 are results for paired t-tests performed
using Method 6B S02, C02, and emission rate data. These tests were performed
to evaluate the statistical significance of observed differences in results
between the colocated sampling trains. This statistical procedure consisted
of calculating the difference between results for Train A and Train B, at both
inlet and outlet locations, for each test run (i.e., each day). For each
pair of trains, and each parameter, mean differences and standard deviations
were calculated. The hypothesis that the mean difference was equal to
zero was then tested at the 95% confidence level using a t-test. The
formula for the t-test is:
where d = mean of the observed paired difference,
u, = hypothesized mean difference, i.e., zero,
S, = standard deviation of the paired differences, and
a
n = number of paired differences in the sample set.
For the hypothesis test, if the calculated value of t is greater than
the table value of t for the sample size n (i.e., n-1 degrees of freedom),
the null hypothesis must be rejected. Rejection of the null hypothesis
(mean difference is equal to zero) would indicate that the difference between
results for the paired trains was statistically significant. If the calcu-
lated value of t is less than the table value, we must fail to reject the
null hypothesis. This is equivalent to saying that there is no reason to
36
-------
believe that observed differences in results between the trains are signifi-
cant (i.e., there is no reason to believe they are not equal to zero).
3.4 QUALITY ASSURANCE AUDITS
A quality assurance audit of measurement efforts associated with this
test program was conducted August 27-29, 1983 at Argonne National Laboratory
in Argonne, Illinois. This audit was performed by the project QA Coordinator
and included performance audits of selected components of the measurement
system, as well as a systems audit of the overall test effort. Audit proce-
dures and results are disscussed in this section.
3.4.1 Performance Audits;
Performance audits for this program provided a direct, point-in-time
evaluation of the capability of the measurement system to generate data of
acceptable quality. In its broadest sense, the measurement system consisted
of numerous components, including the equipment, apparatus, calibration
standards, and personnel used to perform the testing, as well as the asso-
ciated, procedures and techniques used for sample collection, sample analysis,
and data reduction. The primary measurement parameters for this program
were SC-2 emission rate and removal efficiency. These parameters cannot
be measured directly, but rather are calculated based upon measurements
of S02 and COa concentrations in flue gas, and carbon content and gross
caloric value of the fuel. The performance audits were therefore designed
to address the measurement parameters used in calculating SOj emission rates
and removal efficiencies.
The emission rate/removal efficiency "measurement system" may be con-
sidered to have consisted of two subsystems. The primary subsystem was
that used for measuring SOz and COz concentrations in the flue gas. The
other subsystem was that used for determining carbon content and gross
caloric value of the fuel, which were in turn used to calculate the C02
F-factor. Performance audit activities addressed both of these subsystems.
37
-------
Since the fuel analyses represented a relatively minor component of the
overall measurement effort, the performance audit of that subsystem con-
sisted merely of submitting a standard coal sample for analysis along with
one of the actual coal samples. The performance audit of the flue gas S02
and C02 measurements was considerably more involved, commensurate with the level
of effort involved and complexity of the measurement system. As discussed
in Section 5, EPA Method 6B was used for determination of S02 and C02 con-
centrations in the inlet and outlet flue gases. This measurement system as
a whole was audited using standard atmospheres of S02 and C02 in nitrogen.
Major components of the Method 6B measurement system were also audited in-
dividually. These included the analytical phase of the S02 determinations
(i.e., barium-thorin titrations), the dry gas meters used for gas volume
measurements, and the balance used for gravimetric determination of C02.
Performance audit results are summarized in Table 3-3. Audit procedures and
detailed results are presented and discussed in the remainder of this
section.
3.4.1.1 Method 6B Measurements—
The Method 6B sampling and analytical system as a whole was audited by
challenging the system with test atmospheres of S02 and C02 in nitrogen.
These test atmospheres were collected using the four sampling trains (two
trains at the inlet and two at the outlet of the spray dryer/fabric filter
FGD system) in their normal configuration. The only difference between
normal sample collection procedures and those used for the audit test runs
was the sampling interval. For the audit, sampling was performed continuously
over intervals of approximately one hour duration, as opposed to the normal
procedures of intermittent sampling over a 24-hour period.
Test atmospheres for the first two audit runs were generated by blending
two compressed gas mixtures, one containing COa (and 02) in nitrogen, and
the second containing SOa in nitrogen. The third test run used only an
S02 mixture. A total of three gas mixtures were used to generate different
S02 and C02 concentrations for the three test runs. Two S02 mixtures were
used, both of which were EPA Traceability Protocol mixtures obtained from
38
-------
Parameter
SOa (Sampling
and Analysis)
C02 (Sampling
and Analysis)
SOa (Analysis
Only)
Gas Volume
Weight
% Carbon in
Coal
Btu/lb Coal
Analytical
Method Instrument Audit Standard (s)
EPA Method 6B -'- Scott Environmental
Technology S02 Cyl.
//AAL 11426 and Cyl.
//AAL 11470
EPA Method 6B — Scott Environmental
Technology 02/C02 Cyl.
//AAL 6541
Barium-Thorin — EPA Stationary Source
Titration QA Reference Standards
Lot //0980
Dry Gas Meter Singer //K418992 GCA/Precision Scientific
Wet Test Meter //14AES
Singer //H988524 GCA/Precision Scientific
Wet Test Meter //14AES
Singer //H988523 GCA/Precision Scientific
Wet Test Meter //14AES
Singer //H988525 GCA/Precision Scientific
Wet Test Meter //14AES
Balance Mettler PC 4400 Ainsworth 4254-S Class
Serial No. 816571 S Weights Serial No.
36697
ASTM D3178 — Alpha Resources Coal
Standard AR 2781 Lot #315
ASTM D2015 — Alpha Resources Coal
Standard AR 2781 Lot #315
Mean Relative
Error3
0.15%
0.00%
-0 . 20%
-2.80%
-1.30%
-3.91%
-2.55%
±0.02gb
-0.35%
-0.72%
Average percentage error, unless otherwise indicated.
Error range, in grams.
-------
Scott Environmental Technology, Inc. One mixture (Cylinder #AAL 11426)
had an S02 concentration of 1630 ppm while the other (Cylinder #AAL 11470)
contained 612.6 ppm S02. The third mixture (Cylinder #AAL 6541) was a
"certified" standard (analytical accuracy ±2%) also obtained from Scott,
containing 30% C02 and 40% 02 in nitrogen.
The C02/02 mixture was blended with the S02 mixtures using a Radian-
modified Bendix Model 8861-DA gas dilution system. This sytem uses precision
pressure regulators to control flow through a series of capillary flow re-
strictors. Various ratios of two gas mixtures are obtained using different
capillary combinations. Capillary flows (i.e., mixing ratio) were measured
immediately before each test run using an NBS-traceable Hastings HBM 1A soap
bubble flow meter. Audit gas mixtures were introduced to the sampling trains
using a manifold system which incorporated a tee for venting excess flow to
the atmosphere, preventing pressurization of the manifold. Duplicate samples
were collected during each of the three audit test runs using either inlet or
outlet sampling train pairs. Two runs were conducted using the inlet trains
and one using the outlet train, for a total of six samples.
Results for the Method 6B audit runs are presented in Table 3-4. As
discussed in Section 3.3 above, individual values for relative error include
both systematic and random error components (i.e., error due to both bias
and imprecision). By averaging relative error values for a given parameter
to obtain mean relative error, variability due to imprecision tends to be
"averaged out." Thus, mean relative error is the best available estimate
of measurement bias. The 95% confidence interval is a range which takes
into account variability among the observations and the number of observa-
tions in the sample set to define the uncertainty of the bias estimate. It
represents the interval within which we can be 95% confident that the "true"
mean value (i.e., the population mean) falls. If, as is the case for the
Method 6B audit data, the 95% confidence interval for mean relative error
includes zero, a conclusion of bias is not justified and the measurement
data are judged to be accurate within the limits of its precision. Preci-
sion of the Method 6B data is discussed in Section 3.5 below.
