Environmental
Protection Agency
Electric Power
Research Institute
Topics:
Flue gas desulfurization
Sulfur oxides
Nitrogen oxides
Wet scrubbers
Dry scrubbers
Pollution control equipment
EPRI CS-3706
Volume 1
Project 982-31
Proceedings
November 1984
Proceedings: Eighth Symposium
on Flue Gas Desulfurization
Volume 1
Prepared by
Research Triangle Institute
Research Triangle Park, North Carolina
-------
Environmental Topics: EPRICS-3706
Protection Agency Flue gas desulfurization Volume 1
Sulfur oxides Project 982-31
flectric Power Nitrogen oxides Proceedings
esearch Institute Wet scrubbers November 1984
Dry scrubbers
Pollution control equipment
Proceedings: Eighth Symposium
on Flue Gas Desulfurization
Volume 1
Prepared by
Research Triangle Institute
Research Triangle Park, North Carolina
-------
REPORT SUMMARY
SUBJECTS SOX control / NOX control / Solid by-product disposal/reuse / Integrated
environmental control
TOPICS Flue gas desulfurization
Sulfur oxides
Nitrogen oxides
Wet scrubbers
Dry scrubbers
Pollution control equipment
AUDIENCE Environmental engineers / Generation planners
Proceedings: Eighth Symposium on Flue
Gas Desulfurization
Volumes 1 and 2
Timely exchanges of technical and economic information on
flue gas desulfurization (FGD) systems are essential to coal util-
ities that must meet strict emissions standards. These proceed-
ings constitute a valuable resource for utility, architectural-
engineering, and system-supplier personnel who must make
decisions about the design, installation, and operation of FGD
systems.
BACKGROUND
OBJECTIVE
APPROACH
KEY POINTS
Sulfur dioxide (SO2) emissions from coal-fired generating plants must be
carefully controlled to comply with government regulations. Compliance,
however, frequently means that utilities must install expensive and compli-
cated FGD systems. Therefore, utilities faced with limiting S02 emissions
need up-to-date information on this rapidly evolving technology in order to
select the most reliable and cost-effective process.
To provide a forum for exchanging information on the scientific, technical,
and regulatory developments related to SOa control.
The EPA and EPRI cosponsored a four-day symposium that featured the
presentation of 40 technical papers and a major panel discussion. Utility and
industrial users and representatives of FGD system suppliers, research insti-
tutions, and government agencies were invited to contribute papers empha-
sizing progress in SO2 control, recent experience with installed systems, and
pertinent test results. Some 730 persons attended.
In the keynote address, the executive director of the National Acid Precipita-
tion Assessment Program examined the program's purpose, scope, and
status and its focus on providing Congress with a better scientific basis for
legislative and regulatory decisions. Nine other sessions included such
diverse topics as economics, construction materials, absorbent injection,
dual alkali systems, flue gas treatment (combined SOX/NOX), FGD chemistry,
limestone and organic acid, waste disposal and utilization, and dry FGD
systems (both pilot- and full-scale). The role of the architect-engineer in
EPRI CS-3706S Vols. 1 and 2
-------
EPRI PERSPECTIVE
PROJECT
constructing FGD systems for utilities was the topic of a panel discus-
sion conducted by representatives from seven architectural-engineering
firms. The purpose of the architect-engineer, all agreed, was to serve as
an extension of the utility's own engineering staff.
Of eight flue gas desulfurization symposia that have been held, this is
the second EPRI has cosponsored. The meetings, held approximately
every 18 months, bring together FGD vendors, government regulators,
researchers, and architect-engineers. In this relatively new and contin-
ually changing technology, the symposia present an excellent opportu-
nity for a wide-ranging interchange of FGD information and experience.
The meetings are well attended, and the published proceedings provide
a comprehensive and useful source of up-to-date happenings in S02
control. The next symposium is planned for June 1985 in Cincinnati.
RP982-31
EPRI Project Manager: Thomas M. Morasky
Coal Combustion Systems Division
Contractor: Research Triangle Institute
For further information on EPRI research programs, call
EPRI Technical Information Specialists (415) 855-2411.
ORDERING INFORMATION
EPRI Members
Nonmembers
EPRI CS-3706 Vols. 1 and 2, Proceedings, November 1984.
V1,594 pages. V2,578 pages.
If this report is not available from your company libraries or your
Technical Information Coordinator, you can order it from
Research Reports Center
P.O. Box 50490
Palo Alto, CA 94303
(415) 965-4081
You can order this report in print or microfiche from
Research Reports Center.
Price: V1 $41.50; V2 $41.50 Overseas price: V1 $83.00; V2 $83.00
(California residents add sales tax.)
Payment must accompany order.
© 1984 Electric Power Research Institute, RO. Box 10412, Palo Alto, CA 94303. All rights reserved.
-------
Proceedings: Eighth Symposium on Flue Gas
Desulfurization
Volume 1
CS-3706, Volume 1
Research Project 982-31
Proceedings, November 1984
New Orleans, Louisiana
November 1-4, 1983
Prepared by
RESEARCH TRIANGLE INSTITUTE
Cornwallis Road
Research Triangle Park, North Carolina 27709
Compiler •
F. A. Ayer
Prepared for
Environmental Protection Agency
Office of Research and Development
401 M Street SW
Washington, DC 20460
Industrial Environmental Research Laboratory
Research Triangle Park, North Carolina 27711
EPA Project Officer
J. W. Jones
and
Electric Power Research Institute
3412 Hillview Avenue
Palo Alto, California 94304
EPRI Project Manager
T M. Morasky
Desulfurization Processes Program
Coal Combustion Systems Division
-------
ORDERING INFORMATION
Requests for copies of this report should be directed to Research Reports Center
(RRC), Box 50490, Palo Alto, CA 94303, (415) 965-4081. There is no charge for reports
requested by EPRI member utilities and affiliates, U.S. utility associations, U.S. government
agencies (federal, state, and local), media, and foreign organizations with which EPRI has an
information exchange agreement. On request, RRC will send a catalog of EPRI reports.
Research Categories: SOX control
NOX control
Solid by-product disposal/reuse
Integrated environmental control
Copyright © 1984 Electric Power Research Institute, Inc. All rights reserved,
NOTICE
This report was prepared as an account of work sponsored in part by the Electric Power Research Institute, Inc.
(EPRI). Neither EPRI. members of EPRI, nor any person acting on their behalf: (a) makes any warranty, express or
implied, with respect to the use of any information, apparatus, method, or process disclosed in this report or that
such use may not infringe privately owned rights; or (b) assumes any liabilities with respect to the use of, or for
damages resulting from the use of, any information, apparatus, method, or process disclosed in this report.
-------
ABSTRACT
These proceedings are of the Eighth Symposium on Flue Gas Desulfurization,
held November 1 to U, 1983, in New Orleans, Louisiana. The symposium was
sponsored by EPA's Industrial Environmental Research Laboratory, located in
Research Triangle Park, North Carolina, and the EPRI Coal Combustion Sys-
tems Division, located in Palo Alto, California.
The objective of the symposium was to provide a forum for supplier, user,
service, and regulatory groups to discuss the technical and regulatory
aspects of SC>2 control. The emphasis was on progress in SC^ control
technology, recent experience, and test results, not on future plans.
Volume 1 contains 24 papers from days one, two, and three, plus abstracts
from panel members. Volume 2 contains 16 papers from days three and four,
plus 6 unpresented papers.
iii
-------
PREFACE
These proceedings for the Eighth Symposium on Flue Gas Desulfurization
(FGD) constitute the final report submitted to the Industrial Environmental
Research Laboratory; EPA; Research Triangle Park (IERL-RTP), North
Carolina; and EPRI's Coal Combustion Systems Division, Palo Alto,
California. The symposium was conducted at the Sheraton. Hotel in New
Orleans, Louisiana, November 1 to 4, 1983.
The meeting served as a forum for the exchange of technical and regulatory
information and developments regarding systems and processes applicable to
utility and industrial boilers. At the opening session, the keynote
address examined the status and outlook for the National Acid Precipitation
Assessment Program: its present status and outlook for the future. Pre-
sentations were also made on the state of air quality legislation and regu-
lations, current and projected regulations of the Resource Conservation and
Recovery Act, and trends in commercial application of FGD technology. Sub-
sequent technical sessions dealt with economics, construction materials,
dry furnace absorbent injection, dual-alkali, flue gas treatment (combined
SO /NO ). Other sessions included FGD chemistry, limestone/organic acid,
waste disposal/utilization, and dry FGD systems, pilot plant test results,
and full-scale installations. Participants also discussed the role of
architect-engineer as middleman between the utility and FGD suppliers.
Representatives from electric utilities, state environmental agencies,
equipment and process suppliers, coal and petroleum suppliers, EPA and
other federal agencies, and research organizations attended the sessions.
The following people contributed their efforts to this symposium.
