ir 1972 Environmental Protection
Validation of Improved
Chemical Methods
for Sulfur Oxides Measurements
from Stationary Sources
of Research
i,l, ETOiiiMSffliffltsS
Washington. D.IL
-------
EPA-R2-72-105
Validation
of Improved Chemical Methods
for Sulfur Oxides Measurements
from Stationary Sources
By
J. Driscoll, J. Becker,
R. Hebert, K. Horbal and M. Young
Wai den Research Corporation
359 Allston Street
Cambridge, Massachusetts
Contract No. 68-02-0009
Program Element No. All010
Project Officer: Fredric C. Jaye
Division of Chemistry and Physics
National Environmental Research Center
Research Triangle Park, North Carolina 27711
Prepared for
OFFICE OF RESEARCH AND MONITORING
U.S. ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
November 1972
-------
This report has been reviewed by the Environmental Protection
Agency and approved for publication. Approval does not signify that
the contents necessarily reflect the views and policies of the Agency,
nor does mention of trade names or commercial products constitute
endorsement or recommendation for use.
ii
-------
ACKNOWLEDGMENTS
We are pleased to acknowledge the assistance of F. Jaye, R. Larkin,
and R. Statnick of EPA for their help in the planning phase of this pro-
gram. We are indebted to Drs. Miller, Scott and Thompson also of EPA
for performing X-ray fluorescence and emission spectrographic analyses
of selected field samples. We thank the many organizations which pro-
vided the test sites for the SOX validation program.
WALDEN RESEARCH CORPORATION
-------
PROJECT STAFF
Major technical contributions to this report have been made by the
following:
J. Driscoll - Program Manager
J. Becker
J. Ehrenfeld
P. Giever
P. Gravalese
R. Hebert
K. Horbal
R. Maddock
L. Paley
M. Young
1 V WALDEN RESEARCH CORPORATION
-------
SECTION 1
FOSSIL FUEL COMBUSTION SOURCES
WALDEN RESEARCH CORPORATION
-------
TABLE OF CONTENTS
Section Title Page
1.1 INTRODUCTION 1-1
1.2 BRIEF DESCRIPTION OF SOX EMISSIONS FROM FOSSIL FUEL
COMBUSTION 1-4
1.3 SAMPLING SYSTEMS AND ANALYTICAL METHODS 1-10
1.3.1 Probe 1-10
1.3.2 Sampling Trains 1-14
1.3.3 Calibration of Metering Devices and Variation
with Time 1-18
1.3.4 Analytical Methods 1-18
1.4 S02 TEST RESULTS FOR A CONTROLLED COAL-FIRED POWER
PLANT 1-25
1.5 S02 RESULTS FOR AN OIL-FIRED POWER PLANT 1-34
1.6 S02 RESULTS FOR AN UNCONTROLLED COAL-FIRED POWER PLANT .. 1-41
1.7 SOs RESULTS FROM FOSSIL FUEL COMBUSTION SOURCES 1-58
1.8 S02 RESULTS OBTAINED IN THE WALDEN PILOT PLANT AND
LABORATORY 1-61
1.8.1 Description of Pilot Plant 1-61
1.8.2 Supplementary Laboratory and Pilot Plant Studies . 1-61
1.8.3 Effect of NOX on S02 Methods and Accuracy
Studies 1-71
1.8.4 Effect of HC1 on the Accuracy of S02 Methods 1-82
1.8.5 Effect of Organic Acids on S02 Methods 1-93
1.9 SUMMARY AND CONCLUSIONS 1-99
LIST OF ILLUSTRATIONS
Figure Caption Page
1-1 Actual Test Schedule 1-3
1-2 Schematic of Stationary Combustion Systems 1-7
1-3 Probe Module 1-12
1-4 SOX-NOX Sampling Probe and Condenser 1-13
vii
WALDEN RESEARCH CORPORATION
-------
LIST OF ILLUSTRATIONS (continued)
Figure Caption Page
1-5 Sampling Trains 1-15
1-6 Controlled Condensation Collector for $03 1-16
1-7 Maiden Prototype Sampling System 1-17
1-8 Flow Chart of Analytical Procedures 1-21
1-9 Effect of Cation Exchange Resin on the Removal of
Sul fate 1-24
1-10 Sampling Port Locations for Controlled Coal-Fired Power
Plant 1-26
1-11 Schematic of the Oil-Fired Power Plant 1-35
1-12 Detail of Test Ports in an Oil-Fired Power Plant 1-36
1-13 Uncontrolled Coal-Fired Power Plant 1-47
1-14 Sampling Port Locations for the Uncontrolled Coal-Fired
Power PI ant 1-48
1-15 Walden Pilot Plant 1-62
1-16 Pilot Plant Test Section Assembly 1-63
1-17 Pilot Plant Schematic 1-64
1-18 Linearity of the Calibration Curve for the Barium
Chloranilate Method 1-68
1-19 Effect of Flushing Time on the S02 in 80% IPA 1-70
1-20 Equilibrium Distribution of Chlorine Compounds in the
Combustion Gas of a Sewall Coal at 20 Percent Excess
Ai r 1-88
1 -21 Doping System for Organic Acids 1-95
LIST OF TABLES
Table Title Page
1-1 Nationwide Emissions of SOX, 1966 1-5
1-2 Nationwide Emissions of S02 from Fossil Fuel Combustion
Sources (1967) 1-6
1-3 Size Categories for Stationary Combustion Sources 1-8
1-4 Emission Parameters for Fossil Fuel Combustion 1-9
1-5 SOX Control Systems 1-11
WALDEN RESEARCH CORPORATION
-------
LIST OF TABLES (continued)
Table Title Page
1-6 Comparison of Rated and Measured Flow Rates for Some
Mi 11i pore Cri ti cal Ori fi ces 1-19
1-7 Variation of Orifice Calibration with Sampling Time in
the Field 1-20
1-8 Effect of Ion Exchange Step on the Analytical Methods ... 1-23
1-9 S02 Test Results from a Controlled Coal-Fired Power Plant
at the Inlet of a Wet Scrubber 1-27
1-10 Fuel Analysis for Controlled Coal-Fired Power Plant 1-28
1-11 S02 Test Results from a Controlled Coal-Fired Power Plant
at the Outlet of a Wet Scrubber 1-30
1-12 Precision of Duplicate Analysis of S02 for a Controlled
Coal-Fired Power Plant 1-32
1-13 Regression Analyses and Correlation Coefficients for a
Controlled Coal-Fired Power Plant 1-33
1-14 S02 Test Results from an Oil-Fired Power Plant 1-37
1-15 Precision of Duplicate S02 Analysis for an Oil-Fired
Power PI ant 1-39
1-16 Regression Analyses and Correlation Coefficients for an
Oil-Fired Power Plant 1-40
1-17 Fuel Analysis for an Oil-Fired Power Plant 1-42
1-18 Supplementary£ield Test Data for an Oil-Fired Power
Plant 1-43
1-19 Comparison of Measured and Calculated S02 Values for an
Oil-Fired Power Plant, Barium Chioranilate Method 1-44
1-20 Comparison of Measured and Calculated S02 Values for an
Oil-Fired Power Plant, Ba++ Titration Method 1-45
1-21 Comparison of Measured and Calculated S02 Values for an
Oil-Fired Power Plant, NaOH Titration Method 1-46
1-22 S02 Test Results from an Uncontrolled Coal-Fired Power
Plant 1-50
1-23 Precision of Duplicate S02 Analysis for an Uncontrolled
Coal-Fired Power Plant 1-51
1-24 Regression Analyses and Correlation Coefficients for an
Uncontrolled Coal-Fired Power Plant 1-52
1-25 Supplementary Field Test Data for an Uncontrolled Coal-
Fi red Power PI ant 1-53
WALDEN RESEARCH CORPORATION
-------
LIST OF TABLES (continued)
Table Title Page
1-26 Fuel Analyses for an Uncontrolled Coal-Fired Power
Plant 1-54
1-27 Comparison of Measured and Calculated SC"2 Values from an
Uncontrolled Coal-Fired Power Plant, Barium Chloranilate
Method 1-55
1-28 Comparison of Measured and Calculated SO? Values from an
Uncontrolled Coal-Fired Power Plant, Ba+* Titration
Method 1-56
1-29 Comparison of Measured and Calculated S02 Values from an
Uncontrolled Coal-Fired Power Plant, NaOH Titration
Method 1-57
1-30 S03 Test Results for Fossil Fuel Combustion 1-60
1-31 Pilot Plant Specifications 1-65
1-32 Characteristics of the Pilot Plant on #2 Oil and Gas 1-66
1-33 Effect of Flushing Time on the S02 Concentration in the
80% Isopropanol-20% Water Absorber 1-69
1-34 Effect of NOX on Wet Chemical Methods for S02 1-72
1-35 Determination of Collection Efficiency of S02 in Aqueous
Hydrogen Peroxide Solution 1-75
1-36 Pilot Plant Comparison of Theoretical S02 versus ppm S02
Measured by Wet Chemi stry 1 -77
1-37 Pilot Plant Results - Precision with NOX Doping 1-83
1-38 Regression Equations and Correlation Coefficients for a
Gas-Fired Pilot Plant 1-84
1-39 Chlorine Content of Various Coals 1-86
1-40 Combustion of Pulverized Coals 1-87
1-41 Comparison of Measured and Calculated HC1 Concentrations
in the Pilot Plant 1-90
1-42 Effect of HC1 on Wet Chemical Methods for S02 1-91
1-43 Pilot Plant Doping - Precision of HC1 and HOAc Doping ... 1-94
1-44 Effect of Organic Acids on Wet Chemical Methods for S02 . 1-96
WALDEN RESEARCH CORPORATION
-------
1.1 INTRODUCTION
This program was initiated in January 1971. The principal objective
was to validate the wet chemical methods for sulfur oxides which were de-
veloped under EPA Contract No. CPA 22-69-95. A brief description of the
methods is given below:
The barium chloranilate procedure contains the following major recom-
mendations:
(a) SO- is separated from SOp, water, and other interferences by con-
trolled condensation of the flue gas.
(b) S0_ and S02 analyses are performed colorimetrically.
(c) Sampling rates are controlled by critical orifice.
The barium ion-thorin procedure has been used as a standard for com-
parison. This has been described in detail by EPA (1971) and is summarized
in Appendix I. The principal differences in this procedure are that SO, is
collected by absorption in 80% isopropanol-20% water and analysis is, of
course, conducted by titration.
A simplified method, which was not specific for sulfate, was also used
to analyze aliquots of the S02 samples. This method involves titration of
the sulfuric acid with standard sodium hydroxide solution using bromphenol
blue indicator. This procedure is also described in Appendix I.
The reader is referred to the Final Report under EPA Contract No.
CPA 22-69-95, "Improved Chemical Methods for Sampling and Analysis ,"
Volume I - Sulfur Oxides, APTD #1106, PB 209-268, for background material
on the chemistry as well as the collection and analysis methods for sulfur
oxides.
The first six months of the program were initially planned to evaluate
(determine accuracy, interferences, etc.) the collection and analysis
methods in the pilot plant and define an experimental plan for the field
testing program. During the last six months of the program, the revised
WALDEN RESEARCH CORPORATION
-------
methods would then be tested at uncontrolled coal, uncontrolled oil, and
controlled coal-fired power plants.
At the end of February 1972, a meeting was held at EPA in Durham, North
Carolina. Representatives of Walden Research Corp., the EPA Project Officer,
and the EPA source testing group were present. It was decided that the pro-
gram should be modified to enable the data collected in the field to be
utilized in preparing the national emission standards for sulfur oxides.
As a result of the new schedule, major modifications to the program
involved:
(a) Interchanging the position of the pilot plant and field programs -
the field program would now be completed before the pilot plant studies.
(b) Reducing the elapsed time in the field from six months to less
than two. This would be done by overlapping test programs at the various
sites.
(c) Presentation of the test data including statistical analysis by
the last week in April 1971.
The field testing was completed and the report was presented by the pre-
scribed time. Some difficulties were encountered as a result of the
speeded-up program. These will be discussed in the text.
Following the field program, the pilot plant studies were initiated.
Before these were completed,, a request was made by EPA to extend the scope
of the SO validation program to include several Category II and other sta-
/\
tionary sources. These were:
(a) gray iron foundries
(b) kraft pulping
(c) iron and steel
(d) smelting
(e) sulfuric acid
A detailed schedule of the actual resulting expanded test program is given
in Figure 1-1.
]_2 WALDEN RESEARCH CORPORATION
-------
8
m
v>
o
i
s
S
o
FEB MAR APR
MAY
1971
JUNE JULY
AUG SEPT OCT NOV DEC
JAN
1972
FEB MAR
APR
i
j PREPARATION
Or tOUl PM t N T !
i j
CONTROLLED
COAL
!
i
1 OIL
UNCONTROLLED
I COAL
i
j
PILOT PLANT
[ TEbT ING
PREPARATION
FOR NEW SET
TESTS
GRAY IRON
FOUNDRY
KRAFT MILL
SMELTER
Cu
Pb
STEEL MILL
-SO 4
i 2
».
-» |
— >,
i
^
-- • - -
I
-5, -ijw-*sj
fc
— ^
!
f
L * — -
->
i
-^
r
:
i
-+
H
|
1
i
i
t
h
Figure 1-1. Actual Test Schedule.
-------
The results obtained in fossil fuel-fired power plants and in the
Walden pilot plant are given in Section 1. The results for the other (non-
combustion) stationary sources are discussed in Section 2. The appendix
contains a detailed write-up of the analytical methods used, the Pb titra-
tion method developed under this contract, and data obtained on solid sor-
bents for collection of SO .
/\
For each source which was tested, a brief description of the process is
given to enable the reader to understand the sampling and/or analytical con-
figurations used. This section is followed by a description of the sampling
and analytical procedures, the test data including statistical analysis, and
conclusions and recommendations for each source.
1.2 BRIEF DESCRIPTION OF S0x EMISSIONS FROM FOSSIL FUEL COMBUSTION*
A rough guide to method development priorities may be obtained from the
relative contributions of various sources to national SO emissions (Table
1-1).
Since SO emissions are primarily fuel dependent rather than equipment
/\
dependent, the data of Table 1-1 essentially integrate the product of fuel
use and do not distinguish fossil fuel effluents from process emissions.
A somewhat more detailed breakdown, which excludes (direct-fired) proc-
ess emissions, is available from work conducted at Walden under EPA contract
(Ehrenfeld, et al., 1971) and is given in Table 1-2.
As may be seen from the Boiler Category of Table 1-2, 90% of total
(non-process) emissions are attributable to watertube boilers, for which the
smallest size range is about 25,000 pounds of steam per hour (or ca. 3 x 10
Btu/hr). The S02 emissions determination problem is centered in these rela-
tively large sources. Further examination of the categories of Tables 1-1
and 1-2 reveals that the highest priority is clearly to be placed upon
This section was taken from Driscoll and Berger (1971).
_. WALDEN RESEARCH CORPORATION
-------
TABLE 1-1
NATIONWIDE EMISSIONS OF SOY, 1966
/\
10 tons/year
Total from Fossil Fuel Combustion in Stationary
Sources 22.1
By Fuel Coal 18.3
Residual Fuel Oil 3.3
Distillate Fuel Oil 0.5
By Sector Power Generation3 14.0
Industrial 5.6
Residential and Commercial Heating 2.5
aAbout 90% of the power generation contribution to SOX emission is derived
from coal-fired units and about 10% from oil.
WALDEN RESEARCH CORPORATION
I *"D
-------
TABLE 1-2
NATIONWIDE EMISSIONS OF S02 FROM FOSSIL FUEL
COMBUSTION SOURCES (1967)
10 tons/year
Total from Fossil Fuel Combustion in Stationary
Sources
By Fuel Coal
Residual Fuel Oil
Distillate Fuel Oil
By Sector Utilities
Industrial
Commercial
Residential
By Boiler Category (excluding residential)
Watertube ^500,000 pph
Watertube <_ 500,000 pph
Firetube
Cast Iron
18.63
15.60
2.76
0.27
12.58
4.28
1.53
0.24
9.29
6.50
0.88
0.72
1-6
WALDEN RESEARCH CORPORATION
-------
analytical methods and equipment applicable to large boilers since they rep-
resent the largest sources and will involve the first application of control
techniques.
It does not appear desirable to develop specially adapted high preci-
sion methods applicable to the myriad and relatively unimportant small
sources at this time.
A prerequisite for the selection of manual chemical methods for
sampling and analysis of gaseous pollutants from the combustion of fossil
fuels is the description of the environments to be sampled. These will be
discussed in the following paragraphs.
All combustion equipment can be generalized in terms of the schematic
shown in Figure 1-2. Fuel is burned in a furnace to release heat. Combus-
tion products and ash are generated as undesirable by-products. The com-
bustion heat is transferred in a heat exchanger to some convenient thermal
fluid, usually water, steam or air. Domestic heaters are familiar, simple
examples of this generalized system. They are designed to provide unat-
tended service over long periods of time to untrained users. Consequently,
they are characterized by a low level of sophistication in combustion con-
trol, heat transfer (thermal efficiency) or air pollution control. On the
Useful Heat
t
Fuel fr
Furnace
Kir -fc
I
Ash
1
1
1 Heat
Exchanger
1
t
Cool Thermal
Fluid
^ Combustion
Products
Figure 1-2. Schematic of Stationary Combustion Systems.
1-7
WALDEN RESEARCH CORPORATION
-------
other end of the sophistication spectrum are new steam generating units at
large power plants. While the generalized combustion system shown in Fig-
ure 1-2 is still applicable, it is embellished considerably by combustion
air preheaters, economizers, high pressure heat exchangers and pollution
control equipment. In addition, computerized systems are utilized to moni-
tor and control furnace and heat exchanger operation. The combustion prod-
ucts may be modified by the use of additives, either in the fuel, or added
in the furnace or heat exchangers.
For this evaluation, we have divided combustion equipment into three
arbitrary size groups, as shown in Table 1-3, according to several approxi-
mately equivalent criteria. There are relatively few suppliers of equipment
in the large size range (four major sources) with 1968 annual sales of the
order of 75 units. On the other hand, there are several hundred suppliers
of small units with 1967 annual shipments of approximately 1,500,000 units.
Useful life is perhaps 15 and 30 years, respectively, for the small and
large units. Combining the large number of installed units with the dis-
parate designs, range of fuels used, load factors, varying operating prac-
tices, etc., it is clear that it is totally impractical to attempt to
specify the environment for individual cases. Rather, we have reviewed
the range of environments reported in the literature for each of the three
size categories shown in Table 1-3 for oil, coal and gas. A summary of
emission parameters is given in Table 1-4 in consistent units (#/10 Btu).
TABLE 1-3
SIZE CATEGORIES FOR STATIONARY COMBUSTION SOURCES
Btu per hour
Boiler horsepower
Pounds of steam
per hour
Megawatts
Large
>_ 5 x 108
>_ 15,000
^500,000
>_ 50
Intermediate
3xl05-5xl08
10-15,000
350-500,000
< 50 MW
Small
< 3 x 105
< 10
< 330
Not Applicable
WALDEN RESEARCH CORPORATION
-------
TABLE 1-4
EMISSION PARAMETERS* FOR FOSSIL FUEL COMBUSTION3 (pounds/106 Btu)
Velocity
ft/mi n
Mn i c "fi i v»o
Excess Air
Stack Temp.
°F (of
Breeching)
CO
so2
so3
NO
N02
Parti cu-
late
Hydrocarbons
Transients
Large
2000-
3000
15
275-
400
Neg.
4x1 O"4
0.35
0.014
Neg.
10%
GASb
Intermed.
mO <\V X^ Air
Vr U/o AO M 1 r
1 c Aj Ml
15-75
400-
750
4x1 O"4
4x1 O"4
0.2
0.016
Neg.
of Capacity Per
Small
75
750-
900
4x1 O"4
4x1 O"4
0.1
0.017
Neg.
Min.
Large
(Resid.
2000-
3000
20
300-
400
3x1 O"4
1.65
0.03
0.68
0.07
0.022
10%
OILC
Intermed.
) (Resid. /Dist.
10^ 13 n# Y*\ A-iv*
Q1^ f3 9n°^ YC A-J v*
J /O \r C.\J /o AJ r\ 1 I
20-75
400-
750
0.01/0.01
1.65/0.31
0.016/3xlO"3
0.47/0.51
0.15/0.10
0.01/0.01
of Capacity Per
Small
) (Dist.)
75
750-
900
.01
0.31
3x1 O"3
.08
0.06
.02
Min.
Large
2000-
3000
25
300-
400
0.02
3.5
0.8
0.46
(90% cont.
8x1 O"3
5%
COALd
Intermed.
25-75
400-
750
0.1-2
3.5
0.8-0.3
0. 53
) (65%' cont.)
0.04-0.4
Capacity Per Min
Small
75-100
750-900
2
3.5
0.3
0.8
(uncont.)
0.4
•
I
vo
5
TO
O
I
O
O
TO
s
TO
O
z
a. Details in Appendix 1, Final Report, Vol. I
b. 1100 Btu/SCF
From Driscoll and Berger (June 1971).
c. Sdist. 0.39%, 142,000 Btu/gal
Sresid. 1.6%, 152,000 Btu/gal
d. S = 2.5%, 8% ash, 13,000 Btu/#
-------
Details are given in Appendix I from Driscoll and Berger (1971). A summary
of conditions before and after control equipment for the most prominent SO
/\
control systems is given in Table 1-5. From these data, the various re-
quirements for principal factors such as required sensitivity, possible in-
terferences, transients and temperature regimes may be determined. The
major factors are briefly described below.
Sampling equipment must be designed to tolerate "normal" operation as
well as unusual temperature excursions. In general, large units are de-
signed for stack temperatures of 300-400°F. Gas-fired unit emissions, as a
result of their freedom from the effects of corrosive sulfuric acid mist,
run cooler, about 275°F. In domestic units, stack temperatures are much
higher, ca. 700-800°F. Inlet temperatures of the Monsanto SO control proc-
A
ess also run at this level. Intermediate combustion units cover the range
of temperatures between these extremes (400-700°F).
The implications of this temperature variation for sampling are three-
fold:
(1) Materials of construction must be adequate to tolerate the highest
temperature.
(2) Materials must be chosen to eliminate high temperature catalytic
oxidation of S02 to SO, in the sample probe and lines.
(3) Water and acid condensation must be taken into consideration in
design of the sampling train. The sample stream should be kept hot insofar
as possible, up to the collection equipment.
1.3 SAMPLING SYSTEMS AND ANALYTICAL METHODS
1.3.1 Probe
The probe, Figure 1-3, consists of an electrically heated pyrex
insert with a stainless steel jacket. A detailed drawing of the probe and
sampling system is shown in Figure 1-4. The probe could be heated to a
temperature of about 330°F with 120 VAC. A loosely-packed glass-wool plug
1-10
WALDEN RESEARCH CORPORATION
-------
TABLE 1-5
SOX CONTROL SYSTEMS
BEFORE
i
WALDEN RES
TO
a
o
0
33
Developer Process
CE Wet Scrub-
bing (Dol-
omi te )
Monsanto CAT-OX
Bumines Alkalized
Alumina
Reinluft Char.
Absorption
Well man- Alkali-
Lord Sulfite
Absorption
TVA Dry Dolomite
Absorption
(Intermit-
tent)
*
Temp.°F
250-300
850
625
250-300
300
Additive
Rate of
2 Times
Stoichi-
ometric
Require-
ment
Input Temp.°F
Normal 250-300
Double Load
Particulates
Follows 99% 250
+ Elect.
Precip.
Recommend ., 250
0.9 gr/ftj
Recommend ^ 250
0.9 gr/ff3
Only as 250
Req'd by
Power Plant
Equipment
2-3 times
Parti cul ate
Load in Flue
Gas Stream
and Bottom
Ash
AFTER
Output
99% Parti -
cul ate Re-
moval
--
Follows Cy-
clone Sep-
arator
—
+95% Parti -
cul ate Re-
moval by
Special
Scrubber
*
S0x
S03-99% Re-
moved
S02-90% Re-
moved
(Same)
(Same)
(Same)
(Same)
25-50% S09
99% S03 £
s
33
Design efficiencies assumed
-------
FILTER
I
ro
I
o
m
ASBESTOS
INSULATION
HEATING
WIRE
A_
/~
f\
/ \y
' V
f\
' \
v
f\
1
\ t
\s
f\
1 \
4 ^
/
s\
\,
V
/
<\ /\]
V 1
v ~i
1 1 mm DIA
PYREX
|NCH
STAINLESS
TUBE
1/4 DIAMETER
STAINLESS STEEL PROBE
FOR NOviCO. AND 0->
i <-
QUICK TEE
STACK ADAPTER
PYREX SOCKET
JOINT
O
I
8
a
S
TO
Figure 1-3. Probe Module.
-------
I
co
-. ASBESTOS COVERED ._.
(1) i- HEATING WIRE © /~l/4 N°x PROBE
IT ///// /// / / / / / / / ////// I
ASBESTOS PAPER TAPE
•PYREX sox PBOBS
HARD SOLDER
BRASS 3/4
PIPE
-STAINLESS STEEL
PROBE COVER
PROBE- STACK
CONNECTOR
©
3/4" ELEC
' 4" NIPPLE
•BRASS CLAMP
•ELECTRICAL
CONNECTOR'LT
THREADED
SWITCH
r~u
f@
is
CONNECTOR
.CASE
COPPER COIL WITH
ASBESTOS COVERED
HEATING WIRE
PYREX CONDENSER
•3)
m
O o
(id^-
ALUMINUM CHANNEL
(CONDENSER -IMPINGES SUPPORT)
TO
a
o
o
TO
s
JO
Figure 1-4. SO -NO Sampling Probe and Condenser.
x x
z
-------
was used to remove participate matter. The stainless steel probe attached
to the SO probe (Figure 1-4) was used to collect manual NO samples as
X X
well as C02 and 02 for Orsat analysis. When the three trains (described in
the following section and shown in Figure 1-5) were run in parallel, a glass
tee wrapped with heating tape (controlled with a variac) was used to couple
the trains to the probe. The samples from all three trains were collected
simultaneously through the same probe in order to obtain comparisons of the
three sampling methods. See Appendix A (Figure A-5).
1.3.2 Sampling Trains
The sampling procedures are described in detail in Appendix A.
Schematics of the sampling trains used for the tests are shown in Figure 1-5.
The trains are similar in that S0« is collected in midget impingers by ab-
sorption in aqueous hydrogen peroxide solution. The major difference be-
tween the trains was in the collection of SO.,. In train a, SO, was col-
lected by filtration of the HpSO. mist after the flue gas was cooled to a
temperature between 60 and 90°C. This temperature is below the acid dew-
point and above the water dewpoint. The condenser is shown in Figure 1-6
and in detail in Figure 1-4. This controlled condensation collection pro-
cedure should lead to increased precision and/or accuracy of S03 determina-
tions since the problem of oxidation of dissolved S0« is eliminated. In
train b, SO- was collected in midget impingers by absorption in a mixture
of 80% isopropanol (IPA) and 20% water. Any residual SO- which remains in
the IPA (train b) impingers after collection of a sample of flue gas was re-
moved by flushing with an equal volume of clean air. In train c, S0~ was
collected (and not determined) by absorption in a bubbler containing 80%
isopropanol and 20% water. Any HUSO, not removed in the isopropanol solu-
tion was collected on a glass wool plug-foilowing the bubbler. Train c was
not flushed with air since we wanted to determine the extent of loss of dis-
solved SO- in the IPA impinger. The SOp results for this train were ex-
pected to be lower than trains a or b.
The metering systems for trains a and b are critical orifices
while train c utilizes a dry test (integrating) meter. The a and b meter-
ing system is shown pictorially in Figure 1-7. In the first two trains,
1-14 WALDEN RESEARCH CORPORATION
-------
S03 CONDENSER
CRITICAL ORIFICE
DRIERITE TEMP
•I.MPINGERS
WITH f
FOR S02 COLLECTION
CA RBON
VANE
PUMP
NOTE: NO FLUSHING WAS NECESSARY
FILTER
(b)
"4-
•i— •
1
— g —
CE BAT
H
^__
*\
^
*
>
OR
CO
CR I TICAL ORIFICE
IMPINGE'RS FILLED
WITH 80*A I PA FOR
S0 COLLECTION
IMPINGERS
FOR
FILLED WITH
HO
S02 COLLECTION
CARBON
VANE
PUM P
NOTE: THE I PA IMPINGERS
WERE FLUSHED WITH
AN EQUAL VOLUME OF
CLEAN AIR TO REMOVE
RESIDUAL S02
GLASS WOOL PLUG
DRIERITE
COLUMN
MMM
^
i!
f\
K
-«
uwq
I
f
CE BAT
H
'^
•-^
\
t
^
^ nov
r
TEMP
DRY
TEST
METER
DIAPHRAGM
PUMP NOTE,
FRITTED BUBBLER
FILLED WITH 807.
IPA
IMPINGERS FILLED
WITH H202 FOR COLLECTION
Figure 1-5. Sampling Trains.
NO FLUSHING OF
IMPINGERS WAS
CONDUCTED TO
DETERMINE THE
EXTENT OF LOSS
OF DISSOLVED SOp
MN I PA IMPINGERS
1-15
WALDEN RESEARCH CORPORATION
-------
en
I
o
m
z
m
in
'
Figure 1-6. Controlled Condensation Collector for SO,
-------
I
CO
m
O
I
O
O
TO
s
TO
O
Figure 1-7. Wai den Prototype Sampling System.
-------
the time has to be accurately measured since the total volume of air metered
is determined from the product of the sampling rate and the time. In order
to eliminate any differences between the trains caused by the metering sys-
tems, all metering systems were checked with a spirometer initially and at
frequent intervals (every two to four weeks) with a calibrated dry test
meter. The results are discussed in the following section.
1.3.3 Calibration of Metering Devices and Variation with Time
The only "primary standards" for metering flow are the wet test
meter and the spirometer. These devices are laboratory instruments and are
not readily applicable for field use. Therefore, dry test meters and other
portable metering equipment are used for field applications. These devices,
.however, must be calibrated since they are not "primary standards." We have
calibrated the metering devices, used for the field tests, with a spirometer.
A "new" dry test meter calibrated with the spirometer was kept in the labo-
ratory for calibration purposes. This device was used to check the flow
rates of all the orifices and dry test meters on a regular basis. The use
of the secondary standard (dry test meter) was necessary because of the
Targe number of calibrations required and the awkwardness of the spirometer.
»i
Thirty-four different Millipore critical orifices were checked with the dry
test meter. Each result represents an average of three to five individual
runs. The data are given in Table 1-6. Although many of these orifices
are close to the rated values, errors as large as 50%, 15% and 10% were »
found for 0.5, 1.0 and 3.0 liter/min orifices. Note that smaller errors
are found for the larger orifices. Thusf these orifices must be calibrated
prior to use. A number of orifices were checked at regular intervals dur-
ing the test program. The data are given in Table 1-7. No changes in
calibration were observed although these orifices were used to collect as
many as 20 to 50 field samples between calibrations.
1.3.4 Analytical Methods
The analytical methods used in this study include titration with
with Ba ion (thorin indicator), or NaOH (bromphenol blue indicator) and
colorimetric determination via barium chloranilate. The methods are de-
scribed in detail in Appendix A and are briefly summarized in Figure 1-8.
1_-|g WALDEN RESEARCH CORPORATION
-------
TABLE 1-6
COMPARISON OF RATED AND MEASURED FLOW RATES
FOR SOME MILLIPORE CRITICAL ORIFICES
Rated Value (1/min)
Measured Value (1/min)
0.5
1.0
0.54
0.54
0.27
0.39
0.50
0.48
0.45
0.40
0.46
0.35
0.95
1.01
0.97
1.01
0.85
1.09
0.93
1.04
0.98
1.04
0.98
1.01
0.90
1.03
3.0
3,
2,
3,
3,
2,
2,
2.
3,
2.
08
97
11
19
84
66
79
02
84
2.54
1-19
WALDEN RESEARCH CORPORATION
-------
TABLE 1-7
VARIATION OF ORIFICE CALIBRATION WITH SAMPLING
TIME IN THE FIELD*
Initial Flow
Rate (1/min)
2.86
2.66
2.86
2.78
2.94
3.04
2.96
2.86
2.90
0.52
*
The majority
gram, e.g. ,
Recalibrated Flow
Rate (1/min)
2.72
2.63
2.92
2.81
2.96
3.07
2.96
2.85
2.88
0.49
of this data was taken in the second
at Category II sources (see following
Total Sample
Volume (1)
1800
1800
3800
3800
2700
2700
3000
3000
3000
3000
phase of the pro-
Section 2).
WALDEN RESEARCH CORPORATION
-------
SAM PLING TRAIN *
and ANALYTICAL METHOD
(a)*
Ba CHLORANILATE
NaOH
I
rsj
NaOH TITRATION
Ba TITRATION
Ba CHLORANILATE
(c)*
BcTT ITRATION
NaOH TITRATION
z SAMPLI NG RATE
S l//mm
5 SAMPLING TIMES
SAMPLE
DILUTE
- 50ML
TO
SAMPLE
•DILUTE TO
50ML
SAMPLE-
DILUTE TO
250 ML
15-30 m i n
8
TO
I
5
O
PASS THROUGH
EXCHANGE
COLUMN
ADD
REAGENTS
SHAKE
-£>FOR —C>CENTRIFUGE-
20 MIN
-£> TITRATE WITH.02N
NaOH USING BROM-
PHENOL BLUE- INDICATOR
MEASURE
ABSORBANCE'
e. 53Onm
Tl TRATE WITH .02N
NaOH USING BROM-
PHENOL BLUE INDICATOR
ASS THROUGH ^ADJUST
1 ON EXCHANGE ^ PH
COLUMN
SAME AS TRAIN (a)
TITRATE WITH Ba (C I04)2
"USING THORIN INDICATOR
TITRATE WITH .02 N NaOH
"USING BROMPH'ENOL BLUE
INDICATOR
TITRATE WITH
Ba(Ci04)2
USING THORIN
METHYLENE BLUE
INDICATOR
Figure 1-8. Flow Chart of Analytical Procedures.
-------
Note in Figure 1-8 the samples for trains a and b were passed
through a cation (Dowex resin) column to remove interference from metals
present in the particulate matter. After the field test phase of the pro-
gram was completed at fossil fuel-burning installations, we found poor
agreement between measured and calculated results. The measured results
were considerably lower than expected. Ba Chlor and Ba titration gave
lower results on trains a and b than NaOH titration whereas in train c
equivalent SCL results were obtained with Ba and NaOH titration. NaOH
titration gave nearly the same SOp concentrations on all three trains.
Since the major difference between the methods involved the treatment with
the ion exchange resin, we began to suspect the recovery of sulfate from
the resin. Therefore, we ran a comparison of a number of samples with and
without the ion exchange step. The results are given in Table 1-8. The
reduction in sulfate varies from 10 to 60% by either the barium ion titra-
tion or the barium chloranilate methods. Since the percent reduction varied
widely with concentration, we ran some statistical analyses on the data to
determine whether there was an effect of percent removal on concentration.
The correlation coefficient was found to be 0.98 and the regression equa-
tion determined was y = 168.2 + 1.08X (where X = with cation exchange). A
plot of the data is shown in Figure 1-9.
Apparently the peroxide in the samples had some adverse effect
on the cation exchange column. The peroxide causes the ion exchange column
to swell. No swelling was encountered with synthetic H2SO. samples which
contained no peroxide. Unfortunately, all the a and b train samples by
Ba and Ba Chlor were run using the cation exchange column. Therefore, we
applied a correction to the measured data for these two methods in trains a
and b using the regression equation of Figure 1-9. All the a and b train
i i
data for S00 only by Ba and Ba Chlor on controlled coal, oil and uncon-
6
trolled coal have been corrected for the effect of the ion exchange resin.
Any pilot plant data and samples in other industries were not affected
since this problem was corrected prior to taking any samples there.
A detailed write-up of all the procedures is given in Appendix A.
Note that the ion exchange step is not included.
