5153
Babcock&Wilcox
MAGNESIA BASE
WET SCRUBBING OF
PULVERIZED COAL
GENERATED FLUE GAS -
PILOT DEMONSTRATION
RESEARCH AND DEVELOPMENT DIVISION
RESEARCH CENTER ALLIANCE, OHIO
-------
MAGNESIA BASE WET SCRUBBING OF PULVERIZED COAL
GENERATED FLUE GAS - PILOT DEMONSTRATION
PROJECT SPONSORED BY NATIONAL AIR POLLUTION CONTROL ADMINISTRATION
ORDER 4152-01
RESEARCH CENTER REPORT 5153
BY: W. DOWNS
A. J. KUBASCO
SEPTEMBER 28, 1970
THE BABCOCK & WILCOX COMPANY
RESEARCH AND DEVELOPMENT DIVISION
RESEARCH CENTER
ALLIANCE, OHIO
COPY NO
.93
-------
THE BABCOCK § WILCOX COMPANY
RESEARCH AND DEVELOPMENT DIVISION
RESEARCH CENTER
ALLIANCE, OHIO
MAGNESIA BASE WET SCRUBBING OF PULVERIZED COAL
GENERATED FLUE GAS - PILOT DEMONSTRATION
By: W. Downs
A. J. Kubasco
ABSTRACT
Purpose
The purpose of this project was to demonstrate the adequacy of magnesia
base wet scrubbing to remove essentially all of the particulates (99+%) and
most of the sulfur dioxide (97+%) from pulverized coal fired furnaces. Side
effects such as sulfate formation, NC^ absorption, and scrubber operability
were to be determined.
Summary
A wet scrubbing pilot plant consisting of three scrubbers was designed
and constructed. An existing test furnace was modified to burn pulverized
coal at a rate of 500 pounds per hour. These three scrubbers consisted of a
venturi-type particulate scrubber, a venturi-type absorber, and a tray-type
absorber (floating bed absorber). Over 100 short-term tests were performed
to determine the most satisfactory operating conditions for each scrubber.
Following this, several extended tests were performed.
Results
Sulfur dioxide absorption in the floating bed absorber was a maximum at
about 99.21 at 6.5 inches w.g. The venturi absorber performed at 95.5% for
the same pressure drop. S02 absorption in both absorbers was sensitive to
slurry composition including pH and sulfate concentration. For an average
7 inches w.g. pressure drop, the particulate .scrubber collected 99.5+%, the
venturi absorber 99.11, and the floating bed absorber 97.7% of the fly ash.
Sulfate formation averaged between 4 and 8% of the absorbed S02 in both ab-
sorbers. Deposition of MgS03«6H20 in the piping was prevalent under some
operating conditions but did not form when fly ash was present in the ab-
sorber slurry.
Conclusions
The magnesia base wet scrubbing process has been demonstrated to be
superior to other wet processes known to be under development for abatement of
both particulates and S02 from coal-fired furnaces. The process as demon-
strated is compatible with the B§W regeneration concept.
Research Center Report 5153
Order 4152-01
September 28, 1970
Magnesia Base Wet Scrubbing
Project Sponsored by National Air
Pollution Control Administration
-------
ABSTRACT (Cont'd)
Recommendat ions
A regeneration facility should be added to the present pilot plant to
demonstrate its capability to regenerate magnesia for reuse and to produce
a usable sulfur product.
11
-------
SUMMARY
The magnesia base wet scrubbing and regeneration process offers a
potential solution to the problem of sulfur and particulate pollution from
large stationary sources. This process is illustrated in simplest form as:
Coal-
Concentrated
Boiler
— so2 *
Sulfu
Acid
Eleme
Snlfu
/" N
Wet
Scrubbers
V J
*
ric
or
JlLdl ^
r
MgSOs
(Solid) ^
P
MgO
Regeneration
Process
1
Product
Plant
^ J
aU2/n
As a first step in the evaluation of this process, the National Air
Pollution Control Administration has sponsored the subject project for a demon-
stration of the wet scrubbing of pulverized coal generated flue gas with a
magnesia base slurry. Regeneration of the products was not included. This
work sponsored under contract CPA 22-69-162 called for the fabrication of a wet
scrubbing pilot plant behind an existing pilot-scale fossil fuel burning facility.
The furnace, which was designed for burning gas, oil, and special waste;liquors,
was converted to accommodate pulverized coal. This coal was prepared and stored
in existing facilities. The furnace burns approximately 500 pounds of coal per
hour, which is roughly the firing rate of a 1/2 megawatt (electrical) power
boiler. :
The wet scrubbing pilot plant handled all of the flue gas from the furnace
(5400 pounds/hour). This plant included a venturi-type fly ash scrubber, two
gas absorbers (a venturi type and a tray type), chemical makeup and disposal
systems, and auxiliary piping and pumps.
The test specifications called for an investigation of the following
parameters:
111
-------
SO- absorption as a function of absorber type,
chemical composition of the spray slurry,
chemical composition of the makeup slurry,
sump residence time, and sulfate level.
Sulfate formation rate as a function of furnace excess
air, scrubber arrangement, and combustion air
temperature.
Particulate collection in all three scrubbers.
NO absorption in all three scrubbers under normal
™'" ^X
and special conditions.
SO, formation as a function of furnace operation.
In addition to the above parameters, deposition and the means for its
prevention or reduction became an integral part of this program.
The test program was divided into two phases: parameter studies and
extended tests for as long as 4 days. Over 200 tests, of which about 1/3
were short-term parameter studies, were performed. The results demonstrated
that this wet scrubbing process is capable of removing 99+% of both particu-
lates and sulfur dioxide from the flue gas. The maximum sulfation rate was
shown to be less than 8% at 151 excess air, and it was also shown that de-
position can be prevented by proper design and operation.
SO- absorption in the floating bed absorber (FBA) was found to be superior
to that in the venturi absorber for comparable energy levels. In the FBA only,
SO- absorption increased substantially with increasing liquid flow rate and in-
creasing pH up to about 7.0. At lower pH's, and at high sulfate concentrations
in the absorbing slurry, the SO- absorption was adversely affected in both
£*
absorbers. The presence of fly ash in the absorbing slurry had no adverse effects
upon SO- absorption.
Sulfate formation was found to be a function of furnace excess air. The
presence of fly ash in the absorbing slurry contributed substantially to the
sulfate formation rate. The floating bed absorber and venturi absorber were
not substantially different in their sulfate formation rates.
Deposition of magnesium sulfite was a recurring problem. This deposition
was believed to result from the neutralization reaction between bisulfite and
magnesia. Temperature cycling of the scrubbing slurry was also found to contribute
IV
-------
to this deposition. The deposition was prevented by fly ash in the slurry.
Nonwettable materials also reduced the tendency to receive deposits.
Deposition at the wet-dry line in the gas-liquid contacting zone was
not a problem in the particulate scrubber under any condition and was not
a problem in the gas absorbers when the gas was first attemperated by the
particulate scrubber. However, deposition at that point was severe when
the gas absorbers were in direct contact with the hot furnace gases. Evi-
dently, deposition occurs only if the spray slurry contains solubles which
plate out upon evaporation.
Particulate collection of furnace exit flue gas was excellent in all
three scrubbers. Amongst these, the particulate scrubber was best, and the
floating bed absorber was least effective. Neither absorber, however, collected
a measurable quantity of fly ash leaving the particulate scrubber. Evidently,
this fly ash was much too fine to be collected by the relatively low energy
absorbers.
Absorption of oxides of nitrogen was negligible across all three scrubbers
under all operating conditions tested. Injection Of NCL into the gas stream
was designed to promote total NO absorption. However, injection did not
yC
increase NO absorption. In fact, it significantly increased the sulfate
j\.
formation rate.
With proper design and operation, this wet scrubbing process will remove
both dust and sulfur gases more effectively than most systems presently
proposed.
v
-------
TABLE OF CONTENTS
Page
1.0 INTRODUCTION 1-1
2.0 PILOT PLANT EQUIPMENT 2-1
2.1 Pulverizer 2-2
2.2 Coal Separation Apparatus and Bunker -- - 2-2
2.3 Coal Feed System - - 2-3
2.4 Burner and Furnace - 2-4
2.5 Fuel - 2-7
2.6 Wet Scrubbing Apparatus - - -- 2-8
2.6.1 Particulate Scrubbing System --- --- 2-8
2.6.2 Venturi Absorber System — 2-11
2.6.3 Floating Bed Absorber 2-12
2.7 Flue Gas Exhaust Section - --- --- 2-14
2.8 Pilot Plant Monitoring System -- 2-14
2.9 MgO Makeup and Product Disposal 2-14
2.10 Chemical Analysis Lab - - 2-16
2.11 Overall Pilot Plant Schematic 2-16
3.0 TEST APPARATUS AND PROCEDURES -- - - - -- 3-1
3.1 Test Procedure for 1-Day Operation - 3-1
3.2 Test Procedure for Extended Operation -- 3-2
3.3 Special Tests 3-2
3.3.1 High Sulfate Concentration Tests 3-2
3.3.2 NOx Absorption Tests - - --- 3-2
3.4 Sulfur Dioxide Sampling and Analysis - — 3-3
3.4.1 Barton Titrator - --- - 3-3
3.4.2 Reich Method for S02 Analysis --- - 3-5
3.5 Sulfur Trioxide Sampling - - 3-6
3.6 NOx Sampling and Analysis .— 3-7
3.7 Dust Loading -- - 3-8
3.8 Cascade Impactor -- -- 3-10
3.9 Orsat Analyzer - 3-11
3.10 Magnesium Sulfite-Bisulfite Analysis - 3-11
3.11 Sulfate Analysis 3-12
3.11.1 Titration Method -- 3-12
3.11.2 Sulfate Sample for Chemistry Lab Analysis 3-13
3.12 Particulate Solids Sampling -- 3-13
3.13 Magnesium Sulfite Solids Sampling 3-14
3.14 MgO Makeup Slurry Titration -- 3-14
3.15 Other Measurements -- - - 3-15
VI1
-------
TABLE OF CONTENTS CQNT'D
Page
4.0 RESULTS ------------------ - ----- ..... ----------------------- 4-1
4.1 SC>2 Absorption - Particulate Scrubber ---------------- 4-1
4.2 S02 Absorption - Floating Bed Absorber -------- ....... 4-1
4.2.1 Parameter Study -------- ........ -- ...... ------ 4-1
4.2.2 Floating Bed Absorber - Absorption Correlation- 4-2
4.3 SO? Absorption - Venturi Absorber ....... ------------- 4-4.
4.3.1 Parameter Study -------------- ..... ----- ...... 4-4
4.3.2 Venturi Absorber Absorption Correlations ----- 4-5
4.4 Crystalline Deposition ........ - ....... --------------- 4-7
4.4.1 Parameters Affecting Deposition - — ..... ----- 4-8
4.5 Deposition at Wet-Dry Lines ......... ----------------- 4-10
4.6 Sulfate Formation ------------------------------------ 4-11
4.6.1 Floating Bed Absorber ------ ...... - ..... ------ 4-11
4.6.2 Venturi Absorber ------------------------------ 4-12
4.7 Particulate Collection ----- ..... --------------------- 4-13
4.7.1 Particulate Venturi Scrubber ----------------- 4-13
4.7.2 Floating Bed Absorber With Prescrubber - ..... - 4-14
4.7.3 Floating Bed Absorber Without Prescrubber ---- 4-14
4.7.4 Venturi Absorber With Prescrubber ------------ 4-15
4.7.5 Venturi Absorber Without Prescrubber --------- 4-16
4.8 NOx Absorption ---------------------------------- ..... 4-16
4.9 NOX Injection Test ----------------------------------- 4-17
4.10 503 Formation ..... -------------- ..... -- ..... --------- 4-17
4.11 The Floating Bed Absorber Tray Hydraulics ------------ 4-18
5.0 DISCUSSION OF RESULTS -------------------------------------- 5-1
5.1 Data Analysis ---------------------------------------- 5-1
5.2 Error Analysis --------------------------------------- 5-1
5.3 S02 Absorption ----- ----------------- ..... ------------ 5-1
5.3.1 Comparison Between Absorbers ----------------- 5-1
5.3.2 Mathematical Models of SO? Absorption ....... - 5-2
5.3.3 S02 Vapor Pressure from Literature ----------- 5-6
5.3.4 Effect of Sulfates Upon S02 Absorption -- ..... 5-6
5.3.5 Comparison of Absorber Performance with
Literature ------------------- ....... --------- 5-7
5.4 Chemistry of Absorbing Slurry -- ....... - ........ ------ 5-8
5.5 Deposition ..... ------- ...... -- ..... ------------ ...... 5-11
5.5.1 Deposition Mechanisms ........ ----------- ..... 5-11
5.5.2 Method of Deposit Elimination ------ ..... ----- 5-15
5.6 Sulfate Formation ---- ........ ------------- ........... 5-15
5.7 NOx Absorption -------------------- ......... - ..... ---- 5-18
Vlll
-------
TABLE OF CONTENTS CONT'D
Page
6.0 CONCLUSIONS 6-1
6.1 S02 Absorption 6-1
6.2 Sulfate Formation 6-1
6.3 Deposition Within Absorber 6-2
6.4 Others --- 6-2
7.0 RECOMMENDATIONS - 7-1
7.1 Participate Scrubber Operation and Design 7-1
7.2 S02 Absorber Operation and Design 7-1
7.3 Future Work 7-1
8.0 ACKNOWLEDGEMENTS 8-1
REFERENCES R-l
APPENDIX A, DETAILS OF ANALYSIS TECHNIQUES A-l
A.I Sulfur Dioxide Sampling A-l
A.2 SO, Sampling Apparatus § Procedure A-5
A. 3 Dust Loading Apparatus and Procedure A-10
A.4 Modified Palmrose Analysis A-12
A. 5 Determination of Sulfate Ion in Mg-SO^H-O System A-15
A.6 Particulate Solids Apparatus and Procedure A-19
A.7 MgO Makeup Slurry Titration Procedure A-19
A.8 Magnesia Hydration Analysis A-20
APPENDIX B, CORRELATION COEFFICIENTS B-l
APPENDIX C, ERROR ANALYSIS C-l
C.I General Technique C-l
C.2 Gas Sampling Errors C-2
C.3 Chemical Analysis Errors - Liquid C-ll
C.4 Flow Rate Error Analysis C-14
C.5 Material Balances C-19
List of Tables
Table
2.1 Pulverized Coal and Gas Analysis 2-8
4.1 MgO Conversion to Mg(OH)2 4"4
4.2 X-Ray Diffraction Analysis of Crystalline Deposits 4-8
4.3 Degree of Deposition at Various Locations During Periods
of Rapid Formation - -- 4-10
IX
-------
TABLE OF CONTBffS OONT'D
List of Tables Cont'd
4.4 Average Sulfate Formation in FBA for Individual
Test Series 4-12
4.5 Average Sulfate Formation in Venturi Absorber for
Individual Test Series 4-13
4.6 Particulate Collection Performance of Particulate Scrubber 4-15
4.7 Particulate Collection Across FBA With Prescrubber 4-16
4.8 Particulate Collection Across FBA Without Prescrubber 4-16
4.9 Particulate Collection Across Venturi Absorber With
Prescrubber 4-17
4.10 Particulate Collection in Venturi Absorber (Without
Prescrubber) 4-17
4.11 Average NOX Concentrations Throughout Pilot Plant 4-18
4.12 Summary of NOX Results 4-18
4.13 N02 Injection Test Results 4-18
4.14 Sulfur Trioxide Concentration at Furnace Exit 4-19
4.15 Effect of Bed Height on Dry Pressure Drop Across FBA 4-19
4.16 FBA Packing Attrition After 270 Hours 4-19
List of Figures
Figure
2.1 Magnesia Base Wet Scrubbing Pilot Plant 2-1
2.2 Coal Pulverizer 2-2
2.3 Coal Handling System 2-2
2.4 Coal Storage Facilities - 2-3
2.5 Pulverized Coal Sizing Plot --- 2-3
2.6 Coal Screw Feeder 2-4
2.7 Coal Screw Feeder 2-4
2.8 Pulverized Coal Burner 2-5
2.9 Pulverized Coal Burner 2-5
2.10 Basic Combustion Test Furnace 2-6
2.11 Furnace Sectional View 2-6
2.12 Furnace Relief Valve - Open Position 2-6
2.13 Furnace Relief Valve - Closed Position 2-6
2.14 Steam Lances 2-7
2.15 Furnace Control Panel 2-7
-------
TABLE OF CONTENTS COND'D
List of Figures Cont'd
2.16 Particulate Scrubbing System -- 2-8
2.17 Particulate Scrubbing System 2-9
2.18 Particulate System Dimensions 2-9
2.19 Product Discharge Nozzle - 2-10
2.20 Venturi Absorber Flow Control System 2-10
2.21 Venturi Absorber System - 2-11
2.22 Venturi Absorber System Dimensions 2-11
2.23 Floating Bed Absorber 2-12
2.24 Floating Bed Absorber 2-12
2.25 Plastic "Wiffle" Balls Deformed 2-13
2.26 Floating Bed Absorber Dimensions 2-13
2.27 Top of Floating Bed Absorber 2-14
2.28 Moisture Disengagement Region - Sectional View 2-14
2.29 Measurement Locations 2-15
2.30 Makeup System 2-15
2.31 Product Disposal Leach Beds - 2-16
2.32 Pilot Plant Analysis Lab 2-16
2.33 Schematic Arrangement of Test Facilities -- - 2-17
3.1 N02 Injection Apparatus 3'3
3.2 S02 Sampling System -- - 3'3
3.3 Barton Titrators 3'4
3.4 Barton Electrical Circuit 3~5
3.5 Reich Board " "" 3"5
3.6 SCL Collection Apparatus 3'6
3.7 SCL Collection Apparatus - 3"6
3.8 Particulate Sampling Nozzle 3'8
3.9 Particulate Sampling Apparatus 3"8
3.10 Particulate Sampling Apparatus " 3"9
3.11 Thorston Dust Sampler - 3"9
3.12 Cascade Impactor - - 3'10
3.13 Impactor Stage -- - 3"10
3.14 Orsat Apparatus 3"11
3.15 Syringe Filter Holder - 3"12
3.16 Syringe in Septum 3'12
XI
-------
TABLE OF CONTENTS CONT'D
List of Figures Cont'd
Page
3.17 Palmrose Titration Apparatus 3-13
3.18 Sulfate Sample Preparation Apparatus -- 3-13
3.19 Sulfate Titration Apparatus 3-14
3-20 MgO Makeup Titration Curve 3-14
4.1 SO- Absorption - FBA - 4-2
L*
.4.2 S02 Absorption - FBA 4-2
4.3 Correlation for Equation 4-1, SO- Absorption - FBA 4-3
4.4 Pressure Drop - FBA 4-3
4.5 S02 Absorption - FBA Without Prescrubber 4-4
4.6 Comparison of Effect of High Sulfates Upon SO-
Absorption - FBA 4-4
4.7 S02 Absorption - Venturi Absorber --- -- - 4-5
4.8 S02 Absorption - Venturi Absorber - 4-6
4.9 Pressure Drop Effect Upon SO- Absorption - Venturi Absorber 4-6
4.10 S02 Absorption Correlation for Venturi Absorber 4-7
4.11 Comparison of Effect of High Sulfates Upon S02
Absorption - Venturi Absorber 4-7
4.12 The Effect of Fly Ash Upon Deposits Formation 4-8
4.13 Participate Scrubber - Inside View 4-10
4.14 Sulfate Formation - FBA With Prescrubber 4-11
4.15 Sulfate Formation - FBA Without Prescrubber 4-11
4.16 Sulfate Formation - Venturi Absorber With Prescrubber 4-12
4.17 Sulfate Formation - Venturi Absorber Without Prescrubber -- 4-13
4.18 Dust Collection Performance Particulate Scrubber 4-14
4.19 Dust Collection Performance Particulate Scrubber 4-14
4.20 Pressure Drop - Particulate Scrubber 4-15
4.21 Dust Collection Performance - FBA 4-16
4.22 Dust Collection Performance - Venturi Absorber 4-17
5.1 Computer Results 5-1
5.2 Comparison of SO- Absorption in the Venturi Absorber
With Literature 5-8
5.3 Solubility of Magnesium Sulfite - -- 5-10
5.4 Relationship Between Bisulfite Concentration and Slurry pH 5-10
5.5 Example of Coarse Deposit 5-11
5.6 Equilibrium Constants for NO -SO -H,0 Systems 5-19
-X J\. L,
5.7 SO--NO Absorption Process 5-20
L, X
XI1
-------
TABLE OF CONTENTS OONT'D
List of Figures Cont'd
A.I Spinning Syringe Apparatus -
A.2 S03 Collector - A'5
C.I Barton S02 Analysis Error Potential - C-3
C.2 Comparison Between Barton and Reich Methods C-4
C.3 Velocity Profile Downstream of Flow Straightener -- C-ll
C.4 Maximum Probable Error in Palmrose Analysis C-12
C.5 Comparison of Slurry Strength by Two Techniques C-12
C.6 Comparison Between Gravimetric and Titrametric Sulfate Analyses -- C-13
C.7 Flue Gas Material Balance - C-17
C.8 Sulfur Material Balance C-20
Xlll
-------
NOMENCLATURE
Test Series
A-xxx
B-xxx
C-xxx
D-xxx
E-xxx
F-xxx
G-xxx
H-xxx
I-xxx
J-xxx
K-xxx
L-xxx
M-xxx
N-xxx
0-xxx
Purpose
Final Shakedown Tests
Parameter Study, Particulate Scrubber
and Venturi Absorber, Faulty Gas Analysis
Parameter Study, Particulate Scrubber
and Venturi Absorber
Parameter Study, Particulate Scrubber
and Floating Bed Absorber
Parameter Study, Particulate Scrubber
and Venturi Absorber
Extended Test, Particulate Scrubber
and Venturi Absorber
Parameter Study, Particulate Scrubber
and Venturi Absorber, Elevated Sulfates
Extended Test, Particulate Scrubber
and Venturi Absorber
Extended Test, Particulate Scrubber
and Venturi Absorber
Extended Test, Venturi Absorber Only
(Particulate Scrubber Bypassed)
N02clnjection Tests
Extended Test, Particulate Scrubber
and Floating Bed Absorber
Extended Test, Floating Bed Absorber
Only, (Particulate Scrubber Bypassed)
Parameter Study, Particulate Scrubber
and Floating Bed Absorber, Elevated
Sulfates
Parameter Study, Particulate Scrubber
and Floating Bed Absorber, Elevated
Sulfates
Dates
11/18/69-12/5/69
12/4/69-12/9/69
12/12/69-12/19/69
1/9/70-1/26/70
2/2/70-2/6/70
3/9/70-3/10/70
3/18/70
3/23/70-3/24/70
3/31/70-4/2/70
4/14/70-4/16/70
4/9/70
4/27/70-4/30/70
5/12/70-5/15/70
5/27/70
7/21/70
xiv
-------
1.0 INTRODUCTION
Abatement of pollution from large stationary sources such as power plants
is a primary goal of government, industry, and an aroused public. However,
the magnitude of this abatement task is reflected by the huge disposal problem
with which the power industry is confronted. For example, an 800 MW power plant
burning pulverized coal, which produces electricity for roughly one million
people, creates the following yearly emissions: 65 billion pounds of combustion
gases, of which about 15% can be considered alien to the atmosphere, 635 million
pounds of solid noncombustibles (fly ash), and 36,500 billion Btu of waste heat.
In a conventional boiler cycle, the carbon dioxide emissions and the waste heat
are not amenable to treatment. The fly ash disposal is presently accomplished
through landfill. The major emissions which constitute an immediate hazard in-
clude the oxides of sulfur and the oxides of nitrogen. These emissions presently
are being discharged to the atmosphere with only minor treatment. From the
above 800 MW power plant sulfur dioxide is emitted at a rate of 300 million
pounds per year (based on 31 sulfur coal) and nitric oxide (assuming 1000 ppm
at the stack) amounts to 65 million pounds per year. To put the sulfur dioxide
emission in perspective, the rate of sulfur emission per customer serviced by
this power plant is roughly three times the per capita rate of sulfur mined in
the United States.
The major problem in removing sulfur dioxide from flue gases is not the
technical problem of sulfur dioxide absorption or adsorption itself but is how
to do this economically without adversely affecting the business of electric
power production. Handling the volumes of flue gases required by standard
chemical engineering unit operations such as fixed bed reactors or spray ab-
sorption towers constitutes a completely new engineering problem. Thus, a
primary task of the Division of Process Control Engineering of the National Air
Pollution Control Administration, under which this contract was sponsored, has
been to investigate a broad range of research and development processes poten-
tially amenable to this problem.
Wet scrubbing of flue gases by various aqueous absorbents capable of
reacting with SO- has been one of the most vigorously investigated approaches.
The obvious technical advantage of wet scrubbing is the well-established unit
operation of particulate collection and gas absorption by aqueous sprays.
-------
The most widely applied wet process has been the calcium base or limestone
process. This is technically simple and cheapest to install among potential
wet processes. However, this nonregenerative process has two liabilities
that make other wet processes such as magnesia more attractive. First, no sul-
fur products can be produced to defray the cost of the process, and secondly,
the solid disposal load is more than doubled.
The B§W Magnesia Base Wet Scrubbing and Regeneration Process has the
potential of avoiding both of the above liabilities. Obviously, in order to
justify a regeneration plant, the production cost of the sulfur product must
be less than the saleable price of that product. Secondly, the regenerated
magnesia must be reusable because it is expensive compared to limestone and
its disposal would constitute a pollution problem itself. Without a regener-
ation process, however, magnesia has no economic validity in spite of any
potential technical advantages from a wet scrubbing standpoint.
The object of this contract was to demonstrate the capability of only
the wet scrubbing portion of the magnesia base system on pulverized coal
generated flue gas. The technical program was based upon no specific regener-
ation process. Potential interactions between the wet scrubbing and regeneration
steps were not explicitly investigated. The major technical areas studied were:
S02 Absorption
Sulfate Formation
Deposition
Particulate Collection
All of these except deposition were in the original plan. Solution of the
deposition problem was, however, an absolute necessity.
The following pages report the results of over 200 tests under this
contract, CPA-22-69-162.
1-2
-------
2.0 PILOT PLANT EQUIPMENT
Tests were carried out on a pilot plant using a 1/2-megawatt (electrical)
pulverized coal-burning furnace to generate flue gas. A water tube section
was used to cool the gases to a temperature typical of a boiler stack gases.
The coal handling facilities included a pulverizer, dilute phase trans-
port system, and a coal storage bunker. The pulverized coal-burning furnace
was part of an existing combustion test as were the coal handling facilities.
The gas cleanup equipment consisted of a particulate scrubber venturi, a venturi
absorber, and a floating bed absorber; the latter two absorbers were used for
sulfur dioxide removal. This section covers a general description of the pilot
plant equipment and related facilities. A pictorial drawing of the pilot plant
showing equipment layout is shown in Figure 2.1.
FIGURE 2.1 MAGNESIA BASE WET SCRUBBING PILOT PLANT
2-1
-------
2.1 PULVERIZER
FIGURE 2.2 COAL PULVERIZER
The coal pulverizer, Figure 2.2, is
a B§W B 200 Series with a maximum capacity
of about 10 tons/hr. This is a double row
ball-and-race pulverizer where an inter-
mediate revolving grinding ring is interposed
with two sets of balls between the races of
two stationary rings. Raw coal is fed in-
side the grinding ring. As it passes
through the top row of balls, the coal is
partially pulverized. The coarse par-
ticles fall behind the lower row of balls
and are further pulverized. A primary
air blower supplies heated air, which en-
trains, dries, and carries the coal over
to the separation and storage facility.