• Julian W. Jones, Chemical Engineer, Emissions/Effluent
Technology Branch, Utilities and Industrial Power Division,
IERL-RTP, Research Triangle Park, North Carolina; EPA
symposium general chairman and project officer
• Thomas M. Morasky, Manager, Reliability and Nonrecovery
Systems, Coal Combustion Systems Division, Palo Alto,
California; EPRI symposium general chairman and project
manager
• Franklin A. Ayer, Manager, Conference Planning Office,
Center for Technology Applications, Research Triangle
Institute, Research Triangle Park, North Carolina;
symposium coordinator and compiler of the proceedings
Thomas M. Morasky, Project Manager
Coal Combustion Systems Division
-------
TABLE OF CONTENTS
VOLUME 1
Section
Page
SESSION 1: OPENING SESSION
Julian W. Jones, Chairman
Keynote Address: National Acid Precipitation
Assessment Program: Status and Outlook .1-1
J. Christopher Bernabo
Remarks 1-21
Sheldon Meyers
The Resource Conservation and Recovery Act: Current
and Projected Regulations 1-27
Stephen A. Lingle
Trends in Commercial Applications of FGD 1-29
Bernard A. Laseke,* Michael T. Melia, and Norman Kaplan
SESSION 2: ECONOMICS
Thomas M. Morasky, Chairman
Computer Economics of Physical Coal Cleaning and
Flue Gas Desulfurization 2-1
Charles R. Wright,* Terry W. Tarkington, and
James D. Kilgroe
Economic Evaluation of FGD Systems 2-27
Jack B. Reisdorf,- R. J. Keeth, C. P. Robie,
R. W. Scheck, and Thomas M. Morasky
Estimating Procedure for Retrofit FGD Costs 2-47
R. R. Mora, P. A. Ireland," R. J. Keeth, and
T. M. Morasky
Comparative Costs of S02 Removal Technologies 2-63
John 0. Milliken
SESSION 3: MATERIALS OF CONSTRUCTION
Charles E. Dene, Chairman
''^Denotes speaker
VII
-------
Section Page
EPRI Research on Corrosion and Degradation of
Materials for FGD Systems 3-1
Barry C. Syrett
Simultaneous Design, Planning, and Materials of
Construction Selection for FGD Systems 3-15
Alex Kirschner, Norman Ostroff,* R. F. Miller,
and W. L. Silence
Acid Deposition in FGD Ductwork 3-47
Daniel A. Froelich,* Carl V. Weilert, and Paul N. Dyer
In Situ Evaluation of High Performance Alloys in
Power Plant Flue Gas Desulfurization Scrubbers 3-61
R. W. Schutz and Charles S. Young*
SESSION 4: DRY FURNACE ABSORBENT INJECTION
Randall E. Rush, Chairman
Results from EPA's Development of Limestone
Injection into a Low NO Furnace 4-1
X
Dennis C. Drehmel,* G. Blair Martin, and
James H. Abbott
Review of EPRI Research on Furnace Sorbent
Injection SQ^ Control 4-19
Michael W. McElroy
Direct Desulfurization Through Additive
Injection in the Vicinity of the Flame 4-31
M. Yaqub Chughtai* and Sigfrid Michelfelder
SESSION 5: DUAL ALKALI
Norman Kaplan, Chairman
Utility Double Alkali Operating Experience. . . 5-1
Dennis L. Glancy, Richard J. Grant, L. Karl Legatski,*
James H. Wilhelm, and Beth A. Wrobel
Pilot Evaluation of Limestone Regenerated Dual
Alkali Process 5-21
John C. S. Chang* and Norman Kaplan
SESSION 6: FLUE GAS TREATMENT (COMBINED S0x/N0x)
J. David Mobley, Chairman
Status of the DOE Flue Gas Cleanup Program 6-1
John E. Williams
^Denotes speaker
Vlll
-------
Section Page
Status of SO^ and NO Removal in Japan 6-37
Jumpei Ando
PANEL: THE ARCHITECT-ENGINEER - MIDDLEMAN BETWEEN
UTILITY AND FGD SUPPLIER 6-43
A. V. Slack, Chairman
Edward W. Stenby, Gene H. Dyer, Paul R. Predick,
Michael L. Meadows, Douglas B. Hammontree,
Christopher P. Wedig, and Richard Rao, Panel Members
SESSION 7: FGD CHEMISTRY
Dorothy A. Stewart, Chairwoman
Influence of Chlorides on the Performance of
Flue Gas Desulfurization 7-1
William Downs,* Dennis W. Johnson, Robert W. Aldred,
L. Victoria Tonty, Russell F. Robards,* and
Richard A. Runyan
Effect of High Dissolved Solids on Bench-Scale FGD
Performance 7-19
James B. Jarvis,* Timothy W. Trofe, and
Dorothy A. Stewart
Pilot Plant Tests on the Effects of Dissolved Salts
on Lime/Limestone FGD Chemistry 7-37
Dennis Laslo,* John C. S. Chang, and
J. David Mobley
Modeling of SO^ Removal by Limestone Slurry
Scrubbing: Effects of Chlorides 7-57
Pui K. Chan and Gary T. Rochelle*
Influence of High Dissolved Solids on Precipitation
Kinetics and Solid Particle Size 7-79
Frank B. Meserole, Timothy W. Trofe, and
Dorothy A. Stewart*
Effect of Limestone Grinding Circuit on FGD
Performance and Economics 7-105
J. David Colley,* 0. W. Hargrove, Jr., and
Dorothy A. Stewart
VOLUME 2
SESSION 8: LIMESTONE/ORGANIC ACID
J. David Mobley, Chairman
"'Denotes speaker
IX
-------
Section
Process Troubleshooting at a Utility Limestone
FGD System 8-1
J. David Colley, Robert L. Glover,
Temple E. Donaldson,* and Dorothy A. Stewart
Technical/Economic Feasibility Studies for Full
Scale Application of Organic Acid Technology for
Limestone FGD Systems ; 8-23
James C. Dickerman* and J. David Mobley
SESSION 9: WASTE DISPOSAL/UTILIZATION
James D. Kilgroe, Chairman
Full-Scale Field Evaluation of Waste Disposal
From Coal Fired Electric Generating Plants. . . 9-1
Julian W. Jones,* Chakra J. Santhanam, Armand
Balasco, Itamar Bodek, Charles B. Cooper,
John T. Humphrey, and Barry K. Thacker
Operations History of Louisville Gas & Electric
FGD Sludge Stabilization 9-25
Robert P. Van Ness,* John H. Juzwiak, and
William Mclntyre .
Coal Waste Utilization in Artificial Reef Construction 9-37
Jeffrey H. Parker,* Peter M. J. Woodhead, and
Dean M. Golden
Solid Waste Environmental Studies at Electric
Power Research Institute 9-49
Ishwar P. Murarka
Presented by Karen Summers
SESSION 10, PART I: DRY FGD: PILOT PLANT TEST RESULTS
Theodore G. Brna, Chairman
Current Status of Dry SO^ Control Systems 10-1
Michael A. Palazzolo,* Mary E. Kelly,
and Theodore G. Brna
Acid Rain Prevention Thru New SO /NO Dry
Scrubbing Process 10-23
Karsten S. Felsvang,* Per Morsing,
and Preston L. Veltman
'''Denotes speaker
-------
Section
Page
Process Characterization of SO^ Removal in
Spray Absorber/Baghouse Systems 10-41
Eric A. Samuel,* Thomas W. Lugar, Dennis E. Lapp,
Kenneth R. Murphy, Owen F. Fortune, Theodore G. Brna,
and Ronald L. Ostop
Dry Scrubber, Flue Gas Desulfurization on High Sulfur,
Coal-Fired Steam Generators: Pilot-Scale Evaluation 10-61
Bryan J. Jankura,* John B. Doyle, and Thomas J. Flynn
EPRI Spray Dryer/Baghouse Pilot Plant Status
and Results 10-81
Gary M. Elythe" and Richard G. Rhudy
SESSION 10, PART II: DRY FGD: FULL SCALE INSTALLATIONS
Richard G. Rhudy, Chairman
Field Evaluation of a Utility Dry Scrubbing System 10-109
Gary M. Blythe,* Jack M. Burke, Theodore G. Brna,
and Richard G. Rhudy
Overview and Evaluation of Two Years of Operation
and Testing of the Riverside Spray Dryer System 10-131
John M. Gustke, Wayne E. Morgan,*
and Steven H. Wolf
Design and Initial Operation of the Spray Dryer
FGD System at the Marquette, Michigan, Board of
Light and Power - Shiras #3 Plant 10-161
0. Fortune,* T. F. Bechtel, E. Puska, and J. Arello
Start-Up and Initial Operating Experience of the
Antelope Valley Unit 1 Dry Scrubber 10-181
Robert L. Eriksen,* Frederick R. Stern,
Richard P. Gleiser, and Stanley J. Shilinski
Characterization of an Industrial Spray Dryer at
Argonne National Laboratory 10-199
Paul S. Farber* and C. David Livengood
UNPRESENTED PAPERS
An Economic Evaluation of Limestone Double Alkali
Flue Gas Desulfurization Systems 11-1
Gerald A. Hollinden, C. David Stephenson, and
John G. Stensland
^Denotes speaker
XI
-------
page
Developments and Experience in FGD Mist Eliminator
Application ........................... 11-39
Richard T. Egan and William Ellison
FGD Gypsum: Utilization vs. Disposal .............. 11-61
William Ellison
Operating Experience with the Chiyoda Thoroughbred 121
Flue Gas Desulfurization System ................. 11-75
Seiichi Kaneda, Mitsuhiro Nishimura,
Hitoshi Wakui, Ikuro Kuwahara, and
Donald D. Clasen
Operation Experience with FGD Plant II at
Wilhelmshaven Power Plant, West Germany ............. 11-91
B. Stellbrink, H. Weissert, and P. Kutemeyer
The SULF-X Process ........................ 11-111
Edward Shapiro and William Ellison
APPENDIX: ATTENDEES ....................... A-l
XII
-------
SESSION 1: OPENING SESSION
Chairman: Julian W. Jones
Industrial Environmental Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC
-------
KEYNOTE ADDRESS: NATIONAL ACID PRECIPITATION
ASSESSMENT PROGRAM: STATUS AND OUTLOOK
J. C. Bernabo
-------
National Acid Precipitation Assessment Program; Status and Outlook
J. Christopher Bernabo
Executive Director
National Acid Precipitation
Assessment Program
722 Jackson Place, NW
Washington, DC 20506
The National Acid Precipitation Assessment Program is designed to
successively improve our understanding of the causes, effects and possible
solutions to the acid rain problem. The Program includes research, monitoring
and assessment activities that emphasize the timely development of a pro-
gressively firmer scientific basis for decision making by the Congress,
regulatory agencies, private sector managers, environmental groups, and the
public. The National Program consists of over 200 projects and hundreds of
scientists in government, academia, and the private sector.