WALDEN RESEARCH CORPORATION
-------
TABLE 1-8
EFFECT OF ION EXCHANGE STEP ON THE ANALYTICAL METHODS
ppm SO,
With Cation Exchange
Without Cation Exchange
Barium Chi orani late Analysis
52
259
407
646
850
940
1137
1229
Barium Ion Titration
260
256
150
193
160
131
218
237
539
843
834
1128
1138
1480
1390
519
326
314
344
270
296
308
*
Run on aliquots of the same sample
1-23
WALDEN RESEARCH CORPORATION
-------
ro
I
m
TO
8
TO
s
•x
1500
q, I20°
cr>
c
•0
JL.
x 1000
c
o
+-
D
u
OJ
O
£
Q_
a
500
i I
- 155.7+0.925 x
CORRELATION COEFF=0.98
• Ba Chlor
500 1000 1500
ppm SOp without cation exchange
Figure 1-9. Effect of Cation Exchange Resin on the Removal of Sulfate.
-------
1.4 S00 TEST RESULTS FOR A CONTROLLED COAL-FIRED POWER PLANT
—.£
The test results described in this section were obtained at a coal-
fired power plant equipped with a Combustion Engineering wet scrubbing sys-
tem. The sampling port dimensions are given in Figure 1-10. This plant
has a very high particulate loading at the scrubber inlet because of the
injection of limestone for SO control. The resulting fly ash is basic
A
and very reactive. The SO^ test results obtained at the inlet to the
scrubber are shown in Table 1-9. The SOp concentrations obtained in runs
5-10 were much lower than expected on the basis of the fuel analysis (Table
1-10). The calculated S02 concentration from the fuel analysis, assuming
no conversion to CaSO., is about 2000 ppm. The measured values are only a
fraction of this.
In later talks with Combustion Engineering, we found that they had en-
countered similar sampling problems. They were able to eliminate low SO,,
results by sampling at a low rate (0.25-0.5 liter/min) to minimize par-
ticulate buildup. Since we were running three SO trains off a single
A
probe, our sampling rates (3-4 liters/min) were nearly an order of magni-
tude higher than those recommended by Combustion Engineering. The reaction
of CaC03 with SO,, could take place either in the probe (glass wool plug) or
in the impingers. In the latter case, the NaOH titration would give much
lower results than the barium ion titration because of the formation of
Ca(OHL and HpC03 (by hydrolysis of CaC03) and subsequent reaction of
Ca(OH)» with HpSO.. The barium chloranilate method would be unusable be-
cause of the complete precipitation of the acid chloranilate ion by Ca ion,
a cationic interference. Neither of these effects seems to occur, there-
fore, the losses must be the result of reaction of SOp on the glass wool
plug in the probe. Some recent experimental work by Galeano (1972) showed
that a buildup of Na2CO- on a glass wool "plug" in an SO sampling train
reduced the recovery of SOp. This tends to support our conclusions for
this gas stream heavily laden with CaCO~.
Some of the discrepancies between the trains (b and c) may be due to
collection of some fly ash in the impingers. With train a (condenser), a
, pc WALDEN RESEARCH CORPORATION
-------
>
4"
\
^ 3' ^ 5' ^i
i — | r
1
U 5' ^
-] r
2
I i
5' 3' u
~i i i
3 4
I i
NORTH SOUTH
- .._. ?l ' «j
Preheater
t
30"
JL
Probe
Scrubber
a. Inlet to Wet Scrubbing System.
3ort A
b. Wet Scrubber Outlet.
Figure 1-10. Sampling Port Locations for Controlled Coal-Fired
Power Plant.
1-26
WALDEN RESEARCH CORPORATION
-------
TABLE 1-9
S02 TEST RESULTS FROM A CONTROLLED COAL-FIRED
POWER PLANT AT THE INLET OF A WET SCRUBBER
I
ro
Sample No.
5A*
5B*
6A
6B
7A
7B
8A
8B
9A
9B
10A
10B
*
A and B are
Trai
ppm S02
(NaOH)
157
170
13
12
134
125
132
131
34
35
n a
ppm S02
(Ba Chlor)
290
294
184
184
234
256
253
255
192
180
Train
ppm S02
(NaOH)
291
289
1216
1214
305
303
126
120
120
118
101
94
b
ppm SO?
(Ba++)
408
410
1094
1047
459
463
276
271
271
262
238
292
Train
ppm S02
(NaOH)
438
418
272
268
287
288
150
159
137
137
85
82
c
ppm S02
(Ba++)
428
413
313
271
332
287
148
151
149
135
103
99
aliquots of the same S02 sample.
s
t—
o
m
73
m
in
O
I
o
o
33
s
XI
-------
TABLE 1-10
FUEL ANALYSIS FOR CONTROLLED
COAL-FIRED POWER PLANT
Fuel Component
Coal S 3.72
3.61
76.39
76.72
4.36
4.54
0.93
1.08
1-28 WALDEN RESEARCH CORPORATION
-------
mechanical problem was encountered. The clamps attached to the condenser
to seal this device to the impingers did not apply enough pressure to ob-
tain an adequate seal. The condensers were modified in the field to elimi-
nate this problem, and the runs after 10 show better agreement with the
other trains.
Runs 11-20 (Table 1-11) were conducted at the outlet of the wet
scrubber which was being run at about 50% S0? removal. The expected S0?
concentrations at the scrubber outlet were ^ 1000 ppm. The measured S0?
values are in agreement with the stoichiometric concentrations. A major
portion of the particulate (about 90%) is removed by the Combustion Engi-
neering wet scrubbing system, thus, the problem encountered at the inlet
would not be expected here.
No statistical results were calculated for the data in Table 1-9
(inlet).
A measure of the precision of an analytical method may be obtained by
calculating the standard deviation from replicate measurements. The S0?
determinations in Table 1-11 were run in duplicate. Since the mean is
varying, the standard deviation must be calculated by the following
equation:
ss = E [(¥„ - Y.)2 + (Y2. - Y.)2]
where ss is within runs sum of squares
t is the number of runs
Y,. and Y?. are the results of the duplicate analyses for the ith
run
Y . + Y
Y. = ]1 2 2i is the mean of the ith run
Since the mean is not constant and is calculated for each run, the
total number of degrees of freedom in this sum of squares is equal to the
number of runs. Hence, the variance within runs is:
,_2g WALDEN RESEARCH CORPORATION
-------
I
CO
o
TABLE 1-11
S02 TEST RESULTS FROM A CONTROLLED COAL-FIRED POWER
PLANT AT THE OUTLET OF A WET SCRUBBER
O
m
TO
m
in
m
o
i
o
o
TO
s
Sample No.
11A*
11 B*
12A
12B
13A
13B
14A
14B
15A
15B
16A
16B
17A
17B
ISA
18B
19A
19B
20A
20B
*
A and B are
Train a
ppm S02 ppm S02
(NaOH) (Ba Chlor)
936
906
471
475
528
527
685
697
555
557
1054
1063
aliquots of
863
944
565
597
577
631
740
759
653
658
1004
1021
the same S02
ppm S02
(NaOH)
562
551
602
589
612
598
750
729
706
688
737
708
566
546
505
504
449
444
441
445
sample.
Train b
ppm S02
606
592
596
617
637
639
755
746
636
638
729
729
632
618
529
531
502
515
489
484
ppm S02
(Ba Chlor)
604
612
629
636
787
786
692
725
828
804
707
566
469
466
504
563
503
501
Train
ppm S02
(NaOH)
528
530
514
528
548
526
684
677
667
655
490
501
535
533
518
505
354
367
435
449
c
ppm S02
616
609
571
553
608
592
741
735
694
676
542
517
562
626
563
544
411
404
494
497
o
z
-------
variance
•V?
This value provides an unbiased estimate of the standard deviation and
hence is a measure of the precision of the analytical method. The data cal-
culated using the above equation is summarized in Table 1-12. The barium
ion and NaOH titration procedures yield a precision of 1-2% which is typi-
cal at these SCL concentrations when no interferences are present. The pre-
cision of the barium chloranilate method is quite poor, about 5%, indicating
the presence of an interference with the method. McCrone (1971) found that
the particulate matter from this scrubber (for samples collected at the same
time) was predominantly fine particles of CaSO.. Since calcium is known to
be an interference with the barium chloranilate method, this may account for
the poorer precision of the method for this source.
Regression analysis and correlation coefficients were .computed with a
Compucorp 145E programmable calculator. The statistical intercomparison of
the methods is given in Table 1-13. There appears to be better agreement
between different methods on the scone train than between the same analyti-
cal method on trains a, b or c. For example, Ba titration for train c vs
b yields a poor correlation coefficient (0.67) whereas the NaOH titration
vs Ba in either trains b or c gives correlation coefficients of 0.94 and
0.97, respectively. The S02 results for train c are lower than for either
trains a or b. This is due to the lack of flushing to remove residual SOp
from the 80% IPA absorber. The SO must be flushed from the IPA absorber
following the test. The mean S02 concentration by NaOH titration for runs
11-16 is more than 100 ppm lower for train c than for either train a or b.
We also see that the NaOH titration for train b vs c gives a lower cor-
relation coefficient and poor regression equation, again the result of
residual S02 left in the IPA scrubber for train c.
XA
The Ba titration vs NaOH or Ba Chlor on train b yields essentially
similar statistical results. Since methods specific for sulfate or hydro-
gen ion give the same results within any particular train, the differences
between the trains may be the result of sampling rather than the analytical
methods.
WALDEN RESEARCH CORPORATION
1-31
-------
I
OJ
ro
TO
m
(ft
m
33
O
I
O
O
73
-a
O
TABLE 1-12
PRECISION OF DUPLICATE ANALYSIS* OF S02 FOR A
CONTROLLED COAL-FIRED POWER PLANT
Analytical Method
Ba Chi orani late
Ba++ Titration
NaOH Titration
Sampling Train No.
(a)
(b)
(b)
(c)
(a)
(b)
(c)
*
Sampling errors not included (replicate
(a) S03 collected
(b) S03 collected
(c) S03 collected
of Samples
11
9
16
10
4
6
6
analysis)
by condenser
in impinger containing 80% I PA 1
in a bubbler containing 80% IPA (
Mean S02 Concentration
(ppm)
515
632
553
409
561
653
525
In all cases S0« is col
Standard
Deviation (ppm)
23.2
37.4
6.3
6.8
4.5
13.2
9.1
lected in \\fl2
CV (%)
4.5
5.9
1.1
1.7
0.8
2.0
1.7
O
-------
TABLE 1-13
REGRESSION ANALYSES AND CORRELATION COEFFICIENTS FOR
A CONTROLLED COAL-FIRED POWER PLANT*
Methods Compared
Ba Titration (c) [x] and
Ba++ Titration (b) [y]
Ba++ Titration (c) [x] and
NaOH Titration (c) [y]
Ba Titration (b) [x] and
Ba Chi orani late (b) [y]
Ba++ Titration (b) [x] and
NaOH Titration (b) [y]
Ba Chloranilate (b) [x] and
Ba Chloranilate (a) [y]
NaOH Titration (c) [x] and
Sample
Size
20
20
18
20
10
20
Correlation
Coefficient
0.669
0.972
0.935
0.947
0.871
0.784
Regression
Equation
y =
y =
y =
y =
y =
y =
0.619 x +253
0.964 x -29.8
1.27 x -143
1.17 x -128
1.67 x -464
0.903 x +110
NaOH Titration (b) [y]
Outlet only
(a) Sampling train
(b) Sampling train
(c) Sampling train
condenser - two impingers
four impingers
bubbler - two impingers
1-33
WALDEN RESEARCH CORPORATION
-------
For train a, we have a rather limited number of results for comparison
and for train c, the incomplete removal of S02 from the IPA gives low SCL
results. Hence, the apparent poor agreement between trains.
1.5 S02 RESULTS FOR AN OIL-FIRED POWER PLANT
A detailed layout of the 30 MW oil-fired power plant is shown in Fig-
ure 1-11. The sampling port schematic which shows the probe position in the
duct following the electrostatic precipitator is given in Figure 1-12. The
test results obtained at the oil-fired power plant are given in Table 1-14.
These samples were collected with the sampling trains shown in Figure 1-5.
The samples 29a and 29b in Table 1-14 were collected at the same time in
two parallel probes. We found in later tests that the b system had a
cracked probe. This accounted for the lower results obtained with probe b.
The precision of the analytical methods is excellent as seen in Table
1-15. The values for the coefficients of variation range from about 1.5 to
3%. The precision of titration methods seems to be slightly better than
the barium chloranilate colorimetric method, however, all methods are in
the CV range expected when interferences are minimal.
The statistical analysis of the data from Table 1-14 is tabulated in
Table 1-16. The agreement between the methods is considerably better with-
in a train [Ba titration (b) and barium chloranilate (b), Ba titration
(c) and barium chloranilate (c), Ba titration (c) and NaOH titration (c)]
than for comparing the same methods on different trains [Ba Chi or (a) and
Ba Chi or (b), Ba titration (b) and Ba titration (c)]. The one excep-
tion to this is the NaOH titration (b and c). Excellent agreement is ob-
tained since the NaOH titration aliquots were not run through an ion ex-
change column. The sulfate methods on trains a and b but not c were run
through an ion exchange column. As noted in Section 1.3, the peroxide
causes the resin to swell and take up some sulfate. The correction (see
Figure 1-9) was applied to the sulfate methods (Ba and Ba Chlor) on
trains a and b. Since the same correction was applied to the sulfate
methods on train b, good agreement is obtained for Ba (b) and Ba Chlor
(b).
, ,. WALDEN RESEARCH CORPORATION
-------
Furnace
Electrostatic
Precipitator
co
in
I
o
m
TO
m
Fuel
Induced Draft
Fan
o
o
o
TO
3
TO
5
z
Figure 1-11. Schematic of the Oil-Fired Power Plant.
-------
Port B
~ I
Port A
Probe
30"
62"
24"
56"
ID Fan
Figure 1-12. Detail of Test Ports in an Oil-Fired Power Plant.
1-36
WALDEN RESEARCH CORPORATION
-------
TABLE T-14
S02 TEST RESULTS FROM AN OIL-FIRED POWER PLANT
CO
o
m
Z'
33
m
C/J
O
O
TO
-o
O
•33
Sample No.
lAt
IBt
3A
3B
4A
4B
21A
21B
22A
22B
23A
23B
24A
24B
25A
25B
27A
27B
28A
28B
29a-A
29a-B
29b-A
29b-B
30a-A
30a-B
30b-A
30b-B
31a-A
31a-B
Train a
ppm S02 ppm
(NaOH) (Ba
318
323
328
325
325
328
340
341
335
344
303
303
203
215
S02
Chlor)
431
431
455
448
439
450
426
451
368
378
343
351
312
312
ppm S02
(NaOH)
251
256
352
348
316
311
369
354
367
359
364
369
374
377
219
210
356
354
212
205
361
345
173
184
283
277
154
158
348
346
Train b
ppm S02
(Ba++)
432
434
545
550
506
505
557
556
566
567
555
566
563
558
321
324
439
465
339
339
443
435
341
351
361
362
307
311
432
432
ppm S02
(Ba Chlor)
396
401
523
523
498
469
614
592
672
625
569
573
564
568
308
315
452
451
--
338
431
438
324
336
375
370
314
307
429
429
ppm S02
(NaOH)
259
259
62
63
313
319
370
367
373
374
352
347
364
369
360
359
378
405
200
207
337
304
120
112
295
253
164
178
273
277
Train c
ppm S02
(Ba++)
269
261
61
60
309
314
369
368
364
358
381
383
357
357
338
345
377
369
211
199
319
314
114
101
245
255
128
120
252
254
ppm S02
(Ba Chlor)
225
385
80
125
255
315
O
-------
TABLE 1-14 (continued)
I
GO
00
I
0
m
C/5
m
TO
o
o
O
TO
~o
O
TO
Sample No.
31b-A
31b-B
32a-A
32a-B
32b-A
32b-B
33a-A
33a-B
34A
34B
35A
35B
51A
51B
52A
52B
53A
53B
54A
54B
55A
55B
56A
56B
*
These samples
Train a
ppm S02 ppm
(NaOH) (Ba
209
211
311
323
280
281
320
327
264
276
275
280
311
303
289
288
295
292
S02
Chlor)
309
312
363
338
356
332
403
394
383
369
350
371
378
375
341
344
368
383
are titrated with barium
ppm S02
(NaOH)
220
204
306
309
169
164
376
380
295
298
304
316
335
332
319
320
314
318
Train b
ppm SO?
(Ba++)
336
336
410
403
327
339
445
449
390
392
391
394
412
410
307
313
401
407
perch! orate using
ppm S02
(Ba Chlor)
322
325
393
396
396
451
446
393
372
377
395
393
397
271
378
ppm S02
(NaOH)
143
154
298
296
106
95
373
400
211
213
350
357
301
299
284
300
302
304
the thorin-methylene blue
Train c
ppm S02 ppm
(Ba++) (Ba
130
137
276
282
85
83
346
345
214
217
355
354
295
295
279
280
287
285
mixed indicator
S02
++)
352*
368*
233*
222*
356*
362*
293*
299*
287*
281*
284*
288*
,
TA and B are aliquots of the same S02 sample.
-------
C/5
m
TO
o
I
8
TO
s
TO
5
z
TABLE 1-15
PRECISION OF DUPLICATE S02 ANALYSIS* FOR AN
OIL-FIRED POWER PLANT
CO
I
o
m
Analytical Method
Ba Chi orani late
Ba Titration
NaOH Titration
Sampling Train No. of
(a)
(b)
(b)
(c)
(a)
(b)
Samples
19
16
18
11
7
15
(c) 12
Mean S02 Concentration
(ppm)
367
447
444
316
257
282
274
Standard
Deviation (ppm)
10.7
11 .1
5.9
6.0
5.8
6.3
5.1
CV (%)
2.9
2.5
1.4
1.9
2.3
2.2
1.9
Sampling errors not included (replicate analysis)
(a) S03 collected
(b) S03 collected
(c) S03 collected
by condenser
in impingers containing
in a bubbler containing
80% I PA I
80% IPA j
SOp collected in H^Op in
all trains
-------
TABLE 1-16
REGRESSION ANALYSES AND CORRELATION COEFFICIENTS FOR
AN OIL-FIRED POWER PLANT
Methods Compared
Ba Titration (b) [x] and
Ba++ Titration (c) [y]
Ba Titration (b) [x] and
Ba Chloranilate (b) [y]
Ba Titration (b) [x] and
NaOH Titration (b) [y]
Ba Titration (c) [x] and
Ba Chloranilate (c) [y]
Ba Titration (c) [x] and
NaOH Titration (c) [y]
Ba Chloranilate (a) [x] and
NaOH Titration (a) [y]
Ba Chloranilate (a) [x] and
Ba Chloranilate (b) [y]
NaOH Titration (b) [x] and
Sample
Size
28
30
30
6
30
14
14
28
Correlation
Coefficient
0.
0.
0.
0.
0.
0.
0.
0.
769
966
866
834
981
752
846
930
y
y
y
y
y
y
y
y
Regression
Equation
= 0
= 1
= 0
= 0
= 0
= 0
= 2
= 1
.745 x
.12 x
.702 x
.799 x
.949 x
.623 x
.15 x
.03 x
-44
-51.
-15
+61
+22
+60
-361
-18.
.7
8
.9
.2
.4
.4
2
NaOH Titration (c) [y]
(a) Sampling train - condenser - two impingers
(b) Sampling train - four impingers
(c) Sampling train - bubbler - two impingers
1-40
WALDEN RESEARCH CORPORATION
-------
With the NaOH titration (b) and (a) and barium titration (b) and barium
chloranilate (a), lower correlations are obtained due to the correction of
the sulfate methods. On the train c values which do not have to be corrected
for the ion exchange absorption, good agreement is obtained for barium titra-
tion and NaOH titration.
The data indicate that although some differences are observed between
the individual trains, all the analytical methods give reliable results.
The data on fuel composition and flue gas analysis are given in Tables
1-17 and 1-18, respectively, and were used to calculate the stoichiometric
SOp values.
A comparison of the calculated and measured S02 values are given in
Tables 1-19 through 1-21 for the barium chloranilate, barium ion titration
and NaOH titration methods, respectively.
In Table 1-19, the train a results are 14% lower (on the average) than
train b results. The fact that values > 100% are measured is probably due
to the correction for sulfate adsorbed on the ion exchange resin. The large
intercept for the data in Figure 1-9 tends to overemphasize values in the
low concentration range.
Note that in Table 1-20, the train b results are > 100% of theoretical.
This, again, may be due to the correction of the results (as above). Train
c barium titration yields considerably lower results than train b, however,
these % theoretical values are in good agreement with the NaOH values in
trains a-c which were not subject to correction. Thus, the % theoretical
values for the methods may be expected to be in the range of 80-90%.
1.6 S00 RESULTS FOR AN UNCONTROLLED COAL-FIRED POWER PLANT
(1 "" ™~~~ " "~L" ~ ^ ~
S02 samples were collected at a coal-fired power plant which burned
about 1% S fuel. The unit is essentially uncontrolled since the Aerotec
cyclones have an efficiency of approximately 30%. The unit schematic and
sampling port locations are shown in Figures 1-13 and 1-14, respectively.
All the S02 samples were collected on the south side of the duct. The
WALDEN RESEARCH CORPORATION
-------
TABLE 1-17
FUEL ANALYSIS FOR AN OIL-FIRED POWER PLANT
Fuel Component %
Oil S 0.98
0.99
0.95
0.94
0.95
0.99
0.97
0.94
C 85.33
85.18
H 13.62
13.74
C 85.19
86.62
H 13.85
13.72
Date
3/20/71
3/31/72
4/1/72
4/9/71
3/30/71
3/31/71
3/31/72
WALDEN RESEARCH CORPORATION
-------
TABLE 1-18
SUPPLEMENTARY FIELD TEST DATA FOR AN
OIL-FIRED POWER PLANT
Date
3/18/71
3/30/71
3/31/71
4/1/71
4/2/71
4/7/71
Test No.
1
2
3
4
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
51
52 .
53
54
55
56
o2 (%)
7.0
7.5
6.8
7.5
9.0
8.6
8.6
8.7
8.7
8.5
8.5
9.1
9.2
8.3
8.7
9.0
9.2
8.0
9.1
9.8
12.2
10.7
10.5
11.1
rn 1 °/ ^
LU0 I 7o I
d.
8.5
8.5
8.5
7.7
9.4
10.0
9.7
9.0
9.0
9.1
9.1
9.3
9.0
9.5
9.5
9.5
10.0
9.6
9.0
9.5
10.1
8.5
8.4
8.7
8.5
8.1
T (°F)
310
310
310
310
320
320
315
320
320
320
315
315
315
315
315
315
315
315
315
315
320
320
320
320
320
320
1 -43 WALDEN RESEARCH CORPORATION
-------
I
o
m
TO
m
w
m
TO
o
I
8
33
S
TABLE 1-19
COMPARISON OF MEASURED AND CALCULATED S02 VALUES
FOR AN OIL-FIRED POWER PLANT
BARIUM CHLORANILATE METHOD
Sample No.
21
22
23
24
25
27
28
29a
30a
31a
32a
34
51
52
53
54
55
56
ppm S02
Theoretical
394
420
401
378
366
370
370
378
369
390
390
393
423
356
352
364
356
339
Av. l,
Train
ppm SOp
431
451
445
438
373
347
312
311
--
--
--
350
398
375
360
376
342
375
1 Theoretical
a
% Theor
109
107
111
116
102
94
84
83 '
--
--
--
89
94
105
103
103
96
111
100
Train
ppm SO^
603
650
571
566
312
451
338
435
370
429
395
--
448
383
386
395
271
378
b
% Theor
153
155
142
150
85
122
91
115
100
110
101
--
106
108
110
109
76
112
114
o
z
-------
TABLE 1-20
COMPARISON OF MEASURED AND CALCULATED S02 VALUES
FOR AN OIL-FIRED POWER PLANT
Ba++ TITRATION METHOD
o
m
TO
m
OT
m
O
X
o
o
s
TO
Sample No.
21
22
23
24
25
27
28
29a
30a
31 a
32a
34
51
52
53
54
55
56
ppm S02
Theoretical
394
420
401
378
366
370
370
378
369
390
390
393
423
356
352
364
356
339
AV. :
ppm S(
556
566
560
560
322
452
339
439
362
432
406
.
448
391
392
411
410
404
I Theoretical
Train b
)2 % Theor
141
135
140
148
88
122
92
116
98
111
104
—
106
110
111
113
115
119
116
Train
ppm SO,,
369
361
382
357
342
373
205
317
250
253
279
—
346
216
355
295
280
286
c
% Theor
94
86
95
94
93
100
55
84
68
65
72
—
82
61
100
81
79
84
82
-------
TO
O
I
O
o
73
s
TABLE 1-21
COMPARISON OF MEASURED AND CALCULATED S02 VALUES
FOR AN OIL-FIRED POWER PLANT
NaOH TITRATION METHOD
o
m
z
Sample No.
21
22
23
24
25
27
28
29a
30a
31 a
32a
34
51
52
53
54
55
56
ppm S02
Theoretical
394
420
401
378
366
370
370
378
369
390
390
393
423
356
352
364
356
339
Av. I
Train
ppm S02 5
321
327
327
340
340
303
209
--
—
—
—
317
324
270
278
307
289
294
i Theoretical
a
I Theor
81
78
82
90
93
82
56
_-
—
—
--
81
77
76
79
84
81
87
81
Train
ppm SOp '
361
368
366
375
214
355
208
353
280
347
306
--
378
296
310
333
319
316
b
1 Theor
92
88
91
99
58
96
56
93
76
89
78
—
89
83
88
91
90
93
85
Train
ppm S02 '
368
374
350
366
360
392
204
320
274
275
297
--
386
212
354
300
292
303
c
I Theor
93
89
87
97
98
106
55
85
74
71
76
—
91
60
101
82
82
89
84
O
2
-------
ENGINEERING DATA
Efficiency - 88.0% @ 325,000 Ibs. per hour
Working Pressure - 1,025 Ibs.
Steam Temperature - 900° F.
Furnace Volume - 20,000 cu. ft.
Furnace Width - 24'-7'/,"
Furnace Depth - 20'-2y,"
Heat Reltase • 19,600 B.T.U.
Heating Surfaces
Boiler - 20,700 iq. ft.
Water Walls - 12.420 so,, ft.
Air Heater - 72,500 sq. ft.
Figure 1-13. Uncontrolled Coal-Fired Power Plant.
1-47
WALDEN RESEARCH CORPORATION
-------
12'
A South
42"-
Probe
Sampling Ports
Aerotec
Cyclone
Gas
Flow
Figure 1-14. Sampling Port Locations for the Uncontrolled Coal-Fired Power
Plant.
1-48
WALDEN RESEARCH CORPORATION
-------
other ports were welded onto the duct to allow a velocity profile to be
determined.
The test data obtained by running three trains in parallel off the same
probe are given in Table 1-22. The detail of each train is shown above in
Figure 1-5. The analytical methods appear to give good agreement between
trains except for Ba Chi or in train c (Table 1-22). Here the problem is a
very dilute sample. In the analytical scheme in Figure 1-8, train c is
diluted to 250 ml prior to aliquoting. This leads to a low absorbance and
large error for the chloranilate method. The titration methods can compen-
sate for the larger dilution by taking a larger aliquot for titration. This
is difficult to do for the Ba Chi or colorimetric procedure.
The precision for replicate analysis of samples in Table 1-23 indi-
cates no major interferences with the methods. Coefficients of variation
between 1-3.5% are very similar to the results obtained at the oil-fired
power plant (see Section 1.5). The precision appears to be about the same
for each train, e.g., NaOH titration CV = 2 +_ 1% for a, b, or c train, etc.
The regression analyses and correlation coefficients for the data in
Table 1-22 have been calculated and are tabulated in Table 1-24. Again,
Ba titration and Ba Chlor agree well within each train (a and b). The
agreement for the Ba Chlor method on train c is poor compared to other
analytical methods because of the dilution (see above). The NaOH titra-
tion comparisons to sulfate methods for train a and b has a large intercept
because of the corrections applied to the sulfate methods (Section 1.3.3).
The correlations between trains appear to be better on these samples than
on either of the other sites tested. There is no apparent reason for this
since interferences did not seem to be a problem on the oil-fired power
plant results. The data in Table 1-24 indicates that each of the methods
yields good results with no interferences for this coal-fired power plant.
Data on flue gas composition (CO^, 02) and fuel analysis are given in
Tables 1-25 and 1-26. Stoichiometric S0? values were calculated as de-
++
scribed in Section 1.5 and are compared with Ba Chlor, Ba titration, and
NaOH titration in Tables 1-27 to 1-29.
WALDEN RESEARCH CORPORATION
-------
TABLE 1-22
S02 TEST RESULTS FROM AN UNCONTROLLED COAL-FIRED POWER PLANT
I
en
o
TO
m
C/J
m
TO
o
I
o
o
o
TO
•
Sample
No.
37A*
37B*
38A
38B
39A
39B
40A
40B
41A
41 B
42A
42B
43A
43B
44A
44B
45A
45B
46A
46 B
47A
47B
48A
48B
49A
49B
50A
50B
ppm S02
(NaOH)
494
506
437
407
416
408
380
335
347
0
0
340
347
422
412
405
412
372
377
328
319
470
448
481
486
503
501
Train a
ppm S02
/ Q _ TT 1
V DO /
513
--
446
452
463
--
426
424
424
428
--
--
411
398
453
443
483
487
436
439
414
424
485
492
525
524
519
520
ppm S02
(Ba Chlor)
--
443
445
443
453
--
424
410
423
--
--
380
395
443
453
464
476
431
417
418
405
475
431
538
530
530
530
ppm S02
(NaOH)
528
513
467
465
425
425
411
411
408
405
324
326
335
331
398
405
410
404
313
319
335
331
467
500
499
499
500
506
Train b
ppm S02
(Ba++)
968
973
470-
473
484
485
461
450
465
511
408
408
408
. 414
423
424
465
474
443
446
422
427
50Z
503
517
536
505
497
ppm S02
(Ba Chlor)
536
556
465
474
448
483
411
450
410
420
396
396
396
399
404
429
447
477
431
448
410
403
502
482
539
553
490
502
ppm S02
(NaOH)
865
880
397
412
405
387
392
389
388
379
152
170
294
315
420
359
294
330
346
367
312
309
266
429
446
582
580
Train c
ppm S02
989
1005
395
396
363
358
315
308
334
327
148
147
292
284
292
294
376
383
342
327
256
248
455
452
407
400
596
588
ppm S02
(Ba Chlor)
369
332
264
264
254
254
168
252
215
315
146
119
264
352
0
--
223
..
176
__
189
--
363
_-
351
—
496
--
*
Where A and B are aliquots of the same sample.
o
•z.
-------
TABLE 1-23
PRECISION OF DUPLICATE S02 ANALYSIS* FOR AN
UNCONTROLLED COAL-FIRED POWER PLANT
Analytical Method Sampling Train No. of Samples
Ba Chloranilate (a) 11
(b) 14
Ba++ Titration (a) 11
(b) 11
(c) 13
£ NaOH Titration (a) 12
(b) 14
(c) 11
Sampling errors not included (replicate analysis)
(a) S03 collected by condenser
(b) SO- collected in impingers containing 80% IPA
O
1 (c) S03 collected in a bubbler containing 80% IPA
o
m
m
Z)
o
I
o
o
-o
o
TO
Mean S02 Concentration Standard cv /*x
(ppm) Deviation (ppm) ^ '
452 11.8 2.6
456 14.6 3.2
457 5.5 1.2
457 4.8 1.0
342 5.0 1.5
415 9.3 2.2
416 7.4 1.8
418 14.3 3.4
• S00 collected in H000 in all trains
2 22
o
-------
TABLE 1-24
REGRESSION ANALYSES AND CORRELATION COEFFICIENTS FOR
AN UNCONTROLLED COAL-FIRED POWER PLANT
Methods Compared
Ba Titration (a) [x] and
Ba Chloranilate (a) [y]
Ba Titration (b) [x] and
Ba Chloranilate (b) [y]
Ba++ Titration (c) [x] and
Ba Chloranilate (c) [y]
Ba Titration (a) [x] and
NaOH Titration (a) [y]
Ba Titration (b) [x] and
NaOH Titration (b) [y]
Ba Titration (c) [x] and
NaOH Titration (c) [y]
Ba Titration (a) [x] and
Ba++ Titration (b) [y]
Ba Titration (a) [x] and
Ba++ Titration (c) [y]
Ba Chloranilate (a) [x] and
Ba Chloranilate (b) [y]
Ba Chloranilate (a) [x] and
Ba Chloranilate (c) [y]
NaOH Titration (a) [x] and
NaOH Titration (b) [y]
NaOH Titration (b) [x] and
NaOH Titration (c) [y]
NaOH Titration (a) [x] and
NaOH Titration (c) [y]
Ba Chloranilate (a) [x] and
Sample
Size
22
26
16
23
26
27
23
23
23
17
25
26
23
23
Correlation
Coefficient
0.939
0.853
0.813
0.924
0.848
0.945
0.788
0.786
0.859
0.557
0.885
0.776
0.820
0.730
y
y
y
y
y
y
y
y
y
y
y
y
y
y
Regression
Equation
= 1.09 x -50.9
=1.01 x -20.5
= 0.688 x +41.6
= 1.36 x -215
= 1.46 x -266
= 0.776 x +97.0
= 0.731 x +132
= 1.75 x -429
= 0.849 x +72.5
= 0.914 x -113
=1.01 x +5.75
= 0.953 x +2.04
= 1.13 x -62.9
= 0.597 x +198
Ba++ Titration (b) [y]
Ta~J Sampling train - condenser - two impingers
(b) Sampling train - four impingers
(c) Sampling train - bubbler - two impingers
1-52
WALDEN RESEARCH CORPORATION
-------
TABLE 1-25
SUPPLEMENTARY FIELD TEST DATA FOR AN UNCONTROLLED
COAL-FIRED POWER PLANT
Date
4/5/71
4/6/71
4/7/71
Test No.
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Fuel
Coal
Coal
Coal
Coal
Coal
Coal
Coal
Coal
Coal
Coal
Coal
Coal
Coal
Coal
o2 (%)
9.0
7.5
5.8
5.0
4.6
5.4
5.5
5.3
5.3
6.7
6.3
6.5
6.5
6.5
6.1
rn (%}
WV/* \ K> 1
L
12.0
13.5
14.7
13.7
14.5
13.4
13.8
14.3
14.1
13.4
13.0
12.1
13.2
13.2
13.4
T (°F)
350
350
350
350
350
350
350
350
345
345
345
345
345
345
1-53
WALDEN RESEARCH CORPORATION
-------
TABLE 1-26
FUEL ANALYSES FOR AN UNCONTROLLED
COAL-FIRED POWER PLANT
Component
S
C
H
N
%
0.79
0.84
0.90
0.86
1.05
1.07
76.39
76.72
4.36
4.54
0.93
1.08
Date
4/5/71
4/6/71
4/7/71
1-54
WALDEN RESEARCH CORPORATION
-------
I
o
m
•33
m
en
m
TO
o
s
TABLE 1-27
COMPARISON OF MEASURED AND CALCULATED S02 VALUES FROM
AN UNCONTROLLED COAL-FIRED POWER PLANT
BARIUM CHLORANILATE METHOD
en
en
Sample No.
37
38
39
40
41
42
43
44
45
46
47
48
49
50
ppm S02
Theoretical
469
557
536
568
525
541
561
553
639
620
577
630
630
639
Av. I
Train
ppm S02 5
„
444
448
424
416
--
388
448
470
425
412
450
534
530
\ Theoretical
a
I Theor
..
80
84
75
79
--
69
81
74
69
71
71
85
83
77
Train
ppm S02
546
470
465
430
415
396
398
415
462
440
407
490
545
495
b
% Theor
116
84
87
101
79
73
71
75
72
71
71
78
87
77
82
Train
ppm S02
350
264
254
210
265
132
308
--
223
176
189
363
351
496
c
% Theor
75
47
47
37
50
24
55
--
35
28
33
58
56
78
48
o
-------
I
o
m
z
o
o
o
TO
s
TO
TABLE 1-28
COMPARISON OF MEASURED AND CALCULATED S02 VALUES FROM
AN UNCONTROLLED COAL-FIRED POWER PLANT
Ba++ TITRATION METHOD
I
en
Sample No.