40
-------
TABLE 3-4. METHOD 6B AUDIT RESULTS
SO,
Test Train Actual SO, Measured S02 Relative
Run 10 Concentration Concentration Error0
(ppm) (ppm) (X)
1 Inlet A 1311
1 Inlet B 1311
2 Outlet A 425.5
2 Outlet B 425.5
3 Inlet A 1630
3 Inlet B 1630
a
B C \% CO 2
Difference between measured and
n.i_M,,D R , = Measured Cone.
1312
1279
427.1
431.4
1643
1641
Mean Relative Error
Standard Deviation
95Z C.I.e (-1.
from corresponding S02
) x 106, when CB - S02
actual concentrations,
- Actual Cone. „ ,„.
0.08
-2.44
0.38
1.39
0.80
0.68
0.15X
1.34X
19X, 1.56X)
Actual C02
Concentration
(X v/v)
5.86
5.86
9.16
9.16
0.0
0.0
CO,
Measured CO2
Concentration
(X v/v)
5.78
5.74
9.16
9.35
0.0
0.0
Emission Rate3
Relative Actual Measured Relative
Error0 Emission Rate Emission Rate Error0
(X) (ng/J) (ng/J) (X)
-0.08 2838 2879 1.44
-0.12 2838 2826 -0.42
0.00 589 591 -0.34
0.19 589 585 -0.68
0.00 0.00
0.14 0.97
(-0.22X, 0.22X) (-1.55X, 1.55Z)
-8 3
and C02 concentrations using a carbon dioxide F-f actor (F ) of 4.772 x 10 Mm /J, where
concentration, mg/Nm
expressed as a percentage of the actual concentration. I.e.,
Actual Cone.
CEstlmate of bias, or systematic error
Standard deviation of relative error; Indicative of variability about the mean
952 confidence interval for mean relative error
90
S
-------
The emission rate values in Table 3-4 were calculated from the SOz and
_8
COa concentrations using a carbon dioxide F-factor value of 4.772 x 10 Mm /J.
"Actual emission rates" for each run were calculated using corresponding
values for "actual SOa concentration" and "actual COa concentration."
"Measured emission rates" were calculated in the same manner, using S02
and COa results for each run. These data and corresponding error values
are presented to illustrate error propagation in calculating emission rates
from SOa and COa measurements. As shown in the table, the 95% confidence
interval for mean relative error of the emission rate values is larger
than the corresponding intervals for SOa and COa concentrations. At
±1.55%, it is, however, well within the ±7% objective specified in the
Quality Assurance Project Plan (4).
3.4.1.2 S02 Analyses—
The audit test runs described above provide estimates of total measure-
ment error for Method 6B sampling and analysis. A performance audit was
also performed on the analytical phase of the Method 6B SOa determination,
to assess error associated with the barium-thorin titrations. This audit
consisted of submitting, for analysis a set of five Stationary Source Quality
Assurance SOa Reference Samples. These audit samples were obtained from
the U.S. Environmental Protection Agency, Quality Assurance Division, EMSL/
RTP.
Results for the SOa analytical audit samples are presented in Table 3-5.
Although the magnitude of the mean relative error value (i.e., bias estimate)
for the analytical phase of the method is slightly larger than the corres-
ponding value for sampling and analysis (Table 3-5), the standard deviation
and 95% confidence intervals are smaller indicating less variability (i.e.,
better precision). Since the 95% confidence interval for mean relative
error does not include zero, a slight negative bias is indicated. Overall,
however, these results are excellent, with no observed relative error greater
than 0.4%. Using these audit data to calculate statistical tolerance limits,
it may be shown that, at the 95% confidence level, at least 90% of the
42
-------
Sample
Number
(Lot 0980)
8286
TABLE 3-5. S02 ANALYTICAL AUDIT RESULTS
Actual S02 '
Concentration
(mg/Nm3)
381.3
Measured
C one en t ra t ion
(mg/Nm3)
381.1
Relative
Error3
-0.05
4442
762.6
760.7
-0.25
1823
1143.9
1140.2
-0.32
u>
2357
5565
1906.5
2287.8
1902.5
2284.4
Mean Relative Error
Standard Deviation
95% C.I.C
-0.21
-0.15
-0.20
0.10
(-0.32,-0.07)
Difference between measured and actual concentrations, expressed as a percentage
of the actual concentration, i.e.,
_ , „ Measured Cone. - Actual Cone. ,_-.
Relative Error = T—-—1—7; x 100
Actual 'Cone.
Estimate of bias, or systematic error
c
Standard deviation of relative error; indicative of variability about the mean
95% confidence interval for mean relative error
-------
analytical data would be expected to have relative errors in the interval
ranging from -0.62% to 0.23%.
3.4.1.3 Dry Gas Meters—
Each of the four dry gas meters (DGMs) used for Method 6B sampling
were audited using a GCA/Precision Scientific wet test meter (0.1 ft3 per
revolution, Serial Number 14 AES). Two calibration check runs were performed
for each DGM. Nominal flow rates of one liter per minute were used, corres-
ponding to the normal sampling flow rates. A minimum of 8 liters of air was
drawn during each test run. Gas volumes measured by the wet test meter and
the DGMs were used to calculate dry gas meter correction factors (DGMCFs)
for each DGM. These audit values are presented in Table 3-6. As indicated
in the table, audit correction factors for all four meters agreed with the
pretest calibration factors within the ±5% acceptance criterion.
3.4.1.4 Balance—
The Mettler Model PC 4400 top loader balance (Serial Number 816571)
used for gravimetric determination of COz was audited using a set of Class S
standard weights (Ainsworth 4254-S, Serial Number 36697). As shown in
Table 3-7, the balance was accurate within ±0.02 g over the audit range of
0.02 g to 210.00 g. Since the Ascarite® columns were weighed only to the
nearest 0.1 g, errors of less than ±0.05 g would have no measureable impact
upon the CC-2 data. By way of comparison, a weighing error of 0.1 g would
result in an error of about 0.14% CC-2 > for a sample volume of 0.04 Nm3.
With a nominal CC-2 concentration of 10%, this would represent a relative
error of less than 1.5% in the C02 measurement, or the emission rate measure-
ment, as well.