The coal is transported by dilute phase
with 1 pound of coal requiring about
1-1/2 pounds of air.
2.2 COAL SEPARATION APPARATUS AND BUNKER
The pulverized coal leaving the
pulverizer travels about 200 feet to
a cyclone, which separates the con-
veying air from the coal; see Figure 2.3.
Some of the very fine coal remains en-
trained in the air after leaving the
cyclone. This coal is scrubbed out
by a venturi after which the air is vented
to the atmosphere. Thus some of the coal
fines is lost by this separation process.
The separated coal leaves the bottom of
the cyclone and passes into the coal bunker;
see Figure 2.4. The bunker has a capacity
for 30 tons of pulverized coal.
FIGURE 2.3 COAL HANDLING
SYSTEM
2-2
-------
FIGURE 2.4 COAL STORAGE
FACILITIES
Since a part of the coal fines
is lost in the coal-air separation
process, a coal sizing analysis was
performed on a sample from the bunker
to determine whether this coal
was typical of pulverized coals.
Figure 2.5 shows a plot of the
coal sizing analysis. Typical
pulverized coal passes about 70%
through a #200 U. S. standard
sieve. This plot showed that the
coal used in this test, designated
6§W, was similar to the "typical"
pulverized coal from the Ohio
Edison Edgewater Station.
FIGURE 2.5
PULVERIZED COAL
SIZING PLOT
lf\
u
400 32$ 270 200 140 100 80 70 65 SO
U. S. Standard Sieve Designation
2.5 COAL FEED SYSTEM
The pulverized coal feed
system, Figures 2.6 and 2.7,
uses a screw feeder and conveying
air. The screw transports the
coal from the bottom of the hopper
to the feed pipe, where it is
partially fluidized with com-
pressed air. The coal falls a
few inches and is entrained by
the primary air, which carries
it to the burner. Approximately
500 Ib/hr of primary air, measured
by an orifice meter, is used to
transport about 500 Ib/hr of coal.
2-3
-------
2.4 BURNER AND FURNACE
The pulverized coal burner is a
B§W cell-type circular burner with a
natural gas lighter. This lighter is
operated continuously during coal firing
to help maintain stable ignition. A
sectional view of the burner is shown in
Figure 2.8 and a photograph in Figure 2.9.
The furnace, Figure 2.10, consists
of a horizontal cylinder 8 feet in
length by 4-1/2 feet in diameter. This
forms the actual combustion chamber. The
walls of the furnace are formed by a
water jacket, which dissipated heat by
the production of nonpressurized steam.
See Figure 2.11. The inside walls were
originally covered with refractory. How-
ever, prior to running extended tests,
the refractory was removed from all walls
except the front. This was necessary
to prevent excessive slagging of the tube
banks. The furnace continually grows
hotter as it operates because of slag
buildup on the walls, which act as a
refractory. When extended tests were
attempted with refractory in place,
molten slag froze onto the cold tube
banks, thereby causing furnace plug-
gage. Removing the refractory per-
mitted the furnace to run cooler, thus
allowing the slag to solidify to an
ash before contacting the tubes.
FIGURE 2.6 COAL SCREW FEEDER
FIGURE 2.7 COAL SCREW FEEDER
2-4
-------
FIGURE 2.8 PULVERIZED COAL BURNER
FIGURE 2.9 PULVERIZED COAL BURNER
To maintain stable ignition,
the pulverized coal must be fired
with preheated air. This air is
supplied by a forced draft fan and
preheated by two direct fired air
heaters capable of heating 10,000
Ib/hr air to a temperature of 1000 F.
The heat release rate of the fur-
nace averages about six million
Btu/hr.
Combustion gases pass from
the furnace proper through three
tube banks that cool the flue gas
to approximately 450 F. All tubes
are 1-1/2-inch O.D. on 2-1/2-inch
centers with exception of the first
six rows of tubes which are on 5-inch
I
centers. Behind each tube bank is a
duct permitting flue gas recirculation
to the burner. This feature, how-
ever, was not used during the tests.
Flue gas leaving the tube bank passes
through a transition piece to a
15-inch-diameter vertical stack.
Gas flow to the scrubbing system
is taken from the side of this
stack. The vertical stack ends at
a relief valve. The relief valve is
pneumatically operated and is
automatically activated during a
test if furnace pressure becomes
too great. The relief valve can
also be operated manually from the
control panel and is used during
startup and shutdown operations;
see Figures 2.12 and 2.13.
2-5
-------
FIGURE 2.10 BASIC COMBUSTION
TEST FURNACE
FIGURE 2.12 FURNACE RELIEF VALVE
OPEN POSITION
FIGURE 2.11 FURNACE SECTIONAL VIEW
FIGURE 2.13 FURNACE RELIEF VALVE
CLOSED POSITION
2-6
-------
FIGURE 2.14 STEAM LANCES
FIGURE 2.15 FURNACE CONTROL
PANEL
To prevent buildup of slag and
ash deposits in the tube banks, three
steam lances were installed; see
Figure 2.14. During extended tests
the tubes were lanced about every 6
hours. This succeeded in keeping the
tubes free of ash buildup.
The furnace control panel, shown
in Figure 2.15, is fully equipped to
monitor and control the furnace pressures.
A 12-point Speedomax recorder provides
a continuous check of the temperatures
within the furnace and ancillary com-
ponents. Oxygen concentration of the
flue gas is continuously monitored by
a Bailey 02 Analyzer Model A57. From
time to time the Bailey Analyzer was
checked by Orsat analysis and also by
a Bailey electrochemical 0~ analyzer.
Located on top of the furnace,
Figure 2.11 is the steam drum, a steel
cylinder 4-ft. diameter by 6-ft. long.
This drum supplies water to the furnace
water jacket and acts as a steam-water
separator for venting the steam to
atmosphere.
2.5 FUEL
Two types of fuel were used
during the tests, pulverized coal
and natural gas. Pulverized coal was
2-7
-------
the main fuel while natural gas was used
in the lighter and accounted for about
4% (thermal) of the fuel used. A
number of coal analyses were run and
the results reported in Table 2.1.
A typical natural gas analysis obtained
from the Ohio Fuel Company is also
shown in Table 2.1.
2.6 WET SCRUBBING APPARATUS
2.6.1 Particulate Scrubbing System
Flue gas leaving the furnace passed
through the water tube section to the
particulate venturi and cyclone, see
Figures 2.16 and 2.17. The ducts were
arranged in such a way that by changing
blanks the particulate scrubber could
be bypassed. As the gas entered the
venturi throat, its velocity was greatly
increased. It is here that the fly ash
slurry spray was introduced, just
slightly ahead of the throat. The fly
ash particles traveling at high velocity
impacted upon the slower moving slurry
droplets and were thereby trapped.
The fly ash slurry was separated
from the cleansed flue gas in the cyclone
separator; see Figure 2.16. The flue
gas and slurry entered the cyclone tan-
gentially, spinning the slurry to the
walls while the gases moved toward the
center and out the top. The slurry
moved from the cyclone into the sump
located immediately below.
TABLE 2.1 PULVERIZED COAL AND
GAS ANALYSIS
PULVLRIIED CCW, ANALYSES
Ash (Dry) \
Sulfur (Dry) \
Oirbon (Ult.H
Lab. Serial No.
Saaylc Description
Total Moisture, \
Ash. \
Sulphur, \
Bto per Ib. (Dry)
Btu per Ib. (M&A Free)
Coal fron
BCni Hopper
1-9-70
Co»l from
BCTU Hopper
S-1Q-70
JQ00 hn.
CMJ4J7
• |]
NATURAL GAS ANALYSES
Sulphi
Hydrogen Sulfide, gr/100 cf 0.017
H,S Sulphur Kquiv., gr/100 cf 0.01
Msrcaptans - S - J^uiv.
Sulfide SuifJiur, gr/100 cf
Residual Sulphur, gr/100 c
Total Sulphur, gr/100 cf
. Date of Sample
0.007 , Components:
n.007 Nitrogen
0.004 Carbon Dioxiite
U.034 Methane
Ethane
Propane
Iso-Butme
N-Butane
Iso-Pentaiw
Guernsey
2/12/69
Mol t
FIGURE 2.16 PARTICULATE
SCRUBBING SYSTEM
2-8
-------
FIGURE 2.17 PARTICULATE SCRUBBING
SYSTEM
FIGURE 2.18 PARTICULATE SYSTEM
DIMENSIONS
E
i i
i
"r^i
?7'
A'
1 1
\,/
u
:<" l.D.
X
43
.
,,
-
i :"
i .:••
The slurry was pumped from the
sump and recirculated back to the
venturi spray nozzle. To maintain
the desired composition, part of
the slurry was discarded. Fresh water
previously treated by a zeolite bed
was added directly to the sump. Approxi-
mately 23 gal/min of slurry were re-
circulated through the spray nozzle, and
about 5 gal/min were discarded as par-
ticulate product. The overall diminsion
of the venturi and cyclone is shown
in Figure 2.18.
During shakedown (November 1969)
of the venturi and cyclone an unexpected
development occurred. The pressure
drop across the cyclone was about 14
inches of water. This was 10 inches
more than expected. Observation of
the cyclone operating with the top re-
moved showed a large vortex action in-
dicated by excessive liquid spinning.
To break up the vortex a series of
baffles (vortex eliminators) was installed
on the bottom cone of the cyclone; see
Figure 2.16. This reduced the pressure
drop to a normal 4 inches of water.
The particulate venturi spray
nozzle flow was controlled by a pump
and pinch valve. A pneumatically con-
trolled pinch valve (Red Valve Co.) con-
trolled the amount of "short circuit"
recirculation through the pump and
valve loop. Closing the pinch valve
2-9
-------
forced more slurry through the spray
nozzle. This arrangement was used
instead of the normal control method
via a gate or glove valve, because
solids tend to collect behind the
seat, thereby restricting flow. The
product nozzle takeoff, shown in
Figure 2.19 is merely an orifice cali-
brated versus pressure.
At the gas outlet from the particu-
late cyclone, the vortex spin was eliminated
by a flow straightener. This cross-shaped
member 10 inches long by 10 inches in diam-
eter was placed in the cyclone gas exit duct.
A 5-inch slide valve located in the
particulate cyclone exit duct is part of
the dust sampling apparatus. With the valve
closed, particulate sampling probe bolts
to the slide valve. The valve is then
opened so that the probe can be inserted
into the gas stream. This procedure
eliminates sudden pressure upsets in
the furnace.
A gas sampling probe used for both
NO sampling and S09 sampling is positioned
.X L
4 inches downstream from the dust sampling
connection. The probe includes a 6-inch
long 1-3/4-inch pipe welded flush to the
duct wall and a I/8-inch glass tube
located concentrically in this pipe. The
pipe provides a "quiescent" zone in
which the gas flow to the glass probe
is relatively slow. This minimizes the
possibility of slurry carryover from
the cyclone entering into the glass tubing
FIGURE 2.19 PRODUCT DISCHARGE
NOZZLE
FIGURE 2.20 VENTURI ABSORBER
FLOW CONTROL SYSTEM
2-10
-------
FIGURE 2.21 VENTURI ABSORBER
SYSTEM
FIGURE 2.22 VENTURI ABSORBER
SYSTEM DIMENSIONS
which in turn, could result in S0~
and NO sampling errors due to
X
scrubbing of the gas by the slurry
in the probe.
2.6.2 Venturi Absorber System
Gas leaving the particulate
scrubber first passes through an
orifice meter and then can be directed
in either of two ways: through the
venturi absorber or through the float-
ing bed absorber.
The venturi absorber and cyclone
are shown in Figures 2.20 and 2.21.
The absorbent slurry spray nozzle
enters just ahead of the venturi throat.
The venturi promotes gas absorption
by providing a zone with a high liquid-
to-gas specific surface. The gases
enter the cyclone separator, where the
absorbent slurry is removed from the
gas stream. As in the particulate
cyclone, vortex eliminators are in-
stalled on the bottom cone of the
cyclone. The slurry flows by gravity
to the sump located immediately under
the cyclone. The system dimensions
are shown in Figure 2.22.
The liquid level in the sump is
controlled by a level controller,
which maintains a constant level
with treated makeup water. The slurry
composition is maintained indirectly
by adjustment of the product flow
rate, which in turn controls the
water makeup rate via the level
controller.
2-11
-------
Due to problems with solids settling
an air-driven mixer is installed in the
venturi absorber sump to help keep solids
in suspension. The sump has a 12-by-12 inch
Plexiglas observation window, a pressure
equilizing connection with the cyclone,
and a liquid level indicator.
2.6.3 Floating Bed Absorber
The floating bed absorber (here-
after referred to as the FBA) includes
a sump, two contact stages, and a liquid
disengagement section. The FBA is de-
picted in Figures 2.23 and 2.24. This
countercurrent device takes in the flue
gas through the sump. Ancillary components
of the FBA sump include the following:
a liquid level controller, sump observation
window, level indicator, and air-driven
mixer. The sump is fitted with piping to
allow sump recirculation to promote agi-
tation to keep the solids in suspension.
Sump recirculation, however, was abandoned
in favor of the air-driven mixer.
Above the sump the FBA consists of
two stages. Each tray has an effective
flow area of 2 square feet and consists
of a I/8-inch-thick stainless steel plate
perforated with 3/8-inch-diameter holes
on staggered I/2-inch centers. Each
stage is packed with 6 to 8 inches of
"wiffle balls." An inspection of the
plastic balls after all tests showed that
the balls on the lower tray were deformed,
but those on the top tray retained their
original shape; see Figure 2.25. The
FIGURE 2.23 . FLOATING BED ABSORBER
BMISTER SECTION
SPRAY N022LE
FIGURE 2.24 FLOATING BED ABSORBER
2-12
-------
FIGURE 2.25 PLASTIC "WIFFLE"
BALLS DEFORMED
FIGURE 2.26 FLOATING BED ABSORBER
DIMENSIONS
3'-0"
4'-Q"
4'-0"
t'-O"
I'-O"
,,,
£
7
!•" v
-k±
^lir-
30" -
B
X. /
^
T
'
deformed balls were flattened on
the two sides parallel to the seam
where the two halves are fused to-
gether. Evidently the slight
increase in wall thickness along
the seam was enough to prevent
deformation along this plane. The
first time that this deformation was
noticed was after series M extended
tests, in which the FBA was used
for both S02 scrubbing and par-
ticulate collection. Since hot gas
directly from the furnace was enter-
ing the FBA, possibly hot spots de-
veloped on the lower tray, causing
the balls to deform. The overall
dimensions of the FBA are shown in
Figure 2.26.
The spray nozzle located above
the top tray directs the spray of
absorbing slurry onto the top tray.
Since the gas rises countercurrent
through the slurry, it comes into in-
timate contact with the absorbing slurry.
Gas leaving the top tray flows through
an angle iron baffle section which
serves to trap large water drops,
then through a York Demister to
ensure that all remaining droplets
are removed. The Demister is located
at the very top of the FBA and con-
sists of about 6 inches of Teflon
mesh fibers; see Figures 2.27 and 2.28.
Flow to the spray nozzle is
controlled by the aforementioned pinch
valve arrangement, and the slurry
2-13
-------
composition in the sump is controlled
by the product nozzle flow and MgO
makeup rates.
2.7 FLUE GAS EXHAUST SECTION
The cleansed flue gas leaving the
scrubber and absorbers passes downward
through a vertical length containing an
orifice meter, a gas sampling probe, and
a particulate sampling port with slide
valve. The vertical duct ends in a blank
flange; approximately 3 feet up from the
flange is the takeoff for the induced
draft fan. The purpose of this length of
pipe is to trap .any large liquid or
solid particles before the gas enters
the I.D. fan. A pneumatically operated
damper was located at the fan inlet.
After leaving the fan, the gas is released
to the atmosphere.
2.8 PILOT PLANT MONITORING SYSTEM
Several thermometers and thermo-
couples are located throughout the
system at important points; see
Figure 2.29. Also at various loca-
tions are manometers measuring static
pressures as well as pressure differ-
entials. Two water meters measure the
amount of makeup water used in the par-
ticulate cyclone and to the absorption
systems.
FIGURE 2.27 TOP OF FLOATING
BED ABSORBER
FIGURE 2.28 MOISTURE DISENGAGE-
MENT REGION -
SECTIONAL VIEW
LT
2.9 MgO MAKEUP AND PRODUCT DISPOSAL
The magnesium hydroxide makeup slurry
is prepared in the MgO slaking tank. This
2-14
-------
FIGURE 2.29 MEASUREMENT LOCATIONS
FIGURE 2.30 MAKEUP SYSTEM
tank is equipped with steam coils
and a large stirrer. The steam
coils are used only if the slurry
is slaked; otherwise the slurry
is mixed and dumped into the Mg(OH)2
holding tank below; see Figure 2.30.
The holding tank has a large stirrer
and a recirculation pump, which
helps to keep the solids in sus-
pension. The makeup slurry is drawn
from this recirculation line. A
small flexible impeller chemical
feed pump feeds the slurry through
a rotameter to the proper point.
This pump required a variable speed
drive. Note that no valves are used
in this line. See Figure 2.30.
In the early stages of the test,
MgO slurry was added at the suction
side of the venturi spray pump
(recirculation pump). However, when
changes were made in spray rates,
the pressure changes also affected
the MgO slurry flow rate. There-
fore the MgO makeup line was changed
to direct flow into either the venturi
absorber cyclone sump or the FBA sump,
depending on which system was operating.
During the tests with the venturi
absorber, problems were encountered
with deposits forming in the sump. There-
fore, on the last test using the ven-
turi absorber, J Series, it was decided
to add the magnesium hydroxide
makeup at the venturi throat in order
that the makeup stream would be more
completely mixed with recirculated slurry.
2-15
-------
The magnesium hydroxide makeup
slurry was added to the FBA sump until
the beginning of L series extended
tests. During the L, M, and N series
tests, the makeup was added to the
top tray of the FBA for better mixing.
The product from the particulate
scrubbing and the SO- scrubbing systems
was collected in four leach beds. See
Figure 2.31. Each leach bed consisted
of a cylindrical steel tank, 4 feet in
diameter by 5 feet high. Each bed was
filled with sand to a depth of about
1 foot. The particulate product entered
one tank while the sulfite slurry flowed
simultaneously into the other three tanks.
FIGURE 2.31 PRODUCT DISPOSAL
LEACH BEDS
2.10 CHEMICAL ANALYSIS LAB
The laboratory where all chemical
analyses were performed is shown in
Figure 2.32. The lab was located just
behind the furnace and was equipped with
a stainless steel sink, ample table space,
and storage space for samples.
2.11 OVERALL PILOT PLANT SCHEMATIC
Figure 2.33 shows in schematic form
the overall arrangement and the approximate
nominal flow rates of the various streams.
FIGURE 2.32 PILOT PLANT
ANALYSIS LAB
?_
16
-------
FIGURE 2.33 SCHEMATIC ARRANGEMENT OF TEST FACILITIES
CYCLONE SEPARATOR
FLOATING BED
ABSORBER
(FBA)
BASIC COMBUSTION
TEST FURNACE/"
PARTICULATE
SCRUBBER
\
•
t
s *
ll/
^
f*"
ORIFICE \
1
t
}
J
•-1 , £>
F.D. FAN
SEPARATELY FIRED
AIR PREHEATER
WATER SLURRY
IN DISPOSAL
•-r
SLURRY WATCR
DISPOSAL IN
TO STACK
DISPOSAL
- WATER
IN
t
— »
p
?
T
T, .
»—
f3*
"" 6
MBQ. - WATER
HEATING COIL
MgO PRIT.PARATION TANKS
LOCATION
1
2
3
4
5
6
7
APPROXIMATE FLOW
COAL FIRED
FLUE GAS
PARTI CULATE SPRAY
PARTICULATE PRODUCT
ABSORBER SPRAY
ABSORBER PRODUCT
MAGNESIA MAKEUP (10% SLURRY)
RATES
500 LB/HR
5,400 LB/HR (1100 SCFM)
10,000 LB/HR
2,500 LB/HR
30,000 LB/HR
900 LB/HR
400 LB/HR
2-17
-------
3.0 TEST APPARATUS AND PROCEDURES
5.1 TEST PROCEDURE FOR 1-DAY OPERATION
A typical testing day in which four tests were usually completed is
described below.
A test day began with the start-up of the furnace. First, this necessi-
tated firing of the air heaters and allowing them to reach temperature (^750 F).
This usually took about an hour. The coal feed was then ignited. The furnace
gases at this time were vented directly to the atmosphere via the stack relief.
A waiting period of approximately 1 to 2 hours followed until the furnace
ignition was stable.
While waiting for the desired furnace conditions, the particulate scrubbing
system and the appropriate gas absorber were being readied for operation. During
this time the MgO makeup slurry was prepared to the proper concentration. If
slaking of the makeup was required, preparation was made the night before, and
the MgO was allowed to slake overnight. When the furnace was ready, the relief
valve was closed, thereby directing the flue gas to the scrubbing systems.
After the transfer of gas was made, it usually took about an hour or so
after the MgO makeup was started before the chemical composition of the absorber
system reached steady state and the first test could begin. Steady state was
assumed when two or three Palmrose analyses taken at 10 minute intervals showed
that the system had reached the desired operating conditions and also when the
S02 concentration at the pilot plant exit was constant. A typical test lasted
for about an hour, during which a complete set of data was taken. Test data was
usually taken midway in the 1 hour test period.
During a test several samples were taken and put into storage in the event
that they were needed at a later date. If certain test data were found to be
unusual, an analysis of the stored samples could provide a means for determining
what occurred. Samples collected and stored in 8-ounce bottles consisted of the
following: MgO makeup slurry, particulate cyclone slurry, S02 scrubber slurry,
and a makeup water sample.
After a test was completed, the operating conditions were changed for the
next test. Depending upon the operating parameter changed, there was usually
a waiting period of about 1/2 to 1 hour before steady state was again reached and
the next test could begin. Following this procedure, about four 1 hour tests
were completed per day.
3-1
-------
5.2 TEST PROCEDURE FOR EXTENDED OPERATION
The start-up for extended operation did not differ from that of a normal
test day. Test conditions were specified to be constant for each 24-hour
period. Excess air was the only variable scheduled for change at 24-hour
intervals .
Each extended run was designated by a letter and separated into a number
of tests designated by a number preceded by the letter of the extended run.
Each individual test lasted for 1-1/2 hours, during which complete data were
taken consisting of chemical analyses and temperature and pressure measurements.
The 1-1/2-hour tests were attempted consecutively. However, due to some problems
with deposits and the like, continuous operation was interrupted. Selected
tests such as SO.,, NO , and Thorston dust sampling were done at about 5-hour
O X
intervals.
Prior to starting an extended run, the pulverized coal bunker was filled
completely, to 30-ton capacity, which provided enough coal for a typical 4-day
extended test. Magnesium oxide makeup slurry was prepared about every 7 hours
in the slaking tank and then dumped into the holding tank.
5.3 SPECIAL TESTS
5.3.1 High Sulfate Concentration Tests
Two series of tests, K and N, were run with an artificially induced high-
sulfate concentration. These tests were run to determine the effect of high-
sulfate concentration on the absorption of S02. The high-sulfate level was
attained by adding Epsom salt, MgSO.-TH-O, to the magnesium oxide slurry in
the holding tank; enough Epsom salt was added to obtain a saturated solution
in the makeup. Addition of this makeup to the SO- absorber system resulted in
a sulfate concentration of about 151 of saturation in the absorbent slurry.
These tests were run as normal 1-hour tests.
3.3.2 NO Absorption Tests
1 ^ J\.
The NO absorption tests involved injecting nitrogen dioxide into the
J\.
gas stream and relying on the following reaction to occur:
NO + N02 (injected) -»• N20.,
3-2
-------
FIGURE 3.1 N02 INJECTION
APPARATUS
FIGURE 5.2 S02 SAMPLING SYSTEM
to, sun
UXATIC
NC
fc PROBE
IU7T PUNT
BUT
3
>v sunoc
CbCATTW
[^
NS
n
]
,w
u
SM^C ^_.
IflCATtOJ **
_^ PINCH 1
OKDSOE TOPS
This compound nitrogen trioxide,
would then be absorbed by the ab-
sorbent slurry by the reaction:
Mg(OH).
H20
An attempt was made to add
a stoichiometric quantity of
N09 required to react with the
L*
NO from the furnace. It was in-
tended to meter liquid NO- from
a bottle of liquid N09 through a
t*
rotameter and vaporize it in a
cylindrical container over a hot
plate. However the elutriation
tube in the bottled N02 which was
to supply the liquid NO- was absent.
Therefore a suitable rotameter was
obtained, and gaseous N02 was metered,
passed through the vaporizer, and
injected into the system. The
apparatus is shown in Figure 3.1.
The gaseous NO- was injected at
the particulate cyclone exit, just
a few inches upstream of the
straightening vanes. It was thought
that the vortex action of the gas
leaving the cyclone would promote
complete mixing of the injected NO-.
3.4 SULFUR DIOXIDE SAMPLING AND
ANALYSIS
3.4.1 Barton Titrator
One of the most important
analyses was the measurement of the
sulfur dioxide concentration in the
3-3
-------
flue gas. S02 sampling probes were FIGURE 3.5 BARTON TITRATORS
located at the furnace exit duct,
the particulate cyclone exit, and
the pilot plant exit; see Figure
3.2. This analysis was performed
for the most part by a Barton Model
286 Recording Electrolytic Titrator;
see Figure 3.3. The Barton Titrator
works on the principle of coulometry
titration. This method employs elec-
trolysis to generate a reagent which
reacts stoichiometrically with the sub-
stance to be determined. The quantity
of substance reacted is computed via
Faraday's Law from the current. This
is analogous to the volume of standard
solution used in chemical titrations. The details of the sampling procedure
are described in Section A.I.
The instrument consists of three basic modules: a titration cell, a
solid-state electronic control circuit, and a strip chart recorder.
The heart of the unit is the titration cell, which consists of a com-
bination electrode enclosed in a chamber containing hydrobromic acid and free
bromine. The gas being analyzed is bubbled through the solution, where the
sulfur dioxide is allowed to react with the bromine by the following reaction:
S0
2H20
2HBr
(3.1)
The instrument attempts to maintain a constant bromine concentration by electro-
lytically reducing some of the hydrobromic acid to bromine and free hydrogen.
The amount of current required to maintain a steady-state concentration of
free bromine in solution is a measure of the amount of bromine reacted and is
therefore a measure of the amount of SCL reacted. A schematic diagram of the
electrodes and reaction chamber appears in Figure 3.4.
The Barton instrument is extremely sensitive. The accuracy of coulometric
titration has been noted by Lingane *• ^. "The precision and accuracy of coulo-
metric titration are both excellent, and the method is conveniently applicable
3-4
-------
FIGURE 3.4
BARTON ELECTRICAL
CIRCUIT
Voltage
Adjust.
V
/
Meter
bens ing
Current
Solid State
Electronic
Control
6V.
Generating
Current
I
Br?
Sensing
Electrode
Br2
Corranon
Generating
Electrode
Generating
Electrode
(Reference
Electrode)
FIGURE 3.5 REICH BOARD
to amounts ranging from about 100
milligrams down to as little as a few
hundredths of a microgram in volumes of
the order of 10 to 50 cc. Even in the
microgram range titrations with errors
of only a few tenths of a percent are
possible."