37
38
39
40
41
42
43
44
45
46
47
48
49
50
ppm S02
Theoretical
469
557
536
568
525
541
561
553
639
620
577
630
630
639
Average
Train
ppm S02
513
449
463
425
426
--
405
448
485
437
419
488
525
520
Theoretical
a
% Theor
109
81
86
75
81
--
72
81
76
70
73
77
83
81
80
Train
ppm S02 5
„
472
485
455
488
408
418
445
460
445
425
503
526
500
b
6 Theor
..
85
90
80
93
75
75
80
72
72
74
80
83
78
80
Train
ppm SOp
„
395
360
312
330
147
288
293
380
335
252
453
403
592
c
% Theor
..
71
67
55
63
27
51
53
59
54
44
72
64
93
59
o
-------
I
o
m
z
TO
m
o
I
8
TO
s
TO
5
z
TABLE 1-29
COMPARISON OF MEASURED AND CALCULATED S02 VALUES FROM
AN UNCONTROLLED COAL-FIRED POWER PLANT
NaOH TITRATION METHOD
I
en
Sample No.
37
38
39
40
41
42
43
44
45
46
47
48
49
50
ppm S02
Theoretical
469
557
536
568
525
541
561
553
639
620
577
630
630
639
Av.
Train
ppm S02
500
422
412
380
340
--
343
417
410
375
324
460
483
502
% Theoretical
a
% Theor
107
76
77
67
65
--
61
75
64
60
56
73
77
79
72
Train
ppm S02
520
466
425
411
406
325
333
373
407
316
333
480
499
503
b
% Theor
111
84
79
72
77
60
59
67
64
51
58
76
79
79
73
Train
ppm S0£
..
405
395
390
383
160
305
390
315
355
310
266
437
581
c
% Theor
._
73
74
69
73
30
54
71
49
57
54
42
69
91
62
-------
The Ba Chlor method yields results which are about 80% of the theoret-
ical concentration for trains a and b. The results for Ba Chlor in train c
should be disregarded since the sampling procedure required dilution to 250
ml before an aliquot was taken. This led to a very low absorbance and,
therefore, poor accuracy since a small absorbance had to be multiplied by a
large dilution factor. The Ba titration results for trains a and b in-
dicate about 80% of theoretical which is the same as that for Ba Chlor.
Train c was not flushed to remove residual SOp from the IPA absorber (see
Figure 1-8). This can account for the low results since S02 is soluble in
the water-alcohol mixture.
The NaOH titration results for trains a and b (Table 1-24) are slightly
lower than the sulfate methods. It is not clear whether this is real or an
artifact of the adjustment of the sulfate methods. Again, train c produces
lower S02 results because the residual S02 was not removed (by flushing)
from the IPA absorber.
1.7 S03 RESULTS FROM FOSSIL FUEL COMBUSTION SOURCES
The major problem in quantitative collection of SO., in power plant ef-
fluents is that S02, present in a large excess, is easily oxidized, lead-
ing to high S0~ values and/or poor precision and accuracy. Most collection
methods for the sulfur oxides have been based upon physical separation of
S02 and SO- (by differential absorption) to reduce the magnitude of the
oxidation problem. Techniques in which both oxides are collected without
separation (such as absorption in caustic solution) have not been widely
used since accuracy is generally poor due to oxidation of dissolved SO-.
S03 collection techniques may be divided by approach into two differ-
ent classes: absorption and condensation methods. The former depends on
the solubility of HUSO, in aqueous solutions. Inhibitors, such as alcohols,
have generally been added to prevent the oxidation of dissolved S02. Junge
and Ryan (1958) have found that trace metals catalyze the oxidation of S0?
to sulfate. It is very difficult to inhibit the oxidation of S0? in solu-
tion, therefore, the absorption method may lead to poor accuracy. The
condensation method depends upon controlled cooling of the flue gas to a
1_gg WALDEN RESEARCH CORPORATION
-------
temperature where H^SO. aerosol is formed and subsequent collection on a sin-
tered glass frit or filter paper. The important feature of the controlled
condensation method is that the temperature of the collector is maintained at
60-90°C which is above the typical water dewpoint (40-45°C) for combustion
sources. This eliminates the problem of S(L dissolving in an aqueous phase
and being oxidized to SOT.
The controlled condensation method should lead to improved accuracy in
SCL determinations. In order to assess these methods, IPA and controlled
condensation trains were run at each of the sources. The trains are shown
schematically in Figure 1-5 and the condenser in detail in Figures 1-4 and
1-6.
The results obtained for SO- determinations are given in Table 1-30.
The ion exchange step was eliminated in the procedure to prevent losses of
sulfate observed previously for the barium ion titration. The controlled
condensation method (analyzed by Ba Chi or) gives considerably higher results
for the oil-fired plant than the 80% IPA absorption (analyzed by Ba++ titra-
tion). The average SO- concentration results found for the controlled con-
densation method at the coal-fired power plant with the Combustion Engineer-
ing wet scrubbing system is about the same as IPA absorption but the corre-
lation coefficient for the results is only 0.1. At the uncontrolled coal-
fired plant, lower results were observed for the controlled condensation
method. The 80% IPA results obtained for the oil-fired plant give an SO-
concentration which is about 40% of that obtained by the controlled conden-
sation method. For the controlled coal-fired plant, the IPA results are
about 70% of the values obtained by the controlled condensation technique.
In the uncontrolled coal-fired plant, however, the situation is reversed.
The controlled condensation technique yields a mean SO- concentration which
is about 2/3 of the 80% IPA value. The correlation coefficient for the S03
tests for the oil-fired power plant is 0.4 and about zero for the uncon-
trolled coal-fired power plant. The large difference between the SO-
methods is quite alarming and should be investigated in detail.
1 -59 WALDEN RESEARCH CORPORATION
-------
TABLE 1-30
S03 TEST RESULTS FOR FOSSIL FUEL COMBUSTION
I
O
6
33
m
CO
m
>
a)
O
O
O
z>
s
Sample No.
21
22
23
24
27
28
51
52
53
54
55
56
11
12
13
14
15
16
39
40
41
43
44
45
47
48
49
50
Controlled Condensation /
ppm SO-*
1.1
6.3
5.9
5.5
5.9
5.7
7.5
6.2
5.6
6.6
7.3
8.2
3.7
6.1
2.4
1.5
7.9
4.5
0.7
1.6
3.2
1.6
0.5
2.6
6.6
5.0
2.1
3.8
^bSOrPpPmnSOnt8°% IPA Source
2.2 Oil
2.2
2.5
2.7
6.9
3.1
4.0
4.0
3.8
3.2
3.3
4.8
3.4 Controlled Coal
3.7
2.2
6.1
5.1
6.5
8.8 Uncontrolled Coal
9.6
3.4
5.0
2.4
3.2
5.9
4.8
8.8
8.1
Analytical Method - Barium Chioranilate
Analytical Method - Ba++ Titration
O
•2.
-------
1.8 S02 RESULTS OBTAINED IN THE WALDEN PILOT PLANT AND LABORATORY
1.8.1 Description of Pilot Plant
A photograph of the Wai den combustion pilot plant is shown in
Figure 1-15. The unit consists of a 400,000 Btu/hour (Jackson and Church)
furnace with a combination gas/oil burner. The waste heat is discharged
and the exhaust gas from the burner is passed into a series of carbon steel
test sections three feet in length and eight inches in diameter. The flue
gas is cooled down to about 300°F by an air-cooled heat exchanger and
passed into a second series (3) of carbon steel test sections (sampling
areas). The gas is pulled out of these test sections by a Westinghouse in-
duced draft fan and exhausted through corrugated pipe at roof level.
Detailed schematics of the sampling test sections and pilot
plant are shown in Figures 1-16 and 1-17, respectively. Each of the test
sections has four static ports which can be used for particulate or gas
sampling.
The original burner was designed to burn either natural gas or
#2 fuel oil. When firing #2 oil, the excess air was only 50% in excess of
theoretical. Units of this size normally run at 100 to 150% excess air.
The unit has been modified to attain better combustion. A 90 CFM fan was
placed on the air intake to the fire box. Using this auxiliary fan, we are
able to burn #4 fuel oil with good combustion efficiency at the highest oil
flow of 3.6 gal/hr. With these large nozzles, the heat output is increased
beyond the design capacity of the original unit. The Btu output was in-
creased to 650,000 Btu/hr with three 1.2 gal/hour nozzles. The volumetric
flow in the duct was increased to 160 CFM with the auxiliary fan. The
pilot plant specifications are given in Table 1-31. The concentrations of
some gaseous and particulate pollutants are given in Table 1-32 for oil and
gas firing.
1.8.2 Supplementary Laboratory and Pilot Plant Studies
Some laboratory studies were conducted involving linearity of
1-61
WALDEN RESEARCH CORPORATION
-------
CTl
ro
O
m
in
O
O
O
Figure 1-15. Wai den Pilot Plant.
-------
5 FT
F ROM
HEAT
EXCHANGER
1
1
n n '
vj \j |
i
C) O
w ^
\
A
i
i
O O
w w !
^ "
/ i
U)
FLOW
STRAIGHTENER
TEST A
TEST B
SAMPLING
PORTS
o
m
T
8" TO ID
FAN
o
o
o
TO
s
33
Figure 1-16. Pilot Plant Test Section Assembly.
-------
DOPING
SYSTEM
H EAT
EXCHANGER
FU RN ACE
EXTER NAL
READING
METER
SAMPLING
SECTION
FAN
EXHAUST
Figure 1-17. Pilot Plant Schematic.
1-64
WALDEN RESEARCH CORPORATION
-------
TABLE 1-31
PILOT PLANT SPECIFICATIONS
1. FURNACE
200,000-650,000 Btu/hr Input; gas, #2, #4, and possibly #6 fuel
oil
Warm air space heater
Flue gas exit temperature 900°F
2. HEAT EXCHANGER
Duty - cool flue gases from 900 to 275°F
Heat duty 56,000 Btu/hr
3. TEST SECTIONS
See separate drawings for details
Hot test section: Temperature 600-900°F (depending on fuel flow
rate and composition)
Diameter = 8"
Cold test section: Temperature 300°F
Diameter =8"
Gas velocity from ^3 to 10 ft/sec
4. Fl. F2. F3
Flow meters to measure fuel and combustion air flow rates
Fl range: Direct reading meter for measuring flows of #2 and #4
oil from 0.25 to 5 GPH
F2 range: 2000-10,000 SCFH (for flue gas)
F3 range: 500-190 SCFH (of natural gas)
5. £4
Flue gas flow rate - system will operate from 4900 to 7500 SCFH
on oil and from 2500 to 5000 SCFH on gas.
6. Tl_
Temperature control system; controls cooling air
Flow rate to maintain 900°F temperature in exit from the furnace
7. T2
Temperature control system; controls cooling medium flow rate to
heat exchanger to maintain flue gas temperature exit from 325
to 150°F
8. T^P
Temperature and pressure sensors
Temperature: Max. 900°F; min. 150°F
Pressure: Approximately atmospheric
1-65
WALDEN RESEARCH CORPORATION
-------
TABLE 1-32
CHARACTERISTICS OF THE PILOT PLANT
ON #2 OIL AND GAS
Parti cul ate
Hydrocarbon (ppm)
N0x
CO
Maximum Temperature (°F)
(hot section)
Maximum Flow (CFM)
After steady-state conditions
Fuel Oil
0.008 gr/SCF
^ 2-5
50
200-1000
^ 900
150
are attained.
Gas
% 0.5-5
40
50-500
^ 750
120
, _gg WALDEN RESEARCH CORPORATION
-------
the colorimetric method, collection efficiency and required flushing times
for 80% IPA absorption.
Some early laboratory work on the barium chloranilate method
for sulfate indicated that the calibration curve was linear up to 500 yg/ml
of sulfate. Using calibration curves prepared for sulfate concentrations up
to 500 \ig/ml, least square slopes of about 480 yg/ml/abs unit were obtained.
The data in Figure 1-18 yields a slope of ^ 500 yg/ml/abs unit. Since this
difference may amount to as much as 10 to 15% of the SO concentration, a
A
detailed study of the linearity of the barium chloranilate method was con-
ducted. We found that the method started to deviate from linearity between
350 and 400 yg/ml of sulfate. The deviation from Beer's law is not unusual.
The positive deviation above ^ 400 yg/ml indicates an ionic strength effect
(Barney, 1964). This leads to the lower value of the slope (480 yg/ml) if
concentrations above 350 yg/ml are used for preparation of the calibration
curve. The slope is ^ 500 yg/ml/abs unit if extrapolated from sulfate
levels below 350 yg/ml. Thus, the correct slope is 500 yg/ml/abs unit at
pH 5.6 and the method is linear only to 360 yg/ml. This value is in agree-
ment with that of Barney (1964) and Kanno (1964).
Some data on the collection efficiency of HpOp as a function of
S02 concentration was determined. This will be discussed in detail in
Section 1.8.3.
A series of tests were run in the pilot plant with gas firing
and S02 doped up to 600 ppm. Note that no SO. was added. The SOp was
added into the duct at a low temperature (650°F), therefore, no oxidation
is likely to occur. The sampling train consisted of two 80% IPA impingers
in an ice bath followed by two peroxide impingers in series. The isopropa-
nol impingers were removed from the ice bath and, after the appropriate
flushing time, were treated with peroxide to oxidize any dissolved S0? to
sulfate. The sulfate concentration was determined with the barium ion-
thorin titration. These data are given in Table 1-33 and plotted in Fig-
ure 1-19. We see that after flushing the IPA impingers for 15 minutes at
1 liter/min (this represents an equal volume of clean air since 15 liters
]_57 WALDEN RESEARCH CORPORATION
-------
Ok
00
o
m
m
I
8
30
s
TO
1.000
0.800
£
c
o
K)
m
< 0.600
cc
O
^
CD
0.400
0.200
O.OOO1-
LINEAR REGION
JOO
200
300
of sulfate
400
500
Figure 1-18. Linearity of the Calibration Curve for the Barium Chloram'late Method.
-------
TABLE 1-33
EFFECT OF FLUSHING TIME ON THE S02 CONCENTRATION
IN THE 80% ISOPROPANOL-20% WATER ABSORBER
Run No.
09-2
09-4
010-2
010-4
011-2
011-4
012-2
012-4
013-2
013-4
014-2
014-4
015-2
015-4
016-2
016-4
017-2
017-4
018-2
018-4
*
At a rate
Flushing^ Time
(min)
0
0
0
0
0
0
0
0
5
5
5
5
10
10
10
10
16
16
25
25
of 1.0 1/min
ppm S02
in 80% I PA
100
83
82
75
80
80
72
120
93
59
12
26
7
6
14
22
7
5
15
4
in Hp Op
407
387
412
383
438
370
421
342
380
442
338
438
524
428
503
432
427
430
424
372
% S02 in the
80% IPA
25
21
20
20
15
18
15
26
20
12
4
6
1
1
3
5
2
1
3
1
•|_69 WALDEN RESEARCH CORPORATION
-------
2
a
a:
10
UJ
u
CC
LL)
Q.
TOTAL SC£ (PEROXIDE ADDED)
504 BY AIR
OXIDATION
TIME (mm)
Figure 1-19. Effect of Flushing Time on the S02 in 80% IPA.
25
1-70
WALDEN RESEARCH CORPORATION
-------
of flue gas were collected), the initial SCL concentration in the I PA
dropped from 20 to 2% of the SCL concentration in the hydrogen peroxide.
The data in Table 1-33 represent the maximum sulfate concentra-
tions to be expected since all the dissolved SCL was purposely oxidized to
sulfate. However, two runs at ^ 500 ppm SOp which were flushed for five
minutes and analyzed the following day gave 7 and 9 ppm sulfate in the IPA.
No peroxide was added to these samples. This is interesting in view of the
fact the S03 in fossil fuel-fired plants is * 2% of the S02> Here with no
so added, the residual sulfate (due to oxidation of dissolved SO,) amounts
*J _ C.
to 1% of the SOp. Since no metals nor SO- was present, the SOT came from
the oxidation of SOp by air. These results raise some doubt about the ao-
curacy of the SO- determination if 80% IPA absorption is used.
One additional problem which was observed was the difficulty
with the barium ion titration at SOp concentrations below 50 ppm. The end-
point becomes very indistinct and difficult to reproduce with different
operators at these low levels.
1.8.3 Effect of NO., on S00 Methods and Accuracy Studies
X L.
The concentration of oxides of nitrogen may vary from several
hundred to about 1500 ppm in large coal-fired power plants (Driscoll and
Berger, 1971). Nitric oxide could interfere in the NaOH titration by being
oxidized to ML and reacting with water to form HNO-.
The pilot plant was fired on gas and the flue gas was doped
with SOp and NO. The total volumetric flow in the pilot plant was deter-
mined by stoichiometry with SOp and NO being added with calibrated rota-
meters. Initially, the accuracy of the methods were determined as a func-
tion of the SO, concentration with no addition of NO (just the background
£ A
from the furnace). The data are given in Table 1-34. A total sample vol-
ume of 15 liters was collected at a flow rate of 1.0 liter/minute. Samples
60-1 and 60-2 represent results from two parallel trains (peroxide impingers)
run off the same probe.
WALDEN RESEARCH CORPORATION
-------
TABLE 1-34
Sample No.
60-1
60-2
61-1
61-2
62-1
62-2
63-1
63-2
64-1
64-2
65-1
65-2
66-1
66-2
67-1
67-2
68-1
68-2
69-1
69-2
70-1
70-2
71-1
71-2
72-1
72-2
73 1
73-2
74-1
74-2
75-1
75-2
76-1
76-2
77-1
77-2
78-1
78-2
79-1
79-2
80-1
80-2
Ul 1 l_\^ 1 V
ppm NO
^ 40
~ 40
* 40
% 40
* 40
~ 40
~ 40
* 40
~ 40
^ 40
'v* 40
•v* 40
* 40
~ 40
^ 40
^ 40
~ 40
^ 40
'x. 40
^ 40
* 40
^ 40
~ 40
^ 40
^ 40
^ 40
* 40
~ 40
^ 40
-v 40
356
356
356
356
356
356
618
618
618
618
618
618
A
Theoretical
ppm S02
306
306
306
306
306
306
612
612
612
612
612
612
968
968
968
968
968
968
1246
1246
1246
1246
1246
1246
1590
1590
1590
1590
1590
1590
360
360
360
360
410
410
387
387
268
268
300
300
•IIWIL. rit i nuu.
ppm S02
(NaOH)
300
310
351
352
336
318
373
254
272
283
267
271
J 1 wl\ Owrt
ppm S02
326
263
280
274
259
264
524
524
536
556
570
523
799
834
848
806
871
868
1184
1193
1136
1248
1146
1147
1530
1520
1298
1521
1390
1431
299
302
350
355
318
321
383
264
281
291
264
275
ppm S02
(Ba Chlor)
272
240
249
253
267
256
525
534
523
543
543
472
818
800
850
776
811
808
1090
1117
1101
1113
1071
1102
1438
1444
1211
1466
1346
1394
286
288
348
347
308
308
385
259
278
273
267
252
1-72
WALDEN RESEARCH CORPORATION
-------
TABLE 1-34 (continued)
Sample No.
81-1
81-2
82-1
82-2
83-1
83-2
ppm N0x
990
990
990
990
990
990
Theoretical
ppm S0«
333
333
347
347
344
344
ppm S02
(NaOH)
388
336
318
342
339
346
ppm S02
301
342
330
344
341
349
ppm S02
(Ba Chlor)
271
328
314
340
324
340
1-73 WALDEN RESEARCH CORPORATION
-------
The percent of theoretical SCL measured by Ba titration in-
creases from 90% at 300 ppm to 95% above 1000 ppm. The barium chloranilate
method yields slightly lower results with values ranging from 84% at 300 to
92% of theoretical S02 above 1000 ppm. Nitric oxide has no effect on any
of the wet chemical methods as demonstrated from the data in Table 1-34
(runs 75-83). Note that the ratio of Ba++/NaOH or Ba++/Ba Chlor remains
essentially constant over the entire range. "T" tests (to be described
later in this section) indicate no significant difference between any of
the methods as a function of NO level. The average percent theoretical
J\
(accuracy) for the methods at about 350 ppm is as follows:
Ba titration 92% accurate
NaOH titration 92% accurate
Ba Chlor 88% accurate
The "accuracy" appears to increase slightly as the S0? level is increased
++
(runs 69-74), accuracies of 95 and 92% are observed for Ba and Ba Chlor,
respectively.
In the above work, we have shown that the wet chemical methods
for S02 yield about 90% of the theoretical SOp values at ^ 300 ppm and
^ 95% of the theoretical S02 above 1000 ppm S02> One reason for this low
recovery of S02 could be a reduced collection efficiency at lower S02 con-
centrations. Therefore, we titrated the sulfate (S02) in the impingers
separately instead of combining them for a series of tests. The collec-
tion efficiency was determined from pilot plant studies assuming that all
the S02 was collected in the two impingers, and the fraction collected in
impinger 1 divided by the total was the collection efficiency. The data
in Table 1-35 show that the collection efficiency is 98% at both ^ 500 and
1100 ppm S02. Thus, the decreased accuracy cannot be explained on the
basis of reduced collection efficiency.
In another program (Driscoll, et al., 1972), we have investi-
gated the accuracy of the Ba method by sulfur in fuel plus stoichiometn
rather than direct addition of S09. Fuel oils (#2 and #4) were burned at
1-74 WALDEN RESEARCH CORPORATION
-------
TABLE 1-35
DETERMINATION OF COLLECTION EFFICIENCY OF SO,
IN AQUEOUS HYDROGEN PEROXIDE SOLUTION *
ppm
Impinger 1
394
364
415
336
380
434
338
433
524
425
499
432
427
430
424
372
1062
1294
852
1048
1117
1452
1162
1340
1176
1169
1300
1204
1348
1231
1282
1060
so2
H2°2
Impinger 2
44
6
6
6
0
8
0
5
0
3
4
0
0
0
0
0
0
0
0
26
21
19
21
20
48
10
10
3
28
19
8
6
Total ppm SOp
448
370
421
342
380
442
338
438
524
428
503
432
427
430
424
372
1062
1294
852
1074
1138
1471
1183
1360
1224
1179
1310
1207
1378
1250
1290
1066
Collection Efficiency (%)
(S02 Impinger 1 \
Total S02 )
88
98
96
98
100
98
100
99
100
99
99
100
100
100
100
100
100
100
100
98
98
99
98
99
96
99
99
100
98
98
99
100
Sample rate 1.0 1/min
Sampling time 15 min
1-75
WALDEN RESEARCH CORPORATION
-------
various levels of excess air. The results are shown in Table 1-36. The
percent theoretical obtained is 95% although lower SCL levels were present
in the duct. Thus, some of the reduced accuracy at low (300 ppm) S02
levels with gas firing may be due to flow meter fluctuations, etc. It is
not unreasonable, on the basis of the present data, to estimate the accu-
racy of the wet chemical methods as 95 +_ 5%.
The following is a detailed statistical analysis of the pilot
plant data which appeared in Table 1-34. The data were analyzed to deter-
mine the statistical significance of the differences between various' S02
measurement procedures carried out under various conditions. The statis-
tical analyses and their results are presented below.
Comparison of Ba and Ba Chi or as a Function of SO,, Concen-
tration
The data in Table 1-34 represent 30 paired samples. Each pair
represents analyses by each of the two methods on one effluent sample. The
following procedures were used to compare methods.
(a) Hypothesis: y1 = y2
Reject hypothesis when
t 1-t(l-a/2)(29) or t >_ t(l-a/2)(29)
where y-, .u^ are means of the measurements for the two sampling methods;
t(l-a/2)(29) is cumulative t-distribution for significance level a and 29
degrees of freedom; t = D/S^-; [T is the mean of the differences between the
paired samples (Di's); and S^- is the standard deviation of the Di's.
(b) Hypothesis: o, = cu
Reject hypothesis when
F <. F(a/2)(29,29) or F >_ F(l-a/2)(29,29)
where a,,a2 are the standard deviations of the measurements for the
, yg WALDEN RESEARCH CORPORATION
-------
TABLE 1-36
PILOT PLANT COMPARISON OF THEORETICAL S02 VERSUS
PPM SOg MEASURED BY WET CHEMISTRY
Fuel
#2 Oil
0.75 nozzles
#2 Oil
(0.75 nozzles)
change excessive
air
#2 Oil
large (1.2)
nozzles
ppm S02
by Ba++
1
2
3
4
5
6
7
8
9
10
11
12
149
141
137
143
119
123
96
100
79
Tn
79
RO
ou
oc
85
136
134
131
130
116
124
137
133
106
102
133
130
, 145
140
121
98
79
82
135
130
120
135
104
131
Theoretical
ppm SO.
122
122
122
100
100
100
145
145
145
145
145
145
% Theoretical
119
115
99
98
79
82
93
90
83
93
72
90
1-77
WALDEN RESEARCH CORPORATION
-------
TABLE 1-36 (continued)
Fuel
#4 Oil
large (1.2)
nozzles
•v
ppm S02
by Ba++
13 ?94
295
14 235 235
15 274 268
16 261 266
17 272 27Q
~~Icn % Theoretical
ppm bU«
260
260
260
260
260
Average % Theoretical
113
90
103
102
103
= 95
From Driscoll, J. N., Final Report to SWRI, June 1972.
1-78
WALDEN RESEARCH CORPORATION
-------
sampling methods; F(a/2)(29,29), F(l-a/2)(29,29) are cumulative F-distribu-
tions for significance level a and 29 by 29 degrees of freedom; F = ratio of
sample variances of the two methods.
(c) To analyze the comparison as a function of SCL concentra-
tion, the values of the Di's were regressed on the SCL concentration levels
++
(Ba ). A slope of zero would imply that the difference between the methods
is not a function of SCL concentration so the hypothesis here is that the
slope of the regression line (3,) is zero. (This hypothesis was not for-
mally tested.)
Results
(a) Hypothesis: y, = y2
and
t(l-0.05/2)(29) = 2.045
Therefore, cannot reject hypothesis at 5% significance level, i.e., no sig-
nificant difference between methods at 0.05 level.
(b) Hypothesis: a, = a-
^.i.io
and
F(l-0.05/2)(29,29) = 2.07
Therefore, cannot reject hypothesis at 5% significance level, i.e., no sig-
nificant difference between precision of methods at 0.05 level.
(c) Regression coefficient for Di's vs S02 level is 0.045.
Therefore, difference between methods is not a significant function of SOp
levels. (We have not formally tested this hypothesis.)
1_yg WALDEN RESEARCH CORPORATION
-------
Comparison of Ba . Ba Chi or and NaOH Methods as a Function of
NOX Level
The data in Table 1-34 represent 18 samples. Each set of data
represents analyses by each of the three methods on one effluent sample.
Each of the three combinations of paired methods were analyzed by the fol-
lowing procedures.
(a) Hypothesis: MI =.y2
Reject hypothesis when
t <_ -t(l-o/2)(17) or t >_t(l-a/2)(17)
(b) Hypothesis: a, = a?
Reject hypothesis when
F <_ F(a/2)(17,17) or F >_ F(l-a/2)(17,17)
(c) Hypothesis: 6-1=0 (regression of Di's on NO level)
I /\
(We have not formally tested this hypothesis.)
Results
(a) Hypothesis: y^ = y?
(i) NaOH method vs Ba++ method
t = 0/5^ =^£=-0.39
and
t(a = 0.05)(17) = 2.11
Therefore, cannot reject hypothesis, i.e., no significant difference.
(ii) NaOH method vs Ba Chlor method
1 -80 WALDEN RESEARCH CORPORATION
-------
and
t(a = 0.05)(17) = 2.11
Therefore, cannot reject hypothesis, i.e., no significant difference.
(iii) Ba method vs Ba Chi or
t = 0/50 = = 2.67
and
t(a = 0.05)07) = 2.11
Therefore, reject hypothesis, i.e., there is a significant difference (Ba
method higher than Ba Chlor) at the 0.05 level. However, t(a = 0.01) (17) =
2.890 and, therefore, cannot reject it at the 0.01 level.
(b) Hypothesis: a, = cu
(i) NaOH method vs Ba method
r _ 35.12
F -
F(ct = 0.05)(17,17) = 2.66
Therefore, cannot reject hypothesis, i.e., no significant«.difference.
(ii) NaOH method vs Ba Chlor method
F(a = 0.05) = 2.66
Therefore, cannot reject hypothesis, i.e., no significant difference.
(iii) Ba method vs Ba Chlor method
1-81
WALDEN RESEARCH CORPORATION
-------
F(o = 0.05)(17,17) = 2.66
Therefore, cannot reject hypothesis.
(c) Hypothesis:' B-j = 0
(i) NaOH method vs Ba++ method
B1 = 0.015
Therefore, difference between methods is not significant function of NO
X
level.
(ii) NaOH method vs Ba Chi or method
B] = 0.005
Therefore, difference is not significant function of NO level.
/\
(iii) Ba method vs Ba Chi or method
B] = 0.01
Therefore, difference is not significant function of NO level.
/\
Additional statistical data are given in Tables 1-37 and 1-38
on the effect of NO doping. The precision is essentially the same as a
A
function of NO level and the regression analyses and correlation coeffi-
J\
cients (Table 1-38) indicate that all three methods are in good agreement
with each other.
1.8.4 Effect of HC1 on the Accuracy of S00 Methods
r _ ^ _, ,
1.8.4.1 Introduction
Chlorides (such as NaCl and KC1) as well as chlorine
are present in coal and are converted to HC1 during combustion (lapalucci,
Demski and Bienstock, 1969). The effect of HC1 was studied since it was
expected to be an interference with the NaOH titration procedure. The
WALDEN RESEARCH CORPORATION
-------
TABLE 1-37
PILOT PLANT RESULTS - PRECISION WITH NOX DOPING
I
00
CO
I
D
m
3)
o
I
8
TO
Analytic Method
Barium Chi orani late
Ba++ Titration
NaOH Titration
NaOH Titration
Ba Titration
Barium Chi orani late
Runs
60-62
63-65
66-68
69-71
72-74
60-62
.63-65
66-68
69-71
72-74
60-62
63-65
66-68
69-71
72-74
75-77
78-80
81-83
75-77
78-80
81-83
75-77
78-80
81-83
# Samples
6
6
6
6
6
6
6
5
5
6
6
4
5
6
6
6
6
6
6
6
6
6
6
6
ppm NO
•v 40
% 40
*x. 40
^ 40
* 40
^ 40
* 40
% 40
^ 40
* 40
* 40
•x, 40
•v, 40
n, 40
% 40
356
618
990
356
618
990
356
618
990
mean ppm
so2
235
523
811
1099
1384
278
544
838
1161
1448
189
357
480
677
900
328
287
328
324
293
335
278
236
283
Standard
Deviation
7.3
7.7
10.6
9.5
8.7
15.2
15.5
38.7
25.7
27.2
12.0
6.1
10.6
9.8
12.8
2.6
9.5
8.9
2.8
3.4
4.0
4.4
3.6
5.9
CV
3.1
1.5
1.3
0.9
0.6
5.5
2.9
4.6
2.2
1.9
6.3
1.7
2.2
1.4
1.4
0.8
3.3
2.7
0.9
1.2
1.2
1.6
1.5
2.1
TO
O
z
-------
TABLE 1-38
REGRESSION EQUATIONS AND CORRELATION COEFFICIENTS FOR
A GAS-FIRED PILOT PLANT
1
00
"**
s
r-
o
z
TO
C/J
m
TO
O
I
O
O
TO
T)
O
TO
Methods Compared
Ba Titration (x) and
Ba Chloranilate (y)
Ba Titration (x) and
Ba Chloranilate (y)
Ba Titration (x) and
NaOH Titration (y)
Ba Chloranilate (x) and
NaOH Titration (y)
Ba Titration (x) and
Ba Chloranilate (y)
t i
Ba Titration (x) and
NaOH Titration (y)
Ba Chloranilate (x) and
NaOH Titration (y)
Ba Titration (x) and
Ba Chloranilate (y)
Ba Titration (x) and
NaOH Titration (y)
Ba Chloranilate (x) and
NaOH Titration (y)
Sample No.
60-74
75-83
75-83
75-85
85-100
85-100
85-100
101-113
101-113
101-113
Dopant
(just SOJ
L
N0y
x
N0y
x
N0y
x
HC1
HC1
HC1
HOAc
HOAc
HOAc
Correlation
Coefficient
0.997
0.974
0.819
0.961
0.890
0.680
0.810
0.970
0.803
0.766
Regression Equation
y = 0.96 x -11.7
y = 0.929 x -23.3
y = 0.915 x +29.4
y = 1.04 x +31.7
y = 1.07 x -33
y = 0.810 x +96
y = 0.87 x +81
y = 0.879 x -8.75
y = 0.778 x +43.9
y =0.850 x +41.7
O
z
-------
chlorine content of some American coals is given in Table 1-39 and is seen
to vary from 0.01 to 0.5%. British coals are generally higher in chlorine
content than American coals (Smith and Gruber, 1966). The British coals
are classified with regard to chlorine content as follows: high, over 0.3%;
medium, 0.15-0.3%; low, below 0.15%.
Ten coals containing various levels of chlorine were
burned in a pilot plant at the BuMines to determine the behavior of chlorine
in coal combustion (lapalucci, Demski and Bienstock, 1969). A detailed
analysis of the composition of the various coals is given in Table 1-40.
Earlier work had shown that one of the products of combustion of chlorine
containing coals was HC1. lapalucci, Demski and Bienstock (1969) used both
mass spectrophotometric and wet chemical (Volhard titration) techniques for
analysis of flue gas samples. HC1 was the only product detected from the
combustion of chlorine containing coals. They calculated the equilibrium
distribution for a number of chlorine containing compounds using a computer
program.
These data are given in Figure 1-20. Although Cl?,
HOC1, and NaOCl are possible products, none of these compounds were de-
tected. However, analytical methods employed may not have been sensitive
enough to detect them. Sampling at a large power plant showed that only
1.5% of the chlorine was retained in the ash with the remainder (^ 90%)
being emitted as HC1 (lapalucci, et al., 1969).
In Table 1-40, we note that the maximum Cl/S ratio in'
the coal is 0.39. When this is converted to Cl/SO^, the value becomes
^ 0.195.
These coals were selected by BuMines as typical, and
the chlorine contents probably represent the maximum amount of hydrogen
chloride evolution. If we analyze the data in Table 1-40 and determine the
HC1/S02 ratio, we find that for the five high sulfur coals (S > 2%) the cal-
culated HC1 value is 2.5 ± 0.9% of the S02 concentration. Thus, the inter-
ference for an acid-base titrametric procedure would be minimal (less than
3%). For the low sulfur coals (S = 1% or less), however, the Cl/S ratio
1 -85 WALDEN RESEARCH CORPORATION
-------
TABLE 1-39
CHLORINE CONTENT OF VARIOUS COALS*
Source of Coal
State Bed
Ohio Sharon
Illinois No. 6
Indiana No. 4
West Virginia Pittsburgh
Pennsylvania Lower Freeport
Chlorine Content, %
0.01
0.01
0.06
0.07
0.14
Illinois Central Illinois 0.35
Oklahoma Henryetta
Taken from "Atmospheric Emissions
W. S. Smith and C. W. Gruber, Pub.
1966).
0.46
from Coal Combustion," by
0999-AP-24 by PHS (April
_gg WALDEN RESEARCH CORPORATION
-------
TABLE 1-40
COMBUSTION OF PULVERIZED COALS
Coal seam,
county, and
State
ASTM designation, D388-66
Mine
Coal analysis, as received,
wt pet:
Sulfur
Ash
Sodium.
Proximate analysis , wt pet :
Moisture
Volatile matter... . .