3.4.1.5 Proximate/Ultimate Fuel Analyses—
The C02 F-factor used for calculating SQz emission rates from Method 6B
SOa and COz data was calculated based upon proximate/ultimate analyses of
the coal used for firing the boiler. These coal analyses were performed
by Commercial Testing and Engineering, Inc. The performance audit of
these analyses consisted of submitting a standard coal sample for analysis
44
-------
01,
TABLE 3-6. DRY GAS METER AUDIT RESULTS
Meter
S/N
DGM Correction
Factor, Y, from
Pretest Calibration
Measured DGM Correction
Factors (Audit Results)
Run 1 Run 2
Mean Measured
DGMCF
1
1
n
0
Mean g S
Difference ip
<%) •
K418992 0.9846 1.0135 1.0126 1.0130 -2.80
H988524 0.9980 1.0097 1.0126 1.0112 -1.30
H988523 1.0212 1.0666 1.0589 1.0628 -3.91
H988525 0.9924 1.0272 1.0095 1.0184 -2.55
^Difference between pretest calibration DGMCF (Y) and that measured during audit expressed as a
percentage of the mean audit value, where
M «/ mcc Mean Measured DGMCF - Pretest Calibration DGMCF nnn
Mean % Difference = , _.„..-,- x 100
Mean Measured DGMCF
-------
TABLE 3-7. BALANCE AUDIT RESULTS
Standard
Weight
(grams)
210.00
200.00
180.00
150.00
100.00
50.00
30.00
20.00
10.00
5.00
3.00
2.00
1.00
0.50
0.30
0.20
0.10
0.05
0.02
Measured
Weight
(grains)
210.01
200.02
180.01
150.01
100.01
50.01
30.01
20.00
9.99
5.00
3.00
2.00
1.00
0.50
0.30
0.19
0.10
0.05
0.02
Mean Error
Standard Deviation
95% C.I.d
*a
Error
(jrarns)
0.01
0.02
0.01
0.01
0.01
0.01
0.01
0.00
-0.01
0.00
0.00
0.00
0.00
0.00
0.00
-0.01
0.00
0.00
0.00
0.003
0.007
(0.000, 0.007)
a
Error, in grams (measured weight - standard weight)
Average of individual errors, in grams
£
Standard deviation of individual errors, in grams
95% confidence interval for mean error, in grams
46
-------
along with an actual sample. The audit standard (Part Number AR 2781, Lot
Number 315) was obtained from Alpha Resources, Inc.
Results for the fuel audit analyses are presented in Table 3-8.
Although a total of nine parameters are included in the proximate/ultimate
analyses, only two of these, percent carbon from the ultimate analyses and
Btu/lb from the proximate analysis, are used in calculating C02 F-factors.
As indicated in the table, results for both of these parameters were accurate
within ±1.0%. If the actual % C02 and Btu/lb values for the audit sample are
used to calculate a COa F—factor, a value of 1771 standard cubic feet (SCF)
_8 3
per million Btu is obtained (4.757 x 10 Nm /J). Using the measured values,
an F-factor of 1796 SCF/106 Btu (4.824 x 10~8 Nm3/J) is obtained. Thus, the
audit results correspond to an F-factor relative error of about 1.4% (i.e.,
the difference between the two values represents 1.4% of the value obtained
using the "actual" % carbon and Btu/lb values).
3.4.2 Systems Audit
A systems audit is an on-site qualitative review of the various aspects
of a tptal sampling and/or analytical system to assess its overall effective-
ness. It represents a sxibjective evaluation of a set of interactive systems
with respect to strengths, weaknesses, and potential problem areas. The
audit provides an evaluation of the adequacy of the overall measurement
system(s) to provide data of known quality which are sufficient, in terms
of quantity and quality, to meet the program objectives.
A systems audit of the measurement system used for the FGD system
characterization testing was conducted at the time of the on-site performance
audit. Prior to the field audit, a checklist was prepared which delineated
the critical aspects of the test methodology, using the Quality Assurance
Project Plan (4) as a guide. The checklist served as a tool to direct the
focus of the systems audit and to document relevant observations. A copy
of the completed checklist is shown in Figure. 3-1.
47
-------
-p-
OO
TABLE 3-8. PERFORMANCE AUDIT RESULTS FOR COAL ANALYSES
Parameter
Proximate Analysis
% Ash
% Volatile
% Fixed Carbon
Btu/lb.
% Sulfur
Ultimate Analysis
% Carbon
% Hydrogen
% Nitrogen
% Chlorine
Actual
Value
6.03%
42.39%
51.58%
13561
3.14%
74.81%
5.64%
1.52%
0.01%
Value - Actual Value 1 nn
Measured
Value
6.07%
42.37%
51.56%
13419
3.08%
75.07%
5.26%
1.46%
0.01%
Relative
Error3
0.66%
-0.05%
-0.04%
-0.72%
-1.91%
0.35%
-6.73%
-3.94%
0.00%
i:
Actual Value
-------
Site;
METHOD 6B— DETERMINATION OF S02 AND C02
SYSTEMS AUDIT FORM
T'c Uxfe Date ;
Contract :
, e>\*V to
Auditor: D>L.LguQ.-S>
Yes
No
Comments
Operation
Cf
«»aoauu OAtSO CoUft. 2
Figure 3-1.
PRESAMPLING
1. Knowledge of process conditions.
2. Calibration of pertinent equipment,
in particular, the dry gas meter,
prior to each field test.
3. Adequate facilities.
4. Spare parts and support equipment
available.
5. Qualified personnel.
6. Peroxide efficiency test conducted.
7. Ascarite® cylinder properly packed,
with no open spaces or channels.
SAMPLING
1. Proper preparation and addition of
absorbing solutions to impingers.
Sampling performed at constant rate
(±10%).
Pertinent data recorded before and
after sample collection.
Sampling performed at least 2 minutes
continuously during each cycle of
operation.
Minimum of 12 equal, evenly spaced
periods of sampling per 24 hours.
Probe maintained at proper tempera-
ture.
Sampling train shielded from direct
sunlight.
Sample train leak checked at con-
clusion of run.
Total sample volume between 25 and
60 liters.
Systems Audit Checklist
49
-------
Method 6B Systems Audit Form (Continued)
Yes
No
Comments
Operation
POSTSAMPLING
_ 1. Control sample analysis performed.
_ 2. Impingers and connecting tube
rinsed.
_ 3. Proper sample aliquoting technique.
_ 4. Proper titration technique, parti-
cularly endpoint precision.
_ 5. Blank analysis performed with each
sample set.
_ 6. Calculation procedure/check.
£ CJ^P 1 • Barium perchlorate solution stan-
dardized in appropriate manner.
_ 8. Appropriate data recording format.
_ 9. Blanks used to correct field sample
results.
_ 10. Minimum of ten percent of samples
analyzed in duplicate.
COMMENTS;
Figure 3-1. (Continued)
50
-------
The systems audit conducted for this program consisted of observing and
documenting activities atssociated with the overall sampling/analytical system
employed in the FGD system testing. In addition to providing an on-site
evaluation of sampling and analytical procedures and techniques, the systems
audit included review of all record keeping and data handling systems, in-
cluding :
• documentation of equipment calibration and reagent
standardization,
• completeness of sampling data forms,
• data review and validation procedures,
• data storage and filing procedures,
• sample logging procedures,
• sample custody procedures,
• documentation of quality control data (control charts,
etc.)» and
• documentation of equipment maintenance activities.
Overall, the systems audit indicated an effective, well-organized
sampling/analytical effort which was judged to be adequate for achieving
the program objectives. Attention to details of the internal QC program
and careful compliance with specified procedures were generally observed for
both sampling and analytical activities. The quality control chart for SOz
control sample analyses was current and the analytical notebook included
provisions for noting whether acceptance criteria were met for duplicate
analyses and control sample analyses. The only notable deficiency in the
overall test effort was in regard to completeness of the master sample
logbook and the DART logbook, entries in both of which were several days
behind. However, both of these logbooks were used as backup systems.
Sample data were current in both the analytical notebook and on the sampling
data sheets, and the DART hardcopy provided a detailed record of system
operations.
51
-------
3.5 QUALITY CONTROL DATA
Internal QC procedures for this program were designed to control data
quality within acceptable limits and provide a basis for data quality
assessment. A complete overview of the internal QC program is in the Quality
Assurance Project Plan prepared for this program (4). As with any measure-
ment effort, a primary data quality consideration is measurement variability,
or data precision. In this program, quality control sample analyses,
duplicate samples, and duplicate analyses provided the mechanism for quanti-
tating sampling/analytical variability. The use of specific acceptance
criteria for QC analyses provided the mechanism for controlling measurement
data quality. A discussion of the QC procedures used for quantitating
Method 6B measurement data variability (precision) is presented in this
section, along with summaries of the QC data.