The accuracy of the Barton instru-
ment is also increased because coulo-
metric titrimetry involves a fundamental
quantity, current. This measurement is
relatively "absolute" and free from the
uncertainties associated with a normal
chemical titration.
In a study carried out by the U. S.
Department of Health, Education, and
Welfare, Performance Characteristics of
Instrumental Methods for Monitoring Sul-
fur Dioxide, v •* lag and response time
for the Barton instrument were shown to
be superior to those for most of the other
commercial S02 analyzers tested. Sensi-
tivity was also good, being 0.03 ppm S02
with the base level at 1.00 ppm SO-.
Sensitivity was defined as the smallest
change in SO- input concentration that
produces a readable change in recorder
output .
3.4.2 Reich Method for SO,, Analysis
The Reich method for S02 analysis
was used as a spot check on the Barton
SO- analyzer. This method is useful for
the determination of SO- in a gas stream
when H-S is absent. This test consists
3-5
-------
of aspirating the gas through
standard iodine solution until it
is decolorized. Since the amount
of iodine used is known and the
volume of the aspirated gas is
measured, the percentage of S0~
can readily be calculated. A
photograph of the Reich apparatus
is shown in Figure 3.5.
5.5 SULFUR TRIOXIDE SAMPLING
The determination of sulfur
trioxide concentration was accom-
plished by condensation of the sul-
fur trioxide vapor followed by titra-
tion of the condensate. This method
was adopted from H. Goksoyr and
K. Ross.^- ' The titration procedure
was developed in-plant. The sampling
apparatus and method are presented
in Section A.2.
A sampling probe consisting
of a quartz tube wrapped with a
heating element was placed in the
furnace exit duct. The flue gas was
drawn through the quartz tube and
into the sulfur trioxide collector,
which consisted of a helical coil
enclosed in a water jacket. The
purpose of the water jacket was to
keep the helical coil below the dew
point of SO., and above the dew point
of water vapor.
From the collector the gas then
passed through a sulfur dioxide absorber,
drying column, gas meter, and finally
to the vacuum pump; see Figures 3.6
and 3.7. 3.6
FIGURE 3.6 S07 COLLECTION APPARATUS
FIGURE 3.7 SO, COLLECTION APPARATUS
-------
A total of about 2-1/2 cubic feet of gas was sampled, after which the
collector was washed out with 80% isopropyl alcohol. The washings from the
collector were then titrated with barium chloride solution to determine the
amount of sulfur trioxide gas in solution as sulfate.
Sulfur trioxide measurements were attempted at the particulate cyclone
outlet duct. However, they were unsuccessful. When sampling, the sulfur tri-
oxide collector trapped fly ash, liquid carryover, and other solid particles
which escaped collection by the particulate scrubber. Fly ash itself does
contain sulfates, which would cause errors in analysis. Using a glass wool
filter to filter out solid particles was precluded because it would also filter
out the sulfur trioxide, which would have been in the form of a mist at the
temperatures encountered.
3.6 N0x SAMPLING AND ANALYSIS
NO analyses were taken in duplicate and at times in triplicate at three
J\,
points: the furnace exit, particulate cyclone exit, and pilot plant exit.
Sulfur dioxide sampling lines running from these points were fitted with a
glass tee near the probe from which the NO samples were taken. The Phenol-
A.
Disulfonic Acid Method was used for analysis.
A typical NO sampling and analysis consists of the following. To a
1000 ml flask is added an absorbing solution consisting of hydrogen peroxide
and dilute sulfuric acid. The flask is then evacuated; a gas sample is now
ready to be taken. The line to the Barton SO- Analyzer is pinched off, and
one arm of the three-way stopcock is connected to the tee. The stopcock is
opened, and the gas sample is drawn into the flask.
The nitrogen oxides are converted to nitric acid by the absorbent solution
and are then reacted with phenoldisulfonic acid to produce a yellow compound,
which is measured colorimetrically. The color is measured with a photometer
and compared with calibration curves made with a solution containing a known
amount of nitrite.
A more complete and descriptive explanation of the phenoldisulfonic acid
method used in these tests is given in the American Society for Testing and
Materials, Standard Method of Test for OXIDES OF NITROGEN IN GASEOUS COMBUSTION
PRODUCTS (PHENOL-DISULFONIC ACID PROCEDURE), ASTM Designation: D1608-60.
3-7
-------
With the exception of the following
modifications, the ASTM Method was
followed as written:
Modifications to ASIM Method:
1. Absorbent Solution - 2 ml of
H202 (3%) was added to 50 ml
of 0.1N H2SO.. This is about
4 times the peroxide concen-
tration called for in the ASME
Standard.
2. In place of IN sodium hydroxide
solution, IN potassium hydroxide
solution was used.
The modified Saltzman method for
the determination of NO concentrations
Jv
in flue gas was initially selected as
the preferred method because it is
more rapid and easier to use than the
phenol-disulfonic acid method. However,
before accepting the Saltzman method,
bench-scale tests were made to deter-
mine the effects of sulfur dioxide on
the results. These tests showed that
sulfur dioxide, at the expected con-
centrations, would bleach the Saltz-
man reagent color. Rapid bleaching
started within 15 minutes. Therefore
it was decided to use the more accurate,
but more tedious, PCS method for NO.
analysis.
FIGURE 3.8 PARTICULATE SAMPLING
NOZZLE
x
5.7 DUST LOADING
Dust sampling is carried out at
three points in the pilot plant: the
furnace exit, the particulate cyclone
exit, and the pilot plant exit. A
dust probe especially designed to
WHATMAN CELLULOSE
EXTRACTION THIMBU;
1-1/4"
3/4" SCHD 40
304 SS
PIPE
15/16"
"0" RING SEAL
#4 DRILL
304 SS
FIGURE 3.9 PARTICULATE SAMPLING
APPARATUS
3-8
-------
FIGURE 3.10 PARTICULATE SAMPLING
APPARATUS
FIGURE 3.11 THORSTON DUST SAMPLER
accommodate a filtering thimble placed
immediately behind .the sample nozzle
(see Figure 3.8) is used to sample gas
at the particulate scrubber and pilot
plant exits. The thimbles are Whatman
cellulose extraction thimbles used
for this purpose in the past. The
probe is designed to avoid the usual
shortcomings of conventional sampling
normally employing a large filter at
the end of the probe. Deposition of
dust in these probes ahead of the filters
is a recurring problem. The remainder
of the dust sampling system includes
a drying column, a gas meter, and a
vacuum pump, Figures 3.9 and 3.10. The
sampling apparatus is described more
fully in Section A.3.
Isokinetic sampling is performed
by regulating the flow rate to corre-
spond with the calculated velocity at
the sampling point. This method is used
instead of the null balance method
since the gas flow rate through the
pilot plant is relatively constant.
Representative dust sampling at
the exit of the particulate cyclone
separator was potentially difficult be-
cause of the spinning action of the gas
leaving the cyclone. However, this
problem was circumvented by placing a
flow straightener at the cyclone dis-
charge, which stopped the spin and there-
by redispersed the dust.
Use of the dust sampling probe was
attempted at the furnace exit. For this
3-9
-------
attempt we tried using the Whatman cell-
ulose extraction thimble. However, as
might be expected, the furnace exit gas
temperature was high enough to char the
thimble.
A Thorston dust sampler installed
in the furnace exit duct was used. This
device samples much larger gas volumes
than the mini sampling probe. It con-
sists of a wedged-shape nozzle driven
at a constant rotational speed so that
it progressively traverses all areas of
the stack at about 4 rpm; see Figure 3.11,
The sample nozzle area covers one \ of
the duct area. Gas sampling rates were
measured by an orifice. The gas was
filtered through a nylon bag. This
sampler was normally run for an hour,
during which several grams of material
were collected.
5.8 CASCADE IMPACTOR
A cascade jjnpactor, shown in
Figure 3.12, developed in-plant was
used on one occasion to determine
particle size distribution at the
furnace outlet. The cascade im-
pactor consists of a cyclone and
five impactor stages.
The principle of the impactor
is as follows. The cyclone is sized
to collect 2.5-micron particles with
an efficiency of 50%. The remaining
particulate matter passes through the
impactor stages. A typical stage is
FIGURE 3.12 CASCADE IMPACTOR
FIGURE 3.13 IMPACTOR STAGE
ORIFICE
COLLECTION CUP
SPRING
IMPACTOR STAGE
GASKET
3-10
-------
shown in Figure 3.13. The particle laden gas passes through an orifice,
which increases its velocity. The gas stream is deflected by the collection
cup, and the heavier particles fail to make the abrupt change in direction
and impact upon the collection cup. Placing the orifice closer to the
collection cup traps smaller particles on the cup. The five stages are
designed to collect different particle sizes, and the collection cups are
weighed after sampling to determine the relative distribution of particle
sizes.
3.9 ORSAT ANALYZER
A Fisher Orsat apparatus was used from time to time to analyze the flue
gas for oxygen and carbon dioxide. The Bailey analyzer was used more or less
as a check on the Bailey oxygen analyzer Model A57 and also to detect air
leaks into the system. The Orsat apparatus used in this test is shown in
Figure 3.14.
3.10 MAGNESIUM SULFITE-BISULFITE ANALYSIS
A modified Palmrose analysis was used in these tests to determine the
chemical composition of the absorbing slurry. The recirculated slurry used
to absorb SCL in the scrubbers is
FIGURE 3.14 ORSAT APPARATUS
an aqueous mixture of magnesium sul-
fite, magnesium bisulfite, and mag-
nesia. The magnesium sulfite is only
partially soluble at the temperatures
and concentrations encountered in this
application. Thus, the actual major
constituents are MgSO -6HJD, ,, Mg++,
SOj, HSO", OH", MgO, and Mg(OH)r
This procedure is presented in Section A.4.
During each test two different
Palmrose analyses were made from an un-
filtered sample and a filtered sample.
3-11
-------
The analysis on the unfiltered sample
was representative of the total sample,
containing both dissolved and undissoived
species. The filtered sample represented
only those constituents which were dis-
solved. By taking both a filtered sample
and an unfiltered sample, one can obtain
a better idea of the actual makeup of
the absorbing slurry. A photograph of
the syringe filter holder is shown in
Figure 3.15.
The 2 ml sample is withdrawn by
a syringe from the SO- scrubbing system
recirculation slurry line; see Figure 3.16.
This line is fitted with a septum for in-
sertion of the needle into the line. Fig-
ure 3.17 shown the Palmrose titration
apparatus.
3.11 SULFATE ANALYSIS
5.11.1 Titration Method
The analysis of the absorbing
slurry for sulfate concentration was
another important determination re-
quired in these tests. During every
test performed, at least one 10 ml
sample was taken for sulfate analysis.
Samples were taken with a 10 ml syringe
and withdrawn through a septum from
the S02 scrubbing slurry recirculation
line; see Figure 3.16.
FIGURE 3.15 SYRINGE FILTER HOLDER
FIGURE 3.16 SYRINGE IN SEPTUM
3-12
-------
FIGURE 3.17 PALMROSE TITRATION
APPARATUS
FIGURE 3.18 SULFATE SAMPLE PRE-
PARATION APPARATUS
Sulfate analysis was carried out
by a method developed in plant. Sampling
was done so that oxidation was avoided.
Sulfites and bisulfites were expelled
from the sample by boiling in acid
under nitrogen atmosphere, Figure 3.18.
Interfering cations were removed
by a cation (H+) exchange column,
Figure 3.19. The resulting sample was
titrated under controlled conditions
with barium chloride. See Section A.5
for a detailed description of the method.
5.11.2 Sulfate Sample for Chemistry
Lab Analysis
A sample of the absorbent slurry
from the SO- absorber was taken at
various times, and the sulfate concen-
tration was determined by our Chemistry
Department using a standard gravimetric
analysis. This was to provide a check
on the sulfate analysis method developed
in plant. The sample was taken from the
product nozzle into an 8-ounce bottle
containing 75 ml of isopropyl alcohol
to prevent further oxidation of the
sample.
3.12 PARTICULATE SOLIDS SAMPLING
A method was devised to determine
the relative amount of solids, prin-
cipally fly ash, contained in the par-
ticulate scrubbing slurry. This mea-
surement was taken for a general docu-
mentation of data and also for the
3-13
-------
future possibility of relating it to
the "wear rate" of equipment such as
pumps.
The actual test was very simple
and consisted of filtering a known
volume of sample and weighing the col-
lected solids, thus determi-ing the per-
cent solids in the slurry. See
Section A.6 for a more detailed de-
scription.
5.13 MAGNESIUM SULFITE SOLIDS SAMPLING
The method for the determination of
percent solids in the SO- scrubber is
exactly the same as that used in the
particulate scrubber slurry analysis.
The amount of sulfite solids is de-
termined so that a material balance around
the S0? scrubbing system can be made.
See Section A. 6, Particulate Solids
Apparatus and Procedure, for a detailed
discussion. The only deviation from this
procedure was that only about 40 ml of
sample was filtered because of the
greater solids content.
5.14 MgO MAKEUP SLURRY TITRATION
In order to determine the amount
of magnesium hydroxide solids in the
makeup slurry, a simple acid titration
of a known volume of sample was done.
Previous to this a set of MgO slurry
standards was prepared, and a curve
of percent MgO solids versus ml of
acid was drawn. See Figure 3.20
FIGURE 3.19 SULFATE TITRATION
APPARATUS
FIGURE 3.20 MgO MAKEUP TITRATION
CURVE
';!
i
•
§
* i
S
s*
r
\
/
1
/
\
s
/
/
/
/
s
Milliliters of 7.14 Nonwl ItCl
3-14
-------
The curve was based on using a 20 ml sample and titrating with 7N HC1.
See Section A.7 for the procedure.
5.15 OTHER MEASUREMENTS
The density of the various slurries was determined by a hydrometer.
Measurements of pH of the slurries were made with a Leeds-Northrop pH
indicator (Cat. #7664).
3-15
-------
4.0 RESULTS
The results of both the parametric and extended tests are consolidated
into one set of results. The only tests treated separately are the NO in-
A.
jection tests. Because of the great volume of data gathered, the computer
printout for all tests will be included in an addendum to this report. A
consolidating table of summary results will also be included in that addendum.
4.1 S02 ABSORPTION - PARTICULATE SCRUBBER
No S02 absorption in the particulate scrubber was discernible under any
conditions tested. The pH of the water fly ash slurry ran between 2.0 and
3.5. No chemical, additives were used in this scrubber.
4.2 S02 ABSORPTION - FLOATING BED ABSORBER
4.2.1 Parameter Study
The SO- absorption capability of the floating bed absorber (FBA) showed
varying dependence upon the several parameters studied. Qualitatively, this
dependence was as follows:
1. Liquid-to-gas-ratio: Absorption increased significantly with
increasing L/G.
2. Stoichiometric ratio of tygO to SO,,: Absorption increased with
increasing NJgO flow up to a molar ratio of MgO/SO- = 1.0. Above
this value absorption did not increase significantly.
3. pH of scrubbing slurry: Absorption increased with increasing pH
up to a pH = 7. Above this value absorption remained nearly constant.
4. Pressure drop: Absorption increased with increasing pressure drop
c
across the bed.
5. Fly ash: The effect of the presence of high levels of fly ash
(during extended operations when the particulate scrubber was
bypassed) was uncertain.
6. Magnesium sulfate: The presence of high concentrations of dissolved
magnesium sulfate reduced SO- absorption in the low pH range. At higher
pH values, SO- absorption was not adversely affected.
7. Slaking: Conversion of magnesium oxide to the hydroxide before in-
jection into the absorber did not improve S02 absorption.
8. Residence time: Increasing the total residence time of liquid within
the FBA (including the sump) did not improve absorption. This should
4-1
-------
not be confused with the residence
time of liquid in intimate contact
with the gas, which most surely
could affect absorption.
4.2.2 Floating Bed Absorber _-_
Absorption Correlation"
In the presence of less than
0.10 gm fly ash/100 ml solution and
0.5 gm MgSO./lOO ml solution, the
S09 absorption data correlated well
«
in terms of two operating parameters,
L/G and spray slurry composition.
The slurry composition, whidi is
dependent upon the ratio of MgO/SC^
and the solubility of MgSO,, can be
expressed indirectly in terms of the
bisulfite concentration, Mg(HSCL)2 or
slurry pH. Figure 4.1 illustrates
the absorption as a function of
L/G and bisulfite concentration.
This data includes absorption across
the sump, the two stages, and the
mist disengagement section. Figure
4.2 is a similar plot in which the
chemical composition is expressed
in terms of pH.
The results illustrated in
Figure 1 have been correlated by
the following semiempirical ex-
pression using nonlinear regression
analysis:
FIGURE 4.1
S02 ABSORPTION - FBA
In
- 24.97
- 24-97
= 0.633 L/G +
2.26 (4-1)
FIGURE 4.2 S00 ABSORPTION - FBA
Recirculited Liquid pti
4-2
-------
FIGURE 4.3 CORRELATION FOR EQUATION
4-1, S00 ABSORPTION - FBA
Uquld-u-Cks Ratio
IWlb
X J.B • «.l
21.* - i.6
1.1- 1.7
FIGURE 4.4 PRESSURE DROP - FBA
Gu Velocity
13 to II.* fi/M
where y.T = inlet SO- concentration, ppm
y1 = exit SO- concentration, ppm
B = exit bisulfite concentration,
gm/100 ml as S02
L/G = liquid-to-gas ratio
2
The correlation coefficient (R )* is
97.45. The second term in the argument
of the log term represents the vapor
pressure effect. The correlated data
is illustrated in Figure 4.3. The
derivation of the expression is described
in Section 5.3.2.
Although absorption did seem to re-
late to pressure drop, this factor could
not be isolated from the L/G effect as
expressed above. This follows because a
one-to-one correspondence existed between
L/G and pressure drop (the gas velocity
was constant). This relationship is ex-
pressed in Figure 4.4.
An uncertainty about the effect of
fly ash upon S02 absorption exists be-
cause of the high level of absorption
data scatter associated with the final
extended test. Figure 4.5 illustrates
this scatter. This figure is a plot of
measured k A values versus calculated
values based upon a rearrangement of
Equation (4-1). The standard deviation
is 69.0 compared to 17.0 for the cor-
related data in Figure 4.3.
*The statistical meaning behind the
correlation coefficient is explained
in Appendix B.
4-3
-------
Only five data points were
obtained on elevated sulfates. The
sulfate concentration in the scrubbing
slurry was between ^181 and 60% of
saturation. However these points showed
a significant effect upon absorption.
This data is illustrated in Figure 4.6
against the correlated curve for the
normal sulfate situation.
The effect of slaking upon SCL
absorption was minimal partly because
the MgO was sufficiently reactive to
convert a substantial portion to
Mg(OH)2 in cold water. Table 4.1 lists
the fractional conversion of MgO with
various conditions under which this
conversion took place.
TABLE 4.1 MgO CONVERSION TO Mg(OH),,
FIGURE 4.5 S02 ABSORPTION - FBA
WITHOUT PRESCRUBBER
Test
D-030
D-043
D-036
Slaking
Period,
Hours
5-7
5-7
2-3
Slaking
Temp., F
45°
60°
180°
Con-
version, %
54.4
83.3
94.9
4.3 S02 ABSORPTION - VENTURI ABSORBER
4.3.1 Parameter Study
Except for pressure drop, the ven-
turi absorber S02 performance showed
a minimal dependence upon the parameters
examined. The qualitative response is
described below:
1. Liquid-to-gas ratio:
Absorption was largely
unaffected by changing L/G ex-
cept when L/G fell below 2.2 Ib/lb.
I
g 500
§ alll:
FIGURE 4.6 COMPARISON OF EFFECT OF
HIGH SULFATES UPON S02
ABSORPTION - FBA
/
/
;
•
/
/
/
/
/"
^
-<
->"
• • ',
Iff,
Figure 4.2
- 5.9 to 4,1 Ih/lb
«^ —
1 \—
ominal Sul fa
0.2 pii/100 n
•
Ele
Sulfate
l/(
t-f i ->\
1 as E
1
. . | i
ri
- ; 1
~$
ated Sujfates
3.0 gm/100 ml as S02
• J.O Ih/lb -1
pH of Scrubbing Slurry
4-4
-------
2. Stoichiometric ratio, MgO/SO^: Sulfur dioxide absorption increased
slightly with increasing MgO feed rate.
3. pH of scrubbing slurry: Sulfur dioxide absorption increased
moderately with increasing pH.
4. Pressure drop: Sulfur dioxide absorption increased significantly
with increasing pressure drop.
5. Fly ash: The presence of fly ash had no measurable effect upon
SO- absorption performance.
6. Magnesium sulfate: The presence of moderate concentrations of
magnesium sulfate had a retarding effect upon SCL absorption.
This effect was determined only for low pH.
7. Slaking: Conversion of magnesium oxide to the hydroxide before
injection into the venturi did not improve SCL absorption.
8. Residence time of liquid: Increasing liquid holdup (including the
sump) did not improve SCL absorption.
FIGURE 4.7 S02 ABSORPTION - VENTURI
ABSORBER
4.5.2 Venturi Absorber Absorption
Correlations
Since the bulk of the SO- absorp-
tion data ranged only from 65 to 85%,
no strong correlations between absorp-
tion and operating parameters (with the
possible exception of pressure drop)
were determined. The effect of composi-
tion upon absorption for a wide range
of L/G's is shown in Figures 4.7 and 4.8.
Mathematically this data is correlated
as follows:
£n(l-E) = -0.694[l-e1-633^H ' 5'0)] -
iia Sulfite Concentration, g> .SO,/100 ml
0.8779
(4-2)
where E = absorption efficiency,
fractional
pH = acid strength
4-5
-------
The correlation coefficient of 35%
was low primarily because the vari-
ance in the percentage absorption
was narrow. This curve fit applied
only for pressure drops between
1.0 and 2.0 inches w.g.
Although there are only three
data points at elevated pressure
drop, arising by an accidental test
complication, the effect is sufficient
to warrant presentation here. The
correlation between efficiency and
pressure drop is shown in Figure 4.9.
A dotted straight line is used be-
cause there is insufficient data
to justify a linear relationship.
Mathematically this data is
correlated as follows:
m(l-E) = -1.039 - 0.3137 (AP) (4.3)
Correlation coefficient = 67%
£n(l-E) = -0.3298 AP [1.-e0.9547(PH-5.0)]
FIGURE 4.8 S02 ABSORPTION - VENTURI
ABSORBER
Correlation coefficient = 65$
where E = absorption efficiency,
fractional
AP = venturi absorber pressure
drop, "w.g.
pH = acid strength of recir-
culating slurry
Equation 4.3 is a simple straight line
least squares fit, and the second
correlation is a nonlinear regression
analysis curve fit which includes the
99.4
99.2
99
98
i"
•3 %
1
i "
§" 92
S go
S
S(>
70
6D
to
:o
"
i
"•*
-•^
•
-—•
: !
— i —
•
__'
4
r»
I
*
1 <
Liq
»
j '
lb/lb
5.3
f.2 - 4.9
S.S
2.8
ess than Z.
Venturi Pressure Drop
Average i-l.S Inches w.g
..I. .-!-.
•
'_
FIGURE 4.9 PRESSURE DROP EFFECT
UPON S02 ABSORPTION -
VENTURI ABSORBER
.
l|l|
| :_,:
fflS
' I"
BE
Ji;1
•
•
.S:
' •
>
—
'.-' : '
niT]
-^
/
___
': • .
/
'.\\\
/•
:: .
••\\\
: .' :
~~jr
m<
/
f/'
•• \ --.
'
~r
Liquid-to-Gas Rati
pH » 7
4 Test Seris
B Nraninal Te
s - J
sts, F
~p {-'-
, ..^|-
-m
: :
.
P
1 lb/lb r
i
Iture I.I r
10 12
Venturi Pressure Drop, Inches w.g.
4-6
-------
FIGURE 4.10 S02 ABSORPTION CORRELA-
TION FOR VENTURI ABSORBER
ir
X
SO. Inlet
I' Exit . Calculated (FK» Equation 4.4)
FIGURE 4.11 COMPARISON OF EFFECT
OF HIGH SULFATES UPON
S02 ABSORPTION -
VENTURI ABSORBER
slurry pH effect. This latter correla-
tion must be used with caution since the
pH and pressure drop effects were determined
separately. The interaction between low
pH and high pressure drop was not deter-
mined experimentally. Nevertheless Fig-
ure 4.10 shows how Equation 4.4 correlates
all venturi absorption data except that
from the elevated sulfate tests. All of
the venturi absorber S09 absorption data
£t
is tabulated in Table 4.2 of the addendum.
Although only three data points
were obtained on the effect of elevated
sulfate concentration (approximately 10%
of saturation) and all were below a pH
of 5.7, the data was consistent and
showed a definite retarding effect. This
data is shown in Figure 4.11 plotted
against the datum condition as correlated
by Equation 4-2. Whether this effect
would vanish at higher pH, as was in-
dicated in the analogous FBA tests, was
not explored.
4.4 CRYSTALLINE DEPOSITION
Deposition of hard crystalline de-
posits in piping, pumps, sumps, and spray
nozzles varied from not at all to ex-
tremely rapid. The degree and rate of
deposition were not quantified except in
terms of how rapidly the scrubbing system
was adversely affected.
The deposits resulted primarily from
precipitation of magnesium sulfite. The
form of the deposit was identified by X-ray
diffraction and is reported in Table 4.2.
4-7
-------
TABLE 4.2 X-RAY DIFFRACTION ANALYSIS OF CRYSTALLINE DEPOSITS
Location
Venturi Absorber
Sump Outlet Flange
Recirculation
Pump Suction in
Plastic Pipe Section
Venturi Absorber
Slurry Spray Nozzle
Identification
MgSOx-6H70 (Major)
•J L,
(Major)
(Possible Trace)
MgS03-6H20 (Major)
MgS04-H20 (Possible Trace)
4.4.1 Parameters Affecting Deposition
The parameters affecting the formation are described below:
1. Fly ash in spray slurry: Fly ash in concentrations of 1 to 3%
by weight completely eliminated the formation of crystalline
deposits within the scrubbing system. This is illustrated in
Figure 4.12, which depicts the effect of fly ash upon deposits
found in a 2-inch S.S. pipe leading to the spray nozzle.
2. Chemical composition of scrubbing slurry: The rate of deposition
was correlated to the acidity of the scrubbing slurry. On the
strongly acid side, Mg(HSO,,)2 >vL.O gm/100 ml as S02> deposition
FIGURE 4.12 THE EFFECT OF FLY ASH UPON DEPOSITS FORMATION
Without Fly Ash
With Fly Ash
4-8
-------
formed to a sufficient degree to adversely affect scrubber
operation within 2 to 3 hours. If the slurry was maintained
slightly alkaline, the system operated for about 24 hours
without excessive deposition.
Summarizing:
Bisulfite Cone. ,
gm/100 ml as SO., p_H Operating Period, Hours
>6 2-3
<7 ^24
3. I%0 makeup modifications: By relocating the makeup MgO feed location
from the sump to the active mass transfer section (the top plate of
the FBA and the throat of the venturi absorber) , by slaking the MgO
to the hydroxide, and by providing supplemental agitation in the
sumps, the deposition rate was reduced substantially. Prior to these
modifications, 24 hours was about the maximum operating life. These
modifications permitted runs up to 90 hours (the maximum possible be-
cause of a limited coal supply) without operating problems due to
deposition. Examination did show however that deposits had formed
under these operating conditions.