Ash
Proximate analysis, Btu
Recovery in ash, percent:
Sulfur
Atom ratio, Cl/S:
Carbon combustion efficiency,
percent
Beckley ,
Wyoming ,
W. Va.
Lvb
West Gulf
4.5
82.0
1.6
6.3
0.9
0.11
4.7
0.10
0.11
0.02
0.09
3.6
18 5
73.2
4.7
14,470
7.1
8.1
0.12
0.10
95
Upper
Freeport ,
Cambria ,
Pa.
Lvb
4.3
77.9
1.1
3.7
1.3
0.12
11.7
0.05
0.07
0.02
0.15
1.1
18 5
68.7
11.7
13,440
3.9
7.1
0.08
0.05
95
Sewell ,
Fayette ,
W. Va.
Mvb
5.0
85.5
1.6
4.3
0.62
0.27
3.0
0.08
0.05
0.04
0.11
0.7
25 8
70 5
3.0
15,030
4.4
7.9
0.39
0.22
95
Upper
Kittanning ,
Cambria ,
Pa.
Mvb
Piper
4.8
82.2
1.5
3.7
0.7
0.13
7.1
0.08
0.07
0.01
0.12
0.6
20.4
71.9
7.1
14,400
2.2
10.7
0.17
0.03
95
Illinois #6,
Jefferson,
111.
Hvab
Orient #3
5.5
72.1
1.7
14.3
1.1
0.39
5.3
0.17
0.10
0.02
0.04
6.7
36 9
51.1
5.3
12,850
5.6
6.3
0.32
0.28
94
Pittsburgh,
Allegheny,
Pa.
Hvab
Champion #1
5.2
75.3
1.5
8.3
1.9
0.08
7.8
0.12
0.06
<0.01
0.09
1.6
36.1
54.5
7.8
13,520
4.9
5.3
0.38
0.36
95
Lower
Kittanning,
Indiana ,
Pa.
Hvab
4.8
74.2
1.3
6.2
2.3
0.16
11.2
0.15
0.21
0.02
0.08
2.0
28 4
58.4
11.2
13,210
1.9
4.6
0.06
0.03
97
Upper
Freeport ,
Preston,
W. Va.
Hvab
4.9
75.1
1.4
3.7
3.4
0.17
11.5
0.16
0.16
0.02
0.05
0.7
27 4
60.4
11.5
13,440
4.8
6.3
0.045
0.035
94
Upper
Freeport ,
Clarion,
Pa.
Hvab
Bish £3
5.1
72.3
1.4
7.2
2.1
0.13
11.9
0.15
0.27
0.02
0.05
1.6
34.5
52.0
11.9
12,930
1.6
2.3
0.06
0.04
99
Middle
Kittanning,
Clarion,
Pa.
Hvbb
Sunny Hill
4.4
55.6
1.0
9.9
6.5
0.14
22 6
0.24
0.45
0.04
0.1
3.2
35 8
38.4
22.6
10,150
3.5
4.6
0.019
0.015
98
00
1
s
TO
3
From BuMines RI 7260 "Chlorine in Coal Combustion"
o
-------
Coal - West Virginia Sewell seam
( 0. 27percent chlorine)
rj 1.—••»- —i • • •
600 800 1,000 1.200 1400 1,600. 1800 2.000
TEMPERATURE. K \
Figure 1-20. Equilibrium Distribution of Chlorine Compounds in
the Combustion Gas of a Sewall Coal at 20 Percent
Excess Air.
1-f
WALDEN RESEARCH CORPORATION
-------
is considerably higher. The average values calculated from the data in
Table 1-40 is HC1/SCL = 12 +_ 8%. This would represent a significant error
in the S02 measurement if an acid base titration were used for the analyses.
In order to examine this effect, we investigated the effect of HC1 on the
wet chemical methods for SCL in the pilot plant. This is described in the
following section.
1.8.4.2 Pilot Plant Results
The pilot plant results were obtained with the unit
firing gas and doping the effluent with both S(L and HC1. Runs were made
with two sets of impinger trains (filled with aqueous peroxide) in parallel,
For these runs, a total sample volume of ^ 15 liters was collected at a
flow rate of 1.0 liter/min. The data were taken at HC1 concentrations
varying from ^ 0 to 150 ppm with the S0_ concentration at approximately
150 ppm.
We computed the HC1 flow rates for the rotameter from
the air calibration and calculated the expected HC1 concentrations in the
duct from the stoichiometric gas flow, excess air and temperature. The HC1
concentration in the pilot plant flue gas was determined by bubbling a
stream of the effluent through impingers containing distilled water. The
chloride samples were analyzed using an Orion Model 94-17 chloride elec-
trode. The potential was measured with an Orion Model 701 digital milli-
volt meter. The data which were obtained are given in Table 1-41. The
agreement between the measured and calculated HC1 values is excellent.
The SOp levels were kept between 300 and 400 ppm which
are typical of effluents from fuel containing 0.5-1% sulfur.
The data obtained with S0? doping are given in Table
1-42. The barium chloranilate method yields results which are slightly
lower than Ba with no HC1 added. At 150 ppm HC1, no difference is ob-
served or expected between these two sulfate specific methods.
The NaOH titration which is specific for H* rather
than sulfate does show a marked difference compared to Ba between 0 and
1-89
WALDEN RESEARCH CORPORATION
-------
TABLE 1-41
COMPARISON OF MEASURED AND CALCULATED HC1
CONCENTRATIONS IN THE PILOT PLANT
. ppm HC1
Calculated from Measured with an Ion
Stoichiometry Selective Electrode
150 150
. "" 140
.. ... . . 140
90 " ' ' • -90
•',''.' ' -90
40 47
47
, _QQ WALDEN RESEARCH CORPORATION
-------
TABLE 1-42
EFFECT OF HC1 ON WET CHEMICAL METHODS FOR SO,
ppm SO? ppm SO? ppm SOo . ++,
Sample No. ppm HC1 KK * KK ++ KK z A (NaOH-Ba )
(NaOH) (Ba ) (Ba Chlor)
lOOa
lOOb
99a
99b
98a
98b
97a
97b
96a
96b
95a
95b
94a
94b
93a
93b
92a
92b
91 a
91b
90a
90b
87a
87b
86a
86b
85a
0
0
0
0
40
40
40
40
40
40
40
40
90
90
90
90
150
150
150
150
150
150
150
150
150
150
150
340
333
343
404
318
366
-
364
160
180
160
179
305
348
312
392
-
384
334
349
426
395
454
454
464
444
442
337
330
339
392
286
366
-
-
136
168
139
162
264
292
286
350
312
316
288
312
373
331
400
398
414
390
365
318
308
331
366
307
356
354
350
104
162
-
-
237
271
283
321
325
321
235
264
365
328
404
428
415
396
363
3
3
4
12
32
0
-
-
24
12
21
17
41
56
26
42
-
68
56
37
53
64
54
56
50
54
77
HC1 = X
Y = A
A = 5.42 + 0.35X (HC1)
Correlation Coefficient =0.90
1_g-| WALDEN RESEARCH CORPORATION
-------
150 ppm HC1 (Table 1-42). In this table, we have calculated the difference
between NaOH and Ba titrations (A) and compared this to the HC1 concentra-
tion. There is about 4 ppm difference at 0 ppm HC1 and about 60 ppm differ-
ence at 150 ppm HC1. The regression equation for ppm HC1 versus
A(ppm S02 NaOH-ppm S02 Ba++) is:
A(ppm S02 NaOH-ppm S02 Ba++) = 5.42 + 0.35 (HC1)
The correlation coefficient for these values is 0.90.
Although a 1:1 correlation between A and HC1 concen-
tration might be expected, the chemistry of the system suggests a 0.5:1
ratio as follows:
H202 + 2H+ + 2e~ ^ 2H20
2C1" - 2e~ ^ C12
C12 + H20 <=* HOC! + HC1
Overall H909 + 2HC1 + H90 ^ HC1 + HOC1 + 2H90
reaction * L * 6
The above reaction is spontaneous and indicates that the H900 reacts with
_* ^ *•
Cl to produce HOC!. Since this latter species is a weak acid, e.g.,
pK = 7.5, it will not react with NaOH (if bromphenol blue indicator is
used). Thus, the value of the slope of 0.35 is quite reasonable as sug-
gested by the above reaction. The data indicate that HC1 is a potential
interference with an acid base titration and that with low sulfur coals
containing chlorine, errors from 5 to 15% could be expected.
If we examine the regression equations in Table 1-38
for HC1 doping, we note that Ba or Ba Chlor vs NaOH have a considerable
bias in the equations (intercepts of 96 and 81, respectively, and slopes
which deviate considerably from unity). The sulfate method, however, show
reasonably good agreement with each other for HC1 doping.
•*
In another application, we have used Cl ion to decompose aqueous
, _g2 WALDEN RESEARCH CORPORATION
-------
The precision of the barium titration with HC1 does
not change and is about +. 2%. The CV for the NaOH titration and Ba Chlor
methods are slightly higher, e.g., 3.3 and 5.0%, respectively, than the
values found in the field indicating a possibility of an interference or
most probably, a small sample size (Table 1-43).
1.8.5 Effect of Organic Acids of SO,, Methods
Organic acids have been reported as minor constituents in the
effluents of coal-fired power plants. In units with poor combustion of
fuel, high levels of hydrocarbons (partially organic acids) may be present.
The doping system for the organic acids (acetic acid) was set
up in the pilot plant as shown in Figure 1-21. The acetic acid was heated
to a constant temperature (about 43°C). The concentration of acetic acid
could then be calculated from the vapor pressure of acetic acid at 43°C,
the flow rate of air through the impinger, and the total volumetric flow
in the duct (from stoichiometry). The acetic acid concentration was varied
by increasing the flow rate of air through the impinger and keeping the
acetic acid temperature constant. The acetic acid concentration in the
duct was varied from several hundred to 1000 ppm (acetic acid). The re-
sults are given in Table 1-44. Considerable difficulty was observed with
the NaOH bromphenol blue endpoint even with low concentrations of acetic
acid. The indicator color change was very gradual instead of sharp. This
may be due to the buffering action produced by the weak acid. The data in
Table 1-44 show that the barium ion titration and barium chloranilate
methods agree well with each other and do not seem to be affected by the
very high concentrations of acetic acid. The statistical data in Table
1-38 bear this out. The correlation coefficient is high, 0.97, intercept
is small and the slope is close to unity.
In the initial data (0 ppm HOAc) in Table 1-44, the NaOH titra-
tion gives lower SOp values than expected while Ba yields higher values.
The reason for this is not known. As the acetic acid is increased, the
NaOH gives slightly higher SQ~ values than either SOT method. The differ-
ence at about 1000 ppm AcOH is about 50 ppm or 20%. Levels this high are
WALDEN RESEARCH CORPORATION
-------
TABLE 1-43
PILOT PLANT DOPING - PRECISION
OF HC1 AND HOAc DOPING
Analytic Method
# Samples
mean ppm
so2
cv
Dopant
NaOH Titration
Ba Titration
Barium Chioranilate
NaOH Titration
Ba Titration
Barium Chioranilate
24
23
24
23
23
24
375
348
344
226
240
210
3.3%
2.0%
5.0%
3.3%
4.0%
4.3%
HC1
HC1
HC1
HOAc
HOAc
HOAc
1-94
WALDEN RESEARCH CORPORATION
-------
IO
en
m
m
o
i
8
TO
s
o
Greenburg Smith
Impinger
Flow Meter
Sampling Point (250°F)
Heating Tape
Thermometer
AcOH
Hot Test Section (525°F)
Figure 1-21. Doping System for Organic Acids.
Heat
xchanger
-------
TABLE 1-44
EFFECT OF ORGANIC ACIDS ON WET CHEMICAL METHODS FOR SO,
Sample No.
101-1
101-2
102-1
102-2
107-1
107-2
108-1
108-2
109-1
109-2
110-1
110-2
104-1
104-2
105-1
105-2
106-1
106-2
111-1
111-2
112-1
112-2
114-1
114-2
113-1
113-2
ppm HOAc
0
0
0
0
0
0
0
0
173
173
179
179
190
190
230
230
495
495
871
871
940
940
1045
1045
1360
1360
Theoretical
ppm SOp
280
280
320
320
313
313
277
277
260
260
268
268
285
285
346
346
376
376
316
316
341
341
278
278
363
363
ppm S02
(NaOH)
247
109
150
210
136
223
157
262
174
270
206
130
245
205
262
134
218
284
204
261
300
260
380
319
247
ppm SO?
(Ba++)
306
336
148
172
268
180
285
203
322
181
310
223
115
223
182
214
124
152
313
231
316
244
199
352
306
214
ppm S02
(Ba Chlor)
283
258
136
148
222
138
237
160
261
146
265
227
73
183
257
274
189
261
271
194
278
194
199
293
287
181
1_96 WALDEN RESEARCH CORPORATION
-------
not likely to be encountered in a coal-fired power plant. The difference
between the NaOH and the sulfate methods as evidenced in Table 1-44 is un-
doubtedly due to the obscuration of the endpoint mentioned earlier.
In the following paragraphs is a statistical evaluation of the
data in Table 1-44 via the student "T" distribution.
Comparison of Ba , Ba Chlor and NaOH Methods as a Function of
AcOH Level
The data in Table 1-43 represents 26 samples. Each set of data
represents analyses by each of the three methods on one effluent sample.
Each of the three combinations of paired methods were analyzed by the pro-
cedures outlined above. The NaOH data for sample 101-2 were omitted and,
hence, a sample of 25 data points was available for those combinations in-
volving NaOH method. In each case, the Di's were regressed on HOAc levels.
(a) Hypothesis: y, = y2
(i) NaOH method vs Ba++ method
and
t(a = 0.05)(24) = 2.064
Therefore, cannot reject hypothesis, i.e., no significant difference.
(ii) NaOH method vs Ba Chlor method
= t ' °-20
and
t(a = 0.05)(24) = 2.064
Therefore, cannot reject hypothesis, i.e., no significant difference.
(iii) Ba method vs Ba Chlor method
* = U/== 0-015
1-97
WALDEN RESEARCH CORPORATION
-------
and
t(d = 0.05)(25) = 2.060
Therefore, cannot reject hypothesis, i.e., no significant difference.
(b) Hypothesis: o^ = a^
(i) NaOH method vs Ba"1"1" method
f.l"4\2 = 1.06
and
F(d = 0.05)(24,24) = 2.27
Therefore, cannot reject hypothesis.
(ii) NaOH method vs Ba Chlor method
F (a = 0.05)(24,24) = 2.27
Therefore, cannot reject hypothesis.
(iii) Ba method vs Ba Chlor method
F =
and
F(o = 0.05)(25,25) = 2.27
Therefore, cannot reject hypothesis.
(c) Hypothesis: 6, = 0
(i) NaOH method vs Ba"1"1" method
B] = 0.04
Therefore, difference is not significant function of HOAc level
1 -98 WALDEN RESEARCH CORPORATION
-------
(ii) NaOH method vs Ba Chlor method
B] = 0.05
Therefore, difference is not significant function of HOAc level.
(iii) Ba method vs Ba Chlor method
B] = 0.002
Therefore, difference is not significant function of HOAc level.
1.9 SUMMARY AND CONCLUSIONS
The ion exchange step which was included in the procedures for remov-
ing cationic interferences from the sulfate specific methods leads to a con-
siderable loss of S0?. The problem appears to be a swelling of the resin
and subsequent irreversible adsorption of sulfate ions when peroxide is
present. The field data for the sulfate specific methods on trains a and
b had to be corrected for the loss of sulfate. In the pilot plant and
other stationary source measurements (Section 2), the ion exchange step was
eliminated and no corrections were necessary.
One of the three trains (train c) run in parallel which collected S03
by absorption in 80% isopropanol-20% water (IPA) was not flushed to remove
the dissolved S0?. The barium ion titration results from this train were
considerably lower indicating the need for removing residual S0? from the
SO- absorber.
Many of the field tests were conducted using critical orifices (Fig-
ure 1-5) as metering devices. The long-term calibration of a number of
orifices used in the field was checked frequently and was found not to
vary more than a few percent (Table 1-7) although as many as 25 to 50 S02
samples were run between calibrations. These devices appear to be very
well suited for field use.
A major problem with the sampling was encountered at the inlet to a
wet scrubber on a controlled coal-fired power plant. With the high
1-99
WALDEN RESEARCH CORPORATION
-------
sampling rate employed (3 to 4 liters/min), the dust (CaCO,) collected on
particulate filter and adsorbed SCL.
At the outlet of the wet scrubber, no major problems were encountered
with the sampling or analytical methods. The coefficient of variation (pre-
cision of replicate analysis) was 1 to 2% for barium ion and NaOH titration
but 4 to 5% for the barium chloranilate method. This was traced to an in-
A.A.
terference from the particulate matter (Ca ). This species does not inter-
fere with either titration method but would interfere with the colorimetric
procedure. Good agreement was obtained with all methods on analysis of
aliquots of the same SCL sample. Some variation between methods was noted
between trains, however.
At the oil-fired power plant, the coefficient of variation was 2 to 3%
for both NaOH titration and barium chloranilate but only 1 to 2% for barium
ion titration. All the methods are in excellent agreement with each other
indicating, as expected, no potential interferences with any of the methods.
The percent of theoretical SCL measured by Ba (train b) and Ba Chi or is
greater than 100%. The NaOH titration gives about 85% theoretical. The
difference is due to the correction of the sulfate methods for loss on the
ion exchange resin. The measured values of 85% of theoretical S02 seems
reasonable for these low S02 concentrations (400 ppm).
The coefficients of variation of the analytical methods at the coal-
fired power plant were the same as those for the oil-fired source. Good
agreement (correlation coefficients and least squares fits) were obtained
for all the methods. The percent theoretical values for SO, were 80% for
++
Ba or Ba Chi or and 73% for NaOH. Again, some of this difference may be
due to correction of the sulfate methods for the loss on the ion exchange
column.
The SO, results from controlled condensation method and the IPA ab-
sorption method differ considerably. The correlation coefficients vary
from ij 0.1 to 0.4. The reason for this is not known at the present time.
We suggest that additional testing and laboratory work on these S03 col-
lection methods be initiated.
1-100 WALDEN RESEARCH CORPORATION
-------
Tests on the effect of NO , HC1 and organic acids on the wet chemical
A
methods were conducted in the Wai den pilot plant. Nitric oxide has no ef-
fect on Ba and NaOH titrations or Ba Chlor even up to a ratio of NO/SCL =
++
3. The NaOH titration is effected by HC1 whereas Ba and Ba Chlor are not.
The effect would be most pronounced in low sulfur coals where errors as high
as 10 to 20% are possible. Organic acids have no effect on the accuracy of
any of the wet chemical methods, however, the endpoint of the NaOH titration
is gradual rather than sharp, possibly due to the buffering action of the
weak organic acid. The accuracy of the wet chemical methods is estimated
as 93 +_ 5% based on the pilot plant studies.
i_1Q1 WALDEN RESEARCH CORPORATION
-------
SECTION 2
OTHER STATIONARY SOURCES
WALDEN RESEARCH CORPORATION
-------
Section
TABLE OF CONTENTS
Title
Page
2.1 INTRODUCTION 2-1
2.2 GRAY IRON FOUNDRIES 2-2
2.2.1 Brief Description of Foundries 2-2
2.2.2 Sampling and Analytical Results 2-6
2.2.3 Conclusions 2-26
2.3 KRAFT PULPING 2-26
2.3.1 Sources of S02 in a Kraft Mill 2-26
2.3.2 Sampling and Analytical Results 2-30
2.3.3 Conclusions 2-39
2.4 IRON AND STEEL 2-40
2.4.1 Brief Description of the Iron and Steel Industry .. 2-40
2.4.2 Sampling and Analytical Results 2-44
2.4.3 Conclusions 2-62
2.5 SMELTING 2-62
2.5.1 Brief Description of Smelting Processes 2-62
2.5.2 Sampling and Analytical Results 2-66
2.5.3 Conclusions 2-83
2.6 SULFURIC ACID PRODUCTION 2-83
2.6.1 Brief Description of Sulfuric Acid Plants 2-83
2.6.2 Sampling and Analytical Results 2-87
2.6.3 Conclusions 2-96
2.7 RECOMMENDATIONS FOR FUTURE WORK 2-96
REFERENCES 4 2-97
APPENDIX A PROCEDURE FOR SAMPLING AND ANALYSIS A-l
APPENDIX B POTENTIOMETRIC DETERMINATION OF S02 IN FLUE GASES WITH AN
ION SELECTIVE LEAD ELECTRODE B-l
APPENDIX C SOLID SORBENT STUDIES C-l
APPENDIX D PRECISION OF BARIUM ION TITRATION AND BARIUM CHLORANILATE
FOR VARIOUS ANALYSTS D-l
WALDEN RESEARCH CORPORATION
-------
LIST OF ILLUSTRATIONS
Figure f Caption Page
2-1 Sulfur Balance for Gray Iron Foundry .............. ....... 2-7
2-2 Schematic of Gray Iron Foundry .................... . ...... 2-8
2-3 Detailed Layout of Gray Iron Foundry Sampling Sites ...... 2-9
2-4 Schematic of Standard and Modified EPA SOX Sampling
Trains .......... ....... .............. . ...... . _____ ........ 2rl5
2-5 Typical Appearance of Parti culates and Chemical Map of
................................................ ..... 2-21
2-6 Rescanning Electron Micrographs .... .................. _____ 2-22
2-7 Oscilloscope Traces of Si , S and Fe ............... ....... 2-23
2-8 Additional Oscilloscope Traces of Si, S and Fe ....... ---- 2-24
2-9 Typical Flow Diagram for Kraft Pulping ................. .. 2-28
2-10 Gaseous Sulfur Distribution ...................... ........ 2-31
2-11 Schematic of Kraft Mill Operation .. ................. . ---- 2-32
2-12 Detailed Layout of Kraft Mill Sampling Sites ............. 2-34
2-13 Schematic of Coke Oven Site and Sampling Port ....... . ---- 2-47
2-14 Sintering Plant Schematic and Test Site ................... 2-48
2-15 Schematic of Open Hearth Unit and Sampling Ports . ----- .... 2-49
2-16 Schematic of the Copper Smelting Operation ............... 2-68
2-17 Duct Cross-Section Showing Probe Penetration (R&R)
(Copper Roaster and Reverbatory) ....................... . . 2-69
2-18 Schematic of the Lead Smelting Operation .. ...... ......... 2-70
2-19 Simplified Flow Diagram of Typical Lead-Chamber Process
for Sulfuric Acid Manufacture (Based on Use of Elemental
Sul f ur as the Raw Materi al ) ...................... . ....... 2-85
2-20 Gas Flow Diagram for Typical Sulfur-Burning Contact Plant
in which Air Quench is Used for Part of Converter Inter-
stage Cooling ...... ............................... ...... , 2-86
2-21 Sulfuric Acid Plant Layout ........................ . .. ____ 2-89
i
LIST OF TABLES
Table - Title Page
2-1 Dust and Fume Emissions from Gray Iron Cupolas ........... 2-3
1 V WALDEN RESEARCH CORPORATION
-------
LIST OF TABLES (continued)
Title Page
Qualitative Spectrographic Analysis of Two Samples Taken
from a Baghouse Serving a Gray Iron Cupola Furnace 2-5
2-3 Supplementary Data for the Gray Iron Foundry 2-10
2-4 S02 Test Results from a Gray Iron Foundry 2-11
2-5 Statistical Results for S02 Tests from a Gray Iron
Foundry 2-17
2-6 Semiquantitative Spectrographic Analysis of an S02 Im-
pi nger Samples from a Gray Iron Foundry 2-18
2-7 Determination of Trace Metals in SOg Impingers from a
Gray Iron Foundry by Atomic Absorption Analysis 2-19
2-8 Results of X-Ray Analysis 2-25
2-9 Characteristics of Kraft Pulping 2-27
2-10 S02 Emissions from Kraft Pulping Process 2-29
2-11 S02 Test Results from a Pulp and Paper Mill 2-35
2-12 Supplementary Data from the Pulp and Paper Mill 2-38
2-13 Emissions of Sulfur Oxides to the Atmosphere from the
Iron and Steel Industry (1967) 2-41
2-14 Composition of Coke Oven Gas 2-42
2-15 Supplementary Data from the Iron and Steel Plant 2-46
2-16 S02 Test Results from an Iron and Steel Plant (Coke
Oven) 2-50
2-17 S02 Test Results from an Iron and Steel Plant (Coke
Oven) 2-52
2-18 S02 Test Results from an Iron and Steel Plant (Sinterer) . 2-53
2-19 S02 Test Results from an Iron and Steel Plant (Open
Hearth) 2-55
2-20 Statistical Analysis for S02 Test Results from an Iron
and Steel Plant 2-56
2-21 Semiquantitative Spectrographic Analysis of an S02 Im-
pinger Sample from an Iron and Steel Plant (Coke Oven) ... 2-57
2-22 Sulfur Balance for Coke Oven Operation (Based on Produc-
tion of One Net Ton of Coke) 2-59
2-23 Sulfur Balance for Sintering Machine Operation (Based on
the Production of One Net Ton of Sinter) 2-61
2-24 Sulfur Dioxide Emissions 2-64
WALDEN RESEARCH CORPORATION
-------
LIST OF TABLES (continued)
Table Title Page
2-25 S02 Test Results from a Copper Smelter (Roasting and
Reverberatory) ........................................... 2-71
2-26 Supplementary Data for the Copper and Lead Smelters ...... 2-73
2-27 S02 Test Resul ts from a Lead Smel ter ..................... 2-74
2-28 Statistical Analysis for S02 Test Results from Smelters .. 2-76
2-29 Semi quantitative Emission Spectrographic Analysis of
Impinger Samples from Smelting Operations ................ 2-77
2-30 Atomic Absorption Results from Analysis of S02 Impinger
Liquid from Copper and Lead Smelting .............. . ...... 2-78
2-31 S03 Test Results from Cu and Pb Smelting Operations ...... 2-80
2-32 Log of Operations for Copper and Lead Smelters ........... 2-82
2-33 Collection of H2S04 Mist from a Sulfur-Burning Contact
Sulfuric Acid Plant with Fiber Mist Eliminators .......... 2-88
2-34 S02 and $03 Test Results from a Sulfuric Acid Plant ...... 2-90
2-35 S02 and SOs Test Results from a Sulfuric Acid Plant ...... 2-95
WALDEN RESEARCH CORPORATION
-------
2.1 INTRODUCTION
The second phase of this program was started in November 1971 with ad-
ditional funds from EPA. The purpose of this portion of the program was to
investigate the feasibility of using SO methods (developed for sampling
A
and analysis of fossil fuel combustion effluents) in other industries in-
cluding some Category II sources. These included:
Gray iron foundries
Kraft pulping
Iron and steel
Smelting
Sulfuric acid production
The program plan was to run the EPA procedure, evaluate the data, and repeat
with modification of the sampling and/or analytical procedures as required.
The S0? samples were analyzed by three different analytical procedures.
Each was subject to different interferences. For example, the two sulfate
specific methods were Ba and barium chloranilate which have interferences
from sodium and potassium (Ba ) ions, or zinc, calcium, lead (Ba Chlor)
ions, respectively. The NaOH titration does not have any interferences
from metals, however, carbonates, oxides or other basic and strongly acidic
species will interfere.
For any source sampled, one could decide whether modifications were re-
quired on a statistical basis. For example, if high correlation coefficients
with least squares slopes near unity were obtained after intercomparison of
all the methods, we assumed that no interferences were present since each
method has its own characteristic interference.
Early in the evaluation program, we found that a modification to the
sampling train in the field (inclusion of a fine particulate filter) eli-
minated the necessity for changes to the analytical methods which could im-
prove accuracy or precision. The majority of the tests, therefore, were
conducted with and without the fine particulate filter to determine the ef-
fect of interferences on the methods. Following statistical analysis of
the data, we could determine whether any modifications to the original EPA
method were required.
2-1 WALDEN RESEARCH CORPORATION
-------
In the following sections, we discuss, for each source, the character-
istics, the sampling and analytical methods including test data, and the
conclusions and recommendations.
2.2 GRAY IRON FOUNDRIES
2.2.1 Brief Description of Foundries
Gray iron foundries produce metal castings from melted scrap
metal such as automobile parts, pig iron, etc. A number of different
types of furnaces are used in the ferrous and non-ferrous foundry industry
each having a special application and a characteristic contribution to at-
mospheric pollution.
Of the numerous types of furnaces, the cupola is the one most
commonly used. Approximately 90% of the gray iron melting furnaces are
cupolas and only 15 to 18% of these have air pollution control devices in-
stalled (Giever, 1970).
One of the important characteristics of cupola effluent is the
temperature. This is a factor to consider in the design of control equip-
ment and has been one of the primary problems in the collection of satis-
factory data on other characteristics of the effluent. The temperature of
the effluent has been found to exceed 2000°F in some instances. This high
temperature not only hampers evaluation but presents special problems of
control that are difficult to attain, costly to install, and expensive to
maintain. Cupola effluents, when uncontrolled, may be generally character-
ized as consisting of large quantities of solid particles, dark, often oily
smoke, gas and fumes released to the atmosphere at temperatures often too
high to permit the standard types of stack effluent evaluation.
Cupolas without air pollution control devices are reported to
emit an average of 17 pounds of particulates per ton of iron melted based
on 20 tons/hour production (Giever, 1970). Some characteristic emission
data on cupolas is given in Table 2-1.
2-2 WALDEN RESEARCH CORPORATION
-------
TABLE 2-1
DUST AND FUME EMISSIONS FROM GRAY IRON CUPOLAS
Tent N'..'.
Cupuln data
I:isi ?.l) (i
.:o tci -H (i
-• -14 |i
S|KTilic: gravity
1
60
.
7/1
s, .'no
«., soo
1.H85
-
-
_
-
None'
.-
0. 913
-
6S
-
IS. 1
I.. •<
1^. S
JJ. '1
i>>. 3
(.51
J
<7
1, 9"0
f .<.'./!
K. ^n
•', "i^O
1, -100
1^. J
.
-
-
Nonp
-
1. U
.
62.4
-
17. i
*. =•
10. 1
17. '<
•46. •>
I. 7S
3
63
7, ?00
10. 1 /I
19. 100
30, MX)
213
2.B
.
.
-
None
-
0.413
.
10S
-
23.6
4.f-
•1. S
'?. *
^7.9
4
56
-
.-..5/1
J.J.6^0
17.700
ilO
4. 7
12. 7
0
67. *>
Banhouse
1. 33
0.051
1V7
7. 7
96
• <> S . S
6. J
?.. ^
10. d'
^,S.7b
-
5
42
-
9.2/1
14, 000
20,300 '
430
5.2
1 1.8
0. 1
67. 3
Klloc prrcip
afterburnor
2.973
0.0359
184.7
6.2 1
96. 6
-
-
.
6
60
-
9.6/1
S6. 900
2i,cno
222
-
-
_
-'
lla^hr»us(r
0. <92
0.0456
70. (,
^.2
BS. 4
-
-
.
_
-
7
48
-
7.4/1
16. SOO
8.. 130
432
.
- •
.
-
Elcc Proc ip
1. 52»
0. 186
110
I.'.. 2
87. 7
-
-
•
.
-
'' F'rc.m 20 to 50 \i.
''Orcatcr than 50 ^.
From J. A. Danielson, Air Pollution Engineering Manual. U.S. Dept. HEW,
Pub. No. 999-AP40 (1967T
2-3
WALDEN RESEARCH CORPORATION
-------
Particulate material emitted from the collectors will be composed
of coke, limestone, foreign particles associated with the scrap along with
metallic oxides, fumes and sulfides. Additional data on the metallic ele-
ment composition of cupola emissions collected in a baghouse is shown in
Table 2-2.
A study of foundry cupolas in the Los Angeles area (Danielson,
1967) showed up to 24% by weight of the solids were below 30y. It was also
reported that 74% of the total weight was greater than 325 mesh. Particles
of 1/4 to 3/8 inches in diameter may be blown from the stack of an uncon-
trolled cupola from the blowdown operation.
In addition to particulate matter, the effluent from cupolas con-
tains gases which are primarily carbon dioxide and nitrogen with some excess
oxygen and varying amounts of sulfur dioxide, carbon monoxide and hydrocar-
bons.
Most states have recently adopted regulations requiring strin-
gent control of cupola particulate emissions. Once these regulations become
effective, generally by 1974-75, the sampling environment in the effluent
gas stream will change greatly. The two most commonly employed control
techniques used in foundries are wet scrubbers (venturi) and baghouses. In
both the exit gases will be at considerably lower temperatures than are
characteristic of the uncontrolled top gases. The contact with water in the
scrubber inherently produces cooling, with accompanying saturation of the
gases. In order to prevent damage to the filter materials, the top gases
are cooled prior to entering the baghouse in a water quench step.
The final effluent stream will probably be free of carbon mon-
oxide, unlike the top gases where carbon monoxide levels can reach 15% or
more. An afterburner in which the carbon monoxide is oxidized to carbon di-
oxide is usually installed to reduce the explosion hazard and to control
monoxide emissions to the atmosphere. The afterburner can be installed
either in front of or behind the particulate control unit.
The primary source of sulfur in these operations is from the
coke charge. Coke of 0.6% sulfur could result in the emission of 1.6 to 2.5
2-4
WALDEN RESEARCH CORPORATION
-------
TABLE 2-2
QUALITATIVE SPECTROGRAPHIC ANALYSIS OF TWO SAMPLES TAKEN
FROM A BAGHOUSE SERVING A GRAY IRON CUPOLA FURNACEa
Element
Aluminum
Antimony
Boron
Cadmium
Calcium
Chromium
Copper
Gallium
Germanium
Iron
Lead
Magnesium
Manganese
Molybdenum
Nickel
Potassium
Silicon
Silver
Tin
Titanium
Zinc
Approx amount,
%
Sample A
0. SI
0.24
0.050
0. 13
0. 16
0.022
0. 42
0. 017
0.018
6.0
17. 0
0.29
1. 0
0. 0068
0.023
1.5
8.6
0.0091
D. 41
0. 019
7. 1
Approx amount,
%
Sample B
1. 1
0. 24
0. 054
0. 064
0.25
0.019
0. 32
0. 019
0. 015
7. 5
17.0
0. 30
0.81
0.0075
0.022
1. 2
15.0
0. 0089
0. 38
0. 034
5.9
aThesc data are qualitative only and require
supplementary quantitative analysis for actual
amounts of the elements found to be present.
These arc the same samples as given in
Table 77.
These data are qualitative only and require supplementary
quantitative analysis for actual amounts of the elements
found to be present.
From J. A. Daniel son, Air Pollution Engineering Manual, U.S,
Dept. HEW, Pub. No. 999-AP40 (1967).
2-5
WALDEN RESEARCH CORPORATION
-------
pounds of SOp per ton of melt. The concentration of S(L (if all the sulfur
in the coke goes into the gas stream) in the top gases is 300 to 470 ppm
based on an average quantity of gas of 35,000 scf per ton of melt (Environ-
mental Engineering, 1971). It should be noted, however, that 60,000 scf per
ton of melt is a more reasonable estimate. On the basis of a larger gas
volume, the estimated S02 concentration would be 200-300 ppm. A sulfur
balance on foundries is shown in Figure 2-1. The data of Cowan (1971) indi-
cates only ^ 20-40 ppm of SO in the effluent. Cowan's S09 value is con-
/\ C.
siderably lower than that estimated by Environmental Engineering (1971).
The reason for this difference is discussed in some detail in the following
section.
2.2.2 Sampling and Analytical Results
Two different locations were selected for sampling at the gray
iron foundry. These are shown in Figure 2-2. A detailed layout of the
sampling sites is given in Figure 2-3. The wet scrubber is typical of a
control system used on a foundry. (The liquid in the high pressure-drop
venturi scrubber is water.) Some supplementary data for the foundry is
given in Table 2-3. A few SO,, tests were run at the scrubber outlet. The
sampling system was the conventional EPA SO train with no modifications
/\
(Federal Register, 1971). The first five runs in Table 2-4 were taken at
the outlet of the high pressure drop (35" HLO) venturi scrubber. Two trains
were run in parallel off the same probe. These are reported as runs A and B
in Table 2-3. The sampling rate for each train was about 3 liters/min for a
20-30 minute period. The major problem encountered in sampling was that the
glass wool filter in the probe became plugged. The glass wool had to be re-
placed after each run. In addition, a large amount of the particulate matter
was carried into the impingers. The glass wool filter (plug) in the probe
proved to be ineffective for removing the particulate matter from foundry
effluent (see previous data on particle size section).