3.5.1 Control Sample Analyses
The analytical phase of the Method 6B procedure, involving the barium-
thorin titrimetric determination of SOz (as SOi,) is a critical part of the
overall method. In order to control precision and accuracy of these analyses
within acceptable limits, a 0.100. N sulfuric acid (t^SOi*) solution was used
as a control standard. As prescribed in the QC protocol, the analyst analyzed
a control sample prior to analysis of each set of Method 6B impinger solution
samples. Before sample analyses could proceed, the analyst was required to
demonstrate acceptable accuracy by analysis of the control sample. The
acceptance criterion for this control check was agreement of the measured
concentration within ±5% of the actual concentration. Additionally, as a
check on analytical precision, duplicate analyses of a 0.01 N H2SOi, standard
solution, performed prior to each set of sample analyses to standardize the
barium chloride (BaCl2) titrant, were required to differ by less than or
equal to 1% or 0.2 ml of titrant, whichever was less.
Results for the daily control sample analyses were plotted using a
control chart, shown in Figure 3-2. As shown in the figure, measured values
52
-------
. 106
0. 105
0. 104
0. 103
N 0.102
5 0. 101
M
L ® ' *
T 0.099
Y 0.098
0.097
0.096
0.095
0.094
8C
SO2 CONTROL SAMPLE DATA
=
|
s.
1
L * + • *
= •»• +
1 + * +4.
I ' + * +
^
|
r
r-
12 807 812 817 822 827
-------
ranged from 0.0988 N to 0.1013 N, or from -1.2% to 1.3%, well within the
±5% QC limit. Agreement for duplicate standard titrations ranged from a
difference of 0.10 ml to a difference of 0.00 ml for titrant volumes of
approximately 20 ml. Again this was easily within the 1%/0.2 ml titrant
acceptance criterion.
In addition to controlling analytical accuracy within acceptance
limits, the control sample analyses provide a data base for evaluating pre-
cision of the S02 analyses in terms of day-to-day variability (repeatability).
These data indicate repeatability of 0.7%, expressed in terms of the coeffi-
cient of variation (i.e., relative standard deviation).
3.5.2 Duplicate Analyses
As a continuing check on analytical precision, duplicate analyses of
Method 6B impinger solution samples were subject to the same acceptance
criterion as specified for duplicate standard titrations (i.e., difference
less than or equal to 1% or 0.2 ml of titrant volume). All impinger solu-
tion samples were analyzed in duplicate. Evaluation of these data indicate
within-day precision (replicability) of 0.27% for inlet samples and 0.67%
for outlet samples, both expressed in terms of the pooled (i.e., average)
coefficient of variation (CV). These results represent an overall pooled CV of
0.51% for duplicate analyses.
3.5.3 Duplicate Samples
In addition to SOa control sample analyses and duplicate analyses of
all S02 samples, all Method 6B samples (both SOa and COa) were collected
in duplicate using collocated sampling trains at both the inlet and outlet
of the spray dryer/fabric filter FGD system. A major QC function of the
duplicate, collocated sampling trains was to maximize data capture by pro-
viding a backup sampling system at both sampling locations. However, re-
sults for duplicate samples may be used to assess total (i.e., sampling
54
-------
plus analytical) measurement variability for both COa and S02 data, as well
as variability of SOa emission rates calculated using these data.
Results for SOa and COa concentration measurements and for SOa emission
rates calculated using these measurement data were statistically evaluated
using analysis of variance (ANOVA) techniques. This evaluation provides
information pertaining to the following aspects of the Method 6B data:
• Statistical significance of differences between
COa , and emission rate values for the paired sampling
trains (i.e., determination of relative bias between
Train A and Train B at both locations),
• Statistical significance of observed day-to-day
differences (i.e., temporal variability) in SOa, COa,
and emission rate values as compared to differences
between paired sampling trains,
• Magnitude of measurement variability (precision) for
the various components of variability in the SOa > COa ,
and emission rate determinations, and
• Relative contribution of each component of variability
(e.g., analytical, sampling, and temporal) to total
variability for the S02 , C02 , and emission rate
measurements.
Paired t-tests were used to evaluate Method 6B data for both inlet and
outlet locations to determine the statistical significance of observed
differences in results for the collocated sampling trains. This evaluation
indicated that, at the 5% significance level (i.e., 95% confidence level),
there was no significant difference (i.e., no relative bias) between S02 or
COa concentrations measured using Train A and those measured using Train B.
55
-------
Differences between Train A and Train B emission rates were not statistically
significant for the outlet data, but because of interaction between SOa and
C02 results, the differences were significant for the i.nlet data.
The comparison of results for the paired trains is summarized in Table
3-9. The statistical significance of emission rate differences at the inlet
may be attributed to a positive mean difference (i.e., Train A results tended
to be greater than Train B results) for S02 concentration accompanied by a
negative mean difference for COZ concentration. At the outlet, mean differ-
ences for both S02 and COa were negative (i.e., Train B gave, on the average,
higher results for both parameters). Although the relative bias between
Train A and Train B emission rates at the inlet is statistically significant,
the magnitude of this bias is relatively small, with the mean difference of
38.0 ng/J, representing only 1.9% of the mean measured inlet emission rate
(2009 ng/J).
Similar tests, again at the 5% significance level, were performed to
determine if observed day-to-day variation in measured values are statis-
tically significant. In other words, the data were evaluated to determine
whether day-to-day variations were "real" or were due to random variability
arising from sampling and analytical imprecision. In all cases (i.e., for
SOa, COa, and emission rate values at both inlet and outlet) temporal or
day-to-day variability was significantly greater than combined sampling and
analytical variability.
For the Method 6B data as a whole, there were two major components of
measurement data variability. One component was that due to day-to-day
variability in process operation, or temporal variability. The second com-
ponent was that due to random variability in sampling and analysis. Since
the SOa samples were analyzed in duplicate, sampling and analytical varia-
bility may be evaluated as separate components. Using ANOVA techniques,
both the magnitude and relative contribution of each component was deter-
mined for all three parameters (S02, COa, and emission rate) at both sampling
56
-------
TABLE 3-9. SUMMARY OF SIGNIFICANCE TEST DATA FOR PAIRED RESULTS
Ul
—I
Location
Inlet
Outlet
Parameter
S02
CO 2
Emission Rate
SO 2
C02
Emission Rate
Mean Difference
for Paired Trains
(Train A-Train B)
17.9 ppm
-0.07 %
38.0 ng/J
-11.3 ppm
-0.03 %
-14.5 ng/J
Standard
Deviation
70.0 ppm
0.58 %
72.0 ng/J
35.3 ppm
0.28 %
46.4 ng/J
Number of
Observations
23
23
23
20
22
20
Significant
(95% Confidence
Level)
No
No
Yes
No
No
No
-------
locations. Table 3-10 summarizes results for the component of variance
analysis. These data are presented in terms of coefficients of variation
for day-to-day variability, duplicate samples, and duplicate analyses
in Table 3-11.
3.6 METHOD 6B RELIABILITY AND DATA CAPTURE
Overall, the Method 6B sampling/analytical approach used during this
program proved to be a reliable means of collecting S02 removal efficiency
and emission rate data. The use of duplicate, collocated sampling systems
is credited with achieving an overall data capture of 100% (i.e., valid S02
emission rate and removal efficiency data were collected on 24 out of 24
sampling days). Problems did occur on occasion, resulting in loss or invali-
dation of S02 and/or C02 data from one or more of the four sampling trains.
However, in no case was data for both trains at either the inlet or outlet
locations lost or invalidated on the same day.