LEGISLATION AND ORGANIZATION
The Interagency Task Force on Acid Precipitation was established by the
Acid Precipitation Act of 1980 (Title VII of the Energy Security Act: P.L.
96-29*0 to oversee the planning and implementation of the National Acid
Precipitation Assessment Program. Prior to the passage of the Acid Precipi-
tation Act of 1980, several Federal agencies were sponsoring acid rain
research, but more in a piecemeal fashion than a coordinated program.
The Acid Rain Coordinating Committee (ARCC), which was established in 1980,
was an attempt to organize the ongoing Federal efforts. After the Act passed,
the ARCC was reconstituted,and provided the foundation upon which the statutory
National Acid Precipitation Assessment Program was established.
The Task Force is jointly chaired by the Environmental Protection Agency
(EPA), the Department of Agriculture (USDA), and the National Oceanic and
Atmospheric Administration (NOAA). The Council on Environmental Quality
serves as the Executive Secretary of the Task Force. The remaining members
include one high-level representative from each of the other eight Federal
entities in the Program: the Departments of Energy (DOE), Interior (DOI),
Commerce (DOC), Health and Human Services (HHS), and State (DOS); the National
Aeronautics and Space Administration (NASA); the National Science Foundation
(NSF); and the Tennessee Valley Authority (TVA). The Task Force also includes
the directors of Argonne, Brookhaven, Oak Ridge and Pacific Northwest National
Laboratories, as well as four Presidential appointees.
1-1
-------
The Task Force itself functions as the "board of directors" for the
National Program by setting the research goals, identifying the projects
needed to meet these goals, and deciding which agencies are best suited to
conduct the necessary work.
In addition, the primary responsibilities to the Task Force are to:
• Plan and manage the National Acid Precipitation Assessment Program.
• Provide annual reports on the National Program's progress.
• Produce an annual interagency budget for the National Program.
• Coordinate the National Program with the research and monitoring
activities of the private sector groups, environmental organizations,
states, and other nations.
• Maintain an inventory of Federally-funded acid deposition research
projects.
• Disseminate research results, and assessments of their implications.
The Task Force has 10 working-level Task Groups, one for each of the
National Program's nine research categories and one for international
activities. These technical groups include program managers and experts from
all the participating Federal agencies and National Laboratories. They are
responsible for the detailed planning and research in their assigned areas.
The Task Groups with their Coordinating Agencies are:
Task Group Coordinating Agency
A. Natural Sources NOAA
B. Man-made Sources DOE
C. Atmospheric Processes NOAA
D. Deposition Monitoring DOI
E. Aquatic Effects EPA
F. Terrestrial Effects USDA
G. Effects on Materials and DOI
Cultural Resources
H. Control Technologies EPA
I. Assessments and Policy Analysis EPA
J. International Activities DOS
The Research Coordination Council (RCC) oversees and integrates the
efforts of the various Task Groups, and develops draft reports, program plans,
budgets and other recommendations for consideration by the full Task Force.
The joint chairs designated the Task Force's executive director to chair this
council. The RCC includes leaders of all the Task Groups, the chairperson of
the National Laboratory Consortium, and other appropriate agency representa-
tives.
1-2
-------
Budget Process
The Task Force developed its first interagency budget for the Program in
January, 1981 and submitted it to Congress as part of the President's FY 1982
budget request. The role of the Task Force in planning the interagency budget
for the National Program is a highly effective and unique aspect of the
Federal effort. During this interagency budget process, the Task Force
receives initial guidance on the target funding levels for the total program;
it then advises the individual Task Groups of the anticipated increases (or
decreases). Using this as guidance, the Task Groups adjust their research
plans or suggest new initiatives, as appropriate, for the target funding
levels. The Task Groups present their budget package to the Task Force for
review and approval. From this the Task Force develops and submits its inter-
agency budget recommendations; the agency budget requests are thus reviewed as
part of an integrated multiagency program.
By working together through the Task Force, the agencies have established
a research program that addresses national needs while building on the research
expertise of the individual agencies. This strong interagency planning process
ensures an integrated and comprehensive program, with each agency contributing
to specific aspects of the overall national effort.
The interagency budget for the National Program does not include Federal
funds for control technology hardware development because control technology
activies are conducted under preexisting programs at EPA, DOE, and TVA. The
National Program is actively coordinating with the relevant ongoing Federal
control technology efforts to ensure that the concerns relative to acid
deposition are addressed.
The Task Force is vigorously implementing the comprehensive national
research program. The interagency budget for the National Program was $17.4
million in FY 1982, $22.3 million in FY 1983 (Figures 1, 3, & 4), and $27.5
million has been requested for FY 1984 (Figure 2). Federally-funded acid
deposition research has doubled from about $11 million in FY 1980 to $22.3
million for FY 1983, and the President's budget for FY 1984 requests a further
24% increase in funding (Figure 5). The National Program is producing results
at an increasing rate, and disseminating them as soon as they become available.
Annual Report
By law the Task Force is required to issue an annual report to the
President and Congress describing the progress of the National Program. The
Task Force's first annual report was issued in January, 1982; with research
barely underway, the report discussed primarily the development of the manage-
ment structure, and the initial implementation of the Program.
1-3
-------
ACU mciriTAnoBi AssxssMnrr noctAM
rigure * rtl983
(t la thooaanda)
A.
1.
C.
D.
E.
r.
c.
R.
I.
natural Sonrcea
Man-made Soarcea
Ataoapherlc Proceeses
Deposition Monitoring
Aquatic Effect*
Terreatrlal Effect*
Effects oa Materials
Control Technologies
Assessment* aad Policy
Analysis
TOTALS: f
DOA 001
EPA n n BQAA DOS rs ~€s rv
700
1050 300
3838 1030 274 30
1143 404 143 300 666 100 1449
1933 210 330 148 443 75
1468 946 1238 30 229 486
410 333 30
(7600)* (6300)"
1370 220
1330 1613 1062 2430 73
11.436 2963 2230 2060 3567
TOTAL
f 700
1350
5232
4409
3363
4437
995
(14,100)'
17*0
1 22,276
Control Tcchnologlaa flfurca aot Included In total. Ihaao fond* for f*a«ral d*valopa*nt of SO, and HO control
hardwire arc appropriated wider other preeilatlnf prograeja. The Control Technologies Task Croup coordinate* these
efforts vlth the asaesaaent and research activities of the Rational Program.
-------
Figure 2
•ATUMAI. ACIB mcinunoi ASSESSMENT KOGUH
i
Ul
Propoaed revlalona to tlM Prealdaat'e Indict Bequest (May 1983)
A.
>.
C.
0.
B.
r.
c.
B.
i.
OQA
MA 8E tS
Mature! Soureee
Man-eud* Souroi 1090
Ataoapherle Proceaae* 3230
Deposition Monitorial . 171* 274 U«
Aquatic Effect! 228$ 210
Terreatrtel Effect* 1468 946 12M
Ef fecte on Materiel! 605
Coatrol TeclUMloglee (4100) •
AiM*ra*nt* end Folley 1500
Aiulytl*
taktotelei 1220 1614
TOTALIs 8 11,987 MM
n 1984
(t la thoueaadi)
DOI
•OAA DOE PS GS PH ILM TOTAL
933 9 9SS
260 1350
1020 170 6SAO
1130 942 122 1870 80 6333
330 148 445 123 3763
30 229 466 4437
740 130 1493
(7000) • (11.100) •
1173 2673
1239 2931 123 80
9103 3247 4393 8 I7.S68
Control T«ehnol|l*e flgurce ere not Included tn total*. Theee fund* for g«n«r»l development of SO. and M control
hardware ere appropriated under other pre-eilatlng profrae*. The Control Tachnolo|ia* T**k Croup Zoordlnat** theae
effort* «itk the reaearch and aiaeaneat ectUltlea mf. thn ••tleul Piregrea.