Before the impinger samples could be analyzed in the laboratory,
the particulate matter had to be removed by filtration through a 0.45y
Millipore cellulose acetate filter. Due to the low S02 levels (runs 1-5),
the samples were analyzed by the barium ion titration only.
2-6 WALDEN RESEARCH CORPORATION
-------
IN
7%
Refractory
SULFUR
OUT
5% Particulates 4% SO
73% Iron
Figure 2-1. Sulfur Balance for Gray Iron Foundry.
2-7
WALDEN RESEARCH CORPORATION
-------
ro
CO
I
o
m
73
m
o
x
8
TO
s
Sampling
Port
\
// ' '//Y '/ '/// V ' // /
Figure 2-2. Schematic of Gray Iron Foundry.
O
-------
Cupola
Sampling Port
Probe
Probe
Stack
Sampling Ports
Figure 2-3. Detailed Layout of Gray Iron Foundry Sampling Sites.
2-9
WALDEN RESEARCH CORPORATION
-------
TABLE 2-3
SUPPLEMENTARY DATA FOR THE GRAY IRON FOUNDRY
Run No
1-3
4-5
6-8
9-11
12-15
16-20
21-24
25-27
28-33
34-36
70-76
77-84
*
Stack
+Stack
Date
11/15/71
11/16/71
11/19/71
11/23/71
11/24/71
11/29/71
11/30/71
12/1/71
12/2/71
12/3/71
1/12/72
1/13/72
temperature = 160°F
temperature = 300°F
Sampling Location
*
Scrubber Outlet
Scrubber Outlet
Hot Stack+
Hot Stack
Hot Stack
Hot Stack
Hot Stack
Hot Stack
Hot Stack
Hot Stack
Hot Stack
Hot Stack
early morning, >1500°F
% co2
13.0
14.2
16.2
16.2
15.5
15.2
16.8
15.6
15.0
15.8
•
--
midafternoon.
% o2
3.3
3.8
0.5
1.8
0.5
0.4
0.5
0.0
0.5
0.5
--
--
2-10
WALDEN RESEARCH CORPORATION
-------
TABLE 2-4
SCL TEST RESULTS FROM A GRAY IRON FOUNDRY
Sample No.
lAt
IBt
2B
3A
3B
4A
4B
5A
5B
6A
6B
7A
7B
8A
8B
9A
9B
10A
10B
11A
11B
12A
12B
13A
13B
14A
14B
15A
15B
16A
16B
17A
17B
18A
18B
19A
19B
20A
20B
21A
21 B
22A
22B
ppm S02
(NaOH)
27
32
46
46
0
0
96
84
68
51
78
No endpoint
ppm S02
7
5
22
22
12
18
55
41
50
71
55
82
76
0
20
130
112
94
75
116
92
128
142
42
34
76
73
32
34
123
122
9
10
14
16
92
90
0
118
89
102
6
6
ppm S02
(Ba Chlor)
87
55
77
81
32
18
169
157
147
169
186
182
133
71
18
23
77
78
42
51
125
115
9
10
17
17
103
158
0
221
183
226
15
19
WALDEN RESEARCH CORPORATION
-------
TABLE 2-4 (continued)
Sample No.
23A
23B
24A
24B
25A
25B
26A
26B
27A
27B
28A
28B
29A
29B
30A
306
31 B
32A
32 B
33A
34A
34B
35A
35B
36A
36 B
70A*
70B*
71 A*
71 B*
72A*
72B*
73A*
73B*
74A*
74B*
75A*
75B*
76A*
76 B*
77A*
77B*
78A*
78B*
79A*
79B*
ppm S02
(NaOH)
47
54
94
89
26
26
48
53
45
41
25
32
39
43
56
63
ppm S02
/ D ^ TT i
\ DO /
10
10
31
36
4
4
33
34
110
118
70
43
101
60
201
84
82
43
44
88
19
22
6
18
83
70
203
183
76
76
51
53
78
77
28
27
43
47
54
51
19
23
25
25
34
38
ppm S02tf
(Ba Chlor)
11
13
152
89
23
3
65
39
118
127
76
49
228
144
377
101
198
52
56
86
14
17
7
13
60
89
201
171
57
59
51
39
69
67
19
20
31
35
37
42
17
19
19
19
27
28
2-12 WALDEN RESEARCH CORPORATION
-------
TABLE 2-4 (continued)
Sample No.
ppm S02
(NaOH)
tt
ppm SO?
tt
ppm
(Ba Chlor)
80A*
SOB*
81 A*
81B*
82A*
82B*
83A*
83B*
84A*
84 B*
64
69
38
40
83
84
125
143
42
39
65
71
34
34
57
61
109
125
41
40
57
63
19
27
50
55
105
119
47
29
with fine particulate filter
A and B represent results from two parallel trains run off the same probe.
t-f. ++
' These analytical methods (NaOH, Ba , Ba Chlor) were run on aliquots of
the same sample.
2-13
WALDEN RESEARCH CORPORATION
-------
The sample train modification necessary for sampling the hot
(cupola) stack involved replacement of the pyrex liner with a vycor probe
liner since the stack temperatures were generally in excess of 1500°F.
Tests 6 to 11 were analyzed by all three methods. The NaOH titration gave
low results due presumably to the presence of basic components (metal oxides,
metal carbonates, etc.) in the flue gas. This method was discontinued after
run 11.
*
The SO- results obtained by absorption in 80% isopropanol-20%
water were very low and ranged from about 2-5 ppm by either the barium ion
titration or barium chloranilate methods.
The color change in the barium ion titration method was very un-
usual for samples 6-8 in that a green color was observed when the indicator
was added and the titration endpoint was pale yellow. The barium ion titra-
tion endpoints for the remaining runs (9-36) had the normal color change but
were very difficult to detect.
The barium chloranilate method gave considerably higher results
for SOp than the barium ion titration for runs 6-36. The average S02 con-
centration (runs 4-36) was 90 ppm (Ba Chlor) versus 57 ppm for runs 1-36 by
barium titration.
Since we obtained considerably different S02 results by analysis
of aliquots of the impinger samples with two methods each specific for sul-
fate, we decided to modify the sampling train. We anticipated that the prob-
lem could be eliminated by reducing the quantity of particulate matter
reaching the impingers. This would be true only if the reactions leading to
poor precision and accuracy occurred in the condensed phase (in the impinger),
The SOp train was modified by placing a heated high efficiency
(0.3y) glass fiber filter between the probe and the first impinger. The two
trains are shown in Figure 2-4. The results (runs 70-84) are given in
The condenser was not used because of problems (plugging) expected with the
high particulate loadings.
2-14 WALDEN RESEARCH CORPORATION
-------
Inclined
Manometer
Sample Gasi
Metering f
System i
-Type Pitot Tube
Midget
Impingers
Heated, Glass- _x
Lined Probe
Glass Wool Plug —-
Standard EPA SO Sampling Train
A
Vacuum
Pump
Inclined
Manometer
S-Type Pi tot Tube
Sample Gas
Metering
System
Midget
Impingers
• Do CL
Glass Wool Plug.
Heated, Glass-—-
Heated, Glass- Lined Probe
Fiber Filter
STACK
Modified EPA SOV Sampling Train
/\
Figure 2-4
2-15
WALDEN RESEARCH CORPORATION
-------
Table 2-4. With this modification, the agreement between the barium chlor-
anilate and barium ion titration methods is excellent. Note that even the
NaOH titration yields results which agree well with the methods specific for
sulfate. The color change associated with the barium ion titration was ex-
tremely sharp and considerably improved. The endpoint change was even
clearer (at these low SCL levels) than that obtained at other sources where
no filter had been used. The problematic endpoint occasionally noted in the
titration for this method may be due to trace metallic impurities.
Statistical analysis of the results from Table 2-4 is shown in-
Table 2-5. The improvement is clearly demonstrated by the higher correla-
tion coefficients and regression equations with slopes close to unity for
the filtered samples. The change in the CV for these methods is quite dras-
tic. Samples of the impinger liquid (filtered and unfiltered) were analyzed
by emission spectroscopy. The semi-quantitative results are given in Table
2-6. The large CV's (poor precision) for barium chloranilate are due to the
high concentrations of metallic species such as Zn, Pb, Ca which are known
interferences. In the unfiltered samples the sum of the concentrations of
the metallic species is approximately equal to the SOp concentration. The
CV of the filtered sample analyzed by barium chloranilate is still high for
this concentration (3-5% would be expected if no interference were present).
The traces of Zn and Ca still found in the samples may be responsible for
this. The interference in the barium ion titration method is not quite as
severe as the barium chloranilate but there is a marked difference between
filtered and unfiltered samples. A scan of the data in Table 2-6 indicates
that the high Na levels which are reduced with filtration may be the inter-
ference. This is confirmed by the data in Table 2-7 which shows that the
Na concentration is 4-5 times the SCu concentration. The reduction in the
concentration of Na by the fine particulate filter is responsible for the im-
proved CV of the Ba titration. Note that the CV of the filtered sample is
only slightly higher than found for combustion sources if no interference
were present. The NaOH titration which has no metal ion interferences has
excellent precision and agreement with the sulfate methods once the metal
oxides or carbonates are removed by the fine particulate filter.
WALDEN RESEARCH CORPORATION
-------
TABLE 2-5
STATISTICAL RESULTS FOR S02 TESTS FROM A GRAY IRON FOUNDRY
ro
i
^
s
r~
O
m
z
m
m
a
I
n
o
73
TJ
O
TO
Run No. Regression Equation Coefficient
1-36 ppm S02 (Ba Chlor) = 0.826
-3.63 + 1.54 ppm S02 (Ba )
1-36
4-36
70-84 ppm S02 (Ba++) = 0.993
-8.84 + 1.09 ppm S02 (Ba Chlor)
ppm S02 (Ba ) = 0.938
3.68 + 1.08 ppm S02 (NaOH)
ppm S02 (NaOH) = 0.930
1.06 ppm S02 (Ba Chlor) + 12.9
Coefficient of .. ,„ cn
Variation of M*an S°2
Analytical (™'\ Kemarks
Method lppm;
With glass wool
filter only
Ba++ titr. = 57
11%
Ba Chlor = 90
27%
With heated fine
parti cul ate fil
ter and glass
wool filter
Ba++ titr. = 62
3.4%
Ba Chlor = 53
8.8%
NaOH titr. = 58
0.3%
o
z
-------
TABLE 2-6
SEMIQUANTITATIVE SPECTROGRAPHIC ANALYSIS OF AN SC>2 IMPINGER
SAMPLES FROM A GRAY IRON FOUNDRY
PO
1
00
1
o
m
TO
m
-------
TABLE 2-7
DETERMINATION OF TRACE METALS IN S02 IMPINGERS FROM A
GRAY IRON FOUNDRY BY ATOMIC ABSORPTION ANALYSIS
Sample
A
B
C
D
E
Na
210
240
8.5
7.7
BD
Zn
19
*
3.0
*
0.7
*
0.5
*
0.5
ppm
Ca
14
10
BD
BD
BD
*
Pb
20
BD
BD
BD
BD
*
Cu
0.3
0.3
0.3
0.3
BD
Remarks
Without fine parti cul ate
f 1 1 ter
Without fine parti cul ate
filter
With fine parti cul ate filter
With fine parti cul ate filter
With fine parti cul ate filter
*
Quantitative emission spectroscopy
BD = Below Detectable = 3.6 ppm for Na and Ca by AA
= 3.0 ppm for quantitative emission spectroscopy for
Pb
2-19
WALDEN RESEARCH CORPORATION
-------
Some samples of particulate which was removed by filtration from
the S02 samples 1-36 were sent to the Environmental Protection Agency for
electron microprobe analysis. The results are shown in Figures 2-5 to 2-8.
There was a considerable difference in the particulate loading which was re-
flected in the intensity of the sulfur emission line (oscilloscope traces).
The sulfur X-ray scans show only a general distribution of sulfur throughout
the sample. The apparent higher concentration of sulfur across the filter is
due to severe topographical unevenness. The elements present in the partic-
ulate matter are given in Table 2-8. It is interesting to note that sulfur
is the primary element in the particulate matter.
Environmental Engineering (1971) predict SCL concentrations of
300-470 ppm if all the sulfur in the coke were emitted as S02. The material
balance of sulfur in foundries by Cowan (1971) (see Figure 2-1) predicts
that about 10% of the sulfur in the coke is emitted as gaseous SO . Cowan
A
indicates that a considerable amount (5%) of sulfur is retained in the par-
ti cul ate matter. The majority of the sulfur ends up in the molten iron (see
Figure 2-1) and that 20-40 ppm of sulfur oxides are present in the effluent.
The average of the "filtered" samples is about 60 ppm. The slightly higher
SOp values may be the result of the charge in this particular foundry. Auto-
mobile engine blocks are the major starting material. These are quite oily
and the low but significant sulfur content of the lubricating oil may account
for this difference. The agreement with Cowan's predictions is quite en-
couraging.
Ryason and Harkins (1967) have studied a method for reducing SO^
to sulfur with carbon monoxide according to the following reaction:
2CO + S02 + 2C02 + 1 S2
This reaction may account for the elemental sulfur found in the particulate
matter by electron microprobe analysis. If CO is present at > two times the
S02 level, COS could be formed by the reaction:
C0 + 2~ S2 "* COS
2-20 WALDEN RESEARCH CORPORATION
-------
ro
i
TN3
2
»
8
73
s
TO
Scanning Electron Micrograph of
Particulate Material From Iron Smelter
Cupola. Filter #30A
Magnification 950X
Fe X-Ray Scan of Area Shown
In Micrograph
Sulfur X-Ray Scan from Area in Micrograph
Figure 2-5. Typical Appearance of Participates and Chemical Map of Fe.S.
-------
no
i
ro
ro
I
m
TO
m
in
o
x
o
o
•33
TJ
o
73
Scanning Micrograph of Particulates on
Filter 30A
Fe X-Ray Scan of Area in Micrograph
Sulfur X-ray Scan of
Area in Micrograph
Figure 2-6. Rescanning Electron Micrographs.
o
-------
ro
i
oo
EDX Oscilloscope Trace From Blank
EDX Oscilloscope Trace from Filter 14A
I
o
m
in
m
>
3)
o
o
o
33
-a
o
33
O
2
EDX Oscilloscope Trace From Filter 19B
EDX Oscilloscope Trace From Filter 21B
Figure 2-7. Oscilloscope Traces of Si, S and Fe.
-------
EDX Oscilloscope Trace From Filter 27A
EDX Oscilloscope Trace From Filter 30A
Figure 2-8. Additional Oscilloscope Traces of Si, S and Fe.
2-24
WALDEN RESEARCH CORPORATION
-------
TABLE 2-8
RESULTS OF X-RAY ANALYSIS
All Five Filters
Type of X-Ray Analysis_
EDX
Elements Present
Estimated Amount
Si
Cl
Fe
Cr
Ni
Ti
M
BDT
See accompanying photographs.
Code
Primary element - P
Secondary - S
Minor - M
Trace - T
2-25
WALDEN RESEARCH CORPORATION
-------
This species may be present in foundry effluents.
2.2.3 Conclusions
A number of modifications must be made to the EPA procedure for
sampling sulfur oxides from foundries. These include a vycor or quartz
probe liner to withstand the high temperatures and a fine particulate filter
must be used in conjunction with the glass wool plug. One definite advan-
tage is that the barium ion titration procedure gives sharper endpoint de-
terminations. No modifications to the analytical methods are necessary if
the fine filter is employed. Only a fraction (^ 15%) of the sulfur in the
coke appears as SOp. COS may be a product in the effluent of a foundry
based on the reaction of excess CO with S0?.
2.3 KRAFT PULPING
2.3.1 Sources of SO,, in a Kraft Mill
The kraft pulping process is responsible for the major portion
of all pulp manufactured at the present time. The processing which involves
hydrolysis of lignin produces odorous compounds such as sulfides, disulfides,
mercaptans, etc. Some characteristics of the kraft pulping operation are
given in Table 2-9. The recovery of the spent (black) liquor is a very im-
portant facet of the kraft process and is the largest potential source of
S0«. The black liquor is concentrated, burned, and limed as indicated in
Figure 2-9. Referring to the flow diagram (Figure 2-9), the following table
(Table 2-10) from Environmental Engineering (1971a) indicates the approximate
range of S02 emissions.
From Table 2-10, we see that the major source of S02 emissions is
from the recovery furnace where the black liquor is burned. The logical
sampling locations would be before the direct contact evaporator (if pos-
sible) and at the outlet of the recovery furnace. Some typical conditions
in a recovery boiler flue gas are as follows:
WALDEN RESEARCH CORPORATION
-------
TABLE 2-9
CHARACTERISTICS OF KRAFT PULPING
Cellulosic raw material
Almost anv kind of wood, soil or hard
Principal reaction in digester
Composition of cooking liquor
Cooking conditions
Hydrolysis of lignins to alcohols and
acids; some mercaptans formed
12.5% solution of NaOH, NajS, and
NanC03. Typical analysis of solids:
58.6% NaOH, 27.1% N^S, 14.3%
Na^O;;. Dissolving action due to
NaOH and NajS. NajCO, inactive
and represents the equilibrium residue
between lime and NajCC^ in the for-
mation of NaOH
Time 2-5 hr; temp. 340-350 F; pressure
100-135 psi
Chemical recovery
Materials of construction
Pulp characteristics
Typical paper products
Most of process is devoted to recovery of
cooking chemicals, with incidental
recovery of heat through burning
organic matter dissolved in liquor
from wood; chemical losses from svs-
tem are replenished with, salt cake,
Na.2S04
Digesters, pipelines, pumps, and tanks (an
be made of mild steel or, preferably,
of stainless
Brown color; difficult to bleach; strong
fibers; resistant to mechanical refining
Strong brown bag and wrapping; multi-
wall bags, gumming paper, building
paper; strong white papers from
bleached kraft; paperboards such as
used for cartons, containers, milk
bottles, and corrugated hoard
From R. Shreve, "Chemical Process Industries," McGraw-Hill, New
York (1967).
2-27
WALDEN RESEARCH CORPORATION
-------
ro
i
ro
00
o
m
TO
%
5
TO
O
X
8
TO
3
s
o
f° LJ 1 D C
Ln 1 Hb
(3) (6)
,kr>i/-r'~TC"D k ni r\\tii T A MI/ k> f TPiPi^ .- . . A Pi II P
_. . - - WA°iH F R , ...
// i CIRCULATION I
//• \
// \ KRAFT PULPING
/ / il
// \ CHEMICAL RECOVERY
f/ - SALT-CAKE d
f MAKE-UP (2) (5) <\fy
RFfO\/FRY ^ " DIRECT rOMTATT ^d Mill TIPI E EFFECT* STEAM
^reieBri.. J A (^E^SVn^?^
FURNACE ^ EVAPORATORS ^ EVAPORATORS '
(0
TALL OIL
3OAP 3KIMMING * RECOVERY
V(T)
v '/ \
— 1 — r^ MUD Line, j i VJINC. \
S M E I_T !...• i .-i^x"^ CAl|^->T|p|7FR ^'^ l^M M (A\ \MAKF IIP
DISSOLVING " * ((FITHFR)
TANK f | LIME )
L ME
(
FIGURE 2-9
OR SMELT
TYPICAL FLOW DIAGRAM FOR KRAFT PULPING
-------
TABLE 2-10
S(L EMISSIONS FROM KRAFT PULPING PROCESS
Potential Source
S02 Emission
Ib/ADT*
(1) Recovery Furnace (before DCE)
(2) Recovery Furnace (after DCE)
(3) Digester Relief and Blow
(4) Lime Kiln
(5) Multiple Effect Evaporators
BL Oxidation Tower
(6) Brown Stock Washers
.(7) Smelt Tank
0.5-15
0.5-8
Trace-0.01
Trace
0-0.01
0-0.01
0.01-0.02
0-0.1
Air dried pulp
2-29
WALDEN RESEARCH CORPORATION
-------
Flow to stack: T/D x 300 = acfm
Temp to stack: ^ 325°F
SOp concentration
between boiler & evap ^ 150 ppm
after evap ^ 70 ppm
(Note: Dust (NazSOa) load will decrease as one
a
moves from (1) after boiler to (2) after
evap to (3) after electrostatic
precipitator. )
The SOp and total reduced sulfur (TRS) values may vary considerably with
time as shown in Figure 2-10. We see that when the operation of a recovery
furnace was carried out under optimum conditions, the SOp could vary from a
few ppm to as high as several hundred ppm. We should point out, however,
that the values less than 50 ppm are the most typical of a furnace that is
functioning properly.
2.3.2 Sampling and Analytical Results
From the previous section, we have seen that the major SOp emis-
sions in a kraft pulping operation can be ascribed to the recovery boiler.
Since the parti cul ate matter from the recovery furnace consists of fine par-
ticles of Na?SO. and both the barium ion titration and barium chloranilate
methods are specific for sulfate, an additional heated (0.3y) particulate
filter was a necessary modification to the conventional EPA SOp sampling
train (Figure 2-4). The components used for collection of the field
samples are listed below:
a. heated glass lined probe with glass wool plug
b. heated (0.30y) glass fiber filter
c. impingers (2) filled with aqueous HpOp
d. metering system (orifice or rotameter plus dry
test meter)
e. vacuum pump
The train used is essentially that shown in Figure 2-4b except that the
IPA (SO ) impingers were eliminated. A schematic of the kraft mill is showr
O
in Figure 2-11. Two different locations were selected for sampling; one at
2-30
WALDEN RESEARCH CORPORATION
-------
ro
i
CO
J «•»
§
o
S: 3.0
2.0
1.0
0 0
TIME. HOURS
O
m
z
TO
o
I
o
o
X
T)
o
Figure 2-10. Gaseous Sulfur Distribution.
From Environmental Engineering (1971a).
o
2
-------
ro
i
U)
ro
•x
55
TO
a
Electrostatic Precipitator
Recovery Boiler
Stack
Wet (Black
Liquor)
Scrubber
Sampling Port
Figure 2-11. Schematic of Kraft Mill Operation.
TO
-------
the inlet of the precipitator; the other at the outlet of the wet scrubber
(inlet to the stack). A detailed drawing of the two selected sites is given
in Figure 2-12.
For the S0? data in Table 2-11 up to run 38, two peroxide absorp-
tion trains were run in parallel (A & B represent the two different trains)
following the fine particulate filter. The data indicate very low SCL levels
both in the wet scrubber outlet and in the inlet to the precipitator. In
runs 19 and 20, the effectiveness of the filter is clearly demonstrated.
These high "apparent SCL" values were due to a leak in the filter which al-
lowed some particulate matter (Na?S04) to be collected in the impingers.
A Barton titrator was run in conjunction with the wet chemical
samples up through about run 40. The SOp values from the Barton titrator
were negligible (< 5 ppm) in all cases. This finding is in agreement with
the low values obtained with the barium chloranilate and barium ion titra-
tion methods. The volume of stack gas pulled through the peroxide impingers
ranged from 60-90 liters over a 30-minute period.
Since the West-Gaeke (1956) method is specific for S02 and has
been used in kraft mill effluents, we compared this sensitive sulfite spe-
cific method to those which involve oxidation of the SO^ to sulfate.
For runs 39-62, three sampling trains were operated in parallel;
two contained HLO,, absorbing solution and the third was filled with sodium
tetrachloromercurate (TCM). The sampling rates were about 3 liters/min and
0.5 liters/min for the peroxide and TCM trains, respectively. The sampling
times were 20-30 minutes. The data are given in Table 2-11. Some additional
data on temperatures, flue gas composition, etc., are given in Table 2-12.
*
The results obtained by the three methods show reasonably good
agreement. The sulfate methods, however, suffer from a lack of sensitivity
* ++
A fourth method involving potentiometric titration of sulfate with Pb solu-
tion and monitored with a lead ion selective electrode is described in Ap-
pendix B.
2-33
WALDEN RESEARCH CORPORATION
-------
UNIT A + B
•12
1
15"
i.
STACK
a. PRECIPITATC3
INLET
•* 3.'
6' 3'
b. WET SCRUBBER OUTLET
Figure 2-12. Detailed Layout of Kraft Mill Sampling Sites.
2-34
WALDEN RESEARCH CORPORATION
-------
TABLE 2-11
S02 TEST RESULTS FROM A PULP AND PAPER MILL
Sample No.
1A*
IB*
2A
2B
3A
3B
4A
4B
5A
5B
6A
6B
7A
7B
8A
8B
9A
9B
10A
10B
11A
11B
12A
12B
13A
14A
15A
16A
17A
ISA
19A
20A
21A
21B
22A
22B
23A
23B
25A
25B
26A
26B
27A
27B
ppm S02
4
18
4
20
5
6
4
10
4
6
3
8
3
0
4
14
6
8
5
8
4
9
4
8
4
4
3
3
2
3
174
192
4
2
6
4
5
4
1
1
2
2
2
2
ppm S02 ppm SOg
(Ba Chlor) (West-Gaeke)
4
6
4
16
6
4
2
9
2
8
2
6
2
28
3
4
1
2
1
4
4
10
5
7
4
0
0
2
4
4
198 filter
221 leaking
0
0
4
0
4
1
4
3
3
4
2-35
WALDEN RESEARCH CORPORATION
-------
TABLE 2-11 (continued)
Sample No.
28A
28B
29A
29B
30A
30 B
31A
31 B
32A
32 B
33A
33B
34A
34B
35A
35B
36A
36 B
37A
37B
38A
38B
39A
39 B
40A
40B
41A
41 B
42A
42B
43A
43B
44A
44B
45A
45B
46A
46B
47A
47B
48A
48B
49A
49 B
50A
506
ppm S02ft
1
1
1
1
0
0
1
1
1
1
1
0
1
2
0
0
0
0
1
0
0
0
3
2
2
2
1
2
2
2
1
1
1
2
1
1
1
1
0
0
0
1
0
0
1
0
ppm S02
(Ba Chlor)
4
2
4
2
2
0
1
0
2
4
2
2
6
10
1
2
2
4
2
3
1
2
5
4
2
6
2
6
2
2
5
2
2
2
2
0
2
0
0
1
2
2
2
2
1
2
ppm S02
(West-Gaeke)
•
1f\
.8
1f\
.8
I/*
.2
1/*«
.2
1 9
1 . i.
0.4
0^
.7
0.6
0^
.7
0.7
Of
.5
Op
.5
2-36
WALDEN RESEARCH CORPORATION
-------
TABLE 2-11 (continued)
c , w ppm SO? ppm SO?1"1" ppm S02
Sample No. ^a++)Z (g£ ^r} (West-Gaeke)
51A 0 4 n ,
51B 0 5 u'5
52A 0 2 0 4
52B 0 2 u'^
53A 0 2
53B 0 2 U'H
54A 0 2
54B 0 1 U'b
55A 5 2
55B 6 1 U'b
56A 0 0 0 .
56B 0 2 u'^
58A 14 8 , ,
58B 19 11 blb
59A 24 21 9t. ,
59B 38 34 ">0
60A 0 2 0 .
60B 0 0 U'H
61A 4 2 0 4
61B 5 2 U'^
62A 4 2 0.3
These analytical methods (Ba and Ba Chlor) were run on aliquots of
the same sample.
*
A and B represent results from two parallel trains run off the same
probe.
2-37 WALDEN RESEARCH CORPORATION
-------
TABLE 2-12
SUPPLEMENTARY DATA FROM THE PULP AND PAPER MILL
1
co
00
s
1—
D
m
TO
m
w
TO
O
I
0
0
TO
T)
O
TO
Run No.
1-5
6-12
13-18
19-21
22-23
25-33
34-38
39-40
41-50
51-59
60-62
Date
12/16/71
12/17/71
12/20/71
12/21/71
12/22/71
12/30/71
12/31/72
1/4/72
1/5/72
1/6/72
1/7/72
Sampling Location % C02 % 0. Stack Temperature (°F)
Recovery Boiler-Scrubber 13.5 9
Outlet
Recovery Boiler- Scrubber 12.5 8
Outlet
Recovery Boiler-Scrubber 12.5 8.5
Outlet
Recovery Boiler-Precipitator 12 7
Inlet
Recovery Boiler-Precipitator
Inlet
Recovery Boiler-Precipitator 13.5 7.5
Inlet
Recovery Boiler-Precipitator 12 7
Inlet
Recovery Boiler-Precipitator 13 7
Inlet
Recovery Boiler-Precipitator 13 7.5
Inlet
Recovery Boiler-Precipitator 12 6
Inlet
Recovery Boiler-Precipitator 12 6
Inlet
210
205
204
570
620
565
625
649
625
615
540
-------
at these low SCL (about 2 ppm) levels even with the 60-90 liter sample vol-
umes. The West Gaeke method, on the other hand, has adequate sensitivity in
this range and may be more useful for determining low S0? concentrations,
e.g., 1 or 2 ppm.
Statistical analysis of all the data in Table 2-11 (except runs
19 and 20) yields the following results:
ppm S02 (Ba Chlor) = 1.56 + 0.62 ppm S02 (Ba++)
Correlation Coefficient = 0.71
ppm S02 (West Gaeke) = - 0.18 + 0.72 ppm SOp (Ba"1"1")
Correlation Coefficient =0.95
ppm S02 (West Gaeke) = - 1.3 + 0.93 ppm S02 (Ba Chlor)
Correlation Coefficient = 0.98
The three methods agree well with each other over this rather limited and low
range of SOp concentrations.
2.3.3 Conclusions
The S02 concentrations found in a recovery furnace of a kraft
mill were very low. The average S0? concentration was less than 10 ppm with
samples collected randomly over a three week period. The sulfate specific
methods require a fine particulate filter to remove Na2S04 particles in the
recovery boiler stack. The two sulfate specific methods (barium ion titra-
tion and barium chloranilate colorimetric) yield results for S02 which are in
excellent agreement with the sulfite specific method (West Gaeke) in the re-
covery boiler flue gas.
*
The sensitivity of the barium chloranilate method can be increased by read-
ing the absorbance at 330 nm instead of 530 nm.
2-39 WALDEN RESEARCH CORPORATION
-------
2.4 IRON AND STEEL
2.4.1 Brief Description of the Iron and Steel Industry
The major pollutant emitted from the iron and steel industry is
particulate matter. Emissions of sulfur oxides are derived from the sulfur
in the fuel and other raw material processes.
A detailed breakdown of the emission of S(L to the atmosphere
for 1967 from Varga, et al. (1969) is given in Table 2-13 for the iron and
steel industry. With the exception of fossil fuels (oil and coal) used in
various processes, the production of coke from coal and the use of the by-
product coke in various metalmaking processes accounts for the major portion
of S0? emissions. Clearly, these potential sources dictate the necessary
sites selected for S02 sampling.
Iron and steel processing can be divided into three closely al-
lied but separate operations:
a. coking
b. sintering
c. metalmaking process
2.4.1.1 Coking
In the coke ovens, finely ground (1/8 inch) coal is
burned in an enclosure with a reducing atmosphere (inadequate oxygen for com-
plete combustion). The resulting product consists primarily of carbon with
a small amount of sulfur (^ 0.6%). The off-gases from coking consist of car-
bon monoxide, hydrogen, methane, hydrogen sulfide, sulfur dioxide, ammonia,
nitrogen, and many high molecular weight organic compounds. A typical com-
position of the major constituents of coke oven gas is shown in Table 2-14.
The commonly used "slot type" coke oven is a narrow re-
fractory channel which may be up to 60 feet long, 18 feet high and 20 inches
wide. These ovens are grouped in "batteries." The flues between the adjoin-
ing ovens are heated to produce the high temperatures necessary to drive the
volatile matter from the coal to produce coke.
2-40 WALDEN RESEARCH CORPORATION
-------
TABLE 2-13
EMISSIONS OF SULFUR OXIDES TO THE ATMOSPHERE
FROM THE IRON AND STEEL INDUSTRY (1967)
Coke Oven Gas
Coke Breeze
Fuel Oil
Coal
Tar and Pitch
Tons SOg/year Emitted
to the Atmosphere
392,000
48,000
154,000
138,000
16,000
% of Total
52
6.4
21
18.4
2.2
From Varga, et al. (1969).
2-41
WALDEN RESEARCH CORPORATION
-------
TABLE 2-14
COMPOSITION OF COKE OVEN GAS
Component
co2
°2
N2
CO
Ho
2
CH4
C2H4
C6H6
Density = 0.03 #/ft3
Volume %
1.8
0.2
3.4
6.3
53.0
31.6
2.7
1.0
From "Combustion Calculations by Graphical Methods,"
Combustion Engineering (1970).
WALDEN RESEARCH CORPORATION
-------
The ovens are loaded from the top through several
charging ports which are covered and sealed for the coking operation. The
gases generated during coking are collected by a ducting system at the top
of the "slot" oven and sent to a by-product recovery operation. About 40%
of these gases are cleaned and burned to generate heat for the operation of
the coke ovens.
The exhaust from the heating units is usually vented
through a single stack. The combustion of the cleaned coke oven gas gen-
erates some sulfur dioxide since the original gas may have contained as much
as 3 to 5 grains of sulfur (as H,,S) per cubic foot.
At the conclusion of the coking cycle, the end doors
are removed and a mechanical ram pushes the coke out of the oven into a
special car which carries the coke to a quenching operation.
2.4.1.2 Sintering
In order to process iron ore in a blast furnace, it is
necessary for the particle size to be increased to prevent losses out the
top of the furnace.
In the sintering process, iron ore, furnace flue dust,
miscellaneous fines, and other iron-bearing materials are mixed with lime-
stone and coke breeze or coal and fed to a sinter machine where it is com-
busted on a steel conveyor.
The material is ignited and the combustion is self-sup-
porting. Combustion air is drawn through a sinter bed which has the under-
lying sections separated into individual compartments to ensure uniform air
distribution. These compartments, called "wind boxes", are ducted to a
single fan. The ore thus heated is collected in cyclones and electrostatic
precipitators and then sent to the blast furnace.
The sintering operation is not as large a source of
sulfur oxide emissions as the coke ovens but the quantity of sulfur in the .
raw materials and the fuel is still considerable. It has been estimated
2-43
WALDEN RESEARCH CORPORATION
-------
that approximately two pounds of sulfur are introduced in the sintering proc-
ess during the production of one ton of sinter. Approximately one-half of
the sulfur is in the iron ore and the other half is in the coke, fuel and
limestone (Varga, 1969).
A liquid slag which is formed in the sintering process
retains approximately 64% of the sulfur with the remainder (36%) leaving the
system in the combustion gases. This is equivalent to several hundred ppm
of SCL in the stack gases (Environmental Engineering, 1971b).
2.4.1.3 Metalmaking Process
The open hearth furnaces which were once the primary
units in the iron and steel-making industry are being phased out and the
Basic Oxygen Furnaces (BOF) are taking their place. With the open hearth,
about 50% of the sulfur input goes into the steel and slag and about 50%
goes out as SOp in the flue gas (Varga, 1969). The S02 level in the flue
gas could be as high as 250 ppm if all the sulfur in the process feed were
converted to SOp.
2.4.1.4 Summary
1. The principal source of sulfur emissions in the
iron and steel processes is in the coal used for the production of coke, and
the coke used in the sintering and metalmaking processes.
2. The sources to be sampled should include the coking
process as well as the operations using large quantities of coke, e.g., open
hearth and sintering.
2.4.2 Sampling and Analytical Results
The following operations were tested in an iron and steel plant:
sintering
coke oven
open hearth
2-44 WALDEN RESEARCH CORPORATION
-------
Since the flue gas temperatures of these sources were 500°F or less (Table
2-15) the EPA heated probe with a pyrex liner and glass wool filter was used.