Instances of Method 6B data loss during this program may be attributed
to two causes:
Sampling system problems such as leaks and/or high pressure
drops which resulted in invalidation of results for the
affected sampling train, and
Operator error which resulted in loss of samples (and data).
Observed sampling problems were documented on the data collection sheet and
are summarized in the comment section of Table 2-1. Of the total of 96 sampl-
ing train days (i.e., four trains per day for 24 days, excluding the trial
run at the inlet on the first day), S02 and/or C02 data were lost or invali-
dated in five instances, for a sampling/analytical reliability of 94.8%.
Since both inlet and outlet measurement data are required to calculate
removal efficiency, method reliability with respect to removal efficiency
data must consider the relative frequency of data loss for either inlet or
58
-------
Ul
<£>
TABLE 3-10. SUMMARY OF ANALYSIS OF VARIANCE RESULTS
Parameter
S02
C02
Emission Rate
Source of
Variability
Temporal
Sampling
Analytical
Total
Temporal
Sampling/
Analytical
Total
Temporal
Sampling/
Analytical
Total
Inlet
Mean Measured Variance
Value Component8
14008.8
2489.7
17.5
1527 ppm 16516.0
0.199
0.164
9.64* 0.363
14425.3
3202.8
2009 ng/J 17628.2
Percent of
Total Variance
84.8
15.1
0.1
100.0
54.9
45.1
100.0
81.8
18.2
100.0
Outlet
Mean Measured Variance
Value Component
53326.0
650.2
4.5
310 ppm 53980.7
0.179
0.027
9.7Z 0.206
91748.4
1127.4
406 ng/J 92875.8
Percent of
Total Variance
98.8
1.2
<0.1
100.0
86.9
13.1
100.0
98.8
1.2
100.0
!
Variance components are equal to standard deviations squared and thus have units which are the squares of those used for the
corresponding measurement parameter (e.g., ppm2, etc.)
-------
1
TABLE 3-11. COEFFICIENTS OF VARIATION FOR REPEAT AND DUPLICATE MEASUREMENTS
Parameter
S02
CO 2
Emission Elate
Day-to-Day
Variability
(CV)
8.42Z
6.25Z
6.61Z
Inlet
Duplicate Sample
Precision
(CV)
3.28Z
4.20Z
2.82Z
Analytical
Precision
(CV)
0.27Z
—
Day-to-Day
Variability
(CV)
75. OZ
4.68Z
75. OZ
Outlet
Duplicate Sample
Precision
(CV)
8.26Z
0.37Z
8.27Z
Analytical
Precision
(CV)
0.68Z
—
-------
outlet trains of a given system. For system A (i.e., inlet Train A plus
outlet Train A), there were three days of invalidated data (August 4, 9, and
11), for a reliability of 87.5% (21 of 24 days). For System B, samples were
lost on August 3 and 9, for a reliability of 91.7%. This gives an average
system reliability of 89.6%.
Of the five instances of lost or invalidated data, three were due to
problems (e.g., leaks) with a sampling train, and two were due to operator
error during sample recovery. Thus, sampling reliability alone was 96.9%
(i.e., 93 of 96 sampling train days). It is also worth noting that all
instances of lost or invalidated data occurred within the first ten of the
24 sampling days, indicating that operator familiarity with the sampling
system and related procedures was probably a significant factor in method
reliability. It is probably also true that general familiarity with the
method must be combined with familiarity with, the specific sampling systems
used in order to maximize data capture.
61
-------
SECTION 4
PROCESS DESCRIPTION
The FGD system characterization performed by Radian at Argonne National
Laboratory was conducted on the Unit 5 spray dryer/baghouse system. This
section describes the system configuration and sampling locations used
during testing.
4.1 PLANT CONFIGURATION
The Argonne steam plant consists of five boilers that provide 200 psig
steam throughout the entire 1500 acre facility for heating and evaporative
cooling. Argonne1s main boiler (Unit 5) is a coal-fired Wickes (now Com-
bustion Engineering) spreader stoker unit designed to produce a maximum of
170,000 pounds of saturated steam per hour at 200 psig pressure. Control
equipment was required on Unit 5 to comply with State of Illinois SO2. and
particulate emissions standards (refer to Table 4-1) when high sulfur, mid-
western coal was burned in the unit. (A copy of Subpart D, as printed in the
Federal Register, is included in Appendix D of Volume II).
TABLE 4-1. STATE OF ILLINOIS EMISSION LIMITS FOR ANL
BOILER NO. 5
Pollutant Limit
Sulfur Dioxide 1.2 lb/106 BTU (520 ng/J)
Particulate Matter 0.1 lb/106 BTU and <20%
Opacity
62
-------
The system installed, on Argonne's No. 5 boiler to treat the flue gas
is a Niro Atomizer/Joy Ma.nufacturing industrial design. A simplified
schematic of the process is shown in Figure 4-1.
The system consists of two parts, a wet end and a dry end. In the wet
end, pebble lime is held in a 100 ton storage silo with a "live" cone bottom.
From this vessel, lime is fed through a Wallace and Tiernan weighbelt feeder
into the lime slaker. The weighbelt feeder is equipped with a feedrate indi-
cator as well as a totalizer which allows ANL to measure lime consumption.
In the slaker, careful addition of potable water causes the calcium oxide
(CaO) to react and form calcium hydroxide (Ca(OH)2), or milk of lime. The
milk of lime, at about 15% solids, is passed through a rotary screen in order
to remove the "grits," or inert particles, from the milk of lime. From the
slaker, the milk of lime is sent to a covered, agitated, storage tank. This
storage tank has a 30-minute hold time and serves two purposes: (1) to en-
sure completeness of the slaking reaction as well as to even out any incon-
sistencies in slaker operation, and (2) to ensure a temporary lime supply in
case of slaker system failure.
The milk of lime is next pumped to the slurry mix tank. In this agitated
vessel, recycled waste powder, milk of lime, and some dilution water are com-
bined to form an approximately 35-40 percent (by weight) slurry. The mix
ratio of recycled waste powder, milk of lime, and dilution water is con-
trolled to maintain the desired SOz emission rate (<520 ng/J) and outlet
spray dryer temperature (>150°F). The flow of milk of lime to the mix tank
is dependent upon the liquid level in the mix tank and the SOa concentration.
A decrease in the liquid level or an increase in the SOa concentration will
cause the milk of lime flow to increase. Recycle solids flow to the mix
tank is dependent upon the slurry feed density and the SOa concentration.
A decrease in the slurry feed density or a decrease in the SOa concentration
will cause the flow of recycle solids to increase. The dilution water flow
rate is only dependent upon the liquid level in the feed tank.
63
-------
SLURRY FEED OVERFLOW
SLURRY FEED
HEAD TANK
SPRAY DRYER
/~ ABSORBER
i
*T-I
BAGHOUSE
T IT.
STACK
SOLIDS
' DISPOSAL
70A3407
TREATED
WATER
MIX TANK SLURRY
FEED TANK
Figure 4-1. Schematic Diagram of the Argonne National Laboratory Unit 5 Boiler Spray
Dryer/Baghouse Flue Gas Desulfurization System
-------
From the mix tank,, the slurry is transferred to the slurry feed tank
via another rotary screen to ensure removal of any lumps which might clog
the feed slurry piping system. Overflow from the slurry feed tank goes back
into the slurry mix tank, so there is a continual circulation between the
tanks. Slurry from the feed tank is pumped, at a constant high flow rate,
to a head tank located above the atomizer. A control valve regulates the
amount of slurry fed to the atomizer, with the excess being returned to the
feed tank. This returned slurry also passes through the rotary screen that
is filtering the stream from the slurry mix tank.