-------
Figure 3
National Acid Precipitation Assessment Program
EPA
$11,436K (51
3%)
DOA
$2,963K (13.3%)
NOAA
$2,250K (10.1%)
DOI
$3,567K (16%)
DOE
$2,060K (9.2%)
TOTAL: $22.3 Million
Fiscal year 1983 funding by agency.
1-6
-------
Figure A
National Acid Precipitation Assessment Program
Terrestrial Effects
$4,437K (19.9%)
Aquatic Effects
$3,363K (15.1%)
Effects on Materials
$995K (4.5%)
Assessments
$1,790K (8%)
Natural Sources
S700K (3.1%)
^Man-made Sources
.4. „ . -. \ » / $1,350K (6.1%)
Deposition Monitoring '
$4,409K (19.8%)
•x^ . . • . i- • •: ^
Atmos. Processes
$5,232 (23.5%)
TOTAL: $22.3 Million
Fiscal Year 1983 funding by research category.
1-7
-------
Figure 5
FEDERAL ACID DEPOSITION RESEARCH FUNDING HISTORY
i
oo
10 -
27.6
NATIONAL PROGRAM
-------
In June of this year the National Acid Precipitation Assessment Program
issued the FY 1982 annual report to the President and Congress on its first
full7year of research. The report summarizes the status of scientific know-
ledge, the major research findings and accomplishments of the National
Program's first year, and the outlook for continued progress. As the National
Program progresses, future annual reports will provide new findings and in-
creasingly better information to assist decision makers and the public in
continuing to resolve the acid rain issue.
NATIONAL PROGRAM MANAGEMENT ACTIVITIES
In addition to internal planning and management activities, the Task
Force meets with representatives of key non-Federal groups conducting acid
deposition research and monitoring activities. The Task Force is committed to
conducting a National Program that coordinates the Federal efforts with the
activities of the states, private sector groups, research institutes, environ-
mental groups, and other countries, especially Canada. Several specific steps
have been taken to develop and encourage such cooperation and more extensive
joint planning activities are anticipated. Such activities include:
National Plan Workshop
Besides a public review of the draft plan of the National Program, the
Task Force's April 1981 workshop of non-Federal experts began a dialogue with
state and private groups. The workshop report describes the participants'
ideas concerning how coordination and cooperation can best be accomplished.
The Task Force is implementing many of these suggestions and is actively
pursuing continued exchanges of information with non-Federal groups.
Annual Meeting
The Task Force holds an annual meeting for program participants to assess
research progress, to propose future work, and to assess the implications of
existing information. The meeting provides a mechanism to bring together the
full spectrum of research managers and scientists to fascilitate dialogue
about the National Program and foster integration of its various elements, and
thus enhance program unity.
The next annual meeting is proposed for January 16-19, 1981 with the
theme being "research to meet policy needs". This meeting will be an effort
to educate decision makers on the capabilities and the limitations of the
Natfonal Program's research and assessment activities, as well as a prime
opportunity to get feedback on the needs of the decision makers. In addition
to members of the Task Force, the invited participants will include senior
congressional staff; representatives from states, private sector and environ-
mental organizations; and key scientists.
1-9
-------
Federal-State-Private Sector Coordination
The Task Force initiated and encouraged joint sponsorship of an Acid
Deposition Research Coordination Workshop in November 1982. The non-Federal
sponsors included the Electric Power Research Institute, the American
Petroleum Institute, the Gas Research Institute, the Coordinating Research
Council, and the National Council of the Paper Industry for Air and Stream
Improvement. Environmental groups and states were also invited to partici-
pate.
The purpose of the workshop was to bring together key Federal and
non-Federal managers of acid deposition research to review and discuss the
management and coordination of their research planning efforts. Specifically,
the two goals of the meeting were (1) to assess the scope of the nationwide
acid deposition research effort, and (2) to initiate a continuing process for
joint planning and coordination of Federal and non-Federal acid deposition
research efforts.
In addition to developing joint statements on research needs and dissem-
inating information, the state representatives agreed to organize by regions
and to define their specific areas of expertise. The Task Force agreed to
investigate ways to increase opportunities for state and private sector
participation in reviewing the National Program. Follow-up activities include
further discussion between the Task Force, and the private sector and state
representatives to work out the details.
Since the workshop, National Program representatives have participated in
several meetings with the states to help identify the specific research needs
of the region, and to encourage the states to develop research programs on
acid deposition. Initial plans nave also been made for a follow-up meeting
with the original participants and others.
National Program Research Peer Reviews
An integral part of the sound management of the National Program is
periodic peer reviews by a panel of outside experts. These detailed scien-
tific reviews allow the National Program's research to be examined in a com-
prehensive manner rather than as a series of discrete agency projects, and
provide the opportunity to assess the quality of the science. More speci-
fically, the reviews are designed to:
©provide an opportunity for National Program researchers to report on
progress, and to present research results and major findings;
o facilitate the exchange of information, and enhance cooperative
research among scientists involved in the Federal program and others;
e provide an opportunity for those outside the Federal government to
review and comment on the evolving National Program; and
• assist research managers in providing adequate program coordination and
research direction.
1-10
-------
U.S.-Canadian Research Program Cooperation
In April 1982 and July 1983, Task Force representatives met with the
Canadian Federal/Provincial Research and Monitoring Coordinating Committee
(RMCC) to expand bilateral scientific cooperation. Leaders of both the U.S.
and Canadian research efforts emphasized the mutual value of greater cooper-
ation and coodination of activities. The highlights of the July 1983 meeting
include synopses of all bilateral acid deposition research projects currently
underway, as well as listings of suggested projects to be undertaken jointly
by the two countries in the future. The Task Force representatives feel that
much of the suggested research is promising, and the National Program is now
using the list of suggested projects in setting its research priorities for
the FY 1985 budget process.
The 25 to 30 invited reviewers are chosen from the leading scientists in
the United States, Canada and other nations. They critically evaluate all the
research to ensure that each project contributes to the identified objectives,
milestones, and deliverables, and they assess the research progress and plans
as well as make recommendations to guide future research.
The review of aquatic and terrestrial effects research was held in Feb-
ruary 1983, and the review of sources, monitoring and atmospheric processes
research was held in August 1983. The review of assessment activities is
proposed for February, 198*4.
Inventory of Acid Rain Research Projects
Project descriptors for all the Federally-funded research were entered
into the computer this spring, and are presently being verified by each Task
Group leader. The inventory includes information on all ongoing projects with
the performer(s), duration, funding, study area, research goals, methodolo-
gies, and expected outputs. Witn the Federal research projects on-line, work
can begin in FY 1984 on the inventory of all state-sponsored research and
monitoring acitvities. The computer inventory will be maintained and updated
for use by the Task Force, and all interested parties. The private sector has
conducted a similar inventory of its activities; together with the Task
Force's effort, this should provide a complete catalog of all acid deposition
research and monitoring activities in the United States.
Information Dissemination
The National Program served as the catalyst and hosted a meeting on acid
rain information sources in October, 1983. This meeting brought together
sponsors, producers and users of bibliographies and research project inven-
tories on acid rain. Presentations were made on the status and outlook for
these services, with the focus on the scope and purpose, dissemination
availability, and identification of user needs. The interaction between the
participants was encouraging, and greater cooperation is anticipated for the
future.
1-11
-------
MAJOR RESEARCH QUESTIONS
The National Program includes research, monitoring, and assessment
activities to develop answers to major research questions. In addition to
identifying and reducing scientific uncertainties, the National Program is
organizing the scientific information to address policy questions. The Task
Force is establishing a systematic framework for evaluating costs and benefits
of alternative policy options for controlling man-made sources of acidity or
mitigating adverse effects.
Some of the overall research questions the National Program is specifically
addressing are given here with some indication of the type of work underway to
answer the questions.
1. Does acid deposition significantly affect aquatic resources, crops,
soils, forests, materials, or human health? Except for poorly-buffered lakes,
evidence on other actual impacts is inconclusive. The National Program
includes intensive laboratory and field studies to develop a more definitive
understanding of the potential for damage to sensitive resources.
2. Where are the sensitive resources? How much damage has occurred, or
how long will it take for damage to occur? Concern about acid deposition
centers around its potential and known effects. The National Program is
developing inventories of the aquatic, terrestrial, and cultural resources at
risk, and is surveying the resources that have already been affected by acid
deposition.
3. What is the relative contribution of wet versus dry deposition of
acids? Dry deposition may constitute up to 50 percent or more of acidifying
substances, but reliable methods for its routine collection do not yet exist.
Vigorous efforts are underway in the National Program to develop appropriate
techniques for adequately monitoring dry deposition on a routine basis. In
the meantime, interim methods for collecting dry deposition will be used to
provide the best currently available data.