4. Materials : Deposition varied as a function of materials and the
location of these materials within the scrubber and auxiliary com-
ponents. Deposits at no time formed within the vertical PVDC line
from the recirculation pump to the spray nozzle. Deposition was
however a recurring problem at the spray nozzle and product nozzle
(both of stainless steel). Deposition was severe at times within
the recirculation pump and the sump. The plastic pipe (PVDC)
from the sump to the recirculation pump accumulated deposits
differing in appearance from those formed in the stainless steel
pipe. The deposits that formed within the plastic pipe were not
bonded to the surface as they were to the stainless steel surfaces.
No crystalline deposits formed at any time within the scrubbing
sections exposed to flue gas. Deposition seemed to be most
4-9
-------
pronounced at points of extreme turbulence such as the sump
suction and discharge and even along the threaded sections of
plastic pipe. Deposition occurrence is summarized in Table 4.3.
TABLE 4.3 DEGREE OF DEPOSITION AT VARIOUS LOCATIONS
DURING PERIODS OF RAPID FORMATION
Location
Venturi and cyclone: gas-liquid contact zone
Floating bed absorber: gas-liquid contact zone
PVDC recirculation line
Spray nozzles
Recirculation pump (casing and impeller)
PVDC line from sump to pump
Sumps below liquid level
Sumps above liquid level
Iron angle baffle (see Figure 2.28)
Degree of
Deposit Formation
nil
nil
nil
severe
severe
moderate
severe
nil
moderate
4.5 DEPOSITION AT WET-DRY LINES
Deposition of solids where the
flue gases first contact the spray
slurry was a problem which occurred
in various degrees as follows:
1. Hot flue gases* at parti-
culate scrubber inlet: No
deposits formed at any time.
The participate venturi in
Figure 4.13 after several
hundred hours of operation
is shown to be free of de-
posits.
FIGURE 4.13
PARTICULATE SCRUBBER
INSIDE VIEW
*Flue gases from furnace exit at
400 to 650 F.
4-10
-------
FIGURE 4.14 SULFATE FORMATION - FBA
WITH PRESCRUBBER
- L Serlei
• D Series
Pncott Oj In Floe Gu
2. Attemperated flue gases* at FBA and
venturi absorber: No deposits were
formed at any time.
3. Hot flue gases at venturi absorber:
Deposits formed rapidly. The analysis
by X-ray diffraction of these deposits
identified this material to be an
aggregate of fly ash and MgSCL-SHJD.
These deposits formed near the top
of the throat.
4. Hot flue gases at FBA inlet: Deposits
formed around the rim of the inlet
duct and within the horizontal section
leading to the inlet. See Section 2.6.3
for details of the inlet design.
FIGURE 4.15 SULFATE FORMATION - FEA
WITHOUT PRESCRUBBER
- I
J5
fereent Oj In Plus (ks
*Flue gases from particulate
scrubber cooled to
4.6 SULFATE FORMATION
4.6.1 Floating Bed Absorber
Sulfate formation within the FBA was
examined as a function of slurry pH, bi-
sulfite concentration, S09 mass transfer
£*
coefficient, oxygen concentration in the
gas phase, and operation with and without
the prescrubber. The following results
were apparent:
1. With the prescrubber in service
the sulfate formation was a func-
tion of oxygen concentration only.
This data correlated as follows
and is plotted in Figure 4.14.
Mole % sulfate = 0.287 +
1.2647 Yo2
Standard deviation = 1.163
Correlation coefficient = 69.3%
4-11
-------
2. With the prescrubber not in ser-
vice, i.e., with the furnace
exit gases proceeding directly
to the FRA, the sulfate level
was again a function of oxygen
concentration only. However the
level averaged 4.9 mole % higher
than that found in the tests
with the prescrubber in operation.
This data is correlated as follows:
Mole * sulfate = 5.186 + 1.213 Yo2
Standard deviation =1.94
Correlation coefficient = 44.5%
This data is plotted in Figure 4.15.
The averages of the individual tests
are listed in Table 4.4.
4.6.2 Venturi Absorber
The rate of sulfate formation in the
venturi absorber was quite erratic. The
tendency to form sulfates was examined
as a function of slurry pH, bisulfite con-
centration, SO- mass transfer coefficient,
oxygen concentration in the gas phase,
and operation with and without the
prescrubber.
The following results were apparent:
1. With the particulate scrubber
functioning, the rate of sul-
fate formation in the venturi
absorber appeared to be inde-
pendent of the above parameters.
follows:
TABLE 4.4 AVERAGE SULFATE FORMATION
IN FBA FOR INDIVIDUAL
TEST SERIES
Series Molnrity
D 0.0247
L 0.0230
L 0.02S4
L 0.0347
L 0.0386
L 0.0414
M 0.0360
M 0.0527
M 0.0542
Std. DevT Mole I Std. Dev.
0.0075
0.0073
0.0093
0.0056
0.0092
0.0049
0.0108
O.OJ43
0.0089
3.62
2.85
2.94
5.62
5.93
8.01
7.20
10.91
12.1
1.277
0.80
0.739
0.87
1.30
1.31
1.98
2.13
1.25
Oxygen
Cone.
2.4
2.0
3.0
4.0
5.0
5.7
3.2
4.3
5.5
FIGURE 4.16
SULFATE FORMATION -
VENTURI ABSORBER WITH
PRESCRUBBER
• • E Serin
• • C Serlw
» I S«rl«
A- H Serle*
+ - F Series
Percent Oj In Flia Gas
The average formation rate was as
Average sulfate =3.8 mole I of total sulfur gases absorbed. This
data is plotted in Figure 4.16. In terms of individual test series,
the sulfate formation rate is categorized as in Table 4.5.
4-12
-------
TABLE 4.5 AVERAGE SULFATE FORMATION
IN VENTURI ABSORBER FOR
INDIVIDUAL TEST SERIES
Test
Series
C
E
F
H
I
I
I
I Total Sulfur
0.0179
0.0506
0.0443
0.0439
0.01S2
0.0226
0.0237
Std. Dev.
0.0059
0.0196
0.0144
0.018S
O.OOS4
0.0108
0.0091
Hole I"
2.87
6.00
3.76
S.64
2.2S
2.6S
3.097
0.97
2.4S
1.13
2.85
0.736
0.75
0.991
Oxygen
Cone.
2.6
2.6
1.0
6.0
4.0
5.0
6.0
"Mole « of total sulfur gases present as sulutes.
FIGURE 4.17 SULFATE FORMATION -
VENTURI ABSORBER
WITHOUT PRESCRUBBER
2. With the particulate scrubber
removed from the circuit, i.e.,
with the furnace exit gases
taken direct to the venturi ab-
sorber, the sulfate formation
levels showed two significant
effects:
a. The sulfate formation level
was significantly elevated.
b. The sulfate formation rate
showed a possible dependence
upon oxygen concentration.
A least squares linear fit of this
data yielded:
Mole % sulfate = 1.485 + 1.2867 Yn
2
Standard deviation =2.6 mole \
Correlation coefficient (R2) = 251
This data and curve fit are plotted in
Figure 4.17. No other dependence such
as venturi pressure drop or S02 mass
transfer coefficient had any apparent
influence upon sulfate formation.
4.7 PARTICULATE COLLECTION
4.7.1 Particulate Venturi Scrubber
This scrubber operated under a limited
range of conditions listed as follows:
Pressure drop, 5.0 to 8.2" w.g.
Liquid-to-gas ratio, 1.6 to 2.1 Ib/lb
Ash slurry concentration, 0.4 to
1.1% by wt.
Gas velocity at throat, 145 to
165 ft/sec
With the exception of six points believed to be in error because the filter was
contaminated with slurry droplets carried over from the cyclone, the results ranged
from 0.010 to 0.045 grain/DSCF. These results were not correlatable to any of the
above parameters. The results plotted as a function of pressure drop are shown in
4-13
s ,
f%rufit 0. In Flu* C
-------
Figures 4.18 and 4.19. The
average particulate scrubber
dust loading was 0.0248 grain/DSCF
with a standard deviation of 0.0091
grain/DSCF. The equivalent average
efficiency was 99.241 for an average
pressure drop of 6.2" w.g. This data
is listed in Table 4.6.
For those tests in which inlet
loadings to the particulate scrubber
were obtained simultaneously with
exit loadings, the average efficiency
amounted to 99.19% at an average
pressure drop of 6.4" w.g.
The pressure drop relationship
for this venturi for a constant gas
velocity of 136 ft/sec is shown in
Figure 4.20.
4.7.2 Floating Bed Absorber With
Prescrubber
The particulate collection capa-
bility of the FBA upon the fine ash
passing through the particulate scrubber
was too low to measure with confidence.
The average exit dust loading was 0.0256
grain/DSCF compared to an average inlet
loading of 0.245 grain/DSCF. There
appears to be a tendency for the FBA
particulate collection to decrease
slightly with increasing L/G and pressure
drop. This is illustrated in Figure 4.21.
The results are listed in Table 4.7.
4.7.5 Floating Bed Absorber Without
Prescrubber
The floating bed absorber operating
upon the hot furnace gases performed
4-14
FIGURE 4.18
DUST COLLECTION PERFOR-
MANCE PARTICULATE
SCRUBBER
99.7
99.6
Liquid-to-Ota Ratio -pf
Ib/lb
• Z.O • Z.I
. A 1.6
i Pressure Drop. Inches K.R.
FIGURE 4.19 DUST COLLECTION PER-
FORMANCE PARTICULATE
SCRUBBER
Venturi Pressure Drop, Inches w.g.
-------
TABLE 4.6 PARTICULATE COLLECTION PERFORMANCE OF PARTICULATE SCRUBBER
B007
BOOS
B009
B010
C013
COM
GDIS
C016
C017
C018
C019
CO 20
C021
C022
C023
C025
DOW
Dust Loadings
Grains/DSCF
JeTExit
0.0288
.1230 0.0254
0.0237
1.1881
1.8448
3.5799
17.5586
4.1494
S.4823
0.0626
0.1009
0.0579
0.0198
0.0277
0.0328
0.0532
0.0227
0.0272
I Eft.
97.76
96.60
99.35
98.75
98.58
AP
6.4
6.2
6.0
5.8
6.1
5.9
6.1
6.3
6.3
6.3
6.3
5.0
5.0
5.5
5.4
5.7
7.6
L/C
•JL
1.6
1.6
1.6
1.6
1.6
1.6
1.6
2.0
2.0
2.0
2.0
2.1
2.1
2.3
2.1
2.0
2.0
Contaminated Filter
FIGURE 4.20
PRESSURE DROP -
PARTICULATE SCRUBBER
Velocity - 136 ft/ice
moderately well. An average exit
dust loading of 0.0609 corresponded
to an average collection efficiency
of 98.21 at a pressure drop of 6.0
"w.g. The operating parameters are
listed as follows:
Pressure drop, 5.0 to 6.8 "w.g.
Liquid-to-gas ratio, 3.7 to 4.3
Ib/lb
Gas velocity, 13 to 14 ft/sec
The primary results are listed in
Table 4.8 and plotted in Figure 4.21.
4.7.4 Venturi Absorber With Prescrubber
UVU.UH*MI., iwib The particulate collection capa-
bility of the venturi absorber acting
on the fine ash passing through the
particulate collector was too low to measure with confidence. The average mea-
sured dust concentration leaving the venturi absorber was 0.0238 grain/DSCF. Com-
pared to the average 0.0248 grain/DSCF leaving the particulate scrubber, this
corresponds to less than 5% collection, which represents a finer resolution than
that data justifies. The particulate collection results for the venturi absorber
are plotted in Figure 4.22. The parameter ranges and results are sunmarized
as follows:
4-15
-------
Pressure drop, 1.4 to 1.7' w.g.
Liquid-to-gas ratio, 2.8 to
4.9 Ib/lb
Average exit dust loading,
0.0238 grain/DSCF
Average inlet dust loading,
0.0248 grain/DSCF
All of the results are listed in
Table 4.9.
4.7.5 Venturi Absorber Without
Prescrubber
Although only two data points
were obtained, these were consistent
with usual venturi scrubber experi-
ence. These results are summarized
in Table 4.10. These results show
a substantial improvement in par-
ticulate collection as a function of
pressure drop. Note that the venturi
absorber operating at 8.4" w.g. is
about as effective as the particulate
venturi operating at ^6.5" w.g. but
is significantly less effective at
2.7" w.g. This data is also plotted
in Figure 4.22 to illustrate the
composition effect of the exit dust
loading with and without the par-
ticulate scrubber in operation.
4.8 N0x ABSORPTION
The degree of absorption of nitro-
gen oxides across the particulate scrubber
and gas absorber was negligible. The
average NO concentrations measured are
listed in Table 4.11. Although this
FIGURE 4,21 DUST COLLECTION
PERFORMANCE - FBA
3.0
1.0
o.n
0 . -10
II.. (ji
0.10
O.i 'ti
0.04
o.n
:
ffi
j i . ;
; . : .
:-..
Tr
A
I
!
*-•
*
1
ithout Pr esc rubber
•-With Prescrubber
Gas Velocity - )?-M ft/fee
4.0 4.5 S.O S.5 6.0 6.5 7.D 7.
TABLE 4.7 PARTICULATE COLLECTION
ACROSS FBA WITH PRE-
SCRUBBER
Test I
no',8
D042
D043
DOM
1X14"
L149
IXist Loadings
Grains/DSCT
Inlet
0.0240
0.036S
0.0241
0.0145
0.0159
0.0317
Exit
0.0354
0.0302
0.0167
0.0266
0.0179
0.0269
SP
"w.g.
6.5
6.6
5.7
6.6
5.7
3.9
4.2
2.6
4.3
2.6
S.O
TABLE 4.8 PARTICULATE COLLECTION
ACROSS FBA WITHOUT PRE-
SCRUBBER
Test I
Ml 7:,
Ml 81.
Ml";
IO97
M202
Dust Loadings
Grains/DSCF
et Exit
3.9848
1.9907
2.4981
4.9815
0.0502
0.0816
0.0482
0.0638
j Eff.
38.74
95.90
If
'V.g.
0.8
6.4
5.2
5.0
(..4
4.3
6.0
4.0
4.1
3.7
4-16
-------
TABLE 4.9 PARTICULATE COLLECTION
ACROSS VENTURI ABSORBER
WITH PRESCRUBBER
E049
ma
0.010*
0.017.?
0.0247
0.0460
data indicates a reduction in NO across
.X
each scrubber, this is not a valid con-
clusion since the standard deviation in
the NO measurements is larger than the
A
indicated NO absorption. All of the
.X
NO results are listed in Table 4.12.
J\.
FIGURE 4.22 DUST COLLECTION PER-
FORMANCE - VENTURI
ABSORBER
0.0<
0.03
m
m
—
Without
With Prt
Vcnturl Pressure Drop, Indies w.g.
TABLE 4.10
PARTICULATE COLLECTION
IN VENTURI ABSORBER
(WITHOUT PRESCRUBBER)
Test No.
J1JO
J131
JJ32
Dust Loading,
Grains/DSCF
Inlet
3.0191
0.0240
0.1003
Pressure Drop, "w.g.
9.0
S.4
2.7
Liquid/Gas, I/I
4.8
5.2
5.4
4.9 NO^ INJECTION TEST
The feasibility of promoting NO
absorption by injecting nitrogen dioxide
(N02) into the flue gas ahead of the ven-
turi absorber was tested. The principal
results of this test are as follows:
1. The total NO absorption was
/v
not improved. In fact, the
total NO increased from 635 ppm
at the furnace exit to 984 ppm
at the pilot plant exit.
2. The magnesium sulfate concen-
tration increased substantially.
The nominal sulfate concentration
without N02 injection ran approx-
imately 3.8 mole I total sul-
furous gases absorbed. This value
increased to between 15 and 23
mole I of the total absorbed sul-
furous gases when NO- was injected.
The primary results of this test are
listed in Table 4.13.
See Section 3.3.2 for a detailed
description of this test.
4.10
S03 FORMATION
The sulfur trioxide concentration at
the furnace exit varied between 1.10 and
144 ppm and averaged 18.7 ppm. These
4-17
-------
results are tabulated in Table 4.14.
The variance was examined as a
function of excess air and adiabatic
flame temperature but could not
be resolved.
4.11 THE FLOATING BED ABSORBER
TRAY HYDRAULICS
The effect of packing height upon
the dry pressure drop across the FBA
was evaluated. The results are listed
in Table 4.15.
The attrition or weight loss of
the balls after 270 hours of operation
was examined. The results are shown
in Table 4.16.
TABLE 4.12 SUMMARY OF NO RESULTS
TABLE 4.11 AVERAGE NOY CONCEN-
TRATIONS THROUGHOUT
PILOT PLANT
Furnace Exit
Paniculate Scrubber Exit
Pilot Plont Exit
Average N3x,
847
807
792
Standard Deviation,
ppm
140
125
12S
Tost 1
E007
BOOS
B009
B010
BOH
C012
C013
C014
C016
C017
C018
C019
C024
C02S
C026
E048
EOSO
EOS1
EOS2
E05S
F06S
F076
H084
H099
K138
Id 39
HI 76
Ml 84
Pumace
Outlet
875
991
999
1209
1123
903
908
832
-
713
833
769
721
752
ppm NOx Concentrations
Partioilate Outlet
890
849
771
857
1077
1129
926
844
885
761
763
773
774
575
738
752
680
-A
Pilot Plont Outlet
953
899
782
1124
1055
926'
882
690
924
753
769
698
864
696
617
777
807
760
558
701
618
777
763
TABLE 4.13 N02 INJECTION TEST
RESULTS
Test
No.
K-138
K-139
NOx Concentration, ppra
HSr Part. Scrub. Pilot Plant Sulfatc Cone.
Exit Exit Exit CM Moles/Liter
657
740
1609*
1692*
1036
999
0.203
0.157
Ml 86
M9S
M202
LI 69
L157
L148
D037
DO 39
D040
769
727
746
752
648
758
694
723
982
965
866
852
739
759
663
706
754
693
824
764
794
876
706
755
*This NOX concentration is based upon an estimated N02 injection rate as measured
on an uncalibrated rotameter. The estimated N02 injection via the rotameter was
equivalent to 952 ppm. Based upon the sulfate formation increase however, the
estimated NOo concentration would be approximately equivalent to 590 ppm NO-.
4-18
-------
TABLE 4.14 SULFUR TRIOXIDE CONCENTRATION AT FURNACE EXIT
2/24/70
Test I
C-020
C-022
C-023
C-027
D-031
S03-l
S03-2
S03-3
S03-4
SOj-S
SOj-6
S03-7
S03-8
S03-9
903-10
S0,-ll
SO,-12
15.30
3.85
3.43
0.94
1.17
1.41
2.83
1.23
6.42
6.26
4.50
3.86
2.88
3.19
29.6
9.6
45.2
86.5
17.1
52.2
6.4
62.0
144.0
96.3
2.8
2.9
2.5
5.0
5.0
5.0
2.5
2.5
2.2
2.3
2.4
2.2
Chen. Lab. Method, not N2 purged
Chcm. Lab. Method, purged
Chera. Lab. Method, not purged
Chem. Lab. Method, purged
Chem. Lob. Method, purged
Chen. Lab. Method, purged
Chen. Lob. Method, not N? purged
Chan. Lab. Method, purged
Chen. Lab. Method, purged, acidified with IC1 I
Chem. Lab. Method, purged, acidified with HC1 '
Chem. Lab. Method, purged !
Chem. Lab. Method, purged :
Chen. Lab. Method, purged '
Chem. Lab. Method, purged
Goksoyr and Ross Method
3/5/70
Test >
F-1SS
F-160
F-.168
F-177
• 1
12
13
14
IS
16
F078
11088
H100
L141
LI 49
LI 59
L161
L169
L172
Ml 78
Ml 84
raea
M194
2.89
5.00
9.30
22.0
2.31
5.88
6.79
6.40
5.04
21.0
26.6
32.0
22.2
18.5
6.6
C.O
8.7
8.4
10.5
11.4
6.3
16.5
Cocmcnts
Golcsoyr and Ross Method
2.0
6.2
6.0
5.2
4.0
2.0
2.8
2.0
4.0
4.4
S.2
5.8
1.1
TABLE 4.15
EFFECT OF BED HEIGHT
ON DRY PRESSURE DROP
ACROSS FBA
Height of
Bottom Bed,
inches
Height of
Top Bed,
inches
6.5
3.5
0
7.5
3.5
0
' Air Density - 0.074 Ib/ft
Air Velocity - 12.5 ft/sec
4.15
3.40
2.95
TABLE 4.16 FBA PACKING ATTRITION
AFTER 270 HOURS
Average Weight,
Standard Deviation,
Top Tray
Bottom Tray
4.24
4.60
0.16
0.12
16.5
9.1
*A description of these various methods is given in Sections A. 2 and A.2.6 of
Appendix A.
4-19
-------
5.0 DISCUSSION OF RESULTS
5.1 DATA ANALYSIS
The results reported in Section 4.0 included all of the most pertinent
findings from this project. For the sake of completeness, however, all of the
recorded data and computed results are included in an addendum. The addendum
includes the method of calculation of each item, a table of primary results,
and the computer printout from each test. Examples of the computer sheets are
illustrated in Figure 5.1.
FIGURE 5.1 COMPUTER RESULTS
RU« NUMBER
FLUE CAS OATI
DATE *-l-JO
PRIniRr AIR
SECONDARY MR
FUSNACF E1IT
FLTAIH HUH. OEM D«T STAT
CR/DICF f/< POINT FLOu "RES
F I/MR IN.H20
0.00* *6.* 496. If.6
0.00* ST.* *BS8. I0.«
I 9117, 9,4
ABS CVC.EIM
bO.B
-4.9
FURNACE PERrOR«ANCt
I FUtL »i COAL
REC.(FLOW NO*.1 *)6.)
1CRUBAER PtRFOKHANCt PART. VtHI *fll. FLOAT,RED
IULFATE FOHHATION PM1
cowc.. CM-HOLE/I- o-
RUM NUMBER (-Hi Off 4-l-ro
TtRlAL BALANCE!
!»E Of 0*V 01*0
NPUT FOR EO. - KIN.PROG.'
•UBtOIDIISOl.1
NOKQISULIOI
"ISUIFITI
CARHUN O.TII
HVDIOGEN 0.0*8
OIT*NII O.OB
SULFUR O.OIT
*1H 0.07
HATER 0.0*)
P KCO CO«COilllON
•CD PRESLAItO. NO • O, YES
•CIO ITRENCIH,
70.16*
0.116
0.40)
o.ua
"ARTICULATt 1CRUBBER PRODUCT
.I/VOLUME SOLUTION
5.2 ERROR ANALYSIS
The results reported here and in
the addendum are of limited value un-
less their accuracy is determined. To
accomplish a measure of confidence in
these results, several methods were
used to evaluate potential errors.
This error analysis is included in
Appendix C. The analysis shows that
none of the primary measurements except
possibly the sulfate concentrations
were seriously in error. The reported
sulfates might be as much as 60% or so
higher than the actual values. None of
the other potential errors reflect
significantly upon the reported results
and conclusions.
5.3 SO., ABSORPTION
5.3.1 Comparison Between Absorbers
Although both the venturi absorber
and floating bed absorber operated
under similar conditions, their re-
sponses were in some respects substan-
tially different. For example,
5-1
-------
significant differences between these two absorbers were noted in the effect
of L/G. The floating bed absorber was quite sensitive to L/G, but the venturi
absorber performance was almost completely uneffected by that parameter. Theo-
retical rational for this effect will be presented later.
Probably the most obvious difference between scrubbers was the level of
SO- absorption. This, however, is countered in part by the difference in
operating energy level between these two mass transfer devices. The maximum
operating proficiency for both absorbers is summarized as follows:
Total Energy,*
HP/MCFM % Absorption
Floating bed absorber 1.84 99.2
Floating bed absorber 0.80 88.0
Venturi absorber 2.34 98.0
Ventifr-i absorber 0.87 80.0
This data is based upon the following:
Electrical efficiency of both pump and fan, 80%
Liquid-to-gas ratio, 4#/#
Pump pressure, 25 psig
Another apparent difference between the venturi and FBA was the lesser
sensitivity of the venturi to chemical composition of the spray slurry. This
difference arises not from any basic difference between these two devices but
simply because the FBA was operated at such a high level that the equilibrium
SO,, vapor pressure became a significant factor in limiting further absorption.
Note in Figure 4.2 for an L/G of 0.3 #/# that composition did not strongly affect
SCL absorption in the FBA. Thus, one would expect the venturi absorber operating
at high absorption efficiencies (>90%) to have a similar dependence upon slurry
composition.
5.5.2 Mathematical Models of SO^ Absorption
A theoretical analysis of the gas absorbers would be valuable for correlating
the collected data and for delineating the observed trends. However, no theoretical
*Note: As a point of reference, the electrical energy produced from pulverized
coal in a 40% efficient cycle is 3140 HP/MCFM.
5-2
-------
treatment of anything analogous to the FBA could be found in the recent
literature. A theoretical treatment of a sieve tray without downcomers could
have some application here, but this possibility was not explored. A theo-
retical treatment of the venturi absorber *• ' was found and considered in this
regard. However, upon examination, the paper was found to have many fundamental
errors that made it useless.
Since no adequate theoretical treatments are available, and since a
detailed mathematical model development is beyond the scope of this work, some
very simple models of both the FBA and venturi absorber were developed. The
basic material balance is as follows:
-d(Gy) = Kga (y-y*)dVs (5-1)
where G = total gas mass flow, # mole/Hr
y = mole fraction S07
2
K = mass transfer coefficient, # mole/Hr ft
g 23
a = interfacial mass transfer surface area, ft surface/ft space
y* = equilibrium concentration of SCL over liquid = P'*/Pt
dV = differential element of space
Tacitly assumed in this equation is that the system is gas film controlling.
The two following assumptions greatly simplify Equation 5-1.
1. Since SCL is a minor constituent, G is nearly constant.
2. Since the stoichimetric ratio of liquid to absorbed gas in
the scrubber is so large, the chemical composition can be
assumed to be constant so that y* is constant.
Equation 5-1 can now be simplified as follows:
-dy = V
(y-y*)
But, a dV = Total effective interfacial surface area.
If a' is defined as the interfacial area per unit mass of liquid on the tray,
and if a' is assumed constant, then,
5-3
-------
Therefore,
adV = a'H where H = total mass of liquid on the tray
G
(5-3)
In
- y
= K a1 H
8 G
(5-4)
Equation 5-4 is the basis of Equation 4-1, which was used to correlate the
FRA SO- absorption data. It was necessary, however, to modify Equation 5-4
so that the unknowns y* and H could be expressed in terms of known parameters,
This was done as follows:
For a constant gas flow, as was the case here, the holdup
was equated to L/G as,
]i = Cx (L/G) + C2 (5-5)
Since the vapor pressure of SCL over the liquor was unknown,
it was necessary to postulate the vapor pressure relationship
to composition in a manner similar to Johnsfone's^ .
M
(5-6)
where B = magnesium bisulfite concentration, as gm SCL/lOO ml
M = dissolved magnesium sulfite cone., as gm SO^/lOO ml
However, since M was nearly constant for this test condition,
the vapor pressure equation was simplified:
y
* =
(5-7)
Inserting these equations into Equation 5-4 and combining the
constants:
r~* Tt ™
In
where C. = kga'C
C5 = kga'C2
- C3B
= C4 (L/G) + C5 (5-8)
5-4
-------
As reported in Section 4.2.2, this equation was solved by
nonlinear regression analysis with the following results:
C3 = 24.97
C4 = 0.633
C5= 2.26
and the Correlation Coefficient (R2) = 97.451.