Flue gas, exiting the boiler's induced draft (ID) fan, passes into a
modified breeching at the existing stack. A guillotine damper diverts the
flue gas flow into the FGC system ductwork leading to the spray dryer. This
inlet ductwork splits the flue gas into two streams. One stream, with about
60% of the gas flow, is directed into a roof gas disperser, located on the
top of the spray dryer. The remainder of the gas stream enters a central
gas disperser, located in. the middle of the spray dryer. Both gas streams,
upon entering the dryer, are given circular motions with their main direc-
tions of flow being opposed to each other. In the spray dryer, the slurry
droplets contact the hot flue gases where two events happen somewhat simul-
taneously: (1) the sulfur oxides react with the lime to form calcium sul-
fite and calcium sulfate, and (2) water associated with the lime evaporates,
thereby cooling the flue gas. The spray dryer is designed to control the
temperature of the gas exiting the spray dryer to 22°F (or more) above the
dew point. This temperature control is very important for several reasons
including: (1) achieving consistent SOa control; (2) the necessity of pro-
tecting the baghouse from condensation; (3) minimization of the lime stoichi-
ometry required for S02 removal; and (4) preventing the wetting of the walls
of the spray dryer.
Some of the powder formed in the spray dryer settles to the bottom and
is collected by a drag-line conveyer. The remainder of the powder, entrained
in the gas stream, enters the baghouse where it is removed by filtration.
Upon exiting the baghouse, the gas passes through a booster fan and then
65
-------
into the existing stack. Tables 4-2 and 4-3 list some of the parameters
related to the spray dryer and fabric filter.
TABLE 4-2. SPRAY DRYER PARAMETERS
• Niro Atomizer Incorporated
• 10 Second Residence Time
• 25' Diameter, 19' Straight Side
• Rotary Atomizer - 14,000 RPM
• Dual Gas Inlet - Roof and Central Gas Dispersers
r
• Carbon Steel Construction
TABLE 4-3. FABRIC FILTER PARAMETERS
• Joy Manufacturing Company
• 4 Compartment Pulse Jet
• 3.01:1 Air-to-Cloth Ratio
• 280 Bags/Compartment - 6" Diameter - 12' Long
• 16 Ounce Woven Fiberglass Fabric with Teflon® Coating
• 5278 Ft2 Filter Area/Compartment
66
-------
4.2 DESCRIPTION OF SAMPLING POINTS
Flue gas samples were collected at the inlet to the spray dryer and on
the stack downstream of the baghouse (refer to Point Qj and (2) in Figure
4-1). The location and orientation of the two sampling locations are illus-
trated graphically in Figure 4-2.
Flue gas from the ML Unit 5 boiler travels through a vertical duct to
the roof where the duct: splits and the flue gas can either exit through the
stack or enter the FGD system. During normal operation, the guillotine
valve (refer to Figure 4-2) is closed and the double louver dampers are
open, forcing the boiler flue gas through the FGD system.
The FGD inlet sampling location was between the guillotine valve and
the double louver damper and consisted of three, 3-inch, NPT, horizontally-
oriented ports located one above another. Flue gas samples were collected
using the top (Port A) an.d bottom (Port B) ports. During sampling at the
FGD inlet, the two sampling probes were situated approximately 30 inches
(mid-way) into the duct and the probe tips were approximately 36 inches
from each other (refer to Figure 4-3).
Upon exiting the FGD system, the flue gas passes through an induced
draft fan and is then vented to the atmosphere by means of a six-foot diameter
stack. The FGD outlet flue gas samples were collected using the sampling
ports on the stack sampling platform. Figure 4-4 illustrates the approxi-
mate orientation of the FGD outlet sampling ports and the two sampling probes.
Four, 3-inch, NPT ports were located on two perpendicular diameters. Two of
the four ports were used for sample collection. During sampling at the FGD
outlet, the probe used at: Port A was inserted approximately 18 inches into
the stack, while the probe used at Port B was inserted approximately 24
inches into the stack.
67
-------
STACK
a
s
PLATFORM
LADDER -
oo
INDUCED DRAFT
FAN HOUSING
FLUE GAS
FROM FGD-
SYSTEM
SAMPLING
PORTS
FGD OUTLET
1 SAMPLING LOCATION
GUILLOTINE VALVE
V~(NORMALLY CLOSED)
\
DOUBLE LOUVER DAMPER
(NORMALLY OPEN)
r^
t
^-
SAMPLING PORTS
FGD INLET SAMPLING/TN
LOCATION v!^
I
ROOF
70A3408
FLUE GAS FROM
UNIT #5 BOILER
FLUE GAS TO
FGD SYSTEM
Figure 4-2. Schematic Diagram Illustrating the Location of the Inlet and Outlet Flue
Gas Sampling Ports at the Argonne National Laboratory Unit 5 Spray Dryer
Baghouse Flue Gas Desulfurization System
-------
RADIAN
PORT A-
GLASS LINED
HEAT TRACED
SAMPLING PROBE
\
CAPPED!"?—
PORT [I]—
PORTB
J
• 60"-
~36"
60"
70A3409
Figure 4-3. Diagram Illustrating the Relative Location of the Two
Sampling Probes Used in Collecting the Duplicate EPA
Method 6B Samples at the Inlet to the Argonne National
Laboratory Unit 5 FGD System.
69
-------
RADIAN
3' PLATFORM
LADDER
GLASS LINED
HEAT TRACED
SAMPLING PROBE
70A3410
Figure 4-4. Diagram Illustrating the Location of the Two
Sampling Probes Used in Collecting the Duplicate
EPA Method 6B Samples at the Outlet of the
Argonne National Laboratory Unit 5 FGD System.
70
-------
SECTION 5
SAMPLING AND ANALYSIS
In order to determine the SOz removal efficiency of the FGD system at
the Argonne test site, SOa emission rates were measured at the inlet and
outlet of the system. This required measurement of the S02 content and C02
content of the flue gas, along with fuel analysis to derive C02 F-factors
for the emission rate calculation. Sampling and analytical procedures are
described below.
5.1 SAMPLING
The sampling procedure(s) used in the collection of flue gas samples
and coal samples are described in this section.
5.1.1 Flue Gas Sampling
During this program, the SC>2 and COa concentrations of the flue gas
were determined using EPA Method 6B (1). (A copy of EPA Method 6A and 6B
procedures are included in Appendix D of Volume II). The sampling system
is illustrated in Figure 5-1 and briefly described below.
A glass-lined, heat-traced probe was used to extract a gas sample from
the stack. An out-of-stack heated filter removed particulate from the gas
stream prior to entering the impingers. The probe and filter oven tempera-
tures were maintained at about 250°F. Gas exiting the probe entered a
series of three midget impingers. The first two impingers had tapered stems
and contained approximately 20 ml each of 6 percent HzOa for SOa removal.
The third impinger with a straight stem was filled with about 25 grams of
71
-------
TEMPERATURE
SENSOR
6% H2O2
IMPINGERS
DRIERITE
IMPINGER
STACK WALL
•=[
HEATED PROBE
NJ
FILTER
HOLDER
HEATED BOX
THERMOMETER
ASCARITE®
CO2 ABSORBER
PUMP
NEEDLE
VALVE
DRY GAS
METER
SURGE
TANK
ON/OFF ELECTRICAL
CONTACTS
SEQUENTIAL
TIMER
to
g
r-
Figure 5-1. EPA Method 6B S02 and C02 Sampling Train
-------
Drierite® to help prevent condensation in the sample line exiting the im-
pinger train.