Three trains (A, B, and C) were run in parallel off the same probe on the
coke oven (runs 1-16). In the other sources tested, two and sometimes three
trains were run in parallel off the same probe. The sampling trains con-
sisted of two impingers containing 15 ml of 80% isopropanol-20% water fol-
lowed by two impingers containing 15 ml of 3% aqueous hydrogen peroxide with
appropriate metering and air moving devices. During each test, one or two
samples were collected behind a heated fine (0.3y) glass fiber particulate
filter. The sample volume collected ranged from about 80-100 liters over a
thirty minute period. Detailed layouts of the three operations and also the
sampling sites are shown in Figures 2-13 to 2-15.
The results for the coke oven, sintering and open hearth test
sites are given in Tables 2-16 to 2-19. On the open hearth, a number of
runs were eliminated (Table 2-19) because one of the two probes used had a
broken liner. Although this probe was checked (leak tested) before it was
put into the open hearth stack, the leak was not detected, and no S02 was
found in these samples. These samples were, for the most part, those which
had the fine particulate filter in line.
Statistical analysis of the results from Tables 2-16 to 2-19 are
shown in Table 2-20. For the No. 1 coke oven there appears to be no sig-
nificant difference between the filtered and unfiltered samples for Ba or
Ba Chlor. A semi-quantitative emission spectrographic analysis of an SO,,
impinger sample from a coke oven is shown in Table 2-21. No zinc was found
and Cu , Ca and other interferences, if present, are at very low concen-
trations in both the filtered and unfiltered samples. Atomic absorption
analysis indicated no detectable quantities of Zn, Ca, or Na in the samples.
The regression equations and correlation coefficients are ex-
cellent in both cases. The CV for the Ba titration at the No. 1 coke oven
(1.4%) is very similar to that obtained for the coal-fired power plant (CV =
1.5%) although the mean S02 value is slightly lower. The Ba Chlor shows
slightly higher CV's than that found for the uncontrolled coal-fired power
plant.
2-45 WALDEN RESEARCH CORPORATION
-------
TABLE 2-15
SUPPLEMENTARY DATA FROM THE IRON AND STEEL PLANT
ro
i
•c*
CT>
1-
O
m
z
70
m
c/)
5
TO
a
i
O
0
TO
s
a
Run No.
1-10
11-16
17-25
29-34
35-48
49-58
59-70
Date Sampling Location % C02 % 0^ Stack Temperature (°F)
2/23/72 Coke oven 4.2 10.5 300
2/24/72 Coke oven 335
2/25/72 Coke oven (10% oven gas and 16.0 0 160
90% blast furnace gas)
2/28/72 Sinterer 313
2/29/72 Sinterer 0.5 19.0
3/1/72 Open hearth 3.5 15.5 480
3/2/72 Open hearth 2.5 18.0 500
O
-------
ro
i
I
o
m
TO
m
in
m
>
73
a
a
o
73
s
4'
3'
J
8'
CROSS SECTION DETAIL
OF SAMPLING PORT
COKE
OVEN
SAMPLING
PORT
\
COMBUST ION OF1
SPENT GASES
Figure 2-13. Schematic of Coke Oven Site and Sampling Port.
t
STACK
-------
no
i
CO
>
o
o
I
8
3
h-4'
T
40'
J
0
CROSS SECTION DETAIL
OF SAMPLING PORT
SAMPLING
PORTS
'COMBUSTION
GASES
_T
r
SINTERING
BELT"
MULTICLCNE
STACK
t
ELECTROSTATIC
PRECIPITATOR
X.D.
FAN
Figure 2-14. Sintering Plant Schematic and Test Site.
O
-------
h2'
2'
STACK
vo
I
o
rn
m
en
m
>
TO
a
8
TO
s
TO
8'
CROSS SECTION DETAIL
OF SAMPLING PORT
SAMPLING PORTS •
OPEN
H E A R TH
FURNACES
ELECTROSTATIC
PRECIPITATG3
t
Figure 2-15. Schematic of Open Hearth Unit and Sampling Ports.
O
z
-------
TABLE 2-16
S09 TEST RESULTS FROM AN IRON AND STEEL PLANT
* (COKE OVEN)
Sample No.
1A t
IB* t
1C t
2A
2B*
2C
3A
3B*
3C
4A
4B*
4C
5A
5B*
5C
6A
6B*
6C
7A
7B*
7C
8A
8B*
8C
9A
9B*
9C
10A
10B*
IOC
11A
11 B*
11C*
12A
12B*
12C*
13A
13B*
13C*
14A
14B*
14C*
ppm S02
(Ba Chlor)
251
214
249
293
158
281
273
261
307
288
282
319
287
257
281
135
245
313
225
221
300
295
301
334
287
285
315
273
279
317
172
135
192
177
135
165
162
99
115
153
3
6
ppm SO?
(Ba++)
280
258
282
280
158
267
267
267
295
282
268
310
282
250
278
137
238
310
230
223
306
294
291
330
284
282
315
273
279
316
170
136
188
177
140
176
155
116
118
148
16
3
ppm S02
(NaOH)
300
272
302
302
163
269
286
275
315
297
285
330
301
273
304
141
254
321
247
239
327
301
316
358
302
303
339
291
292
327
182
148
202
189
147
180
163
117
118
2-50 WALDEN RESEARCH CORPORATION
-------
TABLE 2-16 (continued)
camnlp Nn ppm S02 ppm S02
Sample No> (Ba Chlor) (Ba++) (NaOH)
15A 214 208
16A 177 183
*
With fine particulate filter
A, B, and C represent results from three parallel trains run from the
same probe.
2-51 WALDEN RESEARCH CORPORATION
-------
TABLE 2-17
SO- TEST RESULTS FROM AN IRON AND STEEL PLANT
* (COKE OVEN)t
Sample No.
17A*
17B*
18A*tt
18B*ft
18C -!"«•
19A
19B
20A*
20 B*
21A
21B
22A*
22B*
23A
23B
24A*
24B*
25A
25B
ppm S02
(Ba Chlor)
51
53
29
33
35
38
39
23
21
29
29
32
32
28
34
36
35
26
28
ppm SO?
(Ba++f
59
64
43
41
44
46
42
28
29
34
34
36
33
31
30
37
36
32
28
With fine particulate filter
90% blast furnace gas, 10% coke oven gas
1.4,
A, B, and C represent results from three parallel trains run from the
same probe
2-52 WALDEN RESEARCH CORPORATION
-------
TABLE 2-18
S09 TEST RESULTS FROM AN IRON AND STEEL PLANT
* (SINTERER)
Sample No.
29A f
29B* t
29C* f
30A
30B*
30C
31A
31 B*
31 C*
32A
32 B*
32C*
33A
33B*
33C*
34A
34 B*
34C*
35A
35B
36A*
36 B*
37A
37B
38A*
38B*
39A
39B
40A*
40B*
41A*
41 B*
42A
42B
43A*
43B*
44A
44B
45A*
45 B*
46A
46B
ppm S02
(Ba Chlor)
25
30
38
22
22
20
18
26
18
10
8
6
8
14
16
13
18
21
24
23
4
10
5
34
13
16
46
44
24
26
28
34
49
52
25
28
16
14
22
22
40
47
ppm S09
(Ba++r
20
24
28
14
20
21
20
20
18
16
12
9
12
12
16
15
19
22
23
28
0
10
0
34
14
18
45
44
24
26
25
29
50
52
28
30
0
14
25
25
44
48
2-53
WALDEN RESEARCH CORPORATION
-------
TABLE 2-18 (continued)
c,~ T KU SO? ppm SO?
SamP1e No- (to Chlor)
47A* 25 28
47B* 24 26
48A 51 51
48B 42 43
*
With fine participate filter
'A, B, and C represent results from three parallel trains run from the
same probe.
WALDEN RESEARCH CORPORATION
-------
TABLE 2-19
SO, TEST RESULTS FROM AN IRON AND STEEL PLANT
* (OPEN HEARTH)
C m«l« No PP111 S02 PP"1 S°2
SamPle No* (Ba Chlor) (Ba++)
49A t 34 36
49B* t 12 14
49C* t 54 56
50A 28 34
50B* 111 106
51A 66 64
51B* 62 58
51C* 70 66
53A 43 44
53B 44 44
55A 36 38
55B 32 36
57A 38 42
57B 42 46
60A 38 39
60B 34 40
62A 8 20
62B 12 20
63A 46 48
63B 48 51
65A 35 42
65B 38 40
67A 50 52
67B 56 55
69A 27 30
69B 32 35
With fine particulate filter
rA, B, and C represent results from three parallel trains run from the
same probe.
2-55
WALDEN RESEARCH CORPORATION
-------
TABLE 2-20
STATISTICAL ANALYSIS FOR S02 TEST RESULTS
FROM AN IRON AND STEEL PLANT
ro
i
tn
o
rn
73
m
m
TO
o
X
8
TO
TJ
O
73
Site Run No. Regression Equation Coefficient
#1 Coke Oven 1-16* ppm Ba++ = 0.97 ppm Ba Chlor + 5.6 0.98
1-16*
1-16*
l-16t ppm Ba = 0.99 ppm Ba Chlor + 2.8 0.99
l-16t
l-16t
#8 Coke Oven 17-25* ppm Ba++ = 1 .17 ppm Ba Chlor - 4.5 0.89
17-25*
17-25* ++
17-25t ppm Ba = 1.00 ppm Ba Chlor + 3.4 0.95
17-25+
17-25t
Sinterer 29-48* ppm Ba++ =1.06 ppm Ba Chlor - 3.2 0.97
29-48*
29-48*
29-48t ppm Ba = 0.81 ppm Ba Chlor + 3.6 0.91
29-48t
29-48+
Open Hearth 49-70* ppm Ba++ =0.93 ppm Ba Chlor + 5.4 0.99
49-70*
49-70*
*
Just glass wool filter
With fine particulate filter and glass wool filter
Coefficient
Variation
Analytic
Method
(Ba Chlor)
(Ba"'"'")
(Ba Chlor)
(Ba++)
(Ba Chlor)
(Ba"*"*")
(Ba Chlor)
(Ba++)
(Ba Chlor)
(Ba"'"'')
(Ba Chlor)
(Ba++)
(Ba Chlor)
(Ba++)
of
of
5.9%
1.2
4.9%
1.6
9.7%
3.6
7.2%
4.2
12.3
2.9
7.2
4.7
8.8
5.0
Mean
257
256
257
212
38
35
33
40
21
28
29
20
37
35
o
-------
I
0
m
z
TO
m
C/5
TO
O
I
O
O
X3
-o
O
XI
5
O
TABLE 2-21
SEMIQUANTITATIVE SPECTROGRAPHIC ANALYSIS OF AN S02 IMPINGER
SAMPLE FROM AN IRON AND STEEL PLANT (COKE OVEN)
rv>
i
01
ppm Al ppm B
With glass wool 0.05-
filter 0.5
With glass wool 0.03-
and fine par- 0.3
ticulate filter
ppm Be ppm Cu ppm Cr
0.1- 0.03-
1 0.3
0.1- 0.003-
1 0.03
ppm Fe ppm Mg ppm Mn ppm Ni
0.1- 0.03- 0.003- 0.01-
1 0.3 0.03 0.1
0.03- 0.03- 0.001- 0.003-
0.3 0.3 0.01 0.03
ppm Ag ppm Pb
With glass wool 0.03- 0.03-
filter 0.3 0.3
With glass wool 0.1- 0.03-
and fine par- 1 0.3
ticulate filter
ppm Sn ppm Ca ppm Si
0.03- 0.03-
0.3 0.3
0.03- 0.03-
0.3 0.3
ppm Ti ppm Zn ppm Na
0.003- — 0.1-
0.03 1
0.003- — 0.3-
0.03 3
-------
The No. 8 coke oven effluent which is ^ 10% coke oven gas and
90% blast furnace gas does show a slight improvement with the fine particu-
late filter but this is not significant because of the low concentrations of
SOp in the flue gas which produces added variability in the methods. The
small difference in CV values for filtered versus unfiltered samples indi-
cate no interference for Ba or the barium chloranilate methods (the in-
creased CV is due to the lower SOp level).
The sintering operation also has low S0? concentrations and the
difference between filtered and unfiltered samples is negligible. The
chloranilate method, however, does show an apparent improvement in the CV
vor the filtered. However, the regression equation is somewhat poorer.
The Ba and Ba Chi or methods agree well for the unfiltered
(without fine particulate filter) samples. The filtered samples (with a few
exceptions) were run on the other probe which had a cracked probe liner.
The CV for Ba is probably typical for these low SOp levels. It is not
clear whether the higher CV for Ba Chi or is due to an interference or the
imprecision of the method at this S0? level.
*
The sulfur balance for a coke oven operation is given in Table
2-22. The calculated SOp value (ppm by volume) if all of the sulfur in the
coke oven gas was oxidized would be about 1000 ppm. The average measured
value for samples 1-10 (by Ba ) is 295 ppm. The measured Op value in Table
2-15 indicates a dilution of the flue gas with approximately one volume of
air. On this basis, the measured values should be adjusted to give about
600 ppm SOp. This is 60% of the calculated value. We have been in Table
2-14 that the flue gas from a coke oven is a reducing atmosphere with sev-
eral percent carbon monoxide. For a blast furnace, which also has a re-
ducing atmosphere, the Bureau of Mines (1968) has reported that no sulfur
oxides are present in the top gases. We have seen previously that in the
foundry which is operated in a reducing atmosphere to prevent oxidation of
*
An actual sulfur balance carried out on the coke ovens which we sampled was
in good agreement with that in Table 2-22.
2-58 WALDEN RESEARCH CORPORATION
-------
TABLE 2-22
SULFUR BALANCE FOR COKE OVEN OPERATION
(Based on Production of One Net Ton of Coke)
Item
Amount,
pounds
Sulfur,
percent
Amount of
Sulfur, poumls
Input
Coal
2950
0.8
23.60
Output
Coke and breeze
By- pr oducls
' 'ok u- oven gas*a'
2121
290
534
0.
0.
1.
700
344
45
Total
14.
1.
7.
23.
85
00
75
60
-------
the metal in the melt, approximately 50% of the sulfur in the flue gas is
emitted as SO and a large portion remains as sulfur in the particulate mat-
/\
ter or dissolved in the melt. Thus, the approximately 60% of theoretical
S02 obtained here is not unreasonable. The CO may react with S02 to form
elemental S (Ryason and Harkins, 1967). The data on runs 17-25 were ob-
tained in a coke oven which consists of 90% blast furnace gas and 10% coke
oven gas. Since the blast furnace contains no S0_, the values would be ex-
pected and are found to be about 10% of the concentration of runs 1-16.
The sulfur balance for the sintering operation is shown in Table
2-23. From the estimated sulfur in the combustion gases and the quantity of
fuel, the S0? concentration is calculated to be approximately 250 ppm. The
++
average measured S0? values obtained from the sintering operation (Ba runs
29-48) is 20 ppm. The C02 and 02 data in Table 2-15 indicates dilution of
the flue gas with about 15 volumes of air. If we now correct the measured
value for this dilution, we obtain about 300 ppm SO,,. This difference be-
tween measured and calculated S02 may be the result of a different sulfur
content in the fuel oil and/or the large dilution factor determined by 02
and C02 measurements. However, in view of the many approximations made, the
agreement is reasonable, e.g., all of the sulfur is emitted as S0?.
The sulfur balance for an open hearth furnace from Varga (1969)
indicates that nearly 50% of the sulfur goes into the slag or steel and the
remainder is emitted to the atmosphere. The calculated value for S0? then
is of the order of 200 ppm. The measured (average) S02 value from the open
hearth is 45 ppm or 180 ppm when corrected for the approximately four-fold
dilution of the flue gas with air (determined by 02 and C02 measurements in
Table 2-15). This value is quite reasonable since it represents approxi-
mately 90% of the predicted S02 level in the open hearth. No fuel, slag or
scrap samples were taken for sulfur analysis.
For the latter two sites, all the sulfur would be converted to
SO since these conditions are similar to a combustion environment.
2-60 WALDEN RESEARCH CORPORATION
-------
TABLE 2-23
SULFUR BALANCE FOR SINTERING MACHINE OPERATION
(Based on the Production of One Net Ton of Sinter)
Amount,
pounds
Sulfur
Content, percent
Amount of
Sulfur^ poumi•
Iron-bearing material
Coke
Limestone
Sinter
i ntt.-r linos
2200
100
50
200
Input
0. 041
0.70
0. 55
0. 049
Output
2000 0.055
Z89 0.055
Total
Total
0.90
0.70
0. 27
0. 10
1.97
1.10
0. 16
0. 7 I
1.97
2-61
WALDEN RESEARCH CORPORATION
-------
2.4.3 Conclusions
No major interferences with the analytical methods appear to be
present in the sources sampled in the iron and steel industry. The sharper
endpoint for Ba titrations obtained with the fine particulate filter in
the sampling train suggest that this be used, however. The limited data sug-
gest that part of the sulfur from the coke ovens is retained in the particu-
late because of the reducing atmosphere, while in the open hearth and sinter-
ing operations, the estimated and measured SCL values show reasonable agree-
ment.
2.5 SMELTING
2.5.1 Brief Description of Smelting Processes
2.5.1.1- Copper Smelting
The primary smelting process in the extraction of copper
from its ore is a four-stage operation: (1) concentrating and drying,
(2) roasting, (3) melting, and (4) oxidation converting.
These operations can and, indeed, are carried out in a
number of ways depending on the characteristics of the ore and the design of
the plant.
Generally, copper ore will assay at only 1.5 to 2.0% Cu.
In order to reduce the volume of material to be handled through the remaining
stages of smelting the first stage, or concentrating step, usually is set to
yield a product of 20 to 30% Cu.
A common practice, particularly for ores high in sulfur,
is a concentrating process which involves grinding and flotation followed by
the three stages of smelting. The flotation concentrate is composed of
copper sulfides, iron sulfides and residual gangue.
In the next process, the concentrated ore is roasted
and a major portion of the sulfur is driven off leaving a product of the same
materials as originally introduced to the roaster but having a substantially
reduced amount of sulfur.
2-62 WALDEN RESEARCH CORPORATION
-------
The temperature of the gas from the roaster may be in
the range of 1200°F and the sulfur content may be from 5 to 10%. These gas
streams are cooled to around 600°F after preliminary solids collection by
air dilution, water sprays or heat exchangers.
In the next step, the roasted ore is melted with an ap-
propriate flux which produces a matte. Natural gas or oil burners oxidize
the iron sulfides while silicates combine with the iron to concentrate it
in the slag and thus separate the iron from the copper. Temperatures here
may be around 2250°F.
Of the total sulfides in the roasted ore, 70 to
ends up in the matte and 20 to 30% in the flue gas (Environmental Engineer-
ing, 1971c). Due to the removal of a major portion of the sulfides in the
roasting operation, the concentration found in the melting effluent should
be less than that from the roasting operation.
In the final step, the molten matte produced in the
previous process is purified by heating. This separates the iron and sulfur
from the matte. No fuel is needed since the oxidation process itself sup-
plies enough heat to maintain the converter at about 2000°F. After the
separation of the slag, the sulfur is removed by oxidation yielding Cu which
is approximately 99% pure.
Some typical S(L levels in the effluent of various
processes in copper smelting are given in Table 2-24 from Environmental
Engineering (1971c).
Secondary smelting or refining of copper is not a sig-
nificant source of SCL since the copper produced originally is 99% pure.
Production of sulfuric acid is the only proven tech-
nology for removing sulfur oxides from smelter gases. Sulfuric acid pro-
duction is practical from an economic standpoint only from the converter or
roasting process. The more dilute gases from other operations must be in-
corporated since they contain more sulfur than permitted by air pollution
regulations.
2-63
WALDEN RESEARCH CORPORATION
-------
TABLE 2-24
SULFUR DIOXIDE EMISSIONS
Process
Typical S02 Concentrations
Copper Zinc Lead
Dryi ng
Melting (Reverb.)
Combined Sintering
Updraft with recycle
Downdraft
Roaster
Multiple hearth
Fluid bed
Coking (Blast)
Converting
Distillation (Retort)
0-0.5%
0.6-2.5
5-10
10-12
14-21
0-0.5%
6.0-6.5
0.1-2.4
4.5-6.5
10-12
0.08
0-0.5%
0.0-0.2
4.0-6.0
0.8-1.8
0.01-0.25
2-64
WALDEN RESEARCH CORPORATION
-------
2.5.1.2 Lead Smelting
Lead ore occurs essentially as lead sulfide (Galena).
It is usually associated with zinc, iron and copper sulfides. The zinc and
copper are present in recoverable amounts but frequently the iron that can
be separated by mechanical means is discarded.
The ore received at the smelter has usually been con-
centrated by mechanical means to 50 to 70% lead, about 6% zinc, up to 4%
copper, and 15 to 20% sulfur. Lead ores often contain cadmium, bismuth,
gold and silver along with numerous other metals in trace amounts.
All primary lead smelters use essentially the same
processing steps with minor modifications to accommodate the equipment used.
The three steps in processing concentrated lead ore are:
(1) sintering, (2) reduction (usually in a blast furnace), and (3) refining.
In the sintering process, the lead sulfide ore is con-
verted to lead oxide and sulfur.
In this operation, the sulfur dioxide may be from 4 to
6% of the effluent gas and the temperature is in the 300 to 500°F range.
The lead oxide formed in the sintering operation is re-
duced with coke in a blast furnace at 1800°F forming an impure metal and a
slag.
In the final operation, the impure lead is refined at
about 1800°F to a product which is 95 to 99% Pb. The sulfur emitted in this
operation is usually less than 1%.
In most operations, the S02 is recovered in HUSO, plants,
The reported uncontrolled SOp emissions from lead smelting is 0.6 ton per
ton of metal produced. Some typical levels of SOp for various lead smelt-
ing operations are shown in Table 2-24.
WALDEN RESEARCH CORPORATION
-------
2.5.1.3 Zinc Smelting
Zinc ore occurs predominantly as the sulfide. Usually
zinc and lead occur together in about equal amounts and a dual smelting ac-
tivity may be carried on.
The ores are low in both zinc and lead content and must
be concentrated before roasting. Concentrating is performed by a differen-
tial slurry flotation enriching the ore to 50 to 60% zinc, 30 to 33% sulfur,
and 5 to 10% iron, with only minor quantities of copper and cadmium. The
lead that co-existed is separated in the differential slurry flotation.
The second step in zinc recovery is the roasting of the
concentrate which is referred to as "calcine." In this operation, most of
the sulfur is driven off as SO^. The concentration of SO^ varies with the
process, from as low as 1% to as high as 10 to 13% (see Table 2-24).
Roaster operations vary from plant to plant according
to the ore and the process used in the subsequent refining operation. The
temperature of the offgas may vary from 1200°F to 1900°F depending on
sampling location.
The subsequent operations in zinc smelting emit some'sul-
fur dioxide but the primary emissions are in the roasting.
The secondary production of zinc, from primary copper
or lead smelters, is small compared to the amount from the ore.
2.5.2 Sampling and Analytical Results
The following operations were tested at the smelter:
Copper - roasting and reverberating
Copper - converter
Lead - total operation since the effluent from all
operations was vented through a single stack
Tests were conducted with and without the fine particulate filter. Unfor-
tunately, the only sampling positions available were following a baghouse or
2-66 WALDEN RESEARCH CORPORATION
-------
precipitator (see Figures 2-16 and 2-18). Thus, we could not test the
methods with high particulate concentrations in the duct. With run A, SO,
was collected by controlled condensation; run B, SO., was collected in iso-
propanol. The samples with an asterisk (*) were collected with a heated
fine (0.3y) particulate filter between the probe and SO train.
/\
The sampling times ranged from 10 to 20 minutes at a rate of ap-
proximately 0.5 1/minute. Note that both the sampling rate and time were
reduced because of the high SO,, concentrations.
The SOp samples (at the copper roasting or reverberatory opera-
tion) were obtained with and without the fine particulate filter at two
sites (A and B, see Figure 2-17) with different probes. For example, A and
B represent duplicate trains, samples 1-10, 12, 14, etc. (Table 2-25) were
obtained at site A without the fine filter while 11, 13, 15, etc., were ob-
tained at site B with the filter.
The S0? test results for the copper smelter are given in Table
2-25. The low values, runs 10 and 12, are due to process shutdown. Some
supplementary data on flue gas temperature, orsat analysis are given in
Table 2-26. The S0? data from the lead smelter are shown in Table 2-27.
Here again, the low ^ 700 ppm S02 value was a result of the operation being
shut down.
Regression analysis of the data from the copper smelter (Table
2-28) gives a high correlation coefficient and excellent regression equations
for both the Ba and Ba Chi or methods with and without the fine particulate
filter. Apparently the high S02 concentrations and the low particulate load-
ing of this flue gas minimize the effect of the fine particulate filter.
There is, however, a substantial effect of the fine filter on the CV of the
barium chloranilate method. If we look at the semi-quantitative emission
spectrographic analysis (Table 2-29), we see that the concentrations of zinc,
calcium, and copper are reduced substantially by the filter. Atomic absorp-
tion results in Table 2-30 (on a limited number of samples) have high con-
centrations of Zn and Ca in the analyzed samples. These species are reduced
by nearly an order of magnitude by the fine filter, hence the improvement in
2-67 WALDEN RESEARCH CORPORATION
-------
COPPER
CONVERTER
ELECTRO-
STATIC
PRECIPITATOR
COPPER
ROASTED
AND
REVERBATORY
GROUND
Figure 2-16. Schematic of the Copper Smelting Operation.
2-68
WALDEN RESEARCH CORPORATION
-------
B
Figure 2-17. Duct Cross-Section Showing Probe Penetration (R&R)
(Copper Roaster and Reverbatory).
2-69
WALDEN RESEARCH CORPORATION
-------
• I'-
DUCT CROSS SECTION
SHOWING PROBE PENETRATION
A B
»-• -•
10'
10'
BAG HOUSE
Pb
SMELTING
OPERATION
GROUND
Figure 2-18. Schematic of the Lead Smelting Operation.
2-70
WALDEN RESEARCH CORPORATION
-------
Sample No.
lAt
IBt
2A
2B
3A
3B
4A
4B
5A
5B
6A
6B
7A
7B
8A
SB
9A
9B
IDA
10B
11A*
11 B*
12A
12B
13A*
13B*
14A
14B
15A*
15B*
16A
16B
17A*
17B*
18A*
18B*
19A
19B
20A*
20B*
TABLE 2-25
S02 TEST RESULTS FROM A COPPER SMELTER
(Roasting and Reverberatory)
ppm S02
(Ba++)
3308
--
3447
3830
1840
1482
__
2310
2200
2229
4774
4735
3491
5438
3699
4970
2232
2693
106
201
2395
3069
184
141
3658
4249
—
--
3100
1562
2856
2530
3376
4446
1980
2468
4139
4112
2186
2738
ppm S02
(Ba Chlor)
4229
5689
3461
4038
2295
1661
—
—
2274
2067
4821
4843
4173
5055
4750
5179
2280
2615
100
157
2989
3865
237
205
3751
4227
2505
2727
3377
1691
2755
2555
3319
4523
1862
2464
4229
4159
2182
2841
2-71 WALDEN RESEARCH CORPORATION
-------
TABLE 2-25 (continued)
Sample No.
ppm SO?
(Ba++r
ppm S02
(Ba Chlor)
21A 2454 2500
21B 2504 2470
22A* 3308 3383
22B* 4308 4592
23A 2430 2510
23B 2591 2724
24A* 3230 3162
24B* 5489 5369
25A 2630 2559
25B 2721 2687
26A* 5418 2995
26B*
*
with fine particulate filter
'A and B represent results obtained in a parallel train from
the same probe.
2-72 WALDEN RESEARCH CORPORATION
-------
ro
i
CO
TABLE 2-26
SUPPLEMENTARY DATA FOR THE COPPER AND LEAD SMELTERS
Run No.
1-9
10-26
21-32
33-40
41-56
57-65
Date
2/1/72
2/6/72
2/7/72
2/8/72
2/9/72
2/10/72
Sampling Location
Copper roaster and reverberatory
Copper roaster and reverberatory
Copper converter
Copper converter
Lead smelter at baghouse exit
Lead smelter at baghouse exit
% co2
0.9
1.4
0.0
0.0
-
-
%o2
16.5
15.5
7.5
5.5
-
-
Stack Temp.
189.
225
122
153
145
162 '
(°F)
I
o
m
TO
m
o
o
O
73
s
TO
6
-------
TABLE 2-27
S02 TEST RESULTS FROM A LEAD SMELTER
Sample No.
41 At
41 Bt
42A*
42B*
43A
43B
44A*
44B*
45A
45B
46A*
46 B*
47A
47B
48A*
48B*
49A
49B
50A*
SOB*
51A
51B
52A*
52B*
53A
53B
54A*
54B*
55A
55B
56A*
56 B*
57A
57B
58A*
58B*
59A
59B
60A*
60B*
61A
61B
ppm SO?
(Ba++)
1071
--
1427
1410
702
1638
1928
1310
2172
4249
4430
4841
2995
5031
3160
2935
631
842
2172
--
2400
4076
4822
--
2508
3991
5253
5014
2581
2320
5357
5218
2347
4197
4554
3901
2696
4368
4400
4168
2205
'
ppm S02
(Ba Chlor)
972
--
1445
1467
565
1391
1710
1132
1946
2985
4642
3430
2740
3462
2948
2741
110
521
2037
•
2131
3616
4696
--
2250
4059
4982
4898
2209
2219
5140
5140
2491
4185
4461
3890
2525
4310
4301
4229
2050
--
2-74 WALDEN RESEARCH CORPORATION
-------
TABLE 2-27 (continued)
r -i M SO? PPm
SamPle No- (Ba++) (Ba Chlor)
62A* 4368 4434
62B* 4249 4376
63A 2480 2353
63B 3527 3532
64A* 4066 41 31
64B* 3230 3173
65A 2631 .2585
65B 2519 2492
*
with fine particulate filter
A and B represent results from a parallel train run off the
camo nrnho
same probe.
2-75
WALDEN RESEARCH CORPORATION
-------
TABLE 2-28
STATISTICAL ANALYSIS FOR S02 TEST RESULTS
FROM SMELTERS
ro
i
en
I
en
m
>
TO
O
I
O
O
TO
s
Site
Cu Roaster
and
Reverbatory
Pb Smelter
*
With just
fWith fine
Run No. Regression Equation
1-26* ppm Ba Chlor = 1.02 ppm Ba++ + 95
1-26*
1-26* ++
l-26t ppm Ba Chlor =0.99 ppm Ba +196
1-26+
1-26+
41-65* ppm Ba Chlor =0.89 ppm Ba++ + 28
41-65*
41-65* ++
41-65T ppm Ba Chlor = 0.96 ppm Ba +13
41-65+
41-65+
glass wool filter
parti cul ate filter and glass wool filter
Coefficient of
Correlation Variation of
Coefficient Analytic
Method
0.98 ++
(Ba ) 1.8%
(Ba Chlor) 6.5
0.95 ++
(Ba ) 1 .6%
(Ba Chlor) 1.8
0.95 ..
(Ba ) 4.0%
(Ba Chlor) 9.0
0.96
(Ba +) 1.5%
(Ba Chlor) 7.7
Mean
3080
3360
3351
3298
3139
2841
4232
3490
-------
TABLE 2-29
SEMIQUANTITATIVE EMISSION SPECTROGRAPHIC ANALYSIS OF S02
IMPINGER SAMPLES FROM SMELTING OPERATIONS ,
ro
i
"
i
o
m
z
JO
m
o>
5
TO
a
i
o
o
10
-o
O
TO
Cu Smelter
(no filter)
Cu Smelter
(with filter)
Pb Smelter
(no filter)
Pb Smelter
(with filter)
Cu Smelter
(no filter)
Cu Smelter
(with filter)
Pb Smelter
(no filter)
Pb Smelter
(with filter)
ppm Al
0.03-
0.3
0.03-
0.3
0.03-
0.3
0.03-
0.3
ppm Ag
0.1-
1
0.1-
1
0.1-
1
0.03-
0.3
ppm B ppm Be ppm Cu
0.3-
3
0.1-
1
0.1-
1
0.1-
1
ppm Pb ppm Sn ppm Ca
0.03- — 0.3-
0.3 3
0.03- 0.03- 0.03-
0.3 0.3 0.3
0.1- 0.03- 0.03-
1 0.3 0.3
0.1- — 0.03-
1 0.3
ppm Cr
0.03-
0.3
0.003-
0.03
0.03-
0.3
0.01-
0.1
ppm Si
0.03-
0.3
0.03-
0.3
0.03-
0.3
0.03-
0.3
ppm Fe
0.1-
1
0.03-
0.3
0.3-
3
0.1-
1
ppm Ti
0.01-
0.1
0.01-
0.1
0.003-
0.03
0.003-
0.03
ppm Mg
0.03-
0.3
0.03-
0.3
0.03-
0.3
0.03-
0.3
ppm In
0.1-
1
0.03-
0.3
0.03-
0.3
0.03-
0.3
ppm Mn
0.001-
0.01
0.001-
0.01
0.003-
0,03
0.001-
0.01
ppm Na
0.3-
3
0.3-
3
0.1-
1
0.1-
1
ppm Ni
0.01-
0.1
0.003-
0.03
0.03-
0.3
0.003-
0.03
o
z
-------
TABLE 2-30
ATOMIC ABSORPTION RESULTS FROM ANALYSIS OF S02 IMPINGER
LIQUID FROM COPPER AND LEAD SMELTING
Source
Copper Smelter
Lead Smelter
Ca
105
<20
<20
<20
ppm (yg/£)
Zn Na
105
<3.5
5
5
<30f
<30
70
<30
- *
Cu
<2
<2
6
6
Remarks
No fine filter
With fine parti cul ate
No fine filter
With fine parti cul ate
filter
filter
By emission spectrescopy
The relatively higher (30 ppm) detection limits are due to the small vol
ume of stack gas sampled (
2-78
WALDEN RESEARCH CORPORATION
-------
CV of barium chloranilate. No significant quantities of Na are present,
therefore, Ba is not affected.
The SCL results for the lead smelting operation are given in
Table 2-27. Here, the fine particulate filter does improve the agreement
between the two methods slightly. The regression slope is 0.89 without the ,
filter and 0.96 with the filter as indicated in Table 2-28. The correlation
coefficients do not change appreciably. The CV's indicate an interference
with the Ba titration which is removed by filtration. Atomic absorption
results in Table 2-30 show that the Na concentration is reduced considerably
between a filtered and unfiltered S09 sample. This may very well account for
++
the improved precision with the Ba titration. The Ba Chlor method, how-
ever, shows a considerable interference which is only slightly diminished
with the filter. From the data in Table 2-30, the trace amounts of both Zn
and Cu are present at about the same concentration in filtered and unfil-
tered samples. This could account for the interference.
In order to check the accuracy of the spectrographic analysis,
some SOp impinger samples were analyzed with a Cu electrode. Chloride was
added to decompose the residual H?0? prior to analysis. Two filtered samples
++
from the Cu smelter gave 0.07 and 0.1 ppm Cu for the emission spectro-
graphic range 0.1 to 1 and several unfiltered samples gave 0.26 and 0.82 ppm
Cu for the spectrographic range 0.3 to 3. The agreement between the two
methods indicates that the emission spectrographic concentrations are in the
right range. We should point out, however, that only a single sample (fil-
tered and unfiltered) were sent out for analysis. Therefore, these data may
not be representative of the actual concentration.
SO- samples were collected at the smelting operations by con-
trolled condensation and 80% IPA absorption. The 80% IPA samples were
flushed with air (passed through a charcoal column) at a rate of approxi-
mately 3 1/min for 10 minutes. The condenser and IPA S0~ samples were
O
analyzed for sulfate by the barium chloranilate method. The results SBT
lected for inclusion in Table 2-31 were those where the S02 values for
trains A and B were nearly equal. For example, if the S02 values differed
by 50%, this might indicate a leak in one of the trains, thus, these data
would not be valid and were not used.
2-79
WALDEN RESEARCH CORPORATION
-------
TABLE 2-31
S03 TEST RESULTS FROM Cu AND Pb SMELTING OPERATIONS
ppm SO-
'!:.'. . * -J
Controlled Condensation
30
6
4-
17 ' •
8
'10
2
7
4
6
11
11
ii
17
1
15
1
7
3
1
10
1
-. 8
1 .
