U.S. Environmental Protection Agency Industrial tnvironmental Research EPA'600/
Office of Research and Development Laboratory . rt^T
Research Triangle Park. North Carolina 27711 May 1977
EPA-600/7-77-050b
FINAL REPORT: DUAL ALKALI
AND EVALUATION PROGRAM
Volume II. Laboratory and
Pilot Plant Programs
Interagency
Energy-Environment
Research and Development
Program Report
-------
•RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. ^
Environmental Protection Agency, have been grouped into seven series.
These seven broad categories were established to facilitate further-
development and application of environmental technology. Elircina i
of traditional grouping was consciously planned to foster technology
transfer and a maximum interface in related fields. The seven series
are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
This report has been assigned to the INTERAGENCY ENERGY-ENVIRONMENT
RESEARCH AND DEVELOPMENT series. Reports in this series result from
the effort funded under the 17-agehcy Federal Energy/Environment
Research and Development Program. These studies relate to EPA's
mission to protect the public health and welfare from adverse effects
of pollutants associated with energy systems. The goal of the Program
is to assure the rapid development of domestic energy supplies in an
environmentally—compatible manner by providing the necessary
environmental data and control technology. Investigations include
analyses of the transport of energy-related pollutants and their health
and ecological effects; assessments of, and development of, control
technologies for energy systems; and integrated assessments of a wide
range of energy-related environmental issues.
REVIEW NOTICE
This report has been reviewed by the participating Federal
Agencies, and approved for publication. Approval does not
signify that the contents necessarily reflect the views and
policies of the Government, nor does mention of trade names
or commercial products constitute endorsement or recommen-
dation for use.
This document is available to the public through the National Technical
Information Service, Springfield, Virginia 22161.
-------
EPA-600/7-77-050b
May 1977
FINAL REPORT: DUAL ALKALI TEST
AND EVALUATION PROGRAM
Volume II. Laboratory
and Pilot Plant Programs
by
C.R. LaMantia, R.R. Lurit, J.E. Oberholtzer,
El. Field, and J.R. Valentine
Arthur D. Little, Inc.
Acorn Park
Cambridge, Massachusetts 01240
Contract No. 68-02-1071
Program Element No. EHE624
EPA Project Officer: Norman Kaplan
Industrial Environmental Research Laboratory
Office of Energy, Minerals, and Industry
Research Triangle Park, N.C. 27711
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, D.C. 20460
-------
ABSTRACT
This report presents the results of the Dual Alkali Program
conducted by Arthur D. Little, Inc., (ADL) for the Industrial
Environmental Research Laboratory, Research Triangle Park
(IERL, RTP) of the U.S. Environmental Protection Agency (EPA).
The purpose of the program was to investigate, characterize
and evaluate the basic process chemistry and the various
modes of operation of sodium-based dual alkali processes.
The work was carried out at three levels of investigation:
• Task I - Laboratory studies at ADL and IERL, RTP.
• Task II - Pilot Plant Operations in a 1,200 scfm
system at ADL.
• Task III - Prototype Test Program on a 20-megawatt
Combustion Equipment Associates (CEA)/ADL
dual alkali system at Plant Scholz, Southern
Company Services, Inc./Gulf Power Company.
Various modes of operating dual alkali systems on high- and
low-sulfur fuel applications were investigated, including:
• Concentrated and dilute sodium scrubbing systems
• Lime and limestone regeneration
• Slipstream sulfate treatment schemes.
In each mode, the objective was to characterize the dual alkali
process in terms of S02 removal, chemical consumption, oxidation,
sulfate precipitation and control, waste solids characteristics
and soluble solids losses.
This is Volume II of the final report covering Tasks I and II,
the laboratory and pilot plant programs. Volume I is the Exec-
utive Summary; Volume III covers the prototype test program,
Task III.
iii
-------
VOLUME II
TASKS I AND II, LABORATORY AND PILOT
PLANT PROGRAMS
TABLE OF CONTENTS
Page
Chapter No.
ABSTRACT iii
ACKNOWLEDGEMENTS xxi
APPLICABLE CONVERSION FACTORS xxiii
I SUMMARY 1-1
A. PURPOSE AND SCOPE 1-1
B. RESULTS AND CONCLUSIONS 1-2
1. Pilot Plant S02 Removal and
Oxidation - General 1-3
2. Concentrated Mode with Lime Regeneration 1-4
3. Concentrated Mode with Sulfuric Acid
Sulfate Treatment 1-7
4. Concentrated Mode with Limestone
Regeneration , 1-8
5. Dilute Mode with Lime and Limestone
Regeneration 1-10
6. Solids Characterization — Dilute and
Concentrated-Lime Regeneration Modes 1-13
II INTRODUCTION II-l
A. BACKGROUND AND OBJECTIVES II-l
B. DESCRIPTION OF CHEMISTRY AND DEFINITION
OF TERMS II-2
III LABORATORY AND PILOT PLANT SYSTEMS III-l
A. LABORATORY METHODS III-l
1. Experimental Apparatus, Operation and
Sampling Procedures III-l
2. Analytical Procedures III-3
-------
TABLE OF CONTENTS (cont)
Page
No.
Chapter
III (cont)
B. PILOT PLANT III-5
1. Description of the Facility III-5
2. Scrubber Operating Characteristics III-5
IV LIME REGENERATION — CONCENTRATED ACTIVE
SODIUM MODE IV~1
A. PRIOR WORK ON LIME REGENERATION AT ADL IV-1
B. LABORATORY STUDIES OF SULFATE
PRECIPITATION WITH LIME IV-2
1. Introduction IV-2
2. Experimental Results IV-3
3. Discussion IV-9
C. PILOT PLANT OPERATIONS — CONCENTRATED
ACTIVE SODIUM MODE IV-11
1. Pilot Plant Test Program IV-11
2. Regeneration Reactor Performance IV-12
3. Summary of Overall System Operation IV-33
D. CONCLUSIONS
V SULFURIC ACID TREATMENT — CONCENTRATED MODE V-l
A. LABORATORY RESULTS V-l
1. Introduction V-l
2. Experimental Results and Discussion V-3
B. PILOT PLANT RESULTS V-7
1. Sulfuric Acid Reactor Performance V-7
2. Integrated System Operation V-13
C. SULFURIC ACID REACTOR MODEL V-17
D. CONCLUSIONS V-21
vi
-------
TABLE OF CONTENTS (cont)
Page
Chapter No.
VI LIMESTONE REGENERATION — CONCENTRATED MODE VI-1
A. INITIAL LABORATORY STUDIES ' VI-1
1. Batch Studies Comparing the Reactivities
of Different Limestones VI-2
2. Effects of Feed Stoichiometry, Sulfate
Concentration, and Temperature on
Reaction Rate VI-5
3. Continuous Reactor Studies of
Regeneration with Limestone VI-7
B. SUBSEQUENT LABORATORY STUDIES OF FACTORS
AFFECTING THE PHYSICAL PROPERTIES OF
LIMESTONE PRODUCT SOLIDS VI-12
1. Effects of Sulfate Level on Settling
Behavior of Solids VI-14
2. Effects of Magnesium on the Limestone
Regeneration Reaction VI-14
3. Studies of Liming for Magnesium Control
and the Use of Flocculants to Improve
Dewatering Properties VI-26
C. PILOT PLANT OPERATIONS VI-35
1. Open-Loop Multistage Testing VI-37
2. Closed-Loop Runs VI-57
D. CONCLUSIONS VI-71
VII LIMESTONE/LIME DILUTE MODES VII-1
A. EPA LABORATORY RESULTS VII-1
B. DILUTE MODE ALTERNATIVES VII-3
C. ADL LABORATORY STUDIES VII-10
1. Studies of Calcium Precipitation from
CSTR Effluents VII-11
2. Measurements of Gypsum Solubility VII-13
3. Additional Batch Reaction Studies of
Dilute Mode Regeneration with Lime VII-15
vii
-------
TABLE OF CONTENTS (cont)
Chapter
Page
No.
VII (cont)
D. PILOT PLANT OPERATIONS
1. Introduction
2. Reactor System Characterization
(Open-Loop Operation)
3. Evaluation of Overall System Performance
(Closed-Loop Operation)
E. CONCLUSIONS
VII-20
VII-20
VII-21
VII-40
VII-53
VIII STUDIES OF THE PHYSICAL PROPERTIES OF
DUAL ALKALI PRODUCT SOLIDS
A. INTRODUCTION
B. CHEMICAL AND PHYSICAL CHARACTERISTICS
1. Chemical Composition
2. Crystalline Morphology
3. True and Apparent Densities of the Solids
C. COMPACTABILITY OF THE SOLIDS —
MOISTURE/DENSITY RELATIONSHIP
1. Apparatus and Procedure
2. Results and Discussion
D. RESISTANCE OF THE SOLIDS TO PHYSICAL
PENETRATION
1. Apparatus and Procedure
2. Results and Discussion
E. UNCONFINED COMPRESSIVE STRENGTHS OF
COMPACTED DUAL ALKALI SOLIDS
1. Apparatus and Procedure
2. Results and Discussion
VIII-1
VIII-1
VIII-1
VIII-1
VIII-3
VIII-7
VIII-10
VIII-10
VIII-11
VIII-13
VIII-15
VIII-15
VIII-20
VIII-20
VIII-20
viii
-------
TABLE OF CONTENTS (cont)
Chapter
VIII (cont)
Page
No.
IX
X
XI
F. PERMEABILITIES OF SOLIDS
1. Apparatus and Procedure
2. Results and Discussion
G. LEACHING OF SOLUBLES FROM DUAL ALKALI SOLIDS
1. Apparatus and Procedure
2. Results and Discussion
VIII-20
VIII-22
VIII-22
VIII-24
VIII-24
VIII-24
H. EXPLORATORY STUDIES OF THE EFFECTS OF CHEMICAL
TREATMENT OF THE PROPERTIES OF; DUAL ALKALI SOLIDS VIII-28
1. Experimental Procedures
2. Results and Discussion
I. CONCLUSIONS
REFERENCES
ANNOTATED BIBLIOGRAPHY
GLOSSARY
VIII-28
VIII-28
VIII-31
IX-1
X-l
XI-1
ix
-------
LIST OF FIGURES (cont)
Figure Page
No. No.
VI-21 Run 123 (High Sulfate) VI-42
VI-22 Run 124 (High Sulfate + Post-Liming) VI-42
VI-23 Run 127 (Solids Recycle) VI-42
VI-24 Run 128 (Extended Holdup) VI-42
VI-25 Reaction Rate vs Limestone Feed Stoichiometry . VI-43
VI-26 Run 125 (Base Case) VI-44
VI-27 Run 126 (Low CaC03 Feed) VI-44
VI-28 Run 127 (Solids Recycle) VI-44
VI-29 Run 128 (Extended Holdup) VI-44
VI-30 Soluble Calcium vs TOS in Reactor 1 VI-45
VI-31 Soluble Calcium vs TOS in Reactor 6 VI-45
VI-32 Relation of Observed Apparent Solubility Products
to Saturation Values for CaS03 VI-47
VI-33 Settling Curves for Run 125 VI-48
VI-34 Settling Curves for Run 120 VI-49
VI-35 Settling Curves for Run 26 VI-50
VI-36 Settling Curves for Run 121 VI-51
VI-37 Settling Curves for Run 123 VI-52
VI-38 Settling Curves for Run 124 VI-53
VI-39 Settling Curves for Run 127 VI-54
VI-40 Settling Curves for Run 128 VI-55
VI-41 Sulfate Precipitation in the Concentrated
Limestone Mode VI-58
VI-42 Process Flow Diagram for Concentrated Limestone
Mode Pilot Plant Operations Continuous Closed-Loop VI-60
xiv
-------
LIST OF FIGURES (cont)
Figure
No.
No.
VI-6 Precipitation of Sulfate by Fredonia Limestone
in 50-Minute CSTR Experiments VI-13
VI-7 Settling Curves for Solids Produced in 50-Minute
CSTR Experiments Using Fredonia Limestone VI-15
VI-8 Effect of Soluble Magnesium on Batch Regeneration
Reaction Rates with Fredonia Limestone VI-17
VI-9 Settling Behavior Observed for Solids Produced in
a 50-Minute CSTR Using Fredonia Limestone VI-18
VI-10 Settling Behavior of Solids Produced During
Regeneration with Limestone VI-19
VI-11 Settling Behavior of Solids Produced During
Regeneration with Limestone VI-21
VI-12 Change in Settled Volume of Effluent Slurry
Solids as a Function of Reactor Operating
Time - Experiments 65 and 66 VI-22
VI-13 Settling Behavior of Solids Produced During
Regeneration with Limestone in Continuous
Reactors VI-25
VI-14 Settling Behavior of Solids Produced in the
Presence of 2,200 ppm Magnesium - Experiment 70 VI-29
VI-15 Settling Behavior of Solids Produced During
Regeneration with Limestone in the Presence
of 300 ppm Magnesium - Experiment 71 VI-30
VI-16 Comparison of Settling Behavior of Solids Produced
During Regeneration with Limestone in the Presence
of 300 ppm Magnesium after 9 Hours of Reactor
Operation VI-31
VI-17 Run 125 (Base Case) VI-41
VI-18 Run 120 (High CaC03 Feed) VI-41
VI-19 Run 126 (Low CaC03 Feed) VI-41
VI-20 Run 121 (High Magnesium) VI-41
xiii
-------
LIST OF FIGURES (cont)
Figure Page
No. No.
VI-6 Precipitation of Sulfate by Fredonia Limestone
in 50-Minute CSTR Experiments VI-13
VI-7 Settling Curves for Solids Produced in 50-Minute
CSTR Experiments Using Fredonia Limestone VI-15
VI-8 Effect of Soluble Magnesium on Batch Regeneration
Reaction Rates with Fredonia Limestone VI-17
VI-9 Settling Behavior Observed for Solids Produced in
a 50-Minute CSTR Using Fredonia Limestone VI-18
VI-10 Settling Behavior of Solids Produced During
Regeneration with Limestone VI-19
VI-11 Settling Behavior of Solids Produced During
Regeneration with Limestone VI-^21
VI-12 Change in Settled Volume of Effluent Slurry
Solids as a Function of Reactor Operating
Time - Experiments 65 and 66 VI-22
VI-13 Settling Behavior of Solids Produced During
Regeneration with Limestone in Continuous
Reactors VI-25
VI-14 Settling Behavior of Solids Produced in the
Presence of 2,200 ppm Magnesium - Experiment 70 VI-29
VI-15 Settling Behavior of Solids Produced During
Regeneration with Limestone in the Presence
of 300 ppm Magnesium - Experiment 71 VI-30
VI-16 Comparison of Settling Behavior of Solids Produced
During Regeneration with Limestone in the Presence
of 300 ppm Magnesium after 9 Hours of Reactor
Operation VI-31
VI-17 Run 125 (Base Case) VI-41
VI-18 Run 120 (High CaC03 Feed) VI-41
VI-19 Run 126 (Low CaC03 Feed) VI-41
VI-20 Run 121 (High Magnesium) VI-41
xiii
-------
LIST OF FIGURES (cont)
Figure Page
No. No.
VI-21 Run 123 (High Sulfate) VI-42
VI-22 Run 124 (High Sulfate + Post-Liming) VI-42
VI-23 Run 127 (Solids Recycle) VI-42
VI-24 Run 128 (Extended Holdup) VI-42
VI-25 Reaction Rate vs Limestone Feed Stoichiometry VI-43
VI-26 Run 125 (Base Case) VI-44
VI-27 Run 126 (Low CaC03 Feed) VI-44
VI-28 Run 127 (Solids Recycle) VI-44
VI-29 Run 128 (Extended Holdup) VI-44
VI-30 Soluble Calcium vs TOS in Reactor 1 VI-45
VI-31 Soluble Calcium vs TOS in Reactor 6 VI-45
VI-32 Relation of Observed Apparent Solubility Products
to Saturation Values for CaS03 VI-47
VI-33 Settling Curves for Run 125 VI-48
VI-34 Settling Curves for Run 120 VI-49
VI-35 Settling Curves for Run 26 VI-50
VI-36 Settling Curves for Run 121 VI-51
VI-37 Settling Curves for Run 123 VI-52
VI-38 Settling Curves for Run 124 VI-53
VI-39 Settling Curves for Run 127 VI-54
VI-40 Settling Curves for Run 128 VI-5S
VI-41 Sulfate Precipitation in the Concentrated
Limestone Mode
VI-58
VI-42 Process Flow Diagram for Concentrated Limestone
Mode Pilot Plant Operations Continuous Closed-Loop vi-60
xiv
-------
LIST OF FIGURES (cont)
Figure Page
No- No.
VI-43 System Operation in Run 411 (After 36 Hrs) VI-62
VI-44 System Operation in Run 430 (at Termination of Run) VI-63
VI-45 Crystals in Reactor 4, 28 Hrs into Run Using
Limestone (Run 430) VI-65
VI-46 Crystals in Reactor 4, 45 Hrs into Run Using
Limestone (Run 430) VI-66
VI-47 Crystals in Reactor 3, After 48 Hrs of Running
with Lime (Run 430) VI-67
VI-48 Crystals in Reactor 3, After 57 Hrs of Running
with Lime (Run 430) VI-68
VI-49 Crystals in Reactor 3, 8 Hrs After Resuming
Limestone Feed (Run 430) VI-69
VI-50 Crystals in Reactor 3, 30 Hrs After Resuming
Limestone Feed VI-70
VI-51 Settling Curves for Run 430 VI-72
VII-1 EPA Limestone Runs 21-30 - Extent of
Under-Saturation VII-5
VII-2 EPA Limestone Runs 21-30 - Extent of CaS03
Supersaturation VII-6
VII-3 Dilute Limestone/Lime Schematic VI I- 7
VII-4 Dilute Lime/Carbonate Softening Schematic VII-9
VII-5 Preliminary Batch Experiments Dilute Mode VII-23
VII-6 Utilization in Continuous, Open-Loop Runs VII-27
VII-7 Hydroxide Concentration vs Calcium Utilization
in Continuous, Open-Loop Experiments VII-29
VII-8 Calcium Sulfate Precipitation as a Function of
Soluble Sulfate Concentration VII-31
VII-9 Calcium Sulfate Precipitation vs TOS Concentration VII-32
xv
-------
LIST OF FIGURES (cont)
Figure
No.
VII-10 Calcium Sulfate Precipitation vs Lime Feed
Stoichiometry
VII-11 Concentration of Soluble Calcium at Saturation
vs Soluble Sulfate Concentration
VII-12 Calcium Supersaturation in Runs Using a CSTR
VII-13 Solids Settling Rate vs Feed TOS Level
VII-14 Concentration of Settled Solids vs Feed TOS Level
VII-15 Process Flow Diagram for Dilute Lime Mode
Pilot Plant Operations (with TOS Oxidation)
VII-16 Stream Compositions and Flows in Run 602
VII-17 Stream Compositions and Flows in Run 620A
VII-18 Top Tray After Run 620
VII-19 Middle Tray After Run 620
VII-20 Calcium Supersaturation vs Suspended Solids Levels
VIII-1 Dual Alkali Calcium Sulfite Solids (950X)
VIII-2 Dual Alkali Gypsum Solids (950X)
VIII-3 Direct Limestone Calcium Sulfite Solids (1000X)
VIII-4 Elements of Dual Alkali Solid Waste
VIII-5 Compaction of Dual Alkali Product Solids
VIII-6 Load vs Penetration of Dual Alkali Sulfite Solids
VIII-7 Penetration Resistance of Dual Alkali Sulfite Solids
VIII-8 Relationship Between Penetration Resistance of a
Proctor Needle and Moisture Content of a Clay Soil
Compared with Dry Density/Moisture Content Curve
Page
No.
VII-34
VII-37
VII-37
VII-39
VII-39
VII-42
VII-46
VII-47
VII-50
VII-51
VII-54
VIII-4
VIII-5
VIII-6
VIII-8
VIII-12
VIII-16
VIII-18
VIII-19
xvi
-------
VOLUME II
LIST OF TABLES
Table page
No. No.
III-l Scrubber System Operating Conditions III-8
IV-1 Sulfate Precipitation in Concentrated Mode Lime
Regeneration Laboratory Continuous Reactor —r
50-Min Residence Time IV-4
IV-2 Summary of Nominal Reactor Operating Conditions IV-13
IV-3 Summary of Soluble Calcium Concentrations IV-15
IV-4 Effect of CSTR pH and Holdup Time on Solids
Properties IV-24
IV-5 Effects on Sulfate Concentration and pH on
Settling Properties of Solids Produced on
a CSTR with a 30 Minute Holdup IV-25
IV-6 Comparative Settling and Filtration Properties IV-28
IV-7 Operating Conditions for Closed-Loop Operations IV-35
IV-8 Summary of Results for Closed-Loop Runs IV-37
V-l Summary of Laboratory Continuous Reactor
Sulfuric Acid Treatment Experiments V-5
V-2 Summary of Sulfuric Acid Slipstream
Treatment Results V-9
V-3 Summary of Closed-Loop Runs V-14
V-4 Model Simulations of Pilot Plant Operations V-19
VI-1 Solution Compositions Observed During Regen-
eration With Limestone in Continuous Reactors VI-23
Vl-2 Compositions and Characteristics of Solids
Produced During Regeneration With Limestone VI-24
Vl-3 Magnesium Control 'by Liming CSTR Product Slurry VI-27
Vl-4 Limestone/Lime Series Reactor Performance VI-32
xvii
-------
LIST OF TABLES (cont)
Table Page
No. No.
VI-5 Limestone/Lime Series Reactor Solids Properties VI-34
VI-6 Summary of Preliminary Open-Loop Runs Using a
Single CSTR With Marblewhite Limestone VI-36
VI-7 Conditions for Open-Loop Multistage Reactor Runs VI-38
VI-8 Summary of Open-Loop Limestone Runs in Multi-
stage Reactor System (Fredonia Limestone) VI-39
VI-9 Summary of Closed-Loop Runs Using Multistage
Reactor Systems VI-61
VII-1 EPA 2-Liter Reactor Continuous Fredonia
Limestone Runs — Summary VII-2
VII-2 EPA Runs - Potential of Lime Post-Treatment Step VII-4
VII-3 Dilute Mode CSTR Experiments VII-12
VII-4 Study of Calcium De-Supersaturation in
Post-CSTR Batch Reactions VII-14
VII-5 Solubility of Calcium Sulfate in Sodium
Sulfate/Hydroxide Solutions VII-16
VII-6 Batch Reactions of Lime With Sodium Sulfate
Solutions at Two Stoichiometries VII-17
VII-7 Batch Reactor Studies of the Effect of TOS
Level on Regeneration With Lime VII-19
VII-8 Experimental Matrix for Open-Loop, Dilute
Mode Reactor Tests VII-25
VII-9 Calcium Supersaturation in Open-Loop
Reactor Runs VII-36
VII-10 General Operating Conditions for Scrubber System VII-43
VII-11 Closed-Loop, Dilute Lime Mode Runs — General
Operating Conditions & Overall System Performance VII-45
VII-12 Summary of Closed-Loop, Reactor Performance VII-52
xviii
-------
LIST OF TABLES (cont)
Table
No.
VIII-1
VIII-2
VIII-3
VIII-4
VIII-5
VIII-6
VIII-7
VIII-8
VIII-9
Chemical Composition of FGD Product Solids
True Densities of FGD Product Solids and
Fly Ash
Water Stability Tests on Dual Alkali Sulfite
Penetration Tests of Dual Alkali Sulfite Solids
Unconfined Compressive Strengths of Compacted
Dual Alkali Solids
Permeabilities of Compacted FGD Sludge Solids
Leaching of Sodium and Calcium From Untreated
Dual Alkali Product Solids
Compressive Strength of Dual Alkali Solids
(Cured at 100% Relative Humidity)
Effect of Treatment on Leaching of Sodium and
Calcium From Dual Alkali Sulfite
Page
No.
VIII-2
VIII-9
VIII-14
VIII-17
VIII-21
VIII-23
VIII-25
VIII-29
VIII-32
xix
-------
ACKNOWLEDGEMENTS
The work under this program was performed over a four-year period from
May 1973 through May 1977, with contributions from many individuals
representing several organizations. Persons involved at Arthur D. Little,
Inc. were:
Principal Investigators
Charles R. LaMantia - Project Manager
Richard R. Lunt - Pilot Plant and Prototype Program Manager
James E. Oberholtzer - Laboratory Program Manager
Edwin L. Field - Data Analysis Manager
James R. Valentine - Chemical Analysis Manager
Contributing Staff
Itamar Bodek
Lawrance I. Damokosh
Bruce E. Goodwin
George E. Hutchinson
Michael lovine
Bernard Jackson
Indrakumar Jaahnani
C. Lembit Kusik
Stephen P. Spellenberg
Robert A. Swanbon
Frank J. Tremblay
Lawrence R. Woodland
The EPA Project Officer for the entire four-year progam, Norman Kaplan,
made continuing and important technical and management contributions to
the program. Michael Maxwell and Frank Princiotta at EPA, through their
involvement in the review and planning, helped to guide the program over
the four-year period. The earlier part of the EPA laboratory program
was conducted under the direction of Dean Draemel, now at Exxon. EPA
laboratory work was carried on and completed by James MacQueen and Robert
Opferkuch of Monsanto Research Corporation under contract to EPA.
The cooperation and important contributions and support of Gulf Power
Company and Southern Company Services, Inc. (SCS) to the prototype test
program were invaluable. Randall Rush, responsible for coordination of
the program at SCS, made important technical contributions to the test
program and to the preparation of this report, in addition to^this con-
tinuing support throughout the program; the value of Mr. Rush's dedica-
tion and commitment cannot be overstated. In addition, we would like to
thank Reed Edwards of SCS and James Kelly of Gulf Power for their on-site
xxi
-------
assistance in the operation of the system. We wish to acknowledge the
cooperation of George Layman of Gulf Power and William Harrison of SCS,
individually and as representatives of their organizations, in making
the prototype system available and for the operation and maintenance of
the system during the program.
The cooperation, support and contributions of Combustion Equipment
Associates, Inc. (CEA) and its personnel were important to both the
pilot plant and prototype test programs. With the cooperation of CEA,
both systems were made available to the program. Tom Frank, the CEA
Project Manager for prototype system, and Richard White, on-site for
maintenance and operations, were importantly involved in the prototype
test program. The cooperation of Richard Sommer is gratefully acknowl-
edged for CEA's participation and support in this program.
xxii
-------
APPLICABLE CONVERSION FACTORS
ENGLISH TO METRIC UNITS
British
Metric
5/9 (°F-32)
1 ft
1 ft2
1 ft3
1 grain
1 in.
1 in2
1 in3
1 Ib (avoir.)
1 ton (long)
1 ton (short)
1 gal
1 Btu
°C
0.3048 meter
0.0929 meters2
0.0283 meters3
0.0648 gram
2.54 centimeters
6.452 centimeters2
16.39 centimeters3
0.4536 kilogram
1.0160 metric tons
0.9072 metric tons
3.7853 liters
252 calories
xxiii
-------
I. SUMMARY
A. PURPOSE AND SCOPE
This report presents the results of the Dual Alkali Program conducted
by Arthur D. Little, Inc., (ADL) for the Industrial Environmental Research
Laboratory, Research Triangle Park (IERL, RTP) of the U.S. Environmental
Protection Agency (EPA). The purpose of the program was to investigate,
characterize, and evaluate the basic process chemistry and the various
modes of operation of sodium-based dual alkali processes. The work
covered a wide range of flue gas conditions, liquid reactant concen-
trations, and process configurations, including:
• concentrated and dilute mode (dilute sodium scrubbing solutions,
active Na+ concentration below about 0.15M)*
• use of lime and limestone for regeneration
• sulfuric acid treatment for sulfate control.
Each of the various modes was evaluated relative to the following per-
formance characteristics:
• SC>2 removal capability
• oxidation and sulfate formation and control
• lime/limestone utilization
• waste solids properties
• sodium makeup requirements and degree of closed-loop operation
• process reliability.
Investigations were carried out at three levels: laboratory, pilot plant,
and 20-megawatt prototype. Accordingly, the program was divided into three
tasks:
• Task I — Laboratory Program — In the ADL laboratory program,
experiments were performed on the regeneration of concentrated
sodium scrubbing solutions using lime or limestone, and the use
of sulfuric acid treatment for sulfate removal. Work also in-
cluded characterization of the chemical and physical properties
of dual alkali solids. Work was performed at IERL, Research
Triangle Park on regeneration using limestone in dilute mode
operation.
*See Introduction and Glossary for dual alkali terminology.
1-1
-------
• Task II — Pilot Plant Program — Pilot plant work was conducted
at the Combustion Equipment Associates (CEA)/ADL pilot facility
in Cambridge, Massachusetts. The following modes of operation
were investigated in the pilot plant program:
concentrated mode using lime for regeneration
concentrated mode using lime for regeneration with
slipstream sulfuric acid treatment for sulfate control
concentrated mode using limestone for regeneration
dilute mode using lime for regeneration.
• Task III — Prototype Test Program — The test program was con-
ducted on the 20-megawatt CEA/ADL prototype dual alkali system
at Gulf Power Company's Scholz Steam Plant in Sneads, Florida,
from May 1975 to July 1976. The prototype system used lime in
a concentrated mode. The system was operated on flue gas gen-
erated from moderately low- to high-sulfur coals, and with
varying particulate loads to the system.
This is Volume II of the final report covering Tasks I and II of this
program. Volume III covers the Prototype Test Program, Task III. An
Executive Summary for the entire program has also been published as
Volume I.
B. RESULTS AND CONCLUSIONS
Laboratory and pilot plant work on dilute and concentrated modes, using
lime, indicates that these modes can be operated in a closed loop with
the following general performance as a minimum:
« S02 removal — 90% or greater.
• lime utilization — 90% or greater.
• waste cakes solids content — 45% or greater.
• sodium makeup requirements — less than 0.05 mols Na2C03/mol
of S02 removed.
The actual performance of any particular dual alkali process will vary
depending upon the S02 and oxygen concentrations in the flue gas, the
design of the system and the concentration of sodium solutions used in
the process. Using lime, some version of the dual alkali process can
generally be designed to far exceed many or all of the above performance
characteristics in most utility applications.
1-2
-------
Aside from concentration differences, the principal difference in the
operating characteristics between dilute and concentrated lime dual
alkali systems is that dilute systems operate at or near saturation
in calcium sulfate, potentially reducing the reliability and ease of
operation of the dilute systems. Dilute systems require the use of
carbonate makeup to provide some softening of the regenerated solution
prior to recycle back to the scrubber.
No viable approach was developed in this program to enable the use of
limestone for regeneration. The problem which remains to be solved is
the production of solids with good settling characteristics over a wide
range of sulfate, magnesium, and iron concentrations in the scrubbing
liquor. These components tend to reduce the rate of reaction of lime-
stone with sodium scrubbing solutions; solids properties tend to deteri-
orate with a decrease in the limestone reaction rate. However, these
problems may be resolvable. There are considerable economic incentives
for the substitution of limestone for lime in dual alkali processes,
justifying further work in this area.
More specific results and conclusions regarding the various dual alkali
modes and pilot plant operations are given below.
1. Pilot Plant 862 Removal and Oxidation - General
The pilot plant S02 removal and oxidation data are specific to the pilot
plant scrubber configuration as influenced by the scrubber operating tem-
perature for the pilot plant flue gas stream. The scrubber operating
temperature of 140-150°F is higher than that normally encountered in con-
ventional boiler flue gas applications (120-130°F). The elevated tempera^
ture in the pilot plant system tends to decrease S02 removal efficiency
due to elevated SC>2 partial pressures for any given solution, and tends
to increase oxidation rates. However, the purpose of the pilot plant
scrubber and its operations was to provide scrubber effluent with an
appropriate composition for use in the various dual alkali modes rather
than to generate basic data on SC>2 absorption using sodium solutions.
Within the above constraints, the scrubber operations did indicate that
S02 removal in excess of 90% is easily accomplished over a range of S02
inlet concentrations from 700-2,800 ppm by adjusting the scrubber feed
stoichiometry. To achieve this removal efficiency, a stoichiometry of
1.1 mols of active Na+ capacity/mol S02 inlet was required at the high
inlet SC>2 range; a stoichiometry of 1.3 was required in the lower inlet
S02 range. In any range of S02 concentration, increasing stoichiometry
increased the S02 removal. There was no important apparent effect of
active sodium concentration within a range of 0.2-0.5M or total dissolved
solids concentration within a range of 5-15 wt %.
Sulfite oxidation is mass transfer limited at active sodium concentra-
tions above 0.2M with the rate of oxidation increasing with the oxygen
content of the flue gas. At lower active sodium concentrations the
1-3
-------
oxidation rate is roughly proportional to the active sodium concentration.
The rate of oxidation decreases with increasing total dissolved solids; by
increasing TDS from 8-15 wt % to 25-35 wt %, the oxidation rate is reduced
by a factor of 2-3. At lower total dissolved solids in concentrated active
sodium systems (0.3-0.6M active Na+, 5-15 wt % TDS) sulfite oxidation can
be expected to be on the order of 100-300 ppm equivalent SC>2 removal for
oxygen concentrations in the flue gas ranging from 4 vol % to 8 vol %.
2. Concentrated Mode with Lime Regeneration
In the concentrated mode using lime for regeneration, calcium sulfate
will coprecipitate with calcium sulfite at sulfate precipitation rates
equivalent to oxidation rates as high as 25% of the SC>2 removal. Solu-
tions remain unsaturated with respect to calcium sulfate and have low
soluble calcium concentrations. Process modes can be operated over a
wide range of sodium solution concentrations achieving high SC>2 removal
(greater than 90%) producing good quality filter cake (45% solids,or
greater) containing low soluble solids (2-5 wt % dry cake basis) with
no sulfate purge required. The performance characteristics of concen-
trated lime regeneration modes are summarized in more detail below.
• S02 Removal — S02 removal efficiencies in excess of 90% were
easily achieved with the removal efficiency a function of sodium
solution feed stoichiometry for any particular absorber design.
In all closed-loop runs the feed stoichiometry (scrubber opera-
ting pH) was controlled to ensure better than 90% removal. For
a given design, a slightly higher feed stoichiometry (or opera-
ting pH) was required for high sodium solution concentrations
(30-35 wt % sodium salt solutions) than for moderate concentra-
tions (10-15 wt % sodium salt solutions) to achieve the same
removal efficiency because of the increase in S02 equilibrium
partial pressure with the increase in sodium sulfite/bisulfite
concentration.
• Lime Utilization — Lime utilization in the range of 95-100% can
be achieved with reactor holdup times of 25 minutes or greater
when regenerating to a pH of 8 or higher. High utilizations
can be achieved at shorter residence times if the regeneration
reaction is not carried beyond neutralization of the bisulfite.
Lime utilization decreases if regeneration is carried much
beyond a pH of 12.5.
• Oxidation/Sulfate Control — At active sodium concentrations
above about 0.2M, calcium sulfate coprecipitates with calcium
sulfite upon reaction of the sodium salt solution with lime.
The sulfate/sulfite content of the precipitated calcium salts
is related to the sulfate/sulfite concentrations in the reactor
liquor by the following relationship:
mols CaSOtj.
= 0.0365
mols CaSOs
'reactor , reactor
solids liquor
1-4
-------
This relationship describes the coprecipitation phenomenon
over the range of sulfite and sulfate liquor concentrations
used in laboratory and pilot plant experiments ( [SOg] > 0.2M,
[SO^/tSOg] = 0-6). This method of sulfate precipitation is
effective for oxidation rates up to about 25%. At any given
active sodium concentration, high sulfate precipitation appears
to be favored by either partial neutralization of the absorbent
solution or regeneration to pH's well above neutrality (>11.5),
thereby reducing the sulfite concentration in the reactor liquor
and maximizing the sulfate/sulfite ratio in the liquor.
In a properly designed concentrated dual alkali loop, the sulfate/
sulfite ratio will self-adjust at steady-state so that the rate of
sulfate precipitation equals the rate of sulfite oxidation. It is
possible to achieve this balance over a wide range of active sodium
and sulfate concentrations in dynamic response to changes in flue
gas rates and oxygen and S02 concentrations.
For dual alkali systems operating with high TDS (in the range of
25-30 wt % sodium salt solutions) oxidation rates can be reduced
by a factor of 2-3 from those encountered at lowered TDS levels
(10-15 wt %). At such high TDS levels, the active sodium con-
centrations as well as the sulfate concentration must be elevated
in order to promote effective regeneration reactions and produc-
tion of solids with acceptable dewatering properties. As a result,
sulfate precipitation capability is limited.
Solids Properties — Single-stage CSTR (continuous stirred tank
reactor) and multistage reactor systems can produce solids, over
a wide range of process conditions, which settle well and filter
to insoluble solids contents of 45 wt % or higher. When using a
CSTR as the regeneration reactor, solids properties deteriorate
as the regeneration reaction is carried to a higher pH range
with the degree of deterioration increasing from pH 7.5 to
pH 12. This effect is worse for reactor holdup times of 60
minutes than for shorter reactor residence times (30 minutes).
Using a CSTR, solids properties also decrease as the sulfate/
sulfite ratio increases in the reactor liquor (at higher oxida-
tion rates). In a single-stage CSTR, it is difficult to produce
solids with acceptable properties (45 wt % insoluble solids) at
process conditions consistent with sulfate precipitation and
sulfite oxidation rates much beyond 15%.
Good quality solids can be produced over a wider range of pH
and sulfate concentration using a two-stage reactor system,
consisting of a short residence time reactor (5-10 minutes)
followed in series by a longer residence time second stage
(20-40 minutes). This multistage system produces good solids
at pH levels up to about 12.5 and at sulfate/sulfite ratios
required for sulfate precipitation rates equivalent to about
25% oxidation.
1-5
-------
Sodium Losses — For a filter cake containing 50% insoluble
solids, the soluble solids content of the cake can be reduced
to 2-3% (dry cake basis) using the amount of filter cake wash
water which would be normally available when operating closed-
loop in a high-sulfur coal boiler application. At TDS levels
in the range of 10-15 wt %, two to three displacement washes
are effective in reducing the soluble content of the cake to
2-3 wt %. Of this material, 0.5-1.0 wt % soluble sodium salts
appear to be occluded in the calcium salt crystals and cannot
be washed regardless of the amount of wash water used. About
two to three displacement washes are available for high-sulfur
coal applications. At high TDS concentrations (30%), four to
five displacement washes are necessary to reduce solubles to
the 2-3% level. With only three displacement washes, solubles
losses at high TDS concentrations can be expected to be roughly
twice those expected when operating at 10-15 wt % TDS levels in
the absorbent solution.
At the lower TDS levels, sodium makeup requirements are on
the order of 2-3% of the total alkali requirement (mol basis).
That is, roughly 2-3% of the sulfur absorbed from the flue gas
leaves the system as sodium salts with the remainder as calcium
salts.
From the above considerations, operating a concentrated lime mode
with TDS in the range of 10-15%, the single-stage CSTR can pro-
duce good quality solids (45 wt % or greater) containing 2-3 wt %
solubles at system oxidation rates up to 15%. When using the mul-
tistage reactor system the operability of the process is extended
to oxidation levels in the range of about 25%. Increasing TDS
reduces oxidation but requires more wash water to produce the
same cake solubles content. At two to three displacement washes,
the solubles content of the cake is proportional to the TDS levels
in the system loop.
System Operability/Reliability — In concentrated modes using
lime for regeneration, soluble calcium concentrations range
from 15-90 ppm with the calcium concentration generally de-
creasing with increasing sulfite concentration. No scaling
or deposition of solids was observed in the scrubber loop
during any of the concentrated mode operations. Scrubber
operation and S02 removal were easy to control. The regen-
eration reaction is stable and easy to control, but should
be kept at a pH below about 8 if operating with a single-
stage regeneration reactor. Increasing the TDS level in
the system raises the sodium salt saturation temperature,
increasing the potential for solid sodium salt crystalliza-
tion in elements of the system which are permitted to cool.
1-6
-------
3. Concentrated Mode with Sulfuric Acid
Sulfate Treatment
The sulfuric acid slipstream treatment scheme is a technically feasible
and reliable approach for removal of soluble sulfates from dual alkali
systems. The basic chemistry of the treatment process is given in the
following simplified reaction equation:
2CaS03 • 1/2H20 (filter cake) + Na2SOi+ (system liquor) + H2SOit
(2)
+ 3H20 -»• 2NaHS03 + 20380^ • 2H20
The treatment produces sulfate in the form of gypsum that can be readily
dewatered to 65 wt % insoluble solids or higher. The scheme adds com-
plexity to any dual alkali mode to which it is applied. The complexity
is reflected in additional capital costs and in increased operating costs
for the sulfuric acid, the additional lime consumed and the additional
solid waste produced.
The amount of sulfuric acid required is important since it directly affects
the overall lime requirement. As the sulfuric acid addition rate increases,
the lime rate must increase accordingly for precipitation of the additional
sulfur value added to the system. The maximum efficiency of the treatment
scheme ((mols Na2S04 removed/mol H^SOi* fed) x 100%) appears to be practi-
cally limited to a maximum in the range of 60-70%. In order to precipitate
sulfate at a rate sufficient to keep up with an oxidation rate of 15% (of
the S02 absorbed) , the lime feed requirement will be increased by 25% for
a 60% reactor efficiency.
The efficiency of the sulfuric acid treatment is importantly affected by
the calcium utilization achieved in the absorbent regeneration reactor in
the main dual alkali loop. As calcium utilization decreases in the main
loop the efficiency of the sulfuric acid slipstream treatment decreases
and acid consumption increases to neutralize unreacted lime in the filter
cake. In order to achieve a 50% efficiency in the sulfuric acid treatment
system, calcium utilization in the main dual alkali loop must exceed 90%.
Because the use of this sulfuric acid treatment scheme may be costly when
applied to systems with high oxidation rates (due to the sulfuric acid and
extra lime requirements) , it may be more appropriate for systems with inter-
mediate levels of oxidation where the rate of sulfate formation cannot be
easily handled in a simpler concentrated sodium mode. The consequences
of using the sulfuric acid slipstream treatment approach for sulfate regen-
eration should, therefore, be carefully evaluated in terms of the overall
process operation. In many cases, where oxidation rates are high enough
that they cannot be easily handled by normal concentrated mode operation,
other dual alkali approaches, such as the dilute lime system described in
Chapter VII, might be more promising than a sulfuric acid treatment scheme.
1-7
-------
4. Concentrated Mode with Limestone Regeneration
No viable approach was found for use of limestone in a concentrated dual
alkali mode. Through the laboratory and pilot plant efforts allocated
to work on the concentrated limestone mode, we were not able to develop
process parameters and reactor conditions consistent with good limestone
utilization and generation of an acceptable quality of waste solids. The
work did, however, uncover important factors influencing the limestone
regeneration reaction that indicated promising areas of future work. Un-
like results from work on use of limestone in dilute modes (Chapter VII) ,
the potential for technical success argues for additional work on the con-
centrated limestone dual alkali mode, especially when the economic incen-
tives (presented in Chapter II) are considered.
Limestone is substantially less reactive toward sodium salt solutions than
is lime, even when reacting with relatively acidic scrubber bleed solutions.
The reaction rate of the limestone regeneration reaction is dependent upon:
• nature of the limestone and its particle size;
• reactor temperature and residence time;
m concentrations of soluble reactants (sodium sulfites, sodium
bisulfites and sodium sulfates); and
• the presence, at low concentrations, of trace constituents
such as magnesium and iron, which influence the reaction rate.
Increase in the reaction rate was generally consistent with improvement
in the dewatering properties of the solids produced and with improved
utilization of limestone.
Three limestones, with similar particle size distribution, were examined —
Fredonia limestone used in the EPA/TVA Shawnee program; another, locally
available, natural limestone; and reagent grade CaCC>3. Of these, the
Fredonia limestone was amorphous, rather than crystalline in nature, and
was considerably more reactive than the other two limestones examined.
The Fredonia limestone, therefore, was used extensively in the laboratory
and pilot plant programs.
Laboratory experiments indicated that increasing temperature importantly
increased the reaction rate. However, the pilot plant was not equipped
for heating the reactors or for heating the reactor feed. As a consequence
pilot plant regeneration was performed at a maximum of about 50°C.
The dewatering properties of solids were generally observed to deteriorate
as the regeneration reactor residence time was increased. Increasing the
reactor residence time results in carrying out the reaction closer to the
equilibrium conditions and consequently at a lower driving force and reac-
tion rate. Use of multistage reactor systems, containing several stages
1-8
-------
with residence times in the range of 15 minutes, were found to produce
solids with a quality superior to that of solids produced in fewer reac-
tors with the same total residence time. Recycle of solids, increasing
reactor solids concentrations from about 2 wt % to 5 wt %, improved lime-
stone utilization but did not appreciably improve the quality of the solids,
Under controlled conditions, with a multistage reactor system operating at
about 50°C, it was possible to produce solids with acceptable dewatering
properties (45% insoluble solids) and to achieve limestone utilizations
on the order of 75%. However, as the sulfate concentration in the loop
rose above 0.7M, or, the magnesium concentration rose much above 300 ppm,
the reaction rate and the resulting limestone utilization and solids
properties all deteriorated.
Sulfate concentration in the reactor liquor had a much more important,
deleterious effect on the reaction rate and solids properties in lime-
stone regeneration reactions than the similar effects of increased sul-
fate concentration observed in concentrated lime regeneration. As in
lime regeneration, the reaction rate is inversely proportional to the
ratio of sulfate/sulfite concentrations in the liquor; but the rate drops
dramatically using limestone as the sulfate concentration exceeds 0.7M at
TOS levels of 0.3-0.5M. Operation at lower sulfate/sulfite ratios tends
to limit sulfate precipitation in this mode and limit the range of oxida-
tion in which limestone regeneration could be operated closed-loop.
Calcium sulfate coprecipitates along with calcium sulfite in concentrated
limestone regeneration reactions in an analogous fashion to the coprecipi-
tation of calcium sulfate observed in the concentrated lime regeneration.
However, pilot plant data indicate that for the same sulfate/sulfite con-
centrations with the same range of TOS in the feed liquor (i.e., [TOS] =
0.3-0.5M), lower sulfate precipitation occurs when using limestone, as
given by the following:
CaSO
- - ) = 0.022 - - (3)
CaS°3 reactor \ [S°3] /reactor
solids liquor
The sensitivity of the reaction to high sulfate concentrations and the
lower sulfate precipitation rates make limestone regeneration less viable
for closed-loop operation than lime regeneration at higher oxidation rates.
The presence of Mg4"1" in solution, introduced into the system in varying
amounts depending upon the magnesium content of the limestone, also can
have a retarding effect on the limestone regeneration reaction rate, re-
sulting in poor solids quality and limestone utilization. This effect
becomes pronounced as the Mg4"4" concentration rises much above a few hundred
ppm. Relatively low magnesium limestones, such as Fredonia limestone
(1.0-1.5 wt % Mg as MgCOs) would result in concentrations on the order
of 'a few thousand ppm, at steady-state, in a concentrated dual alkali loop.
1-9
-------
Laboratory work confirmed that magnesium concentrations could be controlled
by reacting part of the process stream with lime to precipitate Mg(OH>2-
However, such an approach would reduce operating cost savings, requiring
part of the total regeneration to be performed using lime. Use of lime
with limestone would increase the complexity and the capital cost in a
manner similar to that discussed in Chapter VI for dilute limestone/lime
dual alkali systems, eliminating economic incentives.
In pilot plant operations, iron from corrosion of unlined steel equipment
was found to have an effect similar to that of magnesium on the limestone
regeneration reaction at pH's below about 6. At higher pH's, Fe(OH>3 is
highly insoluble, limiting the buildup of iron in solution. By selection
of proper materials of construction and linings and by carrying the lime-
stone regeneration beyond a pH of 6, interference by iron can be eliminated
in concentrated limestone modes.
Future work on limestone regeneration should be directed at increasing
reaction rates at high magnesium levels by increasing sulfite concentra-
tions, reactor temperature and by staging of the reactors.
5. Dilute Mode with Lime and Limestone Regeneration
Use of limestone only for the regeneration of solutions in the dilute mode
(less than 0.15M active sodium) is not viable. The limestone reaction rate
decreases as the ratio of soluble sulfate/sulfite increases in the reactor
solutions. At sulfate/sulfite ratios required for adequate sulfate pre-
cipitation in the dilute mode, reaction rates are poor and result in poor
limestone utilization and poor solids quality.
Use of lime in combination with limestone in a dilute dual alkali mode
was more viable technically. In this approach, the lime regeneration was
carried out in a second reaction system to promote sulfate precipitation.
The limestone/lime process is more complicated than a simple dilute lime
process, resulting in higher projected capital cost. Economic analysis
indicated that operating cost savings which could potentially be realized
in using limestone for part of the regeneration would not offset the ad-
ditional capital cost probably required to enable use of the limestone.
The dilute lime system, using soda ash for softening, was technically and
economically the most viable dilute mode considered. Conclusions based
upon laboratory and pilot plant investigations of this mode are given below.
A dilute lime mode can be operated in a closed loop with sulfate precipi-
tation keeping up with any level of system oxidation. The system can be
operated with high S02 removal (90% or higher) and good lime utilization
(90% or higher) to produce high quality solids (60% insolubles or higher)
with low soluble sodium losses (2% achievable). The process may be more
appropriate for low-sulfur coal applications or in situations where oxi-
dation rates are expected to exceed 25-30% of the S02 removal. The dilute
lime mode is somewhat more complicated than the concentrated lime mode, in-
volving higher liquid rates and larger reactors and associated equipment.
The process is also potentially less reliable than the concentrated lime"
approach.
1-10
-------
The regeneration reaction, carried out at low sulfite levels, results in
the precipitation of calcium sulfate (usually gypsum) to produce a regen-
erated solution of sodium hydroxide and sodium sulfate with soluble calcium
levels which are, at best, at the saturation level of about 700 ppm Ca"1"*".
Even with moderate amounts of soda ash makeup (and resulting softening
by precipitation of calcium carbonate) the solutions have soluble calcium
levels in the range of 600-700 ppm with a high potential for scaling in
the system. Close control of scrubber pH is required to prevent carbonate
or sulfite scaling. High scrubber oxidation rates may create sulfate scaling.
In the dilute mode regeneration reaction, there is a high tendency to produce
solutions which are supersaturated in Ca"1""1" (with respect to gypsum) . Using
a single-stage CSTR with no solids recycle, calcium supersaturation levels
of 100-200 ppm are easily achieved. Special design precautions must be
taken to prevent supersaturation and the resulting scaling throughout the
system. Supersaturation can be reduced in a number of ways, by reactor
system design and by controlling conditions of the regeneration reaction:
• Increased reactor residence time — Allows time for completion of
reaction and desupersaturation. Holdup time of 60 minutes is a
minimum; 90 minutes is preferable.
• Solids recycle — Increases suspended solids concentration and
seed concentration for reaction and desupersaturation. Recycle
of solids to achieve a concentration of 4% or higher suspended
calcium salt solids is required to eliminate supersaturation in
the reactor effluent.
• Oxidation of sulfite in scrubber bleed prior to regeneration —
Lowers the concentration of TOS which tends to retard the lime/
sulfate reaction when TOS is present in the dilute mode concen-
tration range. Oxidation to TOS concentrations of about 0.02M
or lower is desirable.
• Multistage reactor configuration — Solids generated in a short
residence time first stage provide good seeds for completion of
reaction in longer residence second stage. Using a multistage
reactor can reduce supersaturation to within about 50 ppm of
the saturation level. Solids recycle is required to completely
eliminate supersaturation.
Elimination of supersaturation was achieved in the single-stage reactor,^
with 90 minutes residence time; using solids recycle to the minimum of 4%
suspended calcium salts in the reactor; and with oxidation of the reactor
feed solution to TOS levels of 0.02M or lower. Variation in soluble sul-
fate concentrations in the range of 0.50-0.75M had no apparent effect on
the level of supersaturation.
Utilizing these design factors in a dilute mode with lime regeneration
not only reduces or eliminates supersaturation, but also promotes a good
reaction rate which generally improves the overall process performance
1-11
-------
parameters such as lime utilization, sulfate precipitation and solids
properties. More specifically, the performance of the dilute lime mode
relative to the important process performance characteristics is given
below:
• S02 removal — S02 removal of 90% is easily achieved especially
at low to medium inlet SC>2 levels. SC>2 removal is not as effi-
cient as in a concentrated dual alkali mode (with the same scrub-
ber configuration) because of the low active sodium concentration.
The scrubber operation is more difficult to control due to the low
buffering capacity of the dilute mode liquors. Higher calcium con-
centrations (in the range of 600-700 ppm Ca4"1") present potential
scaling problems in the scrubbing system. Operation of the scrub-
ber in a high pH range (9-11) to promote good S02 removal results
in some C02 absorption and potential carbonate scale formation.
Increasing active sodium concentrations to provide more buffering
can result in sulfite scale formation in the pH range of 8-11.
• Lime utilization — Lime utilization of 90% or higher is achievable
when regenerating to hydroxide concentrations of about 0.1M with
solutions containing sulfate in the range of 0.50-0.75M and using
reactors with a minimum total holdup time of 60 minutes. Utiliza-
tion increases as the residence time and sulfate concentration are
increased. Solids recycle also helps increase lime utilization.
However, TOS levels in the feed to the reactor should be 0.02M or
less (by deliberate oxidation if necessary) to prevent retarding
of the reaction rate by the sulfite.
• Oxidation/sulfate control — Complete sulfate control is possible
in this mode of operation at any rate of oxidation in the system.
However, at very high scrubber oxidation rates, sulfite/bisulfite
buffering is minimal and scrubber pH control becomes difficult.
All other aspects of the process operation are improved by high
oxidation rates (i.e., minimal TOS concentration in the feed to
the regeneration reactor). Deliberate oxidation should be used
to maintain TOS levels below about 0.02M. At sulfate concentra-
tions in the range of 0.50-0.75M, calcium sulfate (usually gypsum)
is produced instead of a mixed calcium sulfite/calcium sulfate
crystal, when TOS is maintained at or below 0.02M. At this point
the calcium sulfate content of the solids is no longer limited by
the apparent maximum content of 25-30% in the mixed crystal; 100%
calcium sulfate can be produced.
e Solids properties — It is possible to produce excellent quality
solids containing 60-80% insoluble solids. Good solids properties
are favored by the following conditions:
Low TOS in the reactor feed — less than 0.02M.
High sulfate in the reactor feed — 0.50-0.75M
(high end of range favored).
1-12
-------
",lmproves solids ^ality but increases the
load and ultimately the size of the thickener.
- Multistage reactor system - improves solids quality compared
to same total residence time in a single stage.
- High reactor residence time — 80% insoluble solids can be
produced using a 90 minute residence time reactor.
• Sodium losses — In any application, increasing the insoluble
solids content of the filter cake increases the effective number
of displacement washes for any given amount of wash water avail-
able. By producing 75% insoluble solids in a high-sulfur coal
application, roughly five displacement washes are available (as
opposed to two and one-half displacement washes at 50% solids)
permitting more effective cake washing; in low-sulfur coal appli-
cations even more wash water can be available. Consequently
sufficient wash water should be available to reduce the solubles
content of the cake to under 2%; and down to the range of 0.5-1.5%
solubles in low-sulfur coal applications. In such applications
it may be possible to wash the filter cake to loss levels lower
than those corresponding to sodium carbonate makeup levels re-
quired for softening of the regenerated liquor. A sodium car-
bonate makeup rate of 2.0-2.5% of the S02 removal rate provides
sufficient carbonate to reduce the Ca^ concentration in the re-
generated liquor by about 50 ppm, providing only minimum soften-
ing. Thus, sodium makeup (and ultimately the losses in the cake)
may be controlled by softening requirements rather than by wash
water availability or cake washability.
• System operability/reliability ~ The dilute lime mode is inherently
less reliable and more difficult to control than the concentrated
lime mode. When appropriate care is taken to eliminate supersatura-
tion, the calcium levels in the regenerated solution are in the
range of 700 ppm. Only a minimum of softening is provided at low
sodium carbonate makeup levels. Potential for scaling exists in
the reactor system and associated auxiliaries and piping, and in
the absorber. Absorber operation is less effective and more dif-
ficult to control than in the concentrated mode.
6. Solids Characterization — Dilute and Concentrated
Lime Regeneration Modes
Limited testing was performed to characterize the basic physical and chemical
properties of ash-free waste filter cakes produced in the two most successful
dual alkali modes piloted — concentrated and dilute active sodium modes with
lime regeneration. Testing included: analysis of major chemical constituents;
crystalline morphology via X-ray diffraction and scanning electron microscopy;
unconfined compressive strength and optimum dry density; permeability; leach-
ing behavior; and the effects of treatment with lime (or portland cement) and
fly ash on the physical properties.
1-13
-------
The concentrated mode filter cake that was tested was produced in the
prototype system using the two-stage reactor. The cake was a mixture
of calcium sulfite and sulfate (about 15% calcium sulfate) and contained
55% solids. The crystalline structure of the solids was rosette-like
agglomerates of needles characteristic of the concentrated mode opera-
tion. X-ray diffraction data and chemical analyses indicate that the
calcium sulfite and calcium sulfate were coprecipitated as a mixed crystal
of hemihydrate salts. There was no evidence of gypsum (CaSO^ • 2H20) in
the solids.
The dilute mode filter cake was essentially pure gypsum produced in the
pilot plant under conditions of intentional oxidation. The solids crystals
were monoclinic and the filter cake contained approximately 80% insoluble
solids.
The mixed sulfite/sulfate solids had the appearance and physical proper-
ties similar to a silt-like soil and handled much like a moist powder.
The gypsum, on the other hand, was much more grainy and had the consis-
tency of a sandy soil. The unconfined compressive strengths of both
materials were in the range of typical soils, 10-15 psi; and both had
optimum dry densities in the range of 75% solids. The coefficient of
the permeability of the compacted sulfite/sulfate solids ranged from
about 3 x lO""4 to_5 x 10~5 cm/second. The permeability of dual alkali
gypsum was 2 x 10 ^ cm/second. These values are within the range of
published data on coefficients of permeability of gypsum and sulfite-
rich solids produced in FGD systems.*>2
The treatment of the sulfite/sulfate filter cake was studied using various
mixtures of lime (or portland cement), filter cake, and fly ash. This work
showed that the concentrated mode solids could be treated in a fashion simi-
lar to the treatment of solids from direct lime and limestone scrubbing sys-
tems with similar effects on the mechanical properties. Testing performed
on prototype system concentrated dual alkali solids by IU Conversion Systems
(IUCS) indicates that the coefficient of permeability of Created filter cake
was about 5 x 10~6 cm/second using standard treatment mixes.2
Accelerated leaching tests and elutriate analyses performed on untreated
samples both at ADL and by IUCS showed that the initial and "steady-state"
concentrations of soluble species that can be leached, notably total dis-
solved solids (TDS) and total oxidizable sulfur (TOS), will be very dependent
upon the initial conditions and composition of the solids (degree of cake
washing, ratio of sulfate-to-sulfite, chloride concentration in the gas,
etc.) and the manner of solids handling and disposal. TDS levels in the
initial leachate can range from a few thousand ppm to, about ten thousand
ppm, and "steady-state" concentrations (after the first few pore volume
displacements) can vary from a few hundred ppm to approximately a thousand
ppm. Similarly, TOS levels can range from essentially nil to up to 50 ppm.
These concentrations are consistent with the range of published data for
leachates from solids generated in direct lime and limestone scrubbing
systems.
1-14
-------
Testing performed by IUCS on the treatment of the filter cake indicated
significant reductions in both initial and "steady-state" levels of IDS
in leachates. Depending upon the type of treatment, reductions of 50%
to 80% were observed.
In all physical properties testing performed at ADL, samples were pre-
pared in accordance with standard soil-mechanics testing procedures.
These procedures required, as a part of the sample preparation, the,
drying and rewetting of the filter cake to achieve a desired solids
content. While the samples were dried at temperatures of 83°C to prevent
loss of water of hydration, there is still concern that the drying/rewetting
procedure resulted in some changes in the behavior of the material, particu-
larly in the case of the rosette-like crystals produced in the concentrated
mode operation. However, the results of these limited tests are believed
to be indicative of the general behavior of the dual alkali solids. More
exhaustive testing on both as-received samples and samples prepared in
accordance with standard soil testing procedures is required to assess
the effects of sample preparation on test results.
1-15
-------
II. INTRODUCTION
A. BACKGROUND AND OBJECTIVES
Dual alkali or double alkali are generic terms used to describe flue gas
desulfurization (FGD) processes involving absorption of S02 using a soluble
alkali, followed by reaction of the scrubber effluent solution with lime
and/or limestone, to produce a solid calcium sulfite/sulfate salt and to
regenerate the soluble alkali scrubber feed solution. Many versions of
dual alkali processes are under development. Most of the advanced systems
utilize sodium as the soluble alkali, but ammonia-based and other systems
also are receiving development effort. In addition to the type of soluble
alkali used, another important difference among the dual alkali processes
is the concentration of soluble alkali in the absorption/regeneration loop.
The concentration and the choice of soluble alkali are importantly related
to the reactions involving soluble sulfate produced in the system and to
the control of soluble solids losses from the system. Many of the various
schemes for dual alkali processes and the general chemistry and technology
have been reviewed by Kaplan at a recent EPA FGD Symposium.3 A glossary of
dual alkali terminology is included in this report.
Work in the EPA/ADL Dual Alkali Program was limited to sodium-based dual
alkali systems, covering a wide range of active sodium concentrations
using lime and/or limestone for regeneration. Special attention was given
to the formation and precipitation of sulfate and to the control of the
soluble solids content of the calcium salts which are produced for disposal.
The objectives of the EPA/ADL Dual Alkali Program were to define promising
sodium-based dual alkali modes of operation and characterize those modes
with regard to S02 removal, yields on reactants, sulfite oxidation, sulfate
precipitation, waste solids characteristics, soluble solids losses, and
overall process reliability.
The program was divided into three task areas as follows:
• Task I
- ADL Laboratory Studies — Performed at ADL laboratories,
Cambridge, Massachusetts — Regeneration of concentrated
sodium scrubbing solutions using lime regeneration, lime-
stone regeneration, sulfuric acid treatment for sulfate
removal, detailed characterization of the physical and
chemical properties of solids produced in these regener-
ation modes.
- EPA Laboratory Studies - Performed at EPA's Industrial
Environmental Research Laboratory, Research Triangle Park,
North Carolina - Regeneration of dilute sodium scrubbing
solutions using lime regeneration, limestone regeneration
and lime/limestone regeneration.
II-l
-------
Task II
ADL Pilot Plant Studies — Performed at CEA/ADL Pilot Facility,
Cambridge, Massachusetts — Promising modes of operation using
dilute and concentrated sodium scrubbing solutions with lime
regeneration, limestone regeneration, and sulfuric acid treat-
ment for sulfate precipitation.
• Task III
Test Program — Performed at the prototype dual alkali system,
Plant Scholz, Gulf Power Company/Southern Company Services, Inc. —
Lime regeneration of concentrated sodium scrubbing solutions using
the 20-megawatt CEA/ADL system.
Consistent with the program objectives, the important process characteris-
tics were determined for each of the dual alkali modes investigated in each
task.
B. DESCRIPTION OF CHEMISTRY AND DEFINITION OF TEEMS
In the absorption section of sodium-based dual alkali processes, absorp-
tion of S02 in sodium sulfite solutions occurs to produce a bisulfite
scrubber effluent solution according to the overall reaction:
Na2S03 + S02 + H20 J 2NaHS03 (4)
Depending upon the dual alkali mode being used, the feed to the absorber
may also contain some sodium hydroxide (formed in the regeneration section
or used as sodium makeup) and/or sodium carbonate (used as sodium makeup
to the system). Both sodium carbonate and hydroxide form sodium sulfite
on absorption of S02:
Na2C03 + S02 -> Na2S03 + C02 (5)
2NaOH + S02 -> Na2S03 + H20 (6)
which is used in further absorption to produce bisulfite. The absorber
feed solution will also contain some level of sodium sulfate in solution
and may contain some sodium bisulfite if neutralization is not completed
in the regeneration section. Sodium, identified as associated with anions
involved in S02 absorption reactions, is referred to as "active" sodium
(includes sodium sulfite, bisulfite, hydroxide and carbonate/bicarbonate).
The sulfite/bisulfite content of solutions, or total oxidizable sulfur
content, is also referred to as TOS.
Some oxidation of sulfite to sulfate occurs in the absorber due to reaction
of sulfite with oxygen in the flue gas:
2Na2S03 + 02 ^ 2Na2S04 (7)
II-2
-------
The rate of oxidation or oxygen transfer in the absorber is a function
of the absorber design, oxygen concentration in the flue gas, flue gas
temperature, and the nature and concentration of the species in the
scrubbing solution. For a given set of process parameters, the oxida-
tion rate in mols of sulfite oxidized per unit time is relatively in-
dependent of the S02 removal rate. The oxidation rate can be expressed,
on an equivalent basis, as a .percentage of the S02 removed. As an example,
for a high-sulfur utility boiler flue gas containing about 4-5% 02, at about
2,500 ppm S02 removal, approximately 10% of the S02 removed from the flue gas
will show as sulfate in the scrubbing solution and the remaining 90% will
show as sulfite/bisulfite. At steady state, this sulfate make, equivalent
to 10% of the S02 removed, must leave the system, either as calcium sulfate
or as a purge of sodium sulfate, at the rate at which it is being formed
in the system.