Gas exiting the impingers entered a canister containing about 200
grams of Drierite® for final moisture removal. After final moisture re-
moval, the gas sample passed through a canister containing about 200 grams
of Ascarite® for C02 absorption. A pump and dry gas meter were used to
control and monitor the flow rate of the sample gas. The gas flow rate was
maintained at approximately 1 liter per minute.
A sequential industrial timer regulated the operation of the sample
pump during the 24-hour sampling period. The sample pump was on 2 minutes
every hour resulting in a total daily sampling time of approximately 48
minutes (48 liters of sample gas). The 24-hour sampling period started
at hour 1000 each day and ended at 0959 on the following day.
Prior to sampling, the impingers and Ascarite® canisters were weighed
and the weights recorded. All weighings were made within ±0.1 grams.
Filters were replaced in the heated filter holders every third day on the
inlet samplers and weekly on the outlet sampling trains. This schedule
was established after the first week of testing. The sampling system was
leak-checked before sampling and leaks with rates of greater than 0.02 liters
per minute were eliminated. Heating systems were operated continuously and
cold water was placed around the impingers as needed. All pertinent sampling
data (i.e., meter volumes., impinger weights, temperatures, etc.) were re-
corded on a standardized data form like the one illustrated in Figure 5-2.
During sampling, a tarpaulin cover protected the sampling system from
direct sunlight and adverse weather. The sampling system was visually
checked periodically during the day to ensure proper operation. Problems
encountered during sampling were noted on the data sheet and in the sampling
log notebook.
73
-------
Method 6B
FIELD SAMPLING DATA SHEET
Plant Na
Sampling Location:
Initial Leak Rate _
Meter ID ,
Run ID
Final Leak Rate
DGM Correction Factor:
Sampling Period: Start: Date
Stop: Date
Operator Initials Duration:
Time
Time
Hrs
Minutes
Net Sample Volume (i)
SAMPLING DATA
Final
Initial
Average
Dry Gas
Meter
Reading
Rotameter
Setting
Dry Gas Meter
Temperature
Inlet
Outlet
Barometric
Pressure
Probe
Temp •
"7
Flex
Connector
Temp .
Train
Outlet
Temp.
SAMPLE RECOVERY DATA
Impingers
and Drierlte
Ascarite
Column
Final Wt. (g)
Initial Wt. (g)
Moisture Wt. (g)
C02 Wt. (g)
Impinger Contents Sample ID
H20j Blank Sample ID _____
Sample Recovered By:
Remarks:
Date:
Figure 5-2. Method 6B Field Sampling Data Sheet
74
-------
5.1.2 Coal Sample Collection
Approximately once a month, coal is shipped by barge to ANL. The coal
is trucked from the barge to the ANL power plant for storage and use. While
the coal is still on the barge, a sample is collected by ANL personnel for
ultimate analysis and proximate analyses. No additional coal samples are
collected on a routine basis. Radian originally planned to obtain a portion
of each coal sample collected by ANL personnel and submit this fraction for
independent analysis.
From August 1 through August 23, 1983, the ANL Unit 5 boiler operated
using a subbituminous coal from Kentucky. On the morning of August 24, 1983,
the ANL Unit 5 boiler started using a subbituminous coal from Illinois. At
the end of the on-site sampling/analysis program, Radian tried to obtain a
sample of the Kentucky and Illinois coal from ANL personnel. However, the
two coal samples collected by ANL personnel were no longer available because
they had already been shipped to the subcontractor for analysis. A grab sam-
ple of the Illinois coal used on August 30, 1983 was subsequently collected
from the conveyer belt by Radian personnel. A sample of the original Kentucky
coal could not be obtained.
5.2 SAMPLE ANALYSIS
Flue gas and coal sample analytical procedures used during this program
are discussed briefly in this section.
5.2.1 Flue Gas Analysis
Test personnel performed a final system leak check after sampling. The
Ascarite® canister was weighed to determine the mass C02 collected. The
contents of the first two impingers were quantitatively transferred to a
100 ml volumetric flask, diluted to volume with distilled water, and
75
-------
analyzed for S02 (as SO^) by means of the barium-thorin titration procedure
outlined in EPA Method 6 (2), using barium chloride (BaCl2) as the titrant.
(A copy of the EPA Method 6 procedure is included in Appendix D of Volume II).
The SC>2 and CC>2 concentrations obtained from the analyses were used to cal-
culate the emission rates in ng/J. All emission rate calculations followed
the standardized calculation data forms (refer to Figure 5-3) located on the
back of each EPA Method 6B data sheet.
5.2.2 Coal Analysis
Commercial Testing and Engineering, Inc. (CT&E), located near Chicago,
Illinois, performed ultimate and proximate analysis of the Kentucky and
Illinois coals used during this program, as well as a quality assurance audit
sample. The Kentucky coal sample was collected and sent by ANL personnel to
CT&E for analysis. The Illinois coal sample was collected by Radian personnel
and shipped along with the audit coal sample to CT&E for analysis. The
ultimate and proximate coal analysis results are presented in Table 2-3. The
CC>2 F-factor (F ) for each coal sample was derived from the ultimate and
proximate analysis (dry basis) of the coal.
76
-------
A. Standard Meter Volume, V . ,., Mm
mistd;
_
tn(std) 1000 I
m(avg)
Pbar ""8
°F + 460"R X 29.92 "Hg
528"R
m(std)
528
1000
460
B. CO2 Volume Collected, Standard Conditions, Km3
x DGMCF
29.92
VC02Utd) ' 5"467
vC02(std) '""" * A
C. CO2 Concentration, Z
(M
af
c . COi(std)
00* Vm(std) + VCO»(std)
x 100
"CO 2
x 100 •
D. SO2 Concentration, mg/Nm3
CSQ2 - 32.03 -e- Cb N
CSQ - 32.03
(To convert mg S02/dscm to ppm, multiply by 3.762 x 10 )
E. Moisture Volume Collected, Urn5
m(std)
K VC02(std)
) x x ( / )
+
mg/Sm1
VH20(8td) ' 1'336 X 10" (Mwf - Mwi>
VH20(std) • 1'336 * i0 <
F. Moisture Concentration, Z
H20(atd)
Vm(std) * VC02(std) * VH20(std)
CH20 "
C. SO2 Emission Rate
/ 100 \
ES02 ' CS02 Fc 1CC02 jx 10
so,
Remarks:
xlO6-
_ng/J
Figure 5-3. Method 6B Calculations Worksheet
77
-------
SECTION 6
PROCESS MONITORING PROCEDURES
During this program, Radian collected pertinent boiler and FGD process
data to allow for the evaluation of the system performance and to help
determine the critical operating parameters and the system economics. Table
6-1 shows the parameters monitored during this program.
A Radian DART II data acquisition system provided the means for on-site,
continuous collection of process data. The DART was directly connected
with the main Unit 5 boiler and FGD control panels to obtain the instrument
output signals of interest. The DART generated five minute averages from
instantaneous signals from the monitors. The DART determined hourly averages
and 24-hour averages for each of the parameters of interest from the 5-minute
averages. The DART 24-hour averages coincided with the EPA Method 6B 24-hour
sampling time period. The five-minute averages, hourly averages, and 24-hour
averages were stored on a floppy disk, and the average values were also
printed on site using an on-line printer. The DART used acronyms to represent
each of the parameters of interest and their respective units of concentra-
tion. The DART acronyms are also included in Table 6-1.
Argonne personnel manually recorded boiler and FGD process data once an
hour using prepared process log sheets. The manually recorded process data
proved useful because not all of the pertinent process data were continually
available from the DART. Certain process data were not available to the DART
because of incompatible or nonexistent electronic signal or because the
monitor signal was not available on the main control panel. Also, the DART
malfunctioned several times during the program resulting in no recorded pro-
cess data for short periods of time. Whenever the DART could not provide
78
-------
process data, the Argonne process log sheets provided the data to calculate
a 24-hour average.