4. What is the relative importance of local versus distant sources in
controlling the acidity of deposition at a given location? Neither models nor
measurements currently address this question with a resonable degree of
confidence. The National Program includes a broad atmospheric sciences effort
with models, experiments, and measurements in the field and laboratory
designed to improve our understanding of source/receptor relationships as
rapidly as possible. Information on the characteristics and magnitude of
man-made emissions is being redefined as well.
5. What is the relative contribution of man-made versus natural sources
of acid-forming materials? Significant natural sources exist, but their role
is uncertain. The amount of natural acid precursors arising from sources such
as oceans and marshes is being determined by the National Program to assess
their contribution to acidity over North America. The magnitude, distribution
and character of man-made emissions are also being investigated to more
accurately identify their contribution to acid deposition.
1-12
-------
6. What is the temporal and spatial distribution of acid deposition?
Monitoring efforts before 1977 were fragmented and inadequate for determining
trends. A National Trends Network with standardization and quality control
has been established to answer these questions. Using a variety of information
sources, research is also underway to improve monitoring techniques and to
decipher any past trends.
7. What changes in deposition would result from a given change in sulfur
dioxide, nitrogen oxides, organics, and/or primary sulfate emissions?
Scientists cannot precisely calculate the amount of emission reductions that
would be required to reduce acid deposition below a particular level within a
given area. We need to know more about atmospheric processes to accurately
quantify source/receptor relationships. The National Program's assessment
activities integrate information on sources and atmospheric processes to
address this question.
NATIONAL PROGRAM ACCOMPLISHMENTS
Some selected highlights of progress in the past two years (FY 1982 and
1983) for the National Program's research areas are as follows:
A. Natural Sources
« Delivered the budget survey of natural sulfur and nitrogen emissions
required for the transport/transformation modelers of atmospheric processes.
« Developed and tested instruments for reliable field measurement of
sulfur and nitrogen emissions from natural sources.
© Continued flux measurements of natural sulfur compounds in the ocean
that suggest that emissions from marine sources may contribute significantly
to atmospheric acidity in some areas.
B. Han-made Sources
« Completed preliminary study on emissions sources which suggested that
nearby petroleum combustion, ad well as local and distant coal combustion,
could significantly affect local acid deposition.
• Continued development of models to predict the cost and emissions
changes from utilities and industries based on various possible control strat-
egies.
• Continued to develop emissions inventories for testing the transport/
transformation models with improved spatial, temporal, and source-type
resolution, and initiated studies to better characterize the man-made sources
of acid precursor pollutants.
1-13
-------
C. Atmospheric Processes
0 Improved understanding of atmospheric transport, cloud processes and
scavenging processes.
© Initiated major field experiment to track the movement of tracer gases
released in the Midwest and Canada over hundreds of miles.
o Improved atmospheric models and applied them to gain a better under-
standing of the movement and conversion of pollutants, and the deposition of
acidic material.
® Improved empirical techniques for assessing source/receptor rela-
tionships, dry deposition of SOx and NOx gases and particles, and the trans-
port of sulfur and nitrogen across North America.
D. Deposition Monitoring
» Produced the first comprehensive maps describing the distribution of
major chemical species over North America.
® Established the National Trends Network for measuring background
quantities of wet acid deposition products.
• Developed prototype equipment and procedures for routine monitoring of
dry deposition.
o Improved methods for field checks of pH and conductivity on low-
concentration solutions.
E. Aquatic Effects
o Produced a nationwide map and regional maps of the Northeast, South-
east, South Central, and Upper Midwest, indicating areas where surface waters
are likely to be most sensitive to acidification.
« Began long-term study of causal relationships of acidic effects on
biota for twenty sensitive watersheds.
• Completed preliminary survey of drinking water in the Northeast that
indicates some reservoirs and their watersheds have limited ability to buffer
acid deposition.
• Took lake core sediment samples to examine historical patterns of
deposition.
• Improved understanding of factors controlling the susceptibility of
surface waters to acidification, and improved methods to quantify, assess, and
show trends of acid deposition effects.
1-14
-------
© Provided research information for predicting acidification of water
bodies, and for assessing potential effects of liming to restore or mitigate
acidification damage.
F. Terrestrial Effects
o Reported on the use of moaels to predict changes in forest growth due
to acid deposition.
-------
o Developed burner systems to decrease nitrogen oxides emissions by 75
percent in large coal-fired boilers.
© Began studies of coal cleaning practices, engineering costs of control
technologies, directly emitted acidic materials, integrated control systems,
and passive (non-hardware) control systems.
I. Assessments and Policy Analysis
» Prepared the "Critical Assessment Document: The Acidic Deposition
Phenomenon and Its Effects" with chapters reviewing acid deposition and its
impacts with contributions from 51 scientists from many institutions in this
country and abroad. The document is now undergoing final revision.
o Prepared advanced methods for integrated assessment linking all
components of transport within and exchanges among relevant systems and
receptors, over time and space.
« Developed analytical and statistical frameworks for increasingly
complex modeling assessments, including cost-benefit comparisons.
• Developed the preliminary version of a model for assessing sulfur di-
oxide emissions from the utility sector, and developed empirical assessment
techniques for use until reliable models of source/receptor predictions are
delivered.
o Began investigation of estimating uncertainty in models and the
implications for formulating control strategies.
o Assembled data bases on the quantity, sulfur content, heating value,
ash content and desulfurization potential of U.S. coals to aid in evaluation
of emission control strategies.
« Evaluated advanced simultaneous sulfur dioxide/nitrogen flue gas
treatment processes for cost and performance relationships, and identified
possible emerging processes such as electron beam irradiation.
• Improved wet flue gas desulfurization process by increasing efficiency
of removal, reagent utilization, reliability and waste product disposal.
e Investigated the effectiveness of spray dryer flue gas desulfurization
systems along with the bnefits of fabric filters and electrostatic
precipitators.
J. International Activities
« Continued to cooperate with Canadian counterparts for joint research
proposals.
1-16
-------
» Published "International Directory of Acid Rain Researchers", listing
1,000 scientists in 32 countries.
o Continued to compile international research information on acid
deposition activities worldwide.
More details of these and other recent accomplishments in the various
research categories will be contained in the FY 1983 Annual Report of the
National Acid Precipitation Assessment Program.
OUTLOOK FOR RESEARCH; FY 1985 to 1989
In the next several years the research under the National Program will
produce a number of important results. The following are some highlights of
specific accomplishments the National Program can produce by FY 1989.
A. Natural Sources
® Develop an experimental data base that characterizes the natural
emissions of sulfur, nitrogen, chlorine, and alkaline emissions (ammonia and
others) on both a national/global and regional scale.
B. Nan-made Sources
9 Update the emissions inventory to an FY 1984 base year.
9 Add hydrogen fluoride, hydrochloric acid, ammonia and other alkaline
substances to the inventory. Also add volatile organic compounds (VOC) by
reactivity class, and disaggregate amounts of sulfur and nitrogen compounds as
necessary.
0 Refine, test, and operate the man-made emissions model for the 1987
and 1989 Assessments.
C. Atmospheric Processes
e Provide a definitive mass balance"for compounds of sulfur and nitrogen
in North America.
e Provide a model capable of assessing localized (mesoscale) rates of wet
and dry deposition from nearby sources, and develop an advanced model capable
of improving assessment methods.
• Develop techniques for measuring dry deposition products that contri-
bute to acidification.
1-17
-------
o Conduct a major air transport experiment to help define long-range
patterns of distribution and rates of atmospheric conversion to acidifying
substances, for example, conversion patterns of SO and SCL downwind of major
industrial areas.
« Conduct basic research on cloud processes that affect formation of
acidifying substances.
D. Deposition Monitoring
« Addition of dry deposition monitoring into the National Trends Network
(NTN) protocol, as well as trace metals and select organic substances.
« Expansion of the NTN and Global Trends Network to provide more
reliable patterns of acid deposition.
E. Aquatic Effects
o Develop recommendations for liming acidified water bodies.
e Develop and validate biological response models assessing effects of
acidification over a range of loading rates and terrestrial conditions.
e Report on health effects, fish populations at risk, drinking water
responses, and waterfowl effects as these relate to acid deposition.
F. Terrestrial Effects
• Update reports on acid deposition effects on soils, especially soil-
nutrient changes, and mobilization of aluminum and other metals; also, forest
species effects; and responses of soybeans, corn and other crops.
• Complete a series of experiments and effects determinations to examine
the enhanced susceptibility of agricultural and forest crops to disease and
reduced yields.
G. Effects on Materials and Cultural Resources
e Provide damage functions with acceptable levels of confidence for
materials exposure to SOx, NOx and ammonia.
« Recommend protection methods for materials at risk as an alternative
to stringent controls.
» Complete materials inventories for the United States.
H. Control Technologies
• Report on control of directly emitted acidic materials, control of
volatile organic compounds, industrial process controls, and state-of-the-art
reports on control technologies.
1-18
-------
• Evaluate passive (non-hardware) control strategies, and combinations
of low cost control options, such as limestone injection multistage burner
system (LIMB) and load dispatching, to attain goals.