Under similar conditions of absorber design, oxygen concentration and solu-
tion characteristics, but at a much lower S02 removal rate (as in a low-sulfur
coal application) , a similar rate of oxygen absorption/reaction will result
in a much higher equivalent percentage of the S02 removed — perhaps as high
as 50% of the S02 removed. Also, in industrial boiler applications, where
operation at higher excess air often occurs , high equivalent oxidation rates
can be encountered even in high-sulfur coal applications. As a consequence,
in order to avoid unacceptable sodium sulfate purge requirements from dual
alkali systems , modes of operation capable of precipitating both sulfite
and sulfate are required; the sulfate and sulfite must leave the system
as solid calcium salts with sodium sulfate and sodium sulfite losses minimized.
The regeneration of acid sodium sulfite/sulfate scrubber effluent solutions
can be considered as a sequential reaction first involving neutralization
of the bisulfite using either lime or limestone, to produce a precipitate
of calcium sulfite:
2NaHS03 + Ca(OH)2 •+ Na2S03 + CaS03 + 2H20 (8)
2NaHS03 + CaC03 t Na2S03 + CaS03 + C02i + H20 (9)
In theory, the lime neutralization reaction should go to completion; com-
plete neutralization of bisulfite is not possible with CaC03. Using lime,
the regeneration can be carried beyond neutralization to generate caustic:
Na2S03 + Ca(OH)2 * 2NaOH + CaS03 (10)
to some equilibrium hydroxide concentration. The usual form of calcium
sulfite produced is the hemihydrate, CaS03 • 1/2H20.
Depending upon the concentration of sulfite and sulfate and the pH of the
solution, the following reaction for sulfate removal also occurs simul-
taneously with neutralization and regeneration reactions (8) - (10) using
either lime or limestone:
(11)
II-3
-------
The form of this calcium sulfate was investigated and will be discussed
later in this report.
Thus, the level of sulfate precipitation in the overall scheme is given
by the ratio of calcium sulfate to total calcium/sulfur salts produced:
mols CaSOtf
Sulfate Precipitation = mois caSO^ + mols CaS03
If the level of sulfate formation or sulfite oxidation given by:
mols S03 oxidized
Sulfate Formation
mols S02 removed
is matched by the level of sulfate precipitation, then all sulfur removed
from the flue gas can leave the system as a calcium salt and no soluble
sulfate purge is necessary to maintain a sulfate balance in the system.
In practice, even if such a balance is established, the washed calcium
sulfite/sulfate salts will contain some soluble sodium salts as well as
soluble fly ash constituents which must be purged and some sodium makeup
to the system will therefore be required. Calcium utilization or yield
in the overall process is defined as:
(mols CaS03 + mols CaSO^) generated
Calcium Utilization = : x 100%
mols Ca fed
regardless of whether lime or limestone is used.
Dual alkali modes can be designed to accommodate varying levels of oxida-
tion, with equivalent precipitation, up to an oxidation rate of 100%. For
example, in the limit, a solution of about 0.5 molar sodium sulfate can
be reacted with lime according to the following equation:
Na2SOij + Ca(OH)2 + 2H20 £ 2NaOH + CaS04 • 2H20 (12)
to produce gypsum and a solution containing approximately 0.10-0.15 molar
sodium hydroxide and 0.45 molar sodium sulfate. If this solution, con-
taining this relatively dilute concentration of active alkali, is recycled
to the absorber where 100% oxidation recurs, then a 0.5 molar sodium sulfate
solution is returned to the regeneration system and the cycle is repeated.
A dual alkali system can be designed on this basis.
As the sulfite concentration in the scrubber effluent is increased, in-
creasing the sulfite/sulfate concentration ratio in solution, the percent
sulfate precipitation decreases. However, the higher concentration of
active sodium species (sulfites, hydroxide and carbonates) reduces both
the scrubber liquid feed rate requirement and the feed rate to the regen-
eration system, for a given S02 removal rate. As the concentration of
II-4
-------
actxve sodium increases, the regeneration system size is thereby reduced
as are the liquid rates throughout the system. Thus, as the sulfate pre-
cipitation requirements decrease, the system size and recirculation rates
can be reduced by increasing active sodium concentrations.
Regeneration reactions in the "concentrated mode" produce solutions satu-
rated (or possibly supersaturated) with respect to calcium sulfite; in the
"dilute mode" saturation (or possibly supersaturation) is with respect to
calcium sulfate (gypsum). For purposes of this program, dilute operating
modes are considered to be those involving solutions containing active
sodium concentrations less than or equal to 0.15 molar active Na+, where
active sodium is sodium sulfite/bisulfite, carbonate /bicarbonate or hydrox-
ide. Concentrated modes are those involving solutions containing active
sodium concentrations greater than 0.15 molar active Na+. Soluble calcium
levels in dilute mode regenerated solutions are quite high compared to
levels in the concentrated mode, usually requiring "softening" with car-
bonate
Ca44- + CO" t CaC03 (13)
to prevent scaling by precipitation of calcium salts in the scrubber.
In the absorption of S02 with a regenerated solution, a reduction in pH
occurs, converting S0| to HSO^, thereby increasing the Ca*"1" solubility
relative to CaSOs due to the reduction in S0| concentration. Therefore,
the solution becomes less saturated relative to CaSOa as the pH drops
below about 9. A similar reduction in saturation relative to CaSOif does
not occur until the pH is reduced below about 3, where conversion of SO^
to HSO^ begins to become significant. Scrubbers for sodium dual alkali
systems are generally designed to operate above a pH of 5. Thus, dilute
solutions, saturated (or supersaturated) with CaSO^ do not benefit from
the pH reduction for prevention of scale in the scrubber.
As will be discussed in more detail, the sulfite and sulfate concentra-
tions, besides affecting sulfate precipitation, also have a profound effect
on the physical properties of the calcium/sulfur salts. As a result, the
settling and filtration characteristics of the solids produced in dual alkali
modes are also affected.
As an alternate approach to operating at decreased active sodium concen-
trations, high levels of sulfate precipitation can be achieved using sul-
furic acid treatment for precipitation of sulfate according to the following
equation :
2CaS03 • 1/2H20 + 3H20 $ 2NaHS03 + 2CaSOk • 2H20
This reaction is carried out at a low pH (2-3), where sulfite is converted
to bisulfite, thereby bringing calcium sulfite into solution and exceeding
the solubility product for calcium sulfate. This reaction can be applied
on a process slipstream for sulfate precipitation, as required. However,
in the overall process this scheme utilizes sulfuric acid, requiring
II-5
-------
additional lime or limestone for ultimate neutralization of this acidity
added to the system and generates an equivalent additional amount of waste
calcium/sulfur salts. This scheme is being practiced commercially in Japan
and offers the advantage of operating a concentrated active sodium absorp-
tion/regeneration system with the ability to sustain high sulfate oxidation
levels with equivalent precipitation of calcium sulfate.
The use of limestone rather than lime in the dual alkali process has im-
portant economic incentives, especially for high-sulfur coal applications.
For a typical application:
• Coal
• Load Factor
• Heat Rate
• Lime
• Limestone
• Waste Disposal
3.5% S; 12,000 Btu/lb
80%
10,000 Btu/kwh
$40/ton CaO; 95% utilization
$5/ton CaC03; 85% utilization
$10/ton dry
at 90% S02 removal, operating costs for the lime-based dual alkali system
are 6.8c/MMBtu higher than if using limestone. At a total fixed cost of
20%/year of the capital investment (including maintenance) , the operating
cost savings with limestone would support an additional capital investment
of almost $24/kw relative to capital cost for lime dual alkali. (For a 1% S
coal, all other factors being similar, operating cost savings would offset
an additional capital cost of only about $7/kw.) As a result, considerable
effort was directed at development of a viable limestone dual alkali process
in this program, even if the use of limestone increased the process complexity.
The various dual alkali modes are discussed in more detail in appropriate
chapters of this report. A glossary for the dual alkali terminology is
presented in each volume of this report.
II-6
-------
III. LABORATORY AND PILOT PLANT SYSTEMS
A. LABORATORY METHODS
The effects of experimental variables on the kinetics and equilibria of
several different reactions related to dual alkali regeneration have been
studied during the course of the laboratory experimental program. For
each of the several basic reactions studied, the general approach was
first to conduct a series of batch experiments in which samples of the
reacting mixture were taken periodically and analyzed to determine reac-
tion rates and final equilibrium compositions. Additional studies were
then carried out in a continuously fed stirred tank reactor (CSTR) to
investigate the reaction further under conditions more representative
of those in a continuous process situation.
Although a variety of reactions was studied, the experimental procedures
were generally quite similar, since in most cases a solid reagent was
reacted with a solution to produce a two-phase product slurry. The general
experimental procedures used are presented in this section. In a few in-
stances where a specific procedure was employed for a certain experiment,
the procedural details are discussed along with the experimental findings.
Determination of the chemical, physical and engineering properties of dual
alkali product solids which are presented in Chapter VIII required that a
number of special-purpose tests be performed; the details of those tests
are presented in that chapter.
1. Experimental Apparatus, Operation and
Sampling Procedures
Batch experiments were carried out in a 1-liter, three-necked, round-bottom
flask, immersed in a controlled-temperature water bath. A magnetically
driven stirrer provided mixing. To prevent oxidation during the course
of the reaction, a slow constant purge of nitrogen was maintained over
the reactor headspace.
Batch experiments were initiated by weighing appropriate amounts of the
water soluble reactants (reagent grade materials were used exclusively)
into the flask, adding the required amount of water, initiating the
nitrogen purge, and stirring to effect dissolution. When dissolution
was complete, a sample of the solution was first pipetted from the flask
for analysis to establish the solution composition accurately, and then
the solid reactant was added to the solution to initiate the reaction.
Periodically, the reactor was opened momentarily to remove a sample of
the reaction slurry for analysis; procedures used for phase separation
and analysis are described below.
The CSTR system used was a rather simple apparatus shown schematically
in Figure III-l. The reactor was a 1.5-liter glass vessel, commonly known
as a resin kettle, fitted with an overflow sideann positioned to yield a
reactor volume of about 1 liter. Mixing was performed with an electrically
III-l
-------
H
I
to
CX3
Solution Feed
Tank
Peristaltic Pump
OK)
Slurry Feed
Tank
Feed Sampling Valves
T—
Peristaltic Pump
Thermometer
Baffle
Water Bath
(stirred, electrically heated)
Overflow
FIGURE 111-1 CONTINUOUSLY FED, STIRRED TANK REACTOR SYSTEM
-------
driven stirrer, operating at about 700 rpm. Four rectangular pieces of
Plexiglas (T.M. of Rohm & Haas Co.) about 2.0 cm wide and 0.3 cm thick,
installed vertically around the sides of the reactor, served as baffles
to ensure good vertical mixing. The lower portion of the reactor was
immersed in a constant temperature bath for temperature control.
Reactant feed lines, a thermometer, an inlet for the nitrogen purge, and
the stirrer shaft all entered the reactor through air-tight, ground glass
joints in the lid; the lid and reactor bottom had mating ground glass
surfaces which could be assembled air tight. During operation, the reac-
tor headspace was purged with a flow of nitrogen which exited via the over-
'flow sidearm.
Reactants were prepared and fed from two, stirred, 14-liter glass vessels
with hermetically sealed stainless steel tops. The solid reagent was
slurried with water, and fed from one tank (baffled) and the sodium sul-
fite/bisulfite/sulfate solution from the second. Reactants were pumped
at controlled rates by variable speed peristaltic pumps (Sigma motor T6S
and Masterflex 7545).
A list of all laboratory experiments is presented in Appendix A.
2. Analytical Procedures
The feed and effluent samples were collected and analyzed by procedures
which permitted mass flow rates into and out of the reactor to be deter-
mined for each of the various chemical species of interest. Samples of
solutions and slurries were collected for known periods of time in tared
flasks and the total mass collected was determined gravimetrically- The
slurries were then filtered, and the solids were washed thoroughly with
a saturated calcium sulfate solution. The weight of the resulting wet
cake was determined and a representative portion was taken and dried at
85°C. From the weights determined above, the measured weight loss upon
"drying, and the solution density, the mass flow rates of both phases of
the slurry could be calculated.
Aliquots of liquids and dried solids were characterized by a variety of
analytical techniques, most of them simple gravimetric or titrimetric
procedures. The analytical methods employed were:
Parameter Method
Sulfur (IV), TOS lodimetric Titration
Total Sulfur (Solids) Oxidize, Weigh as BaS04
Total Sulfur (Solutions) . Titrate with Lead Perchlorate
Acidity/Hydroxide NaOH/HCl Titration
Soluble Carbonate Titration of HCO' with HC1
III-3
-------
Solid Carbonate Acidify, Collect C02 in NaOH,
Titrate resulting solution as
above
Calcium, Magnesium EDTA Titration
I _U
Sodium, trace Mg Atomic Absorption
and Ca4"1" Spectrometry
Determination of the amount of Na2SOit which reacted with lime or lime-
stone to precipitate calcium sulfate from sodium sulfite/bisulfite/sulfate
solutions was important in many of the laboratory experiments conducted.
This determination required that the filtered product solids be washed
thoroughly to remove entrained mother liquor without dissolving calcium
sulfate from the solids. It was for this reason that a wash solution
saturated with respect to CaSOi,. • 21^0 was used for washing the solids.
Prior to using this washing procedure routinely, analyses of dissolved
calcium in the fresh and used wash solutions were performed to ascertain
that no additional calcium sulfate was being introduced into the solids
by the washing procedure.
The washed solids were subsequently analyzed for total sulfur, TOS, cal-
cium, and sodium to permit the calculation of the amount of CaSOi,. present.
Even after thorough washing (1 gram of dry solids washed with four, 25-ml
portions of wash solution), a small amount of sodium, usually on the order
of 3 mol percent of the total sulfur present, remained in the solids. That
sodium was attributed to ^250^. entrapped within the calcium sulfite crystals,
and it was excluded when calculating percent calcium sulfate in the solids.
The percent calcium sulfate in the solids was calculated from the number
of millimols of CaSOx and CaSOs per gram of total solids according to
the relation
CaSOx - CaS03
Sulfate Precipitation =
CaSOx
when CaSOx signifies the total of CaS<\ and CaS03 in the precipitate.
Because this calculation involves taking a relatively small difference
between two larger numbers, the result is quite sensitive to small ana-
lytical errors in the determination of CaSOx and CaS03. For example, in
one actual case where calcium sulfate comprised 9.4% of the total calcium/
sulfur salts, a 1% increase in the amount of CaSOjj accompanied by a 1% de-
crease in the amount of CaSOs would have changed the calculated percent
calcium sulfate content to 11.2%. Analytical uncertainty of that order
is not unreasonable and its effect must be kept in mind when interpreting
the experimental findings relating to sulfate precipitation.
III-4
-------
B. PILOT PLANT
1. Description of the Facility
The dual alkali pilot plant at Arthur D. Little, Inc., (ADL) in Cambridge,
Massachusetts, was built jointly by Combustion Equipment Associates (CEA)
and ADL in 1973. The system operates on 2,000 acfm (400°F) of the com-
bustion flue gas from a natural-gas-fired hot air furnace. S02 is added
to the gas to provide any desired S02 concentration up to about 4,000 ppm.
The basic facility incorporates three process sections: gas scrubbing;
absorbent regeneration; and solids separation. In addition, there are
auxiliary process systems and equipment for special slipstream treatment
operations. A schematic flow sheet of the major process units is pro-
vided in Figure III-2.
The scrubbing system includes a variable throat venturi scrubber followed
by an absorption tower fitted with a radial vane demister. The absorption
tower is equipped with removable trays and sprays so that it can be oper-
ated alternatively as a spray tower, a tray tower (with from one to three
trays), or a de-entrainment separator.
The pilot plant has more than a dozen open-top tanks ranging in size from
8 gallons to 700 gallons for use as feed, storage, or hold tanks and reac-
tor vessels. Two lime/limestone feed systems are available — one for
metering reactant as a dry solid and one for feeding reactant in slurry
form.
The solids dewatering and separation system includes a 6-foot diameter
Dorr-Oliver thickener with three overflow levels, and a Dorr-Oliver
rotary drum vacuum filter. The filter has an active cloth area of 4.3
square feet and is equipped with a system of wash sprays which can be
operated in a variety of spray configurations.
In addition to this basic process equipment, the pilot plant has a number
of auxiliary units including: two Pfaudler kettles (one glass lined and
one stainless); and a 6-inch Bird solid bowl centrifuge.
A list of all pilot plant operations is given in Appendix B.
2. Scrubber Operating Characteristics
The standard scrubber system for the EPA test program consisted of the
venturi scrubber followed by the absorption tower with either two or
three sieve trays. The configuration and operation of the tray tower
depended upon the dual alkali mode being tested. In the concentrated
active sodium modes, two trays were used with no recycle of absorbent
liquor around the tray tower. Regenerated liquor was fed to the top
tray on a once-through basis, the flow leaving the bottom tray being
sent directly to the venturi recycle tank. In the dilute active sodium
modes, three trays were normally used and liquor was recycled around the
III-5
-------
Flue Gas
Tray Scrubber
With Demister and
2 - Trays
Scrubbed Gas
t
Sodium
Makeup
Mix Tank
M
M
H
I
Ca(OH)2 or CaCO3
Settler
Reactor System
\ \ \ \ \
f
HoO
FIGURE 111-2 SCHEMATIC FLOW DIAGRAM OF DUAL ALKALI PILOT PLANT
-------
tray tower to control pH and minimize C02 absorption. Scrubber system
details are given in the flow sheets for each mode in the appropriate
sections of the report.
The range of operating conditions of the scrubber system is given in
Table III-l. There was no washing of the radial vane demister either
with water or scrubber liquor in any of the modes; and, since there was
no fly ash in the inlet flue gas, the spray beneath the bottom tray was
not used.
It is important to realize that the primary function of the scrubber
throughout most of the pilot plant testing was to provide scrubber
bleed liquors of compositions within the range desired for the tests
being performed. However, the scrubber system was also used to assess
calcium scale potential (particularly in dilute mode runs) in certain
operational modes to evaluate the ease of operation.
The general operational characteristics of the scrubber system with regard
to S02 removal and absorbent oxidation are reviewed below. As appropriate,
specific operating conditions and performance characteristics of the scrub-
ber section are covered later in the discussions of operations in each mode.
a. S02 Removal
The S02 removal efficiency in the pilot plant scrubber system has been
evaluated in a number of previous studies as well as during this program.
The overall 862 removal has been shown to be a function of the system feed
stoichiometry, the scrubber configuration, the inlet SC>2 level, and, to a
lesser extent, the concentration of TOS in the absorbent liquor and the
rate of sulfite oxidation.
The feed stoichiometry is given by:
mols active Na+ capacity
mols S02 inlet
The active Na+ capacity is the theoretical equivalent of the mols of S02
which can be absorbed by active sodium species:
[Na-+] . capacity = [Na2S03] + 2[Na2C03]
active
+ [NaOH] + [NaHC03J
The relationship between S02 removal and feed stoichiometry for the
venturi/two-tray configuration operating in the concentrated active
sodium mode is shown in Figure III-3. The data indicate that for a
given stoichiometry, lower S02 levels tend to result in slightly lower
removal efficiencies. The plot shows that to achieve a 90+^ removal of
S02 from a gas containing 2,500 ppm, a feed stoichiometry of about 1.1
III-7
-------
TableIII-1 SCRUBBER SYSTEM OPERATING CONDITIONS
Flue Gas Conditions
Temperature Entering Venturi (Adjustable): 365 - 430°F
Inlet Flow Rate (Adjustable): 1,100 scfm
NO Level: 800 - 1,000 ppm
X
S02 Level (Adjustable): 500 - 2,800 ppm
02 Level (Adjustable): 4 - 10% (dry basis)
HaO Level: 18 - 23% (dry basis)
Gas Dew Point: 120 - 130°F
Venturi Scrubber Conditions
Venturi Tank Temperature: 140 - 145°F
Venturi Recycle Flow: L/G = 16 gals/Macfm sat'd
Venturi Pressure Drop: 10 - 14 in. H20
Tray Tower Conditions (Two Sieve Trays)
Tray Tower Feed: L/G = 1.5 - 12 gals/Macfm sat'd
Pressure Drop/Tray: 1.6 - 2.0 in. H20
III-8
-------
100
100
M
M
M
I
95
90
1 85
o
CD
cc
CM
O
en
80
75
High Inlet
Scrubber Operating Conditions
Bleed Liquor Temperature = 140-150°F
Active Sodium = 0.2-0.55M
Total Dissolved Solids = 5-15 wt. %
Sulfite Oxidation = 150—300 ppm SO2 Equivalent
Venturi: AP= 11-14 in. H20
L/G = 15-18 gals/Macf Saturated
Legend:
Active Sodium (M)
Inlet SO2 0.2-0.3 0.3-0.4
700-1,150 D a
2,050-2,750 o 3
0.4-0.55
95
90
85
80
75
70
70
0.7
0.8
0.9 1.0 1.1 1.2 1,3 1.4
Scrubber Feed Stoichiometry (mols Na Capacity/mols Inlet SO.-
1.5
1.6
1.7
1.8
FIGURE 111-3 SO2 REMOVAL VS. FEED STOICHIOMETRY (VENTURI + 2 TRAYS)
-------
is required, while to achieve the same 90+% removal from a gas containing
about 700 ppm, a feed stoichiometry of about 1.3 is required. This trend
would be expected since at higher SO 2 levels there is a greater driving
force for removal; hence, higher removal efficiencies are more easily
attained.
This is somewhat different than the experience in a direct lime/limestone
scrubbing system where S02 absorption is limited by liquid phase and solid/
liquid mass transfer. In sodium absorption above feed stoichiometries of
1.1, gas phase mass transfer becomes the limiting rate.
Within the ranges examined, there was no effect observed of active sodium
or total dissolved solids concentration on S02 absorption.
b. Scrubber Oxidation Rates
In general, oxidation of sulfite in the scrubber system is a function of
the oxygen content of the flue gas, the scrubber configuration, the con-
centration of sulfite, and the total dissolved solids concentration. Under
most conditions, oxidation has been found to be oxygen mass transfer lim-
ited. Thus, the rate of sulfite oxidation increases with increasing gas/
liquid contacting (number or efficiency of contact stages) and increasing
oxygen content of the flue gas, and decreases with increasing dissolved
solids level (decreasing oxygen solubility). The concentration of the sulfite
ion itself (independent of total dissolved solids) was only found to affect
oxidation at low concentration levels. The rate of sulfite oxidation (mols
sulfite oxidized/unit time) was found to be essentially independent of active
sodium concentration between 0.20M and 0.65M active Na+, all other variables
being equal. However, oxidation decreased roughly proportionally to active
sodium concentration at levels below about 0.20M active Na+, indicating
that oxidation was becoming reaction rate limited. Also, the presence of
trace elements such as those from fly ash can promote or inhibit sulfite
oxidation.
Figures III-4 and III-5 show oxidation data taken in the pilot plant using
the venturi/two-tray configuration. Figure III-4 shows the effects of active
sodium concentration at low total dissolved solids levels and 4-5% oxygen in
the flue gas. Figure III-5 shows the effects of both oxygen content in the
flue gas and total dissolved solids levels on oxidation. At intermediate
active sodium concentrations (0.3-0.6M Na+active) and 1.5-2.5M total sodium,
the oxidation rate (mols sulfite oxidized expressed in equivalent ppm of S02
absorbed) at 5% oxygen was 200-250 ppm of S02. This would correspond to
about 8-10% of the S02 removed for 3.5% sulfur coal and 20-25% for 1.5%
sulfur coal. At 2-3 times the total dissolved solids concentrations, rates
of oxidation were reduced by about 50% or more.
The data presented in Figures III-4 and III-5 are a combination of data
taken prior to and during the EPA test program. Each point in Figure III-5
represents the average of from three to six data points for a given run.
In all cases the rate of oxidation across the scrubber was calculated from
sodium, sulfite and sulfate balances across the system and confirmed by
111-10
-------
300
200 -
c O
O OT
O £
100
Venturi + Two Trays
4-5% O2
TDS = 5-15%
Bleed pH = 5.2-6.2
No Fly Ash
J.
0.1
0.2 0.3 0.4
Active Sodium Concentration (Na , M)
0.5
0.6
FIGURE 111-4 SCRUBBER OXIDATION - EFFECT OF ACTIVE SODIUM CONCENTRATION
III-ll
-------
500
400
CN
O
CO
'o
a
a
Venturi + Two Trays
Bleed pH = 5.2-6.3
No Fly Ash
TDS: • = 8-15%
O = 25-30%
300
I
c
g
•M
| 200
O
CO
100
5 6
Flue Gas Oxygen Concentration (Volume '
FIGURE 111-5 SCRUBBER OXIDATION - EFFECTS OF OXYGEN AND TDS CONCENTRATION
111-12
-------
overall material balances and cake analyses (taking into account oxida-
tion occurring in other parts of the system).
It should be noted that these oxidation and S02 removal data were obtained
at scrubber liquor operating temperatures which were 15-20°F higher than
those which would be expected in a scrubber operating on a normal utility
or industrial boiler. The higher temperatures were due to the fact that
the flue gas was generated from natural gas fuel which has a much lower
carbon-to-hydrogen ratio than coal, and which when burned produces com-
bustion product gas with considerably higher humidity levels. These higher
humidity levels result in higher saturated gas temperatures in the scrubber
which affect both S02 removal and sulfite oxidation rates. The equilibrium
vapor pressure of S02 is on the order of 50-60% higher, so it would be ex-
pected that the removal efficiencies observed in the pilot plant would be
lower than those attainable in actual boiler applications. Similarly, the
rate of sulfite oxidation is strongly dependent upon temperature, increas-
ing with increasing temperature. Thus, the oxidation rates observed in
the pilot plant may also be slightly high compared to those expected in
actual applications.
This is confirmed by data obtained in the prototype system at the Scholz
Steam Plant where oxidation rates at 4.5-5.0% oxygen in the flue gas were
found to be 150-200 ppm compared with the 200-250 ppm for the pilot plant
(using the same venturi/two-tray combination). Similarly, oxidation rates
at 7.5% oxygen averaged about 250 ppm in the Scholz system and between 300 ppm
and 350 ppm in the pilot plant.
c. Conclusions
The pilot plant S02 removal and oxidation data are specific to the pilot
plant scrubber configuration as influenced by the scrubber operating tem-
perature for the pilot plant flue gas stream. The scrubber operating
temperature of 140-150°F is higher than that normally encountered in con-
ventional boiler flue gas applications (120-130°F). The elevated tempera-
ture in the pilot plant system tends to decrease S02 removal efficiency due
to elevated S02 partial pressures for any given solution, and tends to in-
crease oxidation rates. However, the purpose of the pilot plant scrubber
and its operations was to provide scrubber effluent with an appropriate
composition for use in the various dual alkali modes rather than to gen-
erate basic data on S02 absorption using sodium solutions.
Within the above constraints, the scrubber operations did indicate that
S02 removal in excess of 90% is easily accomplished over a range of S02
inlet concentrations from 700-2,800 ppm by adjusting the scrubber feed
stoichiometry. To achieve this removal efficiency, a stoichiometry of
1.1 mols of' active Na+ capacity/mol S02 inlet was required at the high
inlet S02 range; a stoichiometry of 1.3 was required in the lower inlet
S02 range. In any range of S02 concentration, increasing stoichiometry
increased the S02 removal. There was no important apparent effect of
active sodium concentration within a range of 0;2-0.5M or total dissolved
solids concentration within a range of 5-15 wt %.
111-13
-------
Sulfite oxidation is mass transfer limited at active sodium concentrations
above 0.2M with the rate of oxidation increasing with the oxygen content
of the flue gas. At lower active sodium concentrations the oxidation rate
is roughly proportional to the active sodium concentration. The rate of
oxidation decreases with increasing total dissolved solids; by increasing
TDS from 8-15 wt % to 25-35 wt %, the oxidation rate being reduced by a
factor of 2-3. At lower total dissolved solids in concentrated active
sodium systems (0.3-0.6M active Na, 5-15 wt % TDS) sulfite oxidation
can be expected to be on the order of 100-300 ppm equivalent S02 removal
for oxygen concentrations in the flue gas ranging from 4 to 8 vol %.
111-14
-------
IV. LIME REGENERATION — CONCENTRATED
ACTIVE SODIUM MODE
A. PRIOR WORK ON LIME REGENERATION AT ADL
Prior to initiation of the EPA/ADL Dual Alkali Program, laboratory and
pilot plant research and development were conducted at ADL on lime re-
generation of concentrated scrubbing solutions. In 1971 ADL conducted
a detailed laboratory study of the regeneration of simulated sodium
scrubbing solutions using hydrated lime. This work, sponsored by the
Illinois Institute for Environmental Quality,1* dealt exclusively with
characterizing the nature of the regeneration- reaction. The objectives
of the laboratory program were the development of reaction conditions
which would produce waste calcium salts with good settling and filtration
characteristics, production of scrubber feed solutions with low concen-
trations of dissolved and suspended calcium salts, and good utilization
of lime in the regeneration reaction.
The objectives of these laboratory studies were satisfied and conditions
for an effective regeneration reaction were determined. Chemical analyses
in the laboratory at that time were limited to liquid phase analyses; be-
cause of the generally high sulfate concentrations in the reactor feed,
changes in sulfate concentration were usually within the experimental
error of the analyses. As a consequence, no significant sulfate pre-
cipitation was detected in the lime regeneration of concentrated sulfite
solutions.
The dual alkali process still appeared attractive because of the highly
effective S02 removal, very low calcium concentrations in the regenerated
scrubber feed, and the effectively complete utilization of lime in the
process. Based upon the laboratory results, the dual alkali pilot plant
was constructed at ADL's facilities. The pilot plant was operated to
generate the design data for the 20-megawatt prototype dual alkali system
to be built by CEA for Gulf Power Company/Southern Company Services.
The results of the laboratory and pilot work conducted prior to the EPA/
ADL program are summarized in the literature.
IV-1
-------
B. LABORATORY STUDIES OF SULFATE PRECIPITATION
WITH LIME
1. Introduction
The observation during ADL pilot plant tests that calcium sulfate was
present in the product solids when regenerating with lime in a concen-
trated active sodium mode aroused a great deal of interest because in
a concentrated mode the removal of sulfate produced as a result of sul-
fite oxidation was foreseen as a possibly significant problem. From a
consideration of the thermodynamic equilibria involved, one would not
expect calcium sulfate to precipitate during regeneration of concentrated
sodium sulfite/bisulfite/sulfate solutions.
In a typical regenerated solution produced by a concentrated mode process,
the pH would be in the range 9-11, and would contain about 0.50M Na2SOi+
and 0.25M ^2803. Equilibrium levels of total dissolved calcium would
be typically less than 100 ppm (0.0025M). In a solution of this ionic
strength (about 2.3M), for calcium sulfate to precipitate as gypsum
(CaSOtt • 2H20), the apparent solubility product, Ksp' (the product of
total dissolved calcium and total sulfate concentrations), is about
1.05 x 10~2. For the 0.50M ^280^. level taken as an example, a dis-
solved calcium level of 0.021M (about ten times greater than the actual
level) would be required to effect the precipitation of gypsum.
Subsequent to observing the presence of calcium sulfate in the solids
produced at the ADL dual alkali pilot plant, Borgwardt^ demonstrated
quite conclusively during pilot studies of direct limestone slurry
scrubbing that calcium sulfite product solids containing significant
amounts of calcium sulfate from liquors that were unsaturated with
respect to gypsum could be produced. A distinct gypsum or anhydrite
(anhydrous CaSOif) crystalline phase in the solids could not be identi-
fied and it was proposed that the calcium sulfate was present as some
sort of solid solution within the CaSOa • 1/2H20 crystal. Further, it
was found that the ratio of CaSO^/CaSOs in the solids could be corre-
lated quite well with the activities of the sulfate ion in the liquor.
The laboratory program which was conducted to study this apparently
similar phenomenon in the concentrated lime dual alkali process con-
sisted of a series of continuously fed, stirred tank reactor (CSTR)
experiments. They were designed first to verify under "normal" opera-
ting conditions that sulfate was, in fact, being precipitated and then
to study the effects of Na2SOtf and active sodium levels on the amount
of sulfate precipitation. Studies were also conducted to determine the
effect on sulfate precipitation of less than complete neutralization of
bisulfite, regeneration with shorter and longer than normal reaction
times, and higher and lower than normal levels of active sodium.
The normal reactor residence time for these experiments was established
at 50 minutes, a time, based on previous ADL experience, that was adequate
for completion of reactions, even if regeneration was carried well into
IV-2
-------
the hydroxide range. The solution fed to the reactor normally contained
about 0.5M active sodium and had a pH in the range from 5.4 to 5.6
(HS03/S03 ratio approximately 10). As noted earlier in the description
of the laboratory apparatus, lime was fed to the reactor as a 5-10% slurry
which diluted the sodium fed by 10-20% upon mixing in the reactor. In the
subsequent discussions, when reactant concentrations are reported, they
are after dilution by the slurry water.
2. Experimental Results
The precipitation of calcium sulfate observed under normal conditions at
several levels of dissolved sulfate is tabulated, along with the associated
solution compositions, in Table IV-1. The first two experiments, A and B,
are essentially identical except for the change in concentration of Na2SOit
from about 0.20M to about 0.55M. In both cases measurable amounts of sul-
fate precipitation were observed — precipitation increased with increasing
sulfate concentration. In experiment B, regeneration was carried out to
the point where the HS(>3 fed to the reactor was just 100% neutralized and
no free hydroxide was generated (sulfate precipitation was 0.091 mols CaSO[f
per mol CaSOx).
Since in certain cases it proved difficult to adjust reagent flows to
achieve exactly 100% neutralization of the HSOs present, and in other
experiments, the degree of regeneration was intentionally varied, the
term "percent neutralization" was used to define the extent to which
regeneration was carried out. When the bisulfite present was not com-
pletely neutralized, the percent neutralization was computed from the
relation
Percent Neutralization =
initial' [HS°3] final
. . . .
initial
x 100%
When regeneration was carried on into the free hydroxide region, the
neutralization was greater than 100% and was calculated from the re-
lation
I + [OH _
initial final
Percent Neutralization =
i
j
initial
x 100%
Experiment C was similar to experiment B except that percent neutraliza-
tion was increased from 100% to 124% to determine the effect of the extent
of regeneration on sulfate precipitation. However, the sulfate precipita-
tion (0.090) was essentially identical to that observed at 10CU neutrali-
zation. In experiment D, the concentration of sulfate was further increased
to about 1.1M, and sulfate precipitation increased to 0.165 of the total
precipitate. Experiment D was carried to only 67% neutralization; however,
results of subsequent studies established that Na2SOH level, rather than
incomplete neutralization, was the more significant factor in the increased
amount of calcium sulfate precipitation observed.
IV-3
-------
Table IV-1 SULFATE PRECIPITATION IN CONCENTRATED MODE LIME REGENERATION LABORATORY CONTINUOUS REACTOR—50 MIN RESIDENCE TIME
Expt.
A
B
C
D
IHE
0.
0.
0.
0.
Con
JOal.M
345
322
322
354
iposition o]
[SOql.M
0.046
0.042
0.030
0.050
: Feed Liq
0.204
0.543
0.531
1.110
uor
£fi
5.5
5.4
5.4
5.4
[HSOgl.M
0.020
nil
nil
0.116
Compos:
[OH~] ,M
nil
nil
0.077
nil
It ion of Res
[S0°] ,M
0.208
0.211
0.161
0.168
ictor Efflus
0.200
0.540
0.518
1.051
;nt
,Ca++
20
60
56
84
] , ppm pH
7.8
8.5
11.5
6.8
%HS03
Neutralized
94
100
124
67
7, Calcium mols CaSQ^/mol CaSOx
Utilized13 in solids
95
103
90
97
0
0
0
0
.023
.091
.090
.165
Composition of feed liquor after dilution by lime slurry water.
Based on (CaSO /Ca_ .) x 100% in solids.
X COC3.J-
-------
The sulfate precipitated in experiments A-D is shown as a function of
sulfate concentration in Figure IV-1. Within experimental error, sul-
fate precipitation was directly proportional to sulfate concentration
in the reactor. (It should be remembered from the discussion of ex-
perimental error in Chapter III that sulfate precipitation is subject
to some experimental error — for a true precipitation of 0.08, experi-
mental error could cause the measured value to be in the range from
0.06-0.10.) Included in Figure IV-1 are the results of additional ex-
periments carried out at much shorter than normal residence times (5
minutes and 8 minutes) , but at the same active sodium level as experi-
ments A-D and all at 100% neutralization. With the possible exception
of the experiment at 0.2M Na2S04 and 5 minute residence time, calcium
sulfate precipitation and its dependence on sodium sulfate concentration
were essentially the same as that observed at the normal, 50 minute resi-
dence time. Also included in Figure IV-1 is the effect of allowing portions
of reactor effluent slurry to stand with gentle stirring for a period of
three hours after emerging from the reactor. In every case, a decrease
in the mol ratio CaSO^/CaSOx in the solids was observed upon standing.
The effect of percent neutralization was studied in detail with the CSTR
operating at a 5 minute residence time followed by a three-hour hold time
to observe the effect of hold time on reaction products, equilibria, and
product stability. Observed sulfate precipitation is shown in Figure IV-2
as a function of percent bisulfite neutralized. In these studies, active
sodium (after dilution) and the pH of the feed were maintained at "normal"
levels, except for the several cases where specifically indicated, the
concentration of Na2SOi,. was between 0.64M and 0.69M. In those cases where
the Na2SOi,. concentration fell outside that range, the arrows indicate ap-
proximate adjustments to sulfate precipitation which would have been expected
at a sodium sulfate concentration of 0.67M (based on the Na£SOif dependence
shown in Figure IV-1). Although there is some scatter in the data, increased
sulfate precipitation seemed to occur at very high and very low percent neu-
tralizations. Also shown in Figure IV-2 is the general decrease in CaSO^/
CaSOx in the solids when the product slurry was allowed to stand and react
for an additional period of three hours.
Sulfate precipitation as a function of the concentration of TOS was studied
at an 8 minute CSTR residence time and the results observed are shown in
Figure IV-3. Except where specifically indicated, a CSTR residence time
of 8 minutes was employed and the concentration of Na2SOi+ was about 0.6M.
Two of the data points were obtained from experiments in which a 5 minute
CSTR was used. The experiment at 0.65M TOS had a higher sulfate concentra-
tion, 0.75M Na2SO^. Based on the relationship of calcium sulfate precipita-
tion to Na2SOif concentration discussed previously, the arrows attached to
the data points for that experiment indicate what the CaSO^/CaSOx mol
ratio should have been at 0.6M Na2SOtt.
The data in Figure IV-3 show a clear inverse proportionality between the
amount of calcium sulfate precipitated and the TOS concentration. Because
of the relatively small number of data points obtained and the scatter in
them, a precise determination of the functional relationship between TOS
IV-5
-------
TJ
"5
C
D
0.20
0.16
0.12
Constant Conditions
tNa+1 Active -0.45M (diluted)
Feed pH = 5.4-5.6
"% Neutralization = 100%
Temperature - 52°C
(70%NeutraU
i—i
f
CTv
CD
CC
O
to
8
o
~^t
O
in
CO
O
VI
O
0.08
0.04
O
Legend
A 50-Minute CSTR
A 3-Hour Reaction After 50-MinuteCSTR
O 5-Minute CSTR
• 3-Hour Reaction After 5-Minute CSTR
OS-Minute CSTR
0.2
0.4
0.6
0.8
[Na2S04], M
1.0
1.2
1.4
1.6
FIGURE IV-1 SULFATE PRECIPITATION AT VARIOUS Na2SO4 CONCENTRATIONS
-------
<
0.20
o 0.16
s
cc
O
in
<3
o 0.08
O
to
ra
O
0.04
Constant Conditions
[Na+] Actives0.45M (diluted)
Feed pH = 5.4-5.6
Temperature = 50°C
[Na2S04l s 0.67M (diluted), except where specified tl.
(0.75M)
20
(0.b6M)'
Legend
°5-MinuteCSTR
•3-Hour Additional Reaction
40
60
80 100 120
Percent HSOT Neutralized
140
160
180
200
FIGURE IV-2
SULFATE PRECIPITATION FOR VARIOUS DEGREES OF LIQUOR REGENERATION
-------
0.12
0.10
t/J
73
"5
in
•i-*
1 0.08
M—
Ul
k_
o
*J
o
ra
o>
QC
£ 0.06
X
8
CD
O
"o
.E
2* 0.04
en
ra
O
(/}
0
_§
0.02
^ Constant Conditions
\ Feed pH = 5.4 - 5.6
. Temperature ^ 52°C
\ fr-6mln, K^,°n6",00%
- O \ ^ Residence Time (T) = 8 min
N. .
\ T
\ (0.75 M SO")
A
X1
\
(0.75 M S04=) \
1 \ •
\ •
« \
Legend: x.
O Initial Study ^*» •
(TOS, Total Sulfur) f "*\^
• Supplementary Study " _^ "^^^^
~ (TOS, SO/) ^ '
1 1 1 1 1 1
0 0.2 0.4 0.6 0.8 1.0 1.2
[TOS], M
FIGURE IV-3 EFFECT OF TOS CONCENTRATION ON CALCIUM SULFATE PRECIPITATION
REGENERATION WITH LIME
IV-8
-------
concentration and calcium sulfate precipitation based only on laboratory
data was not attempted. A general correlation of all laboratory and pilot
plant data is presented in Section C of this chapter.
3. Discussion
The experimental work described was aimed primarily at developing a prac-
tical understanding of what variables are important to, and in what manner
they influence, calcium sulfate precipitation. The experimental results
clearly establish that calcium sulfate can be precipitated when concen-
trated sodium sulfite/bisulfite/sulfate solutions are reacted with lime.
The direct proportionality between one of the two most significant inde-
pendent variables, NazSOi* concentration, and the percent CaSOit in the
product solids is not surprising since, simplistically, increasing the
SOIj;/S03 ratio should favor precipitation (or inclusion) of CaSOif into
the calcium sulfite. The observation that sulfate precipitation de-
creased when the concentration of TOS was increased is also reasonable.
At constant pH, 863 concentration is directly proportional to the concen-
tration of TOS. Finally, the somewhat lower CaSOif/CaSOx ratio observed
at intermediate percent neutralizations is also reasonable because SO^
concentration is relatively higher in that region than at either extreme.
A general decrease in CaSO^/CaSOx mol ratio when the effluent slurry
from the 5 minute CSTR was allowed to continue to react obviously indi-
cates that the solids in the initial effluent were not at equilibrium
with the solution. The initial effluent liquid phase tended to be super-
saturated with respect to CaSOa as evidenced by the subsequent appearance
of solids (largely CaS03 • 1/2H20) when samples of filtered reactor effluent
were allowed to stand. However, the changes in total dissolved calcium which
occurred when the intact slurry samples were allowed to stand could have pro-
duced only about 10% of the change in CaS04/CaSOx observed even if desuper-
saturation had resulted in the precipitation of only calcium sulfite. In
some of the slurry samples, lime continued to react with the solution after
the first 5 minutes of reaction, resulting in the precipitation of additional
calcium sulfite, but that, in itself, could have changed the CaSO^/CaSOx
ratio by only 10-20% of the amount observed. Thus, it was concluded that
although the amount was small and difficult to measure accurately, some of
the CaSC^ which initially had precipitated had, in fact, redissolved as
the mixture was allowed to equilibrate.
In no case were the measured levels of total dissolved calcium and sulfate
high enough to even approach saturation with respect to gypsum. Values of
KSD', the "apparent solubility product", which were observed under a number
of reaction conditions are shown in Figure IV-4. The solid curve represents
the calculated value of Ksp' as a function of ionic strength for solutions
saturated with respect to gypsum. These values of KSp were calculated by
the method of Kusik and Meissner, 7 which was developed to deal with solu-
tions at high ionic strengths.
IV-9
-------
in"
o
to
to
O
Q.
-o
o
Ell
£
c
0)
l_
(0
a.
Q.
10'
10'° L.
10
,-4
10
-5
I I I
Legend:
A 50-MinuteCSTR, Vary [Na2SO4]
D 8-MinuteCSTR, Vary [Na+] Active
O 25% Neut., 5-Minute CSTR
• 25% Neut., 3-Hour Reaction After
5-Minute CSTR
X EPA Saturation Tests
I I I
I I I
12345
ju, Ionic Strength (Molarity Units)
FIGURE IV-4 RELATION OF OBSERVED APPARENT SOLUBILITY
PRODUCTS TO SATURATION VALUES FOR GYPSUM
IV-10
-------
This method involves first estimating the reduced activity coefficient
of pure CaS04 (or CaS03) at total ionic strength, y, of the multlcompo-
nent solution; such a solution normally would be supersaturated. Then
the activity coefficient, Y, of CaS04 (or CaS03) in the multicomponent
solution is obtained which is normally different from the pure solution
value calculated above. After determining the water activity, a^, the
resulting values are substituted into the appropriate thermodynamic solu-
bility product [ (Kt)sp ] expression
(Kt>sp = mCamS0l/aw
where n is the number of waters of crystal hydration and m is the molality
of the calcium and sulfate (or sulfite) ion as indicated by the subscript.
The accuracy of the predicted values of Ksp' is confirmed by the close
agreement between the calculated values as shown by the curve and experi-
mental values derived from EPA experiments in which dual alkali reaction
effluent slurries were saturated with gypsum (Chapter VIII) during satura-
tion tests.
C. PILOT PLANT OPERATIONS — CONCENTRATED
ACTIVE SODIUM MODE
1. Pilot Plant Test Program
The evaluation of the dual alkali process operation in the concentrated
active sodium mode using slaked lime for absorbent regeneration was fo-
cused primarily on the performance of the regeneration system at active
sodium levels above 0.25M. The operation of the regeneration reactor system
ultimately determines the required calcium/S02 stoichiometry as well as
the properties of the waste solids produced, both of which are important
factors in determining the operability and economics of this dual alkali
approach.
The pilot plant test program involved the characterization of two differ-
ent reactor systems: a continuous stirred tank reactor (CSTR); and a mul-
tistage reactor system previously developed by ADL (ADL reactor), which
for this study consisted of two CSTR's in series. The holdup time in the
first stage is usually maintained between five minutes and ten minutes and^
in the second and later stages between 20 minutes and 40 minutes each. While
all reactants are normally fed to the first stage, the reactor has been tested
with split feeds.
Standard operating conditions for the ADL reactor were established for
the two-stage system in pilot plant development work prior to this EPA
program; however, data on the operation of a CSTR in the concentrated
active sodium mode were not complete. A preliminary study of the CSTR,
therefore, was undertaken in the pilot plant to examine the effects o±
various reactor parameters and selected process variables on the performance
IV-11
-------
of the CSTR, and to develop a set of suitable operating criteria. The
preliminary study consisted of short-term, open-loop reactor runs using
simulated scrubber bleed and dry hydrated lime. This initial phase of
testing was then followed by a number of extended, closed-loop continuous
runs to confirm the open-loop results and to evaluate the overall process
operation using both the CSTR and a two-stage ADL reactor.
The lime used in all the pilot plant runs was a 200-mesh hydrated lime
containing 95 wt % calcium (reported as Ca(OH)2) and averaging 91 wt %
available calcium hydroxide. The remaining 4 wt % calcium,' consisting
mainly of CaCQ$ and burnt CaO, was determined to be essentially unavail-
able for reaction above a pH of 8.
2. Regeneration Reactor Performance
Reactor performance in both the open-loop reactor tests and the closed-
loop continuous runs was evaluated in terms of four process criteria:
calcium utilization; effluent soluble calcium concentration; reaction
of sodium sulfate to precipitate calcium sulfate; and the dewatering
properties of the solids generated. Of particular concern in the pilot
plant work were the effects of sulfate concentration in the feed liquor
and the extent of absorbent regeneration (reactor pH) on the reactor
performance.
The composition of the spent scrubber solution (simulated and actual) fed
to the regeneration system varied over the following ranges:
[Na+ .. ] = 0.25-2.30M
active
[SO^j = 0.05-1.70M
pH = 5.7-6.6
Most of the runs were made with active sodium levels between 0.40M and
0.55M and sulfate levels up to l.OM. However, a number of runs, particu-
larly closed-loop, were made at high active sodium (and sulfate) concen-
trations. These runs were designed primarily to examine the effects of
high TDS (total dissolved solids) levels on oxidation in the scrubber
system and of high TOS (total oxidizable sulfur) levels on sulfate pre-
cipitation.
The nominal reactor system operating conditions and solution compositions
examined are listed in Table IV-2. In all cases, the reactor vessels were
open-top, baffled tanks (four baffles). Single propeller-type agitators
were used for mixing. Stirrer speeds generally ranged from 200 rpm to
300 rpm.
IV-12
-------
TABLE IV-2
SUMMARY OF NOMINAL REACTOR OPERATING CONDITIONS
General Conditions:
Temperature
Agitation -
- 105-140°
Propeller
F
(200-300 rpm)
Ca(OH)2 Feed - Dry Solids
rN + ] [SOT]
LNa *otivej 4
0.25M 0.8M
0.45-0.55 0.05
0.5
0.6
0.6
0.75
0.8
0.8
0.9-1-0
2.0 0.25
1.6
Type
CSTR
CSTR
CSTR
CSTR
CSTR
CSTR
CSTR
ADL
CSTR
CSTR
CSTR
Nominal
Holdup
30 mins.
30
30
30
6.0
30
30
5 & 50
30
30
10-15
Extent of
Regeneration
(Reactor
Effluent pH)
7.5
7.5
8.0
7.5 & 12
7.5 & 12
12 & 12.5
7.5, 11 & 12.5
12
8 & 11.5
12
7.5 & 11
IV-13
-------
a. Calcium Utilization
Calcium utilizations in runs with both reactor systems were high — gen-
erally greater than 95% of the available Ca(OH)2 in the lime feed. There
appeared to be no effect of reactor configuration (ADL versus CSTR) on the
utilization of calcium, nor was any effect of sulfate concentration on cal-
cium utilization observed over the range of conditions studied. (In test-
ing prior to this program, though, utilization was found to decrease slightly
at very high sulfate levels.5) There was, however, a slight dependence of
utilization on reactor operating pH and on reactor holdup time.
In the pH range 7.4-11.0;, calcium utilizations generally approached, and
sometimes exceeded, 100% of the available Ca(OH)2 for reactor holdup times
from 25 minutes to 60 minutes. Utilizations in excess of 100% can be at-
tributed to a number of factors including variations in the lime feed com-
position and analytical errors, as well as the reaction of a portion of
the calcium value normally considered to be unavailable at high pH levels
(e.g., CaCOs). The average utilization observed in all runs below a pH
of 11.5 was 98.5% + 3% of the available Ca(OH)2, as determined from the
analysis of the reactor effluent solids. This would be roughly equiva-
lent to utilization of 94-95% of the total calcium in the raw lime.
At pH levels of 11.5-12.5, the average utilization dropped off slightly
to 96% + 1% of the available Ca(OH)2 (or 91-92% of the total calcium
value). During continuous closed-loop operations at reactor pH's greater
than 11.0 and with a reactor holdup of 25-30 minutes, the thickener over-
flow and filtrate were observed to have slightly higher pH levels than
the reactor effluent. This increase in pH, on the order of 0.2 pH units
(confirmed by slightly higher OH concentrations), indicated that the
reactor effluent at the shorter residence times was not quite at equi-
librium and that some additional reaction was occurring in the thickener.
Regeneration to a pH above 8 required a holdup time greater than 20 minutes
for essentially complete reaction.
b. Effluent Soluble Calcium Concentrations
Soluble calcium concentrations in the reactor and thickener overflow
liquors varied inversely with sulfite concentration, as would be ex-
pected, and were apparently independent of reactor type (ADL versus
CSTR) and reator holdup time (30 minutes versus 60 minutes). The mea-
sured ranges of soluble calcium concentrations are summarized in Table IV-3.
At all but the lowest active sodium levels (0.08-0.13M SO^) , soluble calcium
concentrations were well below 100 ppm. At these levels, there was no evi-
dence of scale buildup in the scrubber system during closed-loop operations.
However, in one run at the lowest active sodium concentrations (approaching
dilute mode conditions), soluble calcium levels exceeded 500 ppm and there
was some buildup of calcium sulfite and carbonate in the tray tower result-
ing in an increase in the pressure drop across the tower (trays plus de-
mister) from 4 inches to 6 inches H2<) over the course of four days of
closed-loop operation. The high calcium levels and solids deposition
IV-14
-------
TABLE IV-3
SUMMARY OF SOLUBLE CALCIUM CONCENTRATIONS
, + 1 ren=i Ionic Strength
Reactor Type (Total Holdup), lNa activeJ L 3J y [Ca ]
CSTR (30 mins.) 0.25M 0.08-0.13M ^2.8 600-800 ppm
CSTR & ADL (30 & 60 mins.) 0.4-0.55 0.2-0.3 2.2-4.0 25-95
CSTR (15 & 30 mins.) 1.7-2.3 0.7-0.9 3.2-8.2 15-30
IV-15
-------
suggest significant supersaturation with respect to calcium sulfite,
similar to that experienced at low sulfite concentrations in the dilute
mode testing (see Chapter VII).
In Figure IV-5 the product of the soluble calcium and sulfite concentra-
tions is plotted against the apparent solubility product curve predicted
by the method of Kusik and Meissner.7 All the experimental solubility
product data for both the thickener overflow and reactor effluent fall
above the predicted curve. In general the measured calcium concentra-
tions and experimental solubility product data for the thickener over-
flow tend to be slightly lower than those for the reactor effluent. There
are a number of periods, though, during closed-loop operations when thick-
ener overflow calcium values and concentration products were equal to or
slightly greater than reactor effluent levels.
It would be anticipated that at steady-state, thickener overflow values
would be lower than reactor effluent values. First, the holdup time in
the thickener ranged from six to ten hours in the closed-loop runs, allow-
ing considerable time for desupersaturation. Second, and perhaps more
important, the thickener overflow was diluted by the addition of filter
wash water. In the case of intermediate active sodium levels this tended
to decrease total solubles concentrations on the average by about 5%; how-
ever, at the higher active sodium levels, this decreased solubles levels
by an average of about 25% (due to the lower feed forward rates). Thus,
the experimental solubility products in the thickener overflow at the
higher active sodium levels appear closer to the predicted curve, and
should, at most, represent saturation levels for the indicated solutions.
These results suggest that the degree of supersaturation indicated by the
difference between the predicted curve and experimental data may be an
artifact or an overstatement of any actual level of supersaturation.
There are two possible explanations for the apparent supersaturation
levels. First, the predicted curve may be slightly low, a conclusion
independently reached from dilute mode data (Chapter VII). The relation-
ship was calculated based upon literature data for low ionic strengths
using the method formulated by Kusik and Meissner for calculating acti-
vity coefficients at high ionic strengths. The method is only good to
about 10-20% and any error in the data base would be translated directly
into the predicted curve. Second, the CaSOs • 1/2H20 solubility product
curve may not apply to the product solids where there is an appreciable
degree of sulfate precipitation. The calcium sulfite/sulfate solids may
be present as a mixed crystal, which could have a different solubility
product.
Based upon the system operation in closed-loop, as long as soluble calcium
levels are in the range of or below 100 ppm, there is no potential for scale
formation in the scrubber system. Even if a small amount of supersaturation
exists at these low levels (at most, 50 ppm Ca++), the tendency in the scrubber
system is to increase the solubility of calcium by conversion of sulfite to
sulfate and bisulfite.
IV-16
-------
10-
-7
1.0
2.0
3.0
4.0 5.0 6.0
p., Ionic Strength (Molarity Units)
7.0
8.0
FIGURE IV-5 RELATION OF OBSERVED APPARENT SOLUBILITY PRODUCTS TO
SATURATION VALUES FOR CaSO3 REGENERATION WITH LIME
9.0
IV-17
-------
c. Precipitation of Calcium Sulfate
A careful analysis of the solids produced show that a significant amount
of calcium sulfate precipitation occurs over a wide range of soluble sul-
fate concentration. The amount of calcium sulfate precipitated was found
to be strongly a function of the soluble TOS and sulfate concentrations.
The sulfate dependence is shown in Figure IV-6, in which observed molar
ratios of (CaSO^/CaSOx) in the reactor effluent solids, corrected for the
small amount of occluded or entrained Na2SOit, are plotted against sulfate
concentration in the feed liquor for TOS levels in the range of 0.4-0.5M.
The general relationship indicated by the dashed line agrees quite closely
with that developed from the laboratory program for the same TOS range.
Figure IV- 7 shows a general correlation plot of sulfate precipitation
measured in all laboratory and pilot plant runs over the entire range
of liquor compositions tested in the concentrated mode. Here, the molar
ratio of calcium sulfate-to-calcium sulfite is plotted as a function of
the soluble sulfate-to-sulfite ratio in the reactor bleed liquor. A. number
of other possible relationships were tested covering different ratios (and
powers) of concentrations in both the feed and bleed liquor in order to
establish a reasonable correlation. The simple relationship shown in
Figure IV- 7 provides the closest correlation — one which is also in-
tuitively reasonable.
It should be pointed out that Figure IV-7 applies only to concentrated-
mode operations. In dilute mode operations the sulfate solubility product
is exceeded and the ratio of sulfate to sulfite in the product solids is
governed by the solubility products of hydrated lime and gypsum, and by
the amount of sulfite that must be precipitated prior to reaching the
gypsum solubility product (see Chapter VII) .
The relationship shown in Figure IV-7 can be expressed as follows:
'mols CaSo
mol CaS03
'SvE* V -/ reactor
solids \ / . .
liquor
or,
/™~ic. roon \ o,0365(rso=i/rso=n
liquor
mol CaSOx / 1 + 0.0365([SOU]/[SO,])
' reactor 3 "reactor liquor
solids
In this relationship calcium sulfate precipitation is inversely propor-
tional to sulfite concentration which suggests that the greatest amount
of sulfate can be precipitated when sulfite concentration is at a minimum —
that is, when the scrubber bleed liquor is only partially neutralized or when
IV-18
-------
u.^u
0.16
ts>
-a
~o
to
+^
c
cu
3
**-
? 0.12
0
4->
O
CO
(U
CC
c
\
H *
< O
1 01
S <3 o.os
o
c
c
^t
O
to
<3
"5
-§ 0.04
Conditions >
[TOS] = 0.40-0.47M in Feed Liquor y
Reactor pH = 7.2-11.5 *
V
c/
D .'n
/
/
D O/
O '
-
/
D % 0
~ / u
n ' a
u y
/
/D
/ £3 Legend
S • O Arthur D. Little, Inc., Reactor
/ n CSTR
/
/
/
/
9^
x^B i i i i i i 1 1 1 1 1 1
0 1-'n"1 ' ' 0.5 1.0 1-5
FIGURE IV-6
Sulfate Concentration in Feed Liquor (M)
(CaSO4/CaSOx) RATIO IN REACTOR SOLIDS AS A FUNCTION OF
SULFATE CONCENTRATION - LIME REGENERATION
-------
0.25
Condition:
Active Sodium = 0.2-2.3M
Sulfate = 0.05-1.7M
0.20
2 0.15
u
03
CD
cc
o
<3
J£
O
0.10
0.05
0365
/JSO£]_\
\ [SO,] /
Legend:
• Pilot Plant Data
O Laboratory Data
10
mols Na2SO4/mol Na2SC>3 in Reactor Bleed Liquor
FIGURE IV-7 (CaSO4/CaSO3) RATIO IN REACTOR SOLIDS AS A
FUNCTION OF REACTOR LIQUOR COMPOSITION
IV-20
-------
the regeneration reaction is carried out to a relatively high pH (where
little sulfite exists). Laboratory data on the effect of the extent of
regeneration on sulfate precipitation, as discussed previously, do reflect
such a "U-shaped curve.
While equations (1) and (14) can be used to predict the amount of sul-
fate that can be precipitated, and thereby the amount of oxidation that
can be tolerated, laboratory data on the stability of the calcium sulfate
precipitated indicate that some redissolution of the calcium sulfate may
occur by continual intimate contacting of the solids with the mother liquor.
It would be expected then, that some calcium sulfate redissolution together
with additional calcium sulfite precipitation would occur in the dewatering
system. No such redissolution could be substantiated in the pilot plant
operations; however, a small amount of redissolution was observed in the
prototype system at Gulf Power Company's Scholz Steam Plant (see Volume III).
During one period of extended operation of the prototype system under stable
conditions, the reactor solids contained an average molar ratio of calcium
sulfate to total calcium sulfur salts of 0.13, while the filter cake ran
0.110-0.115. This corresponds to a 10-15% reduction in the amount of cal-
cium sulfate across the dewatering system.
Applying this redissolution factor (assuming that it holds over all sulfate
precipitation rates), equation (14) becomes:
' \ 0.032([SO"]/[S07])
mols CaSOii \ 4 3 reactor liquor
j ,,;
mol CaSO.. / . 1 + 0.0365([SO"]/[SO"]) (15)
x / cake 4 3 reactor liquor
This equation can,be used in conjunction with estimated sodium losses
(occluded in the filter cake) to predict the concentrations of sodium
sulfate and sulfite required to keep up with given levels of oxidation.
For example, assuming that no sodium salts are lost in the cake, a mol
ratio of CaSOtf in the cake to total calcium-sulfur salts of 0.15 would
be required to keep up with a total system oxidation rate of 15% (0.15
mols 303 oxidized/mol S02 absorbed). According to equation (16) above,
such a calcium sulfate precipitation rate would require a molar ratio
of sulf ate-to-sulf ite of 5.65 in the regenerated liquor. That is, to
maintain an active sodium concentration of 0.5M Na+ with 0.2M as sodium
sulfite, the sodium sulfate concentration at steady-state would be about
1.13M.
In actuality, the steady-state sodium sulfate level would be slightly
lower than 1.13M because some sodium sulfate is lost in the occluded
liquor with the filter cake. This reduces the required calcium sulfate
precipitation. For the same oxidation rate of 15% used in the previous
example, but with 2.5% of the dry cake weight as sodium salts (2.0/0 as
sodium sulfate) , the required ratio of calcium sulfate to total calcium-
sulfur salts drops to 0.135. This would decrease the necessary ratio of
sulf ate-to-sulf ite in the liquor from 5.65 to 5.0; so, for the same "'£
sulfite level, the sulfate concentration would only reach l.OM at steady state.
IV-21
-------
Normally, the system chemistry would self-adjust to the sulfate/active
sodium level required to sustain the rate of oxidation being experienced.
The system would simply be charged to a total sodium concentration con-
sistent with the minimum desired active sodium concentration and the
maximum rate of oxidation anticipated. There is evidence,though, from
prior pilot plant work at ADL and operation of the prototype system that
the maximum amount of sulfate that can be precipitated simultaneously
with calcium sulfite may be limited to 25-30% of the total calcium-sulfur
salts. Also, as will be discussed later, solids properties deteriorate
with increasing sulfate content of the solids. This alone may limit tol-
erable oxidation rates to about 25% of the S02 absorbed.
It is possible to reduce oxidation by increasing total dissolved solids
levels (see Chapter III). Ultimately, though, allowable operating con-
centrations will be limited by the solubilities of the sodium salts, since
at very high TOS levels it may still not be possible to precipitate suffi-
cient calcium sulfate without exceeding sodium salt solubility limits. At
the equivalent of 0.9M sulfite concentration (^2M active sodium including
bisulfite or hydroxide), a 1.3M sulfate level or higher will be required
to keep up with an oxidation rate of only 4-5%. Taking into consideration
the presence of a minimal amount of sodium chloride, such a liquor would
have a sodium sulfate saturation temperature of 75-85°F. At such concen-
trations of active sodium, either the tanks and lines must be heated (par-
ticularly during shutdown), or else sodium salts must be intentionally purged
in the filter cake to maintain lower sulfate levels.
d. Dewatering Properties of Solids
Although the relative performance of the ADL two-stage reactor and the
CSTR were found to be comparable with regard to calcium utilization,
effluent soluble calcium concentrations, and precipitation of calcium
sulfate, there was a noticeable difference in the quality of solids
produced. In prior testing and during the EPA program the two-stage
reactor demonstrated a capability for operating over a much wider range
of system conditions (pH and sulfate concentration) than the CSTR with
respect to the production of waste solids with acceptable settling and
filtration properties.
Two parameters are used in characterizing the solids settling properties:
the bulk or initial settling rate, and the density (solids content) of
the settled solids. The settling rate is primarily a measure of the
crystal size (and,to some degree, crystal shape) and is used along with
the compaction time in sizing the thickener. However, both the settling
rate and compaction time are partly a function of the slurry concentration
and, therefore, are not necessarily a direct measure of the quality of the
solids. The density of the settled solids is primarily dependent on the
crystalline form and is a measure of the slurry concentration that can be
achieved in the thickener underflow. The density of the settled solids
is a better indication of the filterability than is the settling rate,
since it is a measure of how well particles compact under the force of
gravity. These settling parameters are determined for the reactor product
IV-22
-------
slurry by measuring the rate of fall of the meniscus (slurry/clear liquor
interface) in a graduated cylinder, as described in Appendix C.
The filterabllity of the solids is characterized by the insoluble solids
content of the filter cake and by the ease with which the occluded sodium
salts in the mother liquor can be washed from the cake. These are used
rather than filtration rate, since it is of principal importance to pro-
duce a material with good handling properties and with minimal environ-
mental impact rather than simply to minimize filter size.