TABLE 6-1. SUMMARY OF ANL UNIT 5 BOILER AND FGD PROCESS PARAMETERS
OF INTEREST (INCLUDING DART ACRONYMS WHEN APPLICABLE)
Parameter
Unit 5 Boiler Load
Unit 5 Boiler Exit Gas
02 Concentration
Baghouse Outlet Temperature
Spray Dryer Inlet Temperature
Spray Dryer Outlet Temperature
Spray Dryer Slurry Feed Rate
Spray Dryer Inlet Pressures
Baghouse Pressure Drop
Lime Milk Flow
Central Gas Disperser Pressure Drop
Baghouse Outlet Dew Point Temperature
Atomizer Motor Amps
Lime Mix Tank Level
Slurry Density
Slaking Dilution Water Flow
Lime Milk Flow
Lime Milk Density
Slurry Dilution Water Flow
Recycle Feed Rate
Contraves Outlet C02
Contraves Outlet S02
Contraves S02 Emission Rate
Slurry Mix Tank Level
Units
10 3 Ibs steam/hour
%
°F
»F
°F
gpm
inches H20
inches H20
gpm
inches H20
°F
amps
%
grams /cc
gpm
gpm
grams /cc
gpm
Ibs/hr
%
ppm
lbs/106Btu
%
DART Acronym
BLOAD/PPH
*
BHOUT/DEGF
SDIN/DEGF
SDOUT/DEGF
SFR/TGPM
SDINP/TI H20
BHDP/TI H20
LTMFR/TFPM
RGDIS/TI H20
*
*
LMTL/%
SFSG
*
*
*
DH20/TGPM
*
C02/T%
SO 2/ PPM
S02/LB BTU
SMTL/%
*Data collected from operator log sheet.
**Data not available.
79
-------
SECTION 7
EXAMPLE CALCULATIONS
Included in this section are examples of calculations performed during
this program.
Standard Meter Volume, V . ..
m(std)
• Dry Standard Cubic Meter (dscm)
/ Actual Meter \ / Barometric W Dry Gas Meter \
\Volume (liters)/ (528°R)\Pressure ("Hg)/\Correction Factor/
= (»F) + "0'R)(29.92"Hg)
Example: Based on ARG-0825-M6B-0/B data,
v = (53.0 liters)(528"R)(29.52"Hg)(1.02)
m(std) (1000) (84°F + 460°R)(29.92"Hg)
V . .. = 0.0518 m3
m(std)
a Volume Collected, Standard Conditions, V . ,.
Dry Standard Cubic Meter (m3)
('
Final mass Initial mass of
of ascarite - ascarite before
after sampling (g) sampling (g)
Example: Based on ARG-0825-M6B-0/B data,
VC02(std) - 5'467 x 1
VC02(std) = 0-0059 m3
C02 Concentration, C Q
• Percent (%)
= CO; Volume Collected, Standard Conditions (m3) x 100
C02 Standard Meter Volume (m3) + CO Volume Collected, Standard Conditions (m3)
Example: Based on ARG-0825-M6B-0/B data,
(0.0059 m3)(100)
C02 = 0.0518 m3 + 0.0059 m3"
Cco2 = 10'2%
80
-------
S02 Concentration, C_,_
S02
Milligrams per Dry Standard Cubic Meter (mg/sm3)
/Volume _ Volume of WNormality of \ /Total Volume Sample (ml)\
. , ._. \Titrant (ml) ~ Blank (ml)/\ Titrant / \ volume of Aliquots (ml) /
L__. s \j£..\)j) — — — ^^—
2 Standard Meter Volume (m3) + C02 Collected at Standard Conditions (ra3)
Example: Based on ARGi-0825-M6B-0/B data,
/T. n•>^ (24.47 ml - 0.0 ml)(.01Q2 N)
C
/100 ml\
S02 /- .0518 m-* + .0059
Ccrv = 1555 mg/sra3
S02
Parts per Million (ppti)
To convert mg S02/smJ to ppm, multiply by 3.762 x 10 million sm mg S02
Example: Based on ARG-0825-M6B-0/B data,
C_Q = 1555 mg S02/sm:l x 3.762 x ID"1 million sm /mg S02
CS02 = 585 ppm
Carbon Dioxide F-Factor, F
• Standard Cubic Feet per Million BTU (scf/106 BTU)
321 x 103 (%C)
c Gross Calorific Value
Example: Based on the results of the analysis of the Illinois coal
presented in Table 6-3, ,
= (321 x 103)(74.1)
c 13,303 BTU/pound
F - 1788 scf/106 BTU
Standard Cubic Meters per Joule (sm /J)
To convert scf/106 BTU to sm3/j, multiply by 2.686 x 10"5
Example: Based on the results of the analysis of the Illinois coal
presented in Table 6-3,
F = (1788 scf/106 BTU) (2. 686 x 10~5 S " " ^
C \ SCr-J
F = A. 804 x 10~8 sm3/J
81
-------
S02 Emission Rate, E
• Nanogram per Joule (ng/J)
/ 100 \
£„ = (S02 Concentration (mg/sm3)) (C02 F-factor (sm3/J)) \ C02 Concentration (%) /
do6)
Example: Based on ARG-0825-M6B-0/B data,
_8 ( 100 \ / 6x
E - (1555 mg/sm3) (4.804 x 10 sm3/J) \10.2tti (10 )
E,n = 730 ng/J
OU2
• Pounds per Million BTU (lbs/10 BTU) lh T
.-3
To convert ng/J to lbs/10 BTU, multiply.by 2.32 x 10 ng_106 BTU
Example: Based on ARG-0825-M6B-0/B data,
E,,- = (730 ng/J)(2.32 x 10"3 6 __..
SU2 \ ng—1U B1U
Ecn = 1.69 lbs/106 BTU
SO 2
S02 Control Efficiency, CEC-
bO2 ""
• Percent (%)
f Average Inlet S02 Average Outlet S02 "I
Emission Rate (ng/J) ~ Emission Rate (ng/J) x
Qfl a
2 Average Inlet S02 Emission Rate (ng/J)
Example: Based on ARG-0825-M6B-0/B data,
rF ([2060 ng/J) - [734 ng/J])100
. ^ S02 2060 ng/J
CE = 64.4%
82
-------
REFERENCES
1. EPA Method 6B—"Determination of Sulfur Dioxide and Carbon Dioxide
Daily Average Emission from Fossil Fuel Combustion Sources," 40 CFR
60, Appendix A, Environment Reporter, Bureau of National Affairs,
Inc., Washington, D..C., March 25, 1983.
2. EPA Method 6—"Determination of Sulfur Dioxide Emissions from
Stationary Sources," 40 CFR 60, Appendix A, Environment Reporter,
Bureau of National Affairs, Inc., Washington, D.C., March 25, 1983.
3. U.S. Environmental Protection Agency, "Quality Assurance Handbook for
Air Pollution Measurements Systems, Volume I, Principles," EPA 600/9-
76-005, Research Triangle Park, NC, January 1976.
4. Lewis, D. L. and L. A. Rohlack, "Industrial Boiler Continuous Emission
Monitoring at the Argonne National Laboratories Test Site—Quality
Assurance Project Plan," EPA Contract No. 68-02-3542, Work Assignemnt
10, Radian Corporation, Austin, Texas, July 1983.
5. Cochran, William G. and Gertrude M. Cox, Experimental Designs, 2nd
Edition, John Wiley & Sons, Inc., New York, 1957.
R-l
------- |