An even simpler model that is employed in the computer program is based
upon the total volume of the active section of the FBA and assumes that the
vapor pressure is negligible.
In
K a
g
K a
dh =
AAh
(5-9)
or
K a = -In (1-E)
o
AAh
(5-10)
where A = flow area = 1.97 ft
Ah = height of active section = 4 ft/section x 2 sections = 8 ft
Although simple, this equation is of limited value since it does not
really delineate any important effects and ignores vapor pressure in an
operating region where vapor pressure evidently has a very significant effect.
The same simplifying assumptions applicable to Equation (5-1) for the
FBA are applicable to the venturi absorber situation so that,
-dy
K a
E
dV
y-y*
(5-3)
If a is constant and independent of position and if the absorption is completed
within the venturi, then,
V7 -\7*fe
In
y
7
-*
(5-11)
where V = total volume of the spray zone in the venturi, ft .
5-5
-------
If the vapor pressure is ignored, then Equation 5-11 reduces to the form
utilized in the computer program and reported in the printout.
K a - "G ln C1^
g Vs (5-12)
Since vapor pressure did not have a great effect upon SO- absorption in the
venturi absorber at E less than 80%, this equation has somewhat more credence
than Equation 5-10 for the FBA.
5.5.3 SO 2 Vapor Pressure from Literature
From the above discussion, it is apparent that reliable vapor pressure
data for the MgO-SO--H-0 system is needed. Without this data, NT, the number
of transfer units, cannot be determined. Therefore a meaningful K cannot be
o
calculated.
Some SC>2 vapor pressure data exists in the literature for the MgO-SO~-
H-0 system *- '' *• '' *• '. However, this data covers a range of compositions
i*
and a vapor pressure range that are not applicable here. Data presented by
f 81
Markant l J covers the lowest vapor pressure available from the literature.
This range and the present operating range are summarized as follows:
Vapor Pressure,"Hg Monosulfite Bisulfite gm SO^/lOOml
Markant 1.0 to 15 0.2 to 0.6 3.8 to 6.0
Pleasant work 0.01 to 0.4 0.8 to 1.2 0 to 3.5
The impetus behind the available data in the literature was technical
support for the magnesium base pulping process. This differs from the present
system in that absorption of SO- into the bisulfite acid is at a much higher
L*
concentration. Vapor pressure data is available for other sulfite-bisulfite
systems, notably containing sodium, calcium, and ammonia*- •*' *• *. However,
although the form of the correlation is generally similar to that employed in
Equation 5-6, the proportionality constants differ substantially. Thus, it
would appear that development of vapor pressure data in the range at hand
would be in order.
5.3.4 Effect of Sulfates Upon SO., Absorption
The magnitude of the adverse effect created by sulfates upon SO- absorption
in both absorbers was unexpected. However, a paper by Chertkov1- ' dealing with
5-6
-------
the effects of sulfates upon the SCL vapor pressure over solutions of ammonium
£
sulfite and bisulfite seems to present a plausible explanation for this observed
trend. In the ammonia base system, Chertkov found the vapor pressure to change
as follows:
' C+S *
P* = P
Pso2 Pso2
(5-13)
where P*™ = actual vapor pressure
PQn = vapor pressure over pure sulfite and bisulfite solution
2
C = molar concentration of sulfite and bisulfite
S = molar concentration of sulfate
Thus, if this relationship held for the magnesia base system, the effect of
sulfate upon the vapor pressure would be quite substantial. Under normal
conditions at a pH = 5.0, C = 2.6 gm S02/100 ml, and S = 0.25 gm S02/100 ml,
then:
P* = P
Pso Pso
2
2.6 + 0.25 I , nnr r> n ro •
= 1.095 P™ or 9.51 increase
2.6 i
-------
2. Although pressure drop across FIGURE 5.2 COMPARISON OF S02 ABSORP-
the scrubber could not be com-
pared directly with that in the
FBA, analogous trends were noted.
3. Dust collection, although at a
comparable level, showed a de-
pendence upon the liquid rate.
This was not observed in the
FBA tests.
The only available data in the litera-
ture for S07 absorption in a venturi ab-
f8"l
sorber is that by Markant k Jfor a magnesium
sulfite and bisulfite system. This work
was performed on a small pilot venturi. The
paper deals with the effect of chemical com-
position upon S02 absorption. Some overlap of the composition range in that
paper and the present work shows that the S02 absorption compares reasonably well.
This comparison is illustrated in Figure 5.2. However, the Markant work does re-
veal that the absorption depends upon the concentration of magnesium sulfite in
solution. This is a factor which could not be explored in the present work since
the system was saturated with sulfite at constant temperature and was therefore
essentially constant.
Absorber ,Spr»y ji!
5.4 CHEMISTRY OF ABSORBING SLURRY
The chemistry of the magnesia base system played an important role in the
absorption of S02 and had a strong influence upon the tendency to form deposits,
The overall reaction for this system is simply:
MgO + S0
H20
MgS03'6H20
(5-14)
The solid phase of the product permits convenient separation from the process
for subsequent removal to a regeneration facility to reconvert to MgO and SO ,
From a mechanistic standpoint, the system can be summarized as follows:
5-8
-------
Slaking Reactions:
++
>• Mg + 20H (5-15)
Mg"1"1" + 2QH'. Mg(OH)- (5-16)
*
S02 Absorption:
S00 > SO- or H-SO_ (5-17)
Z(aq) L 3
CA 4- OH »- H SO^ + H20 (5-19)
++ H20
S07 + Mg — > MgS07 • 6H-0, , (5-20)
•J •* -J £ (s)
The composition of the absorbing slurry was controlled primarily by the
rate of MgO makeup. Three classes of composition of the spray slurry were
possible. These are listed as follows:
Case I - Excess bisulfite — complete neutralization
Composition: Mg"1"1", HSO^, S0~, MgS03 • 6H20
Case II - Excess MgO [or Mg(OH)2] — complete neutralization
Composition: Mg"1"1", S03, MgS03 • 6H20, OH", MgO [or Mg(OH)2J
Case III - Incomplete neutralization (either excess bisulfite or MgO)
Composition: Mg"1"1", SO" MgSO., • 6H90, MgO [or Mg(OH)7], HSO"
«J «J L, t* O
Case III was probably most prevalent because a small fraction of the MgO
is slow to react even under strongly acidic conditions. On the other hand,
reactions 5-17 through 5-19 are ionic and therefore are probably instantaneous.
The fact that preslaking of the MgO did not have a measurable influence
upon SO- absorption resulted from two factors. First, the sulfite, via reaction
5-19 tends to buffer the solution. Secondly, the MgO used, Michigan Chemical
Number 5, was a fairly reactive magnesia which could hydrate quite rapidly even
when preslaking was not attempted. Magnesium oxide slaking has been investigated
n 21
by Smithson^ , who found the following:
MgO reactivity is related to the crystallite size (a particle being
constructed of an array of crystallites). The crystallite size in
5-9
-------
turn is a function of the history
of the MgO including the chemical
form of the raw material and the
calcining temperature. The optimum
temperature for calcining is between
750 and 1100 F. Above calcining
temperatures of about 2200 F the
resulting MgO is essentially un-
reactive and is referred to as
"dead burned". Finally, Smithson
found that the reaction rate can
be correlated in terms of specific
surface area and that the activation
energy is constant at about 14.1 -
0.2 kcal/gm mole.
The solubility of magnesium sulfite
above a pH of about 6.0 agreed fairly well
with the literature. See Figure 5.3. The
average solubility of magnesium sulfite
was 1.1 gm/100 ml as SO2< Since the
sulfite ion concentration was nearly
constant, a one-to-one correspondence
existed between the magnesium bisulfite
concentration and pH. This is illustrated
in Figure 5.4.
In the calcium oxide and calcium
carbonate absorption systems, the slurry
concentration or, more explicitly, the
specific surface of the limestone is a
very important parameter and is possibly
a controlling factor in SO- absorption.
The magnesia base system shows no corre-
sponding dependence on slurry concentration
because of the relatively high solubility
of the magnesium sulfite. The ratio of
the solubility of magnesium sulfite to
calcium sulfite is 485 mole/mole at 140 F.
FIGURE 5.3 SOLUBILITY-OF MAGNESIUM
SULFITE
FIGURE 5.4 RELATIONSHIP BETWEEN BI-
SULFITE CONCENTRATION AND
SLURRY pH
(frcantntlon. pi/100
5-10
-------
Since the sulfite ion acts as a buffer to maintain a nearly constant
hydroxide ion concentration, the desolution rate of the magnesium hydroxide
becomes less important. The calcium base system has no such buffering effect.
5.5 DEPOSITION
5.5.1 Deposition Mechanisms
Although the study of deposition in the two types of SO- absorber apparatus
was not an objective of this project, the solution to this unexpected problem
became a major concern. It is important, particularly for those who are familiar
with deposition problems in the limestone system, to distinguish between the two
different types of deposits. In limestone scrubbers the primary deposit is a
calcium sulfate hydrate (gypsum), which forms a cement-like deposit in regions
where it is subject to the drying influence of the flue gas. The calcium base
deposit would be classed as an aggregate resulting when a portion of the slurry
is permitted to separate or "hide out" within the system. Continuous wetting,
slurry agitation, and design against hide out are some of the approaches used
to combat the problem.
In contrast, the magnesium base de-
posit has been identified as a sulfite
hydrate, MgSO^'GHJD, which forms a hard,
resilient platelike deposit which bonds
tightly to metal surfaces. The deposit
consists of a matrix of large crystals
that appear under the microscope to be
melted together. An example of one of
the coarser deposits removed from the
venturi absorber sump is shown in Fig-
ure 5.5. Unlike the deposits in the
limestone system, this deposit is not
an aggregate of settled solids from
the slurry, but rather appears to be
kind of crystal growth.
Several factors seemed to in-
fluence the tendency to form deposits.
These included the systematic shifting
FIGURE 5.5
EXAMPLE OF COARSE
DEPOSIT
5-11
-------
of the sulfite slurry concentration, the availability of precipitation sur-
face area, the nature of the surface material, and the fluid flow environment.
These factors are discussed individually below.
Systematic shifting or cycling of the solid magnesium sulfite concentration
can occur by two separate mechanisms. The first involves chemical reaction, and
the second occurs from temperature cycling. The chemical reaction mechanism
proceeds as follows:
1. SO- is absorbed in the gas-liquid contacting zone according
to the following successive reactions:
S02 + OH" + HS03" (5-21)
S03= + H20 -> HS03" + OH" (5-22)
MgS03-6H20 -> Mg++ + S03= (5-23)
The third reaction, which proceeds in order to maintain a constant concen-
tration of sulfite ion, causes a net reduction in the suspended solids concen-
tration.
2. The slurry then proceeds to the sump, where the bisulfite is partially
neutralized by the incoming magnesia.
HS03" + OH" + S03= (5-24)
S03 + Mg •* MgS03-6H20 (5-25) .
This last reaction is the mechanism by which the crystal growth can pro-
ceed. Thus, for each mole of SO- absorbed, two moles of sulfite can precipitate
in the sump. The following facts support this mechanism hypothesis:
1. No deposition formed within the gas-liquid contacting zone. Instead
the solid sulfite was actually dissolving in this region.
2. Deposition was most severe in the sump and piping when the system
was operated on the acid side and when the MgO was added directly
to the sump.
3. The deposition rate was substantially reduced when the makeup location
for the magnesia was changed from the sump to the gas-liquid contacting
zone. To ensure that the magnesia would react readily, the magnesia
was preslaked. By this technique, at least in theory, the total de-
viation of the sulfite slurry concentration could be reduced by 501
5-12
-------
depending upon how readily the available hydroxide would react
with the absorbed SO-.
The second potential mechanism for deposit formation is temperature cycling.
Two temperature effects can occur. The first is the temperature drop between
the inside pipe wall and the bulk liquid. The second is a change in the bulk
slurry temperature due to heating by the flue gas and cooling by the makeup
slurry and makeup water. At 135 F the solubility of magnesium sulfite hydrate
changes by 0.0234 gram/100 ml slurry for each degree change in temperature.
If the "deposition potential" is defined as the maximum deviation in the concen-
tration of MgSO,'6H20, then this expression can be used to describe the likelihood
of deposit formation. Thus, the deposition potential for various situations are
summarized as follows:
Effect Deposition Potential
Chemical reaction
Makeup MgO to sump 1.270 gm/100 ml
Makeup Mg(OH)2 to contact zone 0.635 gm/100 ml
Temperature cycling 0.077 gm/100 ml
Pipe wall temperature drop
Stainless steel 0.0044 gm/100 ml
PVDC 0.0007 gm/100 ml
The deposition potential is not an exact indicator of deposition formation
rate, but is simply a qualitative guide. Other factors influencing deposition
can predominate.
The success of fly ash in preventing deposition formation is believed to
be due in part to the high surface area offered by the fly ash for receiving
the precipitating sulfite. This effect is apparent by comparing the specific
surface of the slurry with and without fly ash. Assume the following typical
conditions:
Sulfite Slurry Fly Ash Sulfite Slurry
Average crystal size = 200 y Average fly ash particle size = 12 u
Suspended solids concentration = 5 gm/100 gm Suspended fly ash concentration =
2.5 gm/100 gm
Specific gravity = 2.8 Specific gravity = 2.34
2 2
.'. Specific surface = 5.35 cm /ml .'. Specific surface = 58.7 on /ml
Since the presence of fly ash increased the suspended solids surface area by a
5-13
-------
factor of 11, it is apparent that the deposition on suspended particles should
be increased relative to the deposition on the stationary or wall surface.
Typical values of deposition ratios of suspended surface to wall surface are
summarized as follows:
Deposition Ratio of Suspended Surface to Wall
Surface,
Without Fly Ash With Fly Ash
Venturi absorber sump 29.8 320
2- inch pipe 6.8 66.5
The additional suspended surface created by the presence of fly ash cannot
by itself explain the observed elimination of deposition. Therefore, some
additional property of fly ash must have contributed to the prevention of de-
position. A probable explanation is the abrasive or scouring action of fly
ash upon the stationary surface. Whatever the mechanism, however, the presence
of fly ash in the slurry proved very beneficial from a deposition standpoint.
Even if the stationary surface constitutes a significant fraction of the
available surface area, the capability of that surface to receive deposits
depends upon the affinity of that surface for the deposits. This affinity is
analogous to the wetability of a surface and involves Vander Wai type forces
between the wall, the liquid median, and the precipitating sulfite. Numerically,
this wetability is expressed in terms of the contact angle between liquid and
solid. As this angle increases between 0 and ir, the wetability decreases. Thus,
the contact angle of water on a newly polished automobile would be large compared
to the contact angle of water on a dull finish. Applying this analogy to the
present situation shows that the PVDC surface has a large contact angle, but
stainless steel and brass have very small contact angles. It should be noted,
however, that once a layer of deposit froms, the contact angle is no longer
between liquid and wall material, but is instead between liquid and deposit.
The value of the contact angle concept is that it offers a simple method
of evaluating alternate materials of construction. Thus, two materials can be
compared by measuring the rise or fall of the liquid in capillary tubes made of
the materials in question.
Finally, a phenomenom of deposition in this magnesium base system, which in
some regards is directly contradictory to expectations, is the observation that
the deposition was highly preferential at points of extreme turbulence or shearing.
5-14
-------
These included the primary recirculation pump, the spin vanes of the spray
nozzles, the sump recirculation pumps (used on two occasions), and other
locations subject to high shear stresses. In contrast, deposits did not form
in the recirculation flow nozzle nor in the spray orifice (immediately following
the spin vanes). These are locations of relative low stress. The suggested
mechanism for deposition is some type of shift in the solubility of the sulfite
at localized points of intense eddy circulation, perhaps vapor flashing and
condensing, which could cause temperature cycling.
5.5.2 Method of Deposit Elimination
The relatively high solubility of the magnesium sulfite deposit can be
used to advantage for cleaning a fly-ash-free absorption system. The pilot
plant was cleaned (descaled) on several occasions by simply stopping the mag-
nesia makeup to the system. The remainder of the absorption system operated
normally. The sulfite deposits dissolved readily according to Reactions 5-21
through 5-23. The effect was that for each mole of SO- absorbed, one mole
of the sulfite solid was dissolved. This method could be adopted for cleaning
a commercial absorption system. For the method to be effective, however, the
cleaning cycle must be started before the absorption system is completely plugged.
Although the SCL absorption drops off during the cleaning cycle, the decrease
is gradual and would probably level off at about 501 absorption. The product
from the absorption system could be diverted to a holding tank during the
cleaning cycle and could thereafter be funneled back to the system during normal
operation.
5.6 SULFATE FORMATION
Sulfate formation was one of the more important factors considered in
this project. Since the sulfate concentration leaving the scrubber can vary
widely, depending upon both the sulfate formation rate and the throughput
rate, the sulfation tendency of the system should be expressed so that both
scrubbers can be compared on common grounds. Three possibilities are as follows:
1. Mole fraction of incoming sulfur leaving product liquid as sulfate.
2. Mole fraction of absorbed sulfur leaving product liquid as sulfate.
If the liquid passes through the scrubber on a once-through basis, then the mole
fraction of sulfur as sulfate in the product liquid is equivalent to (2) above.
3. Overall mass transfer coefficient for 0~ absorption.
5-15
-------
The factors which could contribute to sulfate formation in the SO- absorber
are as follows:
1. Absorption of CL followed by chemical reaction with SO-~.
2. Absorption of SCL.
3. Sulfates in makeup water.
4. Dissolved oxygen in makeup water.
5. NO- absorption and chemical reaction with SO., .
6. Carryover of sulfates from particulate scrubber.
7. Carryover of catalytic cations from particulate scrubber or
from fly ash.
The maximum probable contributions from these factors can be summarized as follows:
Mole % Absorbed Sulfur Converted
to Sulfate
SO, absorption 0.07501
NO absorption 0.0250
X
0- in makeup water 0.0825
Carryover from particulate scrubber 0.1090
Sulfate in makeup water 0.000
TOTAL 0.2915%
Measured sulfate levels = -^4.300%
Therefore, from a purely stoichiometric standpoint, (1) appears to be the most
likely mechanism. If this is the case, then the question which arises next
is whether it is liquid phase or reaction rate controlling. If it is liquid
phase controlling, then the following parameters should be important:
1. Mass transfer surface area
2. Mass transfer contact time (between gaseous 0- and liquid)
3. Liquid and gas turbulence
4. 0- concentration in gas
Temperature might also have a second-order effect. If chemical reaction rate
were the limiting factor, then the following might be significant:
1. Sulfite ion concentration
2. Bisulfite concentration or pH
3. Light intensity
4. Catalysts such as multivalent cations
5-16
-------
It is difficult to ascertain whether the diffusion step is limiting for
Q~. However, the same factors which affect 0- diffusion will also affect SO-
diffusion. Since the S02 absorption is measureable, and since SO- appears
to be diffusion controlling, it is likely that, if 0- diffusion is the limit-
ing step, the sulfate formation will be proportional to the S02 absorption.
In neither the venturi absorber nor the floating bed absorber does there
seem to be any relationship between sulfate levels and SO- absorption.
From the above, it appears that the sulfate formation in this system must
be predominantly (but not entirely) reaction rate controlling. The sulfite
ion concentration is essentially constant, and no relationship exists between
bisulfite concentrations and sulfate levels; the only remaining factors are
variable and undetected catalytic effects or analytical errors. The analytical
errors were checked on selected samples for sulfates, and these results are
shown in Section C. 3.2. In support of this hypothesis is the large increase
in sulfates which occurred when the SO- absorbers operated as simultaneous
particulate collectors. The apparent increase in sulfates was substantially
greater than that accounted for by the increase in SO, gas going to the scrubber.
The literature was of little value in delineating the observed effects.
The literature does agree on the mechanism of diffusion of oxygen across the
liquid film followed by catalytic reaction with the sulfite*-13)' '14^' *-15-).
However, the detailed effects of oxygen partial pressure, slurry pH, sulfite
concentration, catalyst concentration, etc. are relatively inconsistent. For
example, Gunn*- ' claimed a zero-order effect of oxygen concentration on reaction
("14")
rate, but Linek *• J found oxygen to have a first-order effect and even a second-
order effect. The effect of catalyst concentration was found by Linek to be
linear, but Ghertkov*- •* found no effect from the catalyst concentration. Chertkov
also found a sixth-order effect of bisulfite concentration.
Although the effects of slurry pH, bisulfite concentration, and temperature
upon sulfate formation rate were examined, no trends could be found. Oxygen
concentration and fly ash were the only parameters isolated.
To discern whether oxygen diffusion or catalytic oxidation might be con-
trolling, the data was analyzed in the following fashion. It was rationalized
that if diffusion were the controlling mechanism and other factors could be
ignored, the oxygen absorption should have been proportional to the sulfur di-
oxide absorption (also diffusion controlling). Therefore, the sulfate formation
rate should have been proportional to SO- absorption. As a check, the percentage
5-17
-------
of sulfate was plotted against the SO- mass transfer rate for both the FBA
and the venturi absorber. No trend was apparent. Therefore, it was concluded
that oxygen diffusion was probably not the controlling mechanism.
Since the many possible interactions which could have an influence upon
the reaction rate of sulfite to sulfate were not determined in any systematic
experimental fashion, it is difficult to ascertain the magnitude or importance
of these effects. As was stated previously, the presence of bulk quantities
of fly ash was the only discernible parameter. If it is assumed that the catalytic
quality of fly ash is derived from the iron (Fe ) in the ash (an unsubstantial
hypothesis), the probable concentration levels of iron in the sulfite slurry
can be summarized as follows:
++ -4
Maxijnum probable Fe concentration with prescrubber = 0.02 x 10 to
0.1 x 10"4 moles/liter
++ -4
Maximum probable Fe concentration without prescrubber = 60 x 10 to
300 x 10"4 moles/liter
Although this data does show the positive effect of fly ash, it is insufficient
to yield any definite correlation.
The need for an exact definition of the sulfate formation rate depends upon
the consequences of that formation. If the sulfate formation constitutes an
ultimate loss from the system, then it is important to be able to describe and
limit this formation. However, if the sulfate is reclaimed in the regeneration
process, then this parameter is less significant. In any event, the sulfate
formation rate appears to be less than that reported by Stone and Webster^ •*
for the venturi and comparable to that reported by Stone and Webster for the FBA.
5.7 N0x ABSORPTION
One of the most significant potential attributes for any SO- removal process
is its capacity for simultaneous S02 - N02 reduction. Such was the purpose be-
hind the NO feasibility test performed on April 9, 1970. This test entailed
the injection of nitrogen dioxide (N02) into the flue gas stream ahead of the
venturi absorber. A schematic of the apparatus appears in Figure 3.1.
As was obvious from the results, as reported in Section 4.9, the above
approach was not compatible with the process at hand. Basic to the present
MgO process is the low inception of sulfates within the scrubbing system. In
contrast, the injection of NO- in the presence of SO^ and subsequent absorption
5-18
-------
into a sulfite solution or slurry is the principle upon which the Tyco SO- -
H-SO. absorption process depends. That process was investiga'ted under HEW
Contract PH86-68-75. The Tyco process relies upon the oxidation of the S09
Lt
by either of the following reactions:
Gas Phase
N02+S02
S03 + H20
Liquid Phase
N00 + SO/
NO + S0
NO + SO,
(5-26)
(5-27)
(5-28)
FIGURE 5.6 EQUILIBRIUM CONSTANTS
FOR NOx-
SYSTEMS
The fact that the sulfate level was significantly elevated during the current
feasibility test supports the validity of these reaction mechanisms. The
equilibrium constant of reaction (5-26) among others is shown in Figure 5.6
as a function of temperature. Note that the sulfate found in the product liquor
accounted for a substantial portion of the NO- added to the gas stream.
In retrospect it appears obvious that, to accomplish significant NO
removal by wet scrubbing with MgO via the NO- injection method, the process
must entail a two-step operation. The SO- must first be reduced in a scrubber
to a level substantially lower than
the NO concentration leaving the
A.
furnace. NO- could then be injected
into the flue gas so that it could
react with NO and residual SO- and
be absorbed in an Mg(OH)- slurry
tower. Such a potential process is
shown in Figure 5.7. A method
similar to this approach was suggested
C171
by W. Bartokv ' for a calcium-base
system. However, no process limits
or requirements were specified,
and no provisions for desulfating
the product from the NO absorber
were suggested by Bartok. A
material balance around the absorber
indicates, however, that in order
t,
jf
/I
^ i
*. a.
® BOj • H^O •
® NOj • SOj - N
® SOj • HjO * II
0.003
1/T. 't
5-19
-------
to have a self-sustaining supply of NCL
to react with all of the NO leaving the
boiler, the absorber must be at least
851 efficient, the nitration step must
be essentially complete, and the NO
oxidation must be at least 951 complete.
Little is known of the requirements
of the nitration step, but the NO oxi-
dation should be quite simple.
The concept of injection of NO-
into an NO- laden gas for the sub-
sequent absorption of both species
has been proposed by several inves-
tigators^' <18>' <19>. The
bulkrgas-phase reactions most often
assumed are as follows:
+ «° 2»
FIGURE 5.7 S02-NOX ABSORPTION
PROCESS
• btun to either Wj or SDj
2(g)
H2°(g) t »°2Cg)
N2°4(g) + H2°(g)
2N0
S0: • le» than 100 pin
(5-29)
(5-30)
(5-31)
(5-32)
N°
(g)
(5-33)
The first reaction, 5-29, is the basis for the N0? injection concept. The
(20)
kinetics of this reaction were studied by Waynev , who expressed the rate
as follows:
d[HN02]
dt
= k[NO][N02][H20]
(5-34)
where k = 7.3 x 104 atm^sec"1 at T = 298 K
Assuming that [NO] = [N02] = 10'3 and [H20] = 0.15 and that there is no
change at 150 F, which is the temperature at the point of injection, then the
forward reaction would yield about 90% conversion during a 1-second reaction time.
5-20
-------
However, because of chemical equilibrium effects (see Figure 5.7), the reverse
reaction would predominate and limit the bulk-gas-phase reaction to less than
101 conversion. Therefore, this bulk-gas-phase reaction would not play a sig-
nificant role in the absorption of NO . However, at the gas-liquid interface,
A
where the product of the forward reaction could be continuously removed from
the gas film, reaction 5-29 could predominate. If this were the limiting
mechanism, then as a very rough estimate a gas contact time on the order of
1 second would be required for substantial NO removal in a scheme such as
.X.
that shown in Figure 5.6.
The effect of reactions 5-30 and 5-31 upon the total NO reduction is
important. If these reactions were to occur more rapidly and completely than
reaction 5-29, reaction 5-29 would not proceed. Thermodynamics answers the
first portion of this concern by showing that the fractional conversion assuming
an NO- concentration of 1000 ppm is less than 1% at 150 F. However, the kinetics
of this reaction are unknown, and the extent to which it might occur in the gas
film when the reaction products can be continuously removed is a potentially
limiting factor. Peters^ ' has shown that this reaction is a significant con-
tribution to NO- absorption in alkali solutions. However, since it has been
CIS)
shown*• J that the mutual presence of both NO and NO- promotes the absorption
of both species, it would appear that reaction 5-29 is probably faster than
reactions 5-30 and 5-31.