*
Collected in parallel trains
' • ' . ' •
'....'', . . • ' * :
80% I PA Absorption
11
n - - •
• 1 5 ' • • •
' 5 '
10 :
10
id •
6 •
10
8
30
;32 ,
19
22
23
23
9
10
12
6
9
• 8 .
12 . .
8.
from the same probe
Source
Cu Smelting
Cu Smelting
Cu Smelting
Cu Smelting
Cu Smelting
'CU Smelting
Cu Smelting
Cu Smelting
Cu Smelting
Cu Smelting
Cu Smelting
Cu Smelting
Cu Smelting
Cu Smelting
Cu Smelting
Cu Smelting
Pb Smelting
Pb Smelting
Pb Smelting
Pb Smelting
Pb Smelting
. Pb Smelting
Pb Smelting
Pb Smelting
2-80.
WALDEN RESEARCH CORPORATION
-------
The mean SO- values for the controlled condensation and IPA are
7.4 and 12.5 ppm, respectively. The regression analysis obtained is:
ppm SO- (IPA) = 10.2 + 0.3 ppm SO- (controlled condensation)
with a correlation coefficient of only 0.30. The IPA is very susceptible to
interference (oxidation of dissolved S02) because of the high concentration
of SO- in the effluent and the presence of metals (catalysts) in the sample.
Note that low so values would be expected due to the stack temperature
O
(^ 140°F) and the sampling site, e.g., after particulate collection device.
The SO- test results on the copper (roasting and reverberatory)
smelter vary from approximately 2000 to 5000 ppm. The CO- and 0- analyses
in Table 2-26 for this operation indicate dilution with approximately 5
volumes of air. On this basis, the SO- values would be approximately 1-2.5%
SO-. This is in reasonable agreement with the predicted values in the pre-
vious section (Table 2-24). The emissions are highly variable. If we con-
sider the reverberatory firing rate in Table 2-32, we note that a 50% drop
was recorded between 11 and 10 o'clock on 2/1. The average SO- concentra-
tion dropped from approximately 3700 ppm to approximately 1700 ppm. How-
ever, with a constant firing rate all afternoon, the S02 increased from 2000
to 5000 ppm and dropped back down again. Thus, one cannot correlate these
results with the firing rate since the S02 depends on variables such as per-
cent sulfur in the ore, the time during the cycle, etc. The low values,
samples 10 and 12, represent periods when the smelter was shut down during
the actual sampling period. There appeared to be some problems with the
sampling trains leaking (#15, 24, etc.) on the copper (R&R) stack but the
system could not be leak-tested without removing the probe from the duct.
The probe was leak-tested each day. It would have been difficult and time-
consuming to leak-test the system between runs.
The fourteen runs at the copper converter (where the CuS produced
in the reverberatory operation is converted to S0? and Cu) yielded SO- con-
++
centrations less than 100 ppm for both Ba and Ba Chlor. Since this opera-
tion accounts for the major portion of SO- emissions from Cu smelters, the
data are certainly not typical of expected levels (SO- concentrations from
2-81 WALDEN RESEARCH CORPORATION
-------
TABLE 2-32 . :
LOG OF OPERATIONS FOR COPPER SMELTER
REVERB AND ROASTER FLUE
Date
2/1
2/6
Time
10:30 a.m.
1 :45 p.m.
4:00 p.m.
12:04 p.m.
1 :52 p.m.
2:40 p.m.
3:05 p.m.
- 11:11 a.m.
- 4:00 p.m.
- 4:32 p.m.
- 1:52 p.m.
- 2:40 p.m.
- 3:05 p.m.
- 5:30 p.m.
Roasters
Operating
4
; 2
2
. 4
2
2
2
Reverb Furnace
Firing Rate
150,000 CFH
70,000 CFH
65,000 CFH
135,000 CFH
135,000 CFH
100,000 CFH
90,000 CFH
CONVERTER FLUE
Date .
2/7
2/8
2/8.
Time
10:56 a.m.
1 :40 p.m.
7:40 a.m.
8:26 p.m.
- 11:45 a.m.' :
- 2:00 p.m.
- 8:10 a.m.
- 9:57 p.m.
Converters Operating
1 converter
1 converter
2 converters
3 converters
LOG OF OPERATIONS FOR LEAD SMELTER
Roasters Operating
Date
2/9
I-"
2/10
1:20
3:00
3:40
5:25
5:40
8:30
Time
- 1 :50 p.m.
- 3:40 p.m.
- 4:35 p.m.
- 5:40 p.m.
- 11 :00 p.m.
- 10:45
1st Stage
2
2
4:
4
2nd Stage
1
1
1
, 2
2
2-82
WALDEN RESEARCH CORPORATION
-------
% 1.5-8% were expected). Orsat analysis of flue gas samples taken through
the probe yielded CL values which were from 5.5-7.5% and COp values of 0%.
The negligible C0? values are reasonable since no auxiliary fuel is used in
this operation. During the testing period (10:00 a.m. to 1:00 p.m. on 2/7
and 7:00 to 10:00 a.m. on 2/8), one to four converters were running. Thus,
the low S02 values cannot be explained on the basis of the process not oper-
ating or a broken or leaky probe.
2.5.3 Conclusions
The utilization of a fine particulate filter in the sampling
train at the copper smelter improved the precision of the barium chloranilate
method considerably. A measurable effect on CV was not observed with the
barium titration and no problem was noted with the endpoint of the titration
(filtered or unfiltered) since the S0? concentrations were very high. On
the lead smelter, however, the fine particulate filter improved the CV of
the barium ion titration. This was attributed to the presence of sodium in
the samples without the fine particulate filter. No change in precision
was seen for the barium chloranilate. Atomic absorption results indicated
traces of Cu and Zn at about the same level in the samples with and without
the fine particulate filter. The S03 concentrations by the controlled con-
densation method are ^ 1/2 the values obtained by 80% IPA absorption. The
controlled condensation values seem more reasonable because the stack tem-
perature is so low and the sampling location follows a particulate collector.
The IPA values are higher due to oxidation of dissolved SO^.
The fine particulate filter does improve the analytical results
for Cu and Pb smelters even when the particulate concentrations are low,
e.g., after an electrostatic precipitator or baghouse.
2.6 SULFURIC ACID PRODUCTION
2.6.1 Brief Description of Sulfuric Acid Plants
Sulfuric acid manufacture is, by far, the largest industrial
chemical process. Production in 1967 amounted to 28,800,000 tons (MCA and
U.S.P.H.S., 1965). Approximately 97% was produced by the contact process,
2-83 WALDEN RESEARCH CORPORATION
-------
and the remainder by the chamber process. The latter process is rapidly de-
creasing in importance, in 1963, chamber process acid accounted for about 97%
of the total (MCA and U.S.P.H.S., 1965). Both .processes are potential
sources of both sulfuric acid mist (SO, or.HLSO.) and sulfur dioxide gas
(SOp). The chamber process also releases nitrogen oxides.
Flow diagrams of the chamber and contact processes are shown in
Figures 2-19 and 2-20, respectively. In both processes, the principal point
of emissions is the exit stack from the final absorption tower, the Gay-
Lussac tower (chamber) or the final absorber (contact). The pollutant spe-
cies in the tail gas stream are indicated at the right of each figure.
Uncontrolled emissions of SO,, range from 1000 to 2000 ppm (by
.volume) in the chamber process and from 1000 to 6000 ppm in the contact proc-
ess (EPA, 1971). This is equivalent to about 15 to 85 pounds of SO^ per ton
of acid (100% H,,S04) produced (EPA, 1971). Emissions of acid mist range
from 5. to 30 milligrams per cubic foot in the chamber process (EPA, 1971).
The efficiency of the final absorption step is generally higher in the con-
tact:process, so that acid mist emissions tend to be lower, 3 to 15 milli-
grams per cubic foot (EPA, 1971). Excursions to about 50 milligrams per
cubic foot have been reported (EPA, 1971 a). The normal range of emissions
corresponds to an output of approximately 0.5 to 3 pounds of acid mist (FLSO.)
per ton of acid produced.
The Federal Standards of Performance for new stationary sources
(EPA, 1971a) require significantly lower emission levels for both SO,, and
acid mist than are customarily found.in uncontrolled plants. For SO,,, the
standard limits emissions to 4 Ib per ton of acid produced, and for acid
mist, 0.15 Ib per ton of acid produced, both averaged over two hours.
Control processes for S0? generally consist of utilizing addi-
tional absorbers (dual-absorption plants). In such plants, the SO,, levels
range from about 90 to 250 ppm (EPA, 1971 a) or from 1.2 to 3 Ib per ton of
acid produced. Alternatively, tail gas scrubbing with sodium sulfite-bisul-
fite can be used to reduce S02 emissions with essentially equivalent results.
2-84 WALDEN RESEARCH CORPORATION
-------
EXIT GAS AIR. SO... ACID MIST.
NO. AND NO.
78% TO ACID STORAGE.
NUTROUS VITRIOL
CHAMBER ACID
fV>
I
CD
cn
SULFUR
AMMONIA
OXIDATION
UNIT
OXIDES OF NITROGEN
SECONDARY AIR
"I! MOLTEN irSN>—
^
^
\*
\
h
i\
_fc
1
r
GLOVER
TCWER
WATER ATOMIZERS
GAS
FAN
SULFUR
.
i_
/??. ,*,\
(One to twenty of
5,000 to 500.000
cu ft capacity each
in plants of various
capacities)
»
ACID COLLECTING PAN
-TON ,fa /^N
• LEAD CHAMBER
ACID COLLECTING PAN
60- 70%
CHAMBER
ACID
SULFUR
BURNER
COMBUSTION
CHAMBER
SUPPLY
TANK
SUPPLY
TANK
PUMP
SUPPLY PUMP
TANK
%
O
m
z
2)
m
•a
o
8
TO
Figure 2-19.
Simplified Flow Diagram of Typical Lead-Chamber Process for Sulfuric Acid
Manufacture (Based on Use of Elemental Sulfur as the Raw Material).
From MCA and U.S.P.H.S. (1965).
o
-------
t\J
I
DRYING <- MOLTEN SULFUR WASTE "EAT
TOWER BOILER
HOT :",Ax
FU '10.
EXIT GAS AIR. SO=, SO,. AND
H.SO, ACID MIST
HOT AIR
EXIT
i OLEUM
"•tjlj TOWER
SO T COOLER •
BLOWER
(When four-pass converter is used, cooling after both second and third passes is done by air quench.)
ECONOMIZER
ABSORBING
TOWER
I
5
TO
m
in
o
8
TO
s
TO
Figure 2-20. Gas Flow Diagram for Typical Sulfur-Burning Contact Plant in which Air Quench is
Used for Part of Converter Interstage Cooling.
-------
Acid mist control can be achieved by demisting with high effi-
ciency fiber filters or electrostatic precipitators. Typical data on the
performance of fiber mist eliminators is given in Table 2-33 (EPA, 1971).
In both cases, the emissions can be reduced to about 0.1 pound per ton (EPA,
1971a). This is equivalent to a concentration of about 0.5 mg per standard
cubic foot. Measurements of acid mist in the effluent of plants where S0?
controls have been installed, indicate that the additional absorption can
reduce acid mist to levels below the emission standard (EPA, 1971a).
2.6.2 Sampling and Analytical Results
The sampling train utilized in the sulfuric acid plant was the
conventional EPA SOp train. No probe was used in the tests. The impingers
were attached to a gate valve which penetrated approximately 6 inches into
the duct (see Figure 2-21). The plant layout is given in Figure 2-21. The
test results for the sulfuric acid plant are shown in Table 2-34. For runs
1-15, two or three SO trains (A and B or A, B and C) were run in parallel
/*
off the gate valve. All the SO- up to run 57 was collected by absorption in
80% isopropanol. Some were analyzed by barium chloranilate and others were
analyzed by barium ion titration as indicated. The "C" trains were usually
(but not always) run with solid sorbents in front of the S09 impingers. The
^ *
low SOp values, 20 to 300 ppm, were obtained with the solid sorbents in the
train while the high SO-, approximately 1000 ppm, represent runs with the
conventional EPA train (IPA plus peroxide impingers for collection of SO-
*3
and SO-, respectively).
The sampling rate was 1.0 1/min for approximately 15 to 30 min-
utes. Statistical analysis of the SOp data in Table 2-34 (runs 1-57) gives
ppm S02 (Ba++) = +32.6 + 0.97 ppm S0£ (Ba Chlor)
with a correlation coefficient of 0.96. No problems (interferences) were ob-
served with the conventional EPA train. The SO- data were split into those
analyses which were done by the two analytical methods and the following
*
These results will be discussed in detail in Appendix C.
2-87 WALDEN RESEARCH CORPORATION
-------
TABLE 2-33
COLLECTION OF H2S04 MIST FROM A SULFUR-BURNING CONTACT
SULFURIC ACID PLANT WITH FIBER MIST ELIMINATORS
.
Contact plant
production
Mist loading" of
gases leaving
absorber and
entering mist,
eliminator,
mg H..SO(/scf
Mist loading"
of gases
leaving mist
eliminator.
ing II..SO,/scf
Particle
collection
efficiency
(3,, and smaller), %
/
Mist Eliminator A1'
99 r! H..SO, and
05%' oleum at
full capacity
99% H,SO, and
25 r', oleum at
75% capacity
30.9
31.8
39.9
6.55
8.75
0.04
1.50
1.4,1
l.f)4
0.124
0.10!)
0.125
95.1
95.3
95.9
98.1
9H.1
1)8.1
Mist Kliminator IV1
99% H,SO, and
.25%. oleum at
full capacity
99% H..SO, and
25% oleum at
60%, capacity
99% H..SO, at
00% capacity
14.4
18.3
19.3
6.88
2.12
0.085
0.112
0.095
0.045
0.014
99.4
99.4
99.5
99.3
99.3
"The mist-loading values are limited to particles 3 microns in diameter and
smaller.
'•Mist eliminator A was designed for 100 percent efficiency for particles larger
than 3 microns diameter and for 95 percent efficiency for particles 3 microns
and smaller, operated at a pressure drop of 3 inches of water. Mist eliminator
B was designed for 100 percent efficiency for particles larger than 3 microns
and for 99 percent efficiency on particles 3 microns diameter and smaller.
operated at a pressure drop of 6 inches of water.
From MCA (1965).
WALDEN RESEARCH CORPORATION
-------
i
oo
VO
TO
m
8
TO
s
TO
SAMPLING
PORT
t
CROSS SECTION CF
SAMPLING PORT
SULFUR
BURNER
CONVERTER
CONTACT
TOWER
ID
FAN
DEMISTER
STACK
/"///'/'/ 7 '/'/ Y '//// '/'/
Figure 2-21. Sulfuric Acid Plant Layout.
-------
TABLE 2-34
S02 AND SOa TEST RESULTS FROM A
SULFURIC ACID PLANT
Sample No.
1A
IB
2A
2B
3A
3B
4A
4B
4Ctt
5A
5B
5Ctt
6A
6B
6C
7A
7B
7C
8A
8B
8C
9A
9B
9C
10A
10B
IOC
11A
11B
11C
12A
12B
12C
13A
13B
13C
14A
14B
14C
15A
15B
15C
ppm SOg
(Ba Chlor)
1130
1136
1177
1189
1080
1122
1026
987
68
1159
1101
0
1080
1110
1097
1089
1099
1115
1048
1026
1035
1155
1188
1179
1186
1171
1179
1163
1160
1151
829
865
1055
1082
1098
1061
1096
1007
1103
1188
1249
848
ppm SO?
(Ba++)
1108
1086
1112
1107
1049
1049
1002
876
12
1006
1080
8
1078
1078
1088
1135
1104
1130
1082
1092
1096
1210
1218
1206
1170
1169
1216
1150
1180
1172
1044
1080
1066
1110
1090
1114
1113
1078
1111
1113
1035
947
ppm S03
(Ba Chlor)
23
33
37
21
46
23
74
109
—
6
15
—
21
21
19
21
24
23
26
25
32
22
18
25
20
20
15
13
8
14
53
5
48
41
18
34
32
18
39
37
26
46
2-90
WALDEN RESEARCH CORPORATION
-------
TABLE 2-34 (continued)
Sample No.
16A
16B
16Ctt
17A
17B
17Ctt
18A
18B
ISCtt
19A
19B
19Ctt
20A
20B
20Ctt
21A
21 B
21C
22A
22B
22Ctt
23A
23B
23Ctt
24A
24B
24Ctt
25A
25B
25Ctt
26A
26B
26Ctt
27A
27B
27C
28A
28B
28C
29A
29B
29C
30A
306
30C
31A
31 B
31 C
ppm S02
(Ba Chlor)
765
1193
151
1145
893
172
1178
1149
243
1192
1199
117
1171
1182
211
1228
1261
1202
1117
965
65
1182
2385
40
1212
1085
10
1032
1191
15
1136
1118
30
1128
1135
1103
990
935
1001
1093
993
1018
1029
1056
1186
1250
1127
972
ppm SO?
(Ba++)
882
1173
199
1167
890
101
1143
1125
109
1133
1097
79
1153
1107
115
1158
1321
881
1062
1025
59
1161
1104
53
1120
1097
73
1054
1141
56
1137
1115
37
1137
1149
1188
1041
1014
1069
1097
1104
1130
1141
1135
1188
1225
1166
1053
ppm SOa
(Ba Chlor)
19
19
--
27
21
--
23
22
__
5
7
--
11
48
—
4
4
48
24
31
—
26
21
--
6
—
—
21
17
—
14
20
--
18
15
16
8
10
17
7
—
__
48t
54t
47t
47t
55t
59t
2-91
WALDEN RESEARCH CORPORATION
-------
TABLE 2-34 (continued)
Sample No.
32A
32B
32C
33A
33B
33C
34A
34B
34C
35A
35B
35C
36A
36 B
36C
37A
37B
37C
38A
38B
38C
39A
39B
39C
40A
40Btt
40C
41A
41 Btt
41 C
42A
42B
42C
43A
43B
43C
44A
44 B
44Ctt
45A
45B
45Ctt
46A
46B
46Ctt
ppm SOg
(Ba Chlor)
1227
1247
1300
1340
1348
1654
1463
1270
1193
1108
1166
1113
555
962
699
692
1006
859
1791
1255
1090
1087
1065
1033
986
38
687
1050
74
1038
1045
972
1112
1098
1143
111
922
1116
53
1170
1141
18
757
1350
38
ppm S02
/ D •> TT 1
v Do 7
1259
1261
1292
1255
1277
1275
1487
1226
1258
1109
1181
1120
430
1090
1084
800
1178
1232
1759
1145
1149
1245
1174
1191
1103
31
827
1127
70
1315
1090
1012
1259
1231
1225
123
351
1219
69
1193
1214
90
653
1325
144
ppm, $03
52
68
54
16
18
34.
19
28
34
28
24
34
1356
29
35
30
37
33
50
36
39
40
49
65
29
21*
32*
35
5473
61
46
76*
32*
48*
47
—
376
43
—
7
12
__
—
'
--
2-92
WALDEN RESEARCH CORPORATION
-------
TABLE 2-34 (continued)
Sample No.
47A
47B
47C -!••:•
48A
48B
48Ctt
49A
49B
49Ctt
50A
BOB
50Ctt
51A
51B
51Ctt
52A
52B
52Ctt
53A
53B
53Ctt
54A
54B
54Ctt
55A
55B
55Ctt
56A
56 B
56Ctt
57A
57B
57Ctt
ppm SOo
(Ba Chlor)
1170
1089
32
1135
1144
25
1064
1055
305
1112
1036
36
1081
1111
22
1243
1234
--
930
1164
40
945
1158
82
1170
1161
104
1139
1156
--
941
1089
102
ppm SO?
(Ba++)
1106
53
87
1137
1139
75
1076
1085
93
1133
1145
144
1205
1145
81
1235
1266 '
83
1236
1246
87
1315
1223
105
1105
1242
95
1215
1208
62
1063
1147
41
ppm $03
(Ba++)
163
235
--
89
39
—
33
23
--
29
28
.
28
23
--
49
41
--
35
: 19
--
48
40
--
68
81
--
24
22
--
45
10
"
t
tt
Sample analyzed by barium chloranilate method.
Sample analyzed by barium titration method.
Represents amount collected in peroxide impingers following a solid sor-
bent cartridge (see Appendix C).
2-93
WALDEN RESEARCH CORPORATION
-------
statistical results were obtained:
Runs 1-29, Ba Chlor Analysis:
ppm S03 (IPA Train B) = -2.4 + 1.03 ppm SO^ (IPA Train A)
Correlation coefficient =0.80
Runs 30-57, Ba++ Titration:
ppm S03 (IPA Train B) = -1.8 + 1.06 ppm SO- (IPA Train A)
Correlation coefficient =0.87
The SO- values for the two methods were in good agreement as expected. The
SO- data showed good agreement for both analytical methods when 80% IPA was
used in both trains. The high correlation coefficients and near unity
slopes are considerably better than any previous SO, comparisons when IPA
was run versus the controlled condensation method. Thus, the problem is not
the analytical method.
The S0? and SO, data in Table 2-35 were obtained approximately
one month later than that reported in Table 2-34. These data were collected
to obtain some comparison between IPA and controlled condensation collection
of SO- for a sulfuric acid plant. All the samples (SO- and SO.) were analy-
zed by barium ion titration. The statistical analysis obtained was:
ppm SO- (Train B) = +3.3 + 0.99 ppm SO- (Train A)
Correlation coefficient = 0.99
ppm SO, (controlled condensation) = 0.13 + 0.45 ppm SO, (IPA)
0 O
Correlation coefficient = 0.80
These data indicate that the SO- data by controlled condensation produce con-
siderably lower SO- values (note slope of 0.45) than IPA although the SO-
data on the same trains produce high correlations and a unity slope. This
is essentially the same as seen for power plants and smelting operations
where other comparisons were conducted for SO-.
2-94 WALDEN RESEARCH CORPORATION
-------
TABLE 2-35
S02 AND S03 TEST RESULTS FROM A SULFURIC ACID PLANT
Sample No.
Train a
ppm SO,
ppm SO,
Train b1
ppm SO,
ppm SO-
59
60
61
62
63
64
65
66
67
68
69
70
71
SO, collected
J IPA
fS03 collected
840
893
868
911
889
887
772
855
907
942
961
797
680
in impingers
by condenser
5
10
6
4
7
5
3
4
1
6
7
3
6
containing 80%
622
885
890
925
849
923
802
875
904
941
889
864
833
S02 collected in
in all trains
4
6
4
1
3
2
3
1
4
1
2
1
2
2-95
WALDEN RESEARCH CORPORATION
-------
2.6.3 Conclusions
The EPA sampling train and both the barium ion titration and
barium chloranilate methods yield reliable data without any modification.
A considerable discrepancy exists between the controlled conden-
sation and 80% isopropanol absorption methods for SCL collection. This
should be investigated in detail.
2.7 RECOMMENDATIONS FOR FUTURE WORK
In Section 1, we have seen that the SCL results from the controlled
condensation method and IPA absorption methods differ considerably. The
correlation coefficients which vary from 0.1 to 0.4 for the field samples
are very poor. Laboratory work with only SOp added to the pilot plant ef-
fluent fired on gas indicates that SOp is soluble in IPA to the extent of
^ 100 ppm (at ice temperatures). This is consistent with the field test
data for coal-oil combustion where train c which not flushed gave SOp
values about 100 ppm lower than train b where the SOp was removed by
flushing. The laboratory studies indicated that after 15 minutes of
flushing the IPA impingers with air, the S0~ background was reduced from
20% (100 ppm) to a minimum value of 1-2%. It is interesting that the SO,
values typically found are about 1-3% of the SOp values. We appear to be
getting conversion to sulfate although no trace metals are present in the
effluent. /
In Section 2 we find results for SO- similar to those in Section 1.
The condenser train has little correlation with the SOg collected in the
IPA. We have not done enough laboratory work to establish which collec-
tion technique gives the correct SO., values although the absorption in IPA
is probably subject to more errors as a result of catalytic oxidation of
the dissolved SOp.
We recommend that additional laboratory and field studies be conducted
to determine the best method for collection of SO-.
2-96 WALDEN RESEARCH CORPORATION
-------
REFERENCES
Cowan, S., "Sulfur Balance on a Gray Iron Foundry," American Foundrymen's
Society Meeting (May 1971).
Danielson, J., Ed., "Air Pollution Engineering Manual," U.S. Dept. HEW,
Pub. #999-AP-40 (1967).
Driscoll, J. N. and A. W. Berger, "Improved Chemical Methods for Sampling
and Analysis of Gaseous Pollutants from the Combustion of Fossil
Fuels," Vol. I, Sulfur Oxides, APTD #1106 (1971).
Driscoll, J. N., J. H. Becker and J. F. McCoy, "Assistance to Southwest
Research Institute on a Collaborative Testing Program," Subcontract
under EPA Contract No. CPA 70-40 (1972).
EPA, "Standards of Performance for New Stationary Sources," Federal
Register (Dec. 23, 1971).
Environmental Engineering, "Background Information for Establishment of
National Standards of Performance for New Sources," for EPA under
Contract No. CPA 70-142, Gray Iron Foundries (1971), Kraft Mills
(1971a), Iron and Steel (1971b), Smelting (1971c).
Galeano, S. F., T. W. Tucker and L. Duncan, "Determination of Sulfur
Oxides in the Flue Gas of the Pulping Processes," J. Air Pollution
Control Assoc. 22., 790 (1972).
Giever, P., "Control of Emissions from Gray Iron Foundries," American
Foundrymen's Society Meeting (October 1970).
lapalucci, T. L., R. J. Demski and D. Bienstock, "Chlorine in Coal Com-
bustion," U.S. Bureau of Mines Report RI-7260 (1969).
MCA and PHS, "Atmospheric Emissions from Sulfuric Acid Manufacturing
Processes," U.S. Dept. HEW, Pub. #999-AP-13 (1965).
McCrone Associates, "Final Report on KPL Field Tests," EPA Contract No.
EHS-D-71-25 (1971).
Ryason, J. and R. Harkins, "A Control Technique for S02 and NOX," J.
Air Pollution Control Assoc. 1_7, 1813 (1967).
Smith, W. S. and C. W. Gruber, "Atmospheric Emissions from Coal Combus-
tion," U.S.P.H.S. Report #999-AP-24 (1966).
Vaijga, A., et al., "A Systems Study of the Integrated Iron and Steel In-
dustry ."""Contract No. PH 22-68-95 (1969).
West, P. and H. Gaeke, "Fixation of S02 as the Disulfitotetrachloromer-
curate Complex and Subsequent Colorimetric Determinations," Anal.
Chem. 28, 1816 (1956).
2-97 WALDEN RESEARCH CORPORATION
-------
APPENDIX A
PROCEDURE FOR SAMPLING AND ANALYSIS
A-l
WALDEN RESEARCH CORPORATION
-------
I. TEST REQUIREMENTS AND PLANNING
The initial step in the experimental procedure is to establish the ve-
locity, pressure, temperature, and moisture content of the stack gases.
Velocity measurements are required to permit integration over the cross
sectional area to compute the total volume of the effluent. Industrial
and power boilers and heaters are usually designed for exhaust velocities
of 2,000-3,000 ft/min, which is a convenient range for pi tot tube use.
The pi tot tube sampling points can be used subsequently for sampling of
gaseous pollutants.
II. SOURCE CHARACTERISTICS - PRELIMINARY ESTIMATES
To estimate sampling rates, the expected SO concentration may be cal-
A
culated since sulfur oxide emissions depend primarily on the sulfur content
of the fuel. Good correlation has been obtained between SOp (the major
species) concentrations in the flue gas calculated by stoichiometry and SOg
measured by manual wet chemical techniques, provided reliable sampling and
analytical methods are used. For oil- and coal-fired units, S0« concentra-
tion may be estimated (within 20%) from the fuel analysis (C,H,S) the fuel
feed rate and the amount of excess air. An exemplary calculation is given
below.
For the typical residual oil analysis given in Table A-l, the stoi-
chiometric oxygen requirement (no excess air) is shown below. The result-
ing flue gas composition for 15% excess air is given in Table A-2.
The concentration of S02 in the flue gas is seen to be 1270 ppm. For
well-controlled combustion at low excess air, SO- concentrations may be es-
timated as about 1% of the S02 concentration.
The total flue gas volume per unit weight of fuel consumed is readily
obtained from the molar volume at standard (or any convenient) temperature.
Volume flue gas at 60°F 56.61 mole v 379 ft3 v 2000# _ .,n nnn -.3
ton fuel ~ 100# fuel x mole x ton " *JU«UUU Tt
WALDEN RESEARCH CORPORATION
-------
TABLE A-l
OXYGEN CONSUMPTION FOR OIL COMBUSTION
Element Wt. %f
C
H
S
*
pound moles/100
typical residual
87.7
10.0
2.3
Ib fuel
oil analysis
Moles
87.7/12
10.0/1
2.3/32
*
Moles CL Needed
7.31
10.0/4 = 2.50
0.0719
TOTAL 9.88
TABLE A-2
FLUE GAS COMPOSITION (WET)
Species
co2
S02
N2b
Moles3
2 x 2.50
9.88 x 0.15
9.88 x 1.15 x 3.77C
TOTAL
7.31
= 5.00
0.0719
= 1 .481
= 42.75
56.61
%b
12.91
8.83
0.127
2.62
75.5
100.0
a. pound moles/100 Ib fuel
b. 15% excess air
c. (nitrogen + argon)/oxygen ratio for dry air
A-3 WALDEN RESEARCH CORPORATION
-------
Thus, by preliminary calculation, estimates of sulfur oxide flue gas
concentrations and emission parameters may be made to determine initial
values for sample volume and flow velocity.
III. SAMPLING COMPONENTS
The following sections describe an integrated modular flue gas
sampling apparatus for collection of S03 either by absorption in a solu-
tion consisting of 80% isopropanol-20% water, or by controlled condensa-
tion of the flue gas and collection of H2$04 by filtration, and collection
of S0? in midget impingers containing dilute hydrogen peroxide solution.
A. PROBE AND PROBE HEATING
The probe (see Figure 1-3) is constructed of a six-foot length
of pyrex tube with a 12/5 socket joint on the downstream end. The other
end of the probe, which protrudes into the stack, is fitted with a 1.5
inch diameter by 4-cm length of pyrex tube, loosely packed with quartz or
pyrex wool for particulate filtration.* The glass probe is inserted into
a stainless steel shell with stack adapter assembly which allows various
probe insertion depths. The glass probe is wrapped with 20-gauge, asbestos-
covered wire to allow heating of the glass insert above the acid dewpoint.
The probe is controlled by 120 VAC. This gives a temperature of 150°C with
room air pulled through the probe. A stainless steel extension tube per-
mits the sampling point to be extended to ^ 12 feet. An example of the
probe setup with parallel trains is shown in Figure A-l.
B. MIDGET IMPINGERS AND/OR CONDENSER
C. CRITICAL ORIFICE METER OR DRY TEST METER
See modifications to the glass wool filter in Section 2, page 2-14.
WALDEN RESEARCH CORPORATION
-------
STACK
QUARTZ WOOL PLUG
\ I
STAINLESS STEEL PROBE
STACK ADAPTOR
HEATED GLASS TRIDENT
Figure A-l. Schematic of Parallel Trains Used for $03 and S02 Collection.
A-5
WALDEN RESEARCH CORPORATION
-------
D. ADDITIONAL EQUIPMENT
Stop Watch - For measurement of sampling duration.
Thermometer - A dial thermometer or thermocouple (200-500°F) for
measuring the stack gas temperature.
Plastic Bottles - Polyethylene bottles for storage of impinger
and condenser samples.
E. REAGENTS
1. Isopropanol-Hater (IPA) for SO- Collection
Prepare 80% isopropanol by mixing four parts alcohol with
one part of sulfate-free distilled water or for rinsing condenser.
2. Hydrogen Peroxide (3%) for SO,, Collection
Prepare 3% hydrogen peroxide by tenfold dilution of Baker
Analyzed (or equivalent) 30% hydrogen peroxide. This reagent should be
prepared fresh daily and stored in polyethylene containers.
3. Distilled Water
Sulfate-free distilled water (see Section 4) is required for
rinsing out the SO- impingers.
IV. FIELD OPERATIONS
A. PREPARATION OF EQUIPMENT
The pressure drop across the orifice should be checked. If AP <
0.5 atm, the carbon vanes in the pump should be replaced. Critical orifice
calibration should be checked in the laboratory with a spirometer or gasom-
eter. Examine the particulate filter (glass or quartz wool) in the probe.
If this is heavily loaded, it should be replaced. Check heating elements
on probe module to ensure that they are operating properly. Check glass
equipment for cracks and replace broken components. Make certain that
WALDEN RESEARCH CORPORATION
-------
glassware to be used in the field is carefully packed. Check sampling
modules for loose connections.
B. SELECTION OF SAMPLING RATES
SO- - Collection of SO, by absorption in 80% isopropanol was in-
vestigated for midget impingers maintained at ice temperatures. The data
show that even at flow rates less than 3 liters/min, collection was ^ 95%
for two series impingers. Sampling rates of 3.0 liters/min or less are
recommended for SO- collection. The collection efficiency of the con-
trolled condensation method for hLSO. has shown to be 97% for flow rates
up to 20 liters/min (Driscoll and Berger, 1971).
S02 - The collection efficiency of SOp absorbed in 3% peroxide
solution in midget impingers was determined as a function of concentration
(200-2000 ppm), temperature (up to 40°C), and flow rate (0.5-5 liter/min)
(Driscoll and Berger, 1971). Over the temperature and concentration range
examined, no change in the collection efficiency was observed. With 15 cc
of 3% peroxide solution, 96% collection efficiency was obtained at sampl-
ing rates of 0.5 and 1 liter/min. Collection efficiency dropped to 90 and
87% at 3 and 5 liters/min, respectively. However, quantitative collection
can still be obtained if two series impingers are used. At the highest
flow rate (5 liters/min) some blowover and loss of sample occurred. At
high concentrations (4000 ppm), a 0.5 liter/min flow rate is recommended
while at S02 concentrations about 200 ppm, 3 liter/min flow rate and an
extended period of time would be recommended.
C. SAMPLE COLLECTION
The probe module is fitted to the stack flue. Power cords are
connected between the SO probe heater and the variable transformer. The
/\
probe is then heated to an operating temperature of 150°C. The SO- im-
pingers are filled with 15 cc of 80% isopropyl alcohol. These two im-
pingers are then immersed in an ice bath. The S0? impingers are each
charged with 15 cc of 3% peroxide solution. After the probe has reached
operating temperature, the pump is connected to the last impinger with
a vacuum hose and started from the switch on the control module. The
/\_7 WALDEN RESEARCH CORPORATION
-------
time is recorded. The operator checks AP across the critical orifice se-
lected (0.5, 1.0, 3.0 liters/min) and records the pressure and temperature
value at 5 minute intervals during sampling or alternatively takes the dry
gas meter and temperature readings at 5 minute intervals. At the end of a
20-60 minute sampling period (depending on the SO- concentration), the
pump is switched off and the time is recorded. The impinger train is dis-
connected from the probe, the ice bath is removed, and clean air is drawn
through the impingers for 15 minutes or the HpSO. is removed from the con-
denser by rinsing with 80% IPA. Collect the S0_ sample in a polyethylene
bottle for transport to the laboratory. Transfer the contents of the two
midget impingers (which contain the S02 sample) into a polyethylene bottle.
Rinse the impingers several times with sulfate-free distilled water and add
these washings to the contents of the polyethylene bottle. After the im-
pingers are charged with 15 cc of 3% hydrogen peroxide and 15 cc of 80% iso-
propanol (or the condenser is at operating temperature 60-70°C), the sys-
tem is ready for collection of additional samples.
V. ANALYTICAL METHODS
The following sections give a detailed description of the laboratory
methods used for S0? and SO- analysis.