Comparative Dewatering Properties
The effects of reactor holdup and operating pH on the solids generated
in the CSTR were initially studied in the preliminary, open-loop reactor
tests using settling rate and the density of settled solids to characterize
the quality of the solids.
Table IV-4 shows the effect of reactor operating pH on the quality of solids
produced in a CSTR with a feed liquor containing 0.6M sulfate and 0.46M TOS.
The quality of the solids produced at a pH of 7.5 was good at both 30 minute
and 60 minute reactor holdup times. As pH was increased to 12.0, the quality
of the solids deteriorated as shown by the marked decrease in the solids
settling rates and in the densities (wt % solids) of the settled solids.
This pH effect was much more pronounced for the 60 minute holdup than for
the 30 minute holdup; however, in neither case was the loss of solids prop-
erties irreversible. Returning the pH to a low level resulted in a return
to acceptable solids. In light of the greater sensitivity of the 60 minute
reactor holdup, the shorter 30 minute holdup provides a more reliable design
basis for a CSTR. Regeneration to a pH above 8 required a holdup time greater
than 20 minutes to achieve good utilization of lime.
The effect of sulfate concentration on solids properties over a range of
pH for the 30 minute CSTR (in both open- and closed-loop runs) is shown
in Table IV-5. Increasing sulfate levels results in poorer solids, an
effect which is aggravated by high pH. The observed decrease in settling
rates is indicative of smaller particle sizes, and the lower densities of
the settled, partially compacted solids implies a change in the form of
the crystal agglomerates generated. Electron microscopy and optical
studies of the solids confirm the difference in the crystals. As shown
in Figure IV-8, the poorer settling solids are generally made up of
smaller particle sizes with more needle-like crystal structures, while
the solids with good settling properties are agglomerates of needles.
These differences in settling properties are also reflected in the filter-
ability of the solids. In general, low settling rates and low densities
of settled solids are consistent with low solids content in the filter
cake and poor wash efficiencies. However, it should be noted that as
suspended solids levels in the reactor effluent increase settling rates
generally decrease (settling becomes "hindered") and the density of the
settled solids increases because of the higher compaction force.
IV-23
-------
TABLE IV-4
EFFECT OF CSTR pH AND HOLDUP TIME ON SOLIDS PROPERTIES
General Reactor Condtions: Reactant Feed - Dry Ca(OH)2
Reactor Temperature - 105-120°F
Agitation Rate - 200-250 rpm
Liquor Feed Composition - [TOS] = 0.46M
[804] = 0.6M
pH = 5.8
Reactor Effluent Solids Settling Properties
Reactor
Holdup (mins . )
< 60
i
N3
•P-
30
Approximate Suspended Bulk Settling Wt % Insolubles in
pH Solids (gms/1) Rate (ft/min.) Settled Solids (2-3 hrs.)
7.2-7.4
12.0
7.5
11.9
30
55
35
50
0.16-0.23
0.05
0.20-0.25
0.15-0.20
26-36
9
24
15
-------
TABLE IV-5
EFFECTS OF SULFATE
CONCENTRATION AND
pH ON SETTLING PROPERTIES
OF SOLIDS PRODUCED ON A CSTR WITH A 30 MINUTE HOLDUP
General Reactor Conditions: Reactor Holdup - 30 mins.
Reactant Feed - Dry Ca(OH>2
Reactor Temperature - 105-120 °F
Agitation Rate - 200-250 rpm
Liquor Feed Composition - [TOS] = 0.45-0.55M
pH = 5.8
Reactor Feed Reactor
Approximate
[TOS] Feed [S07] pH
--1 — " — " ~ " -—--ll
0.45-0.55M 0.05M 7.5
0.5 7.5-8.5
0.6 7-8
11.9
0.75 12.0
12.5
0.8 11-11.5
0.9-1.0 8-10
11.5-12.0
1.7-2.3 0.25 12.0
1.5-1.7 7-9
12.2
Effluent
Suspended Solids
(gms/1)
30
30
30
50
50
50
20
25
25
-
80
85
Solids
Bulk Settling
Rate (ft/min.)
0.25-0.30
0.20-0.25
0.15-0.25
0.15-0.20
0.05-0.10
0.05-0.10
0.10-0.15
<0.05
<0.05
0.03
0.08
0.02
Settling Properties
Approximate Wt % Insolubles
in Settled Solids (2-3 hrs.)
45
25
20-25
15
15-20
10-15
10
5-15
5
5-10
25-30
10
-------
FIGURE 111-8 SCANNING ELECTRON MICROGRAPHS OF SOLIDS
IV-2 6
-------
Table IV-6 compares the settling and filtration characteristics for solids
£o™UCel Concentrated lime mode using both the ADL reactor and the
CSTR. These results show that the density of the compacted solids, rather
than the initial settling rate, is a better indication of the fllterabillty
and washability of the solids. ^^y
The data presented in Table IV-6 show that both reactors are capable of
producing solids with good dewatering properties, and that in both reac-
tors the solids properties deteriorate with increasing sulfate concentra-
tion. However, the CSTR is more sensitive than the ADL reactor to sulfate
concentration and pH. Solids generated in the CSTR at high pH and sulfate
concentration do not settle or filter as well as those obtained in the ADL
reactor under equivalent conditions. This difference is shown in Figure IV-9,
a plot of filter cake solids content versus sulfate concentration at inter-
mediate TOS levels. Since only limited testing with the two-stage (ADL)
reactor was performed during the EPA program, the data shown in Figure IV-9
include results obtained with the ADL reactor in prior work. At solids
contents above about 45% the filter cake is much like a moist powder. At
solids contents below 40%, the cake is wet and relatively easily liquefied.
The effect of sulfate concentration appears to be dampened by increasing
TOS (active sodium) levels as indicated by the data presented in Tables IV-5
and IV-6. This suggests that the effect of sulfate is primarily one of
slowing the reaction rate. Increasing TOS levels results in an increase
in reaction rate, as indicated by the short holdup times required to achieve
good utilization.
It is evident, therefore, that the quality of the filter cake produced is
not a function of the sulfate concentration alone, but rather a function
of the ratio of sulfate and sulfite concentrations. This relationship is
shown in Figure IV-10, in which the solids content of filter cakes produced
in closed-loop runs is plotted against the ratio of sulfate-to-sulfite con-
centrations in the regenerated liquor for the range of solution compositions
tested (and the equivalent rates of oxidation that can be handled as indi-
cated by the solids analyses or as derived from Figure IV-7). As shown,
the multistage reactor results in a filter cake with solids contents (in-
soluble solids) greater than 40% at oxidation rates as high as 20% of the
S02 absorbed; while the filter cake produced using a CSTR for absorbent
regeneration falls below 40% solids content at oxidation rates of between
10% and 15% of the S02 absorbed.
Filter Testing (Washability)
The lowest level of solubles attained in the concentrated active sodium
mode in continuous, closed-loop operations was about 1.2 wt % solubles
(dry cake basis). This was achieved in a run at active sodium foncen-
tration of 0.4M using slightly more than three displacement washes sprayed
onto the cake with two full-cone nozzles in series. As indicated in
Table IV-6, the reduction in soluble solids in the cake is a function
of three variables: dissolved solids levels in the mother liquor the
number of displacement washes; and the quality (solids content) of the
filter cake.
IV-27
-------
TABLE IV-6
COMPARATIVE SETTLING AND FILTRATION PROPERTIES3
Filter Cake Washability
Settling Properties
TOS Reactor System
Level Type — Holdup
(M) (rains . )
0.3-0.5 ADL - 5/45
5/45
8/50
CSTR - 30
30
25
45
1.7-2.1 CSTR - 10-15*
M
1 a
^ All data taken using the
00
[S04]
0.6
0.7
0.8
0.5
0.6-0.7
0.8
0.9-1.0
1.5-1.7
£TO
11.5-12
9-11
11.5-12
7.5-8.5
7-8
11-11.5
8-9
7.5-9
Dorr-Oliver Rotary
3 TSSb Bulk Settling
1 Rate (ft/min.)
30
30
35
30
30
20
25
80
Drum
0.21
0.16
0.15
0.22
0.15
0.12
0.03
Wt % Insolu'oles Filter Cake Solids
in Settled Solids Wt %
(2-3 hrs.) Insolubles
—
25
20
25
20
10
5
0.08 25-30
Vacuum Filter operating under the
52
48
45
45
38-45
30
22
50
following conditions:
Wt % Solubles
(dry basis)
1.5
3-6
6
1.5-3.5
2.5-4
14
31
3.5
Vacuum - 15-17"
Drum Soeed - 75
Approximate
No. of Displace-
ment Washes0
(spray type)d
2.5 (2 FC)
1.5 (1 FC)
1.0 (1 FC)
3.1 (2 FC)
2.5-3.5 (2 FC)
2.3 (1 FC)
0.4 (1 FC)
4.8 (3 FC)
1 Hg
i-90 sees. /revolutii
% Reduction
of Solubles
85
55-75
60-65
65-85
70-80
40-60
5-20
90
on
% Reduction
Displace-
ment Washes
V35
•x-45
^62
-x-25
^25
•^20
VJO
TSS - Total Suspended Solids
cHo. of Displacement Washes = Occluded Water in Cake
Cloth - Polypropylene, multifilament
Wash Water with 30-50 cfm porosity
of Spray Wash - (1 FC)—Single, full-cone spray
(2 FC)—Two, full-cone sprays in series
(3 FC)—Three, full-cone sprays in series
-------
80
N3
VO
60 -
JB
.Q
g 40
20
Legend:
\
O Arthur D. Little, Inc., Reactor (EPA Program), pH = 11.0-12.0
• Arthur D. Little, Inc., Reactor (Prior Work), pH = 11.9-12.9
D CSTR, pH = 7.5-11.5
Arthur D. Little, Inc., Reactor
0.4
0,8
1.2
1.6
2.0
FIGURE IV-9 AVERAGE WT. % INSOLUBLES IN FILTER CAKE VERSUS SOLUBLE SULFATE
CONCENTRATION - INTERMEDIATE SODIUM CONCENTRATION/LIME REGENERATION
-------
75
O
<5
70
65
60
55
50
1
£ 45
tU
5 40
35
30
25
20
Approximate Level of Total System Oxidation Sustained3 (% of SO2 Absorbed)
5 10 15
20
T
Key:
Reactor Type
CSTR, 10-30 Min Holdup, pH = 7-11
Arthur D. Little, Inc., (EPA), 5/40-50 Min Holdup,
pH = 11-12.5
Arthur D. Little, Inc., (Prior), 5/40-70 Min Holdup,
pH = 11.9-12.9
O
\
\
Symbol
D
O
J_
JL
J_
3 4 5
] /[SOg]) Thickener Overflow
Note: aBased upon filter cake data or, where data not available, from Figure IV—7 taking
into account 12% CaS04 redissolution (assumes 1.5% sodium sulfate in cake, dry weight basis).
FIGURE IV-10 SOLIDS CONTENT OF THE FILTER CAKE AS A FUNCTION
OF SYSTEM OXIDATION
IV-30
-------
The practical minimum solubles level that can be achieved was determined
in a special set of tests conducted with the pilot plant filter The
purpose of these tests was not only to determine washability but also
to evaluate the effects of filter control variables on filtration rate
The testing was done using fresh solids produced in the ADL reactor at'
0.4M active sodium and 0.6M sulfate.
These tests demonstrated that the cake can be realistically washed to a
level of 1.2-1.5 wt % soluble solids (dry basis), which corresponds to
90% reduction in solubles at 50% solids and the liquor composition tested.
The reduction in soluble solids achieved as a function of the number of
displacement washes is shown in Figure IV-11. As indicated, about 10%
of the occluded sodium salts (almost all of which are sodium sulfate)
would leave with the cake regardless of the wash rate unless the cake
were repulped. This confirms laboratory data on cake washing. There
is apparently a low, but significant, quantity of liquor in the inter-
stices of the crystalline agglomerates, which is not easily washed from
the cake. The solubles in this liquor amounts to 0.5-1.0% solubles in
the cake (dry basis).
As a comparison, the wash data taken during a continuous run at high TDS
levels are also plotted in Figure IV-11. At very high wash ratios (eight
displacements), the solubles content also approaches 1-2 wt %. However,
the number of displacement washes required at these high TDS levels to
achieve 2.5 wt % solubles is about five, roughly twice that required at
intermediate dissolved solids levels, exceeding the amount of wash water
normally available in the process water balance for high-sulfur coal,
dual alkali applications.
The maximum cake washing capability for a medium- or high-sulfur coal
application would be roughly equivalent to two or three displacement
washes (0.4 gpm/Mw for a cake with 50 wt % solids). The number of dis-
placement washes will be limited by the amount of makeup water available
for washing as determined by the process water balance and possibly by
the operation of the filter. Assuming cake wash behavior similar to that
seen in the ADL pilot plant using two sprays in series, 75-85% reduction
in soluble solids concentration in the occluded cake liquor can be ex-
pected for this level of washing. With a system liquor containing
0.6-0.7M sulfate, 0.5M active sodium, and 0.2M sodium chloride, the
resultant solubles in the cake would then be 2-3% (dry cake basis) for
a cake containing 50 wt % insoluble solids. To achieve this same level
of solubles at high dissolved solids using a wash ratio of 4.5-5 would
require roughly 85-100% of the total makeup water available for the
system (assuming similar washing behavior as in the pilot plant). Such
a large amount of wash water was available in the pilot plant water
balance because the lime was fed to the system as dry hydrated lime,
not as a slurry; and because the rate of evaporation of water to the
gas in the scrubber system was about 30% higher than would be expected
in an actual boiler application due to the high flue gas temperatures
(saturated gas temperatures of 140-150°F). Even assuming that the same
level of sodium losses could be achieved with less water, either by
IV-31
-------
48-55% Insoluble Solids
In Cake
2 -
56789
Number of Displacement Washes
10 11
12
13
FIGURE IV-11 FILTER CAKE SOLUBLES LOSSES
(TWO OR THREE SPRAYS IN SERIES)
IV-32
-------
increasing the solids content of the cake to 60-70% solids (which could
not be attained in the pilot plant) or by increasing the wash efficiency
the resulting wash water requirements could easily exceed the average rite
allowable by the overall system water balance. verage rate
The exact amount of solubles loss depends upon a number of factors in-
cluding dissolved solids levels, the type of wash used (spray configura-
tion), the operating conditions for the filter, and especially the solids
properties. The degree of washing that is desirable will also depend in
large part on the presence of impurities — the rate at which they enter
the system and the tolerable steady-state impurity levels within the system.
Centrifuge Testing
A six-inch solid bowl centrifuge was tested for comparison with the filter.
In general, the centrifuge produced waste solids with 5% to 10% higher
solids content (55-65% solids versus 50-55% solids). However, the solids
were "masticated" by the centrifuge and were discharged as a much wetter
feeling material with considerably poorer handling properties than those
of the filter cake. Also, the pilot centrifuge had no provision for cake
washing. It is not expected that high sodium recovery efficiency can be
achieved using a centrifuge.
e. Sulfite Oxidation in the Reactor System
A significant amount of oxidation occurred in both reactor systems (CSTR's
and ADL) in some of the early runs due to use of reactor vessels outside
their normal operating capacity (i.e., at very low levels where the pro-
peller agitation caused excessive splashing of the solution). In the
initial open- and closed-loop continuous reactor tests, oxidation rates
ranged from 0.01-0.30 mols TOS/liter-minute (equivalent to 10 ppm to 280 ppm
of S02 in the pilot plant system).
Subsequent batch and continuous runs showed that the rate of oxidation was
strongly dependent on the degree of agitation, the exposed surface area of
the vessel, and the total dissolved solids concentration, indicating that
oxidation was primarily a function of gas-liquid contacting. It was found
to be essentially independent of sulfite concentration in this concentrated
regime, except as it decreased oxygen solubility. Use of proper vessel
dimensions (i.e., liquid height-to-diameter ratios of 1:1 and higher) and
control of agitation (stirrer speed and impeller size) in later runs re
duced oxidation to less than 50 ppm of the S02 absorbed. Sealing o± the
reactor system virtually eliminated oxidation.
1. Summary of Overall System Operation
a. General Operating Conditions
Seven closed-loop, continuous runs were made in the pilot plant in the
concentrated active sodium mode using lime for absorbent regeneration
Six of the runs involved a CSTR (10-45 minutes holdup) and one used the
IV-33
-------
ADL reactor. In addition to the different reactor systems, operating
variables included S02 inlet concentration (2,400-2,800 ppm and 850 ppm),
oxygen content of the flue gas (4-7.5%), active sodium and sulfate con-
centrations, and the use of both Na2SOtf and Na2C03 for sodium makeup. In
all runs, hydrated lime was metered to the reactors as a dry solid. Gen-
eral conditions for each run are listed in Table IV-7.
These runs were intended to confirm and extend open-loop data and to test
the system operation under a variety of conditions. Although not all
process sections were operated under optimal conditions during these runs,
the operations did reflect conditions that could be encountered in some
process applications. Since much of the data generated during these runs
has already been reviewed, the following discussion will be focused on the
overall system performance as related to the operating conditions and the
performance of each process section.
In each run, the system was primed with liquor of predetermined concen-
trations and the system operated under a specified set of conditions for
three to six days (four to ten system holdup times). Although steady-state
operation was not achieved in all runs, sufficient data were collected to
characterize the operation of each section for the given set of conditions.
The overall system configuration for these closed-loop runs is shown in
the process schematic in Figure IV-12.
The scrubber system consisted of the venturi (AP = 10-14 inches of water)
followed by the tray tower containing two trays. The regenerated liquor
was fed directly to the top tray and there was no recycle of any liquor
around the tray tower.
b. System Performance Results
Six criteria are used to characterize the system performance:
• S02 removal;
• lime utilization;
• oxidation/sulfate control;
• solids properties;
• sodium losses; and
• system reliability and operability (i.e., resistance
to upsets, ease of operation, scale potential, and
special process problems).
Table IV-8 gives a summary listing of the key results of each of the
closed-loop runs. Figures IV-13, IV-14, and IV-15 show schematic
process block diagrams for three representative runs — 402, 404,
and 421. These block diagrams express the various system inputs'
outputs, and operating characteristics in terms of the S02 absorbed.
IV-34
-------
TABLE IV-7
M
U>
Run No.
Duration of Run (days)
Inlet Flue Gas:
S02 (ppm)
02 (vol. %)
Temperature (°F)
Reactor System:
Type
Holdup (mins.)
Operating pH
Scrubber Operation:
Absorber L/G (gals/
Macf sat'd - avg)
Top Tray Feed pH
Absorber AP -
2 Trays +
Demister ( H00)
f.
Absorbent Liquor
Composition :
Na+active(M)
SO" (M)
Sodium Makeup Form
OPERATING CONDITIONS FOR CLOSED-LOOP OPERATIONS
Scrubber System: Venturi + Two Trays
Venturi AP = 12-12.5" H20
Venturi L/G = 15-17 gals/Macf sat'd
401 402 403 404 420 421 422
3.5 4 2.5 3 7 7.5 3.5
2,650 2,600 850 2,800 2,400-2,500 2,400-2,500 2,400-2,500
4-5 4-5 4-5 4-55 5 7.5
390-410 390-410 390-410 370-390 360-390 380-410 370-390
CSTR ADL CSTR CSTR CSTR CSTR CSTR
40 . 60 25 30 30 10-15 10-15
7-10 11.5-12 11-11.5 7.5-8.5 7-8 & 11-12. 5 7-9b 7.5-11.5
2.7 2.2 1.4 2.8 2.3 1.35 1.3
8-9 11.5-12 11.5-12 7.5-9 7.5-9 & 7-9b 8-11.5
11-12.5
0.55 0.45 0.25 0.45 0.5-0.75 1.8-2.1 1.95-2.1
1.1 0.8 0.8 0.5 0.7-1.1 1.5-1.7 1.6-1.7
».2S04 »a2S04 ^SO, »,2C03 »a2C03 ^ ^00,
Except in run 403—see text.
Operation at a pH range of 11-12.5 was examined for a brief period.
-------
Tray Scrubber
With Demister
And 2-Trays
Flue Gas
Sodium
Makeup
Mix Tank
\\\\\\\\\v
\\\\\ \\N\\\\\\\\\\ \ \ \ \ \A \ \ \ >\ N |N^ \\ \ \ \ \ '
OJ
Ca(OH),
Settler
Reactor System
FIGURE IV-12 SCHEMATIC FLOW DIAGRAM FOR CONCENTRATED LIME MODE
-------
TABLE IV-8
10
Run No.
Operating Conditions
Inlet Gas:
S02 Level (ppra)
0, Level (vol. Z)
Scrubber Operation:
Absorber L/G (gals/Macf sat'd - avg)
Top Tray Feed pH
Total Feed Stoichiometry (avg)a
Scrubber Bleed Liquor:
PH
[Ha+ac|.],(M)
, (M)
Reactor System:
Type (holdup-rains.)
Operating pH
Sodium Makeup Form
_ Key Operating Results _
S02 Removal Efficiency (avg. Z of inlet)
Calcium Utilization (Z of avail. Ca(OH)2)c
Calcium Feed Stoichiometry
mols Ca(OH)2/mol (AS02 + Na2S04 Added)0
Soluble Calcium in Scrubber Feed (ppo)
Sodium Makeup:
mols Na2-H-/mol (4S02 + Na2SO4) -actual
System Oxidation:
Scrubber (ppra S0_)
Reactor -I- Dewatering (ppm SO-)
Total (ppm S02)
Total (Z of aS02 - to nearest O.SZ)
Sulfate Precipitation (% of AS02)
Filter Cake Composition:
Wt% Insoluble Solids
Wt% Solubles (dry basis)
No. Displacement Washes
SUMMARY OF RESULTS FOR CLOSED-LOOP RUNS
401
2,650
4
2.7
8-9
1.25
5.9-6.1
0.45
0.9-1.1
CSTR (40)
8-9
92
97
0.97
70
0.075
160
290
450
18.5
16
22
31
0.3d
402
2,600
4-5
2.2
11.5-12
1.25
5.8-5.9
0.45
0.8-0.85
AM, (5/45)
11.5-12
91
96
1.03
60
0.065
200
290
490
20.5
13.5
45-50
6
ld
403
850
4-5
1.4
11.5-12
1.4
6.1-6.3
0.25-0.35
0.8
CSTR (25)
7.5-11
96
95
0.86
500-700
0.025
240
65
305
37.5
18
29
16
0.7d
404
2,800
4-5
2.8
7.5-9
1.25
5.8-6.0
0.45
0.5
CSTR (30)
7.5-8.5
94
96
0.96
30-70
0.15
210
120
330
12.5
9
45-50
2
3
420
421
422
2,400-2,500 2,400-2,500 2,400-2,500
5 5 7.5
2.3 1.36 1.3
7.5-9 & 7-9b 8-11.5
11-12.5
1.3 1.9 2.3
1.36
7-9b
1.9
6.2-6.4
1.8-2.1
1.5-1.7
6.0-6.2 6.2-6.4 6.3-6.5
0.5-0.75 1.8-2.1 1.95-2.1
0.7-1.1 1.5-1.7 1.6-1.7
CSTR (30) CSTR (10-15) CSTR (10-15)
7-8 & 11-12.5 7-9b 7.5-11.5
94
97
0.98
20
Varied
245
120
365
16
11
40-55
2-3
3
94
97
1.03
15-30
0.04
75
90
165
7
4-5
50-55
3.5
5
96
97
0.99
15-30
0.055
175
90
265
11.5
4-5
50-55
3.5
5
a Feed Stoichiometry £molsNa capacity/mol inlet S02-
Operation In a pH range of 11-12.5 was examined for a brief period.
c Based upon 91Z Ca(OH)2 In raw lime.
More wash water was available in these runs but was not used.
-------
, 250 ppm S02
U)
oo
2,600 ppm SO2
>„_
Na2SO4
6.5% of AS02
pH = 5.8-6.0
Scrubber System
91-92%SO2 Removal
Oxidation = 8-9% of ASO2
lNaj active -'45M
[S04] = 0.8M
** I
Regeneration System
Oxidation - 9-10% of ASOo ^ Active Lime (Ca(OH)2)
Sulfate Precipitation - . <,n ... en
13-14% of AS02 103% °f A S°2 + Na2S04
'
1
Dewatering System
Oxidation - 2.5% of ASO2
pH = 1 1.5-1 2.0
HoO
T
Cake (45% Insoluble Solids)
Average Cake Composition (wt% dry basis):
CaS04 = 13%
CaSO3 =73%
Other Insolubles = 8%
Na2S04= 5%
Na2SO3= 1%
FIGURE IV-13 CONTINUOUS RUN 402 - ARTHUR D. LITTLE , INC., REACTOR
(Na2SO4 MAKEUP)
-------
200 ppm SO2
pH = 5.8-6.1
[Na+
2,800 ppm SO2
Scrubber System
90-95% S02 Removed
Oxidation = 8.0% of AS02
T
active
[so;;] - O.BM
,45M
r \
Makeup Water
Regeneration System
Oxidation = 3% of ASO2
Sulfate Precipitation =
9% of ASO2
Active Lime (Ca(OH)2)
96% of ASO.H
OJ
VO
Na2CO3
Feed Stoich = 1.2
2.5% of A SO2
Dewatering System
Oxidation = 2% of A SO2
pH = 7.5-8.5
H20
Three Displacement Wash
Cake (45% insoluble solids)
Average Cake Composition (wt% dry basis):
CaSO4 = 10%
CaSO3 = 83%
Other Insolubles = 5%
Na2SO4 = 1.5%
Na2SO3 = 0.5%
FIGURE IV-14 CONTINUOUS RUN 404-CSTR REACTOR (Na2CO3 MAKEUP)
-------
150ppmS02
2,450 ppm S02
Scrubber System
94% S02 Removal
Oxidation = 3% of A S02
I
-P-
O
Feed Stoich = 1.9
4% of A SO2
pH = 6.4
[Na+
active
[S0=] =1.7M
= 2.1M
Dewatering System
Oxidation =* 2% of A SO.,
Regeneration System
Oxidation (unsealed) —
2% of A S02
Sulfate Precipitation =
4.5% of A SO0
Active Lime
103% of AS02
pH = 7.5-8
H20
Five Displacement Wash
Cake (50-55% Insoluble Solids)
Average Cake Composition (wt % dry basis):
CaS04 = 4.5%
CaS03 = 87%
Other lnsolubles = 5%
Na2S04 = 2%
Na2S03 = 1.5%
FIGURE IV-15 CONTINUOUS RUN 421 - CSTR REACTOR (Na2CO3 MAKEUP)
-------
S02 Removal
As shown in Table IV-8 there was no difficulty in achieving S02 removal
rates on the order of 90% and higher in any of the runs. The S02 removal
rate was primarily a function of the scrubber pH or feed stoichiometry
(mols active sodium capacity/mol S02 inlet) at which the system was oper-
ated. The scrubber bleed pH's shown in Table IV-8 represent the range of
operation corresponding to the S02 removal efficiencies given. For the
most part the scrubber bleed pH's (and removal efficiencies) were main-
tained in these ranges during the runs.
k"
In the runs at intermediate active sodium concentrations (0.25-0.75M
active Na+), S02 removal efficiency varied from 91% at a bleed pH of
about 5.8% to 96% at a pH of about 6.2. At the high active sodium
levels (runs 421 and 422), slightly higher pH's (or feed stoichiometries)
were required to achieve equivalent S02 removal efficiencies as in the
intermediate active sodium runs. This is attributed to the higher S02
equilibrium partial pressures over the more concentrated solutions.
Lime Utilization
In all runs, lime utilization exceeded 95% of the available calcium
hydroxide in the raw lime feed. Available calcium hydroxide is mea-
sured by titration of the lime with HC1 to a phenolphthalein endpoint.
In general, the raw lime contained 95-96% calcium (reported as calcium
hydroxide) and 91% calcium hydroxide.
The lime feed stoichiometries (mols Ca(OH)2/mol (AS02 + Na2SOlf added) )
typically ranged from 0.96 to 1.03, except in run 403, where extremely
poor solids properties prevented a reasonable approach to steady-state.
In this run, the lime feed stoichiometry was 0.96.
Excluding run 403, the total alkali feed stoichiometries (mols (Na2C03 +
Ca(OH)2)/mol (AS02 + Na2S04 added) ) generally ranged from 1.02 to 1.09
during stable operating periods. While these feed stoichiometries are
good, it should be pointed out that they are still somewhat greater than
would be expected in full-scale operations over extended periods. These
were relatively short-term operations (a few days) and sodium makeup rates
frequently exceeded sodium losses.
Oxidation and Sulfate Precipitation
The amount of oxidation experienced in the system and the rate of calcium
Sulfate precipitation differed in each run depending upon the particular
set of conditions being tested. The range of total system oxidation
examined in the seven closed-loop runs was 7% to 37% of the S02 absorbed.
Since the system performance at low levels of oxidation ( < 10*) is quite
good, the closed-loop runs focused on oxidation rates greater than iu/o.
The one run at 7% oxidation was made at high active sodium concentrations
to evaluate sulfate precipitation.
1V-41
-------
Of importance in these runs is how the system with different types of
reactors (ADL versus CSTR) and different active sodium levels performed
at the different levels of oxidation, and how the system conditions af-
fected the relative rates of oxidation. As discussed previously (Chap-
ter III), the amount of oxidation experienced in the pilot plant scrubber
system (which generally accounts for the majority of the total system
oxidation) is higher than that which would be experienced in full-scale
systems due to particular pilot plant gas conditions.
Calcium sulfate precipitation has been determined by a careful analysis
of the filter cake and reactor effluent solids. Oxidation rates in the
scrubber and reactor systems have been determined from material balances
around each of these process sections individually. Oxidation rates in
the dewatering system (thickener and filter) were determined from overall
system balances over a period of at least 24 hours, taking into account
changes in inventory within the system and sodium makeup and loss rates.
The direct determination of sulfite oxidation in the dewatering system
is difficult, since flows are not constant, and wash water and vacuum
pump seal water are continually added to the thickener.
The first three runs, 401-403, were made at relatively high rates of oxi-
dation. In runs 401 and 402, total system oxidation ran 18-21% of the S0£
absorbed primarily because of the high rates of oxidation in the reactor
system (which were subsequently reduced by adjusting vessel dimensions
and controlling agitation). In run 403, at the same oxygen content in
the flue gas but at a low inlet S02 level, oxidation totalled 38% of the
S02 absorbed. In all of these runs, sodium sulfate was used as the makeup
source. Therefore, the effective oxidation rates (actual oxidation plus
sulfate addition) amounted to 25-27% in runs 401 and 402, and about 40%
in run 403.
At these high rates of oxidation, sulfate levels rose to 0.8M and higher
in order to precipitate the required amount of calcium sulfate. The CSTR
(runs 401 and 403) was not capable of producing solids with good dewatering
properties at these high sulfate levels and steady-state operation was not
achieved. With the multistage reactor (run 402), on the other hand, rea-
sonably good solids were produced. However, as shown in Figure IV-13, the
rate of sulfate precipitation and soluble sulfate losses in the cake (with
minimal cake washing) did not quite equal the rate of sulfate additions plus
sulfite oxidation, and there was a slight increase in the sulfate-to-active
sodium ratio throughout run 402. Assuming the same sodium losses in the
cake, this trend would have continued until the sulfate level reached
1.1-1.2M if active sodium were maintained at 0.45M, or until it reached
about 0.9M if both the active sodium and sulfate were allowed to vary
together keeping total sodium constant (see Figure IV-7).
In runs 404 and 420 with the improved reactor design (minimizing reactor
oxidation), total system oxidation rates were 12.5% and 15.5% of the S02
absorbed respectively. In these runs, using sodium carbonate rather than
sodium sulfate for makeup, the rate of sulfate precipitation and soluble
sulfate losses in the cake approximated the rate of oxidation.
IV-42
-------
In runs 421 and 422 the high total dissolved levels
the rate of sulfite oxidation in the .crubb.r'^St
creased oxidation in the reactor and dewatering systems. Although It la
difficult to accurately determine the amount of oxidation in the reactor
or dewatering systems independently, it is likely that oxidation in the
reactor system decreased and that in the dewatering system remained con-
stant or possibly increased. The high total dissolved solids level in
the reactant liquor in runs 421 and 422 would reduce the rate of oxygen
mass transfer to the reactant slurry. In the dewatering system, on the
other hand, the dissolved solids level of the filtrate would be greatly
diluted by washing of the cake, and oxygen mass transfer in the filter
would not be expected to be appreciably different from that in runs at
lower dissolved solids levels.
Although the high IDS decreased oxidation, the high active sodium levels
also greatly reduced the amount of sulfate precipitation. Calcium sul-
fate levels in the filter cakes in both runs 421 and 422 were analyzed
to be 4-5% of the total calcium-sulfur salts (molar basis). This amount
of calcium sulfate precipitation along with about 2-2.5% sodium sulfate
losses was sufficient to keep up with the sulfite oxidation estimated in
run 421 (at 5% oxygen in the flue gas). However, it was not sufficient
to keep up oxidation in run 422 (7.5% oxygen in the flue gas). This is
confirmed in Figure IV-16, in which normalized active sodium concentration
in the system liquors (active sodium/total sodium) is plotted with time.
As shown, the ratio of active sodium to total sodium increases throughout
run 421, but decreases throughout run 422.
The difficulty that may arise in operating with such high dissolved solids
levels at oxidation rates in excess of 5% is that sodium sulfite is more
soluble than sodium sulfate. Figure IV-17 shows the solubility limits of
mixed sodium sulfite/sulfate solutions as a function of temperature. As
oxidation rates climb, sulfate concentrations increase and sulfite levels
correspondingly decrease. Thus, as oxidation increases, so does the satu-
ration temperature and the probability of crystallizing sodium salts in-
creases. For example, operating with a 5.0M total sodium concentration
and with 5% oxidation, the equilibrium sulfate and sulfite concentrations
in the regenerated liquor would be roughly 1.35M SO^ and 1.15M S03 (accord-
ing to Figure IV-7, taking into account redissolution of CaSO^ in the thick-
ener and assuming 2% solubles losses). As shown in Figure IV-17, the satu-
ration temperature of this solution is about 24°C (75°F). If oxidation
increases to 10%, then the sulfate will climb to about 1.85M and sulfite
will drop to about 0.65M, and the saturation temperature will increase
to 27-28°C (^82°F).
The above discussion of saturation temperatures does not take into account
the presence of chlorides in the solution, which can have a major impact
on saturation temperatures and the potential for crystallization of sodium
salts. The concentration of chloride attained in the process liquor will
primarily depend upon the level of chloride in the coal, the degree to
which the cake is washed, and the total dissolved solids level in the
liquor. For example, a high-sulfur coal containing 0.1 wt % chloride
IV-43
-------
0.4
Flue Gas Oxygen Levels
- 5%02
RUN 421
7.5%
RUN 422
0.39
c
o
I
o
CJ
E
.2 0.38
•o
.O
'^
£
•M
C
§ 0.37
o
0.36
Venturi Bleed
Thickener Overflow
Excluding Total COg
Due To Makeup Soda Ash
0.35
FIGURE IV-16
I
5^8
Days Of Operation
10
11
12
NORMALIZED ACTIVE SODIUM CONCENTRATION
VS TIME IN RUNS 421 AND 422
IV-44
-------
M
-------
will result in soluble chloride concentrations of 15,000 ppm to 30,000 ppm
(2.5 wt % to 5 wt % NaCl) in the liquor for a system operating with 12-15 wt
TDS and the cake washed to 2.5% solubles. In a system operating at 35 wt %
IDS, on the other hand, chlorides levels of 35,000 ppm to 85,000 ppm would
be reached (5.5 wt % to 14 wt % NaCl) if the cake were washed to 2.5% solu-
bles. Thus, the effect of chloride can force operation at lower TDS levels
to prevent crystallization of sodium salts.
Solids Properties and Filter Operation
As discussed previously, solids properties are primarily a function of the
operating pH and the sulfate-to-active sodium ratio in the system liquor
as dictated by the total system oxidation. Increasing sulfate-to-active
sodium ratios in the system liquor results in decreasing solids content
of the filter cake and poorer cake washability. Solids properties also
tended to deteriorate with increasing reactor operating pH, particularly
using a CSTR for regeneration. This effect of pH was most pronounced at
higher sulfate-to-active sodium levels. In runs at high active sodium
concentrations (runs 421 and 422) in which sulfate-to-active sodium levels
were low (<1.0), the CSTR produced reasonably good solids at operating
pH's up to 11.5. Similar solids were not generated in closed-loop runs
at lower active sodium levels (runs 401, 403, and 420) where sulfate-to-
active sodium levels were greater than 1.0.
The filter cake solids content was found to be an accurate measure in
general of the overall solids properties. High solids content was usu-
ally indicative of high filtration rates (Ibs of dry solids/ft2-hr),
minimal cracking on the filter and good wash efficiency, and acceptable
handling properties. In most cases, 45% insoluble solids was adequate
to ensure a cake with good washing characteristics, a reasonable filtra-
tion rate (>25 Ibs dry solids/ft2-hr), and acceptable handling properties,
This means cakes that discharge readily, are relatively dry, and can be
handled and transported without excessive sticking to equipment.
Based upon these criteria the CSTR appears to be suitable for use where
oxidation rates do not exceed 12-15% of the S02 absorbed. The multistage
reactor, on the other hand, appears to be capable of producing good solids
at oxidation rates up to 20-25% of the S02 absorbed (see Figure IV-10).
These data were generated using a multifilament, polypropylene cloth with
a porosity equal to 30-50 cfm of air at a pressure drop of 1/2 inch t^O.
In general, it was necessary to maintain the cake thickness below 1/2 inch
in order to minimize cracking and ensure good wash efficiency. This was
accomplished by adjusting drum speed (0.5 rpm to 1.0 rpm), drum submergence
(5% to 30%), and thickener underflow concentration. The exact conditions
depended upon the quality of the solids. In most cases with reasonably
good solids the thickener underflow was maintained between 15% and 25%
insoluble solids (even though higher slurry concentrations could frequently
be achieved) in order to ensure the proper cake thickness and provide a
pumpable slurry. Subsequent operation of the prototype system refined
these operating conditions and demonstrated that 5-10% higher solids
IV-46
-------
content could be achieved at equivalent process conditions. This can be
attributed to a difference in filter cloth (monofilament versus multim
ment) and better control of the thickener/filter operation.
One area of particular concern in the pilot plant operation was filter
cloth blinding. In most cases blinding only occurred when poor solids
were generated. However, in runs at high IDS levels the solids appeared
to blind the cloth relatively easily, even though the cake solids content
averaged almost 55% insoluble solids. Usually, the cloth did not require
cleaning during a one-week run when solids content approached or exceeded
50% solids; however, during the high TDS runs (421 and 422) the cloth re-
quired cleaning once a day to ensure a reasonable rate of filter cake dis-
charge. Although the reason for this is not known, it may have been due
to crystallization of soluble salts in the cloth fibers.
Sodium Losses
Losses of soluble sodium in the pilot plant operation were primarily in
the cake, with the amount of sodium lost a function of the cake quality,
the wash ratio used and the TDS concentration in the liquor. At inter-
mediate total dissolved solids levels (10-15%) , two to three displacement
washes were usually adequate with good cake quality to reduce solubles
losses below 3%. In some early runs, though, optimal wash ratios were
not used. Notably, in runs 401, 402, and 403, wash water was set at a
low rate, even though more water was available. This resulted in a wash
ratio of 1 in run 402 in which good quality solids were produced, and less
than 1 in runs 401 and 403 in which poor cake was produced. At high total
dissolved solids levels , wash ratios on the order of five or more displace-
ment washes were required. Such wash ratios, as discussed previously, are
unrealistically high for normal closed-loop operation.
While cake losses accounted for most of the sodium lost, there were also
losses due to sampling, liquor entrainment in the outlet gas, and system
leaks. In order to keep such losses at a minimum, all pump seal leakage
and non-analytical samples (settling tests, clarity, and pH checks, etc.)
were returned to the system. While in most runs at intermediate active
sodium levels, such losses were maintained at or below about 1% of the S02
removed. In runs at high total dissolved solids levels, unaccounted for
sodium losses are estimated to have run as high as 3% of the S02 absorbed.
Although small, these losses do inflate the sodium makeup requirements
for a small pilot operation and they also point up the greater care that
must be taken at t:he higher concentrations to minimize and, to the extent
possible, collect and return all system leaks.
One source of sodium loss that was evident in the high concentration runs
was liquor entrainment in the scrubber off-gas stream. The amount of en-
trainment is not known, but there was sufficient buildup of salts on the
exhaust fan to necessitate cleaning and rebalancing the fan each week.
IV-47
-------
System Operability and Reliability
Other than problems with poor solids properties and the buildup of salts
on the fan during high-concentration runs, there were few operational
problems. In general, the system was quite stable and handled upsets
in inlet S02 and lime feed with ease.
In all but one run, no scale buildup of any kind occurred in the scrubber
system. In run 403 where oxidation rates ran about 40% of the S02 removal,
sulfite concentrations fell to 0.1M and below (approaching a dilute-mode
operation), resulting in soluble calcium levels in excess of 500 ppm in
the scrubber feed. In this run there was a small buildup of calcium sul-
fite and carbonate in the tray tower, causing an increase in the pressure
drop across the tray tower from about 4 inches 1^0 to 6 inches 1^0 over
the course of four days. In all other runs soluble calcium levels ranged
from 15 ppm to 90 ppm and there was no evidence of scaling or solids deposi-
tion, even when mechanical problems in the dewatering system caused overflow
of up to a few hundred ppm of suspended solids from the thickener.
One other problem of a process nature occurred in the runs at high IDS
levels. During weekend shutdowns the high IDS levels resulted in pre-
cipitation of sodium salts from the liquor, despite insulation of tanks
and flushing of all piping. The saturation temperature of the liquor in
these runs was 20-25°C (68-77°F) compared with 8-15°C (45-60°F) for the
lower concentration runs and on two occasions solution "froze" in the
piping and at pipe/tank flanges. The pluggage was cleared by heating
and water flushing. In large-scale systems this could be a more severe
problem in high TDS operations, particularly where absorption of chlorides
increases total dissolved solids levels.
D. CONCLUSIONS
In the concentrated mode using lime for regeneration, calcium sulfate
will coprecipitate with calcium sulfite at sulfate precipitation rates
equivalent to oxidation rates as high as 25% of the S02 removal. Solu-
tions remain unsaturated with respect to calcium sulfate and have low
soluble calcium concentrations. Process modes can be operated over a
wide range of sodium solution concentrations achieving high S02 removal
(greater than 90%) producing good quality filter cake (45% solids or
greater) containing low soluble solids (2-5 wt % dry cake basis) with
no sulfate purge required. The performance characteristics of concen-
trated lime regeneration modes are summarized in more detail below.
• S02 Removal — S02 removal efficiencies in excess of 90% were
easily achieved with the removal efficiency a function of sodium
solution feed stoichiometry for any particular absorber design.
In all closed-loop runs the feed stoichiometry (scrubber oper-
ating pH) was controlled to ensure better than 90% removal. For
a given design, a slightly higher feed stoichiometry (or operating
pH) was required for high sodium solution concentrations (30-35 wt %
sodium salt solutions) than for moderate concentrations (10-15 wt %°
IV-48
-------
sodium salt solution) to achieve the same removal efficiency
because of the increase in S02 equilibrium partial pressure
with the increase in sodium sulflte/blsulfite concentration.
Lime Utilization -- Lime utilization in the range of 95-100%
can be achieved with reactor holdup times of 25 minutes or
greater when regenerating to a pH of 8 or higher. High uti-
lizations can be achieved at shorter residence times if the
regeneration reaction is not carried beyond neutralization
of the bisulfite. Lime utilization decreases if regeneration
is carried much beyond a pH of 12.5.
Oxidation/Sulfate Control — At active sodium concentrations
above about 0.2M, calcium sulfate coprecipitates with calcium
sulfite upon reaction of the sodium salt solution with lime.
The sulfate/sulfite content .of the precipitated calcium salts
is related to the sulfate/sulfite concentrations in the reactor
liquor by the following relationship:
mols CaSOi»\ = [so
mols CaS03 / \ [S0~] ,
3 ' reactor * 3J / reactor
solids liquor
This relationship describes the coprecipitation phenomenon
over the range of sulfite and sulfate liquor concentrations
used in laboratory and pilot plant ^experiments ( [S0j]> 0.2M,
[SO^]/[SO^] = 0-6). This method of sulfate precipitation is
effective for oxidation rates up to about 25%. At any given
active sodium concentration, high sulfate precipitation appears
to be favored by either partial neutralization of the absorbent
solution or regeneration to pH's well above neutrality ( >11.5),
thereby reducing the sulfite concentration in the reactor liquor
and maximizing the sulfate/sulfite ratio in the liquor.
In a properly designed concentrated dual alkali loop, the sulfate/
sulfite ratio will self-adjust at steady-state so that the rate
of sulfate precipitation equals the rate of sulfite oxidation.
It is possible to achieve this balance over a wide range of
active sodium and sulfate concentrations in dynamic response
to changes in flue gas rates and oxygen and S02 concentrations.
For dual alkali systems operating with high TDS (in the range
of 25-30 wt % sodium salt solutions) oxidation rates can be
reduced by a factor of 2-3 from those encountered at lowered
TDS levels (10-15 wt %). At such high TDS levels, the active
sodium concentration as well as the sulfate concentration
must be elevated in order to promote effective regeneration
reactions and production of solids with acceptable dewaterlng
properties. As a result, sulfate precipitation capability is
limited.
IV-49
-------
Solids Properties — Single-stage CSTR and multistage reactor
systems can produce solids, over a wide range of process con-
ditions, which settle well and filter to insoluble solids
contents of 45 wt % or higher. When using a CSTR as the re-
generation reactor, solids properties deteriorate as the
regeneration reaction is carried to a higher pH range with
the degree of deterioration increasing from pH 7.5 to pH 12.
This effect is worse for reactor holdup times of 60 minutes
than for shorter reactor residence times (30 minutes). Using
a CSTR, solids properties also decrease as the sulfate/sulfite
ratio increases in the reactor liquor (at higher oxidation
rates). In a single-stage CSTR, it is difficult to produce
solids with acceptable properties (45 wt % insoluble solids)
at process conditions consistent with sulfate precipitation
and sulfite oxidation rates much beyond 15%.
Good quality solids can be produced over a wider range of pH
and sulfate concentration using a two-stage reactor system,
consisting of a short residence time reactor (5-10 minutes)
followed in series by a longer residence time second stage
(20-40 minutes). This multistage system produces good solids
at pH levels up to about 12.5 and at sulfate/sulfite ratios
required for sulfate precipitation rates equivalent to about
25% oxidation.
Sodium Losses — For a filter cake containing 50% insoluble
solids, the soluble solids content of the cake can be reduced
to 2-3% (dry cake basis) using the amount of filter cake wash
water which would normally be available when operating closed-
loop in a high-sulfur coal boiler application. At TDS levels
in the range of 10-15 wt %, two to three displacement washes
are effective in reducing the soluble content of the cake to
2-3 wt %. Of this material, 0.5-1 wt % soluble sodium salts
appears to be occluded in the calcium salt crystals and cannot
be washed regardless of the amount of wash water used. About
two to three displacement washes are available for high-sulfur
coal applications. At high TDS concentrations (30%), four to
five displacement washes are necessary to reduce solubles to
the 2-3% level. With only three displacement washes, solubles
losses at high TDS concentrations can be expected to be roughly
twice those when operating at 10-15 wt % TDS levels in the ab-
sorbent solution.
At the lower TDS levels, sodium makeup requirements are on the
order of 2-3% of the total alkali requirement (mol basis) . That
is, roughly 2-3% of the sulfur absorbed from the flue gas leaves
the system as sodium salts with the remainder as calcium salts.
From the above considerations, operating a concentrated lime mode
with TDS in the range of 10-15%, the single-stage CSTR can produce
IV-50
-------
good quality solids (45 wt % or greater) containing 2-3 wt %
solubles at system oxidation rates up to 15%. When using the
multistage reactor system the operability of the process is
extended to oxidation levels in the range of about 25%. In-
creasing TDS reduces oxidation but requires more wash water
to produce the same cake solubles content. At two to three
displacement washes, the solubles content of the cake is pro-
portional to the TDS levels in the system loop.
System Reliability/Operability — In concentrated modes using
lime for regeneration, soluble calcium concentrations range
from 15-90 ppm with the calcium concentration generally de-
creasing with increasing sulfite concentration. No scaling
or deposition of solids was observed in the scrubber loop
during any of the concentrated mode operations. Scrubber
operation and S02 removal were easy to control. The regen-
eration reaction is stable and easy to control, but should
be kept at a pH below about 8 if operating with a single-stage
regeneration reactor. Increasing the TDS level in the system
raises the sodium salt saturation temperature, increasing the
potential for solid sodium salt crystallization in elements of
the system which are permitted to cool.
IV-51
-------
V. SULFURIC ACID TREATMENT — CONCENTRATED MODE
A. LABORATORY RESULTS
1. Introduction
The inherent sulfate precipitation which was observed during lime regen-
eration of concentrated sodium sulfite/bisulfite/sulfate solutions shows
promise of being able to control sodium sulfate at manageable levels when
system oxidation rates are in the range of 25-30%. An alternate method,
capable of dealing with significantly higher oxidation rates, involves
the acidification of the process liquor with sulfuric acid in the presence
of solid calcium sulfite to precipitate dissolved sulfate as gypsum. The
overall reaction involved in this treatment process is:
Na2S04 + K2SQk + 2CaS03 • l/2H20(s) + 3H20 -> 2CaS04 • 2H20(s) + 2NaHS03 (2)
thus, although additional sulfate is introduced as sulfuric acid, both that
sulfate and the Na2SO^ should be precipitated as gypsum. On the basis of
the 1:1 relationship between sulfate (sodium) to be precipitated and sulfuric
acid required, a reaction efficiency (theoretically 100% based on equation 19)
can be defined by expressing the moles of Na2SOit removed from the process
stream as a percentage of the moles of H2SOit added:
„,-,-. . (mols Na2SOu removed \ ,„„„
Efficiency = . *•„ !* —•=—:— ] x 100/4
I mols H2SOit fed /
The primary acid consuming reaction is the dissolution of CaSO$ • 1/2H20
by converting it to the soluble calcium bisulfite:
_ | I
2CaS03 • 1/2H20 + H2SOlf + H20 •> CaSO^ • 2H20(s) + 2HS03 + Ca (16)
The dissolved calcium ion produced in this reaction can then react with the
sodium sulfate to be precipitated to form gypsum:
f 2H20 -> CaSO^ • 2H20(s) (17)
In addition to the primary reaction discussed above, other equilibria would
likely be involved and would result in reaction efficiencies of less than
100%. At the acidic conditions required to dissolve calcium sulfite
(pH * 2-3), measurable amounts of sulfurous acid and possibly bisulfate ion
could exist in solution due to the acid equilibrium reactions:
+ H •*• H2S03
+ H+ i HSOh (19)
V-l
-------
In practice, the streams to the acid treatment could contain other sul-
fites, carbonates and possibly hydroxide. In such cases, sulfuric acid
consumption would be increased and reaction efficiency reduced.
The most acidic process stream normally available from which sulfate could
be precipitated — scrubber effluent — normally contains on the order of
0.05M Na2S03. Even well washed filter cake could contain calcium carbonate
or unreacted lime. Depending upon washing efficiency, significant amounts
of Na2S03 and NaOH could also be present. If the calcium sulfite were ob-
tained as a thickened slurry from the regeneration section — attractive
because of the ease of obtaining it and transferring it to the acidification
process — significantly larger amounts of NaOH and Na2S03 would be present.
The NaOH and Na2S03 will react with H2SOLj according to the reactions:
2NaOH + E2SOi^ •> Na2SOJ+ + 2H20 (20)
2Na2S03 + H2S04 -> Na2SOit + 2NaHS03 (21)
consuming H2SOIt and producing additional soluble sulfate in solution. Any
lime or limestone present in the calcium sulfite feed will react with sul-
furic acid:
Ca(OH)2(s) + U.2s°i+ •* CaS04 • 2H20(s) (22)
CaC03(s) + 2E2SOk + H20 -> CaSOk • 2H20(s) + C02i (23)
These reactions will consume additional acid but will not add additional
sulfate to the solution because of the accompanying precipitation of gypsum.
As a point of departure for the subsequent laboratory studies, a very ap-
proximate estimate was made of the pH to which the solution would need to
be taken for the regeneration reaction to proceed, based upon very dilute
solution solubility product and acid dissociation constants. For the pre-
cipitation of gypsum to occur, its apparent solubility product in water:
K '= tCa"1"1"] [S0i|] - 2 x 10'^ (24)
°P
would need to be exceeded. If [S0~] were l.OM initially, [Ca4"1"] would
need to exceed 2 x 10 ^ M for precipitation to occur. For the precipitation
of gypsum to continue until [SO^] = 0.1M, the calcium ion concentration in
solution would need to rise further to about 2 x
To achieve calcium levels of this magnitude in the presence of the very
insoluble calcium sulfite, [SO"] would need to be held to a very low level
by converting it to bisulfite ion. Considering the apparent solubility
product of calcium sulfite :
Kgp'= [CaJ [SOi] * lO' (25)
a calcium ion concentration of about 10 would require that the level of
sulfite ion be held to about lO'^M. Now, to achieve that sulfite ion
V-2
-------
level in a solution containing sulfite/bisulfite at about 0.5M the second
ionization constant of sulfurous acid:
[H+] „ -7
-
„
- 10
„
[HSOj] - (26)
would indicate that [H+] - 5 x 10~3, or a pH in the vicinity of 3, would
be required.
2. Experimental Results and Discussion.
Prior to studying the acidification reaction in the laboratory in a con-
tinuous reactor, a number of batch experiments (in which R2SOk was added
to Na2SOLf/CaS03 slurries) were conducted in a sealed reactor to better
define the operating pH range and, more importantly, to establish the S02
partial pressure levels generated within the reactor. The Na2SOi+/CaS03
slurries were charged into the reactor, and the headspace above the liquid
was purged briefly with 862 to eliminate air before sealing the reactor.
Sulfuric acid was then introduced into the sealed system from a pressure-
equalized addition funnel while monitoring the pressure within the reactor.
The batch experiments indicated that the operating pH would, in fact, be
in the range of 2.5-3.0. Throughout the course of sulfuric acid addition,
the pressure within the reactor indicated that the partial pressure of 862
was less than 1 atmosphere, and a sealed reactor capable of withstanding a
positive internal pressure would not be required for the reaction. The
reactor could be operated at ambient pressure vented to the atmosphere as
long as the reactor vent tube was of a sufficiently narrow bore to minimize
mixing between the internal headspace gas and the outside air.
On the basis of the background information obtained from batch experiments,
the effects of a number of experimental variables on the regeneration re-
action were studied in the laboratory using the CSTR described in Chapter III,
Section A. Since it was the only material available in quantity at that
time, the calcium sulfite used in these studies was unwashed calcium sul-
fite filter cake produced in an earlier ADL pilot plant run in which con-
centrated sodium bisulfite/sulfite/sulfate solutions were regenerated with
lime. This material contained relatively large amounts of calcium carbon-
ate and calcium sulfate along with lesser amounts of sodium salts as the
following composition indicates:
Compound Mol Percent
CaS03 69.2
CaC03 15.4
CaS04 10-5
Na2S0lt 3-3
V-3
-------
Compound Mol Percent
NaOH 1.0
Na2S03 0.5
This composition is based upon chemical analyses and material balances
for the pilot plant runs in which the material was produced. This filter
cake was slurried with a simulated acidic scrubber effluent solution con-
taining, in all cases, Na2S03, 0.05M; NaHS03, 0.40M; and Na2SOit at either
0.75M or 0.37M. This slurry comprised one stream to the CSTR and either
4.5M or 9.QM I^SO^ was fed as the other reactant.
Both phases of the reactor effluent slurry were analyzed in detail. In
the liquid phase, in addition to measuring total soluble calcium ion and
sulfate ion, measurements of total oxidizable sulfur (TOS) and total
acidity (H+) were made by titration. The TOS concentration is the total
concentration of all sulfur (IV) species present, and the total acidity
(H+)tot is the sum of all acidic protons (titratable to pH = 9) in the
solution. These relationships may be expressed as follows:
[TOS] = [H2S03] + [HS03] + [S0~]
[H+]tot = [HS03] + 2[H2S03] + [HSOJ + [H+]
By subtraction, the "excess acidity", ( [H ] - [TOS]) equals
- [TOS] = [H2S03] + [BSOlj + [H+] - [S03]
However, under the range of conditions over which the reaction was studied
(see Table V-l) , H2S03 and HS03 are the predominant species present. At
pH 2-3, [SOf] is virtually zero and [H+] is less than 0.01M.
The dilute solution value for the second ionization of sulfuric acid, pK2 ,
is about 2; however, dissociation of HSO^ increases as ionic strength is
increased. In the CSTR experiments performed, the ionic strength was in
two general regions — 2.25 (experiments 6-7) and 1.4 (experiments 9-10).
Marshall and Jones8 report values of pK2 for H2SOi+ of 0.82 and 1.1, re-
spectively, at those ionic strengths. At the conditions studied here,
then, HSO^ exists only in small concentrations — the concentration of
HSOij; is at most 2% of the H2S03 concentration in the solution. The two
measurements H"1" tQt and TOS , allow one to determine the amounts of
H2S03 and HSO^ present in the reaction effluent.
The results obtained from the CSTR studies are shown in Table V-l. In the
three experiments in series 6, the regeneration of 0.75M Na2SOii solution wag
studied as a function of pH ranging from about 2.35 to 3.15 with reactor
residence time and temperature held constant. In series 7, temperature
and residence time were varied. In the final three experiments, 0.37M
Na2SOif was studied as a function of pH.
V-4
-------
Table V-l SUMMARY OF LABORATORY CONTINUOUS REACTOR SULFURIC ACID TREATMENT EXPERIMENTS
Experimental Variables
Effluent Concentrations.(M)
Expt.a>b
6-a
6-b
6-c
7 -a
7-b
7-c
9
< 10-a
<-n 10-b
[Na,SOu]
0.75
0.75
0.75
0.75
0.75
0.75
0.37
0.37
0.37
Residence
Time (mln
30
30
30
15
30
15
15
15
15
Temp
) t"C)
33
34
34
34
34
51
33
33
33
pH
2.9
2.35
2.8
2.85
2.8
2.8
2.7
2.2
[TOS]
1.39s
1.49
1.46
1.52s
1.59s
1.56
1.135
1.41s
1.56
1.51
1.68
1.98
1.68
1.75
1.72
1.29
1.60s
1.84
0.35s
0.29
0.36
0.27
0.25
0.25s
0.14
0.073s
0.055
0.024
0.026
0.014
0.034
0.030
0.032
0.041s
0.069
0.107
[ca++][so1| ;
(x 103)
8.5
7.5
5.0
9.2
7.5
8.2
5.8
5.1
5.9
, a^
0.090
0.15
0.55
0.11
0.11
0.11
0.16
0.16
0.22
Effective
laySQii Removal
0.49
0.57
0.40
0.61
0.63
0.62
0.66
0.80
0.85
Efficiency
c „, ,d
Observed
0.38
0.38
0.25
0.42
0.40
0.37
0.33
0.29
0.22
Corrected
0.70
0.62
0.37
0.70
0.63
0.57
0.47
0.55
0.35
Solution to be regenerated simulated scrubber effluent:
[NaHS03] = 0.4 M
[Na2S03] = 0.05 M
[^280^] = as indicated
pH = 5.4 - 5.6
bCalcium sulfite feed maintained at 110 - 120% of stoichiometric except:
Expt. 6-c was deficient; no CaSOs-^l^O in effluent
Expt. 10-a, 10-b fed at 200% of stoichiometric
cMol fraction of entering N32S04 actually removed.
d(Mols Na2SOi, in - mols soluble SO^ out)/mols H2SO4 charged.
S0bserved efficiency corrected for effect of alkaline impurities
In CaS(>3 solids fed.
-------
The effect on the regeneration reaction of reducing the pH from 3.15 to
2.9 was an increased TOS concentration as additional calcium sulfite dis-
solved at the lower pH. In turn, the concentration of_SO= decreased as
the soluble calcium which was produced reacted with S04 to precipitate
gypsum. As indicated by the lower SO^, the "effective Na^SO^ removal
from the simulated scrubber stream (ANa2SOit/Na2SOit) was higher at the
lower pH (this ratio is corrected for all sulfate removed which entered
as H2SOLf or in the calcium sulfite filter cake).
As the pH was decreased to 2.9, the concentration of TOS increased by
0.1M but the total acidity increased by 0.17M as a result of the higher^
concentration of H2S03 present at the lower pH. The ratio of H2S03/HSO^
had risen from 0.09 to 0.15 as shown in the table. The formation of
H2S03 consumed H2SO^ and should have reduced the efficiency of H2SO^ uti-
lization. Thus the increased effective sulfate removal at pH 2.9 was
compensated for by the decrease in efficiency of H^SO^ utilization to
produce an essentially constant overall "observed efficiency" of H2SOit
utilization — ANa2SOit/total H2SOi4. fed.
Since the calcium sulfite cake which was used contained a large amount
of CaC03 as well as other impurities which would consume I^SO^, the
effects of the non-calcium sulfite species were taken into account and
a "corrected efficiency", which should have been realized if pure cal-
cium sulfite had been fed, was computed. As shown in the table, H2SOit
utilization efficiency would have been in the range of 0.6 to 0.7.
In experiment 6-c, the pH was further lowered to 2.35; but, upon analysis
of the reaction products, it was found that no solid calcium sulfite re-
mained in the effluent slurry. The results of this experiment show the
important consequence of operating "calcium sulfite limited" (less than
the stoichiometric amount of calcium sulfite); both the effective Na2SOit
regeneration and H2SOtt utilization dropped significantly. In experiments
6-a and 6-b, about 20% and 10%, respectively, of the calcium sulfite fed
remained in the reaction solids.
Decreasing the reactor residence time to 15 minutes and raising the reaction
temperature from 34 to 51°C (series 7) did not produce significant changes
in effective sulfate regeneration or reaction efficiency.
When the concentration of Na2S04 was reduced to 0.37M in experiment 9, an
effective Na2SOi,. removal of 0.66 was observed at pH = 2.8, a value slightly
higher than in the corresponding experiment at the higher sulfate concen-
tration (7-a); however, both the observed and corrected H2SO[+ utilizations
were significantly lower. (About 5% of the calcium sulfite fed remained
in the effluent solids.) An important factor in the decreased H2S04
utilization is the fact that as the incoming concentration of Na2SO^ is
decreased while the concentration of Na2S03 in the solution is held con-
stant, proportionately more of the acid is utilized for the acidification
of Na2S03 than for the regeneration of Na2S04. Lowering the pH to 2.2
produced an even larger effective Na2S(\removal but sulfuric acid utiliza-
tion efficiency decreased even further.
V-6
-------
Thus, as is not unexpected, removing a given amount of Na2SOu per pass
through the reactor will result in increasingly poorer utilization of
sulfuric acid as the Na2SOit level in the incoming stream decreases As
a consequence, for a fixed process oxidation rate, the Na2S0lf regeneration
process will be increasingly inefficient as the required steady state
level in the process is lowered.
B. PILOT PLANT RESULTS
The evaluation of the sulfuric acid slipstream treatment process involved
both open-loop reactor tests to confirm and extend the laboratory per-
formance data, and the use of the sulfuric acid treatment system in con-
junction with the complete dual alkali absorption/regeneration loop
operating in a concentrated lime mode. In both sets of the tests, the
sulfuric acid reactor system was operated in a similar manner. Filter
cake slurry was fed continuously to the reactor together with 20-30 wt %
sulfuric acid at a rate sufficient to maintain a prescribed reactor pH
in the range of 2.3 to 3.3.
The filter cake slurry was made up as required on a batch basis. In the
integrated system operation, a portion of the filter cake generated during
the run was slurried with venturi recycle liquor. In the open-loop tests,
the slurry was prepared using filter cake saved from prior dual alkali
operations and either simulated venturi effluent liquor or simulated
thickener underflow liquor.
The sulfuric acid reactor consisted of a baffled 30-gallon Pfaudler kettle
vented to the atmosphere. The level in the reactor was adjusted to main-
tain approximately a 30 minute reactor holdup time. Effluent from the
reactor was sent to a 6-inch solid bowl centrifuge for the separation of
the solids. In the full system operation, the centrate was returned to
the lime regeneration reactor.
A schematic of the overall system operation is provided in Figure V-l.
1. Sulfuric Acid Reactor Performance
The results of the sulfuric acid reactor pilot plant tests confirmed ^ the
basic laboratory findings. The reaction proceeds readily to completion
with formation of a gypsum product with good settling properties. Pre-
liminary pilot tests showed that a 30 minute reactor holdup time is
sufficient for essentially complete conversion; and the optimum pH appeared
to be above 2 . 35 .
Subsequent runs were then made over a range of pH from 2.6 to 3.3 in J-hree
different operating regimes: Na.SO, limited, CaS03 llrnted, and approximately
stoichiometrics ratios of Na2S04 to CaS03. The results of these runs are
summarized in Table V-2 and Figure V-2.