The reactions which are believed to predominate in the liquid film of the
NO -SO -MgO system are as follows:
2N02, , + 20H" J N02 + N03 + H20 (5-35)
N.,0.,, , + 20H" ^ NOl + NO" + H.,0 (5-36)
2 4(g) 232 '
S02 + 20H" J SOj + H20 (5-37)
NO" + SOj J NOt + SO^ (5-38)
HNO- + OH" J NO" + H-0 (5-39)
£t Lt £
HNO.. + OH" J NO" + H-O (5-40)
*J «J £
5-21
-------
S°3(g) + H2°(g) * H2S04(aerosol)
W(aerosol) + 20H" * *>l + H2°
The relatively low pH of the magnesia slurry (^8.0) precludes any appreciable
C02 absorption. Reactions 5-35, 5-36, 5-39, and 5-40 are the only liquid film
reactions likely to be less than instantaneous since they involve the absorption
of a gaseous species into a liquid while the others are simple ionic reactions.
In the case of HNO- + HNCL, however, these constituents are almost completely
soluble at the condition under consideration. With the possible exception of
5-35 and 5-36, the above reactions will be driven almost completely to the right.
The requirements of an NO -Mg(OH)9 scrubber are probably as follows:
X LI
1. Reaction 5-29 must occur more rapidly than reactions 5-30, 5-31, 5-35,
and 5-36.
2. The S00/. •* or S07/. •> concentration must be small compared to the
2(g) 3(aq)
NO- concentration.
3. The gas must be in intimate contact with the liquid for a period
in the order of 1 second or more.
The NO scrubbing aspects of this concept could be readily included in the present
J\.
contract by adding a packed tower following the floating bed absorber.
5-22
-------
6.0 CONCLUSIONS
The magnesia base wet scrubbing process can perform adequately for SO
A.
and particulate cleanup of pulverized coal generated flue gas. Acceptable
operating conditions are in all aspects compatible with envisioned regener-
ation processes. In fact, since it is desirable to carry a fly ash load in
the absorbing slurry, the wet scrubbing process potentially could be simplified
to use a single scrubber as a simultaneous particulate and SO- scrubbing device.
Conclusions regarding specific aspects of this scrubbing process are presented
below:
6.1 SO., ABSORPTION
In terms of the evaluating parameter y*» the floating bed absorber was
superior to the venturi absorber. However, neither absorber has been optimized
from this standpoint. See Section 5.3.1.
The performance of both absorbers was either comparable to or superior to
that of similar hardware presently employed for the calcium base system.
Final selection of gas absorber hardware must not be based solely upon
absorption performance but must include other factors such as capital cost,
power requirements, scaleup difficulties, pluggage trends, turndown, replace-
ment requirements of internals, etc.
6.2 SULFATE FORMATION
The rate of sulfate formation was at no time prohibitively excessive as
long as the regeneration process is capable of thermal decomposition of the
sulfate.
The sulfates as reported (based upon the titration method) were probably
about 60% higher than the actual values (based upon gravimetric analysis).
The difference in rates of sulfate formation was not significant between
the floating bed absorber (long residence time) and the venturi absorber (short
residence time).
The major parameters affecting sulfate formation were fly ash concentration
and oxygen concentration.
*Note: Y is defined as —^— *— where P is the power consumption expressed
as HP/MCFM. p r
6-1
-------
6.3 DEPOSITION WITHIN ABSORBER
Deposition of l^gSCL-GhUO was the most serious problem encountered. It
can be prevented by proper operation and design. Deposition is preventable by
permitting a fly ash load in the absorption slurry. If formed, the deposition
can be removed by simple desolution.
6.4 OTHERS
The energy required to collect fly ash at levels comparable to those for
high performing electrostatic precipitators (99+) is commercially practical.
Potential SO., absorption in the gas absorbers is not a significant factor
in sulfate formation.
No significant NO absorption could be credited since the error potential
Jx
in the PDS analysis is larger than the observed changes in NO concentration.
A.
6-2
-------
7.0 RECOMMENDATIONS
7.1 PARTICULATE SCRUBBER OPERATION AND DESIGN
A venturi type scrubber should be used for this application. The design
and operation should be optimized. However, until this is done, the following
operating conditions are recommended for PC fly ash only:
1. Gas velocity at the throat = 150 ft/sec
2. Liquid-to-gas ratio = 15 gal/1000 cu ft.
7.2 S02 ABSORBER OPERATION AND DESIGN
1. The stoichiometric ratio of MgO/S02 should be maintained at 1.0
or slightly greater.
2. The makeup preslaked magnesia should be added directly at the
active gas-liquid contacting zone.
3. Materials in contact with the absorber slurry (except at the
gas-liquid contacting zone) should be selected based upon
their ability to resist these deposits.
4. The design of the sumps, holding tanks, and horizontal piping
provide for supplemental agitation of the rapidly settling
sulfite solids.
7.5 FUTURE WORK
1. A regeneration facility should be added to the present pilot
plant to demonstrate its capability to regenerate magnesia for
reuse and to produce a usable sulfur product. This regeneration
facility should be compatible with wet scrubbing in the following
important aspects:
a. Should be capable of thermally decomposing magnesium
sulfate in addition to the sulfite.
b. Should permit a fly ash load in the sulfite slurry.
c. The regenerated magnesia must be reactive to be
reusable. Therefore, the thermal decomposition
temperature should be minimized.
d. To prevent a disposal problem, the water cycle
should be closed.
7-1
-------
2. The feasibility of employing a single dual-purpose scrubber
in the wet scrubbing process should be investigated. This
would require the following:
a. Means for separating the fly ash solids from the
relatively coarse sulfite crystals.
b. Collecting 99+l of fly ash and 98+l of S02 in a
single scrubber at reasonable energy levels.
3. A laboratory study should be implemented for determining physical
property data such as vapor pressure of SO- over the sulfite slurry,
£
which is needed for complete analysis of absorber performance in
terms of the number of transfer units; settling rate and particle
size distribution of slurry solids, which is needed for proper
specification of separation equipment; and viscosity and specific
gravity data needed for pump sizing.
4. Materials appropriate for the spray zone, sumps, piping, etc. which
are suitable for construction, are abrasive resistant and which do
not exhibit a tendency to receive the sulfite deposit should be
identified and specified. Extrapolation of specifications from
other wet scrubbing systems is not appropriate and is not recommended.
5. The gas absorbers, both the floating bed type and the venturi type,
should be optimized from an absorption-versus-power-consumption
standpoint. This could potentially reduce the capital and oper-
ating costs of the wet scrubbing components substantially. It
would also help clarify the suitability of the proper scrubber type.
6. The feasibility of adopting the magnesia base scrubbing system to
NO absorption through an add-on scrubber should be investigated.
A.
Such a process would entail N09 injection ahead of the NO absorber.
L X
Magnesia slurry would be used as the contacting medium.
7. A system control study from an instrumentation standpoint is
suggested. It is recommended that a control system be simulated
and then installed on the pilot plant to develop and demonstrate
a continuous, controllable system.
Important analysis of flow streams, such as continuous monitoring of slurry
composition, should be developed. For example, there should be a capability of
analyzing the bisulfite-sulfite-magnesia content (possibly by pH only) and the
magnesium sulfate concentration (possibly by specific gravity).
7-2
-------
8.0 ACKNOWLEDGEMENTS
The authors wish to recognize several persons who aided measurably in
the progress and success of this project. W. L. Sage devised the preliminary
design of the pilot plant, including several innovations which later proved
invaluable to the steady operation of the plant. Mr. Sage also directed the
modification of the test furnace, including the feed system to burn, pulverized
coal. E. D. Scott aided Mr. Sage in these activities in addition to helping
with the actual plant installation. J. T. Eunson directed the detailed design
and erection of the plant, including ordering all materials and directing the
necessary engineering drawings. J. P. Rooney devised a unique means of con-
structing,the floating bed absorber, which saved substantially on its erected
cost. J. M. Gidley helped establish the procedures used in this project for
both gas and liquid analysis. C. W. Goddard developed the sulfate titration
method employed here. Operating personnel included E. D. Scott, B. T. Lucaric,
J. W. Vick, J. M. Gidley, and C. G. Bozeka.
jlr Submitted by: frJ '.
W. Downs
Approved by:
H/P. Markant
8-1
-------
REFERENCES
1. J. J. Lingane, Electroanalytical Chemistry, second edition, Interscience
Publishers, Inc., 1968, Chapter ZO.
2. C. E. Rodes, et al, "Performance Characteristics of Instrumental Methods
for Monitoring Sulfur Dioxide, Laboratory Evaluation", JAPCA, Vol. 19,
No. 8, August 1969.
3. H. Goksoyr and K. Ross, "The Determination of Sulfur Trioxide in Flue Gases",
Research Report M211, Thornton Research Center, 1961.
4. M. D. Kuznetsov and V. I. Oratovskii, "Rate of Chemisorption in a Venturi-
type Apparatus", Inter. Chem. Eng., Vol. 2, No. 2, April 1962.
5. H. F. Johnstone, et al, "Recovery of Sulfur Dioxide from Waste Gases",
Ind. § Eng. Chem., Vol. 30, No. 1, January 1938.
6. W. T. Smith and R. B. Parkhurst, "The Solubility of Sulfur Dioxide in Sus-
pensions of Calcium and Magnesium Hydroxides", J. Am. Chem. Soc., Vol. 44,
1922, pp. 1918-1927.
7. F. H. Conrad and D. B. Brice, "Some Equilibrium Relations in the System
Magnesium Oxide - Sulfur Dioxide - Water at Pressures Below Atmospheric",
J. Am. Chem. Soc., Vol. 70, June 1948, pp. 2179-2182.
8. H. P. Markant, et al, Babcock § Wilcox Co., "Absorption Studies, MgO-SCL
Systems", 47th Annual Meeting of Tappi, February 20, 1962.
9. F. H. Conrad and W. L. Beuschlein, "Some Equilibrium Relations in the System
Calcium Oxide-Sulfur Dioxide-Water at Pressures Below Atmospheric", J. Am.
Chem. Soc., Vol. 56, 1934, pp. 2554-2562.
10. B. A. Chertkov and N. S. Dobromyslova, '.The Influence of Traces of Sulfate
on the Partial Pressure of S02 over Ammonium Sulfite-Bisulfite Solutions",,
Translated from Zhurnal Prekladnoi Khemii, Vol. 37, No. 8, pp. 1718-1723,
August 1964.
11. W. A. Pollock, et al, "Sulfur Dioxide and Fly Ash Removal From Coal Burning
Power Plant", Air Eng., September 1967.
12. G. L. Smithson and N. N. Bakhshi, "The Kinetics and Mechanism of the Hydration
of Magnesium Oxide in a Batch Reactor", Am. J. Chem. Eng., Vol. 47,
October 1969.
13. D. J. Gunn, "Absorption and Liquid Phase Oxidation of Sulfur Dioxide",
Trans. Inst. Chem. Eng., Vol. 48, 1970.
14. V. Linek and J. Mayrhoferova, "The Kinetics of Oxidation of Aqueous Sodium
Sulphite Solution", Chem. Eng. Science, Vol. 25, 1970, pp. 787-800.
15. B. A. Chertkov, "General Equation for the Oxidation Rate of Sulfite-Disulfite
Solutions in the Extraction of S02 from Gases", Translated from Zhurnal
Prekladnoi Khemii, Vol. 34, No. 4, pp. 771-776, April 1961.
R-l
-------
REFERENCES (Cont'd).
16. E. G. Lowrance, "Sulfur Dioxide Scrubbers, Stone § Webster-Ionics Process",
Final Report for Contract No. CPA 22-69-80 for NAPCA, January 1970.
17. W. Bartok, et al, "Systems Study of Nitrogen Oxide Control Methods for
Stationary Sources", Esso Research and Engineering Co., Interim Report on
Contract PH 22-68-55, May 1, 1969.
18. S. N. Gang. "Increased Removal of Nitrogen Oxides from Exit Gases by Alka-
line Absorption", Izv. Vysshikh Uchebn. Zavedonii Khim i Khim Tekhnol.
4, 998-1002 (1961).
19. N. M. Zhovorankov, et al, "The Study of Absorption of Nitrogen Oxides by
Alkaline Solution in Columns with Filler Bodies", Khim Prom. 754.
20. L. G. Wayne and D. M. Yost, J. Chem. Phys. 18, 767-768, 1950.
21. M. S. Peters and J. L. Holman, "Vapor and Liquid Phase Reactions between
Nitrogen Dioxide and Water", Ind. § Eng. Chem., Vol. 47, No. 1, December
1955.
22. J. H. Perry, "Chemical Engineer's Handbook", Fourth Edition, McGraw-Hill Book
Company, Inc., 1963, pp. 3-61.
23. S. Badzioch, "Correction for Anisokenetic Sampling of Gas-borne Dust Parti-
cles", J.Inst.Fuels, March 1960.
R-2
-------
APPENDIX A
DETAILS OF ANALYSIS TECHNIQUES
A.I SULFUR DIOXIDE SAMPLING
Sampling of flue gases for sulfur dioxide determination was accomplished by
passing the sample gas through a Barton Electrolytic Titrator, the output of
which is continuously recorded on a strip chart recorder. Operation of the equip-
ment is simple and requires little of an operator's time, yet it has been found
to be very accurate.
A.1.1 Apparatus
The complete sampling train is shown in Figure 3.2 and consists of the
apparatus described in the section below.
(a) Sampling Probe in Furnace Exit Duct
The sampling probe used at the furnace exit consists of a 1-3/4"
stainless steel pipe 9 inches long, one end of which is capped
and fitted with a 1/4" connector. A plug of glass wool is placed ;d
in the end of the sampling probe to filter fly ash.
(b) Sampling Tube in Particulate Cyclone Exit and Pilot Plant Exit
The sampling tube consists of a one-fourth-inch O.D. glass tube
inserted in a two-inch diameter by eight-inch long stainless steel pipe
nipple. This sampling tube is designed to avoid liquid carryover from
being sucked into the small sampling line by providing a relatively
quiescent zone ahead of the sampling tube. The large volume-to-area
ratio in the nipple reduces the possibility of the gas sample being
scrubbed by the slurry.
(c) Dry Trap
A 500 ml Erlenmeyer flask is placed after the sampling tube to
trap any moisture which may condense in the sampling tube.
(d) Gas Pump
A single-acting diaphragm pump draws the gas sample through
about 10 feet of "Polyflow" tubing.
(e) Excess Gas Vent
A tee located in the sample line is fitted with a 1/4-inch needle
valve and a gas trap. Excess gas, pumped by the diaphragm pump, which
does not flow through the Barton Titrator is vented.
A-l
-------
(f) Titrator Gas Pump
A Cole-Farmer peristaltic pump with variable speed control is
used to pump the gas sample through the Barton Titrator. A dry trap
located just ahead of this pump removes any water from the line
before entering the Barton Titrator.
(g) Model 286 Barton Recording Electrolytic Titrator
The Barton Electrolytic Titrator is a coulometric gas analyzer.
The sample gas is passed through an electrolytic solution in which
a triple electrode is ijnmersed. The sulfur dioxide reacts with part
of the reactant in the electrolyte. Simultaneously the electrode
attempts to maintain the concentration of the reactant constant by
passing a current through the electrolyte. The amount of current
used is proportional to the amount of SO- reacted and hence to the
concentration of SO- in the gas stream.
00 Bubble Tube
A plastic bubble tube is used to measure the gas flow rate. This
consists of a 5/8" inside diameter, clear plastic tube with a support
base. The flow rate is determined by the time taken for a bubble to
rise between two marks which defines a volume..
A.1.2 Procedure
(a) Sampling
(1) Connect the sampling train as shown in Figure 3.2 making sure
that all joints are gas-tight. Make sure that the appropriate
sampling line is open and the NO lines pinched off.
J^.
(2) Switch on the diaphragm pump. Next, turn on the variable speed
peristaltic pump to a flow of 250 ml/min sample gas.
(3) Turn the Barton on and run through the necessary blanking and
preliminary start-up procedures. (See Barton Bulletin Gl-20)
(4) By adjusting both the needle valve on the excess gas vent and
the speed on the variable speed pump, maintain a gas flow rate
of about 250 ml/min through the Barton. If the Barton is set
for its highest range and if the recording pen goes off scale,
then the gas flow rate must be reduced. This is permissible
because this analyzer measures the mass flow rate of S0~ to the
&
instrument. However, it is not advisable to increase the
A-2
-------
sampling rate substantially above 250 cc/min since the
reaction efficiency in the cell can deteriorate.
A.1.3 Determination of Sulfur Dioxide
The Barton instruction manual contains the necessary charts, equations,
and conversion factors to convert the cahrt reading into an SCL concentration
expressed as ppm with the following exception: The manual does not indicate
clearly that the correction factor for decreasing the sampling rate is linear.
However, this was checked and found to be true.
FIGURE A.I.
SPINNING SYRINGE
APPARATUS
A.1.4 Spinning Syringe Calibration
The spinning Syringe apparatus used
for calibrating the Barton Electrolytic
Titrators is shown in Figure A.I. A
description of the apparatus and proce-
dures is given below.
(a) Apparatus
1. Glass Hypodermic Syringe. A
50 ml Hamilton syringe with a
polished ground glass plunger
and barrel assembly capable of
spinning freely with minimum
gas leakage was prepared and
used.
2. Spinning Vanes. A plastic vane structure was constructed of 1/16"
thick plexiglas and epoxy glued to the plunger top.
3. Flow Orifice. A piece of 3 MM glass capillary tubing was heated and
drawn to a fine orifice tip for delivery of the sample gas.
4. Spinning Force. A small nozzle made of 1/4" stainless steel tubing
was connected to compressed air which has passed through a glass wool
moisture trap.
5. Ring Stand and Clamps. Standard laboratory clamps and stands were used
to set up the equipment as shown earlier.
6. Sample Pump. A Cole Palmer peristaltic pump with variable speed con-
trol was used to supply the sample gas to the Barton Titrator being
calibrated.
A-3
-------
(b) Calibration Procedure.
1. Check of Syringe Delivery Rate. With the syringe mounted as shown
and with the plunger spinning, time the delivery of 20 ml using a
stop watch. By breaking off small portions of the glass flow ori-
fice, time of delivery is adjusted until ^30 minutes delivery time for
50 ml is attained. After removing the orifice tip and plugging the
syringe, determine the time for the plunger to descent 5 ml while
spinning, this being the plunger-barrel leak rate. The corrected
delivery time as ml/min is then calculated.
2. Formulation of Calibration Mixture. Using a Hamilton gas tight syringe
of the appropriate size, a sample of pure SCL is withdrawn from a
Matheson gas cylinder whose contents had been previously analyzed by
the manufacturer. This SCL is then injected into the 50 ml glass
syringe as the glass plunger is withdrawn to the 50 ml mark, thus
creating a slight negative pressure which insures no sample loss and
also aids in the mixing of the gases as they are diluted within the
glass syringe. The orifice tip is then placed on the syringe making
it ready for use.
3. Calibration. The Barton Titrator is prepared for sampling with a gas
inlet flow of <250 ml/min in a location where ambient air is free of
sulfur compounds. After blank adjustments of the Barton titration
cell have been completed, the syringe is mounted so that delivery of
the calibration gas is made at the inlet to the pump. The calibra-
tion gas sample is run until a stable reading is obtained and the net
reading is then recorded along with range setting and actual gas flow
through the Barton Titrator. Duplicate runs are made to check repeat-
ability of the sample analysis.
(a) VCTn = V,
Calculations.
Data:
Volume of pure SO-, ml
Temperature, °K
Syringe delivery time, Min
P 273°K
= Vs
— T
= t
STP ~ VS A 760" mm
(b)
V
PPM S02 =
STP
10
-3
t x R x 10
-3
Barometric pressure, mm Hg = P
Flow rate through cell, ml/min = R
Example: V = .1 ml
T = 297°K (240C)
t = 30 min
P = 740 mm
R = 250 ml/min
A-4
-------
V
STP
VSTp = .09 ml
PPM S02 =
.09 x 10
-3
9. x 10
-5
30 x 250 x 10
-3
7.5
= 1.2 x 10"5 = 12.0 ppm
S05 COLLECTOR
A.2 SO, SAMPLING APPARATUS 5 PROCEDURE
The collection of SO, is based on a method by H. Goksoyr and K. Ross. The
titration method for determining the amount of SO., collected was developed in-
plant.
The design of the collector, Figure A.2, FIGURE A.2.
is based on the idea of condensing the
sulfur trioxide from the flue gas as sul-
furic acid "at a temperature sufficiently
below the acid dewpoint to leave a
negligible quantity of acid in the vapor
state but high enough to prevent con-
densation of water vapor".
The method is described by Goksoyr
and Ross as follows: "The flue gases
are drawn through a glass coil and a
Grade 4 sintered glass disc, both
enclosed in a water-filled jacket. The
water is maintained at a temperature
(60-90°C) well below the acid dewpoint
but slightly above the water dewpoint.
Practically all the sulfuric acid condenses on the walls of the coil. Any
acid mist particles left in the gas stream are retained by the sintered glass
filter. The acid is then washed out and determined by titration with standard
barium chloride solution."^ ' They also incorporate into their sampling train
a method for S02 determination. This method was not used as we employed a
Barton Electrolytic Titrator for S09 analysis.
LI
As the SO, vapor enters the cool region of the coil, it will condense to
a fine mist. Due to centrifugal action of the gas going through the coil,
these droplets will be thrown to the glass walls. According to Goksoyr and
Ross, "The size of the droplet centrifuged from the gas stream is dependent
upon the distance between the droplet and the collecting wall at the inlet to
A-5
-------
the coil. When this distance is maximum, i.e., equal to the tube diameter, it
has been estimated that all droplets larger than 0.63 micron in radius formed
at the inlet to the coil will move to the wall. When this distance is less
than the tube diameter, droplets smaller in radius than 0.63 micron will also
be forced to the wall of the coil. In addition, some of these smaller parti-
cles will probably agglomerate and be thrown out of the gas stream."^ ' They
have shown that a very large proportion of the acid is collected on the walls
of the coil. Other workers have shown that Grade 4 sintered glass filter is
an effective medium for retaining small particles of acid. Therefore, Goksoyr
and Ross conclude that a collector of their design will collect all the sulfur
trioxide present in a gas sample.
A.2.1 Sampling Apparatus
(a) Sampling Probe
The sampling probe consists of a quartz tube 1/2" inside diameter
by 18 inches long wrapped with a nichrome wire heating element.
This assembly is covered by refractory and sheathed in a stainless
steel pipe, see Figure 3.7. The sheath is fitted with a pipe cap
which enables the probe to be screwed onto a pipe nipple on the side
of the furnace exit duct.
Cb) Solids Filter
A plug of glass wool in the sampling probe acts as a fly ash
filter.
(c) Gas Thermocouple
A spherical glass elbow joint connected to the sampling probe
is tapped with a thermocouple junction. This permits measurement
of gas temperature entering the collector.
(d) Sulfur Trioxide Collector
The collector used in these tests is shown in Figure A.2.
(e) Sulfur Dioxide Absorber
A 1000 ml Erlenmeyer flask filled with about 500 ml of 3%
hydrogen peroxide is used to absorb the S07 in the gas. This absorp-
Lt
tion is to prevent corrosion in the gas meter, a quantitative deter-
mination of the gas absorbed is not made.
(f) Drying Column
The gas is dried before entering the meter and pump by passing
it through a cylinder filled with indicating silica gel.
A-6
-------
(g) Gas Meter
A Sprague gas meter, maximum capacity 175 scfh, was used in
these tests.
(h) Vacuum Pump
Fisher vacuum pump with a free air displacement of 4 cubic feet
per minute at 15 psig.
A.2.2 Analytical Apparatus
(a) Burette
10 ml microburette Grade A
(b) Beakers
Assorted
(c) Evaporating Dish
White porcelain
(d) Light
Daylight fluorescent
(e) Magnetic Stirrer and Teflon-Coated Bar
(f) Pipette
10 ml
(g) Washings Flask
A.2.3 Reagents
(a) Washing solution consisting of 80% isopropyl alcohol, the remainder
being deionized water.
(b) Barium chloride standard solution - dissolve 1.221 grains of BaCl-^H-
in one liter of deionized water that has been adjusted to pH 3.8 to
4.0 with dilute HC1. Standardize against standard sulfate solution.
Cl ml = 0.500 mg SC>4=)
(c) Ammonium hydroxide (1:1) - Mix one volume of ammonium hydroxide
(NH.OH, sp. gr. 0.90) with one volume of deionized water.
(d) Hydrochloric acid (1:99) - Mix one volume of HC1 (sp. gr. 1.19)
with 99 volumes of deionized water.
(e) Thorin - "SulfaVer" (Hach Chemical Co.) or dissolve 0.2 gram of 2
(2-hydroxy-3, 6-disulfo-l-naphthylano) benzene arsonic acid in 100
ml of deionized water.
(f) Sulfate, standard solution (1 ml = 0.100 mg S04=) - Dissolve 0.1479
grams of anhydrous sodium sulfate (Na^SCL) in deionized water and
A-7
-------
dilute to one liter. Standardize by gravimetric determination of
sulfate as BaS04.
(g) Alcohol - isopropyl (2-propanol).
A.2.4 Procedure
(a) Sampling
(1) Make sure that the helical coil and sintered glass filter of
the collector are clean. To clean these parts, draw acetone
then air through the collector. If the filter becomes contami-
nated with fly ash, it may be cleaned with chromic acid. All
traces of acid must then be washed out.
(2) Connect the sampling train as shown in Figure 3.7. Apply stopcock
grease to all joints to preclude the possibility of leaks.
(3) Turn the water bath pump on and adjust hot plate for a water
temperature of 60-90 C.:
(4) Adjust the rheostat on the sampling probe heating element to
give a gas temperature of about 350 F at the collector inlet.
(5) Turn on the vacuum pump and adjust the sampling rate up to 1/2
cubic foot per minute. When the appropriate volume has been
sampled, turn off the pump.
(6) The following data was taken during a test:
a. Initial and final gas meter readings
b. Meter pressure
c. Barometric pressure
d. Ambient temperature
e. Gas temperature entering collector
f. Sulfur dioxide concentration at furnace exit
(b) Determination of Sulfur Trioxide
(1) Disconnect the collector from the sampling train and connect
it to the washings flash.
(2) Using a wash bottle, squirt about 50 ml of washing solution into
the collector and draw it through slowly using the vacuum pump.
(3) Wash out the glass elbow which lies between the collector and
the probe.
(4) Record the volume of the washings and pipette about 10 ml into
the evaporating dish.
A-8
-------
(5) To perform the titration accurately a total of 40 ml of solu-
tion must be in the evaporating dish. Therefore if 10 ml of
washing solution is pipetted into the evaporating dish then 30
ml of 1001 isopropyl alcohol are added, if 5 ml of washing
solution are used then 35 ml of 100% isopropyl alcohol is
added, and so on.
(6) Add two drops of "SulfaVer" (sulfur indicator) . The pH of the
solution in the evaporating dish is then adjusted by first add-
ing 1:1 ammonium hydroxide dropwise until the solution turns a
deep pink. The 1:99 hydrochloric acid is added until the solu-
tion turns a pale yellow. The solution is now ready for titra-
tion.
(7) The magnetic stirrer is turned on and the solution titrated with
barium chloride solution from a microburette . As the end point
is approached, the solution will begin to turn pink. When the
solution stabilizes to a pink color, the end point has been
reached .