I. Assessments and Policy Analysis
o Develop sophisticated source/receptor assessment models, combining
control and effects mitigation to integrate overall National Program modelling
inputs.
© Devise and perfect decision analysis methods combining cost and
benefit data for effectively managing acid deposition.
• Devise optimal scientifically-sound strategies for managing national
acid deposition control programs.
J. International Activities
• Continue and expand the cooperation with foreign countries and
international resources of acid deposition research.
The National Acid Precipitation Assessment Program made an ambitious start
and broad progress during its first two years of research (FY 1982 and 1983).
It will take a systematic effort over a number of years to adequately address
the major uncertainties about the causes, effects, and management of acid
deposition. The Program has laid the groundwork for continued scientific-
advancement, and a framework has been established to focus the research on
answering the questions most critical for developing sound policies. The
National Program's recurring assessments will provide a successively refined
scientific basis for decision making, and the opportunity to reexamine
research directions throughout the 1980's.
1-19
-------
REMARKS
S. Meyers
-------
REMARKS
By: Sheldon Meyers
Deputy Assistant Administrator
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
401 M Street, S.W.
Washington, DC 20460
I intend to deal simply and directly with my assigned topic today
which is "The Status of the Clean Air Act Reauthorization." My simple and
direct status report to you is that the Clean Air Act Reauthorization is as
yet unenacted.
Now that I have succinctly dealt with my assigned topic, I can go on
to other things.
What I would really like to do is shed some light on why the Congress
has not acted on the Reauthorization so far, and suggest to you some of the
reasons that this assignment is so difficult and tricky. I would also like
to identify some of the areas of the Act .that are causing us at EPA to have
sleepless nights and develop prematurely gray hair, by reciting some of the
elements which should be included in any discussion of a reauthorization.
Unfortunately, I don't have any predictions to offer you about the
nature or timing of Congressional action, but that might be just as well
because last month I confidently predicted the Philadelphia Phillies would
sweep the Orioles in four straight games. Over the years, I have found
that predicting the outcome of the World Series is much easier than calling
the outcome of legislation in the Congress of the United States.
The first question we ought to look at is "Why should the Clean Air
Act be reauthorized?"
The answer is that the Congress intended it that way.
Many of the basic authorities have expired. When the Congress enacted
the 1977 amendments, it obviously envisioned a thorough review of the Act,
an updating if you will, a tightening where necessary, a relaxation where
appropriate. Some of the major innovations of the 1977 amendments were
just that: new, untried, even experimental. Clearly the Congress intended
those programs to be modified, refined or even discarded on the evidence of
how they worked in practice. To aid the Congress, the Administration and
the public in the making of judgments about mid-course corrections, the
1-21
-------
Congress created the National Commission on Air Quality and mandated a
report from that Commission by March 1, 1981. Incidentally, the language
of the statute was very specific: The Commission would cease to exist on
March 1, 1981, including payment of salaries and expenses, if it failed to
file its report on that day, but if the filing took place successfully, the
Commission had until May 1, 1981, to wind up its work. Needless to say,
the Commission successfully met its deadline, and its payroll, that week.
So the Congress, legislating in the '70's but knowing that the Act had
to carry our country through the 1990's and into the 21st century, quite
prudently in my judgment called for a reauthorization of the Act's basic
authorities in 1981 and thus built into the Act a requirement that a future
Congress look at the same problems again, armed with the experience earned
in the field, from the vantage point of a new and different present. The
Congress, it seems to me, was acknowledging that things change.
Let me reinforce that with some examples: When the Clean Air Act
Amendments became law in December 1970, Persian Gulf crude oil was selling
at $2.50 a barrel, FOB Abu Dhabi.
The Shah was laying plans for the celebration of the 2,000th anniver-
sary of Iran.
The prime rate was under 6 percent.
The total Federal budget for FY '71 was $210 billion with a staggering
deficit of $21.9 billion.
President Nixon was thinking about setting up a committee to run his
reelection campaign for 1972.
The Watergate was a hotel.
And a fellow named William Doyle Ruckelhaus was about to be sworn in
as Administrator of a brand new outfit called EPA.
It's a changing world we live in.
But back to the Congressional intent: there was another reason why
the Congress was so adamant about the timely filing of the report of the
National Commission and that had to do with another deadline contained in
the '77 Amendments. The 1970 Amendments to the Clean Air Act originally
called for the attainment of national ambient air quality standards in all
jurisdictions by 1975, although EPA could grant extensions until 1977;
The 1977 Amendments extended the ambient air quality attainment dates
to 1982, but with respect to the transportation related pollutants, ozone
and carbon monoxide, EPA was authorized to push attainment dates beyond
1982, as far out as 1987.
1-22
-------
But there was a catch, as you know: Congress added a system of area-
specific, pollutant-specific, construction prohibitions. To receive an
extension to 1987, however, state plans had to satisfy what we call Part D
requirements, including the adoption of motor vehicle inspection and main-
tenance program and transportation control measures.
If a state failed to submit an approvable Part D plan for a nonattain-
ment area, the law prohibited further construction in that area of new or
modified major sources that emit a pollutant for which the area is in
nonattainment. And if transportation control measures are required, and if
the Governor has neither submitted a plan nor made a reasonable effort
toward devising one, DOT is required to withhold grants for highway con-
struction and EPA may even withhold sewage treatment grants. The con-
struction moratorium is now in effect in over a third of the states and the
highway grant cutoff is in effect in two counties in Kentucky.
Our basic concern with the nonattainment scheme in the '77 Amendments
is that we know all areas will not meet the national standards, even by
1987. We know that certain areas will not meet the ozone standard regard-
less of what reasonable measures they take, regardless of what controls
they impose, between now and then. Congress therefore will have to address
the problem of attainment deadlines in the Clean Air Act Reauthorization.
If nonattainment Part D is complex, Prevention of Significant Deteri-
oration, Part C of the '77 Amendments is downright mysterious. Walter
Barber once told a Congressional Committee that "few people in or out of
government fully understand ..." the program and no one rose to dispute
his finding.
It is probably impossible to write the history of PSD in less than
100,000 words. It was originally the subject of a Sierra Club lawsuit
dating from 1971 against William D. Ruckelshaus in his first incarnation,
based on the idea of protecting air quality in pristine regions, an idea
which unfortunately had no statutory language on which to rest in the 1970
Amendments. The Federal District Court granted Sierra Club's request for
an injunction -to force EPA to mount such a program, but it wrote no opin-
ion, hence offered no guidance. The Court of Appeals affirmed the injunc-
tion, again without opinion. The United States Supreme Court affirmed by
dividing evenly, 4 to 4, but again the Court offered no opinion. EPA
bravely came up with a regulatory scheme- in 1975, which the Congress ulti-
mately approved by incorporating it bodily into the '77 Amendments. EPA
then put out new regulations which were immediately challenged, and the
Court of Appeals overturned them in part and affirmed them in part in the
Alabama Power case. So the pattern is: the suit; the regs; the law; the
regs; the suit and then the current regs.
Nobody I know argues about the propriety of protecting, in a very
strenuous way, the pristine air in and around our National Parks - the
Class I PSD areas today. Sometimes a source must obtain both a Federal and
state permit and occasionally may be subject to both PSD and nonattainment
provisions of the '77 Amendments. Sources are frequently required to get
1-23
-------
extensive and costly field monitoring data which may or may not be useful
to the decisionmaker looking at a final decision up or down on a permit.
Incidentally, the National Commission on Air Quality recommended that
PSD Class III be eliminated entirely, Class II drastically cut back and
that the tracking of short-term increment consumption be terminated. Only
Class I would remain intact, if the Commission had its way.
There is logic in making all new source reviews internally consistent,
and it seems to me you can still be an advocate of simplification and
streamlining this program without being accused of favoring the deteriora-
tion of air quality in clean air areas.
When they are at it, Congress may want to do something about SIP's.
Even its supporters - and I am one - admit that the process is entirely too
complex and too burdensome. The law requires formal Federal approval of
the initial State Implementation Plan and most subsequent amendments or
adjustments. That requirement taxes EPA's resources in both dollars and
people - and in time. EPA must even review SIP revisions calling for
specific emission limits for specific industrial plants.
Under the statute, EPA has only four months to take action on a SIP.
Needless to say there are delays in processing this mountain of paper.
If we really mean what we say about partnership with the states, we
ought to have that reflected in the Act which should entrust the day-to-day
management of the air programs to the states or counties, with a strong
oversight role for EPA. We shouldn't have to go through formal rule making
procedures under the Administrative Procedures Act in approving or disap-
proving an amendment to a SIP which slightly raises or slightly lowers an
emission limit on some factory in some suburb of some city. It also
strains our relations with the states at a time when they should be
strengthened.
It is also no secret that EPA has trouble with Section 112 of the
Act - NESHAPs - National Emissions Standards for Hazardous Air Pollutants.
A hazardous air pollutant is defined as one that may be reasonably antici-
pated to result in an increase in mortality or serious irreversible or
incapacitating illness. These are the toxic pollutants which even in small
doses can be very dangerous to public health and which can be - and fre-
quently are - carcinogenic. The Administrator is enjoined to set a stan-
dard for such a pollutant which provides "an ample margin of safety" to
protect the public health.