As would be anticipated, the runs with Na2SOlf/CaS03 feed ratios Dearest
to the theoretical stoichiometric requirement of 0.5, resulted in the
V-7
-------
<
i
CD
Flue G,r,
Tray Scrubber
With Demistnr and
2 - Trays
Scrubbed Gd
t
Sodium
Makeup
MJx Tank
v r*-
Figure V - 1 Schematic Flow Diagram of Pilot Plant Operation For Sulfuric Acid Treatment Mode
-------
TABLE V-2
SUMMARY
Sulfuric
OF SULFURIC ACID SLIPSTREAM TREATMENT RESULTS -
Acid Reactor Conditions: T
T
pH
Feed Streams:
PILOT PLANT
= 30 min (CSTR) Residence Time
= 80 - 110°F
= 2.6 - 3.3
Solids Source - Washed Filter Cake
Liquor - Scrubber Bleed or Thickener Underflow
Acid Strength = 2.5 - 3.0 M H2SO^
Feed Composition
Operating Run
Regime No.
Na2S04 Limited 056
< 055
VO
Stoichiometric 016-2
016-1
016-3
017-2
CaS03 Limited 050
017-1
052
051
053
a
mols NapSOt^ in - mols
Wt %
Solids
24.
18.5
19,5
19.
19.5
18.5
13.
18.5
10.5
11.
26.5
total E
Liquid
!spXL_
0.63
0.59
0.74
0.76
0.83
0.80
0.58
0.90
0.54
0.57
1,07
tOii out
Composition
_M £H
11.5
6.9
6.9
7.1
6.9
7.2
6.9
7.1
11.0
7.0
11.5
Rf f i r.i pnrv
Solids
CaSC
74
82
78
80
76
70
78
70
74
74
36
Composition, (wt%)
)3 CaSOit
10
7
17
15
18
19
11
19
13
11
52
mols Na2SOit
mol CaSO^
0.32
0.38
0.46
0.48
0.54
0.57
- 0.60
0.67
0.75
0.76
0.95
Reactor
Observed
0.22
0.37
0.52
0.56
0.64
0.61
0.43
0.39
0.22
0.28
0.10
Efficiency^
Corrected"
0.33
0.45
0.55
0.59
0.68
0.70
0.52
0.44
0.27
0.32
0.13
HjSOi,
Efficiency corrected for HaOH, CaCO^ and Na>SO; alkalinity in feed slurry
-------
0.8
b*
CO
to
IN
I
05
CN
to
.
c
01
O
ra
cy
CC
•*
O
CO
CN
T
0.7
0.6
0.5
0.4
0.3
0.2
0.1
Reactor Operating Conditions
pH = 2.6-3.3
Feed Liquor - Scrubber Bleed + Washed Cake
Legend
O-
Corrected For Excess Alkalinity In Feed
Observed
0.2 0.4 0.6 0.8
Feed Stoichiometry (mols soluble SO^ /mols CaSOg) in Feed Liquor
FIGURE V-2 H2S04 REACTOR EFFICIENCY VERSUS FEED STOICHIOMETRY
1.0
V-10
-------
highest reactor efficiencies. (Here, efficiency is defined as the decrease
in mols of sodium sulfate divided by the mols of sulfuric acid used - (mols
Na2S04 in-mols total SO?? out)/mol H2S04 used.) The highest efficiency
attained in the pilot plant operations was 0.64 at a (Na2SOit/CaS03) feed
ratio of 0.54. In cases where the feed ratio was greatly different from
stoichiometric, the reactor efficiencies were found to decrease markedly.
In the sulfate limited regime, acid is wasted dissolving CaS03 which is
not used; and in the CaS03 limited regime, acid is wasted neutralizing and
adjusting the pH of extra feed solution.
The pilot plant data plotted in Figure V-2 cover a range of operating con-
ditions including sulfate and sulfite concentration, reactor pH, ionic
strength, and alkalinity content of the feed slurry. The chemistry of the
system indicates that all of these factors will influence the efficiency
of the reaction. Since all of these effects are confounded to some degree
in the pilot plant results, the correlation of reactor efficiency with
feed stoichiometry contains secondary factors other than analytical error
that contribute to the data scatter. The efficiency/feed stoichiometry
curve given in Figure V-2 should be considered a generalized representation.
An attempt was made to extract the effects of feed alkalinity in order to
clarify the observed performance. This involved adjusting the sulfuric
acid used in each run to account for that acid used in neutralizing CaC03,
NaOH and Na2S03 in the feed. The "corrected" reactor efficiencies based
upon these adjusted acid requirements are listed in Table V-2 and are
superimposed in Figure V-2. As would be expected, the magnitude of the
correction was generally greatest for the runs using simulated thickener
underflow and for runs in the CaS03 limited regime where excess solution
was fed.
The effects of sulfate concentration and reactor pH were specifically
addressed in the laboratory program. This work showed that the optimum
reactor operating pH is roughly 2.5 to 3.0, the range in which almost
all pilot plant runs were made. Furthermore, as sulfate concentrations
increase in the reactor feed slurry, reactor efficiency improves. This
effect is not clearly demonstrated in the pilot plant data due to the
great range of feed stoichiometry, which has a more pronounced influence
on reactor performance.
Finally, the ionic strength of the reactor effluent slurry has a small
effect on efficiency since it determines, along with temperature, the
solubility product of gypsum. As ionic strength increases, the solubility
product should also increase and, therefore, reactor efficiency should
correspondingly decrease. Although the effect on reactor efficiency is
too small to be observed, the increase in the gypsum solubility product
with ionic strength was observed. Figure V-3 shows the apparent solubility
product curve for gypsum (calculated by molar concentrations) predlctea
by ADL along with solubility data generated in both the pilot plant and
laboratory tests. For all but a few runs, the experimentally determined
V-ll
-------
Laboratory
O Pilot Plant
ju, Ionic Strength
FIGURE V-3 EXTENT OF CaSO4 SUB-SATURATION IN H2SO4
REACTOR EFFLUENT - PILOT PLANT AND LABORATORY DATA
V-12
-------
values of the apparent solubility product fall slightly below the predicted
curve (predicted by the method of Kusik and Meissner), indicating that the
solutions may not have been saturated with respect to CaSOk • 2H20 The
validity of this predicted solubility product curve is demonstrated by the
gypsum saturated data presented in Chapter VII.
The pilot plant solubility data shown in Figure V-3 correspond to soluble
sulfate concentrations of 0.1M to 0.4M SO" and calcium levels of 0 01M to
0.04M Ca (400 ppm to 1,600 ppm) . The lowest effluent sulfate levels, and
therefore highest calcium levels, were achieved in cases where the reactor
efficiency was highest.
Although the highest efficiency achieved in the pilot plant tests was
0.64, it is reasonable to assume that refinements in the operating condi-
tions and slight adjustments in the process configuration could raise
efficiencies to above 0.7. For example, the reactor vent gas could be
recovered and used for pre-acidifying the incoming filter cake slurry.
Also, if Na2SOif is used for the makeup sodium value, then this could be
added either as a concentrate or solid directly to the sulfuric acid
reactor feed slurry.
2. Integrated System Operation
Two closed-loop pilot plant tests were run for extended periods with the
sulfuric acid.reactor system operating on a slipstream from the dual
alkali process. The overall system was operated in the concentrated active
sodium mode using hydrated lime for absorbent regeneration. The first
test, run 016, was made at an inlet S02 level of 2,250 ppm. The second
test, run 017, was conducted at an inlet Sp2 level of 650 ppm. The process
conditions and overall results are summarized in Table V-3.
In both runs the operation of the sulfuric acid treatment system was
similar to the open-loop tests as previously described. Washed filter cake
generated during the run was used as the CaS03 source and venturi recycle
liquor provided Na2SOi+. The dual alkali process liquor was originally
primed to a soluble sulfate level of about 0.8M to ensure that sufficient
sulfate was present to produce high reactor efficiency. The rates of feed
of cake slurry and sulfuric acid were initially set such that the total
anticipated oxidation in the dual alkali system could be sustained by the
combined sulfate precipitation in the absorbent regeneration reactor and
sulfuric acid reactor and by the loss of soluble sulfate in the waste
cake.
The performance of the sulfuric acid system in these runs has been discussed
in connection with the sulfuric acid reactor performance in the previous
section. Overall, the results were very good. Efficiencies ranged from
0.38 to 0.64 throughout both tests, and the average efficiency over eacti
test was greater than 0.5. The gypsum produced had good de^erxng prop
erties. Centrifuge cake containing up to 70 wt % insoluble solxds could
be obtained by close control of the centrifuge feed rate.
V-13
-------
TABLE V-3
SUMMARY OF CLOSED-LOOP PILOT PLANT RUNS
Sulfuric Acid Treatment Mode
Run No. 016 017
Dual Alkali Process Operation
Inlet S02 Level, ppm 2,250 650
S02 Removed, ppm 2,050 530
Scrubber Bleed: pH 6.1 6.1
[TOS], M 0.35 0.5
[SO"], M 0.7 0.8
Overall System Oxidation
Equivalent ppm S02 (avg.) 570 400
% AS02 (avg.) 28% ~70%
Sulfate Precipitation in Lime Reactor:
Equivalent ppm S02 (avg.) 250 100
% AS02 (avg.) 12% 18%
Calcium Feed:
mols Ca(OH)2/mol AS02 1.33 1.83
mols Ca(OH)2/mol sulfur input 1.06 1.01
HgSOti Treatment System Operation
H2SOi+ Feed, (mols H2SOtt/mol AS02) 0.26 0.63
Reactor Efficiency 0.52-0.64 0.38-0.61
Sulfate Precipitation:
Equivalent ppm S02 280 180-300
% of AS02 14% 34-57%
Centrifuge Cake Composition:
Insoluble Solids (% total wt) 60-70 53
Soluble Solids (% dry cake) 4-7 15
V-14
-------
The primary effect of the sulfuric acid slipstream treatment system on
the overall process operation in these two runs was in the increased lime
requirements (for ultimate neutralization of sulfuric acid added to the
system). There were no adverse effects on calcium utilization soluble
calcium levels, or properties of solids generated in the absorbent regen-
eration reactor. In all of these respects, the performance of the system
was consistent with that of a concentrated mode dual alkali process without
sulfuric acid treatment.
The consumption of lime, however, rose significantly. In run 016 where
the S02 removal equaled 2,050ppm, sulfite oxidation throughout the system
amounted to about 570 ppm, or 28% of the S02 removal. Approximately half
of this oxidation rate was sustained by sulfate precipitation in the lime
reactor and by soluble sulfate losses in the waste cakes. The remainder,
about 14% of the S02 absorbed, was removed in the sulfuric acid reactor.
Since the reactor efficiency averaged 0.57 (efficiencies ranged from 0.52
to 0.64), the sulfuric acid requirement amounted to about 26% of the S02
absorbed. This acid feed was directly reflected in the required increase
of 26% in the lime feed (1.33/1.06 = 1.26). At the lower S02 inlet level
of run 017, the use of sulfuric acid increased markedly due primarily to
oxidation in the dual alkali system being a higher percentage of the S02
absorbed. In addition, Na2SOi+ was used to make up lost sodium value.
The combined oxidation and sulfate addition resulted in an increase in
the lime feed of 80-85% over that required for S02 removal alone; and a
total sulfuric acid requirement of 63% of the S02 removed.
It is clear from the pilot plant results that the sulfuric acid treatment
scheme is a technically feasible and reliable approach to the removal of
soluble sulfate from dual alkali systems. Because the use of this sulfuric
acid treatment scheme may be costly when applied to systems with high oxi-
dation rates (>40% of AS02) due to the sulfuric acid and extra lime require-
ments, it may be more appropriate for systems with intermediate levels of
oxidation where the rate of sulfate formation cannot be easily handled in
a simple concentrated active sodium mode.
Composite Diagram of Projected Pilot Plant Operations
Figure V-4 summarizes the operational characteristics of the various dual^
alkali process sections for the concentrated lime mode operation with sul-
furic acid slipstream treatment for a high S02 inlet condition. This com-
posite is based upon pilot plant operations and is analogous to the block
diagrams provided in Chapter IV, Section C for the simple concentrated lime
mode. Figure V-4 indicates the various inputs and outputs expressed in
terms of S02 absorbed for a dual alkali system with an overall oxld^on
rate equivalent to 30% of the S02 absorbed. For each of the four P^ess
sections, the observed rates of sulfite oxidation and/or active sodium re
generation are also expressed as a function of S02 absorption.
V-15
-------
2,500 ppm SO2
Na2CO3
5% of AS02
1
Scrubber System
90-95% SO2 Removal
Oxidation = 25% of ASO-
j
H2S04
>fc
20% of ASO2
1
pH = 5.9-6.1
INa+l • ~ 0 5M
md ' active -OIV'
[S04] = 0.8M
i
1 J
r -i
>
L
r
H2S04 Treatment
65% Efficiency
Oxidation - 0% of ASO2
{
Regeneration System
Oxidation = 3% of ASO^ ^ Lime ._
Sulfate Precipitation 1 1 5% of Aso
= 15%ofAS02
i
pH = 7.5-9
r
Dewatering System j_| Q
Two Displacement Wash
1
~ 45%
^
~ 55%
Cake (65% Insoluble Solids)
FIGURE V-4
Cake (45% Insoluble Solids)
Average Composition (wt% dry basis) :
CaSO4 = 1 5%
CaSO3 = 77%
Other Insolubles = 5%
IMa2SO4 = 2%
Na2SO3 = 1%
COMPOSITE PILOT PLANT OPERATION - H2SO4 TREATMENT MODE
Average Composition (wt % dry basis):
CaSO4 = 85%
CaSO3 = 5%
NaHSO3 I
NaS04 f
-------
C. SULFURIC ACID REACTOR MODEL
A model of the sulfuric acid reactor system was developed to simulate the
effects of various dual alkali process conditions on the sulfuric acid
system efficiency. The principal purpose of the model was to provide a
method for estimating reasonable sulfuric acid reactor operating condi-
tions and for determining the sulfuric acid requirements in different
dual alkali system applications. The model is described in detail in
Appendix D.
As developed, the model applies to the operation of the reactor in the
CaS03 limited regime. It is based upon the reaction equations discussed
in Section A of this chapter, and assumes that chemical equilibrium is
achieved in the system. The model determines the amount of sulfuric acid
required to neutralize all alkalinity values in the slurry feed and adjust
the pH into the 2.3 to 3.3 range. The levels of sulf ate and calcium in
solution in the reactor effluent are determined from an estimate of the
apparent solubility product, K^', for CaSOk (Ksp' = tCa**] x [so"] with
concentrations in mols/liter). The value of Ksp used is the average of
those values calculated from the laboratory data for the range of ionic
strengths expected in the pilot plant operations. Since Kg^' is an input
to the model, this can be adjusted for any desired ionic strength.
In order to adjust for pH, it has simply been assumed that enough sulfuric
acid is added to convert up to 20% of the system TOS to H2S03, depending
upon the operating pH. This range of H2S03 concentrations roughly corre-
sponds to data obtained in the laboratory. It would, of course, be possible
to try to determine the H2S03/HS03 ratio from the pH for H2S03. However,
very little data exist which will allow for reasonable estimates of the
reference activity coefficients for the H2S03 and HS03 species at ionic
strengths greater than 1.0. Furthermore, initial tests with the model
indicate that calculated reactor efficiencies were relatively insensitive
to rather large changes in this pH adjustment factor.
No consideration has been specifically given in the model to the formation
of HSO^. At the pH's involved, the amount of HSO^ should be less than a
few percent of the total S(VI) species.
While this model proved adequate for determining the desired range of
operating conditions for pilot plant testing and even, in most cases,
predicting the pilot plant results, it was limited and a more general
model was later developed. This later model was based upon estimates
of the equilibrium composition of the reactor effluent liquor using the
method of Kusik and Meissner? to predict activity coefficients oi the
appropriate species. This later model is described in Appendix D.
Model Application
The model has been used primarily to design pilot P^YT^i^vstem
to determine the sulfuric acid requirement for various dual alkali system
oxidation rates.
V-17
-------
Table V-4 compares the experimental reactor efficiencies from the pilot
plant operations with those predicted by the model. The model tracks the
experimental runs fairly closely and, in most cases, the predicted value
of efficiency is within about 15% of that observed. This agreement is
good, particularly in light of the sensitivity of the model to the solids
level in the feed slurry. A change of 10% in the solids level (from 20%
to 22% solids) changes the efficiency estimate by about 30% (from say 40%
to 52%). Since the solids level is an experimentally determined value
that can vary slightly during a run, this sensitivity is significant.
The effects of other operating conditions on the reactor efficiency tend
to be less than total insoluble solids, with small changes in [OH ], [TOS]
and CaC03 in the cake generally producing small changes in efficiency.
However, this does not mean that the sulfuric acid reactor efficiency is
not sensitive to the levels of these alkaline species; rather it means
that reasonably close estimates of their levels in the slurry feed are
adequate for assessing/predicting reactor efficiency. The impact of
large concentrations of available alkali (either in the solids or liquid)
is illustrated in Figure V-5 in which the model's prediction of reactor
efficiency is plotted as a function of lime utilization. In order to en-
sure 30% reactor efficiency under reasonable system operating conditions,
calcium utilization in the absorbent regeneration system must exceed 75%;
and to achieve 50% reactor efficiency, calcium utilization must exceed 90%.
The effect of changes in sulfate concentration also tends to be small,
although the effect is a function of the actual level of sulfate. As
sulfate concentrations in the feed decrease toward the equilibrium ef-
fluent sulfate concentrations, changes in sulfate concentration become
more important.
Neither the pH adjustment factor nor the value of KSp" was found to have
a strong influence on the model performance over most of the range of
operating pH's and solution compositions tested. A 100% increase in KSp'
from 8 x 10~3 to 16 x 10~3 generally decreases the predicted reactor effi-
ciency by less than 10%. Similarly a 50% increase in the pH adjustment
factor, from 0.2 to 0.3, usually decreases the efficiency prediction by
about 5%.
D. CONCLUSIONS
The sulfuric acid slipstream treatment scheme is a technically feasible
and reliable approach for removal of soluble sulfates from dual alkali
systems. The treatment produces sulfate in the form of gypsum that can
be readily dewatered to 65 wt % insoluble solids or higher. The scheme
adds complexity to any dual alkali mode to which it is applied. The
complexity is reflected in additional capital costs and in increased
operating costs for the sulfuric acid, the additional lime consumed and
the additional solid waste produced.
The amount of sulfuric acid required is important since it directly affects
the overall lime requirement. As the sulfuric acid addition rate increases,
the lime rate must increase accordingly for precipitation of the additional
V-18
-------
TABLE V-4
.MODEL SIMULATIONS OF PILOT PLANT OPERATIONS
Feed Ratio
/moIs Na2SCh+
Vmol CaSQ^
0.54
0.57
0.60
0.67
0.75
0.76
0.95
Experimental
Reactor
Efficiency
0.64
0.61
0.43
0.39
0.22
0.28
0.10
Efficiency
Predicted
By Model
0.72
0.56
0.45
0.43
0.13
0.19
-0.12
V-19
-------
100
80
60
o
c
LU
Oi
cc
*fr
o
to
CM
X
Basis
Slurry Feed-20 wt % Solids
(mols SO4/mols CaSOx)so|ids = 0.1
, Sol'n Cones. - [S04l = 0.75M
[SO|] =0.10M
'l =0,30M
40
20
H2SO4 Feed Concentration - 2.56M
H2SO4 Reactor Efficiency > 30%
20
40 60
Calcium Utili/ation (%)
Req'u
Calcium
Util.
80
100
FIGURE V-5 H2SO4 REACTOR EFFICIENCY VERSUS CALCIUM UTILIZATION
V-20
-------
sulfur value added to the system. The maximum efficiency of the treatment
scheme ( (mols Na2SO«, removed/mols I^SO,, fed) x 100%) appears to be practi-
cally limited to a maximum in the range of 60-70%. In order to precipitate
sulfate at a rate sufficient to keep up with an oxidation rate of 15% (of
the S02 absorbed), the lime feed requirement will be increased by 25% for
a 60% reactor efficiency.
The efficiency of the sulfuric acid treatment is importantly affected by
the calcium utilization achieved in the absorbent regeneration reactor
in the main dual alkali loop. As calcium utilization decreases in the
main loop the efficiency of the sulfuric acid slipstream treatment de-
creases and acid consumption increases to neutralize unreacted lime in
the filter cake. In order to achieve a 50% efficiency in the sulfuric
acid treatment system, calcium utilization in the main dual alkali loop
must exceed 90%.
Because the use of this sulfuric acid treatment scheme may be costly when
applied to systems with high oxidation rates (due to the sulfuric acid
and extra lime requirements) , it may be more appropriate for systems with
intermediate levels of oxidation where the rate of sulfate formation cannot
be easily handled in a simpler concentrated sodium mode. The consequences
of using the sulfuric acid slipstream treatment approach for sulfate regen-
eration should, therefore, be carefully evaluated in terms of the overall
process operation. In many cases, where oxidation rates are high enough
that they cannot be easily handled by normal concentrated mode operation,
other dual alkali approaches, such as the dilute lime system described in
Chapter VII, might be more promising than a sulfuric acid treatment alone.
V-21
-------
VI. LIMESTONE REGENERATION — CONCENTRATED MODF.
A. INITIAL LABORATORY STUDIES
In view of the increasing cost of the energy required to calcine limestone
to lime, operating cost savings would be realized if limestone could be
used directly in a dual alkali process in lieu of lime. Limestone has
been used in FGD systems which employ direct slurry scrubbing, and the
chemical equilibria involved have been characterized extensively. By
analogy, limestone should be applicable to dual alkali regeneration.
However, because there are significant differences between reaction con-
ditions in direct slurry scrubbing and dual alkali regeneration ~ one
of the most important is the significantly higher ionic strengths found
in concentrated dual alkali solutions — a program of laboratory experi-
ments was undertaken to study the reactiqn between limestone and concen-
trated sodium sulfite/bisulfite solutions. In a parallel effort, the
use of limestone to regenerate dilute sodium sulfite/bisulfite solutions
was studied by personnel at the Research Triangle Park laboratories of
the EPA. The results of those studies are discussed in Chapter VII.
Limestone is being used in two commercial dual alkali systems in Japan.
However, both systems utilize special treatment steps for sulfate pre-
cipitation; both ultimately produce gypsum and are relatively complex
processes. They are described in more detail in a paper by Kaplan.
The ADL laboratory program consisted of an initial series of batch ex-
periments in which the effects of the type of limestone, limestone feed
stoichiometry, reaction temperature, and sodium sulfate level on the
kinetics and equilibria of the reaction were studied. In subsequent
continuous stirred-tank reactor (CSTR) experiments, limestone utiliza-
tion and the physical properties of the product solids were studied as
a function of a number of experimental variables.
Limestone will not react with sodium sulfite solutions to produce calcium
sulfite since limestone is less soluble than the calcium sulfite and does
not dissolve sufficiently to react. The bisulfite ion, HSO~, however, is
sufficiently acidic to react with limestone. This neutralization reaction,
which results in the ultimate precipitation of calcium sulfite, can be
written simplistically as follows:
Ca€03(s) + 2NaHS03 -* CaS03(s) + Na2S03 + H20 + C02 t (9)
In fact, the bisulfite initially present is never completely
because at pH's where the reaction will proceed (up to about PH -7) the
species H2C03, HCO' HSO^, and S0= all exist in sf nl
equilibrium with one another. Nevertheless, based on
one can express the amount of limestone fed as the Perce<*
chiometric amount which would be required to neutralize the
present.
VI-1
-------
tmols CaC03
| x 100%
0.5 x mols
Thus, a feed stoichiometry of 100% is obtained when the mol ratio of
CaCO" to HSO~ fed is 1:2. This definition of feed stoichiometry will
be used throughout the discussion which follows.
Regardless of the nature of the species in solution, it is the removal
of soluble sulfur-containing ions — both sulfur (IV), sulfite, and
sulfur (VI), sulfate — from solution as insoluble calcium salts by
which the effectiveness of limestone utilization is realistically
evaluated. In CSTR experiments, utilization was based on analyses
of the reactor product solids and calculated as follows:
mols CaSO in solids \
7
% Utilization = [ ——= =—£—: ^rr- ] x 100%
total mols Ca in solids I
1. Batch Studies Comparing the Reactivities of
Different Limestones
Since the initial batch experiments were comparative in nature, for the
sake of expediency the precipitation of CaSO^ was ignored and limestone
utilization was computed on the basis of the decrease in the concentra-
tion of soluble sulfur (IV) — total oxidizable sulfur, TOS — in solution
as the reaction proceeded. Thus, if limestone is fed at 100% of stoichio-
metry and it is completely utilized, one would expect [TOS] to decrease
by an amount equal to one-half the concentration of bisulfite ion,
initially present. This relationship can be expressed as follows:
TT4-.ii • „• / T°S in - TOS out , ___
Utilization = o.5 x HSO in x (% Stoich/100) X 100%
\
j
Because the formation of CaSOi^ is ignored in this calculation, its use
leads to conservative estimates of actual utilization.
The initial experiments in the laboratory program compared the relative
reactivities of several types of limestone toward concentrated, acidic
sodium sulfite/bisulf ite solutions. In previous studies related to
limestone slurry scrubbing, Drehmel10 found significant differences in
reactivity which were attributed to differences in both the chemical
and physical properties of the wide variety of limestones tested.
Materials from three sources were available in quantity for testing and
subsequent use in this laboratory program. By chemical analysis all were
more than 95% CaC03. The materials included:
VI-2
-------
. Reagent grade CaC03 (precipitated chalk); Fisher Scientific
Company.
• A natural calcite obtained from Pfizer, MPM Division Clifton
New Jersey. This material (Marblewhite 200) was an industrial
grade, low magnesium limestone produced at Adams, Massachusetts.
• Fredonia limestone ground by and obtained from the EPA/TVA
Shawnee Test Facility.
Based on the sieve analyses shown below, the Marblewhite 200 and the
Fredonia limestones appeared to be similar in their particle size dis-
tributions while the reagent CaC03 was uniformly finer.
Mesh
- 300
-200 + 300
-100 + 200
- 60 + 100
Reagent
100%
-
-
-
Limestone Type
Marblewhite 200
87.5%
11.0%
1.5%
0.2%
Fredonia
85.7%
8.9%
3.7%
0.4%
Examination under the scanning electron microscope (SEM) revealed more
subtle differences which were not evident from the sieve analysis. The
reagent grade material was composed of quite uniformly sized cubes about
8 microns (y) on a side. The Marblewhite particles also appeared crystal-
line but were very irregular and jagged in shape with a longest dimension
ranging from 2 y to greater than 50 y; an average length was estimated to
be about 20 y. The Fredonia material, which was similar to Marblewhite
by sieve analysis, appeared very different under SEM examination. It
appeared amorphous; the bulk of the particles were rough, irregular
spheres about 2-8 y in diameter. Particles larger than 10 y were present
but they appeared to be agglomerates of many smaller particles.
The results of batch reactions of the three limestones with sodium sulfite/
bisulfite/sulfate solutions at 50°C are shown in Figure VI-1. The progress
of the reaction as a function of time was followed by measuring the de-
crease in TOS concentration in solution. The substantially higher reac-
tivity of Fredonia limestone (as compared to reagent CaC03) is immediately
evident from the more rapid TOS reduction obtained with the Fredonia mate-
rial. The solution with which the Marblewhite was reacted had a somewhat
higher initial TOS level than the other two solutions.
To permit a better comparison of reactivity, initial reaction rates were
computed and the values included in Figure VI-1. The overall ^action
rates shown are the change in TOS concentration per minute for the reac
tion time period, .0-15 minutes. The computed overall reaction rates tor
VI-3
-------
0.4
0.3
oo
I
-P-
0.2
Conditions
Limestone Feed = 100% of Stoichiometric
Initial pH = 5.5
[Na2SO4] =0.75M
Temperature = 50°C
ATOS/At(x103)
(gm mols/litermin)
0-15 Minutes
O 6.6
A 4.0
D 4.0
Legend
QFredonia Limestone
AMarblewhite Limestone
D Reagent CaCOg
0.1
I
I
20
40
60
80 100
Time (minutes)
120
140
160
180
FIGURE VI-1
COMPARISON OF LIMESTONE REACTIVITIES IN BATCH REACTIONS
-------
Marblewhite limestone and reagent CaC03 were essentially the same
the rate observed for Fredonia limestone was about 50% faster S
of either of the other two limestones.
With the Fredonia limestone, TOS changed very little after the first 4S
?nUt-S ?n«rT^°n'- WUJ ^ °ther tW° materials> * noticeable reduc-
tion in TOS, indicative of continuing slow reaction, was observed through-
out the remainder of the experiments which were terminated after three
hours .
Plotted at the extreme right-hand side in parentheses in Figure Vl-1 are
the TOS levels which would have been obtained if the limestone which was
fed had been completely utilized to precipitate calcium sulfite. The
final TOS level observed for Fredonia limestone was much closer to the
theoretical level than it was for the other two materials. Utilizations,
based on TOS reduction, in solution, even after three hours of reaction,
were markedly poorer for Marblewhite limestone and reagent CaC03 (81%
and 85%, respectively) than for the Fredonia material (96%). Utilization
of the Fredonia material after only 45 minutes of reaction was about 92%,
whereas utilizations of Marblewhite and reagent CaC03 after 45 minutes
were only 54% and 64%, respectively. If precipitation of CaSO^ had been
taken into account, all utilizations would have been higher; in the case
of Fredonia, utilization would have been about complete.
Because of the chemical similarity of all three materials, the observed
difference in their reactivity probably resulted from the fact that the
Fredonia limestone consisted of amorphous particles with a higher acces-
sible surface area than the crystalline particles characteristic of the
other two.
Because of its significantly higher reactivity in batch tests, the Fredonia
limestone was used almost exclusively in the remainder of the batch and
CSTR experiments that were performed. We do realize, however, that further
consideration must be given to the effect of limestone reactivity before
a complete assessment of its potential for dual alkali regeneration in a
wide variety of situations can be made. The batch reaction method em-
ployed seems to be a relatively simple but realistic means to study
reactivities.
2. Effects of Feed Stoichiometry, Sulfate Concentration,
and Temperature on Reaction Rate
The effect of changing the Fredonia limestone feed Stoichiometry on the
rate of the regeneration reaction at 50°C is shown in Figure VI-2. When
the amount of limestone fed was reduced from 100% to 504 of the stoi-
chiometric amount required to neutralize the HS03 initially present,
the reaction rate during the first 15 minutes of reaction also decreased
significantly, although the percent decrease in reaction rate wa .not
as great as the reduction in the feed Stoichiometry. The theore ™£L
minimum TOS level after 45 minutes of reaction was more closely approached
VI-5
-------
Conditions
Initial pH =5.5
[Na2S04l =0.75M
Temperature = 50°C
CO
O
0.2
ATOS/At(x10°)
(gm mols/litermin)
0-15 minutes
A 8.6
O6.6
D4.0
(D)
(O)
Legend
A 200% of Stoichiometric
0100% of Stoichiometric
D50% of Stoichiometric
0.1
I
1
20
40
60
80 100
Time (minutes)
120
140
160
180
FIGURE VI-2
EFFECT OF LIMESTONE FEED STOICHIOMETRY ON
REGENERATION RATE WITH FREDONIA LIMESTONE
-------
at the lower (50%) stoichiometry, and the resulting limestone utilization
(based on TOS reduction only) was about 97% as compared to 92% when"
feed was 100% of stoichiometric. When the limestonp f^ L • , . n
doubled fro, 100% to 200%, the' i^ti.l^^^I^SfS
and the TOS concentration after 45 minutes had fallen essentially to the
minimum theoretical level. Within experimental error, the concentration
of TOS remained constant during the remainder of the experiment. However
none of the limestone added in excess of 100% stoichiometric reacted;
the overall utilization was only about 50%.
The dependence of the reaction rate on limestone feed stoichiometry ob-
served in these experiments indicates that limestone dissolution is one
of the rate-limiting steps in the regeneration reaction with limestone.
The effect on reaction rate of the concentration of Na2SO[t in solution
at 50°C is shown in Figure VI-3. It is immediately obvious that at
1.25M Na2SOi4, the reaction was significantly slower than at the two
lower Na£SOif levels. Examination of the computed reaction rates for
the initial 15 minutes of reaction reveals that initial rates were
significantly different for each of the Na2SOi+ levels studied. Each
increase in the concentration of Na2SOit was accompanied by a corre-
sponding decrease in initial reaction rate. The reaction rate for the
first 45 minutes was essentially the same for Na2SOit levels of 0.25M
and 0.75M, but the corresponding rate at 1.25M Na2SOit was still signif-
icantly lower.
All of the preceding experiments were conducted at a temperature of 50°C
which is representative of the temperature at which a reactor would oper-
ate if scrubber effluent were neither heated nor cooled prior to regen-
eration. The significant decrease in reaction rate at a lower tempera-
ture, 38°C, is shown in Figure VI-4. The 12°C reduction in reactor
temperature produced essentially the same reduction in the calculated
reaction rate as did increasing the concentration of Na2SO[t from 0.75M
to 1.25M in the preceding experiment.
3. Continuous Reactor Studies of Regeneration
with Limestone
The results of the batch experiments indicated that Fredonia limestone
could be used effectively to regenerate concentrated sodium bisulfite/
sulfite solutions. The regeneration with limestone was studied further
in the laboratory CSTR (continuous, stirred tank reactor) to observe
the physical properties of the solids produced, to ascertain whether
or not sulfate would also be precipitated during the course of the
reaction, and to obtain a more realistic estimate of achievable lime-
stone utilization.
The results of the batch experiments indicated that with Fredonia lime-
stone, achieving nearly complete utilization at a temperature o.t au
and with Na2S04 concentrations of 0.75M or less, required 45 minutes
VI-7
-------
I
00
CO
O
0.2
Conditions
Limestone Feed = 100% of Stoichiometric
Initial pH = 5.5
Temperature = 50°c
ATOS/At(x103)
(gm mols/litermin)
0-15 Minutes
A 9.6
Q6.6
Q4.3
Legend
A 0.25M
O 0.75M
D 1.25M
Na2S04
Na2S04
Na2S04
0.1
_L
I
_L
20
40
60
FIGURE VI-3
80 100
Time (minutes)
120
140
160
EFFECT OF Na2SO4 CONCENTRATION ON
REGENERATION RATE WITH FREDONIA LIMESTONE
180
-------
0.4 -
0.3 -
H
t/3
o
0.2
Conditions
Limestone Feed =
Initial pH = 5.5
[Na2SO4l=0.75M
100% of Stoichiometric
ATOS/At(x10°)
(gm mols/litermin)
0-15 Minutes
A 4.6
O6.6
Legend
A 38°c
O 50°C
0.1
_L
J_
J-
20
40
60
80 100
Time (minutes)
120
140
160
180
FIGURE VI-4
EFFECT OF TEMPERATURE ON REGENERATION RATE WITH FREDONIA LIMESTONE
-------
batch reaction time. Assuming a simple, second-order reaction overall,
then in a CSTR, that extent of reaction would translate to a residence
time of roughly 2-2.5 hours. However, the CSTR system available when
these studies needed to be performed had been specifically designed to
operate at shorter residence times (5-50 minutes) for studying regenera-
tion with lime, and would not operate reliably at longer residence times.
For the sake of expediency, an initial set of CSTR studies with limestone
was performed at the 50 minute maximum residence time obtainable. A new,
larger CSTR was constructed somewhat later and comparative runs at longer
residence times were then carried out.
The observed utilizations of limestone (based on CaSOx/total Ca in the
product solids) in the CSTR experiments as a function of limestone feed
stoichiometry are shown in Figure VI-5. Included for reference are the
utilizations achieved after 45 minutes of batch reaction based on TOS
removal only; the batch data are known to be somewhat low because CaSO^
formation was not included. Even at the 50 minute CSTR residence time,
utilizations of Fredonia limestone greater than 80% were achieved with
a feed stoichiometry of about 60%. When the stoichiometric amount of
limestone fed was increased, utilization decreased significantly.
Utilization of the less reactive Marblewhite limestone was also studied
in the 50 minute CSTR at a feed stoichiometry of 100%. Utilization of
this material was only about 55%; utilization of Fredonia under similar
conditions was about 70%.
Included in Figure VI-5 are the results of a number of experiments in
which the CSTR residence time was maintained at 50 minutes but a portion
of the reactor effluent slurry was collected, filtered, and the solids
recycled to the reactor. In once-through CSTR operation, the weight
percent suspended solids in the reactor effluent slurry was typically
about 2%. In several experiments with Fredonia limestone, the suspended
solids level in the reactor was raised to about 5% by recycling solids,
and, as shown in Figure VI-5, utilization of the limestone increased sig-
nificantly over that observed for once-through operation. Operation with
solids recycle also appeared to increase the utilization of Marblewhite
limestone. The two experiments with Marblewhite were conducted at dif-
ferent feed stoichiometries, but when effluent solids were recycled,
Marblewhite utilization was comparable to that obtained during once-
through operation with Fredonia; in once-through operation, Marblewhite
utilization had been significantly lower than that observed for Fredonia.
Recycling solids effectively increased the amount of time available for
the limestone to react by a factor of 2.5-3. Increasing reactor residence
time by the same factor produced essentially the same improved utilization
as shown for the once-through 150 minute CSTR run.
The solids produced in the once-through CSTR experiments with Fredonia
limestone were analyzed for their CaSOij. content to ascertain whether or
not sulfate was being precipitated during the course of the reaction.
The amount of CaSOif found, ratioed to the total CaSO present in the
VI-10
-------
100
80 ~
1 60
a
O
E
i 40
4^
C
O
u
20 -
CSTR
Runs
0.35M
0.6M
5.5
50°C
[TOS]
[Na2S04
pH (Initial)
Temperature
Fredonia Batch-, 45-Minute - 2% Solids
OFredonia, 50-Minute CSTR, =* 2% Solids
Fredonia, 150-Minute CSTR, — 2% Solids
OFredonia, 50-Minute CSTR •=- 5% Solids by Recycle
AMarblewhite 200, 50-Minute CSTR — 6% Solids by Recycle
AMarblewhite, 50-Minute CSTR =? 2% Solids
80 120 160
Percent Limestone Feed Stoichiometry
200
240
FIGURE VI-5 UTILIZATION OF TWO LIMESTONES IN BATCH AND CONTINUOUS REACTORS
VI-11
-------
solids, is shown as a function of the Na2SOj+ concentration in the reactor
in Figure VI-6. Significant amounts of sulfate precipitation were observed
and were similar to the behavior observed when regenerating with lime.
The sulfate precipitation increased linearly with increasing Na2SOif con-
centration in the reactor solution. In fact, for a given Na2SO|+ concen-
tration, the actual sulfate precipitation with limestone was the same,
within experimental error, as had been observed earlier for lime.
Because the matrix of CSTR experiments was limited in size, the sulfate
precipitation shown in Figure VI-6 was obtained from experiments with
differing limestone feed stoichiometries as indicated in the figure. The
fact that no significant deviations from linear dependence on sulfate
concentration were observed for the wide variations in feed stoichio-
metry, suggests that stoichiometry had little or no effect on sulfate
precipitation when limestone was used.
Total dissolved calcium levels measured in the reactor effluent for the
experiments shown in Figure VI-6 ranged from about 70-200 ppm, increasing
with increasing Na2SOit concentration. However, even at the highest sulfate
levels studied, the solutions did not approach saturation with respect to
gypsum. By X-ray diffraction, the only sulfur-containing crystalline
phase which could be detected in numerous samples of product solids was
CaSOs • 1/2H20. These same two observations were made when regeneration
with lime was studied. When taken together with the essentially iden-
tical dependence of sulfate precipitation on Na2SOj+ concentration, they
suggest that sulfate precipitation by lime and limestone occurs by the
same mechanism, possibly involving the formation of some sort of solid
solution of CaSOit in the CaS03 • 1/2H20 lattice.
The solids produced in these initial CSTR experiments all exhibited settling
behavior similar to that of "good settling material" previously produced in
the laboratory and in the pilot plant when regenerating with lime.
B. SUBSEQUENT LABORATORY STUDIES OF FACTORS AFFECTING THE
PHYSICAL PROPERTIES OF LIMESTONE PRODUCT SOLIDS
With the laboratory experimental results up to that time all indicating
that Fredonia limestone could be used to regenerate concentrated sodium
bisulfite/sulfite solutions effectively, two closed-loop tests with Fredonia
limestone were attempted in the ADL pilot plant. In the first run, the lime-
stone reactor was a simple CSTR with a 90 minute holdup; in the second run,
the CSTR was operated with a recycle of solids from the thickener underflow.
Although calcium utilizations ranged from 75% to 85%, the dewatering prop-
erties of the solids generated in the reactors were poor. Extremely low
settling rates resulted in the carryover of a considerable amount of
solids in the thickener overflow until the scrubber feed liquor contained
2-3 wt % solids. The filter cake reflected the poor quality of solids.
The washed cake averaged less than 40 wt % solids (dry cake basis).
During operating periods when the reactor effluent slurry pH fell below
6.8, bubbling began to occur in the thickener, indicating that a significant
VI-12
-------
0.20
0.16
H
M
U)
Si 0.12
_c
*"x
0.08
<3
0.04
Conditions
Initial pH = 5.5
Initial [JOB] = 0.35M (diluted)
Temperature = 50°C
Limestone Feed Stoichiometries shown (on graph)
(120%)
C>
(60%)
(100%) O
o
^
o
(50%)
O
(70%)
O
I
0.1
0.2
0.3
0.4
[Na2S04l, M
0.5
0.6
0.7
0.8
FIGURE VI-6 PRECIPITATION OF SULFATE BY FREDONIA LIMESTONE
IN 50-MINUTE CSTR EXPERIMENTS
-------
amount of additional reaction was taking place. This contributed to
the solids carryover in the thickener overflow.
1. Effects of Sulfate Level on Settling Behavior of Solids
After those initial pilot plant experiences, laboratory studies with
the seemingly negative goal of producing poor settling solids were
undertaken in an effort to explain the pilot plant results. Three
differences between operating conditions in the laboratory and in the
pilot plant were explored.
First, limestone had been fed as a slurry in the laboratory but was fed
dry in the pilot plant. A CSTR experiment was conducted in the labora-
tory with limestone fed dry and no deterioration of settling properties
over those observed in previous laboratory studies could be detected.
Second, the simulated scrubber bleed solutions which were used in the
laboratory studies did not contain dissolved C02- In the pilot plant,
the scrubber effluent did contain dissolved CC>2 which could have impeded
the reaction. However, good solids settling behavior was still observed
when the solution fed to the laboratory CSTR was saturated with C02.
Third, during the pilot plant tests, the Na2SOtt concentration level ranged
from 0.9M to 0.8M. The highest Na2SOtt level fed to the laboratory CSTR up
to that time had been 0.75M which, after dilution by the limestone slurry,
resulted in a maximum Na2SOij concentration of 0.6M in the reactor. Sub-
sequent laboratory CSTR experiments at higher concentrations of Na2SOif
indicated that the higher Na2SOit level in the pilot plant was probably
at least one factor responsible for the poor settling behavior observed.
The dramatic changes in the settling behavior which occurred when the
concentration of Na2SOif in the laboratory CSTR was changed from 0.6M to
l.OM are shown in Figure VI-7. (Here, the "meniscus position" indicates
the position of the clear liquor/slurry interface, or the volume to which
the solids slurry settles, starting with 100 ml of slurry in a 100 ml
graduated cylinder.) At intermediate concentrations of ^250^ (0.75-0.80M),
the settling behavior observed, for samples taken one hour after the reactor
had filled and begun to overflow, was between the two extremes. At inter-
mediate Na2SOi/ levels, it was further observed that settling behavior
tended to improve and approach that observed for 0.6M Na2SOit as the
reactor was allowed to continue to operate for an additional period of
2-3 hours. However, the very poor settling behavior observed at the
highest Na2SOt,. concentration did not change over a period of several
hours of reactor operation.
2. Effects of Magnesium on the Limestone Regeneration Reaction
Having observed the dramatic effect that a change in Na2SOit level could
have on the regeneration reaction when it was operated under otherwise
constant conditions, it was decided that the effect of other extraneous
VI-14
-------
120
o
o
CL
w
i
t_n
Conditions
Limestone Feed = 100% of Stoichiometric
Initial pH = 5.5
Temperature = 50°C
[Na2S04l = 0.98M; ITOS) jn = 0.36M
after 1 hour; stable with time
[Na2S04] = 0.6M; [TOS] jn = 0.36M
after 3 hours; stable with time
I
I
10
20
30
40 50
Time (minutes)
60
70
80
90
FIGURE VI-7 SETTLING CURVES FOR SOLIDS PRODUCED IN 50-MINUTE
CSTR EXPERIMENTS USING FREDONIA LIMESTONE
-------
species on the regeneration reaction should be studied. Limestone can
contain varying amounts of magnesium; chlorides can be absorbed from
flue gases; iron can be corroded by process equipment; and other soluble
species might build up to significant levels during extended periods of
closed-loop operation. The effects of magnesium were studied in detail
in this program.
The effect of magnesium, added as MgSO^, at a concentration of 200 ppm
(mg/liter) on the reaction of Fredonia limestone in a batch experiment
is shown in Figure VI-8, where TOS precipitation as a function of time
with and without the added magnesium is compared. The 200 ppm of Kg"1"4"
reduced the reaction rate significantly. In fact, if the calculated
initial reaction rates observed with 200 ppm Mg"1"1" present are compared
with those shown in Figure VI-3 when the ^280^ level was increased to
1.25M, it can be seen that the presence of the small amount of magnesium
slowed the reaction more than increasing the ^280^ level from 0.75M to
1.25M.
The settling behavior of solids produced during 50 minute CSTR experi-
ments in which about 100 ppm and 200 ppm of Mg"*"1" had been added to the
solution to be regenerated is shown in Figure VI-9. Settling curves
for experiments in which no MgSO^ was added are included for comparison.
The 19 ppm and 24 ppm Mg"*^" levels shown in Figure VI-9 for these experi-
ments were the amounts of magnesium dissolved from the limestone.
The significant change in settling behavior when Mg*"*" was added can be
seen by comparing the settling curves (B, C and E) for the three experi-
ments in which the concentration of Na2SOt| was held essentially constant
at about 0.6M. The effects of Na2SOif and Mg"1"1" levels in solution appeared
to be similar and, after a fashion, additive. When the concentration of
Na2SOit was reduced to 0.43M (curve D), moderately good settling behavior
was observed even with 208 ppm M.g++ present.
Since, in earlier tests in which magnesium was not intentionally added
to the scrubber bleed, it had been shown that a significant improvement
in limestone utilization could be realized by increasing the CSTR resi-
dence time from 50 minutes to 120-150 minutes, it was of immediate interest
to determine whether or not the deleterious effects of a few hundred ppm
of magnesium would be more or less severe at a longer residence time.
Experiment 65 was carried out, in which a simulated scrubber bleed con-
taining about 300 ppm of soluble magnesium was regenerated with limestone
in a 120 minute CSTR maintained at 50°C. The solids in the effluent slurry
from that experiment settled quite well as shown in Figure VI-10. However,
the rate at which the solids settled became slower for samples which were
taken at later times during the experiment.
Since increasing the CSTR residence time seemed to improve markedly the
poor settling behavior which had previously been observed when a few
hundred ppm of magnesium was present, the level of soluble magnesium
next was raised, in experiment 66, to 2,200 ppm, a level which might be
VI-16
-------
0.4
0.3
M
-vj
C/5
O
Conditions
Initial pH = 5.5
[Na2S04] = 0.75M
Temperature = 50°C
Limestone Feed = 100% of Stoichiometric
0.2
ATOS/At(x103)
(gm mols/litermin)
0-15 Minutes
O 6.6
A 3.3
Legend
O No Mg Added
A 200 ppm Mg Added
0.1
_L
20
40
60
80 100
Time (minutes)
120
140
160
180
FIGURE V\-8 EFFECT OF SOLUBLE MAGNESIUM ON BATCH REGENERATION
REACTION RATES WITH FREDONIA LIMESTONE
-------
120
100
80
Conditions
Initial pH = 5.5
Initial [TOS] = 0.36M
Temperature = 50°C
.Lmnestone Feed = 100% of Stoichiometric
I
i-1
oo
c
o
c
0>
60
40
20
Legend:
A = Expt. C 19
B = Expt. C24
C = Expt. C25
D = Expt. C26
E = Expt. C16
• —• (A) 0.98M SOT ; 24 ppm Mg
) 0.62M SO. ; 202 ppm Mg
-(C) 0.63M S04 ; ~ 120 ppm Mg
(D) 0.43M
208 ppm Mg
— (E) 0.60M SO^; 0.19 ppm Mg
10
20
FIGURE VI-9
30
40
50
Time (min)
60
70
80
90
SETTLING BEHAVIOR OBSERVED FOR SOLIDS PRODUCED IN A 50-MINUTE
CSTR USING FREDONIA LIMESTONE
-------
100
80
60 -
<
M
I
s.
40 -
20 -
Experiment 65
CSTR Residence Time = 2 Mrs
Temperature, 50°C
Magnesium, ^ 300 ppm
Reactor Operating Time
12.5 hr
12 hr
10 hr
8hr
- - 4.5 hr
10
15
20
Time (min)
25
30
35
40
FIGURE VI-10 SETTLING BEHAVIOR OF SOLIDS PRODUCED DURING REGENERATION WITH LIMESTONE
-------
reached in a closed-loop system producing a well washed filter cake.
As shown in Figure VI-11, the solids did not settle as rapidly, nor
did they compact as densely in experiment 66 as they had in the pre-
ceding experiment at 300 ppm magnesium. However, the solids still ex-
hibited better settling properties than did those produced in the 50
minute CSTR with only 200 ppm of magnesium present.
In both of the 120 minute CSTR experiments, the settling properties seemed
to deteriorate slowly as the experiment was allowed to continue. Plots
of the final settled volumes observed at various times for both magnesium
levels are shown in Figure VI-12. After 12 hours of reactor operation,
the final settled volume appeared to be stabilizing in the higher Mg run
(experiment 66) while it was still increasing steadily in the lower Mg
run (experiment 65).
In an effort to improve solids characteristics at 2,200 ppm of Mg, three
additional CSTR experiments were carried out with a different, single
parameter changed in each. In experiment 67 the temperature was in-
creased from 50°C to 80°C; in experiment 68 a two-stage ADL reactor
(a 15 minute CSTR followed by a 120 minute CSTR) was substituted for
the single CSTR; and in experiment 69 effluent solids were recycled
to increase the suspended solids in the reactor from 2% to 6%.
Analyses of the effluent liquors from experiments 65 through 69 are shown
in Table VT-1. Based on the drops in TOS and acidity across the reactor,
increasing the magnesium level from 300 ppm to 2,200 ppm resulted in
about a 10% decrease in the extent of the regeneration reaction. Each
of the three changes in operating conditions at 2,200 ppm Mg increased
the extent of reaction. As compared to experiment 66, the improvement
in TOS reduction across the reactor was about 14% with the ADL reactor
and the CSTR with solids recycled, and about 6% for the CSTR operated
at 80°C.
Analyses of the properties of the solids produced in the five experiments
are shown in Table VI-2. On the basis of the solids analysis, the in-
creased magnesium level in experiment 65 resulted in a reduction in
utilization of about 12%. In both experiments 65 and 66, exhaustive
vacuum filtration of the product slurry on a laboratory Buchner funnel
produced a filter cake containing only slightly more than 32% solids.
Utilizations in experiments 67-69 were consistently greater than those
observed in experiment 66 which was also run at 2,200 ppm magnesium.
While all the product slurries filtered rapidly, there were differences
in the moisture contents of the filter cakes. Relative to the solids
from experiments 65 and 66, using the ADL reactor or solids recycle
improved filtration somewhat, but the solids from the 80°C run (ex-
periment 67) could be dewatered best of all.
Settling curves for experiments 65-69 are illustrated in Figure VI-13.
It can be readily observed that the three different changes in operating
VI-20
-------
100
H
I
to
o
o
O_
OJ
Experiment 66
CSTR Residence Time = 2 Hr
Temperature, 50°C
Magnesium,« 2,200 ppm
Reactor Operating Time
12.5 hr
• 12 hr
10 hr
9.5 hr
10-14 hr; no
40 -
20 -
FIGURE VI-11 SETTLING BEHAVIOR OF SOLIDS PRODUCED DURING REGENERATION WITH LIMESTONE
-------
40
30
[Mg"""] 5= 2,200 ppm
Experiment 66
o
20
<
M
10
7
[Mg"""] = 300 ppm
Experiment 65
_L
I
6 8
Reactor Operating Time (hr)
10
12
14
FIGURE VI-12
CHANGE IN SETTLED VOLUME OF EFFLUENT SLURRY SOLIDS
AS A FUNCTION OF REACTOR OPERATING TIME - EXPERIMENTS 65 AND 66
-------
TABLE VI-1
Expt. Expt'l. Conditions
65 Lower Magnesium, 50°C
CSTR, T = 2 hr, 300 ppm Mg"1^
66 Baseline Expt., 50°C
CSTR, T = 2 hr, 2,200 ppm Mg++
57 Higher Temperature, 80°C
CSTR, T = 2 hr, 2,200 ppm Mg4^"
68 ADL Reactor, 50"C
Ti_=15 min;T2= 2 hours,
2,200 ppm Mg"1"1"
69 Solids Recycle to 6 wt. %, 50°C
CSTR, T = 2 hr, 2,200 ppm Mg44"
ms Observed
Stream
Feod
Effluent
Feed
Effluent
Feed
Effluent
Feed
Effluent
Feed
Effluent
During Regeneration with Limestone in Continuous
U'/"*"] |M
.0129
.0145
.0900
.0896
.0905
.0955
.0860
.0885
.0844
.0849
[TOS],M
.365
.251
.350
.251
.353
.243
.340
.227
.339
.225
+
.338
.116
.324
.131
.324
.120
.318
.098
.309
.116
5.
6.
5.
6.
5.
6.
5.
6.
5.
6.
50
55
32
60
25
54
31
69
30
70
[SOiT
0.
0.
0.
0.
0.
0.
0.
0.
0.
_
Reactors3
],M
62
64
63
65
60
64
60
65
60
_
,«.",.»
0
0.0036
0
0.0049
0
0.0023
0
0.0037
0
0.0038
0
0. :2i
0
O.C30
0
0.020
0
0.020
0
__
^Limestone (Fredonia) fed at 100% stoictiiometry (the exact amount required to neutralize HS03~ fed).
-------
TABLE VI-2
COMPOSITIONS AMD CHARACTERISTICS OF SOLIDS PRODUCED DURING REGENERATION WITH LIMESTONE
Expt. Experimental Conditions
Species in Solids (mm/g)
TOS S04
C03
% CaSOi4
CaSO,,
% Solids in
Utilization wet cakec
65 Lower Magnesium, 50°C „
CSTR, T = 2 hours, 300 ppm Mg
66 Baseline Expt., 50°C
CSTR, T = 2 hours, 2,200 ppm Mg"1"1"
67 Higher Temperature, 80°C
CSTR, T = 2 hours, 2,200 ppm Mg++
7.69 0.09 5.39 0.63 2.01
7.86 0.14 4.66 0.54 2.62
7.80 0.18 5.34 0.48
77
65
74
33
32
47
68 ADL Reactor, 50°C
T =2 hours, 2,200 ppm Mg++
7.64 0.14 5.18 0.79 2.20
12
74
40
69 Solids Recycle to 6 wt. %, 50°C
CSTR, T = 2 hours, 2,200 ppm Mg"4
7.77 0.13 5.17 0.56
72
40
a Magnesium analyses by atomic absorption; Fredonia limestone contained 0.17 mmMg/g.
Corrected for sodium sulfate physically trapped; magnesium sulfite and sulfate
not included in ratio.
T-2% total suspended solids, except for the recycle case.
-------
120
100
i-i
to
Ui
c
o
<£
c
01
80
60
40
20
\x ---^
Experiment 69
~6% Solids with
75 ppm
Betz1,100
\ \
\
\
\
V
h
\
I
Experiment 69
2,200 ppm Mg,
~6% Solids by
Recycle, 50°C
Experiment67
2,200 ppm Mg,
Experiment 68
2,200 ppm Mg,
ADL Reactor
50°C
Experiment 66
2,200 ppm Mg,
50°C
Experiment 60
300 ppm Mg,
50°C
10
15
20
Time (min.)
25
30
35
40
Note: All settling samples taken after 12 hours of reactor operation.
Both CSTR and ADL Reactors had 2-hour residence time.
FIGURE VI-13 SETTLING BEHAVIOR OF SOLIDS PRODUCED DURING REGENERATION
WITH LIMESTONE IN CONTINUOUS REACTORS
-------
conditions all resulted in slower initial settling rates and increased
final settling volumes. Of the curves presented, the ADL reactor solids
and solids from experiment 69, after treatment with a flocculant, Betz 1100,
appeared to have the best settling properties. However, neither material
settled as well as that from experiments 65 and 66.
3. Studies of Liming for Magnesium Control and the Use of
Flocculants to Improve Dewatering Properties
Experiments 65-69 indicated that regeneration with limestone could be
carried out successfully in the presence of soluble magnesium ranging
from 300 ppm to 2,200 ppm in a continuous stirred tank reactor (CSTR)
with a residence time of about two hours. Because of the decreased
limestone utilization and, more importantly, the degradation in solids
settling and filtering properties at the higher magnesium level, it was
concluded that some means of magnesium control and/or the use of floccu-
lating agents to improve settling ought to be investigated for inclusion
in a viable limestone-based dual alkali process.
Initial studies of the use of lime to control soluble magnesium were con-
ducted in conjunction with experiments 65 and 66. Several times through-
out the course of each experiment, 300 g to 500 g samples of the effluent
slurry were collected, placed in beakers, stirred, and the pH monitored
as known amounts of lime were added. At each of several pH values the
slurry was allowed to react for 5-10 minutes. Liquor samples were then
removed and the supernatant liquors later analyzed for soluble magnesium.
Data on pH, amount of lime added, and final dissolved magnesium concen-
tration are shown in Table VI-3 for effluents from the CSTR experiments
which contained 300 ppm and 2,200 ppm of soluble Mg**. The results for
the two 300 ppm cases show that much of the Mg"1""*" could be precipitated
by.adding lime to a pH of 10.6. Small increments of lime above that
level (to pH^11.3) removed essentially all remaining magnesium. Since
the initial magnesium level was low (0.014M), a large portion of the
total lime (about 0.1 mol/1 of solution) was used for precipitating
sulfite.
The data for the two 2,200 ppm tests affirmed the previous results. Al-
though only 0.09-0.096 mol of lime per liter of solution should have
been required to remove all of the Mg++, 0.18-0.19 mol of lime was
actually used. The excess lime was again about 0.09M and was used for
removal of sulfite. On the final_2,200 ppm post-liming test, TOS measure-
ments were made and, assuming HS03/S03 = 9:1, the decrease in TOS agreed
with the amount of excess lime needed^ It should be noted that at the
measured solution carbonate levels of 0.02-0.03M, very little lime was
consumed in forming calcium carbonate.
To study these options more carefully, CSTR experiments at 50°C and two
hours of residence with soluble magnesium levels of 2,200 ppm and 300 ppm,
which were duplicates of earlier experiments,were performed. The ability'
of a flocculating agent to improve the solids settling properties of the
limestone reaction product slurry was studied.
VI-26
-------
TABLE VI-3
MAGNESIUM CONTROL BY LIMING CSTR PRODUCT SLURRY
. a Soluble Mg-H- Ca(OH)2 Added
SamPle PH (Mols/1 of Solution) (Mols/1 of Solution)
=300 ppm Mg"1^, (4t) 6.55 0.0140 0.000
10.6 0.0038 0.100
H-2 0.0009 0.103
11-6 0.0003 0.105
=300 ppm Mg ,(5T) 6.5 0.0145 0.000
10.5 0.0066 0.0847
11-3 0.0000 0.0942
11.6 0.0000 0.0970
=2,200 ppm Mg++, (5T) 6.6 0.0896 0.000
11.6 0.0000 0.191
11.9 0.0000 0.196
=2,200 ppm Mg++, (5T) 6.6 0.0958 0.000b
10.6 0.0165 0.181C
11.6 0.0008 0.204
a
Slurry sample taken after reactor operated for the indicated number of
residence times, T.
[TOS] = 0.262
[TOS] - 0.070
VI-27
-------
A second CSTR with a residence time of 30 minutes was placed after the
limestone reactor to verify the results of earlier batch studies in which
lime was used to control soluble magnesium levels. In these studies, we
attempted to reproduce as closely as possible the conditions employed in
experiment 66 (2,200 ppm magnesium) and experiment 65 (300 ppm magnesium)
discussed previously. Solution concentrations entering the reactor (after
dilution by water accompanying the limestone slurry) were: S0=, 0.6M; TOS,
0.35M; pH, 5.4; limestone feed stoichiometry, about 100%.
Regeneration in the presence of about 2,200 ppm magnesium was studied in
experiment 70, and as is evident from Figure VI-14, the solids in the
product slurry settled very poorly. This behavior was in marked contrast
to that observed when the same experiment had been conducted earlier. The
range of final settled volumes observed previously (experiment 66) is in-
cluded in Figure VI-14 for reference. Adding a polyelectrolyte flocculating
agent (Betz 1100) at concentrations ranging from 1 ppm to 20 ppm improved
settling behavior to some extent, more or less in proportion to the concen-
tration at which it was added (dotted curves in Figure VI-14). Figure VI-14
also shows that the solids in the final slurry after reaction with lime
settled no better than those which had emerged from the limestone reactor.
The solids generated in experiment 71, in which 300 ppm of magnesium was
present, also settled more poorly than did those generated in the corre-
sponding earlier experiment (experiment 65) as shown in Figure VI-15. As
in the earlier experiment, the final settled volume increased as the lime-
stone CSTR was allowed to continue to operate. From an unsettled volume
of 100 ml, the final settled volume increased from about 17 ml after six
hours of operation to about 40 ml when the experiment was terminated after
eleven hours. Weight percent solids in the slurry ranged from 1.7% to 1.8%.
The effect of adding Betz 1100 flocculating agent at the 0.05 ppm level to
the final slurry sample taken after eleven hours of operation is shown by
the dashed line in Figure VI-15. While the final settled volume was not
reduced markedly, the initial settling rate was significantly improved.
By visual observation, addition of the flocculating agent seemed to be
particularly effective in coagulating the finer particles which settled
more slowly and were causing a persistent haze in the portion of the
supernate from which most of the solids had already settled.
Figure VI-16 shows that after reacting the limestone regenerated product
slurry with lime to remove magnesium, the initial settling rate of the
combined solids from the final reactor was slower than that observed
after the initial limestone reaction. The final settled volume, however,
was improved by the liming reaction. Addition of 1 ppm of Betz 1100 again
caused a marked change in the initial settling rate of the lime reactor
solids. The final settled volume was essentially not changed by the
addition of the agent.
A summary of the reactant solution concentrations employed in experiments
70 and 71 and the changes in amounts of soluble species in solution across
each of the two reactors is shown in Table VI-4. Because both the lime-
stone fed to the first reactor and the lime fed to the second were fed
VI-28
-------
100
80
1 ppm
5 ppm
-E 60
o
10 ppm
20 ppm
u
u)
"E
40
Legend:
Lime Reactor (untreated)
20
Limestone Reactor (untreated)
.. Limestone Reactor
(treated with Betz. 1100 at
indicated level)
12hrs.
1 Range of Final
V Settled Volume
1 Observed in
? 6 hrs. ' Experiment 66
10
15
20
Time (min)
25
30
35
40
FIGURE V-14 SETTLING BEHAVIOR OF SOLIDS PRODUCED IN THE PRESENCE
OF 2,200 ppm MAGNESIUM - EXPERIMENT 70
-------
100
U)
o
o
«£
3
O
c
CD
Limestone Reaction Solids
After Indicated Reactor
Operating Time
Solids After 11 hrs. Treated
With 0.05 ppm Betz 1100
11 hrs.
.05 ppm Betz 1100
Range of Final Settled
Volumes Observed in
Experiment 65
FIGURE V-15
20
Time (min)
SETTLING BEHAVIOR OF SOLIDS PRODUCED DURING REGENERATION WITH
LIMESTONE IN THE PRESENCE OF 300 ppm MAGNESIUM - EXPERIMENT 71
-------
100
E 60 -
I
OJ
o
=s
u
c
-------
TABLE VI-4
LIMESTONE/LIME SERIES REACTOR PERFORMANCE1
Reactor Feed , Expt. 71
Concentrations (M) (^300 ppm M
TOS
Acidity
S0=
Mg-H-
PH
Limestone
0.322
0.287
0.586
0.014
5.4
115
Expt. 70
(^2.200 ppm
0.325
0.298
0.608
0.089
5.4
115
Change Across
Limestone Reactor (mmols/min)
TOS
Acidity
Effluent pH
Change Across
Lime Reactor (mmols/min)
TOS
Acidity
Mg++
Effluent Solution
Concentrations (M)
TOS
COjL
Mg^4"
PH
-2.60
-4.93
+0.01
+0.35
6.8
-1.7
-2.68
-0.36
0.102
0.0026
0.0082
0.0002
11.75
-2.05
-4.76
-0.09
+0.44
6.6
-3.49
-3.39
-2.15
0.068
0.005
0.006
0.0009
11.95
Initial limestone reactor, 2 hour CSTR; lime reactor
0.5 hour CSTR.