A. 2. 5 Calculation
Calculate the SO, concentration by the following equation:
V x V x CR x (0.689)
ppm SCL by volume of dry gas =
x PS
where V~ = total volume of washings, ml
VR = amount of BaCl2 used in the titration, ml
CR = S04~ equivalence of BaCl2, mg S04~/ml BaCl2
VT = volume of washings titrated, ml
V~ = volume of gas sampled, cf of dry flue gas corrected to 0°C and
760 mm Hg
p_ = density of flue gas, #/cf
&
(10 gm/mg) x 10 ppm/mole fraction
0 689 -(96 grc/gro mole) C454 gm mole/# mole)
Iff roole dry gas
30# dry gas
A. 2. 6 Other SO, Methods Used
Before the Goksoyr-Ross Method was adopted, an in-plant method and alter-
nates thereto were used. This method referred to in Table 4.14 as the Chemistry
Laboratory Method (No. 206) involves bubbling the sample through an 80% isopropyl
A- 9
-------
alcohol - 20% water solution followed by a colormetric determination of the
solution using barium chloranilate. A detailed description of the method is
available upon request. The three forms of this method referred to in Table 4.14
are as follows:
1. Chem Lab Method, not N^ purged - No modifications
2. Chem Lab Method, purged - After the sample was collected in the iso-
propanol solution, nitrogen was purged through the sample to strip
any potentially absorbed SO- before it would have time to oxidize
to the sulfate. The purging involved passage of 100 cc/min of bottled
nitrogen for 30-45 minutes
3. Chem Lab Method, purged, acidified with Hcl - This method is similar
to 2 above except that the absorbing solution is modified to include
a 2 ml concentrated HC1 to 100 ml of sample. The hope here was to
further reduce the possibility of SCL absorption and oxidation.
A.3 DUST LOADING APPARATUS AND PROCEDURE
A special probe was designed with the filter placed immediately after the
sampling nozzle. This feature eliminates the problem with some dust probes which
use externally placed filters in which dust settles out along the probe length
before it reaches the filter. Dust laden gas is drawn through the probe and
the volume recorded by a gas meter. The weight of the dust collected is
determined and from this a dust loading value can be obtained.
A.3.1 Apparatus
The assembled dust sampling train is shown in Figure 3.10 and consists of
the apparatus described below.
(a) Sampling Probe
The sampling probe is shown in Figure 3.8. The nozzle is approxi-
mately 2 inches long and is screwed into the probe body. The
Whatman cellulose extraction thimble fits over the back end of the
nozzle. The thimble is held in place by an 0-ring. When the noz-
zle is screwed into the probe, the 0-ring is compressed and makes
a gas-tight seal around the thimble. The dust therefore travels
only a couple of inches into the probe before it is filtered.
(b) On-Off Valve
A ball valve provides for on and off control to the
vacuum pump.
A-10
-------
(c) Drying Column
The gas is dried before entering the gas meter by passing it
through a column of indicating silica gel. This also acts to remove
corrosive SO-.
(d) Manometer
A mercury manometer is used to indicate the gas pressure at
the meter.
(e) Gas Meter
A Sprague 175 scfh gas meter is used to measure the total
volume of gas sampled.
(f) Vacuum Pump
A Fisher vacuum pump capable of delivering 3.2 cubic feet of
air per minute at 15 psig was used to draw the gas sample through
the probe.
(g) Slide Valve Air Lock
Located at the particulate scrubber and pilot plant exit duct
is a 5-inch flanged slide valve on to which the probe is bolted.
The probe is bolted in place with the slide valve closed; then
the slide valve is opened and the probe inserted into the duct.
This arrangement is an air lock which eliminates pressure upsets
in the duct.
A.3.2 Procedure
(a) Sampling
(1) Connect the sampling train as shown in Figure 3.10 making sure
that all joints are gas tight.
(2) With the ball valve in the closed position, open the slide
valve and insert the probe into the duct.
(3) With the bleed valve on the vacuum pump wide open, turn on the
pump. Immediately open the ball valve and adjust the bleed
valve on the vacuum pump for a flow rate of approximately
0.5 cfm.
(4) The volume of gas sampled should be about 5 cubic feet. This
will provide a measurable amount of collected dust.
(b) Analysis
(1) Before the extraction thimble is used, it is numbered for
identification and then weighed to four decimal places. It
A-ll
-------
is then placed in a desiccator for 24 hours and reweighed.
If the reweighing agrees to within 0.0002 grams, the filter is
considered acceptable. Otherwise, it is held in lieu of
reweighing. All filters are kept in individual desiccators
until used. Drying by heating in an oven was found to be com-
pletely unsatisfactory since apparently some of the paper
oxidized.
(2) After the extraction thimble was used, it would be subjected to
the same weighing procedure as above. Care in this procedure
is vital since the mass collected was often less than 1% of
the filter weight. However, experience has shown this proce-
dure to yield highly reproducible results.
A.4 MODIFIED PALMROSE ANALYSIS
The modified Palmrose analysis was used to determine the composition of
of magnesia-base scrubbing slurry except the sulfate and flyash constituents.
Two samples were analyzed; one of the total slurry, and the other of a filtered
sample. A syringe filter was used. See Figure 3.15, The filtered sample pro-
vided a means of determining the composition of the dissolved species which
are of importance from an S02 absorption and vapor pressure standpoint while
the filtered sample provided a means of determining the suspended solids con-
centration and was needed to complete the sulfur and magnesium material balances.
The total concentration of SCL including both SO," and HSO,~ in the solu-
tion is determined by oxidizing the sample with potassium iodate in which the
following reactions control:
+ SKI + 6H+ •* 6K+ + 3H0 + 31(A-l)
3S03 + 31, + 3H,0 -»• 6H+ + 61" + JOW4 (A-2)
3HSOj L 3HSO^
Before the titration is performed, an excess of potassium iodide is added
to the sample. The sample is then acidified with sulfuric acid to a methyl red
end point. This insures that there is an excess of H so that reaction (A-l)
will proceed rapidly during the titration. The sulfuric acid must be added
carefully because the solubility of S0~ is reduced as the pH drops thereby
increasing the chance of stripping SO, from the sample before it can be oxi-
dized to the sulfate.
A-12
-------
During titration, reaction (A-2) occurs instantaneously while reaction
(A-l) is somewhat slower. This titration is performed to a starch end point.
Thus the amount of KIO, consumed is a measure of the total SCU in solution.
The convention adopted for this analysis by the pulp and paper industry is to
report all concentrations as equivalent grams of SCL per 100 gm sample. More
realistically, the specific gravity correction is usually ignored and the
data is actually reported as gms SCL per 100 ml sample. Thus the total SCL
concentration is calculated as follows:
gm Total S02 ml KIOj x Normality KK>3 32 gm SCL. 1
100 ml sample = Volume of Sample x Eq S02 x TO" ^A"3^
The 1/10 enters the equation because Normality is expressed in terms of 1000 ml
whereas the Total is expressed in terms of 100 ml; this is also the case with
the calculation for FREE S02.
As shown previously, the actual oxidizable sulfur constituents of the
absorbing slurry are Mg(HSO,)2 and MgSCL. These can be expanded and written
in yet another way as MgSCL-H-SO, and MgSO,. This acid portion of the magnesium
bisulfite namely H2SO, is referred to as the Free SCL concentration and is
by convention considered to be 1/2 of the molar concentration of HSCL expressed
as SCL.
The Free is a measure of the acidity of the solution, the acidity being
the result of the HSO" ion concentration. It can be easily determined by titra-
tion with NaOH. Recalling that the first titration converted HSO^ to HSO^,
the following simple reaction controls this second titration:
HSO" + OH" + SO^ + H20 (A-4)
When the titration corrected for the amount of acid added to the sample before
the first titration, the Free is calculated as follows:
gm Free S02 1
100 ml sample 2
ml NaOHnet x N. of NaOH 81 gm HSO^ 64 g S02
ml of sample 1 eg HSO- 81 g HSO" 10
(A-5)
The concentration of magnesium in the absorbing solution is easily cal-
culated from the above results since the molar concentration of Mg is equal
to the molar concentration of SOZ plus one half the HSOZ concentration. The
magnesium concentration is therefore calculated as follows:
A-13
-------
Mg gm
100 ml sample
Total gm S02
100 ml
Free gm S02
100 ml
24 gm Mg
64 gm S02
(A-6)
The first term on the right has been adopted as an indirect measure of
the magnesium concentration and is referred to as the Combined SCL.
Combined SO
gm
2 100 ml
Total gm S02 Free gm S02
100 ml 100 ml
(A-7)
A.4.1 Apparatus
(a) 2-1/2 ml Leverlok disposable syringe with 18 gauge needle
(b) 250 ml beaker
(c) 25 ml burettes, 4 used
(d) magnetic stirrers, 3 used
A.4.2 Reagents
(a) Potassium lodate (KIO-j): N/8
(b) Sodium Hydroxide (NaOH): N/8
(c) Sodium Thiosulfate (Na^O,): 31
(d) Sulfuric acid: N/8
(e) KI - Starch solution (100 gm KI + 200 gm Thiodene * 800 gm water)
(f) Methyl red indicator
(g) Deionized water
A.4.3 Procedure
(a) Obtaining and preparing sample
(1) Unfiltered sample - Using the 2-1/2 ml syringe, exactly 2 ml of
sample are drawn. Care is taken here for if excess sample is
taken and if the excess is discarded by drawing the plunger
back to 2 ml, the solids may partially settle and what remains
is no longer representative.
(2) Filtered sample - The sample is drawn into the syringe slowly
to prevent tearing of the filter paper. Excess sample may be
drawn here and the excess discarded.
(3) The sample is then injected into a beaker containing 60 to 75 ml
of demineralized water. The diluted sample is redrawn into the
syringe several times to completely wash or purge the sample.
(4) A few drops (3 to 5) of methyl red indicator are added to the
sample. This will turn the solution yellow.
A-14
-------
(b) Total S02 Titration
(1) 5 ml of starch - KI solution are added to the sample.
(2) Pipette 1/8 normal sulfuric acid into the sample until it
turns slightly red. Do not exceed the end point since this
may contribute to stripping of S02 from the sample. Note the
amount of H-SO. added.
(3) Titrate with 1/8 normal potassium iodate (KICK,). Do not stir
the sample until a blue color starts to appear. Titrate until
one drop produces an intense blue color. Premature stirring
would aggrevate the S02 stripping problem. Note the volume
of KI03 used.
(c) Free S02 Titration
(1) To bring the sample back from the deep blue to a red color, add
a couple drops of sodium thiosulfate. If more than a few drops
are required to effect the color change, the KIO, end point was
exceeded and the entire process should be started over.
(2) Titrate the sample to a bright yellow end point with 1/8 normal
NaOH. Note the volume of hydroxide titrated.
A.4.4 Calculations
If this procedure is followed exactly, the composition is calculated as
follows:
Total S02 = ml KI03 x 0.2 = gm SCyiOO ml
Free S02 = (ml NaOH - ml H2SC>4) x 0.2 = gm SCL/lOO ml
Combined = Total - Free = gm SO-/100 ml
A. 5 DETERMINATION OF SULFATE ION IN Mg-SCyE^O SYSTEM*
A.5.1 Scope and Principle
This method covers a rapid volumetric determination of sulfate ion in
Mg-S0--K,0 slurry systems over the range of 20 to 1000 ppm.
L t*
Sampling is performed in the absence of oxidizing media. Sulfites and
bisulfites are decomposed by treatment with acid under inert gas atmosphere.
Interfering cations are removed by ion exchange and the sulfate ion is titrated
in an alcoholic solution under controlled acid conditions with a standard bar-
ium chloride solution using thorin (SulfaVer) as the indicator.
*Modified version of ASTM:D516-63T, Nonreferee Method B
A-15
-------
A.5.2 Interferences
Cations and anions may cause coprecipitation errors with the barium
sulfate precipitate. Potassium, iron, aluminum, fluoride, and nitrate inter-
fere. Phosphate does not interfere appreciably below 300 ppm. Most metallic
ions form colored complexes with thorin. Cation interference is eliminated by
ion exchange. Cr and Zr may form anion complexes with the sulfate ion.
Fluorides and nitrates cause no serious interference up to 2 and 50 ppm, respec-
tively. Sulfite must be eliminated or determined and subtracted if its pres-
ence is suspected. Chlorides obscure the end point when the sulfate is low.
End point detection is better the higher the sulfate concentration. Sulfides,
if present, must be removed. Chromium present as chromates and dichromates
must be reduced and removed by ion exchange.
A.5.3 Apparatus
(a) Beakers - assorted
(b) Burette - 1-50 ml (Na2S04 solution)
1-5 ml micro, 0.01 ml graduations (BaCl2 solution)
(c) Erlenmeyer flask - 50 cc capacity and rubber stopper fitted with
gas inlet glass tubing reaching the flask bottom. Vertical gas
outlet tubing venting flask.
(d) Evaporating dish, white porcelain
(e) Jones reductor column - 50 cm long x 2.06 cm in diameter. Insert
a plug of glass wool in the bottom of the column and fill approxi-
mately 12 cm high with "HachVer" No. 252 resin (IR-120, 20 to 25
mesh can be used). Regenerate the resin after each sample by passing
(1:4) HC1 through and washing thoroughly with deionized water.
(f) Light"- daylight fluorescent
(g) Magnetic stirrer and teflon coated bar
(h) Pipette - 1 ml volumetric type
2 ml
3 ml
5 ml
10 ml
(i) Volumetric flask - 2-1000 ml (BaCl2 and Na2S04 solutions)
1-50 ml (Sample)
A-16
-------
A.5.4 Reagents
(a) Alcohol - isopropyl C2-propanol). Ethyl alcohol (95%), methyl alco-
hol or "HachVer Nonaqueous Solvent" (Hach Chemical Co.) may be used.
(b) Ammonium hydroxide - reagent grade.
(c) Barium chloride (BaCl2'2H20) - reagent grade.
(d) Hydrochloric acid (HC1) - 37-38%, sp gr 1.188-1.192, reagent grade.
(e) Ion exchange resin - "HachVer", No. 252 or IR-120 cation exchange
resin, 20-25 mesh.
(f) Nitrogen gas
(g) Sodium sulfate - reagent grade
(h) Thorin - "SulfaVer" (Hach Chemical Co.) or 2 (2-hydroxy-3, 6-disulfo-
1-naphthylazo) - benzene arsonic acid.
A.5.5 Solutions
(a) Ammonium hydroxide (1:1) - Mix one volume of ammonium hydroxide
(NH.OH, sp gr 0.90) with one volume of deionized water.
(b) Barium chloride standard solution - dissolve 1.221 grams of BaCl9-2H90
t* Lt
in one liter of deionized water that has been adjusted to pH 3.8 to
4.0 with dilute HC1. Standardize against standard sulfate solution.
(1 ml = 0.500 mg SO/).
(c) Hydrochloric acid (1:4) - Mix one volume of concentrated hydrochloric
acid (HC1, sp gr 1.19) with four volumes of deionized water.
(d) Hydrochloric acid (1:99) - Mix one volume of HC1 (sp gr 1.19) with
99 volumes of deionized water.
(e) Sulfate, standard solution (1 ml = 0.100 mg S04=) - dissolve 0.1479
grams of anhydrous sodium sulfate (Na^SO.) in deionized water and
dilute to one liter. Standardize by gravimetric determination of
sulfate as BaS04,
(f) Thorin - "SulfaVer" (Hach Chemical Co.) or dissolve 0.2 gram of 2
(2-hydroxy-3, 6-disulfo-l-naphthylazo) benzene arsonic acid in 100 ml
of deionized water.
A.5.6 Procedure
(a) Place 2 ml of concentrated HC1 in the 50 ml Erlenmeyer flask.
Place stopper fitted with inlet and outlet glass tubing in flask.
Attach tubing supplying nitrogen gas. Purge flask with continuous
nitrogen flow.
(b) Add 10 ml sample by hypodermic syringe through the outlet tube vent,
see Figure 3.18.
A-17
-------
(c) Place flask on hot plate (with N? flowing) and heat solution to
slow boil until all trace of SCU is absent. (About ten minutes.)
(d) Remove flask from hot plate, wash inlet glass tube with a small
amount of deionized water. Cool.
(e) Add NH.OH (cone.) dropwise until a precipitate appears and HC1 (cone.)
dropwise until the precipitate just dissolves. Cool sample.
(f) Pour sample into ion exchange column. Wash flask with minimum amount
of water. Pass sample through column slowly and collect in 50 ml
volumetric flask. Wash column with water until flask is filled to
mark. Do not allow liquid level to fall below top of resin in the
column.
(g) Shake sample and, using appropriate aliquot, pipette sample into white
porcelain evaporating dish. Place dish on magnetic stirrer. Stir
gently.
(h) Add 40 ml of nonaqueous solvent and 2 drops of thorin indicator.
(i) Adjust the pH to 3.8 to 4.0 by carefully adding NH.OH (1:1) dropwise
until the solution just turns pink. Then add HC1 (1:99) until the
pink disappears and the solution changes to yellow. If NH.OH is
added too fast, it is possible to overrun the color change from yel-
low to pink and the sample stays yellow. It is then impossible to
develop by NH4OH addition.
(j) Titrate the solution with standard BaCl2 solution (1 ml = 0.500
mg SO/") to a stable pink which deepens to a reddish pink on over-
titration. A blank of sulfate-free water is treated in the same
manner. (Allow a time lapse of 3-5 sec between additions of the
last few increments of Bad-.) The color change can best be seen
using constant stirring and a daylight fluorescent light.
SO =
^4 »
gm
ml sample
CRVR
VT
M
vs
where: CR = Conversion factor for titrating reagent, gm SO./ml
VR = Volume of Reagent titrated, ml BaCl2
VT = Volume of sample titrated, ml
Vc = Volume of diluted sample, ml
V = Volume of original sample, ml
A-18
-------
A.6 PARTICULATE SOLIDS APPARATUS AND PROCEDURE
The deteimination of percent solids in the particulate scrubber slurry is
accomplished by simple filtration of a known volume of slurry and subsequent
weighing of the solids. The following sections will describe the actual appara-
tus and procedure.
A.6.1 Apparatus
(a) 100 ml graduated cylinder
(b) Millipore filters (solvent resistant), Solvinert, UL, 1.5 y
(c) Millipore filter holder and filtering flask
(d) Vacuum pump
Ce) Desiccator
A.6.2 Procedure
(a) From the product nozzle of the particulate scrubber, take less than
100 ml of slurry in a graduated cylinder; note the volume of sample.
Be sure to use the entire sample drawn.
(b) Filter the sample through a millipore filter while applying a vacuum
to lieIp filtration.
(c) Towards the end of the filtration wash the solids with a few milli-
liters of methanol or acetone. This helps to remove most of the
remaining surface water.
(d) Remove the filter paper containing the solids from the filter holder
and place it in a desiccator.
(e) Allow the sample to dry over night and then weigh.
A.7 MgO MAKEUP SLURRY TITRATION PROCEDURE
The percent MgO solids in the makeup slurry was determined as described in
the following sections.
A.7.1 Apparatus
(a) Pipette with 1/8-inch O.D. tip calibrated to hold 20 ml
(b) 150 ml beaker, 500 ml beaker
(c) Magnetic stirrer with teflon-coated bar
(d) Burette with 25 ml capacity
A.7.2 Reagents
(a) 7N hydrochloric acid
Cb) Methyl red indicator
A.7.3 Procedure
(a) Using a 500 ml beaker take a sample of the slurry from the holding tank.
A-19
-------
(b) While stirring the slurry to keep a uniform suspension of solids,
draw exactly 20 ml into the pipette. Do not overdraw and drain
excess.
(c) Pipette the sample into a 150 ml beaker.
(d) Add a few drops of methyl red indicator, the solution turns yellow.
(e) Titrate with 7N HC1 until the solution remains a clear red color.
A.7.4 Determination of Percent Solids^
Noting the number of milliliters of acid used, go to the standard curve,
see Figure 3.20, and read the percent solids.
A.8 MAGNESIA HYDRATION ANALYSIS
This analysis was used to determine the extent to which magnesia was con-
verted to the hydroxide. The apparatus and procedure are outlined as follows:
A.8.1 Apparatus
(a) Butner Funnel - 500 cc capacity
(b) Funnel Flash
(c) Filter Paper
(d) Drying Oven
.(e) Muffle Furnace
(f) Crucibles
A.8.2 Procedure
Approximately 300 to 500 ml of magnesia slurry are sampled. The slaking
time and temperature are noted. The sample is immediately filtered and rinsed
with acetone. This removes most but not all of the surface water. It is
probable (but not experimentally substantiated) that no further slaking would
take place. The sample is then set aside for later analysis. The actual analy-
sis consists of drying the sample (about 30 grams) at 105°C for 2 hours. The
sample is then cooled in a desiccator and weighed. The weight is noted. The
sample is then calcined at 400°C for at least 15 hours, cooled, and weighed.
The mole % conversion to the hydroxide is calculated as follows:
-o slaked = 100 X
where X = fractional weight loss
= 1"o~
and m = initial sample weight
nv, = Final sample weight
A-20
-------
APPENDIX B
CORRELATION COEFFICIENTS
The quality of a mathematical expression to a fit of experimental data
is usually expressed in terms of a correlation coefficient and a standard
deviation. As an example, the performance of a particulate scrubber might
correlate as follows:
log (OD) = 1.0295 - 0.09136(AP) + 0.000720(AP)2
Correlation coefficient = 97.0%
Standard deviation = 0.012
where OD = outlet dust loading, grains/DSCF
AP = venturi pressure drop, inches H20
The data is actually treated in the following manner. Set log(OD.) = y.
and AP = x-. This gives an array of N data (y., x.). A function which the
engineer feels will fit the data is then proposed. In this case y = f(x) =
k. + k-x + k.,x . The constants are then determined by any number of mathemati-
cal techniques such as a least squares fit. The correlation coefficient,
2
sometimes referred to as R , is then determined by the following expression:
N
where y = the observed value of y @ x.
y = the calculated values of y @ x^ as found by yc = f(x)
y = the arithmetic mean average of all values of yQ
N = total number of data points
K = total number of constants in f(x)
As such, the above expression is a measure of how much better the correlation
y = f (x) fits the observed data y than does the correlation y = y. Showing
this graphically,
B-l
-------
y
Both terms are squared to eliminate the negatives.
The second correlation factor which can be used to determine confidence
bands around the data is based upon the following expression:
N
E
2
a =
(y0 - ycr
N - k
where a = standard deviation
N = number of data points
k = number of coefficients in the function y = f(x)
Although the meaning of this expression may not be obvious, o is a measure of
the variance or scatter in the data not rationalized by the correlation y = f(x).
It should be noted that if k >_ N, this term becomes meaningless. In other words,
to obtain "true" correlations one should always obtain several more data than
the number of coefficients to be used in f(x).
B-2
-------
APPENDIX C
ERROR ANALYSIS
C.I GENERAL TECHNIQUE
Any experimental program, this one included, is subject to numerous
opportunities for error. The magnitude of the errors can, of course, cloud
the validity of the results being sought. Many kinds of errors are possible
including instrument malfunction, human interpretation, and gross errors in
technique or procedure. It is the experimentalist's task to minimize these
errors by calibration and recalibration of instruments, selection of proper
procedures, diligence in execution, and cross checking of measurements where-
ever possible. Nevertheless, errors will occur and as such it is advantageous
to determine the magnitude of these potential errors so that the validity of
the results can be better ascertained. In addition, this analysis helps
interpret what role errors play in the observed variance of the measurement
in question. The following is a discussion of the "error potential" of sev-
eral of the more important measurements of this program.
The general mathematical approach used to evaluate errors is to adapt the
definition of the exact differential as follows:
dy
6x
dx, +
where: y = f(x,, x~,
and x are the measurements making up the calculated value of y
dx are the measurement errors.
The total possible error is then:
"u
Ay =
AX,
5y_
fix.
AX,
6y_
6x,
Ax,
6y
The maximum probable error is:
Ay' =
6x
AX
6y
AX.. +
6x
n
6y
Ax
'n
(C-2)
6x
n
AX
n
1/2
(C-3)
Adopting either equation, C-2 or C-3, provides a convenient means for computing
the magnitude of error which might arise from individual measurement error.
This technique is used below along with cross check experimental procedures to
assess the role of error upon the results of this test program.
C-l
-------
C.2 GAS SAMPLING ERRORS
C.2.1 S02 Measurements
Errors in the SCL concentration determination can arise from six separate
measurements, errors in the calibration factors, loss of or dilution of SO^ in
the sample line, or malfunction of the Barton Titrator. The contribution of
measurement and calibration errors can be expressed as follows:
(CHART -
|A(TGBART)|
TGBART + 460
BLANK) A '
| A* A (XMWDG)
D (18 +
CELLC z
+ |XMWDG*AA
A*XMWDG)
CONF * '
|AVBT|
b V '
VBT
TBART H
| A (XMWDG) |
XMWDG °
APB **
(C-4)
where: CHART = Recorder output reading
BLANK = Residual level (current leakage from titrator)
CELLC = Cell constant (a correction factor on the calibration)
CONF = Calibration factor, relates current output to mass of
bromine generated
TBART = Time of travel of bubble in bubble tube, sec
TGBART = Temperature of gas in bubble tube, °F
A = Humidity of gas in bubble tube
XMWDG = Molecular weight of dry gas in bubble tube
V^rp = Volume of bubble tube
Dl
PB = Barometric pressure of gas at bubble tube
Terms (2) and (3) relate to the instrument calibration. When the Barton
titrators were purchased, these factors were checked by the spinning syringe
technique described in Section A.1.4. All of the other terms represent measure-
ment errors. Two errors arise from assumptions which we made. First, the
temperature of the gas was assumed to be constant at 70 F and secondly, the
humidity was assumed to be negligible. The errors induced by these assumptions
are represented by terms (5) and (6). Assuming the following variance in the
measured variable as follows:
**Note the terms on the right are numbered for identification purposes.
C-2
-------
A(TBART) = 0.5 sec
A(TGBART) = TGBART - 70
AA = A Ib/lb
A(XMWDG) =0.5 Ib/lb mol
BUB. TUBE^ := °'5 ml
A(PB) = 0.1 "Hg
A(CHART) =1.0 division
A(BLANK) = 1.0 for range = 100
1.0 for range = 30
1.0 for range = 10
2.0 for range = 3
4.0 for range = 1
The maximum probable errors due to measurement variance are depicted in
Figure C.I. The major contributions are from the recording errors due to
blank level uncertainty and the error due to the potential humidity effect.
As a typical example, for the following FIGURE C.I. BARTON S02 ANAL
set of conditions, the individual con- ERROR POTENTIAL
tribution can be summarized:
Range = 3
Blank = 4.0
TGBART = 80°F
A = ASAT = 0.023 Ib/lb
XMWDG = 30 Ib/lb mol
VBUB. TUBE = 10° ml
CHART = 20
PB = 30 "Hg
Error Due To:
CHART and BLANK = 18.75%
A = 3.751
TGBART = 1.90%
TBART = 2.08%
VBUB. TUBE == 0>5%
XMWDG = 1.66%
PB = 0.33%
Max. Probable Error = [Ex.2]1/2 = 19.
The error due to calibration shift was examined experimentally by two
independent means. The high concentration range (>1000 ppm) was examined by
a conventional wet chemical technique described as the Reich Method in
Section 3.4.2. This test was performed through the same sampling apparatus.
C-3
-------
COMPARISON BETWEEN
BARTON AND REICH
METHODS
The results of these tests are illustrated FIGURE C.2.
in Figure C.2 and show the agreement to
be good.