/\_g WALDEN RESEARCH CORPORATION
-------
A. LABORATORY REQUIREMENTS FOR BARIUM CHLORANILATE
1. Equipment
a. Shaker, Burrell wrist-action
b. Centrifuge, small clinical type (capable of 2800-3000
rpm)
c. Analytical balance
d. Spectrophotometer for use in the visible region (@ 530
nm)
2. Glassware
a. 2, 5, 10 ml volumetric pi pets
b. 25, 50, 100, 1000 ml volumetric flasks
c. 50 ml buret
3. Reagents
a. Deionized water
b. 0.025M H2S04. Add 10 ml of 18M H2$04 (ACS reagent grade)
to 600-700 ml of water in a 1000 ml volumetric flask and mix by swirling
the flask. Dilute to the mark with water and mix well. Dilute 145 ml
of this 0.18M solution to 1000 ml and again mix well. Standardize
against anhydrous sodium carbonate.
c. Barium chloranilate (Fisher "certified" or equivalent)
d. Isopropanol (ACS reagent grade)
e. pH 5.6 -buffer. Add 50 ml of 0.2M acetic acid (11.4 ml
99% acid in 1000 ml of distilled water) to 500 ml of 0.2M sodium acetate
(27.2g reagent grade NaC2H303 • 3H20 in 1000 ml of water).
A-9 ' WALDEN RESEARCH CORPORATION
-------
f. (Approximately) IN NaOH. Slowly add 40 grams of NaOH
(AR grade) pellets to 800-900 ml of water in a 2-liter beaker with
stirring until all pellets are dissolved. Dilute to 1000 ml with water
and mix well. Store in a polyethylene or polypropylene container.
g. (Approximately) IN HC1. Add 86.3 ml of 11.6M HC1 (ACS
reagent grade) to 800-900 ml of water in a 2-liter beaker with stirring.
Dilute to 1000 ml with water and mix well. This solution can be stored
in glass.
h. 0.05% phenolphthalein. Dissolve O.OBg phenolphthalein
in 50 ml ethanol and dilute to 100 ml with water.
i. Anhydrous sodium carbonate (AR grade)
j. Methyl orange indicator. 0.1%. Dissolve O.lg in 100 ml
of water.
B. LABORATORY PROCEDURE FOR BARIUM CHLORANILATE
1. Standardization
a. 0.025M HpSO.. Heat approximately 3 grams of AR grade
anhydrous sodium carbonate for 4 hours at 250°C to decompose any re-
sidual bicarbonate and to remove water. Cool in a dessicator. Accu-
rately weigh 0.115 +_ 0.005g of the dried sodium carbonate into each of
three 250 ml Erlenmeyer flasks and dissolve the samples in 50 ml of
deionized water. Add 2 drops of 0.1% methyl orange and titrate with
the 0.025M H?SO. in a 50 ml buret to a color change from yellow to
red-orange. A blank of 50 ml deionized water plus indicator should
be determined with each set of titrations.
The normality of the 0.025M H2SO. solution is calculated
as follows:
Wt. of Na2C03 in g
N = (A-B)(oto530)
WALDEN RESEARCH CORPORATION
-------
where A = ml of titrant needed to reach the endpoint
B = ml of titrant needed for the blank
b. Absorbance. A standard curve is prepared by pipetting
0.5, 1, 2, 5, and 10 ml of 0.025M H2$04 into 50 ml volumetric flasks.
Add water to the first four to bring all volumes up to about 10 ml. Add
1 drop of phenolphthalein solution, then add IN NaOH dropwise to the ap-
pearance of a pink color. Add IN HC1 dropwise to the disappearance of a
pink color. (This will usually require just one drop.) Pipet 5 ml of
pH 5.6 buffer into each flask. Pipet 25 ml of isopropanol into each
flask. Mix well, then bring to the mark with water, stopper, and again
mix well. Pour the contents of each flask into a corresponding 125 ml
screw-top shaker flask containing 0.2-0.3g of barium chloranilate. Shake
for 20 minutes on a wrist-action shaker, then centrifuge 10-15 ml of this
suspension for five minutes at 2800-3000 rpm. Decant the centrifugate
into 1 cm cells and read the absorbance versus water at 530 nm. The
blank (no sulfate) versus water should read no more than 0.01 to 0.03 ab-
sorbance units. Plot the absorbance versus sulfate concentration in yg/
ml final solution.
2. Analysis for Sulfur Oxides
Quantitatively transfer the samples collected to a 50 ml
volumetric flask (25 ml vol. flask for S03). Pipet a suitable sized
aliquot into a 50 ml volumetric flask (25 ml volumetric flask for SO.,).
Add 1 drop of phenolphthalein solution to the flask, then add NaOH drop-
wise until the solution just turns pink. Add 1 drop of IN HC1 to return
the solution to colorless. Pipet in 5 ml of pH 5.6 buffer (2.5 ml for
SO- analysis), then add 25 ml isopropanol (12.5 ml for SO- analysis) and
mix well. Dilute to the mark with deionized water. Pour contents into
a 125 ml screw-top shaker flask containing 0.2-0.3g of barium chloranilate.
Shake for 20 minutes on a wrist-action shaker. Centrifuge 15 ml of the
solution at 2800-3000 rpm for six minutes, decant the solution into 1 cm
cells, then read the solution absorbance versus a water blank at 530 nm.
The concentration of SOp (SO.,) by volume is calculated as
follows:
A-ll WALDEN RESEARCH CORPORATION
-------
SO (SO ) = (A)(S)(50)*(24.1)(50/X)*
3^2 *OU3' 95 y
where X = volume of sample analyzed
A = absorbance of sample less absorbance of blank
S = slope of calibration curve (yg/ml/Abs unit)
V = sample gas volume (STP) = (sampling rate of orifice con-
verted to STP x time)
*
for SO- analysis use 25 and 25/X, respectively.
/\_12 WALDEN RESEARCH CORPORATION
-------
C. LABORATORY REQUIREMENTS FOR NaOH TITRATION
1. Equipment
a. Magnetic stirrer and stirring bars
b. Analytical balance
2. Glassware
a. 5, 10 ml volumetric pi pets
b. 25, 50, 1000 ml volumetric flasks
c. 50 ml beakers
d. 50, 100 ml graduates
e. 50 ml buret
3. Reagents
a. Deiom'zed water
b. 0.02N NaOH. Slowly add 0.8g sodium hydroxide (AR grade)
to 800-900 ml of deionized water. After the pellets are dissolved,
dilute to 1000 ml with deionized water and mix well.
c. Bromphenol blue indicator. Dissolve O.lg bromphenol
blue in 7.5 ml of 0.2N NaOH and dilute to 250 ml.
d. Potassium acid phthalate - primary standard
e. Phenolphthalein indicator. Dissolve 0.5g of phenol-
phthalein in 50 ml of ethanol and add 50 ml of water.
WALDEN RESEARCH CORPORATION
-------
D. LABORATORY PROCEDURE FOR NaOH TITRATION
1. Standardization
Weigh 2-3g of primary standard potassium acid phthalate
(KHP) into a weighing bottle and dry in an oven at 120°C for 2 hours.
Cool in a dessicator. Accurately weigh three 0.1 j^O.OOSg portions of
the dried KHP into each of three 250 ml Ehrlenmeyer flasks and dissolve
the sample in 50 ml of water. A blank containing no KHP should be run
with each set of samples. Add 2 drops of phenolphthalein indicator and
titrate with the NaOH to a faint pink color. Compute the normality of
the NaOH as follows:
M = 0 KHP
(0.204)(A-B)
where A = ml of titrant needed to reach the endpoint
B = ml of titrant needed for the blank
2. Analysis for Sulfur Oxides
Quantitatively transfer the sample collected to a 50 ml
volumetric flask; dilute to the mark with deionized water. Pipet a 5 ml
aliquot into a 50 ml beaker, add 2-3 drops of bromphenol blue indicator
and titrate to the blue endpoint using the standardized NaOH solution.
A blank of 10 ml deionized water and 2 drops indicator should be run
with each series of samples.
The concentration of SOp (SO.,) by volume is calculated as
follows:
ppm S02 (S03) - (A-B)(N/2)(24.1)(103)(50/X)
where A = ml of titrant needed to reach endpoint
B = ml of titrant needed for blank
N = normality of NaOH solution
. , , WALDEN RESEARCH CORPORATION
A-14
-------
X = volume of sample analyzed
V = sample gas volume (STP) = (sampling rate of orifice con-
verted to STP x time)
TE WALDEN RESEARCH CORPORATION
-------
E. LABORATORY REQUIREMENTS FOR Pb++ TITRATION .
1. Equipment
a. Lead electrode (Orion 94-82A)
b. pH/MV meter (Orion 701)
c. Double junction reference electrode (Orion 90-02)
d. Magnetic stirrer and stirring bars
2. Glassware
a. 2, 5, 10, 100 ml volumetric pi pets
b. 50, 1000 ml volumetric flasks
c. 250 ml beakers
d. 50 ml buret
e. 50 ml graduate
3. Reagents
a. 0.1M Pb(C104)2 (Orion 94-82-06)
b. Methanol (ACS reagent grade)
c. Zinc (metal)
d. Deionized water
e. (Approximately) 1M NaN03. Add 8.5g NaN03 to 60 ml of
deionized water in a 100 ml vol. flask; swirl to dissolve; dilute to
mark.
A-16 WALDEN RESEARCH CORPORATION
-------
f. Potassium hydrogen phthalate buffer (pH 4). Add 1.0
ml of 0.1M HC1 to 500 ml of 0.1M potassium hydrogen phthalate.
g. Orion 90-00-02 filling solution
F. PROCEDURE FOR Pb++ TITRATION
1. Calibration and Standards
Standards are prepared via a series of tenth dilutions of
the standard 0.1M Pb(C104)2 (Orion 94-82-06). Standardizing solutions
more dilute than 10 M must be prepared daily. The lead electrode (Orion
94-82A) is used with a double junction reference electrode (Orion 90-02)
filled with 1M NaN03 in the outer chamber and Orion 90-00-02 solution in
the inner chamber. The electrode is calibrated by measuring the poten-
tial of lO^M through 10"5M Pb(C104)2 solutions. The potential (MV) is
graphed versus the concentration on semi-log paper. The calibration of
the electrode should be checked in the desired range prior to beginning
analyses.
2. Determination of Sulfur Oxides
Quantitatively transfer the sample collected to a 50 ml
volumetric flask and dilute to the mark with deionized water. Pi pet a
suitably-sized aliquot (2, 5, or 10 ml) in a 250 ml beaker. If pH of
solution is greater than 1, acidify with dilute HC1. Add 0.1-0.2g zinc;
when pH of sample is 5-6 (5 to 10 minutes), add 5 ml of potassium hydro-
gen phthalate buffer (pH 4), sufficient deionized water to make a total
volume of 50 ml, and 50 ml of methanol (ACS reagent grade). Titrate the
sample with standard 0.01M PbCClO.Jp, monitoring the titration with the
lead electrode. Add the titrant in increments of 0.5 to 1.0 ml in the
beginning of the titration and about 0.1 to 0.20 ml in the endpoint re-
gion. Record the potential after each addition. Continue the titration
4-5 ml past the endpoint. Stir continuously throughout the titration.
Plot ml of Pb(C10.)2 versus MV readings on standard coordinate graph
paper. The point of greatest inflection is taken as the endpoint. A
typical potentiometric titration curve is shown in Figure A-2.
A-17 WALDEN RESEARCH CORPORATION
-------
i
oo
m
55
o
o
o
30
s
TO
150 -
250
16
Ml Pb
Figure A-2. Potentiometric Titration of Sulfate with a Pb Ion Selective Electrode.
-------
The concentration of S(L (SO.,) by volume is calculated as
fol1ows:
ppm S02 (S03) = ml(M)(0.01)(24.1)(103)(50/X)
where X = volume of sample analyzed
V = volume gas sampled (STP) = (sampling rate of orifice x time
converted to STP)
M = molarity of Pb(C104)2 solution
A-19 WALDEN RESEARCH CORPORATION
-------
G. LABORATORY REQUIREMENTS FOR Ba++ TITRATION
1. Equipment
a. Magnetic stirrer and stirring bars
b. Analytic balance
2. Glassware
a. 2, 5, 10 ml vol. pi pets
b. 100 ml beakers
c. 25, 50, 100, 1000 ml vol. flasks
d. 50 ml buret
3. Reagents
a. Deionized water
b. (Approximately) IN NaOH. Slowly add 40 grams of NaOH
(AR grade) pellets to 800-900 ml of water in a 2-liter beaker with
stirring. Dilute to 1000 ml with water and mix well. Store in a poly-
ethylene or polypropylene container.
c. (Approximately) IN HC1. Add 86.3 of 11.6M HC1 (ACS
reagent grade) to 800-900 ml of water in a 2-liter beaker with stirring.
Dilute to 1000 ml with water and mix well. This solution can be stored
in glass.
d. Sulfuric Acid, 0.01N. Dilute 36N ACS reagent grade
acid by pipetting 10 ml into a 1000-ml volumetric flask containing 800-
900 ml of water. Swirl, then dilute this to the mark with additional
water. Accurately dilute 30 ml of this 0.36N solution to 1000 ml by
adding the 30 ml to 800-900 ml of water contained in another 1000 ml
A-20 WALDEN RESEARCH CORPORATION
-------
vol. and, after swirling, make this up to the mark with additional water.
Stopper the flask and mix the contents. Standardize the 0.36N solution
against anhydrous sodium carbonate.
e. Anhydrous sodium carbonate - AR grade
f. Methyl orange indicator, 0.1%. Dissolve 0.1 gm of the
solid in 100 ml of water and filter, if necessary.
g. Isopropanol - ACS reagent grade
h. 0.01M barium perch!orate. Prepare 0.01M barium per-
chlorate by dissolving 3.9 grams of the solid Ba(C104). • 3FLO in 900 ml
of 80% isopropyl alcohol contained in a 1000-ml volumetric flask. Ad-
just the solution pH to 3.5 with dilute perchloric acid (several drops
of the dilute perchloric usually suffices), then dilute to the mark and
mix well. Standardize against 0.01N H2SO..
i. Perchloric acid, dilute. Add 1 volume of 70% AR grade
perchloric acid to two volumes of sulfate-free water,
j. Mixed indicator. Two parts 0.2% thorin to one part
0.0125% methylene blue, both solutions in sulfate-free water. Prepare
thorin by dissolving 0.2 gm of the solid in 100 ml water. Prepare
methylene blue (water soluble variety) by dissolving 0.0125 gm of the
solid in 100 ml water.
H. LABORATORY PROCEDURE FOR Ba++ TITRATION
1. Standardization
a. 0.36N HpSO.. Heat approximately 5 grams of AR grade
anhydrous sodium carbonate for 4 hours at 250°C to decompose any re-
sidual bicarbonate and to remove water. Cool in a dessicator. Accu-
rately weigh 0.860 ^0.001 g of the dried sodium carbonate into each of
three 250 ml Erlenmeyer flasks and dissolve the samples in 50 ml of de-
ionized water. Add 2 drops of 0.1% methyl orange and titrate with the
0.36N HpSO. in a 50 ml buret to a color change from yellow to red-
A-21 WALDEN RESEARCH CORPORATION
-------
orange. A blank of 50 ml deionized water plus indicator should be de-
termined with each set of titrations.
The normality of the 0.36N H2$04 solution is calculated as
follows:
Wt. of Na2C03 in g
Nl = (A-B)(0.0530)
where A = ml of titrant needed to reach the endpoint
B = ml of titrant needed for the blank
The normality (N^) of the diluted H^SO. solution is thus
30 x N,
1 -N,
1000 "2
b. 0.1M Ba(C104)2. Pi pet 10 ml of standard 0.01N H2$0.
into each of three beakers. Add 40 ml of isopropanol and two drops of
modified thorin indicator to each beaker. Titrate with the 0.01M
Ba(C10.)2 to a color change of yellow to pink. A blank of 40 ml iso-
propanol , 10 ml deionized water, and two drops of indicator should be
determined with the set.
The molarity of the 0.01M Ba(C104)2 is calculated as
follows:
M =
10(2)
where A = ml of titrant needed to reach the endpoint
B = ml of titrant needed for the blank
N2 = normality of H^SO. solution
2. Analysis for Sulfur Oxides
Quantitatively transfer the sample collected to a 50 ml
volumetric flask (25 ml vol. flask for SO-) and dilute to the mark with
deionized water. Pipet an appropriate aliquot (2, 5 or 10 ml) of the
A-22 WALDEN RESEARCH CORPORATION
-------
dilute sample into each of two beakers. Add 5 ml deionized water, 40 ml
isopropanol, and two drops of modified thorin (for S03 analysis use 0.2%
thorin) to each beaker. Titrate to the pink endpoint with the standard
0.01M Ba(C10.)2< A blank should be run with each series of samples.
The concentration of S0_ (SO.J by volume is calculated as
follows:
ppm S02 (S03) - (A-B)(M)(103)(g4.1)(50/X)'
where A = ml of titrant needed to reach the endpoint
B = ml of titrant needed for the blank
M = molarity of Ba(C10.)2 solution
X = volume of sample analyzed
V = sample gas volume (STP) = (sampling rate of orifice x time)
converted to STP
. for S03 use (25/X) factor.
/\_23 WALDEN RESEARCH CORPORATION
-------
APPENDIX B
POTENTIOMETRIC DETERMINATION OF S02 IN FLUE GASES
WITH AN ION SELECTIVE LEAD ELECTRODE
WALDEN RESEARCH CORPORATION
-------
I. INTRODUCTION
Sulfate can be determined potentiometrically by monitoring its titra-
tion with Pb with a lead ion selective electrode (Orion 94-82A). The
sample, in 50% methanol, is titrated with standard lead perchlorate, and
the electrode potential is measured throughout the titration with an Orion
Model 701 digital millivolt meter. The endpoint is determined by plotting
ml of titrant added versus the electrode potential. An example of the re-
producibility of the lead electrode is given in Table B-l. Note that the
columns represent seven different samples. The standard deviation (pre-
cision) is quite
centrations are:
cision) is quite good. For example, the a's for the different Pb con-
1 x 10"6 Pb++ a = 3.2 mV
1 x 10"5 Pb++ a = 1.3 mV
1 x 10"4 Pb++ a = 0.5 mV
II. CALIBRATION
The lead electrode is calibrated with a series of standards prepared
by the dilution of Orion standard 0.1M Pb (C104)2 solution (94-82-06) with
deionized water. The calibration curve (Figure B-l) is obtained by plotting
the mV reading versus the concentration on semilogarithmic graph paper. The
electrode response is Nernstean at lower levels of Pb , but deviates at
higher concentrations due to the increasing difference between activity
and concentration.
III. POTENTIOMETRIC TITRATIONS
Standard 0.02396M H^SO. was analyzed using the lead electrode to
study the accuracy of potentiometric titrations for sulfate determinations,
All titrations were run in 1:1 mixtures of methanol and water to decrease
the solubility of the PbSO. formed. A typical curve is given in Figure
B-2. The HpSO. solution was determined to be 0.02400M which agreed well
with the 0.02396M value determined with both sodium carbonate standardiza-
tion and titration with standard NaOH.
B-2
WALDEN RESEARCH CORPORATION
-------
I
o
m
yj
m
(S>
o
I
8
TO
TO
O
TO
o
TABLE B-l
REPRODUCIBILITY OF THE ORION ION SELECTIVE (Pb)
ELECTRODE ON CALIBRATION STANDARDS
"mud
1 x
1.1
w 1.1
co
Sample
*
it
10"6M Pb++
x 10"5M Pb++
x 10"4M Pb++
1
-236f
-222
-186
-159
2
-275
-225
-188
-160
3
-262
-224
-188
-160
Millivolts
4
-268
-226
-189
-160
5
-266
-226
-188
-160
6
-240
-217
-186
-160
7
-255
-224
-186
-159
AmV
36
36
28
*
Blank containing 50% deionized water-50% methanol
Not used in calculation of A
-------
-ICC
-150
>
£
-2CCh-
-250h
-275
ACT ! VITY
f /
•CONCENTRATION
moles per liter Pb
"1"1"
,o-6
"5
io'' 10°
Figure B-l. Calibration Curve Pb Electrode.
B-4
WALDEN RESEARCH CORPORATION
-------
CD
CJI
150
170
190
210
230
I
5
m
m
3J
O
o
O
aj.
s
250
8
Ml Pb"1
J
10
12
14
16
Figure B-2. Potentiometric Titration of Sulfate with a PbTT Ion Selective Electrode.
-------
IV. APPLICATION TO ANALYSIS FOR SULFUR OXIDES IN STATIONARY SOURCES
In addition to the various cations present in effluent samples which
interfere with titrations monitored by the lead electrode, e.g., Cu , Hg ,
+ J.O -L.1.
and Ag at any concentration, and Fe and Cd at concentrations greater
than that of Pb , there is also a very major interference inherent in the
sample collection liquid, e.g., HLCL. Sulfur dioxide is collected in 3%
hydrogen peroxide which oxidizes the SOI to SOT; H?0? will also oxidize
the surface of the electrode, causing erratic response and non-reproducible
results.
Initially, heating the solutions was attempted as a means of decom-
posing the peroxide; this did not decompose all of the peroxide, since sub-
sequent analyses of the solutions indicated that oxidation of the surface
of the electrode was still occurring.
A half reaction was sought which would have sufficient potential to
reduce the peroxide to water, but not interfere with the Pb titration.
The Zn/Zn couple was selected (Table B-2). Zinc metal (0.1-0.2g) was
added to a sample prior to its analysis. Completion of the reduction of
peroxide was noted by a decrease in pH from approximately 1 to 6. Potas-
sium hydrogen phthalate buffer (pH 4) was added to the sample to prevent
precipitation of Zn (DHL, which occurs at pH 6. The Zn/Zn also will
reduce the other cations known to poison the lead electrode (Table B-2),
thereby removing all major interferences to the titration. Titrations
performed using this method yielded'accurate and reproducible results
which compared well with values found using standard methods for SO^
analysis (Table B-3). A typical curve for a sulfuric acid plant sample
is seen in Figure B-3.
An alternate method of assimilating data is the use of Gran's plots
(Gran, 1952). These plots involve plotting ml of titrant vs potential (mV)
on semi-antilog paper which is designed to accommodate the volume change
which occurs during the titration. The basis for this selection of axes
is Nernst's equation:
g_g WALDEN RESEARCH CORPORATION
-------
TABLE B-2
STANDARD HALF-CELL ELECTRODE POTENTIALS
Ion-Electron Half Reaction
H202 + 2H+ +
Hg++ + 2e =
Ag + e = Ag
r +++ ,
Fe + e =
Cu++ + 2e =
Pb++ + 2e =
(Ca++ + 2e =
Zn++ + 2e =
2e = 2H20
Hg
r ++
Fe
Cu
Pb
Cd
Zn
E°, volts
+1.77
+0.854
+0.799
+0.771
+0.337
-0.126
-0.403)
-0.763
B-7
WALDEN RESEARCH CORPORATION
-------
TABLE B-3
S02 TEST RESULTS - COMPARISON OF ANALYTIC METHODS
co
i
00
o
m
z
W)
a
o
o
TO
s
TO
Sample No.
38
7A
13B
14A
15C
16A
25A
36C
40C
52B
54A
56A
50A
52A
54B
6B
HA
11B
12B
12C
13B
13C
29C
41B
58A
59A
59B
F13C
F15C
Endpoint
ppm S02 ppm S02 ppm SO?
(Pb++)t (Pb-H'-Gran's Plot) (Ba++)
..
--
1115
--
--
866
--
975
--
1165
1100
1183
1691
4754
4769
217
--
147
--
—
--
--
--
--
13
35
34
498
612
determined
1089
825
1149
1113
620
1228
811
1202
227
1435
1216
1244
2639
—
--
--
237
146
264
274
146
109
319
155
--
--
--
--
— —
potentiometrically
1049
1135
1090
1113
947
882
1054
1084
827
1266
1315
1215
2172
4822
5014
238
170
136
140
176
116
118
28
29
14
4
38
600
635
ppm S02
(Ba Chlor)
1122
1089
1098
1096
848
765
1032
--
687
1234
945
1139
2220
4696
4898
245
172
135
135
165
99
115
38
34
8
21
24
--
--
Source
Sulfuric Acid Plant
Sulfuric Acid Plant
Sulfuric Acid Plant
Sulfuric Acid Plant
Sulfuric Acid Plant
Sulfuric Acid Plant
Sulfuric Acid Plant
Sulfuric Acid Plant
Sulfuric Acid Plant
Sulfuric Acid Plant
Sulfuric Acid Plant
Sulfuric Acid Plant
Pb Smelter
Pb Smelter
Pb Smelter
Iron & Steel Plant ,
Iron & Steel Plant
Iron & Steel Plant -
Iron & Steel Plant
Iron & Steel Plant
Iron & Steel Plant
Iron & Steel Plant
Iron & Steel Plant
Iron & Steel Plant
Kraft Mill
Kraft Mill
Kraft Mill
Coal -Fired Power Plant
Coal -Fired Power Plant
o
-------
MV
-150 -
-IfeO-
-170-
-iec-
-190—
-200 —
-2IOU
-220
-230-
-240-
-250
I I I I I I I I
I I
01 2 34 56 78 9 10 II
ml Pb*
Figure B-3. Potentiometric Titration Curves for $62 Samples from a
Sulfuric Acid Plant.
B-9
WALDEN RESEARCH CORPORATION
-------
= E° + r log
Rearrangement yields
E°
-RT
nFE /nFE . , n r.
•- - + 1og [A
The ion sensed can be made a linear function of the electrode potential by
taking the antilog of both sides:
antilog (E/S) = E] + [A~]
where S = nF/RT and EI = antilog (E°/S)
The change in volume is corrected for, as previously mentioned, by shift-
ing the axis up 10% from left to right making additions of up to 10% of
the original sample volume possible. Gran's plot paper has a slope of
58 mV for a monovalent electrode and 29 mV for a divalent electrode. Cor-
rections for electrodes whose slopes differ from 58 mV can be achieved by
the use of a blank.
A major advantage to the use of Gran's plot paper (available from
Orion Research Corp.) is that the plot is linear, hence fewer points are
needed and the time per determination is reduced considerably. A straight
line is drawn through the points and the endpoint is found by extrapola-
tion to the x axis. With this method, points well beyond the endpoint can
be used to obtain the straight line and thereby the endpoint.
One difficulty which arises from the addition of zinc to the sample
to be analyzed is that Pb is reduced as well as the interfering cations.
This will cause points after the endpoint to be questionable. This could
affect the potentiometric curve by varying the slope in the endpoint re-
gion. Apparently, this reduction does not occur rapidly at low levels of
excess Pb and the potentiometric curve still yields accurate results.
Regression equations using the Ba titration and Ba Chloranilate methods
as standards give excellent correlation coefficients, small intercepts,
and slopes near unity (Table B-4). A regression of the Ba titration vs
the Ba Chloranilate method yields comparable results.
WALDEN RESEARCH CORPORATION
-------
TABLE B-4
REGRESSION EQUATIONS
Methods Compared
Regression Equation
Pb Titration (Gran's Plot)(x)
vs Ba Titration (y)
Pb++ Titration (Gran's Plot)(x)
vs Ba Chioranilate (y)
Pb4"1" Titration (x)
vs Ba++ Titration (y)
Pb"1"* Titration (x)
vs Ba Chioranilate (y)
Pb"1"1" Titration (x)
vs Pb Titration (Gran's
Plot)(y)
Ba"1"1" Titration (x)
vs Ba Chioranilate (y)
y = 0.854 x +95.6
y = 0.847 x +62.4
y = 1.04 x +41.4
y = 1.02 x +3.6
y = 1.47 x -233
y = 0.976 x -18.2
0.926
0.936
0.997
0.995
0.939
0.998
B-ll
WALDEN RESEARCH CORPORATION
-------
Gran's plots are, however, much more sensitive to the points after
the endpoint; slight variations can greatly affect endpoint determination
which is based upon extrapolation. This causes a loss in accuracy which
can be seen in the regression equations (Table B-4). The correlation coef-
ficients while still high are lower than those for the Pb titrations with
the endpoints determined potentiometrically. The slope varies ^ 15% from
unity indicating a loss in accuracy of the SCL levels determined using
Gran's plots. The use of an alternate reducing agent for peroxide, which
would not reduce Pb , might improve the feasibility of using Gran's plots
in stack gas S(L analyses.
WALDEN RESEARCH CORPORATION
-------
APPENDIX C
SOLID SORBENT STUDIES
C-l WALDEN RESEARCH CORPORATION
-------
The efficiency of several solid sorbents was investigated in the lab-
oratory by passing a gas mixture (SCL in nitrogen) through a packed tube
containing PbCL or ascarite. The sorbent tube was followed with an im-
pinger containing aqueous 3% hydrogen peroxide. The collection efficiency
was determined by
ppm SCL In-ppm SCL-impinger
- ppm S02 In - xl°°
and is given in Table C-l . Both the PbCL and ascarite show very high col-
lection efficiency at room temperature. The Rhom and Hass Amberlite XE309
resin was investigated at a sulfuric acid plant. The inlet S02 concentra-
tion was calculated from the average of the two peroxide impingers run in
parallel with the train. The S0? passing the ion exchange resin was
measured by placing a peroxide impinger behind the sorbent tube. These
data are given in Table C-l.
The purpose of the solid sorbent tests at the sulfuric acid plant was
to investigate the feasibility of using the ion exchange resin to collect
SCL and SCL simultaneously. Some preliminary tests by Allied Chemical
Corp. with this sorbent system spurred this work. In their work, they
found that SCL could be desorbed by passing hot gas over the sorbent. The
resulting gaseous SCL was collected in aqueous hydrogen peroxide. The SCL
was desorbed by pouring the ion exchange resin in an aqueous solution.
They analyzed the sulfate (SCL and SCL) by titration.
In our tests, we found that the SCL could be collected quantitatively
but we could not quantitatively desorb the SCL from the resin. We tried
passing hot nitrogen over the sample and recovered 5 to 10%. We could not
recover the remaining S00 by either leaching with water overnight or leach-
Ci
ing with boiling water.
The recovery for the laboratory tests on PbOg was also low as indi-
cated in Table C-2. Only the recovery with ascarite was quantitative.
Some of the problem with quantitative recovery was the minimal amount of
time allotted to this task. We would expect that with some additional ef-
fort, S02 could be quantitatively removed from these other sorbents.
WALDEN RESEARCH CORPORATION
-------
o
I
OJ
I
6
m
TABLE C-l
COLLECTION EFFICIENCY DATA FOR SOLID SORBENTS
Sorbent
Pb02
Ascarite
Amberlite XE309
Sampling rate = 0.
Sorbent tubes were
T = about 20°C
Length (cm) ,
15
10
5
15
10
5
20
15
10
5 liter/min for 15
made of 8 mm o.d.
Efficiency
99
-99
98
99
99
98
98
92
91
minutes
glass tubing
No. of Runs
3
3
3
3
3
3
2
3
5
Inlet S02
Concentration
330
330
330
330
330
330
1100
1100
1100
Site
Laboratory
Laboratory
Laboratory
Laboratory
Laboratory
Laboratory
Sulfuric Acid
Sulfuric Acid
Sulfuric Acid
Plant
Plant
Plant
TO
a
o
o
•»
s
TO
O
-------
TABLE.. C-2
RECOVERY FROM SORBENTS
Sorbent
S09 Recovery (%)
Procedure
Amberlite XE309
10
20
Ascarite
105
(a) .hot Ng then absorption in
. H2°2
(b) aqueous leaching
Boiling with Na?CO~ followed by
acidification, evaporation and
Ba Chlor analysis.
Leached with aqueous H?0?,
neutralized with HC1, then analy-
sis by Ba Chlor.
C-4
WALDEN RESEARCH CORPORATION
-------
These tests were only preliminary but the collection of various gases
on solid sorbents appears promising. More work is needed on the analytical
methods and quantitative desorption of these species from the sorbents.
C-5 WALDEN RESEARCH CORPORATION
-------
APPENDIX D
PRECISION OF BARIUM ION TITRATION AND BARIUM
CHLORANILATE FOR VARIOUS ANALYSTS
D-l WALDEN RESEARCH CORPORATION
-------
The precision of the Ba titration for six different analysts is
given in Table D-l. The precision for two analysts (A and C) for foundry
effluents is 3.1 ^0.3% for S02 analysis from foundry effluents. Similarly.
A and B have CV's of 1.5 and 1.95%, respectively, for copper smelters. A
and F have CV's of 1.5 and 2.0%, respectively, for the uncontrolled coal-
fired power plant. Note, however, that B and E have considerably differ-
ent CV's for the sulfuric acid plant. The CV's of 1.6 and 2.4 by analysts
B and D are well within the precision expected at 1000 ppm while the CV of
3.4 by operator E is very poor compared to any other operators or sources
at that concentration of SO-. This analyst appears to have a major prob-
lem with analytical technique, not just the barium ion titration since the
same operator (operator C, Table D-2) has a very high CV on the sulfuric
plant with the barium chloranilate where the judgment error associated with
the titration is eliminated.
The precision of the other five operators falls between 1.5 and 2.8%
regardless of source and/or concentration. The pooled CV (weighted for
samp
D-l.
sample size) for the Ba titration was 2.1% for the 693 samples in Table
The precision of the barium chloranilate as a function of analyst is
given in Table D-2. Here we don't have as many comparisons on the same
source as we did with the barium titration. For an iron and steel plant,
analysts A and B get the same precision although the SOp levels are dif-
ferent. Analyst C's results (CV's) on the sulfuric acid plant are con-
siderably higher than those obtained by analyst B. This is apparently a
problem with analytical technique as mentioned above. The CV for the
barium chloranilate method range from 1.7 to 3.5% when no interferences
are present or the concentration is greater than 100 ppm. The pooled CV
weighted for sample size for 179 samples in Table D-2 is 2.7%. The
values below 100 ppm were dropped along with the data from the iron and
steel plant. This value of 2.7 is very close to the average values ob-
tained at all the power plant sites by the barium chloranilate method.
The data in this section do provide some limited information on the
precision of both methods as a function of operators.
D-2
WALDEN RESEARCH CORPORATION
-------
PORA"
o
TABLE D-l
COMPARISON OF PRECISION OF Ba++ TITRATION FOR S02 BY DIFFERENT ANALYSTS
Analyst Site
A Foundry
Cu Smelter
Uncontrolled Coal -Fired Power Plant
B Cu Smelter
Cu Smelter
Iron and Steel Plant
Iron and Stee.l Plant
o Pb Smelter
w Sulfuric Acid Plant
C Foundry
D Sulfuric Acid Plant
Pilot Plant (HC1 doping)
E Sulfuric Acid Plant
F Controlled Coal -Fired Power Plant
I Oil -Fired Power Plant
g Uncontrolled Coal -Fired Power Plant
TO
ni
c/> ^
> With fine parti cul ate filter in sampling train
o
-T
No. of Samples
24
30
11
42
24
52
32
48
78
28
198
12
56
10
22
26
mean ppm S0?
62
3181
432
3234
3538
256
214
3678
1109
64
1170
607
862
727
393
406
CV (35)
2.8*
1.5*
1.5
2.0
1.9*
1.2
1.5*
1.6*
1.6
3.4*
2.4
2.0
3.2
2.2
2.3
2.0
-------
TABLE D-2
COMPARISON OF PRECISION OF Ba CHLORANILATE METHOD FOR S02 BY DIFFERENT ANALYSTS
3
i
S
>
r-
O
m
z
3)
m
in
m
>
7i
o
X
o
0
m
TJ
O
73
>
Analyst Site
A Iron and Steel Plant
Iron and Steel Plant
B Iron and Steel Plant
Sulfuric Acid Plant
Cu Smelter
Cu Smelter
C Sulfuric Acid Plant
D Pilot Plant (HC1 doping)
E Controlled Coal -Fired Power Plant
Oil -Fired Power Plant
Uncontrolled Coal -Fired Power Plant
*
With fine parti cul ate filter in sampling train
No. of Samples
72
150
32
56
16
42
226
17
,5
24
19
mean ppm S0?
37
39
223
1119
3434
3777
1124
220
701
406
460
CV (%)
4.3*
3.7
4.6*
2.3
1.7*
2.3
5.6
3.8
4.3
3.4
3.1
------- |