After dilution by water in limestone slurry fed.
% of stoichiometric amount equivalent to bisulfite fed.
VI-32
-------
as slurries, account needed to be taken for the diluting effect of the
slurry water. To make them more understandable, the changes in concen-
trations in each of the two reactors are expressed as fluxes in units
of mmols/min. In experiment 71, in which about 300 ppm of magnesium was
removed during the lime reaction, about 60% of the total TOS removed by
the combined reactors in series occurred during the reaction with lime-
stone. In experiment 70, where a significantly greater amount of magne-
sium was removed, only about 40% of the total TOS which was precipitated
occurred in the limestone reactor.
The reaction taking place in the lime reactor was essentially composed
of two steps. The acidity remaining in solution after the limestone
reaction was first neutralized — for every two equivalents of acidity
neutralized, one mol of calcium sulfite was precipitated.
Ca(OH)2(s) + 2HSO~ -> CaS03(s) + S0= + H20
After neutralization was completed, the hydroxide produced from the con-
tinued dissolution of lime reacted with magnesium to precipitate magnesium
hydroxide.
Ca(OH)2(s) + Mg*"1" + SO^ -> Mg(OH)2(s) + CaS03(s)
The calcium ion produced reacted with sulfite to precipitate additional
calcium sulfite. The amount of TOS removed during neutralization re-
mained relatively constant regardless of the amount of magnesium present,
while the TOS removal accompanying magnesium hydroxide precipitation was
directly proportional to the amount of magnesium removed. Thus, the
2.68 mmols/min reduction in acidity observed in experiment 71 (Table VI-4)
was accompanied by the removal of 1.34 mmols/min of TOS from solution.
The remaining TOS which was removed during the lime reaction was essen-
tially equal to the 0.36 mmols/min of magnesium removed.
The characteristics of the solids produced in the initial limestone reactor
and the combined solids after treatment with lime in the second reactor are
shown in Table VI-5. The 78% utilization observed in experiment 71 was
quite similar to utilizations observed previously in two hour CSTR ex-
periments at low magnesium levels. The decrease in utilization to 68%
at the higher magnesium level reflects the inhibition of the regenera-
tion reaction which was also observed previously.
The amounts of magnesium in the well-washed limestone reactor solids in-
creased as the soluble magnesium concentration increased. The magnesium
remaining in the solids is a significant fraction of the magnesium present
in the limestone reagent. The Fredonia limestone used in these experiments
had a magnesium content of 0.017 mmols of magnesium per mmol of calcium.
The Mg/Ca mol ratio in the solids from experiment 71 was 0.012. Thus,
VI-33
-------
TABLE VI-5
LIMESTONE/LIME SERIES REACTOR SOLIDS PROPERTIES
Species in
Limestone Reactor Expt. 71 Expt. 70
Solids (mmols/g) (V300 ppm Kg**) (%2,200 ppm Mg++)
Ca44" 7.70 7.85
Mg++ 0.09 0.14
TOS 5.60 4.94
S0| 0.49 0.42
CDs 1.80 2.38
(CaSOtO/CCaSOx) 7% 7%
Limestone Utilization 78% 68%
Wet Cake Solids 34% 30%
Species in
Combined (LS+Lime)
Reactor Solids (mmols/g)
Ca++ 7.51 7.46
Mg"4^ 0.46 1.82
TOS 5.10 4.90
SOTJ 0.43 0.52
CO! n.a. 1.27
OH- 0.30 2.45
(CaSOtt)/(CaSOx) 7% 8%
Wet Cake Solids 37% 38%
a
Not Analyzed.
VI-34
-------
with about 300 ppm of soluble magnesium in the reactor, approximately
two-thirds of the magnesium which entered with the limestone remained
in the solids. In experiment 70, the Mg/Ca mol ratio can be calculated
to be 0.018; with 2,200 ppm of magnesium in the reacting solution, some
of the soluble magnesium was actually being purged in the limestone cake.
Analyses of the solids from the two-reactor sequence indicated that the
CaSOtf/CaSOx ratio of about 7% was maintained during the lime treatment.
The amount of hydroxide in the final combined reactor solids does not
represent unutilized lime, but, rather, the solid magnesium hydroxide
present. In fact, in both experiments, the amounts of hydroxide found
were lower than would be expected for the amounts of magnesium precipi-
tated. The low hydroxide results are caused by the insolubility of
magnesium hydroxide and resulting difficulties in completely titrating
it.
C. PILOT PLANT OPERATIONS
As in the evaluation of the concentrated active sodium mode using lime
for absorbent regeneration, pilot plant testing in the concentrated lime-
stone mode was focused on the performance of the regeneration system. Of
primary concern was the determination of the effects of principal reactor
operating and design parameters (i.e., operating pH, reactor configuration,
and reactor holdup time) and system variables (i.e., sulfate concentration
or sulfate-to-active-sodium ratio and type of limestone) on reactor per-
formance. The purpose was to develop a viable reactor design and establish
the range of operating conditions that would ensure better than 80% utiliza-
tion of limestone and waste solids that could be filtered to a minimum of
45% insolubles.
Preliminary open-loop runs in the pilot plant confirmed laboratory results
regarding the difficulties associated with implementing a limestone dual
alkali system. Three runs made using a single CSTR and a natural calcite
(Marblewhite 200 obtained from Pfizer) with a feed liquor comparable to
that in a concentrated-lime mode operation (containing no soluble mag-
nesium) showed very low limestone utilization and generally poor solids
settling properties. Table VI-6 summarizes the general conditions and
results of these three runs. Even operating with a reactor holdup time
of two hours or increasing the calcite feed rate to more than twice that
required to neutralize all the acidity present in the feed liquor, the
pH in the effluent from the reactor could not be increased above 6.05.
Based upon these unfavorable results and the extensive laboratory testing,
it was decided to utilize a multistage reactor system in subsequent pilot
plant operations. Such a system would approach plug flow or batch reactor
performance and would allow close monitoring of the reactor performance as
the reaction proceeds. The best performance obtained in laboratory studies
had been in batch reactor studies; the multistage approach to the plug flow
reactor should produce similar results as the batch reactor but in continu-
ous operation. Fredonia limestone was also used rather than Marblewhite 200
in all subsequent runs, as laboratory testing had shown Fredonia limestone
to be the more reactive.
VI-35
-------
TABLE VI-6
SUMMARY OF PRELIMINARY OPEN-LOOP RUNS USING A SINGLE CSTR WITH MARBLEWHITE LIMESTONE
Feed Liquor: pH =5.8
[TOS] = 0.48M
[804] = 0.65M
[Mg**] = 0
lea**] = 0
[C02] = 0
Temp. = 48-52°C
Performance
Reactor Conditions
Holdup
(hrs.)
1
2
1
Solids
Recycle
No
No
to ^10%a
Limestone Feed
(% to Neutralize HSO^)
155
135
215
Effluent Liquor
-ES-
5.9
6.0
6.0
[TOS]
(M)
0.41
0.39
—
[Ca^]
(ppm)
210
190
440
Limestone
Utilization
10-15%
15-20%
<20%b
Sulfate
Precipitation
mols CaSOi
mol CaSOy
.15
.11
.14
Settling
Time
(mins . )
8
10
>60
Final
Volume
(ml)C
70
115
'v-SOO
washed filter cake from previous runs with lime
Estimated based upon analyses of product solids and recycled filter cake used.
Q
1,000 ml graduate used for pilot plant settling tests.
-------
1. Open-Loop Multistage Testing
A total of nine runs was made using a reactor system consisting of six
CSTR s arranged in series. The six reactors were identical in design.
Each was a polyethylene tank fitted with four baffles, and a center
entering propeller-type stirrer. The six tanks were arranged for gravity
liquor flow with discharge baffles around the overflow ports to ensure
no short-circuiting of the feed liquor. Slurry from the last reactor
was sent directly to the pilot plant rotary drum vacuum filter for solids
dewatering. It was hoped that the filter operation would help to charac-
terize the solids properties.
Table VI-7 lists the basic test conditions and the range of variables
explored. In all runs the composition of the reactor feed liquor used
was made to simulate that anticipated for a system operating in a con-
centrated active sodium mode under closed-loop conditions.
Run 125 is considered to be the base case. It is a repetition of run 122
(not included in Table VI-7) in which an agitator failure led to question-
able results. In the other seven runs the effects of four process variables
on limestone utilization and solids properties were studied: sulfate (0.65M
to 0.9M) and magnesium (25-2,250 ppm) concentrations in the simulated scrub-
ber bleed; limestone feed stoichiometry (68-184%); and solids holdup time
(by increasing reactor residence time or recycling product solids). In
one run the liquor from the last (sixth) reactor was fed to another
reactor where it was treated with lime to increase pH and precipitate
magnesium (post-liming mode). Throughout all of the runs the temperature
of the feed liquor was set at 52 + 4°C (125 + 7°F).
All runs were started with the reactors empty. Liquor and limestone feed
to the first reactor were started simultaneously. The reactor liquors
were then monitored with time to ensure that steady-state operation was
achieved. Most of the runs lasted about five mean holdup times, a suffi-
cient duration to closely approach steady-state liquor conditions.
Specific operating conditions for each run and the principal results are
summarized in Table VI-8. Four process criteria have been used to charac-
terize the reactor system operation and performance: calcium utilization;
levels of soluble calcium and magnesium in the reactor effluent liquor;
sulfate precipitation; and dewatering properties of the waste solids
produced. These are discussed in the following subsections.
Reaction Rate and Limestone Utilization
Over the range of process conditions, limestone utilization varied from
20% to approximately 90-95% (based upon the average of the solids and
liquid analyses). The effects of variations in process conditions
confirmed the results of the laboratory program. In general, limestone
utilization increased with increasing solids residence time. The highest
utilization was achieved in run 127 in which product solids filtered from
the last reactor liquor were recycled to the first reactor. Figures VI-17
VI-37
-------
TABLE VI-7
CONDITIONS FOR OPEN-LOOP MULTISTAGE REACTOR RUNS
Operating Conditions;
Run 125
Feed Liquor Base Range
pH 5.5 ± 0.2
[TOS], (M) 0.34 ±0.02
[804], (M) 0.7 ± 0.05 0.65 - 0.9
[Mg"1"1"], (ppm) 350 ± 65 25-2,250
[Ca"1"1"], (ppm) 80 ± 20
[C02], (M) 0.02 ± o.Ol
Temp., (°C) 52 ± 4
Limestone Feed Stoichiometry, (%)a 85 * 5 68 - 184
Runs Summary;
Run No. Variable
120 High CaC03 Stoichiometry
121 High magnesium
122 Base run
123 High sulfate
124 High sulfate and post-liming
125 Base run (repeat of 122)
126 Low CaC03 Stoichiometry
127 Solids recycle
I28 Extended reactor holdup time
Calculated as available CaC03 feed/2 x HSO~ feed.
VI-38
-------
TABLE VI-8
M
Lo
VO
SUMMARY OF OPEN-LOOP LIMESTONE RUNS
Run
No.
125
(& 122)
120
126
121
123
124
127
128
Run Variable
BASE
High CaC03 Feed
Low CaCOj Feed
High Magnesium
High Sulfate
High Sulfate &
Post Liming
Solids Recycle
High Reactor
Holdup
Reactor
Configuration
No. of
RXTR's x Min.
6
6
6
6
6
6
1
6
6
x 15
x 15
x 15
x 15
x 15
x 15 +
x 30a
x 15
x 26
Feed Liquor
Reactor Conditions
Liquid Feed
Solids
Recycle
No
No
No
No
No
No
to 4%d
No
[S04
(M)
0.70
0.65
0.71
0.8
0.9
0.9
0.71
0.71
(ppm)
415
290
325
2,250
300
25
345
355
IN MULTISTAGE REACTOR SYSTEM (FREDONIA LIMESTONE)
: pH
[TOS] =
[C02] =
Temp . =
: Temp . =
TSS
Limestone
Feedb
% to
Neutralizi
Bisulfite
82
189
68
79
87
83
95
92
5.3-5.6
0.32-0.36M
60-100 ppm
0.02M
48-56°C
40-48°C
1.2-2% (except Runs 120 & 127)
Limestone Utilization
Effluent Liquid Based on
2 [TOS]
(M)
0.239
0.160
0.272
0.299
0.268
0.260C
0.208
0.215
[Acidity]
(M)
0.142
0.067
0.196
0.267
0.238
0.201C
0.086
0.093
pH
6.2
6.65
6.4
5.7
5.9
6.0
6.7
6.4
Solids
Analyses
74
—
64
24
20
44
91
76
Liquid
Balance
mo Is CaSO,
mols CaSO,
Settling
j Time
(min.)
75
54
72
24
23
37
98
84
0.13
~
0.16
0.24
(0.15)
0.18
0.12
0.13
8-10
12-14
5-6
15-17
14-16
12-14
Very Poor
Poor (30-40)
aPost liming was carried out in a seventh reactor having 30 minutes residence time.
Based on 93% available calcium carbonate in the raw limestone feed.
°Liquor from the post liming reactor had the following composition: pH = 11
[TOS] = 0.:.24M
[Mg-H-J = 0
Solids recycled to first reactor were filtered solids from last reactor.
-------
through VI-24 show the change in TOS concentration and acidity through
the successive reactors in these eight runs.
The rate of reaction is also dependent on the availability of calcium
from the limestone (or quantity of limestone present). The effect of
calcium availability can be seen from the dependence of reaction rate
on limestone feed stoichiometry in runs 120, 125, and 126. The average
initial rate of reaction in these runs as determined from the decrease
in TOS across the first reactor is plotted in Figure VI-25. As shown,
the rate of reaction roughly doubled with almost a threefold increase
in the limestone feed rate. This confirms similar laboratory results
and indicates that limestone dissolution is one of the rate-limiting
steps.
However, the mechanism and kinetics of calcium sulfite precipitation are
important too. This is clear from the effects of solids recycle on the
overall limestone utilization. Recycling product solids (which contained
only about 10% unutilized calcium carbonate) increased utilization from
about 75% in run 125 to over 90% in run 127. The increase in reaction
rate is also apparent in Figures VI-17 and VI-23. It should be noted,
though, that part of the difference in the initial TOS decrease between
runs 125 (Figure VI-17) and 127 (Figure VI-23) is also partly due to the
increase in limestone feed stoichiometry (82% in run 125 versus about
100% in run 127) and to the liquor occluded with the recycled solids.
However, these factors account for only a small fraction of the overall
difference in rates.
Soluble Calcium Concentrations
The rate of reaction is also reflected in the soluble calcium concentra-
tions in the reactor system. Figures VI-26 through VI-29 show the con-
centrations of TOS and soluble calcium through successive reactors in
the last four runs (calcium concentrations in the first four runs were
measured only in the last reactor), and Figures VI-30 and VI-31 show
soluble calcium concentration as a function of TOS in the first reactor
(last four runs) and in the last reactor (all runs).
As shown, the soluble calcium concentration is related to the decrease
in TOS (or increase in sulfite concentration by neutralization of bi-
sulfite). Calcium levels increase to a high level in the first reactor
where sulfite concentrations are low (but TOS high), and then decrease
through the remaining reactors, tracking TOS (as sulfite concentrations
increase).
It is not possible to correlate calcium concentrations directly with
sulfite because sulfite is not measured directly and there are uncer-
tainties in carbonate analyses and the bicarbonate/carbonic acid split
that preclude accurate estimates of sulfite from TOS, acidity, and
carbonate analyses. Since the estimates of sulfite concentration have
some uncertainty, it is difficult to accurately determine the apparent
solubility products ( [Ca++] x [S0~] ) in these solutions. However, in
VI-40
-------
0.4
0.1
0.4
0.1
[SO4~] = 0.7 M
] =415ppm
123456
Reactor Effluent Number
FIGURE VI-17 RUN 125 (BASE CASE)
[804] = 0.7 M
[Mg*] = 325 ppm
012345 6
Reactor Effluent Number
FIGURE VI-19 RUN 126 (LOW CaC03 FEED)
0.4
0.3 <
c
o
0.2 -
u
c
0.1 -
123456
Reactor Effluent Number
FIGURE VI-18 RUN 120 (HIGH CaCO3 FEED)
£ 0.2
u
0.1
Acidity
[SO* ] = 0.8 M
[Nig*] = 2,250 ppm
_L
_L
J_
_L
12345 6
Reactor Effluent Number
FIGURE VI-20 RUN 121 (HIGH MAGNESIUM)
VI-41
-------
0.4
c
o
o
c
o
c
o
o
[S04 ] = 0.90 M
= 300 ppm
0.1 h
23456
Reactor Effluent Number
FIGURE VI-21 RUN 123 (HIGH SULFATE)
0123 4567
Reactor Effluent Number
FIGURE VI-22 RUN 124 (HIGH SULFATE + POST- LIMING)
[SO4 ] = 0.7 M
= 355 ppm
2345
Reactor Effluent Number
FIGURE VI-23 RUN 127 (SOLIDS RECYCLE)
2345
Reactor Effluent Number
FIGURE VI-24 RUN 128 (EXTENDED HOLDUP)
VI-42
-------
2 4
X
c
1
(120)
O q
E J
0)
EC
I 2
iZ
c
i
tn
(126)
(125)
Run No. Shown In Paretheses
I
50
100 150
Limestone Feed (% of HSOs Neutralization)
200
FIGURE VI-25 REACTION RATE VS. LIMESTONE FEED STOICHIOMETRY
VI-43
-------
0.4' i
o
- 100
012345
Reactor Effluent Number
FIGURE VI-26 RUN 125 (BASE CASE)
0.4
0.3
to 0.2
O
0.1
Ca++
800
700
600
500
400
300
200
100
a
a.
Q
012 3456
Reactor Effluent Number
FIGURE VI-28 RUN 127 (SOLIDS RECYCLE)
- 400
- 300
- 200
- 100
0123456
Reactor Effluent Number
FIGURE VI-27 RUN 126 (LOW CaCO3 FEED)
2 3456
Reactor Effluent Number
FIGURE VI-29 RUN 128 (EXTENDED HOLDUP)
VI-44
-------
600
500
I 400
a
300
200
100
0
0.15
J_
0.20
0.25
[TOS], (M)
0.30
0.35
FIGURE VI-30 SOLUBLE CALCIUM VS TOS IN REACTOR 1
600
500
400
£
a.
a.
300
5 200
100
0
0.15
0.20
0.25
[TOS], (M)
0.30
0.35
FIGURE VI-31 SOLUBLE CALCIUM VS TOS IN REACTOR 6
VI-45
-------
all reactors the apparent solubility products are higher than those pre-
dicted by the method of Kusik and Meissner,7 as in the concentrated lime
mode (see Figure VI-32). This could reflect supersaturation, inaccuracy
in the predicted curve, and/or the fact that the solubility product curve
for pure calcium sulfite does not apply to the combined calcium sulfite/
sulfate solids produced.
It is interesting that the apparent solubility product for the high mag-
nesium run is greater than those of any of the other runs even when low
estimates of sulfite concentration are used. This may result from inter-
ference of magnesium in calcium sulfite precipitation.
Dewatering Properties of the Waste Solids
The settling curves for the solids produced in these runs are shown in
Figures VI-33 through VI-40.
In general, the settling properties of the solids produced deteriorated
with increasing solids holdup time, as seen in runs 127 and 128. The
solids properties in run 127 (with solids recycle) were particularly bad.
In 24 hours the solids in the slurry from the last reactor settled only
20%, and the family of settling curves shown in Figure VI-39 show that
the settling rates for the solids from all reactors were equally low.
The properties of the solids were also reflected in the consistency of
the filter cake. The cake produced in run 127 was quite thixotropic in
nature, rather than gritty and porous as in most of the other runs, al-
though in many cases the gritty, porous nature of the cake was due to
the presence of large quantities of unreacted limestone.
Unfortunately, the relative solids content of the various filter cakes
did not follow the trends in the settling data. Most of the filter cakes,
including that in run 127, ranged between 35% and 40% solids. However,
the solids content data for the filter cake is of questionable signifi-
cance in these runs, since the unthickened effluent slurry was fed directly
to the filter. Because of the low flow and the low solids levels, there
were usually not enough solids to completely cover the filter cloth.
Therefore, air leakage through open cloth area and thin solids layers
frequently resulted in low vacuum and poor dewatering.
The deterioration of settling properties with increasing solids holdup,
as seen in runs 127 and 128, is consistent with behavior observed in
laboratory batch experiments. Settling time in the batch experiments
tended to increase as the run progressed even after most of the reaction
was complete. It would appear that the major effect is one of holdup time,
but it is difficult to divorce limestone utilization or extent of reaction,
since these tend to go hand-in-hand.
The effects of high magnesium and sulfate concentrations in these runs
are primarily on reaction rates. Since there was little reaction, the
solids produced were primarily limestone and therefore exhibited reason-
ably good settling rates. However, in all cases of high sulfate or
VI-46
-------
ID"0
10"
II 00
O
CO
(0
O
to
O
to
<3
6
1
a.
•
I
s
a
10-
10-'
O
O
1.0
2.0
O
O
Legend:
O 90 Minute Reactor Holdup
150 Minute Reactor Holdup and Solids
Recycle
rj High Magnesium
3.0 4.0
, Ionic Strength
5.0
6.0
FIGURE VI-32 RELATION OF OBSERVED APPARENT SOLUBILITY
PRODUCTS TO SATURATION VALUES FOR CaSOg
VI-47
-------
1,000
O Reactor 1
Reactor 2
0 Reactor 3
• Reactor 4
• Reactor 5
Reactor 6
20 30
Time, Minutes
40
50
FIGURE VI-33 SETTLING CURVES FOR RUN 125 - BASE RUN
VI-48
-------
1,000
Reactor 2
ReactorS
Reactor 4
Reactor 5
Reactor 6
Reactor 6
Reactor 5
Reactor 4
Reactor 3
Reactor 2
10
20 30
Time, Minutes
40
50
FIGURE VI-34
SETTLING CURVES FOR RUN 120
VI-49
-------
1,000
Settling Curves for all
Reactors in this Range
Settled Volumes After 15 min.
Reactor No.
1
2
3
4
5
6
Vol (ml)
~40
50*
~50*
~50*
~60*
~100
* Appreciable C02 Offgassing.
I
10
20 30
Time, Minutes
40
50
FIGURE VI-35 SETTLING CURVES FOR RUN 121
VI-50
-------
1,000
800
c
o
o
O_
(/>
I
'c
-------
1,000
800
c
o
C
CD
600
400
200
I
10
O Reactor 1
O Reactor 2
[D Reactor 3
• Reactor 4
• Reactor 5
• Reactor 6
A Reactor 7
I
20 30
Time, Minutes
40
50
FIGURE VI-37 SETTLING CURVES FOR RUN 124
VI-52
-------
1,000 k-
800
o
1
'c
0)
600 r~
400
200
Reactor 1
Q Reactor 2
O Reactor 3
• Reactor 4
• Reactor 5
Reactor 6
10
20
30
Time, Minutes
40
FIGURE VI-38 SETTLING CURVES FOR RUN 126
VI-5 3
-------
1,000
800
o
c
0)
600
400
O Reactor 1
O Reactor 2
EJ Reactor 3
• Reactor 4
+ Reactor 5
• Reactor 6
200
Time, Hr
17
FIGURE VI-39 SETTLING CURVES FOR RUN 127
18
VI-54
-------
1,000
800 —
c
o
•35 600 —
u
V
400 —
200 —
O Reactor 1
Q Reactor 2
O Reactor 3
• Reactor 4
• Reactor 5
Reactor 6
10
20 30
Time, Minutes
FIGURE VI-40 SETTLING CURVES FOR RUN 128
VI-55
-------
magnesium (runs 121, 123 and 124) the solids settling rates were notice-
ably slower than in runs 125 and 126, where limestone utilizations were
significantly higher, and were only comparable to that in run 120 (high
limestone feed) because the high solids content of run 120 resulted in
hindered settling.
In all runs except 120, 127, and 128 there was appreciable off-gassing
of C02 in all of the reactors. This off-gassing was also apparent in
the settling test performed on the reactor effluent slurries. For the
most part, the degree of off-gassing (or bubbling) decreased with in-
creased bisulfite conversion. Thus, in runs 121, 123 and 124 there
tended to be more bubbling in the last few reactors than in runs 125
and 126. This bubbling suggests that better than 75% utilization of
limestone will be required in the reactor system to avoid poor clarity
in the thickener overflow. Otherwise, special provisions will have to
be made to circumvent solids carryover to the scrubber system.
Magnesium Dissolution
Since soluble magnesium has an important deleterious effect on reaction
rate, magnesium dissolution from the limestone and the projected steady-
state soluble magnesium concentration is of concern. The soluble mag-
nesium concentration may depend upon a number of factors including the
amount of magnesium in the limestone, its rate of dissolution, the purge
rate of soluble magnesium in the filter cake, and the solubility of
magnesium sulfite.
The Fredonia limestone used in these runs contained 0.14-0.19 mols of
magnesium per gram of dry limestone. If all of this could dissolve,
then the concentration of magnesium attained in the liquor would be
determined by the solubility of magnesium sulfite and could exceed
3,000 ppm. However, the amount of magnesium dissolved varied with the
magnesium concentration in the liquor and solids holdup. In all runs
at less than 500 ppm magnesium and 90 minutes total holdup, 30% to 50%
of the magnesium in the limestone dissolved, resulting in increases of
20 to 40 ppm of magnesium in the liquor. At the same inlet liquor con-
centrations but with solids recycle or a 180-minute reactor holdup (runs
127 and 128) about 75% of the soluble magnesium dissolved, resulting in
increases in magnesium in the solution of about 60 ppm across the reactor
system. And at 2,250 ppm magnesium with 90 minutes holdup there was no
appreciable dissolution of magnesium (run 121), although there was also
very little reaction of the limestone, which may have limited dissolution
of magnesium.
These data suggest that with Fredonia limestone operating at 0.3-0.4
M TOS and at a temperature of 50JI 5°C, the soluble magnesium concen-
tration will not exceed about 2,000 ppm in steady-state operation.
However, at these concentrations, utilization is seriously reduced.
One possible measure that can be taken to decrease magnesium levels, as
discussed previously, is to treat the limestone reaction effluent with
lime to raise the pH above 10. This will precipitate magnesium as
VI-56
-------
Mg(OH)2 and prevent the buildup of soluble magnesium. The merits of such
an approach depend upon the amount of lime required and the associated
system costs. As a test of post-liming in the open-loop runs, product
liquor was reacted with lime to a pH slightly above 11 in a 30 minute
CSTR (run 124). The solids produced had acceptable settling properties
and the magnesium level in the solution was decreased essentially to zero.
However, the pH was difficult to control and in a large-scale system care
would have to be taken to fine-tune lime feed rates, or an intermittent
lime feed system would have to be used.
Calcium Sulfate Precipitation
The precipitation of calcium sulfate in the concentrated limestone mode
is even more important than in the concentrated lime mode. The limestone
utilization is much more sensitive to high sulfate concentration than is
lime utilization, and limiting sulfate concentration, thereby limiting
calcium sulfate precipitation rates, can severely restrict the applica-
bility of a limestone dual alkali system.
In Figure VI-41 the (CaS04/CaS03) ratio in the product solids is plotted
as a function of the ratio of ([SO^]/[S03]) in the reactor system effluent
liquor. Included in this plot are a number of points from the closed-loop
runs. Over a reasonable operating range of soluble sulfate (up to sulfate-
to-sulfite ratio of about 10:1) the typical calcium sulfate precipitation
can be approximated by the following equation:
CaS04 \ / [SO"]
I = 0.022 I —-
CaS03 I V [SOg]
/reactor solids reactor liquor
This rate of calcium sulfate precipitation is only about 60% of that in the
concentrated lime mode as shown in Figure VI-41 and by comparison of equa-
tion (3 ) with equation ( 1 ) . The lower calcium sulfate precipitation may
be due to the slower reaction rate over the range of TOS studied ([TOS] =
0.3-0.4M) than in the lime regeneration. This slower rate may allow a closer
approach to equilibrium. (In this regard, it is recalled that in the concen-
trated lime mode there was a slight redissolution of calcium sulfate with
long solids-liquid holdup times following the reactor system.)
The implication of this lower level calcium sulfate precipitation is clearly
that limestone dual alkali systems operating in this range of TOS will be
limited to applications where low oxidation rates are expected, unless means
are found to increase limestone reaction rates (which may also increase sul-
fate precipitation).
2. Closed-Loop Runs
Four closed-loop runs in the concentrated limestone mode were made with
active sodium levels ranging from 0.27M to 0.45M and inlet S02 levels
ranging from 2,100 ppm to 2,800 ppm. All of these runs involved multi-
stage reactor systems consisting of anywhere from two to five CSTR s in
VI-57
-------
0.30
I
00
Sulfate Precipitation Correlation For
Concentrated Lime Mode (See Figure IV-7)
Conditions:
Feed [TOS] = 0.3-0.4 M
Feed [Mg++] = 300-2,500 ppm
Temperature = 45—50°F
Reactor Holdup = 50 min — 2 hrs
Key:
O Open-Loop, Pilot Plant
• Closed-Loop, Pilot Plant
D Laboratory
!
I
-L
10
15
20
[SO4] /[S031 in Reaction Effluent Liquor
FIGURE VI-41 SULFATE PRECIPITATION IN THE CONCENTRATED LIMESTONE MODE
-------
series. Except for the reactor system, the equipment configuration and
utilization were identical to that of concentrated lime mode operations.
Figure VI-42 gives a schematic of the system configuration and Table VI-9
summarizes the general operating conditions and key performance parameters
for the four runs.
While overall limestone utilization and S02 removal were reasonably high
in all runs, none of the runs can be considered successful. In every
case the solids produced exhibited poor settling properties. After
about one day of operation in each run, solids began overflowing the
thickener and were carried forward to the scrubber system. The quantity
of solids in the overflow increased continuously with time to the point
that the system became a combination of direct limestone/dual alkali
process. The presence of high levels of suspended solids in the scrubber
system undoubtedly accounts for the very high limestone utilization
factors, as evidenced by the analyses of the solids in the scrubber
bleed slurry. Not only does some unutilized limestone react in the
scrubber system, but the carryover of solids into the scrubber results
in an overall solids recycle throughout the system, allowing higher
solids residence time and longer solids/liquid contact.
Figures VI-43 and VI-44 show process operating diagrams for runs 411 and
430, two runs in which there were sufficiently stable periods when the
system operation and performance could be evaluated quantitatively. It
must be stressed that the data in these figures represent "snapshots"
of the process operation rather than steady-state conditions. In all
runs the overflow of solids from the thickener increased with time,
eventually necessitating shutdown of the system.
The causes for the poor settling solids and carryover in the thickener
overflow differed according to operating conditions. In general,
though, there were four factors of obvious significance:
• High soluble iron concentrations — In run 430 high soluble iron
concentrations (up to 400 ppm) apparently from dissolution of
the iron oxide coating in the thickener and feed forward surge
tank initially retarded reaction rates to the extent that it was
not possible to regenerate to a pH greater than 5.8 even with four
reactors; and limestone utilizations were less than 35%. This
resulted in excessive bubbling in the thickener (C02 off-gassing)
due to continuing reaction.
The iron was virtually eliminated by switching from limestone
to lime. After three days of operation with lime, the solids
properties noticeably improved. The solids settling rate
increased; the level of suspended solids in the thickener
overflow decreased to less than 500 ppm; and the filtration
rate increased from a few pounds of cake per hour to over 200
pounds per hour. The solids content of the filter cake also
rose to about 45%.
VI-59
-------
Gas Scrubber Out
i
o>
o
Reactor System
(See Table VI-9)
\
.— -'
1
1
1
Na2S04
or
Na2CO3
or
NaOH
H2O
Waste Solid
Filtrate Receiver
FIGURE VI—42 PROCESS FLOW DIAGRAM FOR CONCENTRATED LIMESTONE
MODE PILOT PLANT OPERATIONS CONTINUOUS CLOSED-LOOP
-------
TABLE VI-9
<
M
I
SUMMARY OF
Run No.
Run Duration, (days)
Operating Conditions
Inlet Gas:
S02, (ppm)
02, (vol. %)
Temperature, (°F)
Scrubber Bleed:
PH
[TOS], (M)
[804], (M)
[TSS], (%)
Reactor System:
No. Reactors x Holdup, (mins.)
Solids Recycle to Reactor
Chemical Makeup:
Sodium
Limestone
Key Results
S02 Removal, (% of inlet)
Sodium Feed Rate, mols Na •HV(mol (AS02 + Na2S04))
Calcium Feed Rate, mols CaC03/(mol (AS02 + Na2S04))
(% to neutralize HSO^)
Calcium Utilization, (% of CaC03)
Soluble Calcium in Reactor Effluent, (ppm)
Scrubber System Oxidation, (ppm of S02)
Filter Cake:
Wt % Insoluble Solids, (avg.)
Wt % Soluble Solids, (avg. -dry basis)
No. Displacement Washes, (typical)
mols CaSO^/mol CaSOx, (avg.)
CLOSED-LOOP RUNS USING
410
2.5
2,450
4
370-415
5.5-6.4
0.4
0.85
<1
1 x 7 + 1 x 70
No
Na2S04
Fredonia
86
0.10
1.1
105
60
110
T-200
33
13
0.5-1
0.13
MULTISTAGE REACTOR SYSTEMS
411
2.5
2,800
5
370-415
5.6-5.9
0.37-0.4
0.8-0.85
up to 3,5a
1 x 10 + 1 x (60-80) 3
Intermittent
Na2S04
Fredonia
90-95
0.07
0.95-1.05
125-1453
80-90
60-140
_b
35
11
0.09
412
3
2,100
4.5-5.5
360-375
5.0-6.0
0.25
0.7
<2
x 15 + 1 x 30
No
Na2C03
Fredonia
85
0.03
1.0-1.15
135
92
140
_b
40-45
1.5
"-3
0.09
430
6.5
2,100
4-5
355-390
5.3-6.3
0.3C
0.6-0.7C
2-3
1 x 6 + (2 or 3) x 17
No .
NaOH
Fredonia
90
0.10
0.95-1.0
75
95
_b
30
2.1
1.5-2
0.11
a During Run 411 there was significant carryover of solids in the thickener overflow due to "the poor settling properties of the solids. The amount
of solids increased during the run. When the system was shut down, the venturi bleed contained 3.5% suspended solids. This carryover of solids
undoubtedly accounts for the high utilizations achieved since unused limestone reacted in the scrubber system. This also accounts for the artifi-
cially high CaC03/(HS03) feed ratio to the reactor, since some of the S02 precipitated in the scrubber (reducing the liquor feed forward rate).
b Oxidation rates could not be estimated due to the presence of a variable amount of solids in the scrubber system.
c Average concentrations maintained by a number of systems repriuies and Charging of the Na2S03 during the run.
-------
•^^ 230 ppm S02
• H20
2,800 ppm S02 Venturi
^^^^W +
i^^^ 2 Irays
5%02
AS02 = 92%
pn - o.o
"* [TOSJ = 0.23IV
[SOT] = 0.84M
TSS =15 gms/l
3.5—4 gpm
pH = 5.7
[TOS] =0.41M
[SOT] = 0.82M
TSS — 25 gms/l
1 1
Dry Limestone:
ao/ci L.dOU3 ^"
2% MgC03
R1
10 min
mols CaC03/mol ASO2
-0.95-1.05
Liquor Comp. Scrubber R1
pH 5.7 6.1
[TOS] , M 0.409 0.341
[SO^LM 0.819 0.817
TSS, gms/l 40 61
Iron, ppm — -
[Mg+ ], ppm ~ 125 —
Solids Comp:
CaSOx/Ca ~ 1.0 0.72
CaS04/CaSOv 0.14 0.12
I
i
Hold
T
ank
Ma
iNa2
S04 -
mnlc Ma ^O /mrtl AQO 'Sj n f\"7
20
r~
\ >
\. .
7
R2
60-80 min
R2
6
0.
0.
70
—
—
0
~ 0.
5
285
804
80
1
y— i
i ,
f
f
Thickener
i
ii
Wash Water
No. Displacement
Washes ** 3
t
Typical Cake Composition
Insoluble Solids -40%
Soluble Solids = 1.5% (dry basis)
FIGURE VI-43 SYSTEM OPERATION IN RUN 411 (AFTER 36 HRS)
VI-62
-------
215 ppm SO2
H2O
<
M
I
U)
| H20
2, 100 ppm SO2
•
Scrubber
Venturi
•••MH^.1 i -i-™,,.
4-5% 02
AS02 = 90%
pH =
[TO
[SO
TSS
Dry Limestone: i '
93% CaCU3
2% MgCO3
R1
5—7 min
R
S]
pH=6.6
k
I luaj - u. i /M I i I
[504] = 0.70M | ^ J,
~6gpm
.6
= 0.265M
H = 0.69M
\ CaSO /Ca = 0.964
^25 gms/l } CaE
(
R2
15-20 min
O4/CaSO -
' 0.075
R3
15-20 min
rm*
r~
\
\
NaOH
rnnlc Ma "*^/mnl A^f
~ 0 10
1 II
Thickener
mols CaCog/mol A SO2
T
Wash Water
Filter -«
No. Displacement
Washes = 1.5-2.0
= 0.95-1.0
Liquor Comp.
PH
[TOS], M
[SO'], M
TSS, gms/l
Iron, ppm
[Mg ], ppm -
Solids Comp:
CaSOx/Tot. Ca
CaSO4/CaSOx
Scrubber
R1
R2
R3
5.6
0.265
0.69
25
8
244
6.1
0.234
0.705
39
—
„—
6.4
0.187
0.71
42
-
—
6.5
0.184
0.70
42
3
262
0.964
0.075
0.92
0.0635
0.96
0.084
Cake Composition
Insoluble Solids = 30%
Solubles = 2.1% (dry basis)
FIGURE VI-44 SYSTEM OPERATION IN RUN 430 (AT TERMINATION OF RUN)
-------
At this point the lime feed was stopped and the limestone feed
resumed. Solids properties again deteriorated. Solids began
overflowing the thickener, and the filter cake solids content
fell to about 30%. However, the reactor system effluent pH
was maintained above 6.1 and soluble iron rose only to a maxi-
mum of 8 ppm. Throughout run 430 sodium hydroxide was used
for sodium makeup rather than sodium sulfate or soda ash in
the hope that the additional alkalinity would minimize soluble
iron.
The progression of solids quality prior to, during, and after
the lime feed is shown in Figures VI-45 through VI-50. During
lime feed, agglomerates of needles were produced which settled
reasonably well; however, before and shortly after liming, the
needles did not agglomerate, and settled poorly.
High sulfate-to-active sodium ratios — In all the closed-loop
runs there was a relatively high sulfate-to-active sodium ratio.
This was not by design. The sulfate levels in all runs tended
to increase due to a combination of factors: makeup with sodium
sulfate (runs 410 and 411); low calcium sulfate precipitation
rates compared with oxidation; and, apparently, slightly higher
oxidation rates in the system due to the additional reactors
and pumps.
The combination of the relatively high sulfate levels and the
soluble magnesium (which ranged from 100 ppm to 300 ppm) may
have been sufficient, along with a small amount of soluble
iron, to affect solids properties. In none of the runs,
though, did magnesium or sulfate approach steady-state levels
due to the occasional attempts to improve solids properties
by repriming the venturi recycle tank and reactors. What is
the effect?
Inappropriate reactor design — In runs 410 and 411 the
reactor system consisted of only two tanks in series, a
reactor with a short holdup (5-10 minutes) followed by one
with a long holdup (60-80 minutes). As shown by laboratory
testing conducted after these runs, a long holdup CSTR does
not produce good settling solids. In fact, as shown in the
laboratory tests and the open-loop pilot plant runs, the
solids properties deteriorate with increasing reaction time
(batch or CSTR) either because of increased solids holdup
or increased limestone conversion.
Recirculation of poor solids — The onset of recirculation of
solids from the thickener overflow through the scrubber and
back to the reactor system marked the beginning of deterioration
of settling properties in the reactor system. Poor settling
solids were already present, though in the thickener, apparently
because it is like a poorly stirred CSTR with a 4-6 hour holdup
VI-64
-------
FIGURE V-45 CRYSTALS IN REACTOR 4, 28 MRS INTO RUN USING
LIMESTONE (RUN 430)
VI-65
-------
FIGURE V-46 CRYSTALS IN REACTOR 4, 45 MRS INTO RUN USING
LIMESTONE (RUN 430)
VI-66
-------
if-
*»•
FIGURE V-47 CRYSTALS IN REACTOR 3, AFTER 48 MRS OF
RUNNING WITH LIME (RUN 430)
VI-67
-------
FIGURE V-48 CRYSTALS IN REACTOR 3, AFTER 57 MRS OF RUNNING
WITH LIME (RUN 430)
VI-68
-------
FIGURE V-49 CRYSTALS IN REACTOR 3, 8 MRS AFTER RESUMING
LIMESTONE FEED (RUN 430}
VI-69
-------
FIGURE V-50 CRYSTALS IN REACTOR 3, 30 MRS AFTER RESUMING
LIMESTONE FEED
VI-70
-------
in which a large fraction of the reaction was occurring. There
was also precipitation of some calcium-sulfur salts in the
scrubber under non-optimal conditions and possibly some re-
dissolution at low pH's, both of which contributed to the poor
quality of the solids feeding into the reactor system.
Figure VI-51 shows the difference in the settling properties of
solids from reactors 1 and 3 just prior to and just after an
appreciable quantity of solids began overflowing the thickener.
This phenomenon occurred in all runs and thwarted attempts to
reestablish good settling solids by draining and repriming the
venturi recycle tank and reactors (but not the thickener).
In addition to adjusting reactor configuration and holdup,
repriming tanks and using intermittent lime feed, four other
methods were tried in an effort to improve solids settling
properties:
• addition of fly ash to the first reactor (to speed CC>2
off-gassing);
• recycle of thickener underflow to the reactor system
(to grow better crystals);
• use of flocculants (to increase settling rates); and
• adjusting operating pH and the limestone feed rate.
None of these proved successful. The flocculant did increase
the settling rate, but the final settled volume did not change.
Based upon these results, additional work is required to develop
an operable concentrated limestone mode.
D. CONCLUSIONS
No viable approach was found for use of limestone in a concentrated dual
alkali mode. Through the laboratory and pilot plant efforts allocated to
work on the concentrated limestone mode, we were not able to develop process
parameters and reactor conditions consistent with good limestone utilization
and generation of acceptable quality waste solids. The work did, however,
uncover important factors influencing the limestone regeneration reaction
that indicated promising areas of future work. Unlike results from work on
use of limestone in dilute modes (see Chapter VII), the potential for tech-
nical success argues for additional work on the concentrated limestone dual
alkali mode; especially when the economic incentives (presented in the Intro-
duction, Chapter II) are considered.
Limestone is substantially less reactive toward sodium salt solutions than
is lime, even when reacting with relatively acidic scrubber bleed solutions.
The reaction rate of the limestone regeneration reaction is dependent upon:
VI-71
-------
o
o
Q-
c
til
1,000
800
600
400
200
0
• —e-- Reactor 1
—-•— Reactor 3
-i—i-f-i-9
10 20 30 40
Time, Minutes
50
60
Settling Curves 1 Hr.
Before Noticeable Solids
In Overflow
1,000
-§ 800
o
3
U
w
'c
-------
• nature of the limestone and its particle size distribution;
• reactor temperature and residence time;
• concentrations of soluble reactants (sodium sulfites,
sodium bisulfites and sodium sulfates); and
• the presence, at low concentrations, of trace constituents,
such as magnesium and iron, which influence the reaction rate.
Increase in the reaction rate was generally consistent with improvement
in the dewatering properties of the solids produced and with improved
utilization of limestone.
Three limestones, with similar particle size distribution, were examined —
Fredonia limestone used in the EPA/TVA Shawnee program; another, locally
available, natural limestone; and reagent grade CaCOa. Of these, the
Fredonia limestone was amorphous, rather than crystalline in nature, and
was considerably more reactive than the other two limestones examined.
The Fredonia limestone, therefore, was used extensively in the labora-
tory and pilot plant programs.
Laboratory experiments indicated that increasing temperature importantly
increased the reaction rate. However, the pilot plant was not equipped
for heating the reactors or for heating the reactor feed. As a conse-
quence pilot plant regeneration was performed at a maximum of about 50°C.
The dewatering properties of solids were generally observed to deteriorate
as the regeneration reactor residence time was increased. Increasing the
reactor residence time results in carrying out the reaction closer to the
equilibrium conditions and consequently at a lower driving force and
reaction rate. Use of multistage reactor systems, containing several
stages with residence times in the range of 15 minutes, were found to
produce solids with a quality superior to that of solids produced in
fewer reactors with the same total residence time. Recycle of solids,
increasing reactor solids concentrations from about 2 wt % to 5 wt %,
improved limestone utilization but did not appreciably improve the
quality of the solids.
Under controlled conditions, with a multistage reactor system operating
at about 50°C, it was possible to produce solids with acceptable de-
watering properties (45% insoluble solids) and to achieve limestone
utilizations on the order of 75%. However, as the sulfate concentra-
tion in the loop rose above 0.7M, or the magnesium concentration rose
much above 300 ppm, the reaction rate and the resulting limestone
utilization and solids properties all deteriorated.
Sulfate concentration in the reactor liquor had a much more important,
deleterious effect on the reaction rate and solids properties in lime-
stone regeneration reactions than the similar effects of increased sulfate
concentration observed in concentrated lime regeneration. As in lime
VI-73
-------
regeneration, the reaction rate is inversely proportional to the ratio
of sulfate/sulfite concentrations in the liquor, but the rate drops
dramatically using limestone as the sulfate concentration exceeds 0.7M
at TOS levels of 0.3-0.5M. Operation at lower sulfate/sulfite ratios
tends to limit sulfate precipitation in this mode and limit the range
of oxidation in which limestone regeneration could be operated closed
loop.
Calcium sulfate coprecipitates along with calcium sulfite in concentrated
limestone regeneration reactions in an analogous fashion to the coprecip-
itation of calcium sulfate observed in the concentrated lime regeneration.
However, pilot plant data indicate that for the same sulfate/sulfite con-
centrations with the same range of TOS in the feed liquor (i.e., [TOS] =
0.3-0.5M) lower sulfate precipitation occurs when using limestone,
as given by the following:
[S0=]
= 0.022
.
C*S03 [so"]
reactor solids \ 3 reactor iiquor
The sensitivity of the reaction to high sulfate concentrations and the
lower sulfate precipitation rates make limestone regeneration less viable
for closed- loop operation than lime regeneration at higher oxidation rates.
The presence of Mg"*"1" in solution, introduced into the system in varying
amounts depending upon the magnesium content of the limestone, also can
have a retarding effect on the limestone regeneration reaction rate, re-
sulting in poor solids quality and limestone utilization. This effect
becomes pronounced as the Mg""" concentration rises much above a few
hundred ppm. Relatively low magnesium limestones, such as Fredonia
limestone (1.0-1.5 wt % Mg as MgCOa), would result in concentrations
on the order of a few thousand ppm, at steady-state, in a concentrated
dual alkali loop.
Laboratory work confirmed that magnesium concentrations could be con-
trolled by reacting part of the process stream with lime to precipitate
Mg(OH>2. However, such an approach would reduce operating cost savings,
requiring part of the total regeneration to be performed using lime. Use
of lime with limestone would increase the complexity and the capital cost
in a manner similar to that discussed in Chapter VII for dilute limestone/
lime dual alkali systems, eliminating economic incentive.
In pilot plant operations, iron from corrosion of unlined steel equipment
was found to have an effect similar to that of magnesium on the limestone
regeneration reaction at pH's below about 6. At higher pH's, Fe(OH)s is
highly insoluble, limiting the buildup of iron in solution. By selection
of proper materials of construction and linings and by carrying the lime-
stone regeneration beyond a pH of 6, interference by iron can be eliminated
in concentrated limestone modes.
VI-74
-------
Future work on limestone regeneration should be directed at increasing
reaction rates at high magnesium levels by increasing sulfite concentra-
tions, reactor temperature and by staging of the reactors.
VI-75
-------
VII. LIMESTONE/LIME DILUTE MODES
As explained in the Introduction, operating with dilute sulfite/bisulfite
(TOS) solutions represents one possibility for increasing the proportion
of sulfate in the cakes produced during regeneration. Against this pos-
sible benefit must be balanced the cost penalties (represented by larger
equipment and higher pumping rates) caused by operating with more dilute
TOS solutions.
A. EPA LABORATORY RESULTS
In its Research Triangle Park laboratories, EPA had already performed
some preliminary scouting runs on the regeneration of dilute TOS solu-
tions with limestone and lime. These were summarized by Draemel,11 wno
showed that the CaS03 exhibited a pronounced tendency to supersaturate;
i.e., the reactions:
CaC03 + 2NaHS03 •* CaS03(s) +Na2S03 + C02*+ H20 (9)
Ca(OH)2 + 2NaHS03 -> CaS03 (s) +Na?S03 + 2H20 (8)
did not proceed rapidly to completion in dilute TOS solutions. The pre-
liminary runs indicated that this tendency could be counteracted by _ re-
circulating the product solids through the reactor, and that maintaining
a level of 5% solids in the reactor might permit reasonable utilization
(80-90% with lime, 50-70% with limestone), with one-hour holdup times.
EPA extended this work by performing a series of continuous limestone
runs in a 2-liter reactor followed by a 5-liter clarifier, with slurry
recycle to the reactor. These were long-term runs (36-50 hours in length)
at about 50°C. Satisfactory material balances were calculated for all,
using an ADL computer program adapted for the purpose; this Program
performs balances around designated sections of the circuit for total
weight, total volume, and the weight of water, sulfur, sulfate, sodium,
and calcium.
.
feed solution, with a one-hour residence time in the reactor
S
solids concentrations shown in the table.
THe total -Hate extent of the 11 »asn
fo vaa "tiiition One *«»£*=.«.- can b.
noted by observing the cake sulfate content in runs 21, 24, 25,
VII-1
-------
Table VII-1 EPA 2-LITER REACTOR CONTINUOUS FREDONIA LIMESTONE RUNS—SUMMARY
Feed Solution
Composition, M
Run [TOS] [SO^t]
21
22
23
24
25
26
27
28
29
30
0.14
0.14
0.14
0.13
0.13
0.04
0.12
0.07
0.05
0.04
0.25
0.52
0.41
0.26
0.26
0.25
0.27
0.27
0.53
0.73
Percent Solids
in Reactor
2.1
2.2
2.9
4.6
6.0
2.7
3.5
3.3
2.0
2.0
Calcium Feed
S t oi chiome t ry
1.13
1.10
1.19
1.02
1.20
0.88
1.25
0.78
0.72
0.84
Calcium i
Utilization '
TOS Basis, a %
90
64
73
116
106
133
78
110
55
52
/ CaSOij \
\ CaSOx / n.,
x x 'solids
10.1%
8.5
6.0
6.1
0.2
6.3
4.5
11.1
22.2
8.2
a. 100 x mols TOS Removed/mol Calcium Added
-------
all conducted with essentially the same feeds and the same nominal stoi-
chiometry, but differing in percent solids level in the reactor. There
appears to be a downward trend in cake sulfate level as percent solids
level (and thus effective solids residence time) in the reactor is in-
creased.
On the other hand, lowering the TOS and raising the sulfate levels in
solution did render a lime post-treatment more effective in precipitating
sulfate from solution. In runs 26 through 30, samples of the thickener
overflow were agitated with excess lime for one hour, and the sulfate
disappearance from solution was monitored. Table VII-2 shows the results
for these runs, in-terms of the ability of the system to reject sulfate
as a fraction of total sulfur species precipitated, both for the lime
step only and for the combination of the limestone and lime step (assum-
ing all limestone step overflow to go on to lime treatment) . The lime
treatment solids in runs 29 and 30 were analyzed and, as the table shows,
the check with sulfate precipitation from solution is quite good.
The calcium and sulfate levels in the limestone reactor overflow and
thickener underflow solutions of runs 26 through 30 were multiplied for
comparison with the apparent molal solubility product for CaSO^ • 2H20
(predicted by the method of Kusik and Meissner7) for the ionic strength
involved. Figure VII-1 shows that in all runs (runs 21 through 30) levels
were not high enough to precipitate CaSO^ • 2H20 as such; however, whenever
crystals of CaSO^. • 21^0 were shaken with samples to test for calcium
sulfate saturation, the product of the calcium and sulfate levels in the
resulting solution rose to the predicted curve almost exactly (the squares
on Figure VII-1). Figure VII-2 shows similar calculations for the calcium
times the sulfite, compared with the analogous curve for CaS03 • 1/2H20.
In all runs, CaS03 supersaturation is indicated.
B. DILUTE MODE ALTERNATIVES
The results of this work have been utilized to help us project possible
flow sheets for dilute operation at 35% sulfate precipitation rates. We
have compared
• using limestone, with lime for added sulfate precipitation, and
• using lime, with sodium carbonate for added Ca softening.
The first of these has been called the "double loop-sulfite softening"
mode because the calcium level leaving the lime reactor/clarifier is
lowered by recycling the solution to the limestone reactor, in which
the precipitation of CaS03 lowers the Ca level.
Figure VII-3 shows how one would operate a dilute limestone/lime system
to precipitate sulfate to sustain a 35% oxidation level, based on the
results of EPA run 30. About one-quarter volume of regenerated liquor
would have to be sent to a lime loop (for sulfate precipitation) for
VII-3
-------
Table VII-2 EPA RUNS - POTENTIAL OF LIME POST-TREATMENT STEP
Concentrations
in Feed Liquor, M
LIME STEP ONLY
Run 26
27
28
<
'-| 29
M ^*
1
*- 30
LIMESTONE PLUS
Run 26
27
28
29
30
[SO^]
0.2721
0.2845
0.2511
0.5349
0.7389
LIME OVERALL
0.2420a
0.2660
0.2626
0.5208
0.7183
[TOS]
0.0184
0.0648
0.0418
0.0326
0.0313
0.03903
0.1183
0.0712
0.0443
0.0401
Concentrations in
Treated Liquor. M
[S01J]
0.2925
0.2886
0.2523
0.5125
0.6999
0.2925
0.2886
0.2523
0.5125
0.6999
[TOS]
0.0090
0.0180
0.0214
0.0100
0.0106
0.0090
0.0180
0.0214
0.0100
0.0106
ASOi,
A(SO~ + TOS)
(0)
(0)
(0)
49.8%
65.3
(0)
(0)
17.1%
19.5
38.4
\
Jliquor
CaSOx/ ...
x/solids
41.9%
64.5
a. These (and following) adjusted for dilution of feed with limestone slurry.
-------
"o*
E
3
(A
a
•a
I
c
-------
A Reactor Overflow, 1 Hour
O Thickener Underflow, After
2 Hours Additional
-7 I I I I I I I I I I I I III
III
10
34
, Ionic Strength
FIGURE VII-2 EPA LIMESTONE RUNS 21-30 EXTENT OF Ca SO, SUPERSATURATION
VII-6
-------
H20 Evaporation H^O Make-Up
- 1.00 g moi
131
501
.730JV1 SO.,
.040 TOS
.011 Ca
Limestone
Reactor
Thickener 1
Filter 1
131
.700 jw so.,
010 TOS
.1 OH
.016 Ca
J"\-
63!
.723JM SO.,
.027 TOS
.011 Ca
Solids
05
-------
every volume sent to the scrubber. At these conditions, the material
returned to the scrubber would be undersaturated (with respect to
CaS04 • 2H20) by about 150 ppm Ca. The expected utilizations are only
50% for limestone and 80% for lime, based on experimental results. If
one tries to keep up with higher levels of oxidation by sending higher
fractions through the lime loop, the calcium undersaturation ("safety
factor") decreases as the material leaving the lime reactor, which is
saturated in CaSO^ • 2H20, begins to overpower the sulfite softening
occurring in the limestone reactor; indeed, it does not appear possible
to keep up with more than 40% oxidation without having to accept a re-
generated scrubber liquor with a calcium level approaching within less
than 100 ppm Ca of saturation with respect to gypsum.
Figure VII-4 shows how one might operate an alternative flow sheet using
lime only, relying on sodium makeup with Na2C03 for the necessary Ca
softening. Again a 35% oxidation level has been taken as a basis. The
expected Ca(OH)2 utilization is about 80%. The amount of sodium in the
Na2C03 needed for softening to a level 100 ppm below saturation is seen
from Figure VII-4 to be less than 1% of the weight of the dry waste
solids. Since soluble Na losses in well washed cakes have been in the
range of about 1%, the normal makeup of Na should introduce more than
enough carbonate for softening.
The mode of operation in Figure VII-4 has a much smaller scrubber liquor
feed; while the mode of Figure VII-3 requires 50 1/g mol SC>2 absorbed,
that of Figure VII-4 requires 9.1 1/g mol S02 absorbed. The chief un-
certainties in the mode of Figure VII-4 are:
1) amount of solids recycle required to produce good settling
and to eliminate CaSO^ supersaturation in the lime reactor,
2) ability to produce good-settling carbonate solids in the
carbonate treatment thickener, and
3) ability to bypass some liquor around the carbonate treatment
step to save on the cost of the thickener (see dashed lines
in Figure VII-4).
Based upon the schematic process flow sheets developed from the labora-
tory data, rough economic analyses were made comparing the two dilute
systems: limestone/lime with sulfite softening and lime with soda ash
softening. The limestone/lime system involves higher capital costs
because of the higher flow rates through the absorption/regeneration
system, but it has lower raw material costs since half of the regenera-
tion is carried out using limestone, as opposed to complete regeneration
using the more expensive lime in the alternate dilute system. It was
assumed that both systems would have the same effective cake washing
efficiency and require approximately the same sodium makeup. Because
of lower yields on raw materials, the limestone/lime system generates
more waste for disposal.
VII-8
-------
H20 Evaporation HjO Make-Up
Na, CO., .027 g mol
.700MS04
.01 TOS
.1 OH
.016 Ca
Solids
.355 g mol SO.,
.645 TOS
.027 CO,
.455 OH
1.25 Ca
Basis: 1 9 mt>l of S02 absorbed
35% Oxidation Rate (all assumed in scrubber tor convenience)
FIGURE VII-4 DILUTE LIME/CARBONATE SOFTENING SCHEMATIC
VII-9
-------
On the basis of the following analysis, it was found that the raw material
cost savings for the limestone/lime system were not sufficient to offset
the expected capital cost increases and increases in waste disposal costs.
Two cases were examined^or the economic analysis:
• a 3.5% S coal application requiring 90% S02 removal, and
• a 1.0% S coal application requiring 80% S02 removal.
In both cases a relatively large cost difference was assumed between lime
and limestone:
• $40/ton, CaO (80% utilization)
• $5/ton, CaC03 (ground, 50% utilization)
This large cost difference in materials favors the more capital-intensive
sulfite softening approach. A waste disposal cost of
• $10/ton waste (dry basis)
was used in all cases. A total annual fixed cost of 20% of the capital
investment (including maintenance costs) was used to calculate annual costs.
Using this basis, we calculated the minimum capital investment difference
at which the annual capital cost savings of the simpler lime approach
equaled the annual operating cost savings of the limestone/lime approach.
For the high-sulfur coal case, a capital investment savings of about $7/kw
or more would offset operating cost savings. In the low-sulfur coal situa-
tion, the corresponding figure dropped to only $1.8/kw. In each case, we
would expect the actual capital investments for the limestone/lime system
to easily exceed those of the lime system by more than these calculated
values. As a result, lime regeneration with soda ash softening was se-
lected as the basic approach to be investigated for dilute modes in future
pilot operations under this program.
C. ADL LABORATORY STUDIES
The results obtained during the EPA laboratory research program showed
that if a lime or soda ash softening step were included, limestone could
be used to perform some of the regeneration in dual alkali FGD systems
operating at dilute active sodium concentrations. However, because it
was estimated that the capital costs associated with a dual reagent (lime-
stone/lime) system would be substantially higher as compared to a system
using only lime, the remaining studies of dilute mode systems at ADL were
focused on the latter.
The several unit operations which would comprise dilute mode schemes were
first studied in a matrix of open-loop tests in the ADL pilot plant. The
individual unit operations were then to be assembled and subjected to
VII-10
-------
extended closed-loop testing. During the open-loop tests in the pilot
plant, unexpectedly high effluent soluble calcium levels were observed
in some of the runs. Saturation tests indicated that the solutions were
supersaturated with respect to calcium sulfate (gypsum). In an attempt
to better understand the phenomena responsible for calcium supersatura-
tion, a program of laboratory experiments was undertaken in parallel
with the studies being performed in the pilot plant.
Initially, a series of experiments was carried out in a laboratory scale
continuous stirred tank reactor (CSTR). Batch experiments were then
carried out, in which the rates of decrease of soluble calcium concen-
trations in the presence of different amounts and types of suspended
solids were monitored. Measurements of the equilibrium solubility of
gypsum at various sodium sulfate concentrations, hydroxide concentra-
tions, and temperatures were also performed to obtain fundamental data
against which to compare experimental observations.
1. Studies of Calcium Precipitation from CSTR Effluents
The effect of total oxidizable sulfur (TOS) concentration on the compo-
sition of the regenerated solution was studied at 50°C in a 1-liter CSTR
operating at a residence time of 30 minutes. The pH of the feed solution
was maintained in the range of 5.4-5.6 in all cases and 300 ppm calcium
was added to the feed solution to simulate a dilute mode scrubber bleed
stream. The lime feed was adjusted so that the amount fed was 110% of
the amount theoretically required to produce 0.12M hydroxide in the
reactor effluent.
The results obtained from three CSTR experiments in which TOS levels
(after dilution by the lime slurry) were about 0.041M, 0.023M, and O.OOM
are shown in Table VII-3. In all cases, the concentration of hydroxide
in the effluent reached only about 0.1M, indicating that a significant
portion of the lime had not reacted. For the intermediate TOS level,
the effluent hydroxide concentration was about 10% less than that ob-
served in the other two experiments. In both experiment 47 and experi-
ment 50, the concentration of TOS in the reactor effluent was about 0.02M,
approaching the sulfite saturation level for these solutions. The solids
from experiment 47, in which a very small reduction in soluble TOS was
observed, contained only 6% of the calcium/sulfur salts as calcium sulfite.
In all three cases, the reactor effluent solutions were supersaturated
with respect to gypsum. With TOS present in the feed, soluble calcium
levels were higher than in experiment 48 where no TOS was fed.^ Equilib-
rium saturation levels of calcium with respect to gypsum at 20°C, as
determined by equilibrating the supernatant liquid with excess gypsum
crystals, were essentially the same for all three experiments. The same
test, but carried out at the reaction temperature of 50 C, indicated that
gypsum solubility was greater at that higher temperature. Consequently,
conducting the saturation test at a temperature lower than the reaction
temperature produced results which would indicate a greater degree of
supersaturation than actually existed under reactor conditions.
VII-11
-------
TABLE VII-3
DILUTE MODE CSTR EXPERIMENTS'
Feed Solution
[TOS],
[Ca^L
[so=],
(M)
, (M)
(M)
Expt. 50
.0407
.0075
.557
Expt. 47
.0227
.0075
.504
Expt. 48
0^0
.0075
.517
Reactor Effluent
Solution
[TOS], (M)
[Ca"*4"], (M)
.0203
.0265
.0220
.0275
[Ca ], Gypsum
Saturation, (M)
[SO'], (M)
[OH-], (M)
.0164 (20°C)
.486
.106
.0168 (20°C)
.0183 (50°C)
.458
.097
0.0
.0230
.0157 (20°C)
.0170 (50°C)
.478
.107
Reactor Solids
Mol % CaSO,/CaSO
4 x
43.4
94.0
100.0
Q
CSTR residence time, 30 mins.; temperature, 50°C; feed solution pH,
5.4-5.6; limestone feed stoichiometry, 110% of amount required to
produce 0.12M hydroxide.
VII-12
-------
In an attempt to gain insight into the effect of various combinations
of reactor residence time and suspended solids concentrations on effluent
soluble calcium levels, a series of batch experiments was conducted in
which portions of effluent slurry from the CSTR were allowed to react
for additional periods of two hours in the presence of varying amounts
of added suspended solids. Experiment 48 (no TOS) was repeated and the
CSTR was operated for a time sufficient to allow a quantity of reaction
solids to be produced and collected. After the solids were collected,
CSTR operation was continued and portions of the effluent slurry were
taken directly from the reactor overflow and placed in stirred flasks
held at 50°C. Quantities of additional solids (either reactor product
solids or reagent grade gypsum crystals) were added to the flasks imme-
diately after the slurry was collected from the CSTR. Samples of slurry
were then taken from the stirred flasks and analyzed at intervals through-
out the two-hour reaction period. The results from this set of experi-
ments are shown in Table VII-4. The weight percent solids listed in
Table VII-4 are the total suspended solids which were present in each
flask, i.e., including the 0.5 weight percent solids initially present
in the reactor slurry itself.
For the sample which had no additional solids added, the calcium con-
centration decreased to essentially the saturation value of 0.017M after
sixty minutes. With a total of 3% solids present during the batch
reaction, the soluble calcium levels after two hours were close to the
saturation value with either reactor solids or gypsum added. However,
the rate at which the calcium level fell was markedly greater in the
case when the additional 2.5% solids were reagent-grade gypsum. With
6% total suspended reactor solids, the soluble calcium levels remained
essentially constant for the duration of the two-hour experiment, but
in the case where the added solids were gypsum, the levels were uniformly
lower—essentially at saturation. With 9% reactor solids, the soluble
calcium level again remained constant and was essentially the same as
the level with 6% reactor solids.