But we all know there are no absolutely safe levels for some health
effects. The absence of a clear threshold for many adverse effects is
inconsistent with the concepts of "safe" levels and "ample margins" of
safety. It is understandable, I think, that EPA has not acted quickly in
listing pollutants under this section. In the Reauthorization, we will be
looking for clearer guidance from the Congress^ about how we can best deal
with the problem, given the limits of our scientific knowledge, the limits
1-24
-------
of our resources - people, money, and time. No one disputes the need to
control hazardous air pollutants. The question is not "if" but "how" we
should go about it.
Transboundary air pollution is another area of lively interest for all
parties in the Reauthorization, particularly since acid rain is on every-
body's tongue - and we know that makes for acid tongues.
The problem is that the current statutory authorities most likely to
help, don't: Section 115 allows EPA to act on reports from an interna-
tional agency to determine if air pollutants emitted in the United States
cause or contribute to pollution which endangers public health in another
country. To trigger Section 115, the foreign country must provide for
reciprocal statutory authorities. EPA can then require the state in which
the pollution originates to take action to correct the problem.
Section 126 deals with • interstate pollution. An aggrieved
state - that is, a state which believes that emissions from another state
upwind are preventing it from achieving or maintaining standards - the
State on the receiving end of the pollution may petition EPA to limit such
emissions in the state where they originate.
Our problem with both 115 and 126 is to make a causal link between a
specific source giving off emissions somewhere and the pollution that
occurs hundreds, or even thousands, of miles away from that source. Our
limited understanding of long-range transport, atmospheric chemical trans-
formations , imprecise models all come together to prevent us from making
the causal connections the Act requires before remedial action can occur.
This is one of the areas the Congress will have to look at, regardless of
what approach it ultimately takes on acid rain.
There are a dozen other areas of the Act that the Congress will look
at in its reauthorization process - big issues in the mobile source area
for instance: auto emissions, particularly NO , high altitude emissions,
diesel particulates, heavy duty truck standards. There are tough issues in
stationary sources, as the members of this symposium know very well - visi-
bility protection, percentage reduction, noncompliance penalties, and NSR
and RACT, LAER and BACT, etc.
This litany of problems, however, should not obscure the fact that the
Clean Air Act has worked.
It may be one of the most complex pieces of legislation ever devised
by the hand or mind of Congress. It may be at times unwieldy, at times
inflexible, now and then even cantankerous, but it works. The air is
measurably cleaner around our country because it works. The public health
is better served now than it was before the Act became law in 1970.
The recitation of the difficulties we have with the Act is not, I
hope, understood as totally negative criticism. On the contrary, it is
offered with a positive goal in mind: we want to see something that is
1-25
-------
good, made better. We want to see the Act made more effective by dis-
carding some of the duplications and red tape that slow us down now. We
want the Act to be flexible enough to carry us through the '80's and 90 "s
into the next century. We want to get the burdensome requirements eased
and the nonessential elements deleted completely. We don't want to see
economic growth and development stymied in our country by requirements that
have little to do with assuring clean air.
The issues are tough because our society is so complex, so varied, so
diverse. But that is also one of our great strengths, and one of the
reasons why far-reaching statutes like the Clean Air Act are, almost by
definition, controversial: they deal with so many interests, cut across so
many industries, involve such great social and economic costs and generate
such a wide range of conflicting claims and counterclaims.
I said at the beginning that I hoped to shed some light on why the
Reauthorization was so difficult and so time consuming, and identify some
basic elements that any proper reauthorization had to take into account. I
hope I have done that.
I thank you for your kind attention.
1-26
-------
THE RESOURCE CONSERVATION AND RECOVERY ACT:
CURRENT AND PROJECTED REGULATIONS
S. A. Lingle
-------
THE RESOURCE CONSERVATION AND RECOVERY ACT:
CURRENT AND PROJECTED REGULATIONS
by
Stephen A. Lingle
Chief, Technology Branch
Hazardous Industrial Waste Division
U.S. EPA, Washington, DC 20460
(Abstract and paper were not submitted)
1-27
-------
TRENDS IN COMMERCIAL APPLICATIONS OF FGD
B. A. Laseke, Jr.,M. T. Melia, N. Kaplan
-------
ABSTRACT
PEDCo Environmental, Inc., has been monitoring and reporting
on the status of utility flue gas desulfurization (FGD) technology
since 1974. During the period of 1974 to 1982, this effort was
supported by the U.S. Environmental Protection Agency (EPA) under
the direc-
tion of the Industrial Environmental Research Laboratory - Research
Triangle Park. Starting this year, this effort is now jointly
sponsored by EPA and the Electric Power Research Institute (EPRI).
Project direction from EPRI is provided by the Coal Combustion
Systems Division.
Information for this program is obtained by visits to plants
having operational FGD systems and through regular contacts with
the owner/operator utilities who are presently operating or plan-
ning FGD installations. Supplemental information is also solic-
ited from FGD system and equipment suppliers, design/engineering
firms, research organizations, and regulatory agencies.
The information collected in this program is stored in the
Flue Gas Desulfurization Information System (FGDIS), which is a
collection of computerized data base files containing descriptive,
design, performance, and cost data for all the FGD systems iden-
tified in FGDIS. FGDIS has the dual capability of generating the
periodic survey report (now available through EPRI) as well as
permitting immediate access to the data files via remote terminal.
This latter feature allows private and government users to access
directly FGDIS at any time in order to conduct custom-designed
data analyses, examine detailed data that may be too specific to
be conveniently included in the survey report, or review informa-
tion that has been loaded into the system but not yet published
in the quarterly report.
This paper summarizes the status of FGD technology as of
June 1983 and highlights the status of the electric utility power
industry and projected growth of coal-fired power generation, the
present status and future trends in the growth of FGD, develop-
ments in system design and application, current operating ex-
perience, and costs.
As of June 1983, there were 114 FGD systems in service
representing 45,750 MW (gross) of equivalent power generating
capacity. Another 100 systems representing 59,324 MW were under
construction and planned. Approximately 16 percent of the present
coal-fired generating capacity is controlled by FGD. This figure
is projected to rise to 34 percent during the next 10 years.
1-29
-------
NOTES
1. Company Names and Products.
The mention of company names or products is not to be con-
sidered an endorsement or recommendation for use by the U.S.
Environmental Protection Agency.
2. Consistency of Information.
\
The information presented was obtained from a variety of
sources including the owner/operator utilities, FGD system
and equipment suppliers, design/engineering firms, research
organizations, and regulatory agencies. The information was
acquired through a variety of methods including plant visits,
telephone contacts, written communications, and other tech-
nical reports. Accordingly, the consistency of information
on a particular system and between the several systems
discussed may be lacking. The information presented is
basically that which was voluntarily submitted by users and
developers with some interpretation by the authors. The
order of presentation of information or the amount of infor-
mation presented for any one system should not be construed
to favor or disfavor that particular system.
3. Units of Measure.
EPA policy is to express all measurements in metric units.
When implementing this practice will result in undue cost or
affect clarity, conversion factors are provided for the
non-metric units. This paper uses British units of measure.
The following equivalents can be' used for conversion to the
metric system:
British Metric
5/9 (°F-32) °C
1 ft 0.3048 m
1 ft2 0.0929 m2
1 ft3 0.0283 m3
1 grain 0.0648 gram
1 Ib (avoir.) 0.4536 kg
1 ton (long) 1.0160 m tons
1 ton (short) 0.9072 m tons
1 gal. 3.7853 liters
1 lb/106 Btu 429.6 ng/J
1 Btu/kWh 1055.056 J/kWh
1-30
-------
PROGRAM OVERVIEW
For approximately 10 years, PEDCo Environmental has moni-
tored and reported on the development and growth of flue gas
desulfurization (FGD)* technology for .utility fossil-fuel-fired
boilers in the United States. From 1974 to 1982, this program
was solely supported by the U.S. Environmental Protection Agency
(EPA) under the direction of the Industrial Environmental Research
Laboratory-Research Triangle Park. In 1983, this program was
continued under a joint sponsorship agreement between EPA and the
Electric Power Research Institute (EPRI). Project direction
within EPRI is provided by the Coal Combustion Systems Division.
The overall objective of the program still remains the same:
to provide an objective and current perspective of FGD technology
as applied to utility fossil-fuel-fired boilers and facilitate,
through information dissemination, improvements in system design,
performance, and costs for present and future facilities.
Utilities, system and equipment suppliers, design/engineering
firms, research organizations, regulatory agencies, and others
all volunteer the information for this program. This voluntary
approach facilitates timely dissemination of pertinent information
in this key technological area.
All information gathered for this program is housed in a
computerized data base called the Flue Gas Desulfurization Informa-
tion System (FGDIS). Design and performance data for both the
operational and planned domestic utility FGD systems are stored
in FGDIS. Also stored are data on operational domestic scrubbers
for removal of particulate matter and data on operational FGD
systems applied to coal-fired utility boilers in Japan.