A second procedure used to check the
f i
low range of the Barton was a spinning
syringe technique. This apparatus was i *
adapted to the Barton sampling train
at the pilot plant exit gas sampling
location. This technique was used to
assess the potential effect of the con- { '
densate traps and pumps upon the S02
accuracy at low concentrations.
The spinning syringe technique can
be described as follows:
The syringe feeds S09 gas at a constant rate to the titrator. The barrel
of the syringe is fitted with vanes which are spun by compressed air. The
spinning of the barrel eliminates static friction between the barrel and the
wall which permits uniform gas discharge. By knowing both the SO- gas discharge
rate through a calibrated capillary tube and the air flow to the Barton Titra-
tor, the SO- concentration of the gas can be accurately calculated.
The results of these tests are tabulated in Table C.I as follows:
TABLE C.I COMPARISON OF S09 CONCENTRATIONS BETWEEN BARTON AND
SPINNING SYRINGE
Ma Range ppm Calculated S02 (Barton) ppm Calibrated S02 (Syringe)
2000 3000 4000 5000
SO, Concentration by Reich Method, ppa
.3
1.0
3.0
.3
1.0
3.0
3.78
3.49
3.51
0.22
0.54
0.52
2.26
2.26
2.26
0.57
0.57
0.57
These results show that the Barton response is sufficiently sensitive to
detect SO- in the less than 10 ppm range but that excessive errors are possible
in this low range. The average error at the 2.26 ppm level was +591, and the
average error at the 0.57 ppm level was -25%. It should be noted that this
C-4
-------
is not solely a calibration error but also encompasses the measurement errors
mentioned above.
Another potential interference in the S02 measurement is absorption of
S02 into condensate in the sampling line between the duct and titrator. This
loss can be estimated by assuming that all of the moisture in the flue gas
condenses and becomes saturated with S02. The SO- absorption can be expressed
as follows:
ASO
where: H = SCL solubility in liquid water, gm/gm H_0
A = Humidity of sampled gas, gm H-O/gm D.G.
= SCL concentration in flue gas, ppm
Now for a system between 70 and 90 F, the SCL solubility can be obtained from
6
Perry1 J . With the appropriate units adapted and extrapolating to 10 atm,
the solubility can be correlated as follows:
H = (9 x 10~6)(S09)0-500 (C-6)
A (SCL)
= (4.22)(S02) U'bA (C-7)
the humidity averaged about 0.12 Ib/lb dry gas at the pilot plant exit. There-
fore, condensation and absorption maximum error varied from 16.0% at 10 ppm to
1.6% at 1000 ppm.
In summary, except for gross errors such as air leakage into the sample
train or depletion of hydrogen bromide in the cell, we expect that the SCL
measurements are accurate to within 15% at the 1000 ppm range and to within
50% at the 20 ppm range. Below 20 ppm the "error potential" seems to be
excessive.
C.2.2 SO, Error Analysis
Attempts at measuring SO, concentration leaving the furnace were troubled
with significant variations. Whether these variations were real or the result
of analytical errors is questionable. The likelihood of real variations are
probable. SO, is believed to be formed in the furnace by both homogenous gas-
phase reactions and by catalytic formation with solid constituents. In addition,
C-5
-------
metal surface and ash deposits can have a strong affinity for SO,. It is quite
possible to have variations in the SO., concentration at the furnace exit
even when the flow rates and temperatures seem to be relatively steady.
Measured SO, at the furnace exit varied between 1 and 150 ppm. The degree of
variance that the furnace can produce is difficult to predict. For the pres-
ent, this factor is left unanswered. The potential for SO, variance due to
analytical errors are numerous. The principal sources of error are as follow:
1. Incomplete collection of the acid mist.
2. Loss of mist in the sample probe because of low temperature.
3. Absorption of SO, by the probe ash filter.
4. Absorption of S02 and subsequent oxidation to SO,.
5. Contamination by highly sulfated ash.
6. Catalytic oxidation within the probe.
The first three of the above effects would produce measurements lower than
actual, while the final three would yield higher values. Adoption of the
Goksoyr sampler eliminated (1) and (4) from the list of possibilities. The
potential effects of (2) and (6) are minimized by using a heated quartz tube.
While the temperature of the gas was monitored, the temperature of the probe
surface was not. .
Some direct measurements of SO, concentration leaving the particulate
scrubber were attempted as described in Section 3.5. These measurements were
voided because of the presence of liquid carryover from the cyclone separator
into the sample tube.
The nominal concentration of SO, leaving the furnace ran about 18 ppm.
At that level, SO, from the furnace would not have possessed the potential for
significant effects upon the S02 absorber by way of the formation of undesired
sulfates. Even if all of the SO, were to bypass the prescrubber (an unlikely
occurrence) it would have amounted to less than 15% of the nominal sulfate
found in the absorber liquor products. The pH of the prescrubber slurry ranging
between 2.0 and 3.5 indicated that a substantial portion of the SO, was prob-
ably collected at that location.
Errors induced by measurement errors in the SO, sampling and analysis can
be treated in similar fashion to the error analysis for SO- measurement.
Combining and differentiating the equation relating SO, concentration to the
measured variables yields the following:
C-6
-------
AS03 |A(MS03)| |AT| |A(METCOR)| |A(vso3)| |APB| + |APS
~~ = +' + + • *
SO~T MS03 l ' T + 460 ' METCOR VS03 •* (PB + PS)
A(MS0) |AV | |ACJ AV AVT AV
- * * *
where: S03 = Sulfur trioxide concentration, ppm
MS03 = Mass of S03 collected, mg
T = Gas temperature at gas meter
METCOR = Meter correction factor
VSC>3 = Volume of gas collected during SCL sampling, ft
VR = Volume of reagent titrated, ml
C, = Conversion factor for reagent, mg SOT/ml
VD = Dilution volume, ml
VT = Titrated volume of sample, ml
Vg = Volume of original sample
Average values and expected maximum measurement errors for the above
parameters are listed as follows:
Average Values Maximum Expected Errors
T = 90°F AT = 5°F
METCOR = 1.0 A(METCOR) = 0.05
VS03 = 6.0 ft3 A(VS03) = 0.2 ft3
PB = 29.0 "Hg A(PB) = 0.1 "Hg
PS = -18.0 "Hg A (PS) = 0.5 "Hg
V,, = 1.0 ml AVD = 0.15 ml
K K
CR = 0.5 mg SO^/ml ACR = 0.01
VD = 50 ml AVD = 0.5 ml
VT = 1 ml AVT = 0.05 ml
Vg = 10 ml AVg = 0.6 ml
Combining these estimates gives the following:
AMS03
Most probable maximum error for 4. = 16.6%
C-7
-------
ASO
Most probable maximum error for -F— = 18.4%
This analysis reveals two factors: the first being that the dominant major
measurement error arises from the deviation in the titration end point. The
second conclusion is that the expected measurement errors cannot account for
the large scatter in the SO, measurements. This means that the scatter arises
either out of "real" variations or gross experimental errors such as air leak-
age or sample contamination.
The primary conclusions drawn from the SO, measurements are first that
insufficient data were obtained to characterize these concentrations in terms
of furnace operation (this was not a goal of the contract) and secondly,
assuming that the measured nominal level was truly representative, the SO,
could not have played a very substantial role in the formation of sulfates
within the S02 absorbers.
C . 2 . 3 NOy Error Analysis
Although the Phenyldisolfonic Acid method for NCL determination is a
recognized ASTM standard, this procedure is quite tedious, time consuming, and
contains a number of opportunities for experimental error. Probably, the major
source for error is the chemical analysis procedure' for developing the color.
This involves transferring the sample from the collection flash to an evapor-
ation dish, evaporating the sample, a filtration step, and a final dilution
step. Errors made during this process are procedural and are minimized with
careful technique and with experience. However, these errors are not readily
subject to numerical examination.
A number of errors are possible from experimental measurements. These
errors have been analyzed by the technique previously described. This analysis
yields the following results:
ANOX
The major contributions to the 7.51 error are the uncertainty in the slope
of the color absorbance curve and the uncertainty in the exact occupied volume
»
of the collection flask.
An additional error which evidently happened on occasion was leakage of
air into the evacuated collection flask either before or during the sampling
C-8
-------
period. This probably accounts for some of the unexpectedly low NCL values
experienced on occasion.
One additional factor that should be noted is that virtually all of the
potentially significant errors would tend to produce negative errors, i.e.,
to produce lower NCL concentrations than the "true" value. This is important
when attempting to interpret scattered MX data which is in fact not randomly
scattered but is biased by unreal low values. It is the writer's opinion that
the most nearly correct value listed in Table 4.12 for any given test is the
maximum value for that test.
As a check on the PDS procedure, a prepared NCL gas standard was purchased
and sampled. This gas contained a 900 ppm concentration of NCL in a nitrogen
atmosphere. Three samples of this gas were analyzed with the following results:
Sample NCL, ppm
1 738
2 810
3 777
Avg. 775 ppm
Average error for 900 ppm = 13.9%
Average deviation from 900 ppm = 128 ppm
Two factors are worthy of note. All three samples analyzed low and
secondly, the average deviation of these samples is comparable with the average
deviation of the NCL measurements reported in Table 4.11. This data confirms
the trends evident in the bulk of that NCL test.
C.2.4 Dust Loading Error Analysis
The dust loadings performed at the furnace exit and at the exit of the
scrubbers were performed by different techniques. The Thorston sampling
apparatus as described in Section 3.7 is a large volume high mass sampler.
The potential errors with this sampler include nonrepresentative sampling
because of anisokinetic effects plus a number of measurement errors. An
analysis of the effect of measurement errors lead to the following net error
potential:
AQDLriH.fr) _ ,, ~
DLITH+ -1 '
C-9
-------
The error analysis reveals that the largest potential error arises from
the deviation in the gas temperature at the flow orifice. All other measure-
ment errors are small in comparison.
The other potential source of error here arises from possible anisokinetic
sampling situations. The estimated isokinetic sampling rate error is approxi-
mately -15 to -20%. This error would tend to produce dust loadings biased
slightly upwards. The magnitude of error from anisokinetic sampling can be
estimated by a method described by Badzioch^ '. He relates the percentage
error as follows:
CVS - V)
e = —r (100)a (C-10)
S
where e = % error in dust concentration measurement
Vg = Velocity at which gas was sampled
V = The isokinetic sampling velocity
a = An inertial parameter depending upon the flow patterns around the
sampling nozzle.
'Xy ^
For normal pulverized coal flue gas, a = 1/2 and therefore e = 7.5% to 10%.
In any event, the maximum possible value of a = 1.0 such that the error due
to anisokinetic sampling would certainly be less than 20%.
The average deviation of furnace exit dust loadings was 36%. Since the
expected experimental error is probably less than 13%, the deviations are
probably real. In any event, the anisokinetic error indicates that most of
these dust loadings might be approximately 10% too high.
The dust loading sampling method employed at the exit of the particulate
scrubber and the pilot plant exit was a mini-sampler using simple filtration
in a paper thimble. This method as described in Section 3.7 differed consider-
ably from the Thorston method. The total sample weight was around 0.01 gram,
and the sampling rate at about 0.5 CFM. Primary sources of error were mea-
surement errors with a small possibility of error from nonrepresentative sam-
pling. Nonrepresentative sampling seems possible only if a substantial portion
of the fly ash particles leaving the scrubber were larger than 4 microns.
If the particle size distribution entering the particulate scrubber approximated
a normal distribution for P.C. fly ash, then in order to have a substantial
fraction of fly ash in the 4 micron range, the measured dust loading would have
to be in error to the extent of over 1000%. This would seem very unlikely.
C-10
-------
FIGURE C.3.
VELOCITY PROFILE
DOWNSTREAM OF FLOW
STRAIGHTENER
On the other hand, only single point, center line sampling \vas utilized
at the particulate cyclone exit. The sampling point was located 2-1/2 pipe
diameter from a flow straightener described
in Section 2.6.1. One might argue that
this distance is insufficient. To check
this point, a pitot tube velocity pro-
file was determined. It was reasoned
that if the particle size of the ash at
this point was at least nearly sub-
micron, then the particulate concentra-
tion would not be any more maldis-
tributed than the gas. Figure C.3
illustrates this profile which is seen
to be of the normal turbulent flow
D 1 « 0 I IU
shape with only a slight skewness. It «,,«.,„«,. •«».
was concluded from this profile and
from the expected particle size distribution that nonrepresentative sampling
was not a factor contributing to the error potential for the outlet dust loadings.
The effect of anisokinetic sampling was determined as above. For a
4-micron particle, a = 0.0224 and the maximum deviation from isokinetic sampling
rate was ^25%. Therefore, for a 4-micron particle, the potential error would
only be ^0.56%. The value of a was determined assuming that the flow distur-
bance upstream of the sampling nozzle was 8 on. This value is an extrapolation
of the data presented by Badzioch. Again, the anisokinetic sampling factor is
probably not significant.
Errors arising from sampling measurements were determined in the usual
manner with the following results:
ADL
Most Probable Maximum Error:
"EL
= 10.6%
This error is dominated almost completely by the filter weight errors.
C.3 CHEMICAL ANALYSIS ERRORS - LIQUID
C.3.1 Palmrose Analysis
The Palmrose technique for chemical analysis of the composition of the
MgO-S09-FLO system was developed for application in the pulp and paper industry.
Lt £,
This technique is a wet chemical method involving two titrations. Potential
errors arise from sampling, titration, and premature stripping of sulfur dioxide.
C-ll
-------
Since sampling was performed with
a syringe (using a #18 gauge needle),
the question of whether the slurry was
sampled representatively was investi-
gated by comparing samples obtained in
this manner with those obtained with a
sample flask from the product spray.
This test showed that the syringe did
sample representatively.
Errors arising during chemical analy-
sis involve volume error in the syringe
(2-1/2 ml syringes were used) titration
errors, and errors in the chemical rea-
gents .
The' maximum probable errors for each
analysis are shown in Figure C.4. For the
most part, these error potentials are not
too significant from the standpoint of
their effects upon the absorption system.
Possible exceptions are the dissolved mono-
sulfite concentration and the bisulfite
concentration. Both these parameters could
have significant effects upon the SCL
absorption capability of the slurry. It
was, therefore, desirable to know these
concentrations with some degree of cer-
tainty.
The precision of the palmrose analysis
to determine the suspended solids concen-
tration is illustrated in Figure C.5. It
is a plot of the monosulfite solids con-
centration determined by the Palmrose analy-
sis against the total solids determined by
filtration. This plot shows that the agree-
ment is fairly good over the full range on
a majority of the tests but that deviations
are biased towards the Palmrose analysis.
C.4. MAXIMUM PROBABLE ERROR
IN PALMROSE ANALYSIS
. O.I
a"
»O.
8
I ) 4 S b
Toul SOj. ta/100 «1 " SOj
0 O.S 1.0 t.S 2.0
lFn»| oVIM U u SO;
Total SOj, p^lOO ml u SOj
S o.i
I .
s-,.
«•»
Frw
(E)
0 & 10 IS 20
\ Solid* (Acbal) by wt.
C.5. COMPARISON OF SLURRY
STRENGTH BY TWO TECHNIQUES
3. :<
a
§ i.
• 16
,,
1 »
£ 10
| ,
S *
1 '
i
0
V
/
»
/*
1
/
•
— -
1
•
/
-
m
1
^/
u
,
•/
14
%X
id
/•
11
/
a
>/
•
12
/
U
/
u,
Sutpendad Salldi Cone, by Crav. Aral.. p^lOO •!
C-12
-------
C.3.2 Sulfate Analysis Errors
The magnesium sulfate concentration was a most important measurement.
The errors associated with this analysis are similar to those encountered in
the SCL analysis with an additional source arising from potential oxidation
of the lOcc slurry sample from sulfite to sulfate before analysis was com-
pleted. The maximum probable error arising from measurement deviation amounts
to:
ASulfate
Sulfate
= 17.051
I I
Commit*
The major contributor to this error is the vague titration endpoint. The
tendency was to overtitrate, that is, to exceed the true endpoint.
As an independent check upon the sulfate analysis technique, samples were
analyzed from time to time by conventional C.6. COMPARISON BETWEEN
gravimetric methods. Although a qualitative SmS^A^^SSsES
comparison was exhibited, quantitative agree-
ment was poor. This comparison is plotted f
in Figure C.6. On the average, the titrated ?
§ '
sulfates are about 64% higher than the }
gravimetric sulfates. The 171 experimental jj'
error can explain only part of this devia- f,
tion. The remainder must be due to una- j
voided sulfation of some of the sulfite f
in the titrated sample. See Section A.5
for description of titration method.
C.3.3 MgO Makeup Composition and Concentration
The concentration of the magnesia slurry whether slaked or not was deter-
mined by titration with 7 N sulfuric acid. Titration errors and representative
sampling errors are the primary problems. Sampling of the slurry is a problem
because the magnesia wants to settle very rapidly meaning that drawing of a
representative sample into the pipette can be difficult. This was circumvented
in part by cutting the capillary end off the pipette which permits rapid with-
drawal of the magnesia slurry from the sample beaker. Assuming a representative
sample was obtained, the maximum probable error in the magnesia concentration
eatnttaa br ritmtric *MJr»l». pt/100 ml u 3
A(XMGO) =
XMGO
= _
'"
The composition of the magnesia, i.e., the ratio of M (OH)- to MgO is important
o
from the system chemistry standpoint. The method for determining the extent of
C-13
-------
hydration of the magnesia was described in Section A. 8. Errors in this analysis
arise principally from the need to remove uncombined water from the sample before
calcining. If all of this water is not removed, then the fraction slaked will
be too high. If this drying process is too severe, some of the combined water
may be released yielding fractional slaking values too low. The total maximum
probable error in the magnesia conversion for a 1% error in the precalcined
weight is:
where S = The mole % magnesia converted to
One final point regarding the magnesia composition is that the results
indicated that preslaking was not important from an S02 absorption point but
may have been important as a deposition consideration. As is explained in
Section 5.4, even in the case of deposition, the important factor is the con-
centration of slowly reacting magnesia which can, if present, continue to
combine with bisulfite outside of the gas-liquid contacting zone. Preslaking
may or may not have a bearing on the level of this material. It may be that
the magnesia used actually had a spectrum of reactivity with some very reactive
material that hydrated in cold water and some relatively unreactive material
which resisted efforts to preslake. What really is needed to delineate this
point is an analysis method which would be sensitive enough to detect magnesia
in the presence of bisulfite in the absorption slurry. If the Palmrose analysis
were sufficiently accurate, it would in fact provide this information by the
difference in bisulfite concentration between the filtered and unf iltered analy-
sis. Noting Figure C.4(b), however, it is obvious that the method is not
sufficiently accurate. However, an automatic titrator might be. This point
was not explored.
C.4 FLOW RATE ERROR ANALYSIS
C.4.1 Coal
The feed rate of coal to the furnace was not a directly measured quantity.
Rather, it was calculated based upon an assumed constant composition and upon
the measured combustion air rate. It is also assumed that the combustion is
complete. This last assumption will be discussed later. The assumed coal com-
position and the maximum expected variation in that composition are listed as
follows :
C-14
-------
Assumed _ Variation
Carbon 71.5% by weight 70 to 74%
Hydrogen 4.85% 4.5 to 5.5%
Oxygen § Nitrogen 8.17% 7 to 9%
Sulfur 3.7% 3 to 5%
Ash 7.4% 6 to 11%
Moisture 4.38% 3 to 7%
100.00%
Based upon the above assumptions, the coal feed rate possessed an error potential
- 0.10 or lot
The major contributor to this error was not the uncertainty in the coal
composition, but rather the deviation in the excess oxygen measurement.
C.4.2 Air and Flue Gas
The air to the furnace is monitored by standard ASME design orifices. The
errors arising from measurements were minimal and amounted to only about 3%
maximum probable error.
One air stream to the furnace however is not accounted for in the total air
flow calculation. This source is the compressed air which fluidized the coal
falling from the end of the screw. A rotometer indicated a constant flow of
about 0.6#/min. This amounts to about 6% of the primary air and 0.6% of the
total air.
The flue gas flowrate at the furnace exit was not measured but was calcu-
lated from the furnace material balance. Therefore, the deviation in this flow-
rate reflects the deviations in the input flows to the furnace. The only furnace
input measurement which reflects a potential significant error is the combustion
air flowrate. This yielded a probable error of about 3%. In addition, a condi-
tion sometime existed whereby air could leak into the furnace or duct thus
diluting the flue gas. This condition existed when the static pressure at the
furnace exit went strongly negative relative to the atmosphere. Normal operation
called for maintaining the static pressure at this location close to neutral.
However, this was not possible because at times pluggage occurred within the
tube banks of the furnace cooling section. This leakage was difficult to esti-
mate and occurred primarily around probe penetration into the furnace stack. As
a check against leakage throughout the pilot plant, an Orsat analysis was
C-15
-------
performed at various locations. These tests at no time produced any sign of
air leakage.
The flue gas flowrate was measured by a standard ASME orifice in the hori-
zontal duct between the particulate scrubber, and the gas absorber consistently
indicated a gas flow about 400 to 600 pounds per hour less than the calculated
flowrate at the furnace exit. This represents about a 101 difference. This
difference is difficult to rationalize. The possibility of actual loss via
leakage of this quantity of gas seems very unlikely. Primary opportunities for
leakage include the emergency relief apparatus (See Figure 2.12) and an expan-
sion joint. However, these sources were checked on several occasions with no
leakage detected. The maximum probable error in the flowrate amounts to
A[FWDG (9)] ^ fi -,
FDGW (9)D'/0
This potential error is not sufficient to account for the observed deviation.
Recalling the pitot tube traverse described in Section 6.2.4, that test showed
the following agreement between orifice and pilot tube traverse.
Pilot Flowrate - 1295 ft3/min
Orifice - -1340 ft3/min
This is better than 96% agreement and reduces the probability of any large errors
in the orifice calibration.
The flowrate measurement at the pilot plant exit orifice indicated a flow-
rate consistently (400 to 600 #/hr) higher than the flowrate at the intermediate
orifice (between particulate scrubber and gas absorbers as referred to above).
However, good agreement was established between this orifice and the calculated
dry gas flowrate at the furnace exit. A plot of this material balance is shown
in Figure C.7.
Although any unbalance in flue gas flowrate is undesirable, the only really
significant factor that this situation suggests is air leakage into the pilot
plant and the resulting effects that might have on the sulfate formation. How-
ever, as we stated previously, upon investigation via an Orsat analysis, no
leakage could be found. Thus, it is felt that the unbalance must arise from
anomolies concerning the orifice location relative to bends, etc.
C.4.3 Makeup Water
The makeup water rate to all through scrubbers was established through
volume displacement meters calibrated down to 0.1 gal. The makeup water which
C-16
-------
C.7. FLUE GAS MATERIAL BALANCE
v
Rmace Exit ••
Pilot Plant Exit -A
Abiorber Inlet • •
Rimue Exit ••
Pilot Plant Liit -A
TS 8? 86 90 96 98 102 106 110 IK 110 122 126 130 134
Test Aaber
6000
Dry Flue Gu. Iti/nr
i I i i i
\4
N
^
y
.-"
^
V
^i
_-)
, ,>
Pumace Exit -•
Pilot Plant Exit -A
Ataorter lnl«t • •
<,
[ 1
i (
/
•>
~-H
s
>j
—J
/
/--'
r>
^~
r-~ '
\^
— 1
>XX
•'*<
h 4
/
i — -^
'"
47 U^> 150 ISA 138 162 166 170 17i 1 78 182 186 190 194 198 2
C-17
-------
was controlled by the level controller on the scrubber sumps was therefore
an unsteady flow. The average rate was established by metering the volume
displaced over a period of several minutes. The degree of error in this mea-
sure was dependent upon how long the integrating period was established. The
shorter the period, the greater the probable error. For the particulate
scrubber, the probable maximum error was:
= 0.2% @ AT = 5 min. or 0.03% @ AT = 30 min.
ow
and for the gas absorbers:
? = 3% g AT = 5 min. or 0.51 @ AT = 30 min.
Flow
C.4.4 Makeup MgO Slurry
The makeup MgO slurry was monitored by a calibrated flowrator. Although
at times this flow needed careful manual control, it was normally fairly steady.
The error potential of this stream is summarized as follows:
AFMGO n nit T TO
= °-072 or 7-2*
The major contributor is the variation in float position.
C.4.5 Venturi Absorber - FBA Product FlowrateVj
The method of monitoring the product f low' from,. the. SCL absorbers is
described in Section 2.6.1 and illustrated in Figure 2.20. This system was a
source of considerable error in flow measurement. The primary course of error
was partial pluggage of the product nozzle. The flowrate was correlated to
the nozzle pressure. Thus, as the nozzle orifice plugged, the flow was
restricted but the pressure still indicated an undisturbed flow. The maximum
probable error can be summarized as follows:
= 32.51 with pluggage
without
AFlow
Flow , .
product
The pluggage error yields an apparent flowrate higher than actual.
C-18
-------
C.5 MATERIAL BALANCES
C.5.1 Sulfur Material Balance
The sulfur material balance variances are a reflection of all the errors
involved in determining concentrations and flowrates of sulfur-containing
streams. The sulfur material balance is illustrated in Figure C.8. This
plot shows the total sulfur flow accounted for at the furnace inlet, leaving
the furnace, and leaving the pilot plant. A good balance is evident up to
test 48. From there to test 102, a large discrepancy existed between all
three points. From test 102 to 128, the pilot plant exit flow continued to
be in wide disagreement. From thereon, however, agreement was good. The
maximum variance that would be expected at each location from measurement
errors is summarized as follows:
Furnace Input; ^| = 28.8%
Furnace Exit; ^| = 15.3%
Pilot Plant Exit; £| = 40%
Examining Figure C.8, it is apparent that experimental error can account
for the variance in runs 28-48 and 128-205 but not between 48 and 127. The
reason for this greater variance is due to a severe pluggage condition from
sulfite deposits in the product nozzle. The restriction to flow must have
been higher than the 25% estimate used in calculating the expected error in
the product flowrate. In fact, the sulfur balance is a good indicator in
itself of the extent of deposition.
The difference between the sulfur input to and output from the furnace is
a more difficult discrepancy to account for. The largest expected variances
in individual measurements that can be reasonably expected cannot account for
the difference evident between runs 60 and 102. The only explanation which
seems plausible is as follows: During the period from about March 1 to April 1,
a truck strike prevented delivery of coal to the test site. To insure con-
tinuation of the test program, coal which is stock piled in-plant was used
instead. This coal was very wet (being exposed to the elements over a period
of time). As such, it did not dry completely in the pulverizer. Thus, the
coal lacked its usual fluidity. Therefore, when this coal was fed to the fur-
nace, it did not disperse as well as P.C. coal normally does. The result, the
C-19
-------
C.8. SULFUR MATERIAL BALANCE
9 5
JbH
Punace Inlet •
Pilot Plmt Exit -
Punu« Diit -
13— in— in
\t**
A!!:
Furnace Inlet - •
Pilot Plant Exit -•
Rimace Exit -A
"lifl 162 166 170 174 178 182 186 190
198 202
C-20
-------
author believes, is that much of the coal fell to the furnace floor as large
flakes whereupon it would slowly char liberating the more volatile sulfur
constituents. In support of this hypothesis is the fact that during this
period, furnace deposition was severe and rapid. The ash collected by the
Thorston sampler contained as much as 20% carbon. Finally, a yellowish
crystalline deposit which formed on the outer side of the P.C. hopper during
this period was analyzed and found to be an iron sulfate hydrate.
C-21
------- |