The reason for the apparently faster equilibration, but to a higher final
calcium level at 6% and 9% reactor solids, was not immediately obvious —
particularly when that behavior was contrasted to the behavior observed
when gypsum was added. In the latter case, the equilibrium calcium level
appeared to be independent of the total suspended solids concentration.
One possible explanation was that since the reactor solids contained un-
reacted lime, adding further reactor solids resulted in a larger amount
of unreacted lime, which led to the higher calcium levels. The results
of the experiments do suggest, however, that under the conditions studied,
additional reaction time may be as effective, or even more effective, in
lowering the effluent soluble calcium level than recycling reaction solids.
2. Measurements of Gypsum Solubility
Since the gypsum solubility was observed to increase with temperature
in the CSTR saturation tests and since there was a possibility that
hydroxide concentration could also affect gypsum solubility, a series
VII-13
-------
TABLE VII-4
STUDY OF CALCIUM DE-SUPERSATURATION IN POST-CSTR BATCH REACTIONS5
Time After No Additional 3% Total Solids 6% Total Solids 9% Total Solids
Slurry Removed Solids (0.5%) Reactor Solids Gypsum Reactor Solids Gypsum Reactor Solids
from Reactor ,..,_,. ++++++
(mins.) Ca (M) OH~(M) Ca (M) OH~(M) Ca (M) OH~(M) Ca (M) QH"(M) Ca (M) OH"(M) Ca (M) OH~(M)
15 .0205 — .0201 — .0220 -- .0194 — .0174 -- .0190
30 .0188 .112 .0198 .112 .0176 .115 .0191 .118 .0179 .125 l.0221]b .125
60 .0166 — .0189 — .0181 — .0188 — .0169 -- .0190
120 .0176 — .0178 — .0176 .115 .0191 .115 .0173 — .0194 .122
aReactor solids and slurries obtained from repeat of CSTR experiment 48.
Conditions: 50°C, 30 mins. res!
Reactor effluent solution: Ca"1""1"
b
Conditions: 50°C, 30 mins. residence, TOS = 0, SO^ = 0.5, stoichiometry = 110% of lime required for 0.12M OH"
'f = .0218M, Ca"*4 gypsum saturation = 0.0170M, OH~ = .112M
Doubtful value.
-------
™nrrn ** * functlon of s°dium sulfate
concentration, hydroxide concentration, and temperature were made for
solutions equilibrated in contact with excess solid gypsum.
The experiments were carried out by adding excess (about 5 g/100 ml solu-
tion) reagent grade CaSO^ - 2H20 to sodium sulfate/hydroxide solutions
at room temperature. The mixtures were stirred, brought to temperature,
and stirred at temperature for one-half hour. The solids were then al-
lowed to settle for one-half hour at temperature, and finally a portion
of the supernate was removed and centrifuged using temperature equili-
brated glassware. The calcium and sulfate concentrations in the centrates
were determined by the usual titrimetric methods.
The results of this series of measurements are shown in Table VII-5. In
addition to a general increase in calcium concentration with an increase
in temperature from 20°C to 55°C, a significant increase in soluble cal-
cium was observed when the concentration of hydroxide was increased from
0-0.1M to 0.15M. The 0.017M concentration of calcium measured in experi-
ments G and H at 55°C corresponds quite closely to the equilibrium levels
observed in the batch and CSTR experiments described above. The fact
that the concentration of hydroxide needed to be raised to 0.15M to
achieve a soluble calcium level of 0.019M in these solubility experi-
ments suggests that the 0.019M level observed in the earlier experiments
(where hydroxide concentration was never greater than 0.125M) was prob-
ably due, in part, to the presence of unreacted lime.
These results show that erroneously high gypsum supersaturation can be
indicated if solutions are allowed to cool before gypsum crystals are
introduced for determination of saturation levels.
3. Additional Batch Reaction Studies of Dilute
Mode Regeneration with Lime
To explore further the possibility that unreacted lime might have been
responsible for the elevated calcium levels observed in some of the ex-
periments described earlier, two batch reactions were carried out in
which 0.5M N32S04 solutions were reacted with different amounts of lime.
In one experiment (80% stoichiometric), the amount of lime added was 80%
of that which would have been required to produce 0.12M hydroxide in
solution. In the other, 130% of the amount of lime required to regen-
erate to 0.12M hydroxide was added. The measured levels of soluble
calcium and hydroxide as a function of time in these experiments are
shown in Table VII-6. For the experiment at 80% of stoichiometric, the
soluble calcium level had approached the saturation value of 0.0165M after
about 45 minutes. In the other experiment where excess lime was added,
the soluble calcium level was higher and remained essentially constant
throughout the experiment. The 0.019M level observed was essentially
the same as the level observed in the previous batch experiments carried
out with 6% and 9% reactor solids, suggesting that excess lime may have
produced the higher calcium levels.
VII-15
-------
TABLE VII-5
M
I
SOLUBILITY
OF CALCIUM SULFATE IN SODIUM SULFATE/HYDROXIDE SOLUTIONS
Sodium Salt Species
Expt.
Series
A
D
B
F:
H
C
F
G
H
J
(Mol Per
Na0SO,a
0.267
0.287
0.506
0.535
0.509
0.988
1.04
0.536
0.527
0.526
Liter)
NaOH
0
0
0
0
0
0
0
0.100
0.100
0.150
Soluble Calcium - Mol Per Liter at Noted Temperature
VL5°C 20°C
0.0113
0.0102
0.0128
0.0126
0.0125
0.0155
0.0160
0.0150
0.0144
0.0156
40°C
0.0117
0.0118
0.0140
0.0130
0.0129
0.0163
0.0164
0.0154
0.0160
0.0173
55°C
0.0120
0.0118
0.0143
0.0136
(0.0124)b
lost
0.0172
0.0171
0.0169
0.0191
70°C
(0.0160)b
0.0113
0.0147
0.0140
0.0141
____
0.0177
(0.0160)b
0.0185
0.0190
The sulf ate values noted are the average for the experimental series ; a separate sulf ate measurement
was made for each calcium measurement.
b
Values in parentheses are thought to be erroneous.
-------
TABLE VII-6
M
I
BATCH
Time After Addn.
(mins . )
15
30
45
60
90
120
REACTIONS OF LIME WITH SODIUM SULFATE SOLUTIONS AT TWO STOICHIOMETRIES3'
80% Stoichiometry
130% Stoichiometry
. Ca"1"^" Gypsum , .
Ca (M) Saturation(M) @ 5Q°C OH~(M) Ca (M)
.0200 .0160
.0182 .0175
.0176 .0157
.0170 .0176
.0168 .0152
.0168 .0168
Average . 0165
.092 .0193
.0935 .0194
.094 .0197
.094 .0190
.095 .0187
.095 .0190
OH"(M)
.120
.120
.122
.122
.124
.124
a!00% Stoichiometry defined as amount of lime required to produce 0.12M OH in solution.
bn
Temperature, 50°C;
i, 0.5M.
-------
Several additional batch experiments were carried out to investigate
the kinetics of the regeneration of dilute mode liquors with TOS con-
centrations in the range of 0.02M to 0.05M. Regeneration behavior in
that region of TOS concentration was of interest because it represented
a region of transition between: the regeneration of solutions containing
very little or no TOS (produced by intentionally oxidizing scrubber ef-
fluent prior to regeneration) in which the solid product is very nearly
pure gypsum, and the regeneration of solutions containing higher levels
of TOS in which no distinct gypsum phase can be observed in the product
solids.
The results obtained when solutions containing 0.025M TOS and 0.047M TOS
were regenerated with lim'e in batch reactions at 50°C are shown in Table
VII-7. Included for reference from Table VII-6 are the results of the
experiment in which no TOS was present. In all cases, the amount of lime
was 130% of the amount which would have been required to produce 0.12M
free hydroxide in solution. The concentration of sodium sulfate was 0.5M,
and when TOS was present, the ratio of bisulfite:sulfite was 9:1.
Three distinctly different types of behavior were observed in the three
experiments. When no TOS was present, the reaction was very nearly com-
plete after 15 minutes. With 0.025M TOS present, the concentrations of
calcium, hydroxide, and TOS in solution remained essentially constant
for samples taken between 5 minutes and 30 minutes after the start of
the experiment, indicating that no reaction was taking place. After that
apparent induction period, the reaction proceeded during the interval
between 45 minutes and 90 minutes with a resultant increase in OH con-
centration. However, the concentration of TOS continued to remain essen-
tially constant, and the solid reaction product must have been primarily
gypsum. Whether the apparent decrease in TOS during the first 5 minutes
of reaction was real or due to experimental error is difficult to ascer-
tain because the starting solution, which was prepared with known weights
of reagents, was not analyzed before the lime was added. The reduction
in acidity after 5 minutes of reaction was greater than the observed
increase in calcium concentration, suggesting that some calcium sulfite
might have been precipitated.
In the third experiment, the initial TOS concentration of 0.047M was
confirmed by analysis before lime was added. A substantial decrease
in TOS occurred during the first 5 minutes of reaction, and TOS was ob-
served to continue to decrease throughout the experiment reaching 0.011M
after 180 minutes — well below the final TOS level in the preceding ex-
periment, and below the saturation level for TOS in this solution. This
possibly may be due to coprecipitation of calcium sulfite with calcium
sulfate in a manner analogous to the coprecipitation of sulfate with
calcium sulfite in concentrated dual alkali modes.
VII-18
-------
TABLE VII-7
BATCH REACTOR STUDIES OF THE EFFECT OF TOS LEVEL
ON REGENERATION WITH LIMEa
Time
(mins . )
0
5
10
15
30
45
60
90 .
120
180
Initial [TOS] = 0
tCa"1"1"], (M) [OH-], (M)
0.0 nil
—
—
0.0193 0.120
0.0194 0.120
0.0197 0.122
0.0190 0.122
0.0187 0.124
0.0190 0.124
—
Initial [TOS] = 0.025M
tea"""], (M)
0.0
0.0475
0.0455
0.0475
0.0475
0.0340
0.0250
0.0213
0.0208
—
[OH~], (M)
(0.0225)b>C
0.0860
0.0880
0.0890
0.0885
0.1025
0.1175
0.1230
0.1230
—
[TOS], (M)
0.025°
0.0190
0.0186
0.0168
0.0187
0.0199
0.0185
0.0162
0.0184
—
Initial [TOS] = 0.047M
[Ca*\afd> «
—
0.0163
0.0166
0.0164
0.0166
0.0177
0.0173
0.0175
0.0175
~
' tCa"14"], (M)
0.0
0.0438
0.0340
0.0345
0.0348
0.0325
0.0313
0.0273
0.0250
0.0223
[OH~], (M)
(0.0423)b
0.0904
0.0980
0.1010
0.1022
0.1022
0.1048
0.1110
0.1128
0.1192
[TOS], (M) [
0.047
0.029
0.022
0.018
0.017
0.017
0.013
0.012
0.012
0.011
Ca++]sat'd' (M)d
—
0.0163
0.0171
0.0168
0.0165
0.0163
0.0164
0.0169
0.0170
0.0177
temperature, 50°C; [Na-SO,], 0.5M; lime feed, 130% of amount required to produce 0.12M free hydroxide.
Acidity equivalent to [HS07].
CBased on weighed reagents, not analyzed.
Soluble calcium concentration after equilibrating supernate with excess gypsum at 50°C.
-------
D. PILOT PLANT OPERATIONS
1. Introduction
As discussed previously, operation of dual alkali systems at dilute
active sodium concentrations offers one possibility for increasing the
proportion of calcium sulfate in the waste cake. Dilute dual alkali
modes are defined as those in which the reactor solutions are saturated
with respect to calcium sulfate. Under normal operating conditions this
occurs with active sodium concentrations roughly equal to or less than
0.15M Na+.
There are two basic approaches to operating in the dilute mode using lime
for regeneration. One approach is to oxidize the scrubber effluent liquor
intentionally to convert all the sulfite and bisulfite to sulfate prior
to regeneration. The resulting waste solids, therefore, are calcium
sulfate, usually in the form of gypsum, with no calcium sulfite. The
second approach is simply to regenerate the scrubber bleed liquor di-
rectly, without oxidation. The waste cake is then a mixture of calcium
sulfite and sulfate with the ratio of the two salts dependent upon the
rate of oxidation in the scrubber system and the concentrations of sodium
salts in the liquor. The calcium sulfate can be produced as a dihydrate
(gypsum) or hemihydrate mixed with sulfite, or both, depending upon the
system conditions.
Both of these approaches were included in the evaluation of dilute modes
in the pilot plant. Testing was performed in two phases. The first
phase consisted of batch and continuous open-loop runs of specific unit
operations in the regeneration system. These included tests of the ab-
sorbent regeneration reactor, the carbonate softening reactor, and an
aeration system for oxidation of the scrubber bleed liquor. Emphasis
in this phase of testing was placed on the regeneration reactor. Runs
were geared toward characterizing reactor performance over a range of
conditions in order to develop suitable design and operating parameters.
Testing of the oxidation of scrubber bleed liquor and carbonate soften-
ing was included primarily to ensure that no major difficulties would be
encountered in implementing these operations in the pilot plant system.
For the most part, the softening and oxidation operations were straight-
forward and not considered to present undue problems in full-scale systems.
The second phase of testing consisted of a number of closed-loop runs
combining gas scrubbing, absorbent regeneration, and waste solids sep-
aration and dewatering. The object of these runs was to confirm the
open-loop results; to optimize and characterize, to the extent possible,
the performance of the integrated dilute mode operation; and to identify
major problem areas or limitations of dilute mode operations. The runs
lasted from a few days to up to a few weeks, depending both on the type
of data required and the success of the system operation.
Throughout all of this work, hydrated lime was used for regeneration.
The hydrated lime was metered to the reactor system as a dry solid
VII-20
-------
(90% through 200 mesh). On the average, the lime contained approximately
87% available calcium as calcium hydroxide. The hydroxide content of
different batches of lime, though, ranged from 80% to 92%.
2. Reactor System Characterization (Open-Loop Operation)
As in the testing of the concentrated lime mode, the characterization
of the reactor system operation was based upon four process criteria:
(1) calcium utilization, (2) level of soluble calcium in the reactor
effluent liquor, (3) sulfate precipitation, and (4) the dewatering prop-
erties of the waste solids produced. More than 35 batch and continuous
flow runs were made over a range of reactor feed liquor compositions and
reactor operating conditions. The purpose of these runs was to determine
how major process variables affected the reactor performance in order to
establish a set of operating conditions that would ensure:
• better than 90% calcium utilization with regeneration to
hydroxide concentrations above 0.08M;
• less than 50 ppm of supersaturation of calcium with respect
to gypsum in the reactor effluent liquor; and
• waste solids which could be filtered to a minimum of 50%
solids and would provide a clear liquor overflow upon
settling.
The limitation on soluble calcium levels is intended to minimize plugging
of lines in the reactor and dewatering systems and reduce scale potential
in the scrubber. A minimum of 50% solids in the filter cake would also
ensure acceptable cake washing and handling properties.
a. Batch Experiments
The batch runs were made to verify that an equilibrium hydroxide level
in excess of 0.10M could be achieved by reacting 0.5M Na2SOtf with hydrated
lime and to confirm that supersaturation with respect to gypsum could be
controlled. A concentration of 0.5M Na2SOif was chosen because previous
work by ADL9 and General. Motors9 indicates that hydroxide levels of about
0.15M should be achievable. Hydroxide levels greater than 0.10M are
desirable in order to reduce pumping rates and minimize tank sizes.
A total of 12 batch experiments was conducted using lime/Na2SOi,. feed
stoichiometries ranging from 0.05-0.25 mols Ca(OH)2/mol Na2SOi+. In a
number of the runs at the higher stoichiometries, product solids were
also added to study the effects of seeding. All .runs were made at
temperatures in the range of 45°C to 50°C.
Most experiments were limited to 30-minute reaction times, starting with
reactant liquor pH levels of 3.8 (adjusted by sulfuric acid). A few runs
were made starting at a pH of 7.4, and in one run the reaction time was
extended to 60 minutes. Longer batch reaction times would translate into
VII-21
-------
excessively long holdup times for continuous flow reactor systems , par-
ticularly those involving single CSTR's. Such long holdup times would
adversely affect system design and capital costs.
Equilibrium Hydroxide
Figure VII-5 shows the effects of lime/Na2SOit feed stoichiometry and the
addition of product solids on the level of hydroxide produced in these
batch tests. The product solids added in these runs were analyzed for
available calcium hydroxide and the calcium hydroxide contained in the
solids has been included in the calculation of the lime/Na2SOit feed
stoichiometry. In all experiments at lime/Na2SOi+ feed stoichiometries
below 0.17, lime utilization was on the order of 90% or greater. The
presence of additional nucleation sites created by product solids
noticeably increased the utilization of lime and the level of hydroxide
achieved. Extending the reactor holdup time to 60 minutes did not appear
to increase lime utilization significantly.
The maximum level of hydroxide attained in these batch experiments was
0.145M, which was reached when product solids were added to the reaction
mix. It is apparent from the results of these experiments that 0.145M
closely approaches the equilibrium hydroxide concentration for the ex-
perimental conditions. It is also consistent with the results from
previous ADL work.
For the sulfate regeneration reaction:
Ca(OH)2 + Na2SO[t + 2H20 t CaSO^ • 2H20 + 2NaOH (12)
the equilibrium constant, K Q, is given by:
K
where the activity of the hydroxide and sulfate ions are:
and
Here the activity coefficients, Y0H~and Yso=, are a function of the
hydroxide and sulfate ions, respectively; the total ionic strength,
and the temperature.
VII-22
-------
c
o
§
c
0.20
General Feed Liquor Conditions:
[SOI] = 0.5-0.55 M
[TOS] = 0
0.15
M
S3
OO
o
CT
•o
HI
o>
c
CD
DC
0.10
O 30 min Batch Test
• 60 m in Batch Test
0.05
> 0.5-1.
2% Solids
30 min Batch Test with Product
Solids Added to Attain 4-6% Solids
0.05
0.1 0.15 0.2
Reactor Feed Stoichiometry (Ca(OH)2/Na2SO4)
0.25
0.30
_FIGURE VII-5 PRELIMINARY BATCH EXPERIMENTS
DILUTE MODE
-------
The equilibrium ratio of hydroxide and sulfate concentrations, then, is
given by:
[OH ]2
[S0=]
Y
so"
(Y -
U
2 ° eq
where K ' is a function of ionic strength and temperature.
eq
Data previously obtained in batch experiments performed by ADL for the
Illinois Institute for Environmental Quality^ indicate the sensitivity
of the reaction equilibrium to temperature and sulfate concentration.
These data are summarized below along with the data from this study.
Equilibrium Concentrations (M)
Study Temperature (°C)
IIEQ 30
EPA 45
IIEQ 71
IIEQ 71
[SO^]
0.47
0.45
0.47
0.25
[OH ]
0.155
0.145
0.110
0.090
tea**"]
0.017
0.018
0.020
0.0165
[OH~]2
[S0=]
0.051
0.047
0.026
0.032
Calcium Supersaturation
In almost all runs the concentration of soluble calcium in the product
slurries ranged from 680-750 Mg Ca4^"/!. This corresponded in most cases
to less than 50 ppm Supersaturation with respect to gypsum. (Calcium
saturation levels were determined by introducing an excess of gypsum
crystals into the product slurries.) Only in the run where the reaction
was started at a pH of 7.4 (with a reaction time of 30 minutes) was sig-
nificant Supersaturation observed. The Supersaturation may be attributable
to the slow generation of seed crystals. In an analogous run in which
product solids were added to raise the suspended solids level from about
0.8% to about 3.5%, there was no Supersaturation after 30 minutes of
reaction.
b. Continuous Flow Reactor Runs
The matrix of experimental conditions covered in the open-loop, continuous
flow reactor runs is given in Table VII-8. This matrix of experiments was
designed to evaluate absorbent regeneration in the two different dilute
mode options previously discussed ~ the direct regeneration of scrubber
bleed liquor, and the regeneration of liquor after oxidation of TOS. Two
nominal reactor feed TOS levels were selected as representative of each
of these two options — 0.0-0.01M TOS for operation with oxidation and
0.04-0.05M TOS for operation without oxidation. Two additional runs were
also made at 0.07M and 0.13M TOS. The range of compositions examined is
summarized below.
VII-24
-------
M
N3
Ul
TABLE VII-8
EXPERIMENTAL MATRIX FOR OPEN^LOOP, DILUTE MODE REACTOR TESTS
Nominal Feed
[SO'], (M)
.q_ _. >
0.25
0.25
0.25
0.25
0.25
0.25
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.75
0.75
0.75
Liquor Composition
[TOS], (M)
0.05
0.05
0.05
0.05
0.05
0.05
0
0
0
0
0
0
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.15
0
0
0.05
pH
5.8
5.8
5.8
5.8
5.8
5.8
4
4
4
4
4
4
5.8
5.8
5.8
5.8
5.8
5.8
5.8
5.8
4
4
5.8
Run
No.
502
501
507
505
504
506
510
513
522
521
525
524
511
515
516
517
518
519
520
512
514
523
503
Reactor Configuration^
Holdup /Type Solids
(min . ) /- Recycle
30/CSTR
30/CSTR
30/CSTR
5 + 30/ADL
60/CSTR
5 + 60/ADL
30/CSTR
30/CSTR
30/CSTR
60/CSTR
60/CSTR
60/CSTR /
30/CSTR
30/CSTR
5 + 60/ADL
5 + 60/ADL
5 + 60/ADL
60/CSTR /
120/CSTR /
30/CSTR
30/CSTR
60/CSTR
30/CSTR
Lime Feed Stoichiometry
(% of req't for 0.15M OH")
21
35
67
52
115
57
73
110
77
78
95
88
60
75
42
40
58
66
75
72
115
82
83
-------
A sulfate level of 0.5M was chosen as the basis for this continuous flow
reactor study. A few exploratory runs were also made at nominal sulfate
levels of 0.25M and 0.75M in order to determine the effect of sulfate
concentration on reactor performance. The pH of the feed was adjusted
depending upon the concentration of TOS. At 0.05M TOS in the reactor
feed, the feed liquor pH was adjusted to 5.8, which represents a bisul-
fite- to-sulfite ratio of about 5:1. In all but one run involving
little or no TOS, the pH was adjusted to approximately 4. This corre-
sponds to an acidity of roughly 0.015M, a level that would be achieved
with partial neutralization prior to, or during, oxidation of the scrubber
bleed. Thus, the basis for the matrix of tests shown in Table VII-8 is
summarized as follows:
Nominal Composition of Reactor Feed Liquors
Component Without Oxidation With Oxidation
TOS 0.05, 0.07, 0.13 0, 0.01
pH 5.8 4 and 6.5
SO" 0.25, 0.5, 0.75 0.5, 0.75
Ca*"1" 0 and 500 500
In addition to the range of reactor feed compositions, a number of reactor
parameters were also varied: reactor type (ADL two-stage reactor system,
CSTR), solids recycle (to 4% solids), reactor holdup (30-120 minutes),
and lime feed stoichiometry (20% to 115% of that required to regenerate
0.15M OH~, the equilibrium level of hydroxide determined in the batch
experiments for reaction with 0.5M sulfate). This equilibrium level of
hydroxide is used as a normalizing factor to put the range of lime feed
rates used in all runs on a common basis. However, it is realized that
the equilibrium level of hydroxide will vary with ionic strength and
changes in the apparent solubility products of gypsum and calcium
hydroxide.
The results of this series of continuous flow tests and the implications
of these results to reactor design and operation are discussed in the
following sections in terms of the four performance criteria previously
cited: calcium utilization, sulfate precipitation, calcium supersatura-
tion, and waste solids properties.
Calcium Utilization
Figure VII-6 summarizes the utilization of calcium observed over the range
of liquor compositions and reactor operating conditions. In this plot of
lime utilization as a function of lime feed stoichiometry there are dis-
tinctly different functional relationships apparent for 0.25M S0= and
0.5-0.75M SO". The utilization of calcium is consistently higher at the
VII-26
-------
100
90
I
o
3
a)
1 80
'<5
>
(0
o
.t! 70
|
|o
<3
60
50
— — VP* ^5^ lc» — — ,
502 ^-^ 517 516 "-^^518 —
^
A
501
General Conditions:
Temp = 45-50°C
pH = 4 (no TOS)
5.5-6 (0.5M TOS)
Feed
[SO']
Symbol (M)
A 0.25
A 0.25
O 0.5-0.75
1
d
•
I
D 0.55-0.75
H I
• , I
V.
\
\
"S.^ 520
^519
\J ^_
\ 515 \ •
\ \524
\ H \
A^» CO1 V
\ O \
506 A \ 503 Nr-»~0.5-0.75MSOJ
5°7 ^v D\
^ 510 N.
Liquor
[TOS]
(M)
0.05
0.05
0.05
0.05
0.05
0
0
0
I
\ H\
\ 525 \
0.25MSO4 -* ^
Reactor Holdup \
(min) \
\
30 \
60 \
30
60
60 with solids recycle
30
60
60 with solids recycle
I I 1
\
\
\
\
\
\
\H513B
v \
\ rVtcio/v
\ \ A
\ ,504
\ \
\ i
' \
514Q.
20 40 60 80 100
Lime Feed Stoichiometry (% of Ca(OH)2 required to reach 0.15IVI OH~)
120
FIGURE VII-6 CALCIUM UTILIZATION IN CONTINUOUS, OPEN-LOOP RUNS
VII-27
-------
higher sulfate concentrations for equivalent lime stoichiometries. This
can be attributed to reduced rates of reaction at the lower sulfate level,
due to the lower equilibrium levels of hydroxide. The results show that
at 0.25M sulfate, calcium utilization suffers when the lime feed rates
are greater than that required to produce hydroxide levels above 0.05M.
This is shown more clearly in Figure VII-7, in which lime utilization is
plotted as a function of the hydroxide concentration achieved. Thus, if
it is necessary or desirable to regenerate to 0.08M OH or higher, the
sulfate concentration must be on the order of 0.5M.
At 0.5M sulfate, better than 90% lime utilization can be quite readily
attained with regeneration to hydroxide levels slightly higher than
0.10M OH~. However, in order to ensure 90% utilization with reactor
holdup times on the order of 60 minutes or less, the lime feed stoichio-
metry must be maintained at or below 85% of that required to regenerate
0.15M hydroxide. _That is, a reasonable practical limit for regeneration
is about 0.12M OH . This is clearly indicated in Figure VII-7.
Lime feed stoichiometry and sulfate concentration were found to be the
two most important factors affecting lime utilization. By comparison,
the four other variables covered in these runs appear to have little
discernible effect on lime utilization — reactor type (ADL versus CSTR),
liquor holdup time in the reactor (30 minutes versus 60 minutes), TOS
concentration (0-0.13M), and recycle of product solids. The data in
Figure VII-6 suggest that some gains in utilization can be made by in-
creasing the reactor holdup time from 30 minutes to 60 minutes (or higher)
and by recycling product solids. The data suggest that at 0.5M sulfate,
operation with 60-minute holdup and with solids recycle might increase
lime utilization by 5% over that achieved with a 30-minute holdup and
with no solids recycle.
The fact that there is little effect of TOS concentration up to 0.13M TOS
on lime utilization is somewhat surprising. The runs that were made with
TOS in the reactor feed liquor also had high acidity levels in the feed
liquor compared with those runs involving little or no TOS. For example,
a typical reactor feed containing 0.05M TOS had a pH of about 5.8 and on
the order of 0.045M titratable acidity, while a typical reactor feed with
no TOS had a pH of about 4 but had only about 0.015M titratable acidity.
In regenerating to a hydroxide level of 0.12M, the liquor with TOS would
consume about 30% of the lime in neutralizing the acidity, while the
liquor with no TOS would consume about 10% of the lime for neutralization.
Since the neutralization reaction is virtually immediate, it would be ex-
pected that the runs with 0.05M TOS would exhibit higher lime utilizations.
The results do not confirm this; if anything, there is a tendency for the
runs with TOS to have slightly lower utilizations (Figure VII-7).
This agrees with laboratory results which indicate that low TOS levels
retard the reaction rate. The higher lime utilization in runs without
TOS may be related to the fact that in the presence of low levels of TOS,
the solids are a mixture of calcium sulfite and sulfate which could super-
saturate more easily. Such supersaturation, if not avoided, can retard
VII-28
-------
100
General Conditions:
Temp = 45-50°C
pH = 4 (no TOS)
5.5-6 (0.5M TOS)
Feed Liquor
[804] [TOSJ Reactor Holdup
(M) (mini
0,25 0.05 30
0.25 0.06 60
30
60
60 with solids recycle
30
60
60 with solids recycle
0.5-0.75M S04
75
0.05
0.10
Effluent [OH~],(M)
0.15
0.20
FIGURE VII-7 HYDROXIDE CONCENTRATION VS CALCIUM UTILIZATION
IN CONTINUOUS, OPEN-LOOP EXPERIMENTS
VII-29
-------
the rate of lime dissolution and thereby the overall rate of reaction,
resulting in lower lime utilization. The question of supersaturation
is discussed later.
Sulfate Precipitation
The potential for sulfate precipitation during absorbent regeneration
is substantially greater in dilute mode operations than in concentrated
mode operations. In fact, since sodium hydroxide can be regenerated from
pure sodium sulfate by reaction with lime, the opportunity exists for
producing a relatively pure calcium sulfate waste product by oxidizing
the TOS prior to regeneration.
The effects of sulfate and TOS concentration in the reactor feed liquor
on the level of sulfate in the waste solids are shown in Figures VII-8
and VII-9. Figure VII-8 shows a strong dependence of sulfate precipi-
tation on soluble sulfate concentration at soluble sulfate levels greater
than 0.5M. Figure VII-9 shows that, for reactor feed liquors containing
0.5M S0~, there is a strong sensitivity of sulfate precipitation to soluble
TOS concentration at TOS levels below about 0.04M. Each of these relation-
ships in Figures VII-8 and VII-9 represents but one of a family of para-
metric curves.
Similar relationships developed for the concentrated active sodium regime
in previous laboratory and pilot plant test programs had indicated that
sulfate precipitation equivalent to as much as 25% of the total calcium-
sulfur waste salts could be reasonably achieved. This would be equivalent
to keeping up with overall system oxidation rates on the order of 25-30%
of the S02 removal (taking into account Na2SOit losses in the waste cake).
Concentrated mode data on crystalline morphology and water of hydration
indicate that the calcium sulfate at levels of up to about 25% of the
calcium-sulfur salts is present as a solid solution or mixed crystal
with calcium sulfite, and not as gypsum. The results shown in Figures
VII-8 and VII-9 corroborate these concentrated mode results. The sulfate
precipitation behavior suggests that there may be a change in the sulfate
precipitation mechanism when conditions drive sulfate levels in the solids
to greater than about 30%. It is likely that at below 30% the calcium
sulfate is present as a solid solution or mixed crystal with calcium
sulfite, and that above 30% the sulfate is precipitated as gypsum.
The exact liquor concentrations in the reactor ( [SOJj / [SOg] and total
ionic strength) will dictate the point at which gypsum will begin to pre-
cipitate. While the fluctuations in the soluble TOS data on the reactor
effluent preclude an accurate estimate of where this change may occur,
the feed liquor data (Figures VII-8 and VII-9) suggest that this break
point occurs at a feed ratio of sulfate-to-TOS of roughly 15-20 in the
range of ionic strengths investigated.
The advantage of the dilute mode, therefore, lies in its potential for
keeping up with oxidation rates in excess of 30% of the S02 removal.
VII-30
-------
0.70
0.60
Conditions:
[TOS] = 0.04-0.06M
Reactor Holdup = 30—60 min.
0.50
8
o
O
3
—
o
0.40
0.30
0.20
0.10
0.2
0.4
[SOJ],
0.6
0.8
1.0
FIGURE VI I -8
CALCIUM SULFATE PRECIPITATION AS A FUNCTION
OF SOLUBLE SULFATE CONCENTRATION
VII-31
-------
X
O
CO
<3
o
E
d
CO
0
"5
1.00
0.90
0.80
0.70
0.60
0.50
0.40
0.30
0.20
0.10
0
1
\
' General Feed Conditions:
\ [SOj] = 0.5 M
-, pH='5.6-5.8
' • - 30 min CSTR
\ 0-60 min CSTR
~ \
\
\
\
\
\
\
\
^N^°
A
—
1 I 1
0.05
0.10
[TOS], M
0.15
0.20
FIGURE VII-9 CALCIUM SULFATE PRECIPITATION VS
TOS CONCENTRATION
VII-32
-------
Similarly, at 0.5M SO^, oxidation rates of 30% would result in scrubber
effluent TOS concentrations less than 0.05M. In a dilute mode applica-
tion the system would probably be charged to a total sodium concentration
consistent with the operational level of sulfate desired. If the scrubber
bleed liquor were not intentionally oxidized, then the proportions of
active sodium (TOS) and sulfate would vary according to the level of oxi-
dation experienced. For example, with system oxidation ra|es of 50%, it
would be necessary to operate the system at about 0.75M S0~ if TOS levels
of 0.05M were desired (Figure VII-8). On the other hand, if a 0.5M SO"
level were desired, then TOS concentration in the scrubber bleed would
be about 0.02M (Figure VII-9).
The data plots given in Figures VII-8 and VII-9 exhibit considerable
data scatter. Some of this can be attributed to slight variations in
sulfate and TOS concentrations. However, the data also cover a range
of lime feed stoichiometries_from 20% to 115% of that theoretically
required to achieve 0.15M OH~. The effect of lime stoichiometry on
sulfate precipitation is shown in Figure VII-10 for sulfate levels of
0.25M and 0.57M. Although there is also scatter in these data which may
amount to + 15%, the data suggest some dependence of sulfate precipita-
tion on lime stoichiometry at 0.57M SO^. Sulfate levels in the solids
generated with a lime feed stoichiometry of 40% are 10-30% higher than
the sulfate levels above 60% stoichiometry. Here, 40% stoichiometry is
roughly equivalent to regeneration to a hydroxide level of 0.02M. The
effect of stoichiometry is similar to the one observed in the labora-
tory work in the concentrated lime mode. However, the number of data
points is small, and the fact that a similar relationship is not seen at
0.25M sulfate indicates that more data would be required to confirm and
explain these results.
Calcium Supersaturation
The levels of soluble calcium present in the reactor effluent liquor
depended upon four factors: the concentration of TOS in the reactor
feed liquor, the extent of regeneration, the solids holdup time in the
regeneration reactor (both nominal fluid holdup and solids recycle), and
the type of reactor system employed. The extent of regeneration (or lime
feed stoichiometry) was of importance only in those runs in which there
were appreciable levels of TOS in the reactor feed liquor. In these
runs the reactor effluent liquor was unsaturated or just saturated with
respect to gypsum when lime feed stoichiometries were low. At low stoi-
chiometries the only sulfate being precipitated was in a solid solution
or mixed crystal with calcium sulfite and the soluble calcium level in
the liquor was controlled by the TOS concentration. However, when lime
was fed at a rate sufficient to regenerate to above 0.06M hydroxide (a
level which could not be reached by reaction with TOS alone), the effluent
liquors tended to supersaturate with respect to gypsum.
In most cases, it is desirable to operate dilute mode systems at the
highest hydroxide concentrations that can be realistically achieved
in Order to minimize liquor flows. A high priority in the pilot plant
VII-33
-------
0.40
0-30
0.10
Conditions: Symbol [S04]
• 0.25 M
o
0.57 M
[TOS]
0.04-0.05 M
0.05-0.06 M
O
c/J
8
O
*t
o
CO
O
0.20
\
\
O
o
0.57 M SO4
0.25 M SO*
_L
_L
-L
_L
_L
20 40 60 80 100
Lime Feed Stoichiometry (% of that required to reach 0.15 M OH~~)
120
FIGURE VII-10 CALCIUM SULFATE PRECIPITATION VS LIME
FEED STOICHIOMETRY
VII-34
-------
work was therefore placed on establishing reactor conditions that would
allow regeneration to greater than 0.06M OH~ without resulting in ex-
cessive supersaturation. Such supersaturation can cause scale deposi-
tion not only in the scrubber system, but also in the reactor system,
dewatering equipment, and interconnecting piping. At hydroxide levels
much below 0.06M OH", pumping costs and tank sizes become excessive.
Table VII-9 lists the operating conditions and soluble calcium levels for
runs at nominal sulfate concentrations of 0.5M and 0.75M in which hydroxide
levels greater than 0.06M were achieved. As previously discussed, such
levels of hydroxide could not be satisfactorily achieved operating with
0.25M sulfate. The soluble calcium data shown in Table VII-9 represent
averages of two or three concentrations measured during each run. In
most cases the individual measurements differed by less than 0.0015M Ca"1"1"
(about 60 ppm).
The level of calcium supersaturation was determined from measurements of
the soluble calcium concentrations in samples of the reactor slurry before
and after mixing the slurry with an excess of gypsum crystals. The calcium
saturation levels determined by this procedure are plotted as a function
of sulfate concentration in Figure VII-11. Not surprisingly, the calcium
saturation level increases with increasing sulfate concentration over the
range of 0.25M to 0.75M sulfate. This increase is due to the marked effect
of ionic strength on the apparent gypsum solubility product.
The data scatter in Figure VII-11 arises from slight differences in tempera-
ture at which the saturation measurements were made and from variations in
ionic strength due to different TOS and hydroxide concentrations. The error
in the calcium determinations themselves is estimated to be on the order of
+ 20 ppm.
The soluble calcium levels in the runs shown in Table VII-9 ranged from
720 ppm to about 950 ppm, a range which corresponds to about 100 ppm to
350 ppm supersaturation. The highest soluble calcium levels occurred in
runs using a CSTR with a 30 minute holdup time. The lowest calcium levels
occurred in runs using either a CSTR with a 60 (or 120) minute holdup time
and recycle of product solids (to 4%), or in the ADL reactor system with a
60 minute holdup time. However, supersaturation occurred in all of the
open-loop runs in which regeneration was carried out to above 0.06M
hydroxide.
The diagrammatic representation of the calcium data in Figure VII-12 sum-
marizes the relative effects of feed liquor composition, holdup time, and
solids recycle on calcium supersaturation in a CSTR. The decrease in
supersaturation with increasing fluid holdup time and product solids
recycle is clear. There appears to be little effect, if any, though, =
of sulfate concentration on supersaturation over the range 0.5-0.75M SO^.
However, the data suggest that low concentrations of TOS increase super-
saturation slightly over the equivalent case where there is no TOS. This
may actually be due to the crystalline form of product solids rather than
simply the presence of TOS. Runs in which a significant fraction of the
VII-35
-------
TABLE VII-9
CALCIUM SUPERSATURATION IN OPEN-LOOP REACTOR RUNS
M
I
00
Nominal
Feed Liquor
Composition
[SO^] [TOS]
(M) (M)
0.5 0.05
0.5 0.05
0.5 0.05
0.5 0.05
0.5 0.05
0.5 0.01
0.5 0
0.5 0
0.5 0
0.5 0
0.75 0
0.75 0
0.75 0
3 S = CaS03 • 1/2H2
Run
No.
511
515
518
519
520
525
510
513
521
524
514
522
523
0; P
Reactor
Holdup /Type
(min.)/-
30/CSTR
30/CSTR
5 + 55/ADL
60/CSTR
120/CSTR
60/CSTR
30/CSTR
30/CSTR
60/CSTR
60/CSTR
30/CSTR
30/CSTR
60/CSTR
Operation
Solids
Recycle
._
—
—
to 4%
to 4%
—
—
—
—
to 4%
—
—
—
Effluent
[os-]
(M)
0.07
0.09
0.06
0.07
0.095
0.105
0.11
0.095
0.095
0.12
0.08
0.10
0.11
Concentrations
Avg. [Ca"1"1"]
(mg/1)
960
850
720
800
780
920
780
950
760
720
- 800
880
860
Avg . Ca"1""1"
Supersat.
(mg/1)
320
240
120
200
180
260
200
360
120
80
200
200
180
Predominant
Solids Forma
S & P
S & P
S & P
S & P
S & P
P
G
P
G
G
G
G
P
& G
& G
= mixed sulfite/sulfate or CaSO, • 1/2H20; G = CaSO,
-------
I,UUO
If
—cs
15
3 y
+-* Q)
(D Q
(/}
i- 4 = 0.75M
TOS = O.OM
Calcium Supersaturation (mg/l)
100 200 300
_J 1 h-
-I
100 200
Calcium Supersaturation (mg/l)
300
400
H
CSTR (30 min)
CSTR (60 min)
CSTR + Recycle
CSTR (30 min)
CSTR (60 min)
CSTR + Recycle
(60 min)
CSTR (30 min)
CSTR (60 min)
400
FIGURE
VII-12 CALCIUM SUPERSATURATION IN RUNS USING A CSTR
VII-37
-------
solids were in the form of hend-hydrates (either pure calcium sulfate or
mixed calcium sulfite/sulfate) showed greater degrees of supersaturation
than corresponding runs in which gypsum (calcium sulfate dihydrate) was
formed. In the presence of appreciable concentrations of TOS (>0.04M),
the hemihydrate form is clearly due to the dominance of the mixed sulfite/
sulfate product and its seeding effects. The reason for the formation of
calcium sulfate hemihydrate when there were very low levels of TOS or no
TOS (runs 513, 523, and 525) is not clear. It may have been due to varia-
tions in temperature. Both the hemihydrate and dihydrate forms are known
to exist in the temperature range of theee runs (45-55°C). The hemihydrate
form is generally considered to be metastable at these temperatures and
its existence could simply have been a transitory phenomenon.
Although saturation was not achieved in any of the runs, the results
indicate what steps can be taken to minimize supersaturation. These
are:
• fluid holdup times of 60 minutes or more;
• elimination of TOS prior to regeneration (by oxidation
of the scrubber bleed);
• recycle of product solids to achieve levels of 4% or
greater suspended solids; and/or
• use of a multistage reactor system.
All of these were, in fact, tried in the closed-loop runs. As will be
discussed later, supersaturation was eliminated by the recycle of product
solids to suspended solids levels greater than 5% in conjunction with a
holdup time of 90 minutes. The ADL reactor system without solids recycle
did not eliminate supersaturation but did maintain calcium levels within
100 ppm of saturation (typically about 50 ppm higher than saturation).
Solids Properties
Two parameters were used as a measure of the dewatering properties of the
solids: the initial settling rate of the solids, and the density of the
settled solids layer after one hour of settling. These two parameters
are plotted in Figures VII-13 and VII-14 for the open-loop runs. It
should be noted that these data are of importance only in terms of the
relative values over the range of conditions. The absolute values of
the parameters should not be compared with similar data obtained in the
concentrated lime mode. The slurry concentrations are different, and
slurry concentration can strongly affect settling rates (hindered settling
versus free settling, etc.).
In general, both the settling rates and densities of settled solids de-
crease with increasing TOS concentration. This is consistent with the
widely held belief that calcium sulfite does not settle or dewater as
well as calcium sulfate and that the presence of calcium sulfite in
VII-38
-------
0.5
0.4 [3-
a
c
0.3
\
\
•5
, Reactor Holdup
30 min 60 min Recycle
0.5 M Na2SO4 O O •
0.75MNa2S04 D H —
S04, 30 min
,,
l
•-O-- O
O~ 0.5M SO4, 30 min
0.05
0.10 0.15
Feed [TOS], (M)
0.20
FIGURE VII-13 SOLIDS SETTLING RATE VS FEED TOS LEVEL
20
I
I
<
15
se ( I
i 10
o
CO
1
O
O
S
§ 0
N
Reactor Holdup
30 min 60 min Recycle
0.5MNa2S04 O 3 •
0.75MNa2S04 D H "
.7S M
1
'5 M S°4=' 30 min
0.05 0.10 0.15
Feed [TOS], (M)
0.20
FIGURE VII-14 CONCENTRATION OF SETTLED SOLIDS VS. FEED TOS LEVEL
VII-39
-------
calcium sulfate cakes reduces the filterability of waste solids. However,
it should be noted that exactly the opposite effect is true in the concen-
trated lime mode — the dewatering properties of the solids deteriorate
with increasing sulfate concentration.
The effect of sulfate concentration on solids properties in the dilute
mode cannot be determined based upon the results in Figures VII-13 and
VII-14. The initial settling rate of the solids appears to increase
by increasing sulfate concentration from 0.5M to 0.75M, but the density
of the settled solids tends to decrease. Such behavior might suggest
that slightly larger crystals are formed at 0.75M, but that these do
not pack as well under a one gravity force.
The effect of fluid holdup time on solids properties is similar to that
observed in the concentrated lime mode. Solids properties tend to dete-
riorate when the reactor holdup time is increased from 30 minutes to 60
minutes. Recycling product solids, on the other hand, tends to enhance
solids properties, apparently by allowing crystals to grow. The fact
that the initial settling rates were slower in runs with solids recycle
is due primarily to the fact that the suspended solids level was eight
times higher in these runs. The presence of this quantity of solids
hindered the normal free fall observed in the thinner reactor slurries.
In runs with solids recycle the settling rates of solids per unit area
(tons/ft2-hr) were equivalent to or greater than those in runs without
solids recycle, but the increase in the settling rates was not sufficient
to compensate for the greater quantity of solid that the thickener would
have to handle in a solids recycle mode. Therefore, use of solids re-
cycle would increase the required thickener area by a factor of two to five.
3. Evaluation of Overall System Performance
(Closed-Loop Operation)
The results of the open-loop testing of the reactor system demonstrated
that in a dilute mode operation the sulfate concentration should be main-
tained at about 0.5M or higher so that hydroxide levels of 0.1M can be
relatively easily achieved, and that elimination of all soluble TOS prior
to regeneration (by oxidation of sulfite to sulfate) improves solids prop-
erties and reduces the potential for supersaturation. These two conditions,
then, provided the basis for the closed-loop dilute mode runs.
a. Description of System Operation
Three closed-loop runs were made at inlet S02 levels ranging from 700 ppm
to 1,100 ppm. Dilute mode operations are most applicable to low-sulfur
coal operations where high rates of sulfite oxidation occur that cannot
be readily handled by concentrated modes.
The first two runs, 601 and 602, were relatively short — each lasting
only a few days. These runs were exploratory in nature, being geared
toward verifying the results of the open-loop tests and establishing
appropriate operating conditions for the third 500-hour closed-loop run
V1I-40
-------
(run 620). In each run the system was charged with liquors with composi-
tions based upon the expected steady-state conditions of each run.
Except for the type of reactor system used, the process configuration in
all of the closed-loop runs was essentially the same. Figure VII-15 shows
a schematic of the basic equipment arrangement.
The scrubber system consisted of a venturi followed by a tray tower con-
taining either two or three trays. Two trays were used in runs 601 and 602,
and three trays were used in run 620. The general operating parameters and
flue gas conditions are listed in Table VII-10.
Regenerated liquor was fed to the tray recycle tank which served as a hold
tank for liquor recirculated across the trays. A bleed from the recircu-
lating liquor was sent forward to the venturi recycle tank. The rate of
this bleed was set manually based upon pH of the venturi liquor (or S02
removal) and this rate set the overall feed forward through the integrated
system. , For the most part, the pH of the venturi recycle in all runs was
typically in the range of 5.3 to 6.0.
A bleed stream of spent liquor from the venturi recycle line was sent to
the air-sparged oxidation tank where most or all of the TOS was converted
to sulfate. The rate of withdrawal of the bleed stream was controlled on
the level of liquor in the venturi recycle tank.
The design and operation of the oxidation tank were based upon the results
of a number of open-loop tests. Air was introduced at the bottom of the
tank through a ring-shaped sparger at a rate equivalent to an oxygen:TOS
stoichiometry ranging from 5:1 to 25:1 depending upon the concentration
of TOS in the scrubber bleed. The pH of the oxidation tank was controlled
between 4.5 and 10 using clarified regenerated liquor in order to prevent
S02 off-gassing. The open-loop tests had shown that without at least
partial neutralization of the oxidation tank liquor, the pH of the liquor
would fall to below 4 causing appreciable S02 off-gassing equivalent to
as much as 15% of the TOS fed to the oxidation tank. Partial neutraliza-
tion in the oxidation tank, as shown in the open-loop testing of the re-
generation reactor system, has very little, if any, effect on the reactor
system performance and avoided the need for a sealed aeration tank vented
to the scrubber inlet.
Two different aeration systems were used. In runs 601 and 602 the aera-
tion tank had about a 30 minute holdup and pure oxygen was introduced
into the liquor feed line to the system to assist in oxidation. In
run 620 a tank with a two hour holdup time was used with no oxygen fed
into the liquor line.
The oxidized scrubber bleed was fed by gravity to the regeneration reactor
system where the liquor reacted with hydrated lime. The lime was metered
to the reactor in the form of a dry solid. The lime feed rate was adjusted
according to the pH of the reactor effluent in order to maintain the hydrox-
ide concentration in the regenerated liquor in the 0.08M to 0.12M OH range.
VII-41
-------
I
-e-
N5
Flue Gas
Venturi
Scrubber
Carbonate
Feed System
Venturi
Recycle
Tank
Ain
TOS
Oxidizer
Waste
Solids
Filtrate
l*~~lReceiver
FIGURE VII-15 PROCESS FLOW DIAGRAM FOR DILUTE LIME MODE PILOT PLANT
OPERATIONS (WITH TOS OXIDATION)
-------
TABLE VII-10
GENERAL OPERATING CONDITIONS FOR SCRUBBER SYSTEM
Inlet Flue Gas:
S02 Level, (ppm) 700-1050
Q£ Level, (ppm) 5-9.5
Dry-Bulb Temperature, (°F) 340-400
Dew Point, (°F) 120-130
Venturi Operating Parameters:
AP, (inches H20) 11-12
L/G, (gals./Macf sat'd) 10-20
Tray Tower Operating Parameters:
AP, (inches H20/tray) 1.4-1.8
L/G, (gals./Macf sat'd) 5-10
VII-43
-------
The configuration of the reactor system was the primary difference among
the closed-loop runs. In both runs 601 and 602 a simple CSTR with a two-
hour holdup time was used. The long reactor holdup time was to ensure
good utilization of lime and, hopefully, at the same time, allow suffi-
cient time for desupersaturation of calcium. Solids recycle was also
tested for a few hours at the end of run 602 using filter cake as the
source of the recycle solids.
Run 620, which lasted about four weeks, was broken into two parts, each
corresponding to the use of an entirely different reactor configuration.
In the first part, run 620A, a CSTR with a 90 minute holdup time was
used along with the recycle of solids from the thickener underflow. The
rate of recycle was varied to study the effects of suspended solids levels
on calcium desupersaturation. In run 620B the ADL reactor system with a
90 minute to total holdup time was used without solids recycle (first
stage holdup = 10 minutes; second stage holdup = 80 minutes).
The waste solids generated in the reactor were dewatered using a standard
thickener and rotary drum vacuum filter combination. The filter cake was
washed to recover most of the sodium value occluded with the cake. The
combined filtrate and wash water was returned to the thickener. A portion
of the clarified overflow from the thickener was drawn off for use in
neutralization of the oxidation tank. The remainder was returned to the
scrubber system either directly or through the softening reactor.
The soda ash makeup to the system was added to the softening reactor where
it reacted with the soluble calcium in the clarified liquor producing cal-
cium carbonate. The softening reactor was a simple CSTR with a holdup
time of roughly 30 minutes. After settling out the calcium carbonate
the softened liquor was passed to the tray recycle tank.
The overall effect of the soda ash softening step was to reduce the soluble
calcium concentration in the combined liquor returned to the scrubber system
by about 50 ppm. The amount of softening that could actually be accomplished
in any run was dependent upon the rate of sodium loss in the filter cake,
since this determined the rate of soda ash makeup.
b. Summary of System Performance
Table VII-11 summarizes the general operating conditions and overall system
performance in the closed-loop operations, and Figures VII-16 and VII-17
show the compositions and flows of the process streams in runs 602 and 620A.
In general, the system performed reasonably well as long as the level of
TOS in the system liquor was at or below 0.02M prior to absorbent regen-
eration. There were, however, a number of problems with scaling and scale
potential. These are discussed along with the various aspects of system
performance in the following sections.
VII-44
-------
M
M
I
*-
Ln
CLOSED-LOOP, DILUTE
Run No.
General Operating Conditions
Inlet Gas:
S02 (ppm, dry)
02 (vol%, dry)
Temperature (°F)
Scrubber Operation:
Absorber L/G (gals/Macf sat'd)
Top Tray Feed pH
Total Feed Stoichiometry (Avg)a
Number of Trays
Absorber AP (inches of H 0)
Scrubber Bleed Liquor:
pH
[TOS], (M)
[SO'], (M)
Regeneration System:
Reactor Type
Holdup , (min)
Extent of Regeneration, ([OH~],M)
Key Operating Results
S02 Removal (% of inlet)
Lime Feed Stoichiometry (mols Ca(OH)2/mol AS02)
Sodium Makeup (mols Na CO /mol ASO ) :
Actual
Required
System Oxidation Rates (% of ASOj) :
Scrubber
Oxidizer
Calcium Supersaturation in Reactor Effluent
Filter Cake Properties:
% Insoluble Solids
% Soluble Salts (dry basis)
CaSO. /CaSO
4 x
Scaling & Solids Deposition
LIME MODE RUNS GENERAL OPERATING CONDITIONS
601 602
1,000 900
5-6 9-9.5
375-400 340
5.3-6.1 5.3-5.8
0.05-0.08 0,03-0.05
0.6 0.5-0.55
CSTR CSTR
150 150
0.10 0.10
>90% 90-91%
1.1.0 1.04
0.05
0.02
20-25% (200ppm) 50%K400ppm)
25-30% (250ppm) 45%(^350ppm)
120
<30 37C
1.8
0.95
CaCO-j in
Tray Tower Loop
& OVERALL SYSTEM PERFORMANCE
620A
700-800
6.5-9 (typ.,7)
335-390
5.1-6.2
0.01-0.045 (typ.,0.04)
0.45-0.55
CSTR w/recycle
60-90
0.09-0.12
90-95%
1.05-1.15
0.017
0.01
60-65% (lAOOppm)
30-35%(-v-200ppm)
(50-250 at <4% solids (
|0 at >5% solids (
65-80
0.5-1.5
>0.98
j*CaC03 and CaS03 - 1/2H20 in Tray
i CaS04 - 2H20 in Reactor Piping a
( Weir
620B
'
750-850
6-7.5
375-400
5.1-5.8
0.02-0.05 (typ., 0.035)
0.4-0.5
ADL
80-90 * ~ 7I1a^
R2 = 72-80)
0.08-0.10
90-95%
1.05-1.15
0.015
0.01
40-45XCx.300ppm)
50-55%(^350ppm)
0-100
70-80
0.5-1.5
>0.98
Tower Loop I
nd Thickener f
3 Feed Stoichiometry AmolsNa capacity/mol inlet SO,.
Pressure drop includes trays, ducting and demister.
c Filter cake was mixture of solids produced in 601 and 602. Solids produced in
run 602 alone would have filtered to better than 50% -lolids.
-------
85 ppm SO2
i
Outlet
Gas
Inlet Gas
1,100scfm(wet)
920 ppm SO2
Scrubber
System
Flow
pH
[TOS]
M
.p-
= 13.251pm
= 5.3
= 0.043M
= 0.52M
= 0.013M
H20
4
1.1 Kprn
[OH"] = 0.096M
[TOS] = 0.009M
[S04=] = 0.50M
[Ca++] = 0.013M
Flow = 4.5 8pm
0.85 Epm
O2/min)
Dry Lime
90 gms/min
85% Ca(OH)2
(1.03 gm mols Ca(OH)2/min)
= 0.096M
= 0.012M
= 0.48M
= 0.0200M
82% CaS04
3% CaSO3
2%Ca(OH)2
8% CaC03
5% Incrts
\
D
cr
50 ml/min
[Na2C03l =1.013M
(0.051 gm mols Na2CO3/min)
CaC03 (inventory in settler)
[OHT] = 0.097M
[TOS] = 0.011M
[so:
0.50M
lCa++] = 0.0195M
0.5 8pm
FIGURE VII-16
STREAM COMPOSITIONS AND FLOWS FOR RUN 602
a 205 gms Dry Cake/min
38% Insoluble Solids
1.8% Soluble Solids
-------
60 ppm SO2
Outlet
Gas
Inlet Gas
~1,100scfm (Wet)
750-800 ppm
Scrubber
System
S02
Flow
pH
[TOS]
[SOj
10.412pm
5.6
0.045M
0.47M
<3
M
H20
1.55£pm
[TOS] = 0.105M
[TOS] = 0.003M
[SO] = 0.44M
= 0.007M
Flow = 2.65 8pm
15 ml/mi n
Flow = 8.70 8pm
Flow=* 3.4 Spm
[Na2CO3] = 0.95-1.06M
(0.015 gm/mols Na2CO3/min)
I »~
[OH~]
[TOS]
[S0=]
0.105M
0.0025M
0.44M
0.0145M
PH
[TOS]
[804]
7.7
0.004M
0.50M
0.55 £pm
Dry Lime
74 gms/min
88% Ca(OH)2
(0.88 gm mols Ca(OH)2/min)
FIGURE VII-17 STREAM COMPOSITIONS AND FLOWS IN RUN 620A
185gms Cake/min
65-75% Insoluble Solids
Dry Cake Comp:
87% CaS04
9% CaCOo
1.4%(Na2S04-
2.5% Inerts
2H20
-------
Scrubber System Performance
Over the range of inlet S02 levels covered in these runs (700-1,050 ppm)
there was no difficulty in achieving S02 removal efficiencies of 90% or
better. As would be expected, in run 620 (A and B) where three trays
were used, S02 removal efficiencies were slightly higher (and outlet
S02 levels correspondingly lower) than in runs 601 and 602 in which two
trays were used. The fact that only about 90% removal was achieved with
two trays and a venturi reflects the particular conditions under which
the pilot plant was operated. The saturated flue gas exiting the scrubber
was at a temperature of about 140-145°F, which is about 20-25°F higher
than would normally be experienced in a normal boiler application. The
higher operating temperature increases the vapor pressure of S02, making
removal more difficult. The lower flue gas temperature in a full-scale
system, plus the better stage efficiencies in a larger tray absorber
(due to better gas/liquid distribution), would result in somewhat better
S02 removal efficiencies than that observed in the pilot plant.
Except in run 601, oxygen levels in the flue gas were also set higher
than would normally be experienced in an efficient boiler operation.
The intent in these runs was to induce oxidation; the higher the level
of oxidation in the scrubber system, the lower the load on the oxidation
system. As shown in Table VII-11, the oxidation in the scrubber system
varied directly with the oxygen content of the flue gas — from an
equivalent oxidation of 200 ppm of S02 at 5-6% 02 in run 601, to about
400 ppm of S02 at 9-9.5% 02 in run 602. The low oxidation rates expe-
rienced in the scrubber system in run 601 actually caused operational
problems in the reactor system performance.
The major problems associated with the operation of the scrubber system
were the control of pH and prevention of or minimization of potential
for scale formation. These problems are closely related to the amount
of oxidation occurring. In runs 620A and 602, for example, there was
an appreciable amount of oxidation in the scrubber system (55-65% of
the S02 removed). As a result, there was very little buffering in the
tray recirculation liquor (very little TOS). In order to keep up the
S02 removal capacity, it was necessary to operate the system with a
neutral or slightly basic pH in the tray loop. Because of the lack of
buffering in this liquor, the pH of the tray feed frequently drifted up
to 11-12 for extended periods. This resulted in C02 absorption and pre-
cipitation of CaC03 in the tray loop, forming a scale throughout the tray
tower circuit. The amount of scale formed varied with the duration of
such excursions. Unfortunately, the formation of the scale was a cumula-
tive effect, since CaCOs dissolves very slowly at pH's above 6.5. This
scale forced shutdowns on two occasions for cleaning of piping, pumps,
and valves.
The scale formation on the trays is evidence of the severity of this problem.
Only the top tray and radial vane demister showed appreciable deposition of
CaC03. At the conclusion of run 620 there was a 1/2-3/4 inch scale of GaC03
(calcite with a small amount of aragonite) covering the entire surface of
VII-48
-------
^°P /%' SCale increased ^e tray tower pressure drop. The
middle and bottom trays remained relatively clean, except for the inlet
e
T Where S°me CaC°3 deposition occurred. Figures
and VII-19 show photographs of the top and middle trays, respec-
tively, at the conclusion of the run.
At slightly lower oxidation rates in the scrubber system in run 620B there
was more buffering capacity in the tray tower circulation loop due to the
higher levels of TOS. This allowed better control of the pH of the liquor
fed to the top tray. However, operation at a pH of 8-11 resulted in the
precipitation of calcium sulfite. Although no direct scaling was observed
after about 100 hours of operation, the presence of these solids indicates
a scale potential.
It is apparent from these results that there is a fine line of operating
conditions that should be maintained in order to prevent scaling in the
scrubber system. Although not tested, the most promising approach may be
to operate a dilute mode scrubber at a pH of about 7 and provide sufficient
gas/liquid contacting to ensure good S02 removal efficiencies. Such an
operation may still be difficult to control if there is enough oxidation
to reduce TOS levels in the scrubber system to below 0.02M.
Regeneration System Performance
Table VII-12 summarizes the performance of the various reactor systems
used in the closed-loop runs . In all runs , with all reactor configura-
tions, lime utilization exceeded 90%. In runs 601 and 602 with holdup
times of 2-3 hours, lime utilization was generally 95% or higher. In
run 620, with 90 minute holdup times, lime utilization was slightly
higher — typically about 93% of the available Ca(OH)2- The pH of the
reactor feed liquor in run 620 was also slightly higher.
The dewatering properties of the solids produced in the closed-loop runs
were also quite good as long as TOS levels in the reactor feed liquor
were maintained in the range of 0.02 or lower. At TOS levels above 0.03M,
as in run 601, the properties of the solids deteriorated markedly due to
high reactor operating pH and long holdup times used. In run 601 the total
oxidation across the combined scrubber system and oxidation tank amounted
to less than 50% of the S02 removal because of the low flue-gas oxygen
levels and insufficient aeration in the oxidation tank. The solids gen-
erated in the reactor system, therefore, were primarily calcium sulfite,
and previous work in the concentrated lime mode showed that a high reactor
pH ( > 10) and a long holdup time ( > 30 minutes) are unfavorable conditions
for producing good calcium sulfite solids with a CSTR.
The quality of the solids produced is indicated both by the settling rates
of the solids and the filter cake characteristics. In run 601 the cake
contained less than 30% insoluble solids and the solids virtually did not
settle. Under identical conditions in run 602, except that the TOS in
the feed liquor was reduced to less than 0.01M TOS, the quality of the
solids improved markedly, as evidenced by the settling data. However,
VII-49
-------
I
Ui
o
FIGURE VII-18 TOP TRAY AFTER RUN 620
-------
V
FIGURE VII-19 MIDDLE TRAY AFTER RUN 620
-------
TABLE VII-12
SUMMARY OF CLOSED-LOOP. REACTOR PERFORMANCE
Ul
Run
Reactor Parameters:
Type
Nominal Holdup
Solids Recycle
Reactor Feed Liquor:
PH
[TOS], (M)
[804], (M)
Reactor Effluent:
[OH-], (M)
tCa4*], (M)
[Ca++], (ppm)
[Ca] Supersaturation, (ppm)
% Solids
Solids Characteristics:
Initial Settling Rate, (ft/min.)
Settling Parameter, (tons/ft^ day)
Density of Settled Solids, (wt% solids)
Filter Cake~%Solids
601
CSTR
2-3 hrs
No
5.5
0.05
0.63
602A
CSTR
2-3 hrs
No
5.5-6.6
0.01
0.53
602B
CSTR
2 hrs
Yes
0.003
0.53
62 OA
CSTR
1.5 hrs
Yes
6-10
0.002
0.57
620B
ADL
1.5 hrs
No
7-10
0.003
0.51
0.05-0.10
0.018-0.024
720-960
20-250
0.8
0.095
0.0200
800
120
0.8
0.085
0.0170
680
0
3.5
0.115
0.0178
710
0
8.5
0.09
0.0186
745
50
0.9
Would not
settle
<30
0.36
0.15
15
—
0.05
0.12
20
-
0.14
0.7
55
65-80
0.3
0.14
30
65-80
Calcium Utilization
100
95-97
^93
^93
-------
since run 602 was a continuation of run 601, the filter cake was a mix-
ture of solids produced during both runs and, therefore, did not reflect
the better dewatering properties.
In both parts of run 620 (with the ADL reactor, and with the CSTR and
solids recycle) the solids properties were excellent. The filter cake
was consistently higher than 60% solids and frequently ran as high as
75-80% solids. An analysis of the solids produced showed that they were
essentially pure gypsum, as were the solids produced in run 602.
The need for solids recycle to prevent calcium supersaturation in the
reactor liquid (and scrubber system feed) was clearly indicated by these
closed-loop tests. The soluble calcium concentrations measured at the
different solids levels in runs 602B and 620A are plotted in Figure VII-20.
The results show that in order to achieve saturation (approximately 700 ppm
Ca ), the solids level in the reactor slurry must be greater than 4% solids.
Unfortunately, it was difficult in the open-loop runs previously discussed
to achieve greater than 4% solids without actually closing the loop. Thus,
reduction of calcium concentration to the saturation level was never
achieved except in the closed-loop tests.
The level of calcium observed using the multistage reactor without solids
recycle is also indicated in Figure VII-20 (run 620B). With the ADL
reactor, soluble calcium concentrations ran about 770 ppm, or about
50 ppm higher than saturation. This is considerably better than that
achieved using a CSTR without solids recycle.
As in the scrubber system, the major process problem in the reactor system
was scaling. Throughout both parts of run 620, gypsum scale built up in
the reactor bleed piping, and thickener overflow weir and piping. This
scaling forced numerous shutdowns to clean and flush piping and equipment.
The scale was probably due to supersaturation of calcium in the reactor
effluent. Even when the system was run with solids recycle to the reactor,
there were extended periods when the solids level in the reactor slurry
was below 4% due either to operation according to a prescribed set point
or due to inability to set and maintain the underflow recycle rate accu-
rately. It is possible that had a level of 5% solids or greater been
maintained in the reactor slurry, scale formation would have been minimal.
E. CONCLUSIONS
Use of limestone only for the regeneration of solutions in the dilute
mode (less than 0.15M active sodium) is not viable. As indicated in
Chapter VI, the limestone reaction rate decreases as the ratio of soluble
sulfate/sulfite increases in the reactor solutions. At sulfate/sulfite
ratios required for adequate sulfate precipitation in the dilute mode,
reaction rates are poor resulting in poor limestone utilization and poor
solids quality.
VII-53
-------
M
1,000
900
en
If 800
o
I
J
700
<3
^ 600
500
-~ 250 ppm Supersat'd
• Run 620A - CSTR, 90 min
O Run 602B - CSTR, 150 min
(JRun 620B - ADL, 90 min
• ~ 135 ppm Supersat'd
-~ 65 ppm Supersat'd
» Solids Recycle
I I I
20
40
J_
J_
I
I
I
60 80 100 120 140
Suspended Solids Level in Reactor Liquor, (gms/liter)
160
180
200
FIGURE VI 1-20
CALCIUM SUPERSATURATION VS SUSPENDED SOLIDS LEVELS
-------
Use of lime in combination with limestone in a dilute dual alkali mode
was more viable technically. In this approach, the lime regeneration
was carried out in a second reaction system to promote sulfate precipi-
tation. The limestone/lime process is more complicated than a simple
dilute lime process, resulting in higher projected capital cost. Eco-
nomic analysis indicated that operating cost savings which could poten-
tially be realized in using limestone for part of the regeneration would
not offset the additional capital cost probably required to enable use
of the limestone. The dilute lime system, using soda ash for softening,
was technically and economically the most viable dilute mode considered.
Conclusions based upon laboratory and pilot plant investigations of this
mode are given below.
A dilute lime mode can be operated in a closed loop with sulfate precipi-
tation keeping up with any level of system oxidation. The system can be
operated with high S02 removal (90% or higher) and good lime utilization (90%
or higher) to produce high quality solids (60% insoluble or higher) with
low soluble' sodium losses (2% achievable). The process may be more appro-
priate for low-sulfur coal applications or in situations where oxidation
rates are expected to exceed 25-30% of the S02 removal. The dilute lime
mode is somewhat more complicated than the concentrated lime mode, in-
volving higher liquid rates and larger reactors and associated equipment,
The process is also potentially less reliable than the concentrated lime
approach.
The regeneration reaction, carried out at low sulfite levels, results in
the precipitation of calcium sulfate (usually gypsum) to produce a re-
generated solution of sodium hydroxide and sodium sulfate with soluble
calcium levels which are, at best, at the saturation level of about
700 ppm Ca"1"*". Even with moderate amounts of soda ash makeup (and re-
sulting softening by precipitation of calcium carbonate) the solutions
have soluble calcium levels in the range of 600-700 ppm with a high
potential for scaling in the system. Close control of scrubber pH is
required to prevent carbonate or sulfite scaling. High scrubber oxida-
tion rates may create sulfate scaling.
In the dilute mode regeneration reaction, there is a high tendency to
produce solutions which are supersaturated in Ca"1"1" (with respect to
gypsum). Using a single-stage CSTR with no solids recycle, calcium
supersaturation levels of 100-200 ppm are easily encountered. Special
design precautions must be taken to prevent supersaturation and the re-
sulting scaling throughout the system. Supersaturation can be reduced
in a number of ways, by reactor system design and by controlling con-
ditions of the regeneration reaction:
• Increased reactor residence time — Allows time for completion of
reaction and desupersaturation. Holdup time of 60 minutes is a
minimum; 90 minutes is preferable.
e Solids recycle — Increases suspended solids concentration and seed
concentration for reaction and desupersaturation. Recycle of solids
VII-55
-------
to achieve a concentration of 4% or higher suspended calcium salt
solids is required to eliminate supersaturation in the reactor
effluent.
• Oxidation of sulfite in scrubber bleed prior to regeneration —
Lowers the concentration of TOS which tends to retard the lime/
sulfate reaction when TOS is present in the dilute mode concen-
tration range. Oxidation to TOS concentrations of about 0.02M or
lower is desirable.
e Multistage reactor configuration — Solids generated in a short
residence time first stage provide good seeds for completion of
reaction in longer residence second stage. Using a multistage
reactor can reduce supersaturation to within about 50 ppm of
the saturation level. Solids recycle is required to completely
eliminate supersaturation.
Elimination of supersaturation was achieved in the single-stage reactor,
with 90 minutes residence time, using solids recycle to the minimum of 4%
suspended calcium salts in the reactor, and with oxidation of the reactor
feed solution to TOS levels of 0.02M or lower. Variation in soluble sulfate
concentrations in the range of 0.5-0.75M had no apparent effect on the level
of supersaturation.
Utilizing these design factors in a dilute mode with lime regeneration not
only reduces or eliminates supersaturation, but also promotes a good reac-
tion rate which generally improves the overall process performance param-
eters such as lime utilization, sulfate precipitation and solids properties.
More specifically, the performance of the dilute lime mode relative to the
important process performance characteristics is given below:
e S02 removal — S02 removal of 90% is easily achieved especially at
low to medium inlet S02 levels. S02 removal is not as efficient
as in a concentrated dual alkali mode (with the same scrubber
configuration) because of the low active sodium concentration.
The scrubber operation is more difficult to control due to the
low buffering capacity of the dilute mode liquors. Higher calcium
concentrations (in the range of 600-700 ppm Ca"1"*") present potential
scaling problems in the scrubbing system. Operation of the scrubber
in a high pH range (9-11) to promote good S02 removal results in
some C02 absorption and potential carbonate scale formation. In-
creasing active sodium concentrations to provide more buffering
can result in sulfite scale formation in the pH range of 8-11.
• Lime utilization — Lime utilization of 90% or higher is achievable
when regenerating to hydroxide concentrations of about 0.1M with
solutions containing sulfate in the range of 0.5-0.75M and using
reactors with a minimum total holdup time of 60 minutes. Utiliza-
tion increases as the residence time and sulfate concentration
are increased. Solids recycle also helps increase lime utiliza-
tion. However, TOS levels in the feed to the reactor should be
VII-56
-------
0.02M or less (by deliberate oxidation if necessary) to prevent
retarding of the reaction rate by the sulfite.
Oxidation/sulfate control ~ Complete sulfate control is possible
in this mode of operation at any rate of oxidation in the system.
However, at very high scrubber oxidation rates, sulfite/bisulfite
buffering is minimal and scrubber pH control becomes difficult. All
other aspects of the process operation are improved by high oxida-
tion rates (i.e. , minimal TOS concentration in the feed to the re-
generation reactor). Deliberate oxidation should be used to maintain
TOS levels below about 0.02M. At sulfate concentration in the range
of 0.5-0.75M, calcium sulfate (usually gypsum) is produced instead
of a mixed calcium sulfite/calcium sulfate crystal when TOS is main-
tained at or below 0.02M. At this point, the calcium sulfate content
of the solids is no longer limited by the apparent maximum content
of 25-30% in the mixed crystal; 100% calcium sulfate can be produced.
Solids properties — It is possible to produce excellent quality
solids containing 60-80% insoluble solids. Good solids properties
are favored by the following conditions:
Low TOS in the reactor feed — less than 0.02M.
High sulfate in -.he reactor feed — 0.5-0.75M (high end
of range favored).
Solids recycle — improves solids quality but increases
the solids load and ultimately the size of the thickener.
Multistage reactor system — improves solids quality compared
to same total residence time in a single stage.
High reactor residence time — 80% insolubles solids can be
produced using a 90 minute residence time reactor.
Sodium losses — In any application, increasing the insoluble solids
content of the filter cake increases the effective number of displace-
ment washes for any given amount of wash water available. By producing
75% insolubles solids in a high-sulfur coal application, roughly five
displacement washes are available (as opposed to two and one-half dis-
placement washes at 50% solids) permitting more effective cake washing;
in low-sulfur coal applications even more wash water can be available.
Consequently, sufficient wash water should be available to reduce the
solubles content of the cake to under 2%, and down to the range of
0.5-1.5% solubles in low-sulfur coal applications. In such applica-
tions it may be possible to wash the filter cake to loss levels lower
than those corresponding to sodium carbonate makeup levels required
for softening of the regenerated liquor. A sodium carbonate makeup
rate of 2-2.5% of the S02 removal rate provides sufficient carbonate
to reduce the Ca++ concentration in the regenerated liquor by about
50 ppm, providing only minimum softening. Thus, sodium makeup (and
VII-57
-------
ultimately the losses in the cake) may be controlled by softening
requirements rather than by wash water availability or cake wash-
ability.
System operability/reliability — The dilute lime mode is inherently
less reliable and more difficult to control than the concentrated
lime mode. When appropriate care is taken to eliminate supersatura-
tion, the calcium levels in the regenerated solution are in the range
of 700 ppm. Only a minimum of softening is provided at low sodium
carbonate makeup levels. Potential for scaling exists in the reactor
system and associated auxiliaries and piping, and in the absorber.
Absorber operation is less effective and more difficult to control
than in the concentrated mode.
VII-58
-------
VIII. STUDIES OF THE PHYSICAL PROPERTIES
OF DUAL ALKAT.I PRODUCT SOT.TBfi
A. INTRODUCTION
In the preceding chapters of this report, several viable dual alkali oper-
ating modes were discussed. An important requirement for a dual alkali
process is that the solids which are produced have good dewatering prop-
erties so that a tight liquor loop can be maintained. If the product
solids can be consistently dewatered to a sufficient degree, it might
be possible to dispose of them in a reclaimable landfill instead of in a
conventional sludge pond. Consequently, as a first step in assessing
that possibility, a series of laboratory tests, which are normally used
to characterize the behavior of soils in fills and embankments, were
applied to samples of dual alkali product solids — one high in calcium
sulfite and the other predominantly gypsum. The behavior of a sample of
solids from a direct limestone slurry system was compared to the dual
alkali solids in some of the tests.
The principal physical characteristic studied was the behavior of compacted
product solids when a load was applied to them. The permeability of the
compacted material to water and the concentration of soluble substances
in samples of leachate were also determined. Because there has been a
good deal of discussion about the possible use of "chemical treatment"
to improve the physical properties of FGD sludges, e.g., to increase
their load-bearing strength and to decrease permeability, the changes
in the properties of dual alkali product solids when they are treated
by mixing them with fly ash and either lime or portland cement were
studied. Measurements of load-bearing strength, permeability, and leach-
ate composition were made to characterize the efficacy of such treatment.
B. CHEMICAL AND PHYSICAL CHARACTERISTICS
1. Chemical Composition
The results of chemical analyses of each of the three materials studied
are shown in Table VIII-1. The sample of high calcium sulfite dual alkali
product solids, which will hereafter be referred to as "dual alkali sulfite",
was obtained during July 1975 from the CEA/ADL dual alkali prototype system
at the Gulf Power, Scholz Steam Plant located in Sneads, Florida. It was
composed of about 75% by weight CaS03 • 1/2H20, 15% to 20% calcium sulfate,
about 3.5% inert material and essentially no fly ash. Examination of the
dual alkali sulfite material by X-ray diffraction (XRD) indicated that
only a trace of gypsum was present. It is quite likely that the trace
of gypsum detected by XRD resulted from oxidation of calcium sulfite during
handling and storage of the material. In a parallel XRD study conducted
for Southern Company Services by IU Conversion Systems (IUCS), no gypsum
could be detected in a sample of dual alkali sulfite obtained from the
same process.
viii-i
-------
TABLE VIII-1
Material
Dual Alkali Sulfite
Dual Alkali Gypsum
Direct Limestone Sulfite
M
M
M
I a
1
CHEMICAL COMPOSITION OF FGD PRODUCT
SOLIDS
Chemical Composition (mmols/e")
Wt
Acid
. Percent
Insolubles Ca"1"1" TOSa
3.5 7.44 5.12
nil n.a. <0.01
25 6.48 3.04
SOi* Na+ Mg"1"* Cl~ OH~
1.28 0.32 0.14 0.025 n.a.b
n.a. 0.13 n.a. n.a. 0.38
0.64 0:03 0.07 0.09 n.a.
Total Oxidizable Sulfur, calculated as sulfite.
n.a.; not analyzed.
-------
In the XRD studies of dual alkali sulfite materials carried out during
this program, the diffraction patterns of a number of solids composed
primarily of calcium sulfite, but also containing varying amounts of
calcium sulfate, were examined with a great deal of care. The sulfite/
sulfate mixtures contained no lines to indicate the presence of a cal-
cium/sulfur solid phase other than CaS03 • 1/2H20. That, plus the fact
that the^CaS(VCaS03 ratio was found to vary when the S0=/S0= concentra-
tion ratio in the process liquor was changed, suggested very strongly
that the CaSO^ was present as a solid solution within the CaS03 • 1/2H20
crystal. Lattice mismatch must have been small because no evidence of a
change in lattice parameters with CaSO^ content could be detected.
The "dual alkali gypsum" was essentially pure gypsum, CaSO^ • 2H20, on
the basis of both X-ray diffraction and wet chemical measurements. It
contained a small amount of unreacted lime, but no TOS could be detected.
The material was produced during run 620 in the ADL pilot plant in which
dilute mode operation, with intentional sulfite oxidation prior to regen-
eration, was studied.
The direct limestone FGD sludge, which will be called "direct limestone
sulfite", was a sample obtained on June 17, 1975, from the EPA test
facility at the Shawnee Steam Plant, Paducah, Kentucky. Calcium sulfite
and a small amount of unreacted limestone were the only solid phases
which could be detected in that material by X-ray diffraction. It con-
tained about 25 wt % acid insoluble material, the bulk of which was fly
ash which is not crystalline and is not detected by X-ray diffraction.
2. Crystalline Morphology
The three product solids chosen for characterization were first examined
under the scanning electron microscope (SEM) to observe the detailed
morphology of the individual solid particles. As can be seen in the SEM
photomicrographs reproduced in Figures VIII-1 and VIII-2, the two dual
alkali materials were very different physically. Both of the photomicro-
graphs were taken at a magnification of 950X. The dual alkali sulfite
shown in Figure VIII-1 was composed of agglomerates of small, thin needles
which are typical of the agglomerates which are found in high calcium
sulfite dual alkali solids that settle and filter well. Although the
ultimate crystallites which make up each agglomerate were quite small,
the agglomerates themselves were reasonably large — ranging from 10-20 ym
in diameter.
The dual alkali gypsum crystals shown in Figure VIII-2, on the other hand,
were large, well formed rectangular single crystals about 50 ym in length.
In addition to being long and quite wide, the dual alkali gypsum crystals
were 3-4 ym thick.
The direct limestone sulfite solids exhibited another distinctly different
morphology as the photomicrograph in Figure VIII-3 shows. The magnifica-
tion in that photomicrograph was 1000X, which is close enough to the mag-
nification used in the other two cases to permit direct comparisons. The
VIII-3
-------
FIGURE VIII-1 DUAL ALKALI CALCIUM SULFITE SOLIDS (950X)
VIII-4
-------
FIGURE VIII-2 DUAL ALKALI GYPSUM SOLIDS (950X)
VIII-5
-------
FIGURE VIII-3 DIRECT LIMESTONE CALCIUM SULFITE SOLIDS (1000X)
VIII-6
-------
direct limestone sulfite crystals appear to be quite regular rectangles
that are not agglomerated. Some were only 2-3 ym long, while others were
nearly as long as gypsum crystals. A key characteristic that set the direct
limestone sulfite crystals apart from the dual alkali solids was the fact
that they were individual crystals (not agglomerated) and were very flat
and thin ~ only a few tenths ym thick. The spheres in Figure VIII-3 were
particles of fly ash which were collected along with the S02 in the direct
limestone scrubber.
3. True and Apparent Densities of the Solids
As shown in the -diagram in Figure VIII-4, the solids are composed of as
many as three distinct phases — the solids, entrained liquor, and en-
trained air. The actual bulk density of the product solids as they emerge
from the process depends upon the proportion of each phase which is present
along with the true densities of the solid and liquid phases. The true den-
sities of the solids which made up the samples studied in this work were
measured with an air pycnometer and the results are shown in Table VIII-2.
Gypsum is less dense than CaS03 • 1/2H20; however, if it is dried at too
high a temperature, the gypsum (CaSO^ • 2H20) is converted to plaster of
Paris (CaS04 • 1/2H20), which has a higher density.
Using the definition of bulk density shown in Figure VIII-4, a typical
dual alkali sulfite filter cake composed of 60 wt % solids and 40 wt %
diluted liquor would have a bulk density of about 1.6 g/cc or about
100 Ib/cu ft. In addition to the definitions of true density, bulk
density, and dry bulk density, four other parameters which are useful
in characterizing FGD product solids are defined in Figure VIII-4.
Weight percent solids, the percentage of the total sample weight which
is due to the solid phase, is quite straightforward. However, percent
moisture content, which is a term used frequently by civil engineers
who work in soil mechanics, is computed in a manner which often appears
unusual to chemists and chemical engineers. Percent moisture content is
defined as the weight of the liquid phase divided by the weight of the
solid phase and expressed as a percent. Thus, for materials which have
a great deal of liquid phase present, it is possible to have percent mois-
ture contents of 500% or more. Throughout the discussion which follows,
we will use the term weight percent solids exclusively when discussing
the proportion of a material which is comprised of liquid and that which
is solid.
Also included in Figure VIII-4 are definitions of the void ratio, e, and
the degree of saturation, S. Both of these parameters are used when dis-
cussing the permeability of solid materials. The void ratio is the ratio
of the total void volume, the volume through which a permeating liquid or
gas can move, to the solid volume through which such movement cannot take
place. The degree of saturation is the percentage of the total void volume
which is, in fact, occupied by liquid phase.
VIII-7
-------
V = Liquid Volume
w
W = Liquid Weight
w
V = Air Volume
V = Solid Volume
s
W = Solid Weight
Total Void Volume
Wo
p = true density = s
V
s
,,, , . W + W
y = bulk density = s w
V + V + V
was
= dry bulk density =
W
V + V + V
was
W
Weight percent solids = s x 100%
W + W
w s
W
Percent moisture content = w x 100%
W
e = void ratio = w a
S = degree of saturation = Vw x 100%
V + V
w a
FIGURE VIII-4 ELEMENTS OF DUAL ALKALI SOLID WASTE
VIII-8
-------
TABLE VIII-2
TRUE DENSITIES OF FGD PRODUCT SOLIDS AND FLY ASH
,a
Sample True Density (g/cc)
Dual Alkali Sulfite 2.60
Dual Alkali Gypsum (83°C) 2.39
Dual Alkali "Gypsum" (125°C) 2.82
Direct Limestone Sulfite 2.65
Fly Ash 2.11
3A11 samples dried at 83°C prior to measurement except for the
one sample of gypsum which was dried at 125°C.
VIII-9
-------
C. COMPACTABILITY OF THE SOLIDS — MOISTURE/DENSITY
RELATIONSHIP
When soils are compacted with varying amounts of moisture present, it has
been found that as the moisture content increases from zero, the dry den-
sity of the compacted material (it is dried after compaction but before
determining the density) goes through a maximum. Furthermore, the load-
bearing strength of the compacted material is usually a maximum at the
maximum dry density. This occurs when small to moderate amounts of water
are present in the soil and the water is adsorbed as a film on the surface
of the individual soil particles. When the material is compacted the water
acts as a lubricant, allowing the solid particles to slide easily against
one another and to pack more densely. However, when more than the optimum
amount of water is present, the excess water occupies space that solid
particles could occupy if the water were not there and prevents the solid
materials from being compacted to the maximum density.
As a first approximation, if these solids are considered to be similar
to soils, then they too should exhibit greatest load-bearing strength
when they are compacted at the optimum moisture content required to
produce the greatest compacted dry density. Thus, it was deemed appro-
priate to study the moisture/density relationship for compaction of the
three product solids to determine the moisture content that would be
required to produce the maximum compacted density and then to compare
those density values with typical values for compacted soils.
1. Apparatus and Procedure
Because the density to which a material can be compacted depends on the
means of compaction, the key to obtaining meaningful moisture/density
relationship data is to compact the material using a standard, repro-
ducible technique. One standard method of compaction which is widely
used is called the "Standard Proctor Compaction" test. This test, which
was employed with a minor modification, is described in ASTM D 698-70,
Standard Methods of Test for Moisture/Density Relations of Soils.
The version of the Proctor Compaction used in this work differed from the
standard method primarily in the sample size used. Because the available
molds were smaller, the volume of solids used was about one-half that
described in the ASTM procedure. However, by making other appropriate
changes so that the compactive energy per unit volume was maintained at
the standard value of 12,400 ft-lb/cu ft., the results from this work
should be directly comparable to those obtained with the standard tests.
The sample mold used was a plastic cylinder (2.7 In. diam., 4.6 in. high)
fastened to the laboratory bench. A 2.8 Ib rammer was then positioned
12 in. above the solid surface in a guide sleeve. The weight was allowed
to fall freely 25 times, with the blows evenly distributed over the surface;
each sample was made up of three layers with the procedure described above
repeated on each layer. After compaction, the sample was weighed and the
moisture content determined.
-------
Preparation of the material to be compacted was an important part of the
test. Following the standard procedure used for soil studies the filter
cake was first oven-dried at about 83°C until the solid was completely dry.
An appropriate amount of dry material was then weighed out and water was
added to obtain a particular desired nominal moisture content. The soil-
water mixture was then allowed to stand in a covered container overnight
to ensure that the entire mixture was homogeneous. By varying the water
added, it was possible to obtain a complete curve of dry density versus
percent moisture content. This method of sample preparation was used in
all physical testing.
2. Results and Discussion
The results of the Proctor compaction tests which were carried out on
samples of the three product solids are shown in Figure VIII-5. For both
the dual alkali sulfite and the dual alkali gypsum materials, the dry
density of the compacted sample passed through a maximum when the mate-
rial being compacted contained about 75 wt % solids. The position of
the maximum for the dual alkali sulfite was quite well defined. How-
ever, with only three data points available for the dual alkali gypsum
material, it was not possible to define the maximum with the same degree
of certainty. It is possible that the actual maximum lay at a slightly
higher weight percent solids and the corresponding dry density might have
been somewhat higher than the 77 Ibs/cu ft. that was measured at 75 wt %
solids. However, regardless of this uncertainty, the dual alkali gypsum
could be compacted to a significantly higher density than could the dual
alkali sulfite, even though the true density of the dual alkali gypsum
crystals was about 8% less than that of the dual alkali sulfite solids.
The lower compacted density observed for the dual alkali sulfite is a
likely result of the significant difference in morphology of the two
materials (compare Figures VIII-1 and VIII-2). The agglomerates of
small crystallites that comprise the dual alkali sulfite solids con-
tain a good deal of void space which is responsible for the good de-
waterability and washability of the material. Unless the void space
was eliminated by fracturing the agglomerate during compaction, however,
it would lead to a lower compacted density. The dual alkali gypsum
crystals, on the other hand, are quite regular and could align them-
selves into a rather dense mass upon compaction. For comparison, one
sample of the direct limestone sulfite material was compacted at 75 wt %
solids. With only one data point, it is obviously impossible to deter-
mine if the density of 79 Ibs/cu ft. was the maximum dry density which
could have been attained.
For reference, a silty clay soil can typically be compacted to a maximum
density of about 100 Ibs/cu ft. if compaction is carried out at about
83% solids. Although compactability alone is not a sufficient criterion
for judging disposal acceptability, the FGD product solids were compactable
to dry densities not drastically different from silty clay, and the latter
is quite amenable to inclusion in landfill operations.
vin-ii
-------
Legend:
Dual Alkali Sulfite
No Fly Ash
75
70
. - 43-50% Fly Ash
12-25% Fly Ash
Pilot Plant Gypsum
No Fly Ash
Direct Limestone
Sulfite
65
M
M
&
60
55
50 -
— 25
60
65
70
75
Weight % Solids
80
85
90
FIGURE VIII-5 COMPACTION OF DUAL ALKALI PRODUCT SOLIDS
-------
For the dual alkali sulfite material, some additional dewatering over and
above that which can be produced by vacuum filtration would be necessary
to reach the moisture content where maximum compacted dry density can be
achieved. One means of increasing the solids content above that which
can be produced by filtration is to add dry fly ash to the filtered mate-
rial. The effect of fly ash addition on the moisture/density relationship
for the dual alkali sulfite sample was studied at two levels of added fly
ash and the resulting moisture/density curves are shown by the broken
lines in Figure VIII-5. For mixtures composed of dual alkali sulfite
and fly ash, the weight percent fly ash present in the sample (dry basis)
is indicated by the circled number next to each data point.
The addition of fly ash had two distinct effects on the moisture/density
relationship. It shifted the optimum weight percent solids to a lower
value and it resulted in a decrease in the maximum compacted dry density
which could be achieved. The reduction in compacted dry density was not
completely unexpected, since fly ash is considerably less dense than dual
alkali sulfite — 2.11 g/cc versus 2.60 g/cc. In fact the maximum dry
densities measured for the mixtures containing 25 wt % (dry basis) and
50 wt % (dry basis) fly ash are very close to the densities one would
predict by adjusting the maximum dry density of solids without fly ash,
for the amount of fly ash added.
Although the preceding studies showed that if they were dewatered suffi-
ciently well, dual alkali solids could be compacted, an important question
remained. What might happen if solids were dumped and compacted and then,
at a later time, saturated with water? Might they reslurry and become
plastic? Table VIII-3 shows the changes which occurred when three samples
of dual alkali sulfite solids were stored under water for two weeks after
being saturated by evacuating air and pulling water through.
The compacted sample was prepared by the Proctor method and the molded
sample was prepared by randomly pushing a glass rod into the solid to re-
lease trapped air; the uncompacted sample was merely placed in a sample
cylinder and left untouched. The compacted and rodded samples showed no
observable change in total volume after saturation and soaking, indicating
that the void ratio remained constant. However, there was a noticeable in-
crease in the degree of saturation as more air voids were filled with water
during the saturation process and soaking period. Of most significance was
the observation that the compacted sample, which had a hard surface prior
to soaking, did not swell or reslurry during the period of soaking.
D. RESISTANCE OF THE SOLIDS TO PHYSICAL PENETRATION
One result of a search for test procedures to characterize the load-bearing
strengths of the solids, particularly after they were compacted, was a
technique called penetrometry. In that technique, one measures the load
required to force a blunt probe of a relatively small diameter into a test
specimen at a constant rate.
•VIII-13
-------
TABLE VIII-3
WATER STABILITY TESTS ON DUAL ALKALI SULFITE
i
h-»
-P-
Sample
Compacted
Rodded
Uncompacted
Final Dry Density
After Soaking
(Ibs/cu. ff.)
65
47
46
Volume Change
Degree of Saturation3
Initial Final
-25
69
42
41
85
71
70
Appearance
After Soaking
Hard dry surface, no
observable change in
characteristics.
Firm moist surface, not
pourable, sample slid from
mold intact.
Soft wet surface, fairly
dense, not pourable, but
surface deforms upon
changing position.
a See Figure VIII-4.
-------
1. Apparatus and Procedure
Penetration tests were carried out according to ASTM method D 1558-71,
Standard Method of Test for Moisture-Penetration Resistance Relations
of Fine-Grained Soils. The tests were carried out using an Instron
Universal Testing Machine (Model TTD), equipped with a needle having
a diameter of 5/16 in. The samples to be tested were contained in
cylindrical plastic molds 2 5/8 in. in diameter. The experiments were
performed at two constant penetration rates: (1) 20 in./min — a rela-
tively rapid rate similar to that used in field measurements of penetra-
tion resistance, and (2) 2 in./min — a slower rate to simulate more closely
a static loading.
2. Results and Discussion
The general shape of the load-penetration curves obtained is shown in
Figure VIII-6. The yield load is the point at which the needle requires
maximum force to overcome the initial solid resistance. It is the yield
load which is of value in comparing the resistance of materials to pene-
tration. After yield, the load drops off until it reaches a minimum termed
the compaction load, and then, due to the limited height of the samples,
there is an increase again in load with further penetration.
Loads and penetrations at yield for five differently treated samples
of dual alkali sulfite solids are shown in Table VIII-4. As would be
expected, the yield load was greater at the higher penetration rate.
Samples 1 and 2, which had been compacted by the Proctor method, had
significantly greater penetration resistances than the three uncom-
pacted samples. The lubricating effect of water produced the lower
yield loads for sample 2 compared to sample 1, even though they had
similar dry densities. Comparison of sample 1 with samples 3 and 4 —
all at about the same saturation — shows the effect of dry density.
The higher density produced by compaction resulted in a substantial
increase in yield load. Sample 5, which was uncompacted, unsaturated,
as-received material, had an anomalous load/penetration curve. No
maxima or minima in the load curve were observed; rather, load in-
creased monotonically with penetration distance.
The yield loads shown in Table VIII-4 are plotted as a function of weight
percent solids in Figure VIII-7. For comparison, the dry densities after
standard Proctor compaction were taken from Figure VIII-5 and included in
Fijure VIII-7. At less than the optimum percent solids, both dry density
Sf yield load increase with decreasing moisture content. Shown in Figure
mi- are stilar curves from the literature for a clay soil reversed
since the abscissa is moisture content). (Kef. ™~2' > ™* ^207)
similar in shape, but the optimum moisture content for the clayJ* *«£
VIII-15
-------
E
A
°
B D
Penetration (in.)
Legend:
A = Yield Load
B = Yield Penetration
C = Compaction Load
D = Compaction Penetration
E = Load at 3 in. Penetration
3 in.
FIGURE VIII-6
LOAD VS. PENETRATION OF DUAL ALKALI
SULFITE SOLIDS
VIII-16
-------
TABLE VIII-4 PENETRATION TESTS OF DUAL ALKALI SULFITE SOLIDS
<
M
1^ |
M
1
h-1
*^l
Degree of
Sample Saturation %
1. Compacted to 57
64 Ibs/cu.ft.
Unsaturated
2. Compacted to 35
65 Ibs/cu.ft.
Saturated
3. Rodded to 71
47 Ibs/cu.ft.
Saturated
4. Uncompacted 70
46 Ibs/cu.ft.
Saturated
5. Uncompacted 33
37 Ibs/cu.ft.
Unsaturated
Speed of Pene-
tration (in/min.)
20
2
20
2
20
2
20
2
20
2
Yield Load
(psi)
860
717
535
456
70
33
52
33
c
ND
ND°
Yield
(in.)
1.50~
1.25
1.00
1.00
0.25
0.25
0.10
0.10
ND
ND
Penetration
(% Sample Height)
35
29
24
24
6
6
2
2
»
~
Sample height was approximately 4.25".
Penetration needle diameter, 5/16"; area, 0.0767 sq. in.
A maximum in load indicating the point of yield was not detectable.
-------
75
70
65
M
M
M
I
I-1
oo
•s 60
c
V
Q
55
50 -
60
Penetration Yield
Load at 20 in./min.
65
Dry Density
After
Proctor
Compaction
1000
800
600
CO
O
400
200
70
Weight % Solids
75
80
85
FIGURE VIII-7 PENETRATION RESISTANCE OF DUAL ALKALI SULFITE
SOLIDS
-------
Weight % Solids
88 82
11U
106
4-5
S-
4 102
£
1
>•
Q
94
90
/
Ai
0
Pen
V /
\jf
\
\
fi
1
""" Dryd
t
etration resi
^S
\
\
\
\
\
ensity
stance
\
\
X*-^
-» _-* J>
o •&• co ro "ro c
8888?
Penetration Resistance — lb./sq. in.
14 18 22
Moisture Content — %
26
FIGURE VIII-8 RELATIONSHIP BETWEEN PENETRATION RESISTANCE OF A PROCTOR
NEEDLE AND MOISTURE CONTENT OF A CLAY SOIL COMPARED WITH
DRY DENSITY/MOISTURE CONTENT CURVE
VIII-19
-------
penetration. The compacted dual alkali sulfite sample had a yield load
of about 800 psi at a weight percent solids less than the optimum, and
the yield load seemed to be increasing in the same fashion as the one in
Figure VIII-8. Thus, from standpoint of penetration resistance, the com-
pacted dual alkali sulfite material performed as well as, or better than,
the clay.
E. UNCONFINED COMPRESSIVE STRENGTHS OF COMPACTED
DUAL ALKALI SOLIDS
Although the technique is usually reserved for "stronger" materials like
concrete, unconfined compression tests can be performed on soil-like mate-
rials. Triaxial compression tests are more appropriate for studying soil-
like materials; however, access to the necessary equipment was not available
within the time and budget constraints of this program. Since the apparatus
for unconfined compression tests was readily available, measurements on com-
pacted dual alkali sulfite and dual alkali gypsum were carried out.
1. Apparatus and Procedure
Samples were prepared for unconfined compression tests by compacting them
in cylindrical molds by the Proctor compaction technique described earlier
(Section C-l). They were compacted at approximately the optimum weight
percent solids at which maximum dry density could be achieved.
After they were compacted, the molds were removed and the samples tested on
an Instron Testing Machine (Model TTD) according to ASTM Method D 1633-63. The
samples were compressed at a constant speed of 0.04 in./min until they fractured.
2. Results and Discussion
The unconfined compressive strength of a sample is the stress (force/unit
area) which is being exerted at the onset of failure. Since the compacted
dual alkali solids were not hard or brittle, failure did not occur cata-
strophically; rather, failure was observed as a slow decrease in stress
at a constant loading rate. The decrease in stress was accompanied by
cracking and subsequent fracture of portions of the specimen.
The stress at failure observed for the two dual alkali solids is shown
in Table VIII-5. The unconfined compressive strengths of 10 psi and
15 psi are quite low in comparison to those of "strong" materials like
concrete which have unconfined compressive strengths in the range of
thousands of psi. However, an unconfined compressive strength of
10-15 psi compared quite closely to that of many natural solids.
F. PERMEABILITIES OF SOLIDS
If these solids are to be disposed of in a landfill operation, the permea-
bility of the disposed mass to groundwater or surface water can be an im-
portant factor in the impact which the landfill operation has on the
VIII-20
-------
TABLE VII1-5
M
M
S3
Sample
Dual Alkali Sulfite
Dual Alkali Gypsum
UNCONFINED COMPRESSIVE STRENGTHS OF COMPACTED DUAL ALKALI SOLIDS
At Failure
% Solids
:ite 73.7
mm 75.6
Dry Density
(Ibs/cu ft.)
67.6
77.3
Test Specimen
Stress
Diam. (in.) Height (in.) (psi)
2.69 5.25 15
2.69 5.63 10
Strain
(in.)
0.019
0.045
-------
groundwater in the immediate vicinity. If the permeability of the mate-
rial being disposed can be reduced, then although the concentration of
soluble substances in the leachate leaving the disposal area may not have
changed, the rate of pollutant discharge would be reduced in proportion
to the reduction in permeability.
1. Apparatus and Procedure
The permeabilities of FGD product solids were measured using a constant-
head permeability apparatus as described in ASTM D 2434-68, Standard Method
of Test for Permeability of Granular Soils (Constant Head). All of the
samples which were studied were first compacted in cylindrical plastic
molds by the standard Proctor method discussed previously. Then tight-
fitting end caps with tubulations to which plastic tubing could be attached
were sealed to both ends of the molds. Next, the entrained air was evac-
uated from the samples, and they were saturated with de-aerated, distilled
water. Finally, a constant-head water supply was connected to the sample
cell and the volume of water which flowed from the bottom of the cell was
measured as a function of time.
The coefficient of permeability, k, in units of cm/sec, was then calcu-
lated from the formula
k = -SSL
thA
where Q = total volume of water measured
L = length of sample
t = elapsed time for volume measurement
h = total head of water
A = cross-sectional area of sample.
2. Results and Discussion
The coefficients of permeability measured for two samples of dual alkali
sulfite and one sample each of dual alkali gypsum and direct limestone
sulfite are given in Table VIII-6. The dual alkali solids ranged in
permeability from 0.2 x W~k to 3.0 x I0~k cm/sec. The dual alkali
gypsum had the lowest permeability. The five-fold difference between
the permeabilities of the two dual alkali sulfite samples was probably
due to the combined effects of a lower compacted dry density (more void
volume through which water can pass) and a greater degree of saturation
(relatively more of the void volume actually participating in the permea-
tion process) in the second sample.
The range of measured coefficients of permeability for the dual alkali
solids is similar to that exhibited by relatively pervious silts. Coarse
grained sands would be more permeable. Clays would have lower coeffi-
cients of permeability. Some very impervious clays have coefficients
of permeability in the range of 10~7 to 10~8 cm/sec.
VIII-22
-------
TABLE VIII-6
PERMEABILITIES OF COMPACTED FGD SLUDGE SOLIDS
Material
Dual Alkali Sulfite
Dual Alkali Sulfite
Dual Alkali Gypsum
Direct Limestone Sulfite
H
H
M
1
ro
OJ
Compacted Dry
Density
% Solids (Ibs/cu.ft.)
70.0 66.0
76.7 61.4
72.0 72.0
75.0 79.6
Degree of
Saturation (%)
70
79
74
95
Coefficient
of Permeability
(x 10^ cm/sec)
0.56
2.96
0.21
(n.d.)3
A valid permeability could not be obtained due to permeameter blockage; see text.
-------
A valid measure of the permeability of the direct limestone sulfite sample
could not be obtained. When the sample was first installed in the perme-
ameter and saturated with water, it was firm and the flow of water from
the permeameter corresponded to a permeability of 0.3 x 10" cm/sec. How-
ever, after it had been in the permeameter for a short time, it showed
signs of reslurrying, became quite fluid, and the apparent permeability
fell to 0.06 x lO"4 cm/sec. That behavior suggested that at the very
bottom of the permeation cell, a few layers of the very thin, flat, direct
limestone sulfite platelets had oriented themselves one on top of the
other, effectively blocking the flow. It is likely that a phenomenon
such as that is responsible for the fact that sludges consisting of in-
dividual platelets are often very difficult to dewater by vacuum filtra-
tion. Certain types of clay which are also composed of very thin, flat
platelets are known to be equally difficult to dewater by filtration.
G. LEACHING OF SOLUBLES FROM DUAL ALKALI SOLIDS
Several relatively simple experiments were carried out to study the way
the major cations, sodium and calcium, leached from compacted samples
of dual alkali sulfite and dual alkali gypsum. The aim of the experi-
ments was to obtain some initial data about the leaching behavior of
untreated materials, both for its own sake and for use as baseline data
to which the leaching behavior of treated materials could be compared.
1. Apparatus and Procedure
Leaching tests were carried out using compacted samples contained in the
same type of cylindrical plastic molds fitted with end caps which were
used for the determinations of permeability. De-aerated, distilled water
was passed through the solids in the test cylinders and fractions of the
leachate were collected for analyses of sodium and calcium by atomic ab-
sorption spectrometry.
In one extended test, which was allowed to proceed for a period of 80 hours,
the flow of water through the cylinder was controlled at about 0.25 ml/min.
The total amount of leachate collected during that period corresponded to
about 4.4 times the total void volume of the solids in the test cylinder.
Two accelerated leaching tests were carried out at considerably higher
flow rates of about 2.7 ml/min for comparison. Leachate equivalent to
about three void volumes was collected during each of those tests.
2. Results and Discussion
The concentrations of sodium and calcium in each of the seven leachate
fractions which were collected during the extended leach test of dual
alkali sulfite are shown in Table VIII-7. Included in the table are
estimates of the total dissolved solids (TDS) concentrations, which were
arrived at by converting the measured concentrations of sodium and calcium
to equivalent concentrations of sulfate salts and summing them. The maximum
sodium concentration was found in the first fraction collected, which corre-
sponded to the displacement of about 0.4 void volumes of leachate. In the
VIII-24
-------
TABLE VIII-7
to
LEACHING OF SODIUM AND CALCIUM FROM
Material
Dual Alkali
Sulfite
Dual Alkali
Sulfite
Dual Alkali
Gypsum
Eluate
Fraction
1
2
3
4
5
6
7
1
2
1
2
3
UNTREATED DUAL
Volume of
Fraction (ml)
114
248
144
276
142
-198
112
396
312
131
362
316
ALKALI PRODUCT SOLIDS
Cum. Void
Volume
0.41
1.30
1.82
2.81
3.32
4.03
4.43
1.56
2.78
0.51
1.93
3.17
[Na ] ppm
2,748
2,200
950
540
330
198
120
3,100
600
520
580
54
[Ca ] ppm
496
596
552
592
604
604
604
520
598
1126
1155
1293
IDS ,
Estimated (ppm)
10,100
8,820
4,810
3,680
4,700
2,660
2,420
11,300
3,890
5,430
5,720
4,560
a Total Dissolved Solids concentration; calculated as the sum of the sodium and
calcium concentrations expressed as sulfate salts.
-------
second fraction, which included the remainder of the first void volume
displacement and an additional 0.3 of the second displacement, the sodium
concentration had fallen to 2,200 ppm. A continued rapid drop in sodium
concentration was observed in succeeding fractions. The calcium con-
centration remained more or less constant at about 600 ppm.
The fact that the calcium concentrations in all of the eluate fractions
remained more or less constant while sodium decreased by more than a
factor of 20 is probably the result of the presence of a small amount of
gypsum in the dual alkali sulfite material which was tested. If one
assumes for the sake of simplicity that all of the sodium and calcium
in fraction 1 were present as sulfate, the ionic strength of that solu-
tion would have been about 0.23M, and the apparent solubility product
for calcium sulfate, Ksp' (see Chapter IV), would have been about 9 x 10"1*.
(The product of the calcium and estimated sulfate concentrations in mols/1.)
This apparent solubility product agrees with the value predicted for gypsum
in a solution of that ionic strength. The concentration of sodium de-
creased as the sodium salts, primarily Na2SOit, were washed from the sample
cylinder. The calcium concentration remained essentially constant, how-
ever, because of two counter-balancing changes which occurred. In frac-
tion 7, which had a significantly lower ionic strength (about 0.07M), the
apparent solubility product for calcium sulfate had fallen to about 2 x 10"1*,
which is not unreasonable. But, since the concentration of sulfate ion had
fallen similarly, the calcium level remained relatively constant in accor-
dance with KSp'.
From the standpoint of the pollution potential of the dual alkali sulfite
leachate, if one again makes the simplifying assumption that all of the
salts were present as sulfates, the total dissolved solids (TDS) in frac-
tion 1 would have been about 10,000 ppm. The leachate at the end of the
experiment would have contained about 2,400 ppm TDS. One would expect
the concentration of TDS in the leachate to remain constant at about
2,000 ppm as long as gypsum was present in the solid phase. With no
gypsum present, and under strictly anaerobic conditions, one would ex-
pect the concentration of TDS leachate from calcium sulfite to be a great
deal lower — in the range of 100-200 ppm.
The expectation that TDS levels would be considerably lower when a
distinct gypsum phase was not present in the sludge was confirmed by
the results of the parallel studies conducted for Southern Company
Services by IUCS2. When the sample of dual alkali sulfite studied
in that work, a material shown to contain CaSO but no gypsum, was
subjected to five successive batch washes (1 g dry solids per 4 g water),
the TDS level in the final wash was 682 ppm. Since it was not stated
that those washes were carried out under strictly anaerobic conditions,
it is likely that TDS would have been even lower if all oxygen had been
excluded.
The dual alkali sulfite solids studied in the extended leaching experi-
ment contained 0.32mmol/g of sodium. After the extended leaching ex-
periment had been terminated, samples of the leached solids were analyzed
VIII-26
-------
and were found to have a sodium content of O.llmmol/g, suggesting that
only about two-thirds of the total sodium had been leached? When the
total sodium measured in each of the seven fractions collected during
the leaching experiment was totaled, about 90% of the 0.21 mmol/g of
sodium leached, based on the solids analyses, could be accounted for
zn the leachate.
The extended leaching experiment suggested that a portion of the total
sodium in the dual alkali sulfite solids was very difficult to leach.
This phenomenon was examined further by performing three batch leaches
in which 5-gram samples of dual alkali sulfite solids were stirred in
three, 100-ml portions of water for 30 minutes, 60 minutes, and 120
minutes, respectively. After stirring, a portion of each slurry was
centrifuged and the supernatant liquid was analyzed for sodium. The
concentration of sodium in the supernatant solutions was independent
of stirring time. In each case the amount of sodium found in solution
corresponded to 0.19 mmol/g of sodium leached from the calcium sulfite
solids. By difference, the leached solids should have contained 0.13 mmol/g
sodium, a concentration which agreed quite closely with the 0.11 mmol/g resid-
ual sodium found in the solids after the extended column leach test.
The sodium balance observed in both the column and batch leaching experi-
ments indicated that about 30% of the total sodium in the solids was not
available for leaching. The most plausible explanation for the unavail-
ability of that sodium is that it was either occluded, or in true solution,
within the calcium sulfite/sulfate crystals.
An accelerated leach of dual alkali sulfite in which the eluting water
was passed through the column at a ten-fold higher rate than in the ex-
tended leach was carried out for comparison. The leachate from that
test was collected in two fractions, and the concentrations of sodium
and calcium measured in each are included in Table VIII-7. The behavior
of the dual alkali sulfite was not appreciably different in the acceler-
ated leach than in the extended test. The sodium concentration in the
first fraction was somewhat higher than had been measured in the initial
fraction which was collected during the extended leach, but the concen-
tration in the second fraction was comparable to those measured in
fractions 3 and 4 of the extended leach, which corresponded to about
the same void volume increment. The fact that the concentrations found
in the accelerated test were not substantially lower indicate that at
the higher flow rate, leaching was not rate limited.
A sample of dual alkali gypsum was also subjected to accelerated leaching
at a flow of about 2.7 ml/min. Analyses of the three fractions of leach-
ate which were collected during the test are included in Table VIII-7.
The sodium concentrations in the first fractions collected were lower
for the dual alkali gypsum than for the dual alkali sulfite, because the
gypsum solids initially contained less total sodium (0.13 mmol/g) than
did the dual alkali sulfite (0.32 mmol/g). The difference in total sodium
content was due to the fact that the gypsum was produced during "dilute
mode" operation where total sodium concentrations in the process liquor
were lower and more cake washing applied than used for the production
VIH-27
-------
of the dual alkali sulfite. An examination of the total amount of sodium
collected in the leachate from the gypsum leach test indicated that only
about half of the 0.13 mmol/g total amount of sodium initially present
had been leached even though the sodium concentration was very low in
the last fraction collected. Again, this suggests that sodium might
have been entrapped within the gypsum crystals too.
H. EXPLORATORY STUDIES OF THE EFFECTS OF CHEMICAL TREATMENT
ON THE PROPERTIES OF DUAL ALKALI SOLIDS
Chemical treatment by addition of lime and fly ash, or one or another
proprietary ingredient, has been proposed as a means for improving the
physical and chemical properties of FGD sludges to make them better
suited for disposal in landfills or in ponds which can ultimately be
covered and reclaimed. Most chemical treatment has involved sludges
produced by direct slurry scrubbing systems. Since the effectiveness
of treatment seems to depend upon arriving at the optimum technique for
a particular sludge, it was decided to perform a few initial exploratory
studies to determine the improvement, if any, that chemical treatment
might impart to the physical and chemical properties of dual alkali solids.
1. Experimental Procedures
The treatment techniques studied involved the addition of fly ash and a
source of lime to the solids to be treated. The silica in the fly ash
undergoes the well-known pozzolonic reaction with calcium from the lime
to produce an insoluble, cementitious calcium silicate matrix which has
been shown in other studies to increase the strength of the solids and
decrease their permeability. The source of lime was either quicklime
(CaO) or portland cement; the latter is a source of both lime and sili-
cate. Treatment consisted of mixing samples of dual alkali sulfite or
gypsum with varying proportions of fly ash and CaO or portland cement.
A few of the treated mixtures were prepared in an excess of water, poured
into molds, allowed to settle, and then left to cure under water. Other
mixtures were prepared by mixing the three ingredients in the presence
of the required amount of moisture for optimum Proctor compaction as
described previously. After compaction, the samples were removed from
their molds, placed in a 100% relative humidity chamber at room tempera-
ture, and allowed to cure for either 14 or 28 days.
The primary measure of treatment effectiveness was the increase in un-
confined compressive strength imparted to the samples by the various
treatments. Single samples of treated dual alkali sulfite were also
subjected to permeability and leaching tests.
2. Results and Discussion
The unconfined compressive strengths of the various treated dual alkali
solids are shown, along with the kinds and amounts of additives used
for each, in Table VIII-8. As noted earlier, the actual compressive
strength of the sample is the stress at the onset of failure; the strain
VIII-28
-------
TABLE VIII-8
COMPRESSIVE STRENGTH OF DUAL
(CURED AT 100%
Sample Composition Prior to
Sample
1
2C
3
4
5
6
< 7
M '
H 8
N>
VO
9
10
11
12
Source of
Solids
Scholz
Scholz
Scholz
Scholz
Scholz
Scholz
Scholz
PP-620
(gypsum)
Scholz
Scholz
Scholz
PP-620
(gypsum)
a Samples 1 and 2
Weight %
Fly Ash
68.9
47.6
47.5
42.5
23.8
21.2
0.0
0.0
40.0
40.0
40.0
40.0
Solids" CaO
24.2 6.9
33.3 19.1
47.5 5.0
42.5 15.0
71.2 5.0
63.8 15,0
100.0 0.0
100.0 0.0
40.0 20. Oe
40.0 20. Oe
40.0 20. Oe
40.0 20. Oe
%Solids
__
—
73.0
74.4
66.3
69.7
73.7
75.6
72.0
72.0
61.1
70.0
ALKALI SOLIDS
RELATIVE HUMIDITY)3
Curing Test Specimen
Dry Density Diam. Height
(Ibs/cu ft.) (in.) (in.)
50. 8d
69. ld
59.3
60.6
63.1
66.8
67.6
77.3
61.5
62.6
58. 3f
68.0
were mixed with an excess of water, poured
2.88
2.38
2.69
2.69
2.69
2.69
2.69
2.69
2.69
2.69
2.69
2.69
into a mold
2.50
2.00
5.75
5.63
5.75
5.75
5.25
5.63
5.25
5.38
5.31
Curing
Time
(Days)
28
28
14
14
14
14
14
14
14
28
28
Unconfined
Compressive
Stress Strain
(psi)
80
199
19
56
27
39
15
10
97
213
165
5.50 28 244
, allowed to settle,
0.048
0.067
0.025
0.024
0.042
0.017
0.019
0.045
0.014
0.021
0.022
0.009
Secant Modulus
of Elasticity
at Compressive
Strength
(psi)
1,665
2,970
760
2,335
640
2,295
790
220
6,930
10,140
7,420
28,400
and left to cure under water for four weeks.
FGD product solids, ash-free, dried.
c Contains 19.1% portland cement.
After curing.
e Contains 20% portland cement, equivalent to 12-13% CaO.
Mixture was plastic and could not be compacted; it was rodded to eliminate air.
-------
is the deformation at failure. The modulus of the elasticity is the
stress divided by the strain at failure. The modulus is an indication
of the sample's resistance to deformation; i.e., it is the stress nec-
essary to produce a unit strain.
Included for comparison in Table VIII-8 are the measured unconfined com-
pressive strengths for untreated, compacted samples of dual alkali sulfite
and gypsum. These can be considered base cases to which the characteris-
tics of the various treated samples can be compared. Samples 1 and 2,
which were cured under water for 28 days, exhibited compressive strengths
of 80 psi and about 200 psi, respectively. While those strengths clearly
indicated that treatment had had an effect, the fact that the test speci-
mens were only about two inches high could have resulted in erroneously
high apparent unconfined compressive strengths. The remaining samples
had height:diameter ratios of about 2:1, which was more appropriate for
unconfined compression testing.
Samples 3-6 which contained varying proportions of fly ash and quicklime,
and which were not cured under water, showed small increases in unconfined
compressive strength over the untreated materials. Samples 4 and 6, which
contained the larger amount of quicklime, had the higher strengths. At a
constant 15% CaO, doubling the amount of fly ash also resulted in a higher
strength.
Samples 9-12 contained 20 wt % portland cement as a calcium source instead
of quicklime. By comparing samples 9 and 10, the increase in strength
with additional curing time can be seen clearly. During the second
fourteen days of curing, the compressive strengths more than doubled.
Sample 11 had the same dry composition as sample 10, but it was prepared
in the presence of more moisture, so that it could not be compacted be-
cause of its plasticity. Consequently, the mixture could only be rodded
in the mold. After two days of curing it had become sufficiently rigid
so that it could be removed from the mold and allowed to cure at 100%
relative humidity with the other samples. However, its strength after
twenty-eight days was somewhat lower than sample 10 which had been com-
pacted. A sample of dual alkali gypsum was treated in a manner identical
to that of sample 10; after twenty-eight days of curing it exhibited a
somewhat greater unconfined compressive strength than did the dual alkali
sulfite.
In summary, the chemical treatment procedures tested produced small, but
significant,increases in the unconfined compressive strengths of the dual
alkali solids. As a point of comparison, concrete generally has uncon-
fined compressive strengths in excess of 2,000 psi and moduli of elas-
ticity greater than 2,000,000. Thus, the treated solids, while considerably
stronger than untreated materials, did not have the concrete-like charac-
teristics required for materials of construction.
An attempt was made to measure the permeability of one sample of treated
dual alkali sulfite (identical to sample 10 in Table VIII-8). For some
reason, an apparent permeability of 2 x 10~4 cm/sec (not substantially
VIII-3Q
-------
different from untreated material) was measured in that test It could
nf^6^^6^111^ ?et?er the treated "^r^1 w*s, in fact, that permeable,
or whether the high value was due to experimental error. However, perme-
abilities of 5.5 x 10-6 and 5.6 x 10-6 cm/sec have been reported for
treated dual alkali sulfite after eight days and twenty-nine days of
curing, respectively, in the parallel study conducted for Southern
Company Services*.
Another treated sample, identical to the one whose permeability was
measured, was subjected to the accelerated leaching test described in
Section G. As shown in Table VIII-9, the concentration of both sodium
and calcium in the two fractions of leachate collected from the treated
material was considerably lower than for the untreated dual alkali sulfite.
The reductions in the concentrations of sodium by factors of three and
two, respectively, in the first and second leachate fractions could have
been the result of the dilution of the calcium sulfite solids by the addi-
tion of fly ash and portland cement. However, the cementation reaction
seemed to reduce the concentration of dissolved calcium quite markedly.
If both the sodium and calcium are considered to be present as sulfates,
the concentration of TDS in the first fraction of leachate collected
from the treated material corresponds to about 3,500 ppm, a significant
reduction from the 11,000 ppm of TDS in the first fraction collected
from the untreated material.
In the parallel effort mentioned previously2, samples of dual alkali
sulfite were treated, cured, and tested for leachability by shaking
samples of the solids, either treated or untreated, with four times
their weight of water for 48 hours. Treatment reduced the concentra-
tions of TDS in the wash liquors by factors ranging from 2 to nearly 5.
I. CONCLUSIONS
Limited testing was performed to characterize the basic physical and
chemical properties of ash-free waste filter cakes produced in the two
most successful dual alkali modes piloted — concentrated and dilute
active sodium modes with lime regeneration. Testing included: analysis
of major chemical constituents; crystalline morphology via X-ray diffrac-
tion and scanning electron microscopy; unconfined compressive strength;
compaction moisture/density relationship; permeability; leaching behav-
ior; and the effects of treatment with lime (or portland cement) and
fly ash on the physical properties.
The concentrated mode filter cake that was tested was produced in the
prototype system using the two-stage reactor. The cake was a mixture
of calcium sulfite and sulfate (about 15% calcium sulfate) and contained
55% solids. The crystalline structure of the solids was rosette-like
agglomerates of needles characteristic of the concentrated mode operation.
X-ray diffraction data and chemical analyses indicate that the calcium
sulfite and calcium sulfate were coprecipitated as a mixed crystal of
hemihydrate salts. There was no evidence of any appreciable amount of
gypsum (CaSOif • 2H20) in the solids.
VIH-31
-------
TABLE VIII-9
M
M
Sample
Treated
Dual Alkali Sulfite'
Untreated
Dual Alkali Sulfite
EFFECT OF
Eluate
Fraction
1
2
1
2
TREATMENT ON LEACHING
FROM DUAL ALKALI
Volume of
Fraction (ml)
339
338
396
312
OF SODIUM AND CALCIUM
SULFITE
Cum. Void
Volume
1.33
2.66
1.56
2.78
[Na ] ppm
1,120
300
3,100
600
[Ca ] ppm
39
100
520
598
Treated identically to sample 10, Table VIII-8.
-------
The Dilute mode filter cake was essentially pure gypsum produced in the
£±Vlan^der conditions of intentional oxidation. The solids crystals
were monoclinic and the filter cake contained approximately 80% insoluble
solids.
The mixed sulfite/sulfate solids had the appearance and physical proper-
ties similar to a silt-like soil and handled much like a moist powder.
The gypsum, on the other hand, was much more grainy and had the consis-
tency of a sandy soil. The unconfined compressive strengths of both
materials were in the range of typical soils, 10-15 psi, and both had
optimum dry densities in the range of 75% solids. The coefficient of
the permeability of the compacted sulfite/sulfate solids ranged from
about 3 x 10~lf to 5^x 10~5 cm/second. The permeability of dual alkali
gypsum was 2 x 10~5 cm/sec. These values are within the range of pub-
lished data on the coefficients of permeability of gypsum and sulfite-
rich solids produced in FGD systems;*1
The treatment of the sulfite/sulfate filter cake was studied using various
mixtures of lime (or portland cement), filter cake, and fly ash. This
work showed that the concentrated mode solids could be treated in a fashion
similar to the treatment of solids from direct lime and limestone scrubbing
systems with similar effects on the mechanical properties. Testing per-
formed on prototype system concentrated dual alkali solids by IU Conversion
Systems (IUCS) indicates that the coefficient of permeability of treated
filter cake was about 5 x 10~^ cm/sec using standard treatment mixes.
Accelerated leaching tests and elutriate analyses performed on untreated
samples both at ADL and IUCS showed that the initial and "steady-state"
concentrations of soluble species that can be leached, notably total dis-
solved solids (TDS) and total oxidizable sulfur (TOS), will be very de-
pendent upon the initial conditions and composition of the solids as
affected by the degree of cake washing, ratio of sulfate-to-sulfite,
chloride concentration in the gas, etc. and the manner of solids handling
and disposal. TDS levels in the initial leachate can range from a few
thousand ppm to about ten thousand ppm, and "steady-state" concentra-
tions (after the first few pore volume displacements) can vary from a
few hundred ppm to approximately two thousand ppm. Similarly, TOS levels
can range from essentially nil to up to fifty ppm. These concentrations
are consistent with the range of published data for leachates from solids
generated in direct lime and limestone scrubbing systems.
Testing performed by IUCS on the treatment of the filter cake indicated
significant reductions in both initial and "steady-state" levels of TDS
in leachates. Depending upon the type of treatment, reductions of 50%
to 80% were observed.
In all physical properties testing performed at ADL, samples were pre-
pared in accordance with standard soil-mechanics testing procedures.
These procedures required, as a part of the sample preparation, the
drying and rewetting of the filter cake to achieve a desired solids
content. While the samples were dried at a temperature of 83°C to
VIII-33
-------
prevent loss of water of hydration, there is still concern that the
drying/rewetting procedure resulted in some changes in the behavior
of the material, particularly in the case of the rosette-like crystals
produced in the concentrated mode operation. However, the results of
these limited tests are believed to be indicative of the general be-
havior of the dual alkali solids. More exhaustive testing on both as-
received samples and samples prepared in accordance with standard soil
testing procedures is required to assess the effects of sample prepara-
tion on test results.
VIH-34
-------
IX. REFERENCES
1. Leo, P. P., and J. Rossoff. Control of Waste and Water Pollution
from Power Plant Flue Gas Cleaning Systems: First Annual R and D
Report, EPA-600/7-76-018, October, 1976.
2. Edwards, R. Personal Communication. I.U. Conversion Systems to
Reed Edwards of the Southern Company Services, Inc., April, 1976.
3. Kaplan, N. Introduction to Double Alkali Flue Gas Desulfurization
Technology, Proceedings of the Sixth Flue Gas Desulfurization
Symposium, Environmental Protection Agency, New Orleans, Louisiana,
March 8-11, 1976, pp. 387-422.
4. LaMantia, C. R., E. L. Field, T. J. Lamb, J. E. Oberholtzer, and
J. R. Valentine. Sulfur Dioxide Control Process Study - Sodium
Scrubbing with Lime Regeneration, A Report to the State of Illinois,
Institute of Environmental Quality, Arthur D. Little, Inc., Cambridge,
MA, January, 1972, p. 64.
5. LaMantia, C. R., R. R. Lunt, and I. S. Shah. Dual Alkali Process
for S02 Control, Proceedings of the Sixty-sixth American Institute
of Chemical Engineers Symposium on Air: II. Control of NOX and SOX
Emissions, Philadelphia, Pennsylvania, 1975, Series #148, 71: 324-9-
6. Epstein, M., L. Sybert, S. C. Wang, C. C. Leivo, A. H. Abdul-Sattar
F. T. Princiotta, J. E. Williams, R. H. Borgwardt, R. M. Statnick,
and D. C. Drehmel. Preliminary Report of Test Results from the EPA
Alkali Scrubbing Test Facilities at the TVA Shawnee Power Plant and
at Research Triangle Park, Presented at an Environmental Protection
Agency Public Briefing at Research Triangle Park, N.C., December 19,
1973.
7. Kusik, C. L., and H. P. Meissner. Int. J. of Mineral Proc. _2,
105-115 (1975).
8. Marshall, W. L., and E. V. Jones. Second Dissociation Constant of
Sulfuric Acid from 25-350° Evaluated from Solubilities of Calcium
Sulfate in Sulfuric Acid Solutions, J. Phys. Chem., 70 (12):
4028-40, 1966.
9. Kaplan, N. An Overview of Double Alkali Processes for Flue Gas
Desulfurization, Proceedings of the Fifth Flue Gas Desulfurization
Symposium, Environmental Protection Agency, Atlanta, Georgia,
November, 1974, pp. 387-422.
10. Drehmel, D. C. Wet Limestone Scrubbing of Sulfur Oxides: Limestone
Selection, Proceedings of the Second International Lime/Limestone
Wet Scrubbing Symposium, New Orleans, Louisiana, November 8-12, 1971.
IX-1
-------
11. Draemel, D.C. Regeneration Chemistry of Sodium-Based Double-Alkali
Scrubbing Process EPA-R2-73-186, U.S. Environmental Protection
Agency, Research Triangle Park, North Carolina, March 1973, 37 pp.
12. Fitch, B. Batch Tests Predict Thickener Performance, Chemical
Engineering, August 23, 1971, pp. 83-88.
IX-2
-------
X. ANNOTATED BIBLIOGRAPHY
Ainsworth, Richard G.
DISSOCIATION CONSTANT OF CALCIUM SULPHATE FROM 25 to 50°C
J. Chem. Soc. Faraday Trans., Part I 69 (Pt. 6): 1028-32, 1973
Alper, E.
KINETICS OF OXIDATION OF SODIUM SULFITE SOLUTION
Trans. Inst. Chem. Eng., 51 (2): 159-61, 1973
The kinetics of oxidation of aqueous solutions containing cobaltous
sulfate as a catalyst were studied in a stirred cell and two different
packed columns.
Ando, Jumpei
STATUS OF FLUE GAS DESULFURIZATION AND SIMULTANEOUS REMOVAL OF S02 AND
NO IN JAPAN
X
Proceedings of the Sixth Flue Gas Desulfurization Symposium, Environ-
mental Protection Agency, New Orleans, Louisiana, March 8-11, 1976,
pp. 53-78
In Japan, twenty-four double alkali type flue gas desulfurization process
plants have been put in operation since the beginning of 1973. The
largest number of plant sites have been developed by Showa Denko. These
systems are designed to use Na2S03 as the absorbent and CaC03 as the
precipitant. Another large-scale developer, a combine of Kureha Chemical
Industry and Kawasaki Heavy Industries, have completed plants at four
electric power plant sites.
Bernard, R.E.; Teague, R.K.; and Vansickle, G.C.
THE CALSOX SYSTEM DEVELOPMENT PROGRAM
Proceedings of the Fifth Flue Gas Desulfurization Symposium, Environ-
mental Protection Agency, Atlanta, Georgia, November 1974, pp. 1127-49
A description of the pilot plant operations conducted in a joint program
between Monsanto Enviro-Chem. Systems, Inc. and Indianapolis Power and
Light Company. Ethanolamine is used as the alkaline agent for absorp-
tion in this dual alkali type process, since it can be readily regener-
ated with lime.
Barren, Charles H. and O'Hern, Harold A.
REACTION KINETICS OF SODIUM SULFITE OXIDATION BY THE RAPID-MIXING METHOD
Chemical Engineering Science, 21: 397-404, 1966
X-l
-------
Betts, R.H. and Voss, R.H.
KINETICS OF OXYGEN EXCHANGE BETWEEN THE SULFITE ION AND WATER
Can. J. Chem., 48 (13): 2035-41, July 1, 1970
Oxygen of mass 18 was used as a stable tracer to measure the rate of
exchange between the sulflte ion and water as a function of pH and total
sulfite concentration.
Betts, R.H. and Libich, S.
OXYGEN-18 TRANSFER IN THE SYSTEM THIOSULFATE-SULFITE-WATER: EXAMPLE OF
A SET OF CONSECUTIVE REVERSIBLE FIRST ORDER RATE PROCESSES
Can. J. Chem, 49 (2): 180-6, January 15, 1971
Bittrich, H.J. and Leibnitz, E.
INFORMATION ON THE SYSTEM Na+CA++OH~S04H20 Part I. CAUSTIFICATION OF
SODIUM SULFATE
J. Prakt. Chem., 3 (4): 126-36, 1956
Bloss, H.E.; Wilhelm, James; Holhut, W.J.
THE BUELL DOUBLE-ALKALI S02 CONTROL PROCESS
Proceedings of the Sixth Flue Gas Desulfurization Symposium, Environ-
mental Protection Agency, New Orleans, Louisiana, March 8-11, 1976,
pp. 545-62
The process designed by Envirotech Corporation is available in two modes.
The dilute mode for use with low sulfur coals utilizes NaOH in the
scrubber, Ca(OH)a to regenerate the scrubber liquor, and a softening
agent to reduce the calcium ion level in the liquor before it is returned
to the scrubber. In the concentrated mode sodium sulfite/bisulfite is
used to remove SO . Design modifications for use with high chloride coal
are presented.
Borisek, R.; Balhar, L., Schmied, J.; and Vesely, J.
INHIBITION OF THE OXIDATION OF SULFITES IN AQUEOUS SOLUTION BY AIR
Czech Patent 145, 747, October 15, 1972, 2p.
A method for inhibiting the oxidation of sulfite ions to sulfate ions
through the use of a solution of formaldehyde which forms an aldehyde-
bisulfite complex thereby blocking the formation of oxygen radicals.
X-2
-------
Brady, J.D.
SULFUR DIOXIDE REMOVAL USING SOLUBLE SULFITES
Presented at Rocky Mountain States Section, Air Pollution Control Assoc.,
Colorado Springs, Colorado, April 30, 1974, 34P.
A comparison of the process chemistry of double-alkali systems. The
author describes the method which uses NaOH as the absorption solution
as dilute double-alkali and the FMC Corp. system in which the absorbing
solution is sodium sulfite as concentrated double alkali since the
sulfite and bisulfite concentration typically totals above 1.0 molar.
Capital and operating cost data are reported for these two systems.
Bunn, C.W.
ADSORPTION, ORIENTED OVERGROWTH AND MIXED-CRYSTAL FORMATION
Proc. Roy. Soc. (London) A141: 567-93, 1933
Research on mixed-crystal formation shows that where a secondary
substance — an impurity — precipitates on a certain plane or planes of
the primary crystal there is modification of the crystal habit.
Chen, T. and Barren, C.H.
SOME ASPECTS OF THE HOMOGENEOUS KINETICS OF SULFITE OXIDATION
Ind. Eng. Chem. Fundam., 11 (4): 466-70, 1972
A study of the homogeneous kinetics and catalysis of sodium sulfite
using a rapid-mixing method. This technique allowed the reaction of
already dissolved oxygen, thus eliminating possible errors due to the
interphase transfer of oxygen. The experimental findings showed that:
(1) the reaction rate was independent of oxygen concentration; (2) the
reaction order was 1.5 with respect to sulfite concentration; and
(3) the reaction rate was proportional to the square root of the total
concentration of cobalt added to the reacting solution.
Cornell, C.G. and Dahlstrom, D.A.
SULFUR DEVELOPMENTS: SULFUR DIOXIDE REMOVAL IN A DOUBLE-ALKALI PLANT
Chem. Eng. Prog., 69 (12): 47-53, 1973
Pilot plant operations of a double alkali system are described. The
calcium sulfate waste is disposed of by means of landfill. The system
was designed by Envirotech Corp. and the facility is located at a Salt
Lake City power station.
X-3
-------
Counterman, P. and Altwicker, R.
INHIBITION OF OXIDATION OF SULFUR DIOXIDE DURING ALKALINE SCRUBBING.
Preprint of paper presented at the 164th National American Chemical
Society Meeting, New York City, August 28-September 1, 1972, pp. 43-9
Results of laboratory studies on the use of antioxidants to inhibit the
formation of sulfates in alkaline solutions of sulfite and bisulfite
ions. The inhibitors listed are benzyl alcohol, phenol, and hydroquinone.
Dahlstrom, D.A. and Cornell, C.F.
SULFUR DIOXIDE SCRUBBING PROCESS
U.S. Patent 3,873,532, March 25, 1975, assigned to Envirotech Corporation,
Menlo Park, California
This process utilizes a sodium-based aqueous scrubbing solution to
absorb stack gases, followed by a regeneration of the solution with
slaked lime. This solution is then reacted with sodium carbonate, to
reduce calcium ion concentration and solids content, before it is
returned to the absorption apparatus.
Devitt, T.W.; Isaacs, G.A.; and Laseke, B.A.
STATUS OF FLUE GAS DESULFURIZATION SYSTEMS IN THE UNITED STATES
Proceedings of the Sixth Flue Gas Desulfurization Symposium, Environ-
mental Protection Agency, New Orleans, Louisiana, March 8, 1976, pp. 13-
51
A discussion of numbers and types of FGD systems, operational, under
construction, or planned. Reports on a number of individual facilities
presents values for process designs and operating parameters. Study is
on-going and further reports will be generated, including data on dual
alkali systems.
De Wall, K.J.A. and Okeson, J.C.
THE OXIDATION OF AQUEOUS SODIUM SULPHITE SOLUTIONS
Chemical Engineering Science, 21: 559-72, 1966
A method is suggested for the determination of the interfacial area
between a gas and a liquid and of the surface renewal rate of this
interface. This method is based upon the oxidation of a concentrated
aqueous sulphite solution with gaseous oxygen in the presence of
cobaltous sulphate as a catalyst.
X-4
-------
Dingo, T.T.
°F M ^DUSTRIAL SIZE DUAL ALKALI S02
Proceedings of the Fifth Flue Gas Desulfurization Symposium, Environ-
mental Protection Agency, Atlanta, Georgia, November 1974, pp. 517-37
A review of the process chemistry, equipment and performance character-
istics of the General Motors double alkali S02 Control System.
Draemel, D.C.
REGENERATION CHEMISTRY OF SODIUM-BASED DOUBLE-ALKALI SCRUBBING PROCESS
EPA-R2-7 3-186, U.S. Environmental Protection Agency, Research Triangle
Park, North Carolina, March 1973, 37 pp.
*
A report of the results of a study of the reactions of calcium hydroxide
and calcium carbonate with the aqueous (sodium, sulfite, bisulfite, and
sulfate) system. The objectives were to study various reactions of
importance in the sodium-based double alkali process and to define
possible operating modes for the process.
Draemel, D.C.
AN EPA OVERVIEW OF SODIUM-BASED DOUBLE ALKALI PROCESSES
PART I. A VIEW OF THE PROCESS CHEMISTRY OF IDENTIFIABLE AND ATTRACTIVE
SCHEMES
Proceedings of the Fourth Flue Gas Desulfurization Symposium, Environ-
mental Protection Agency, New Orleans, Louisiana, 1973, pp. 997-1018
The process chemistry of double alkali scrubbing is presented, and five
double alkali operating schemes are outlined. Sulfur oxides are absorbed
into a sulfite/bisulfite buffer solution, shifting the pH down and
increasing the bisulfite concentration. The liquor from the scrubber may
then be treated with limestone to precipitate calcium sulfite and neutral-
ize the bisulfite, and the liquor from the scrubber or the limestone
reaction vessel may be treated with lime to precipitate calcium sulfite
and possible calcium sulfate. The oxidation of absorbed sulfur requires
the regeneration of sulfate. Calcium ion concentrations in the scrubber
can be controlled by softening steps in the liquor loop.
Drehmel, D.C.
WET LIMESTONE SCRUBBING OF SULFUR OXIDES: LIMESTONE SELECTION
Proceedings of the Second International Lime/Limestone Wet Scrubbing
Symposium, New Orleans, Louisiana, November 8-12, 1971
A report of tests made on different types of limestone to determine their
dissolution rates in acid media and their sulfur oxide removal effi-
ciencies in a batch scrubber.
X-5
-------
Elder, H.W.; Princiotta, F.T.; Hollinden, G.A.; Gage, S.G.
SULFUR OXIDE CONTROL TECHNOLOGY
VISITS IN JAPAN - AUGUST 1972
U.S. Government Interagency Report, Muscle Shoals, Alabama, October 30,
1972, 113 pp.
During the period August 2 to August 16, 1972, a U.S. Government team
visited organizations in Japan to evaluate technology for control of
sulfur dioxide emissions. The Tennessee Valley Authority, the EPA and
the Office of Science and Technology (Executive Office of the President)
were represented. They visited 14 companies and the Japan Environmental
Agency. The primary purpose of the trip was to observe full-scale
scrubbing systems in operation on coal fired power plants. Systems
discussed include dual alkali.
Ellison, W.; Heden, S.D.; and Kominek, E.G.
SYSTEM RELIABILITY AND ENVIRONMENTAL IMPACT OF S02 SCRUBBING PROCESSES
Proceedings of The Coal Utilization Symposium-Focus S02 Emission Control,
National Coal Association, Louisville, Kentucky, 1974, p.130-52
A discussion of the major problems of system design and operation related
to scrubbing processes. It includes information on (1) the chemistry of
throw-away systems e.g. lime/limestone and double alkali, (2) reliability
design with regard to major equipment for these processes and (3) system
development for the double-alkali technique of scrubbing.
Epstein, M.
EPA ALKALI SCRUBBING TEST FACILITY: SUMMARY OF TESTING THROUGH OCTOBER
1974
EPA-650/2-75-047, Bechtel Inc., San Francisco, California, June 1975,
506 p.
This report on the EPA test facility at Shawnee Power Station includes
data on sodium carbonate (soda ash) scrubbing. Models are presented for
predicting S02 removal in sodium carbonate.
Epstein, M.; Sybert, L.; Wang, S.C.; Leivo, C.C.; Abdul-Sattar, A.H.;
Princiotta, F.T.; Williams, J.E.; Borgwardt, R.H.; Statnick, R.M.; and
Drehmel, D.C.
PRELIMINARY REPORT OF TEST RESULTS FROM THE EPA ALKALI SCRUBBING TEST
FACILITIES AT THE TVA SHAWNEE POWER PLANT AND AT RESEARCH TRIANGLE PARK
Presented at an Environmental Protection Agency Public Briefing at
Research Triangle Park, N.C., December 19, 1973
In June, 1968 EPA initiated a program to test a prototype lime and lime-
stone wet-scrubbing system for removing S02 and particulates from flue
gases.
This report presents the results, through early December, 1973, of (1)
limestone and lime reliability verification and long-term reliability
testing at the Shawnee Prototype Facility and (2) limestone and lime
testing at the EPA Pilot Facility at Research Triangle Park, N.C.
X-6
-------
Field, E.L.; LaMantia, C.R.; Lunt, R.R.; Oberholtzer, J.E.; and
Valentine, J.R.
MULTISTAGE PROCESS FOR REMOVING SULFUR DIOXIDE FROM STACK GAGES
U.S. Patent 3,994,649, March 16, 1976, assigned to Combustion Equipment
Associates, Inc., New York, NY
A method and apparatus for removing sulfur dioxide from stack gases by
means of an alkali scrubbing solution which is regenerated by treatment
with calcium compounds in at least two separate reaction stages to
produce calcium sulfate and calcium sulfite prior to recycling the
scrubber solution.
Fitch, B.
BATCH TESTS PREDICT THICKENER PERFORMANCE
Chemical Engineering, August 23, 1971, pp. 83-88
Discussion of the advantages and limitations of several methods for
using batch settling data to design thickeners. The paper includes a
review of the fundamental equation of solids flux derived by Coe and
Clevenger, as well as several thickening models and design criteria.
Frazier, J.F.
REMOVAL OF SULFUR OXIDES FROM INDUSTRIAL BOILER FLUE GASES
National Eng., 75 (8): 6-9, 1971
A discussion of costs of the chemical requirements and cost of equipment
for the dual-alkali method for desulfurizing flue gases. Also has a
table of factors for estimating quantities of chemicals used in the
process.
Galeano, S.F. and Harding, C.I.
SULFUR DIOXIDE REMOVAL AND RECOVERY FROM PULP MILL POWER PLANTS
J. Air Pollut. Control Assoc., 17 (8): 536-9, August 1967
The use of soda ash liquor to scrub S02-rich power plant flue gases was
studied using an Airetron pilot scrubber with a maximum capacity of
3000 cfm. The relative effects of the major operating variables —
temperature, soda ash concentration, and the gas/liquid flow ratio •— on
the absorption phenomenon were determined.
Harris, I.F. and Roper, G.H.
THE ABSORPTION OF OXYGEN BY SODIUM SULPHITE ON A SIEVE PLATE
Can. J. Chem. Eng., 42: 34-7, February 1964
The oxidation of sodium sulfite on a single sieve plate indicates that
the reaction rate constant is substantially greater than that reported
in the literature.
X-7
-------
Horlacher, W.R.; Barnard, R.E.; league, R.K. and Hayden, P.L.
FOUR SULFUR DIOXIDE REMOVAL SYSTEMS
Chem. Eng. Progr., 68 (8): 43-50, 1972
A description of equipment and methods of operation developed by Mon-
santo Enviro-Chem Systems, which includes double alkali type process.
Howard, H. and Stantial, F.G.
METHOD OF RECOVERING SULFUR DIOXIDE FURNACE GASES
U.S. Patent #1,271,899, July 9, 1918, 2p.
Initial patent on the method of scrubbing flue gases with a sodium salt
solution followed by reaction of the scrubber effluent with calcium
hydroxide to precipitate calcium sulfite and regenerate the scrubber
feed.
E.; Ciabettari, E.; Wolff, R.A.; and Bernstein, I.
POLLUTION CONTROL BY AIR OXIDATION OF WASTE SULFITE LIQUORS
Industrial and Eng. Chem., 51 (10): 1301-4, October 1959
Semi-pilot plant investigation ,of the oxidation of sulfite wastes with
tests of catalysts and their effect on the oxidizing rate. The equip-
ment used in the experiment is described.
Johnstone, H.F.; Read, H.J.; and Blankmeyer, H.C.
RECOVERY OF SULFUR DIOXIDE FROM WASTE GASES: EQUILIBRIUM VAPOR PRESSURES
OVER SULFITE-BISULFITE SOLUTIONS
Industrial and Eng. Chem., 30 (1): 101-9, 1938
Partial vapor pressures were measured over a temperature range from
35° - 90°C for an extensive series of concentrations and compositions of
sulfite-bisulfite solutions of sodium and methylamine. Sufficient data
for a number of other sulfite-bisulfite solutions are reported to show
the effect of the nature of the solution on the temperature coefficient
of the vapor pressure of sulfur dioxide.
Kaplan, N.
INTRODUCTION TO DOUBLE ALKALI FLUE GAS DESULFURIZATION TECHNOLOGY
Proceedings of the Sixth Flue Gas Desulfurization Symposium, Environ-
mental Protection Agency, New Orleans, Louisiana, March 8-11, 1976,
pp. 387-422
A detailed description of the sodium/calcium based double alkali pro-
cesses to date. Special terminology which has evolved with the tech-t
nology is defined and discussed. These include such terms as absorp-
tion/regeneration chemistry, active alkali, active sodium and TOS (total
oxidizable sulfur). Significant process and design factors with some
cost information are carefully reviewed. EPA's Research and Development
program on double alkali technology is outlined.
Xo
— o
-------
Kaplan, N.
AN OVERVIEW OF DOUBLE ALKALI PROCESSES FOR FLUE GAS DESULFURIZATION
Proceedings of the Fifth Flue Gas Desulfurization Symposium, Environ-
mental Protection Agency, Atlanta, Georgia, November 1974, pp. 387-422
The chemistry and process design considerations applicable to sodium/
calcium double alkali systems are presented. Technical terminology
associated with these systems is defined.
The developmental efforts and full scale applications of the technology
by Envirotech, FMC, General Motors, Zurn Industries, Arthur D. Little/
Combustion Equipment Associates, Kawasaki Heavy Industries/Kureha Chem-
ical Industry and Showa Denko KK are discussed with reference to appro-
priate flowsheets. Planned applications of technology by these companies
are also discussed and tabulated.
Kaplan, N.
AN EPA OVERVIEW OF SODIUM BASED DOUBLE ALKALI PROCESSES. PART II -
STATUS OF TECHNOLOGY AND DESCRIPTION OF ATTRACTIVE SCHEMES
Proceedings of the Fourth Flue Gas Desulfurization Symposium, Enviromental
Protection Agency, New Orleans, Louisiana, 1973, pp. 1019-60
Criteria for evaluating double alkali schemes are described and include
sulfate removal, scale prevention, water balance, waste product washing,
S0£ removal and capital and operating costs. Flowsheets for potentially
attractive schemes of operation are presented and compared. Status of
the technology is reviewed. The EPA/A.D. Little double alkali develop-
ment program plan and general philosophy are described.
Kawamoto, K.; Tsuno, T. and Namiki, T. (Mitsubishi Chemical Industries
Co., Ltd.)
ANTIOXIDANTS FOR SULFITE SOLUTIONS
German Patent #2,243,201, March 15, 1973, 16p.
A patent covering the use of the following classes of antioxidants and
their effects on solutions of sodium bisulfite, sodium sulfite, potassium
bisulfite and potassium sulfite. (1) substituted phenols, (2) tris
(alkyphenyl) phosphates, (3) trialkyl phosphites and (4) alpha glycerol
esters. Methods for their use are also in this patent.
X-9
-------
Kerr C P.
AQUEOUS SODIUM SULFITE, BISULFITE AND SULFATE EQUILIBRIA
AIChE, J.., 20 (6): 1213-5, November .1974
A description and formulae for use in calculating parameters for
scrubber design. The subject areas explored are the equilibrium S02
partial pressure over the solution and liquid composition with the
following specifications: temperature, total pressure, sodium ion
concentration, total dissolved sulfur concentration, COa partial
pressure, and sulfate ion concentration.
LaMantia, C.R.; Bangel, E.R.; Phillips, R.; and Lamb, T.J.
EMISSION CONTROL FOR SMALL SCALE FACILITIES
A Chemical Engineering Progress Technical Manual, Amer. Institute of
Chem. Engineers, New York, NY, 1971, pp. 142-50
A study to evaluate the various approaches for control of S02 according
to the following criteria: emission reduction, technical risk., reli-
ability, appropriateness of the process and capital investment and
operating costs. Dual alkali process is among these considered.
LaMantia, C.R.; Lamb, T.J.; Sommer, R.S.; Shah, I.S.; and Falco, J.M.
SULFUR DIOXIDE EMISSION CONTROL FOR INDUSTRIAL BOILERS
Paper presented at Industrial Coal Conference, Purdue University,
October 5, 1972, 18 pp.
A discussion of the overall problem of S0£ control for industrial scale
boilers and description of three appropriate scrubbing processes (1)
once-through sodium, (2) direct lime/limestone and (3) sodium with lime
regeneration (includes cost data).
LaMantia, C.R.; Field, E.L.; Lamb, T.J.; Oberholtzer, J.E.; Valentine, J.R.
SULFUR DIOXIDE CONTROL PROCESS STUDY - SODIUM SCRUBBING WITH LIME
REGENERATION
A Report to the State of Illinois, Institute of Environmental Quality,
Arthur D. Little, Inc., Cambridge, Massachusetts, January 1972, 64 p.
An experimental laboratory investigation of the regeneration step of the
sodium scrubbing process with lime regeneration. It includes an analysis
of the experimental results, a conceptual process design, and a prelim-
inary estimate of capital and operating costs for a representative size,
full-scale, industrial boiler installation.
X-10
-------
LaMantia, C.R. and Raben, I. A.
SOME ALTERNATES FOR S02 CONTROL
> ^ouisville, Kentucky,
Technology of three systems of gas scrubbing — sodium, alkali fly ash,
and dual alkali being used on an industrial scale are described. The
technology is the result of cooperative efforts between Arthur D. Little,
Inc. and Combustion Equipment Associates, Inc.
LaMantia, C.R.; Lunt, R.R.; Oberholtzer, J.E.; Field, E.L.; and Kaplan, N.
EPA-ADL DUAL ALKALI PROGRAM — INTERIM RESULTS
Proceedings of the Fifth Flue Gas Desulfurization Symposium, Environmental
Protection Agency, Atlanta, Georgia, November 1974, pp. 549-665
A report of the laboratory and pilot plant results obtained on lime and
limestone regeneration of concentrated and dilute sodium scrubbing
solutions as well as work performed on the sulfuric acid treatment scheme
for soluble sulfate precipitation.
LaMantia, C.R.; Lunt, R.R.; and Shah, I.S.
DUAL ALKALI PROCESS FOR S02 CONTROL
Proceedings of the Sixty-sixth American Institute of Chemical Engineers
Symposium on Air: II. Control of NO and SO Emissions, Philadelphia,
Pennsylvania, 1975, Series #148, 71? 324-9 X
Laboratory development work is discussed including batch reactor
experiments and continuous reactor experiments. Pilot plant facility
is described and performance data on the scrubber system and regeneration
reactor system are presented.
LaMantia, C.R.; Lunt, R.R.; Rush, R.E.; Frank, T.M.; and Kaplan, N.
OPERATING EXPERIENCE—CEA/ADL DUAL ALKALI PROTOTYPE SYSTEM AT GULF
POWER/SOUTHERN SERVICES, INC.
Proceedings of the Sixth Flue Gas Desulfurization Symposium, Environ-
mental Protection Agency, New Orleans, Louisiana, March 8-11, 1976,
pp. 423-69
A description of the 20-megawatt dual alkali S02 control facility at
Gulf Power Company's Scholz Steam Plant. The system was developed,
designed and installed by Combustion Equipment Associates, Inc. and
Arthur D. Little, Inc. for Gulf Power/Southern Services, Inc. and tested
as part of EPA's Industrial Environmental Research Laboratory program.
Operating history and system performance are reviewed from initial
start-up through the first year of operation.
X-ll
-------
Larson, J.W.
THERMODYNAMICS OF DIVALENT METAL SULFATE DISSOCIATION AND THE STRUCTURES
OF THE SOLVATED METAL SULFATE ION PAIR
J. of Phys. Chem., 74 (18): 3392-6, September 3, 1970
Data for the heats and entropies of dissociation of divalent metal
sulfate ion pairs are reported. These calculations made use of the very
precise heats of dilution measured by Lange or his coworkers.
Legatski, L.K.; Johnson, K.E. and Lee, L.Y.
THE FMC CONCENTRATED DOUBLE-ALKALI PROCESS
Proceedings of the Sixth Flue Gas Desulfurization Symposium, Environ-
mental Protection Agency, New Orleans, Louisiana, March 8-11, 1976,
pp. 471-502
This process utilizes Na2S03 in the scrubber. In the regeneration step,
which is pH controlled, Ca (OH)2 is employed. Development efforts such
as scrubber performance, sulfate formation rate studies, and lime
reactor performance show strategy employed. Plant demonstration
descriptions include some capital and operating cost data.
Leo, P.P. and Rossoff, J.
CONTROL OF WASTE AND WATER POLLUTION FROM POWER PLANT FLUE GAS CLEANING
SYSTEMS: FIRST ANNUAL R AND D REPORT
EPA-600/7-76-018, October 1976
This report summarizes and assesses the state of research and development
as of 1975 in the fields of nonregenerable flue gas cleaning waste
treatment, utilization and disposal, as well as water reuse technology
for coal-fired utility power plants.
Lewis, P.M.
OPERATING EXPERIENCE WITH THE ZURN DOUBLE ALKALI FLUE GAS DESULFURIZATION
PROCESS
Proceedings of the Sixth Flue Gas Desulfurization Symposium, Environ-
mental Protection Agency, New Orleans, Louisiana, March 8-11, 1976,
pp. 503-14
This system in use at a plant of the Caterpillar Tractor Co. utilizes a
dilute NaOH solution for gas scrubbing. The spent solution is regen-
erated with Ca(OH)2, after which a "softening" agent Na2C03 is added to
further precipitate calcium before the sodium hydroxide solution is
returned to the scrubber.
X-12
-------
Linek, V.
THE OXIDATION OF AQUEOUS SULPHITE SOLUTIONS
Chem. Eng. Science, 26: 491-4, 1971
A discussion of data in the literature concerned with the absorption
rate of oxygen in aqueous sulfite solutions and the effects of impurities,
pH, and inert gas on the absorption rate.
Linek, V. and Tvrdik, J.
GENERALIZATION OF KINETIC DATA ON SULPHITE OXIDATION SYSTEMS
Biotechnol, and Bioeng., 13 (3): 353-69, May 1971
A study of mass transfer in gas-liquid systems utilizing the oxidation
of aqueous sulfate solutions in the presence of a cobaltous sulfate
catalyst. The derived data was used as a basis for a method of kinetic
data processing.
Linek, V. and Mayrhoferova, J.
THE KINETICS OF OXIDATION OF AQUEOUS SODIUM SULPHITE SOLUTION
Chem. Eng. Science, 25: 787-800, 1970
The kinetic data of the reaction were calculated from the absorption
rate of oxygen into the mechanically agitated sulphite solutions. It
was found that the reaction order in oxygen depends on the oxygen con-
centration in the liquid phase at the interface. The reaction is first
order for oxygen concentrations higher than approximately 6. 10~'t kmol.itT3
at the interface and second order for lower oxygen concentrations.
Lunt, R.R.; Rush, R.E.; Frank, T.E.; LaMantia, C.R.
STARTUP AND OPERATION OF THE CEA/ADL ALKALI PROCESS AT GULF POWER/
SOUTHERN SERVICES
Presented at the Sixty-eighth Annual Meeting of the American Institute of
Chemical Engineers, Los Angeles, California, November 16-20, 1975, 36 p.
A description of the 20-megawatt, dual alkali S02 control process at
Gulf Power Company's Scholz Steam Plant and summary of the results of
the initial months of operation. The system was developed, designed and
installed by Combustion Equipment Associates, Inc./Arthur D. Little, Inc.
for Southern Services, Inc. at Gulf Power Company's Scholz Plant in
Sneads, Florida,and started up in February 1975.
X-13
-------
Marshall, W.L. and Jones, E.V.
SECOND DISSOCIATION CONSTANT OF SULFURIC ACID FROM 25-350° EVALUATED
FROM SOLUBILITIES OF CALCIUM SULFATE IN SULFURIC ACID SOLUTIONS
J. Phys. Chem., 70 (12): 4028-40, 1966
A documentation of values from second dissociation quotients and con-
stants for sulfuric acid determined from extensive solubility measure-
ments of calcium sulfate and its hydrates in aqueous sulfuric acid at
specific concentrations and temperatures.
Marshall, W.L. and Slusher, R.
DEBYE-HUECKEL CORRELATED SOLUBILITIES OF CALCIUM SULFATE IN WATER AND IN
AQUEOUS SODIUM NITRATE AND LITHIUM NITRATE SOLUTIONS OF MOLALITY 0 TO 6
MOLE KG"1 AND AT TEMPERATURES FROM 398 TO 623°K
J. Chem. Thermodyn., 5 (2): 189-97, 1973
McGlammery, G.G.; Faucett, H.L.; Torstrick, R.L.; and Henson, L.J.
FLUE GAS DESULFURIZATION ECONOMICS
Proceedings of the Sixth Flue Gas Desulfurization Symposium, Environ-
mental Protection Agency, New Orleans, Louisiana, March 8-11, 1976,
pp. 79-99
Cost data for the double-alkali systems—Envirotech, CEA/ADL, and FMC
are reported briefly. Study shows total investment and revenue require-
ments of the three systems are relatively close. More intensive evalua-
tion of the most promising systems is planned.
Morita, T.
ANTIOXIDATION TREATMENT OF BISULFITES AND SULFITES
Japanese Patent 72, 20, 455, June 10, 1972, 3pp.
A patent covering the use of the following antioxidants as inhibitors
in solutions of sulfites: p-aminophenol, p-aminophenol & sodium
nitriloacetate, or the combination of p-aminophenol & sodium nitrilo-
acetate & p-phenylenediamine.
Morita, T.
POLLUTION CONTROL. 3. AIR POLLUTION CONTROL. WASTE GAS DESULFURIZATION
BY THE SODIUM SULFITE-GYPSUM PROCESS
Kagaku, Kogyo, 23 (7): 915-21, 1972 (In Japanese)
X-14
-------
Morita, T.
VVT , KUREM SODIUM SULFITE-GYPSUM PROCESS
Sekka To Kekkai No. 121, 317-21, 1972 (Japan)
A review of a process for absorbing sulfur dioxide in a sodium sulfite
solution and then treating the resulting solution with limestone to
produce calcium sulfite and recover sodium sulfite — the calcium
sulfite is then oxidized to calcium sulfate.
Nilsson, G.; Rengemo, T.; and Sillen, L.
SOME SOLUTION EQUILIBRIA INVOLVING CALCIUM SULFITE AND CARBONATE
I. SIMPLE SOLUBILITY EQUILIBRIA OF C02, S02 , CaC03 AND CaS04
Acta Chemica Scandinavica, 12 (5): 868-72, 1958
Onda, K.; Takeuchi, H. and Maeda, Y.
THE ABSORPTION OF OXYGEN INTO SODIUM SULPHITE SOLUTIONS IN A PACKED
COLUMN
Chem. Eng. Science, 27: 449-51, 1972
A study of the absorption of oxygen into sodium sulphite solutions in
the presence of cobalt sulphate as catalyst with the investigation
directed to the influence of oxygen pressure on absorption rate and
interfacial areas.
Onozuka, M.; Nomoto, K.I.; and Morita, T.
METHOD FOR PRODUCING CALCIUM SULFITE SEMIHYDRATE
U.S. Patent 3,848,070, November 12, 1974, assigned to Kureha Kagaku
Kogyo Kabushiki Kaisha, Tokyo
A method comprising the use of an aqueous solution of sodium sulfite to
absorb S02 from stack gases and subsequently reacting this solution with
calcium carbonate under controlled conditions to form the product.
Edwards, R.
PERSONAL COMMUNICATION, I.U. CONVERSION SYSTEMS TO REED EDWARDS OF THE
SOUTHERN COMPANY SERVICES
April, 1976
A report to Reed Edwards from IUCS, regarding limited chemical analyses
and physical properties testing performed on waste solids generated by
the prototype systems at the Scholz 'Steam Plant.
X-15
-------
Phillips, R.J.
SULFUR DIOXIDE EMISSION CONTROL FOR INDUSTRIAL POWER PLANTS
Proceedings of the Second International Lime/Limestone Wet-Scrubbing
Symposium, Environmental Protection Agency, New Orleans, Louisiana,
May 14-17, 1973, 36p.
A description of the research and development done at General Motors on
caustic wet scrubbing with lime regeneration from laboratory scale tests
to pilot plant operations.
Piasecki, E.J.
EQUIPMENT PERFORMANCE WITH A COMMERCIAL DUAL-ALKALI S02 REMOVAL SYSTEM
Proceedings of the Fifth Flue Gas Desulfurization Symposium, Environ-
mental Protection Agency, Atlanta, Georgia, 1974, pp. 539-48
A discussion of the GM dual alkali system and equipment design and
performance.
Rand, M.C. and Gale, S.B.
KINETICS OF THE OXIDATION OF SULFITES BY DISSOLVED OXYGEN
Proceedings of the Fourth Conference on Principles Appl., Water Chem.,
Rudolphs Res. Rutgers State University, 1965, pp. 380-404
A review of the technical results in scientific publications in this
subject area which appeared in the literature from 1897 to 1960.
Rawa, R.T.
S02 CONTROL FOR SMALL BOILERS
Pollut. Eng., 4 (1): 22-3, January/February 1972
An overview of caustic wet scrubbing for small scale industrial boilers
with boiler steam capacities from about 50,000 to 500,000 Ib/hr.
Reinders, W. and Vies, S.I.
REACTION VELOCITY OF OXYGEN WITH SOLUTIONS OF SOME INORGANIC SALTS
III. THE CATALYTIC OXIDATION OF SULPHITES
Rec. Trav. Chem., 44: 249-68, 1925
Details of a study on the oxidation of sulphite without and with a
catalyst and with copper salts as catalysts. The reaction-velocity
curve with respect to the pH was traced and the_optimum velocity was
found to lie at pH of 10. Two constituents (SOs and Cu++) determine
the velocity constants and both are strongly dependent on the pH.
Lowering the pH decreases the velocity because the SOg ion concentra-
tion decreases with the formation HSOs and H2S03 so that at pH 3
scarcely and 803 ions exist and the velocity is very small.
X-16
-------
Reith, T. and Beek, W.J.
THE OXIDATION OF AQUEOUS SODIUM SULPHITE SOLUTIONS
Chem. Eng. Science, 28: 1331-9, 1973
New experimental work on the kinetics of the absorption of oxygen in
an aqueous solution of sodium sulphite with cobaltous sulphate as a
catalyst. The data presented is compared with already published data.
Rengemo, T; Brune, U.; and Sillen, L.G.
SOME SOLUTION EQUILIBRIA INVOLVING CALCIUM SULFITE AND CARBONATE
II. THE EQUILIBRIUM BETWEEN CALCIUM SULFATE AND CALCIUM SULFITE IN
AQUEOUS SOLUTIONS
Acta Chemica Scandinavica, 12: 873-7, 1958
Rochelle, G.T.
ECONOMICS OF FLUE GAS DESULFURIZATION
Proceedings of the Fourth Flue Gas Desulfurization Symposium, Environ-
mental Protection Agency, New Orleans, Louisiana, 1972, pp. 103-32
A summary of results and conclusions of an analysis of the costs of
flue gas desulfurization for fossil fuel boiler plants, includes data
on dual alkali process costs.
Shah, I.S.
REMOVAL OF SULFUR DIOXIDE FROM GAS STREAMS
U.S. Patent 3,775,532, November 27, 1973, Assigned to Chemical
Construction Corporation, New York, NY
A process for wet scrubbing of waste flue gas which employs an aqueous
solution of sodium sulfite as the scrubber liquid, resulting in an
aqueous sodium sulfite and sodium bisulfite mixture. Portions of this
solution are treated in two separate procedures to precipitate calcium
sulfite and calcium sulfate and thus prevent undesirable solids build-
up in the circulating system.
Spalding, C.W. and Han, S.T.
ABSORPTION WITH CHEMICAL REACTION FROM A DILUTE GAS IN PACKED TOWERS
Tappi, 45 (3): 192-9, 1962
A review of the absorption in a liquid, of a dilute component of a
gaseous mixture by chemical reaction on the basis of both the film and
the penetration theories. The design procedure and operating character-
istics of packed towers are discussed in the light of the theories and
illustrated with S02-H20 and S02-NaOH-H20 systems.
X-17
-------
Srivastava, R.D.; McMillan, A.F.; and Harris, I.J.
THE KINETICS OF OXIDATION OF SODIUM SULPHITE
Can. J. Chem. Eng., 46: 181-4, June 1968
A study of the kinetics of the reaction of sodium sulphite in aqueous
solution without and with a catalyst of cobaltous sulphate. The
reaction rate was found to be first order with respect to oxygen
and variable order with respect to sulphite, and the accelerating effect
of cobaltous catalyst was proportional to its concentration.
Wall, B.I.; Harris, H.C.; and Arnuad, J.L.
PROCESS FOR SEPARATING SULFUR OXIDES FROM GAS STREAMS
U.S. Patent 3,911,084, October 7, 1975, assigned to FMC Corporation,
New York, NY
A scrubber solution of sodium sulfite-sodium bisulfite maintained at
pH 6 is regenerated in a single stage step by addition of calcium
hydroxide to precipitate calcium sulfite which is removed as waste.
Solution is then returned to the scrubber liquor.
Wen, C.Y. and Uchida, S.
ABSORPTION OF S02 BY ALKALINE SOLUTIONS IN VENTURI SCRUBBER SYSTEMS
EPA-650/2-73-003, U.S. Environmental Protection Agency, West Virginia
University, July 1973, 169pp.
A description of studies on S02 absorption from flue gases by water
and alkaline solutions in venturi scrubbers, including recycle of the
scrubbing liquor. Mathematical models are proposed for process momentum,
heat and mass transfer in S02-NaOH-H20 and S02-CaC03~H20 systems. The
momentum, heat and mass balances are used to describe the reactions
taking place in the venturi scrubber.
Wesselingh, J.A. and Van't Hoog, A.C.
OXIDATION OF AQUEOUS SULPHITE SOLUTIONS: A MODEL REACTION FOR MEASURE-
MENTS IN GAS-LIQUID DISPERSIONS
Trans. Instn. Chem. Engrs., 48: T69-T74, 1970
A documentation of research methodology, oxygen pressure, specific
absorption rate of oxygen versus pH, influence of impurities and also
hydrodynamic conditions. Comparisons are made with existing kinetic
data showing reasons for the discrepancies.
X-18
-------
Whitney, R.P.; Han, S.T.; and Davis, J.L.
ON THE MECHANISM OF SULPHUR DIOXIDE ABSORPTION IN AQUEOUS MEDIA
Tappi, 36 (4): 172-5, 1953
A discussion of the absorption of sulphur dioxide in aqueous media
water, sodium carbonate solution, and sodium sulphite solution.
Yagi, S. and Inoue, H.
THE ABSORPTION OF OXYGEN INTO SODIUM SULPHITE SOLUTION
Chem. Eng. Science, 17: 411-21, 1962
An investigation of the kinetics of absorption and the mechanism of
absorption of oxygen in a sulphite solution.
Yeatts, L.B. and Marshall, W.L.
APPARENT INVARIANCE OF ACTIVITY COEFFICIENTS OF CALCIUM SULFATE AT
CONSTANT IONIC STRENGTH AND TEMPERATURE IN THE SYSTEM CaSO^-NaSO
H20 TO THE CRITICAL TEMPERATURE OF WATER. ASSOCIATION EQUILIBRIA
J. Phys. Chem., 73 (1): 81-90. 1969
X-19
-------
XI. GLOSSARY
Active Sodium - Sodium associated with anions involved in S02 absorption
reactions and includes aulfite, bisulfite, hydroxide and carbonate/
bicarbonate. Total active sodium concentration is calculated as
follows:
fNa+:iactive = 2 x (tNa2S03] + [Na2C03]) + [NaHS03] + [NaOH] + [NaHC03]
Active Sodium Capacity - The equivalent amount of S02 which can be theoreti-
cally absorbed by the active sodium, with conversion to NaHS03.
Active sodium capacity is defined by:
^Na+^active caPacity = [Na2S03] + 2 x [Na2C03] + [NaOH] + [NaHC03]
Calcium Utilization - The percentage of the calcium in the lime or lime-
stone which is present in the solid product as a calcium-sulfur salt.
Calcium utilization is defined as:
mols (CaS03 + CaSOiJ generated
Calcium Utilization = x 100%
mol Ca fed
Concentrated Dual Alkali Modes - Modes of operation of the dual alkali
process in which regeneration reactions produce solids containing
CaS03*%H20 or a mixed crystal containing calcium sulfite and calcium
sulfate hemihydrates, but not containing gypsum. Active sodium con-
centrations are usually higher than 0.15M Na+ in concentrated mode
solutions.
CSTR—Continuous Stirred Tank Reactor - A well-agitated, baffled reactor
vessel having a uniform concentration of species throughout. At
any time the concentrations in the effluent from a CSTR are equiva-
lent to those within the vessel.
Dilute Dual Alkali Modes - Modes of operation of the dual alkali process
in which regeneration reactions produce solids containing gypsum
(CaS04-2H20). Active sodium concentrations are usually lower than
0.15M Na+ in dilute mode solutions.
XI-1
-------
Sulfate Formation - The oxidation of sulfite. The level of sulfate forma-
tion relative to 862 absorption is given by:
mols SO^ oxidized
Sulfate Formation = x 100%
tnol SC>2 removed
Sulfate Precipitation - The formation of CaSOtf'XH20 in soluble solids.
The level of sulfate precipitation in the overall scheme is given
by the ratio of calcium sulfate to the total calcium-sulfur salts
produced :
mols
Sulfate Precipitation =
tnol CaSOY
IDS—Total Dissolved Solids - Equivalent to the sum of all soluble species.
TOS~Total Oxidizable Sulfur - Equivalent to the sum of all sulfite and
bisulfite species.
XI-2
-------
APPENDIX A
LIST OF LABORATORY EXPERIMENTS
A-l
-------
LIST OF LABORATORY EXPERIMENTS
Experimental
Series
1-10
PN1-PN17
A-N
C1-C32
35-45
46-65
66-83
T1-T30
Description
Sulfuric Acid Treatment, Continuous
Reactor
Concentrated Lime (Partial Neutrali-
zation) , Continuous Reactor
Concentrated Limestone, Batch
Reactions
Concentrated Limestone, Continuous
Reactor
Concentrated Lime (Effects of TOS
on Sulfate Precipitation), Contin-
uous Reactor
Dilute Lime Batch and Continuous
Reactor Runs
Concentrated Limestone, Supplemen-
tal Batch and Continuous Reactor
Runs
Proctor Compaction Runs for Studies
of Density, Strength, Permeability,
Leachability of Treated and
Untreated Product Solids
Dates Text
Conducted References
7/73-10/73
10/73-1/74
2/74
3/74-8/74
9/74-11/77
12/74-1/75
2/75-6/75
7/75-2/76
V-A
IV-B
VI-A
VI-A
IV-B
VIT-A
VI-B
VIII
A-3
-------
APPENDIX B
LIST OF PILOT PLANT RUNS
B-l
-------
TABLE B-l
test Series
000
100-400
I
UJ
500-600
Mode Test Set
Scrubber Evaluation 001-014
Sulfuric Acid Treatment 015-019
050-057
Concentrated Lime/Limestone 101-105
120-129
201-212
301-304
401-404
410-412
420-422
430
450-452
460-475
Dilute Lime 501-525
580-591
601-602
620
LIST OF PILOT PLANT RUNS
Type of Operation
Open-loop, scrubber only
Closed-loop, integrated system
Open-loop I^SOi^ reactor
Open-loop, concentrated lime, continuous reactor
Open-loop, concentrated limestone, continuous reactor
Open-loop, concentrated lime, batch reactor
Open-loop, concentrated lime, continuous reactor
Closed-loop, concentrated lime, (short-term)
Closed-loop, concentrated limestone (short-term)
Closed-loop, concentrated lime (long-term)
Closed-loop, concentrated limestone (long-term)
Open & closed loop centrifuge testing (concentrated lime & limestone)
Open & closed loop filter testing (concentrated lime)
Open-loop, continuous reactor
Open-loop, batch reactor
Closed-loop (short-term)
Closed-loop (long-term)
Dates Conducted
10-11/73
12/73
10-11/73, 1/74
1/74
7-8/75
2-3/74, 8/74
3/74
4/74, 9/74
5/74,12/75
10-11/75
1/76
4-5/74
7/74
6/74 - 1/75
12/74
1/75
4-5/75
Text Reference
III-B
V-B
V-B
IV-C
VI-C
IV-C
IV-C
IV-C
VI-C
IV-C
VI-C
IV-C
IV-C
VII-C
VII-C
VII-C
VII-C
-------
APPENDIX C
BATCH SETTLING TEST METHODS
C-l
-------
APPENDIX C
BATCH SETTLING TEST METHODS
C-l
-------
APPENDIX C
BATCH SETTLING TESTS—EXPLANATION OF PARAMETERS
In order to assess the dewatering properties of the solids produced in
the laboratory and pilot plant tests of various dual alkali regeneration
reactor schemes, batch settling tests were performed on the reactor
effluent slurries. The batch settling tests simply involved allowing
a sample of reactor effluent slurry, initially at uniform concentration,
to settle in a graduated cylinder. The settling produced a slurry-
supernatant interface (meniscus). The position of the meniscus was
recorded as a function of time and the meniscus/time data were then
evaluated as described by Fitch12to determine settling parameters that
would quantify the settling behavior and provide a basis for comparison
of the quality of solids produced under the various reactor conditions
tested.
The three parameters determined from the batch tests performed in this
program were the bulk (initial) settling rate, the maximum solids flux,
and the density or insoluble solids content of the settled, partially
compacted solids approximately two hours after the completion of the
settling test. The bulk settling rate was estimated from plots of the
meniscus position versus time. Three different types of settling plots
can be employed: log-log (logarithm of meniscus position versus the
logarithm of time), semi-log (logarithm of meniscus position versus time),
and linear. In any of these plots the curves typically exhibit three
characteristic regions—a constant rate section at the beginning of
settling, followed by a first falling-rate section (transition zone),
and finally a second falling-rate section (compression zone). Figure C-l
shows these three regions in a semi-log plot of settling data taken during
the testing of the ADL two-stage reactor in the concentrated lime mode
(run 402).
The constant rate section of the plot is frequently referred to as "free
settling". During free settling, the rate of meniscus fall is a function
of the slurry concentration and the type of solids (density, particle
size, and crystalline morphology) and, to a lesser extent, the density of
the mother liquor. For a given solids material, say CaS03/CaS04 mixed
solids, the free settling behavior will be determined primarily by the
particle size distribution and the slurry concentration. The rate of
fall during free settling is commonly called the bulk (or initial) set-
tling rate and is a principal parameter in the sizing of clarifiers.
The first falling-rate section is the result of the collapse of solids
against the bottom of the column. Solids begin to back up, and the zones
of free settling concentrations cannot transmit solids at the rate at
which they are settling in from above. The time at which such transition
settling occurs and its duration is strongly a function of the quantity
of solids in the initial slurry and the rate at which they settle. As
shown in Figure C-l, one of the settling curves exhibited essentially no
transition zone settling.
C-3
-------
Conditions:
Reactor System - Arthur D. Little, Inc., (Run 402)
[SO^] =0.8 M
Susp'd Solids = 33 gms/liter
Sp. Gravity = 1.105
Cylinder K = 840
3 Sets of Data Taken During Run
100
I
J
1
10 12 14 16 18 20
Settling Time (min)
22 24
26 28
FIGURE C-1 REPRESENTATIVE SETTLING CURVE -
CONCENTRATED LIME MODE
C-4
-------
at th^T^ rg^ SeCtl°n ±S the Period when solids ^ compacting
at the bottom of the column. The point of discontinuity between the
first and second falling-rate sections of the curve is called the "com-
pression point '. This is a critical parameter in the determination of
the maximum solids flux that can be passed through a thickener. Hence,
the Compression point is the basis for sizing thickener areas. In a
semi-log plot the compression point time is at the intersection of the
tangents drawn through the first and second falling-rate sections of the
r*iiif*\T£*
curve.
The maximum solids flux for sizing the thickener is expressed in terms
of the compression time as follows:
where Gt is the flux in tons of solids /(sq. ft.) (day); W is
grams of solids in the batch tests; tu is the "underflow"
time; and K is the graduated cylinder constant (cc/ft. of
cylinder height). Here, tu is the time at which the desired
underflow concentration is achieved and is related to the
compression point time by the following equation:
0 (Hc - Hu)
(C-2)
2.3 (Hc - HJ
where tc is the compression point time in minutes ; Hc is
the meniscus height at the compression point; Hu is the
meniscus height at the desired underflow concentration;
HOO is the final meniscus height; and 0 is the time interval
required for the tangent line constructed to the first
falling-rate section to cover one log cycle.
Assuming that the desired underflow concentration is equal to that achieved
at the compression point — a reasonable assumption both for designing thick-
eners as well as comparing the quality of solids from different tests--
then the maximum solids flux becomes:
45W ,„ ov
Gt " ^K ' (C~3>
The third parameter, the density of the settled solids, is probably the
best measure of the filterability of the solids, since it is an indica-
tion of how well the particles compact under the force of gravity. The
density of the settled solids is estimated at H. (actually at two hours
following the compression point time). The density is calculated using
the following equation:
C-5
-------
w
Settled Density: ~ r~~ , (C-4)
where V^, is the volume of the settled solids; and ps and p£
are the specific gravities of the solids and liquor, respec-
tively.
Table C-l lists the values of the three parameters discussed above calcu-
lated for the settling curves shown in Figure C-l.
C-6
-------
TABLE C-l
SETTLING PARAMETERS FOR CURVES IN FIGURE C-l
Curve
1
2
3
Bulk Settling Rate
(ft/min)
0.21
0.16
0.12
Maximum Solids Flux, Gt
(tons/ft^ day)
0.21
0.21
0.21
Density of Settled Solids
(wt%)
22
22
25
o
-------
APPENDIX D
SULFURIC ACID MODEL
D-l
-------
'APPENDIX D
SULFURIC ACID REACTOR MODEL
A model of the sulfuric acid reactor system was developed to provide a
method for estimating reasonable sulfuric acid reactor operating condi-
tions for pilot plant testing, by determining the sulfuric acid require
ments in different dual alkali system applications.
The model is based upon the following overall reaction:
2CaS03 • %K2Q + 3H20 -> 2CaS01| • 2H20 + 2NaHS03 (D-l)
Since this reaction proceeds at pH's below 4 (in order to dissolve calcium
sulf ite) , sulfuric acid is also consumed in neutralizing the alkalinity
in the cake and acidifying the incoming liquor. These neutralization and
acidification reactions are included in the model as follows:
Na2C03 + H2S04 -»• Na2S04 + C02t (D-2)
2NaOH + H2S04 -> Na2S04 + 2H20 (D-3)
2Na2S03 + H2SOtt •* Na2SO^ + 2NaHS03 (D-4)
CaC03 + H2SOi+ + H20 -»• CaS04 • 2H20 + C02+ (D-5)
Ca(OH)2 + H2SOit -> CaSO^ • 2H20 (D-6)
All of these usages of sulfuric acid are parasitic in the sense that they
do not reduce the sodium sulf ate content of the system liquor.
ORIGINAL MODEL
The model is formulated to determine the amount of sulfuric acid required
to neutralize all alkalinity value in the slurry feed and adjust the pH
into the 2 3-3 3 range. The model assumes that chemical equilibrium is
achieved throughout the system. Levels of sulfate and calcium in solution
in the reactor effluent are determined from an estimate of the apparent
solubility product for gypsum (K - = [Ca++] x [SO*] with concentrations
in mols per liter). The value ofPKSp- used is the average of the values
calculated from the laboratory data for the range of ionic strengths
expected in the pilot plant operations. In this model, Ksp^ is an input
and can be adjusted for any desired ionic strength.
D-3
-------
In order to adjust for pH, it has been simply assumed in the original
model that enough sulfuric acid is added to convert up to 20% of the
system TOS to H2SOs (depending upon the operating pH) . This is a rough
estimate based upon laboratory results, and is used because of the lack
of accurate estimates of the dissociation constant for sulfurous acid at
high ionic strengths. No consideration has been given in this original
model to the presence of H2SO^. At the pH's in most of the tests per-
formed, the amount of HSO^ should be less than 5% of the total S(VI)
species.
The model is based upon treatment of settler underflow and scrubber bleed
in a system similar to that illustrated in Figure D-l, where code numbers
for the various streams and their components are as shown. For example,
X18,5 w°uld be the sodium sulfate concentration (in molarity) in the
liquid fraction of the settler underflow slurry.
The model involves four rather simple relationships shown below. Once
all inlet concentrations have been converted to mols per liter of liquid
phase, the model solves for all flow rates in liters per hour to keep up
with the given oxidation rate producing S mols of "fresh" sulfate per
hour. The performance of the sulfuric acid reactor is then defined in
terms of the soluble sulfate removed per mol of sulfuric acid used. The
original model applies only to the CaS03~limited regime.
Basic Relationships
Letting S = Oxidation rate of TOS within system in mols/hr.
1. CaS03 req'd = S + (H2SOit)added ~ (H2SOit for CaC03 and Ca(OH)2)
+ (Dissolved Ca Out Less Ca from CaSOi^ in Settler Underflow)
Ql8x18,12 = S + Q2i).x24j6 - Qie(xi8,10 + X18,10) + (k^25 ~ Ql8x18,13J
where kx = [Ca++][SO^] , apparent solubility product
k2 = [SOlHreactor effluent
2. H2SOtt req'd = (Acid to neutralize OH~, SOf, COf, CaC03 , and Ca(OH)2)
+ (Acid for precipitation of
+ (Acid for pH adjustment)
D-4
-------
FIGURE D-l
SCHEMATIC OF lSOif REACTOR
tf
Ui
H20
Settler Underfl
H?SOu
Scrubber Effl.
H2SOif
REACTOR
To
Centrifuge
H = mass flow
Q = vol. flow
I = molarity
W = Wt. fraction
Components:
1 - H20
2 - Na2S03
3 - NaHS03
4 - N2S03
5 - Na2SOit
6 - H2SOit
7 - NaOH
8 - Na2C03
9 - NaHC03
10 - CaC03
11 - Ca(OH)2
12 - CaS03 - 1/2 H20
13 - CaSOi^ - 2 H20
14 - Impurities in solids (ash, etc.)
-------
6 = (°-5Ql8x18,7 + 0.5Qi8x18>2 + °-5QlOx10,2 + Ql8x18,8 + Ql8x18,10
+ Ql8x18,ll) + (°-5Ql8x18,7 + 0.5Q18x18j2 + 0.5Q10xi0,2
+ Ql8x18,8 + Ql8x18,5 + QlOx10,s) + k3(Q18x18,2 + QlOx10,2
+ QlOx10,3 + Ql8x18,3 + Ql8x18,2)
where k3 = a factor to allow for the formation of H2S03 and
the precipitation of the SO^ added to get the H2S03.
3. Soluble SO^ Feed = S + (Soluble SO^ effluent in excess of CaSOi^ from
settler underflow)
Ql8x18,5 + QlOx10,5 = S + (k2Q25 ~ Ql8x18,is)
4. Volume of Liquid Reactor Effluent = (Volume of Settler Underflow)
+ (Volume of Scrubber Effluent)
+ (Volume of Sulfuric Acid)
- (Volume Increase of Solids in Effluent)
- (Apparent Volume Disappearance of H20
taken up as water of crystallization)
F k4 172(x18>10 + xi8,ll + X18,12 + X18,13) + 60x18,l»t
Q25 " Ql8 + - Ql8 + -- -
1 ~ F X2l|»6 (2. 34) (1000)
18(X18,10
1000
where F = Qio/(Ql8 + Qio)
X18,7 + X18,2 + 2x18,8 + X18,10 + X18,5
+ k3(x!8,2 + X18,3 + X18,12)
X10,2 + Xi0,5 + k3(xio,2 + X10,3)~|
1 - F L 'J
172 = MW of CaSOlf-2H20
60 = MW of Si02 ("impurity")
18 = MW of H20
2.34 = Sp. Gr. of mixed CaSO^^H^ and of impurity
D-6
-------
It can be shown that the above equations all condense to provide the key
flow rate for stream 18 (the settler underflow required):
+ (Jk?~- ^
Ql8 = S —'
2k8
where
kj, k2, ka, k4, F, and S have been defined and
F k4 172 x18>10 + x18>12 + x18>13 + 60x18jll+
1 - F x2Lfj6 (2.34)(1000)
18 X18,10 + ]
1000
k5 = k^ - x18jlo - x18)13
F
ky = X18,5 + X18,13 + -
1 - F
2
k8 = k7x18}i2 - kek7 ~ klk5
kg = k6 - k7 -
Once Q18 is known all the other flows follow from the relations under (4)
above and the efficiency can be calculated:
mols N32SOit removed
Efficiency = - ~
mol H^SOij added
The degree of success of the sulfuric acid treatment system was thus
evaluated in terms of the ratio of NazSO^ removed (from that originally
present in the scrubber bleed and thickener underflow) to the amount of
sulfuric acid fed. Obviously it is desirable that this ratio be as high
as possible, realizing that it must be below unity. The model was used
to estimate the effects of different process variables on this efficiency
ratio. Variables studied included: lime/limestone utilization, solids
concentration in the underflow slurry-scrubber bleed composition, and the
ratio of scrubber bleed to thickener underflow.
D-7
-------
The results of these parametric studies were used as a guide in choosing
satisfactory regimes of reactor operation for pilot plant testing. As an
example of the use to which it was put, see Figure D-2, in which the
effect of high CaCOs content in the feed solids (low lime/limestone utili-
zation) is shown to be quite severe in limiting E2SO^ reactor efficiency.
Note that since the liquor composition in this example is the same for
both the scrubber effluent and the settler underflow, proceeding to the
right is merely equivalei t to lowering the percent solids in the under-
flow, starting from 20% solids at the left and ending with about 11%
solids on the right. It is clear in this example that at 20% solids
there is already enough Na2SOt,. in the solution to be equivalent to the
CaSOs present, and that introducing more solution simply increases the
volume of reactor effluent, raising the reactor effluent sulfate loss and
lowering the efficiency as shown.
MODEL REFINEMENTS
While the original model was quite accurate in the operating regime for
which it was designed, it was limited, in that it:
• was restricted to the calcium sulfite-limited regime;
• assumed a split (20%) for H2S03/TOS;
• required an independent estimate and input of the value of
the apparent Ksp- for gypsum;
• used a simple adjustment factor to account for changes in
water of crystallization; and
• did not take into account the formation of HSO^.
A more general model can be developed to overcome these deficiencies.
The model would be based upon the neutralization equations of the simpli-
fied model, but would utilize the method of Kusik and Meissner7 to
predict the equilibrium composition of the effluent liquor from estimates
of appropriate activity coefficients. The activity coefficient estimates
would be used to calculate the dissociation constants of sulfurous acid
(H2S03), bisulfate (HSOt^), and the solubility product of gypsum. Such a
model would apply to all modes of operation at pH's ranging from approxi-
mately 2.0 to 4.0.
The reaction equations appropriate to the refined model are as follows:
cof + (H2soi,.) •* soi; + H2o + co2t (D-?)
20KT + (H2SOit) -> S0~ + 2H20 (D_8)
2SO| + (H2S0lt) -> 2HS03 + S0° (D_9)
CaC03 + (H2SOi,.) -»• Ca++ + SO^ + C02f (D-10)
Ca(OH)2 + (H^Oi,) •*• Ca++ + SO^ + 2H20 (D-ll)
2CaS03 • %H20 + (H2SOit) •*- 2Ca++ + 2HSO| + S0~ + H20
D-8
-------
Liquid Compositions:
Scr. Effluent Settler U'flow
Na2S04 .750 M .750
Underflow Contains 20% Solids with
CaS03/CaS04 =
= 0.004, k3 = 0.20
10%CaC03 In Solids
20 30
Scrubber Effluent
Scrubber Effluent + Settler Underflow
FIGURE D-2 EFFECTS OF LIME/LIMESTONE
FltoUHC u UT,UZATION AND FEED SOLIDS CONCENTRATION
ON THE SULFURIC ACID REACTOR EFFICIENCY
D-9
-------
The equilibrium equations governing the composition of the effluent solu-
tion would be:
[H2S03]
[HSOIJ]
where, [H+] = 10 pH, and
y = ionic strength
The success of this simulation is, of course, dependent on the validity
of the assumptions and the functional relationships derived for the above
equilibrium equations. As a first approximation, it can be assumed that:
• The ionic strength of the liquor will be between 1 and 4.
• The predominant species (dissolved compounds) in solution
are Na2S04 and NaHSOs . The reduced activity coefficients
of these species are about equal and values for Na2SO^ can
be used.
• Other ionic species in solution in much lower concentra-
tions are HSOi* and SOJ and trace value activity coeffi-
cients can be used for these electrolyte species.
• The amount of undissociated H2S03 is relatively small
and its activity coefficient is approximately 0.1
The equilibrium equations then become:
[H+][HSOi]
0.00576 y
[H2S03]
:i
= 0.047
[Ca++][SO=] * 0.003e°'2^ (for 1 < y < 5, based
upon laboratory results)
where all concentrations are expressed in mols/liter.
D-10
-------
Given the desired pH of operation and the compositions of the feed streams
(filter cake or underflow slurry, sulfuric acid and scrubber bleed), these
equilibrium equations can be solved in conjunction with the neutraliza-
tion reaction equations and overall material balances by an iterative
procedure to determine the effluent liquor composition and reactor
efficiency (with respect to utilization of sulfuric acid). While the
model was not programmed for the computer, it was checked against the
results of a number of representative runs made in the pilot plant.
Table D-l shows a comparison of pilot plant results with those predicted
using both the original and refined models. It has been assumed in using
the models that there is no off gassing of S02, no oxidation of sulfite,
and that all calcium sulfite is dissolved (which has been verified by
pilot plant tests).
In general the predicted efficiencies using the original model tended
to fall below the observed reactor efficiencies, while those of the
refined model tended to exceed the observed efficiencies. In both cases
this appears to be due to the inaccuracies in the quantity and composi-
tion of the feed solids. Predicted effluent concentrations of TOS based
upon the feed liquor composition were consistently higher than observed
values. Thus, the models are estimating higher levels of CaSC>3 dissolving
than actually occurring. With the original model, where the additional
acid required to form l^SOs is estimated as a fraction of the total TOS
present, this tends to overestimate acid requirements.
For the refined model where estimates of H2S03 are based upon the
observed pH, the higher levels of TOS have less of an effect on the
acid requirement, but do calculate higher levels of calcium sulfite
available to precipitate sulfate (and thereby increase efficiencies).
This confirms the results of the variables study which showed that
estimates of reactor efficiency are particularly sensitive to the amount
of CaS03 fed to the reactor (in terms of either the level or composition
of insoluble solids).
D-H
-------
TABLE D-l
K3
COMPARISON OF PILOT PLANT RESULTS
WITH ORIGINAL
EXPERIMENTAL
FEED RATIO
(mols Na2SOit/mol CaSOs) Run No.
<0.5 056
055
>0.5 050
051
054
Effluent
PH [C
2.6 0.
2.7 0.
2.9 0.
2.8 0.
2.3 0.
Liquor
a++]
00 9M
015
017
014
008
[s
0.
0.
0.
0.
0.
AND REFINED MODELS
RESULTS
Composition
Of]
21M
13
17
29
60
1
0
1
0
0
0
TOS]
.95M
.05
.75
.59
.60
Efficiency
0.22
0.
0.
0.
-0.
37
43
28
01
PREDICTED EFFICIENCIES
Refined Mode
Efficiency
0.31
0.
0.
0.
0.
42
51
45
12
Original
Model
-
0.45
0.19
<0
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
jy>A-600/7-77-050b
3. RECIPIENT'S ACCESSION-NO.
LEANDSUBT1TLE FINAL REPORT: DUAL ALKALI TEST
AND EVALUATION PROGRAM; Volume H. Laboratory
and Pilot Plant Programs
5. REPORT DATE
May 1977
|6. PERFORMING ORGANIZATION CODE
AUii-tUHtS) f~t _, _. _, _— .
C.R. LaMantia, R.R. Lunt, J.E.Oberholtzer
E. L. Field, and J. R. Valentine '
8. PERFORMING ORGANIZATION REPORT NO.
ND ADDRESS
Arthur D. Little, Inc.
Acorn Park
Cambridge, Massachusetts 01240
10. PROGRAM ELEMENT NO.
EHE624
11. CONTRACT/GRANT NO.
68-02-1071
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final; 5/73-4/77
14. SPONSORING AGENCY CODE
EPA/600/13
^.SUPPLEMENTARY NOTES jERL-RTP project officer for this report is Norman Kaplan
Mail Drop 61, 919/549-8411 Ext 2915.
16. ABSTRACT
Volume II of the report covers Tasks I and II of a three-task program to
investigate, characterize, and evaluate the basic process chemistry and the various
operating modes of sodium-based dual alkali scrubbing processes. The tasks were:
I, laboratory studies at both Arthur D. Little, Inc. (ADL) and IERL-RTP; n, pilot
plant operations in a 1200 scfm system at ADL; and IE, a prototype test program on a
20 MW dual alkali system at Plant Scholz. Dual alkali system operating modes on high
and low sulfur fuel applications investigated included: concentrated and dilute dual
alkali systems, lime and limestone regeneration, and slipstream sulfate treatment
schemes. For each mode, the dual alkali process was characterized in terms of SO2
removal, chemical consumption, oxidation, sulfate precipitation and control, waste
solids characteristics, and soluble solids losses.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/
= EN ENDED
n Field/Group
Air Pollution
Alkalies
Sodium
Scrubbers
Desulfurization
Sulfur Dioxide
13. DISTRIBUTION STATEMENT
Unlimited
Form 2220-1 (9-73)
Calcium Oxides
Limestone
Sulfates
Tests
Pilot Plants
Air Pollution Control
Stationary Sources
Dual Alkali Process
13 B
07D 08G
07B
07A 14B
131
19. SECURITY CLASS (ThisReport)
Unclassified
21. NO. OF PAGES
20. SECURITY CLASS (Thispage)
Unclassified
22. PRICE
------- |