The design data fields contained in FGDIS encompass the
entire emission control system and the power-generating unit to
which it is applied. General descriptive data fields include
plant location, standards limiting emissions of S02 and particu-
late matter, power-generating capacity, boiler and stack informa-
tion, average fuel analyses, and other more general data. The
design data fields assigned for the FGD system range from general
information such as process type, system supplier, and initial
system start-up date to more specific component design information
and operating parameters such as absorber type, gas and liquid
* FGD refers to post combustion dry or wet processes for SO
emission control. x
1-31
-------
flow rates, pressure drop, and materials of construction. Also
included in the data are descriptions of the methods of solids
concentration and waste disposal, reagent preparation and handling,
flue gas reheat, mist elimination, and information on capital
costs and annual revenue requirements.
For the operational FGD systems, FGDIS stores comprehensive
performance data, including periodic dependability parameters and
the service times from which they are calculated. Where available,
actual S02 and particulate matter removal efficiencies are included
(and qualified). Problems encountered with system operation and
the solutions implemented to correct them are coded and described.
The performance of the FGD-equipped boiler is described in terms
of service time, production level, and capacity factor.
In addition to being used to generate a periodic survey
report, FGDIS is also available for direct on-line access. This
important function not only provides interested parties with an
opportunity to examine data that are too specific for convenient
inclusion in the report, but it also provides immediate access to
information that has been loaded into the system but not yet
published (i.e., information that has become available during the
period between reports). Information is gathered, reduced, veri-
fied, and loaded into FGDIS on a continual basis to ensure that
the files remain current and complete.
Access to the FGDIS data files and management of these data
are accomplished via a computer software package known as System
2000® developed by the Intel Corporation. This comprehensive
data base management system offers extensive data retrieval
capabilities. The set of user-oriented commands provided are
flexible enough to satisfy virtually any information need.
Utilization of system functions (average, standard deviation,
summation, maximum value, and minimum value) can be used to
analyze the numerical data contained in the files. In addition,
the data requested through the available commands can be selec-
tively limited by a set of criteria included in the commands.
This feature facilitates examination of design or performance
parameters for a specific unit or a specific process type, and so
on. The retrieval and analytical possibilities are limited only
by the needs and imagination of the user.
The FGDIS files are stored at EPA's National Computer Center
(NCC) in Research Triangle Park, North Carolina.
1-32
-------
TECHNOLOGY OVERVIEW
CURRENT STATUS
1 2
United States Electric Utility Power Industry '
The generating capacity of the electric utility power indus-
try in the United States can be examined in two ways: installed
capacity and production capacity. Installed capacity represents
the full-load continuous rating of a unit as installed, irrespec-
tive of how it is operated. Production capacity represents the
generating output of a unit for a period of time, typically a
year.
As of the end of 1982, the electric utility power industry
in the United States contained a total of 10,818 power-generating
units representing a combined installed power-generating capacity
of approximately 656 gigawatts (GW). A breakdown of these in-
stalled totals by primary energy source is provided in Table 1.
For the period of January to December 1982, the electric
utility power industry in the United States produced a total of
approximately 2,241,000 gigawatt-hours (GWh). This production
level represents an industry capacity factor of 39 percent, based
on the maximum attainable output available from the total in-
stalled power-generating capacity. A breakdown of the 1982
production level by primary energy source is provided in Table 2.
As indicated in Tables 1 and 2, coal-fired power generation
accounts for 1347 units and approximately 278 GW of installed
capacity and approximately 1,188,000 GWh of production capacity.
It is noteworthy that although coal represents approximately 42
percent of the installed capacity, it accounts for approximately
53 percent of the production capacity.
The coal-fired capacities represented in Tables 1 and 2 can
also be regarded in terms of fuel consumption. The 1982 produc-
tion capacity represents approximately 636 million tons of coal
consumption.
United States Electric Utility Flue Gas Desulfurization
The status of FGD technology as applied to coal-fired utili-
ty boilers in the United States is presented in Table 3. This
table summarizes the number and installed capacity of FGD-equipped
units in service, under construction, and firm planning as of
June 1983.
1-33
-------
TABLE 1. STATUS OF INSTALLED CAPACITY OF UNITED STATES ELECTRIC UTILITY
POWER INDUSTRY AS OF DECEMBER 1982
Primary
energy source
Coal
Oil
Gas
Nuclear
Hydro
Other
Total
Installed
capacity,
MW
278,159
151,377
75,320
68,963
78,843
3,696
656,358
No.
1,347
4,399
1,640
84
3,291
57
10,818
TABLE 2. STATUS OF PRODUCTION CAPACITY OF UNITED STATES ELECTRIC UTILITY
POWER INDUSTRY FOR JANUARY-DECEMBER 1982
Primary
energy source
Production
capacity,
GWh
Coal
Oil
Gas
Nuclear
Hydro
1,187,842
134,473
313,770
291,356
313,770
Total
2,241,211
1-34
-------
TABLE 3. NUMBER AND INSTALLED CAPACITY OF UTILITY FGD SYSTEMS
Status
Operational
Under construction
Planned:
Contract awarded
Letter of intent
Requesting/evaluating bids
Considering only FGD
systems
TOTAL
No. of
units
114
27
18
7
6
42
214
Total
controlled
capacity, MW
45,750
14,712
11,494
6,060
3,400
23,658
105,074
Equivalent
scrubbed .
capacity, MW
42,135
14,441
11,386
6,060
3,400
23,480
100,902
Summation of the gross unit capacities (MW) brought into compliance by the
use of FGD systems regardless of the percentage of the flue gas scrubbed
by the FGD system(s).
^Summation of the effective scrubbed flue gas in equivalent MW, based on
the percentage of flue gas scrubbed by the FGD system(s).
1-35
-------
As indicated in Table 3, there are 114 FGD-equipped units in
service representing an installed generating capacity of 45,750
MW (gross). Comparing these figures with the figures presented
in Table 1 for coal-fired power generation reveal that approxi-
mately 8 percent of the units and approximately 16 percent of the
installed capacity are equipped with FGD systems. Moreover, if
production capacity is analyzed, it appears that FGD-equipped
units account for approximately 241,000 GWh of production which,
when compared with that for coal-fired power generation (Table
2), represents approximately 20 percent of the overall production
level for 1982.
GROWTH TRENDS1'6
FGD technology has been available for service on utility
boiler flue gas for approximately 50 years. The most significant
period of development has occurred since 1966 when limestone
injection/tail-end wet scrubbing was first introduced in the
United States and tested on coal-fired boiler flue gas. Since
that time, there have been more than 125 FGD systems operated on
utility boilers, primarily as commercial facilities and second-
arily as experimental prototype or demonstrational facilities.
A comparison of coal-fired power generation with FGD-equipped
coal-fired power generation with respect to projected growth
reveals a number of interesting trends.
If growth of the installed capacity of coal-fired power gen-
eration is compared with FGD-equipped installed capacity during
the next 10 years, the relationship illustrated in Figure 1 is
observed. Coal-fired power generation is expected to post a net
gain of 121 units and approximately 69,500 MW.* This represents
an increase of 9 and 25 percent for the number of units and in-
stalled capacity, respectively. The net increase for FGD-equipped
power generation during the same period is projected at 132 units
and approximately 73,250 MW of installed capacity, representing
an overall growth of 115 and 160 percent, respectively. The
FGD-equipped growth rate slightly exceeds the coal-fired growth
rate. This occurs because the FGD-equipped population generally
comprises newer units and therefore sees a lower rate of unit
retirement than that for the entirety of the coal-fired population.
Also, FGD installed capacity will represent approximately 34
percent of the total installed coal-fired capacity in place in 10
years, up from approximately 16 percent at present, based on
plans submitted by the utility industry.
If the growth of coal consumption is analyzed, projected
changes in fossil fuel requirements for utility power generation
for the next 10 years show coal use expanding by approximately 36
percent - from 636 million tons to 862 million tons.
Net gain accounts for the retirement of older units.
1-36
-------
450
400
350-
300
250
i r
COAL - FIRED CAPACITY
FGD - CONTROLLED CAPACITY
75 76 77 78 79 80 81 82 83 84 85 86 87
YEAR*
YEAR-END TOTALS
89 90 91
Figure 1. Growth of coal-fired installed capacity and FGD installed
capacity from 1975 through 1991.1-6
1-37
-------
If the growth of FGD is examined in and of itself, a number
of interesting trends are evident regarding year of projection,
utility experience, and system supplier experience.
With respect to year of projection, a historical review of
the annual status of FGD technology in the utility industry is
illustrated in Figure 2. This figure displays the operating,
under construction, and planned installed capacities which existed
at year's end for the period of 1970 to mid-1983. This figure
reflects a significant growth in operating capacity since 1972.
A rise is also noted in the capacity under construction from 1970
to 1977 and a fairly stable growth pattern thereafter. Planned
capacity also exhibits significant growth through 1980; however,
it sharply declines thereafter.
With respect to utility experience, 66 separate utility
companies have direct experience in operating FGD systems.
Moreover, many of these companies have multiple system experience
(32), or single system experience with additional system(s) under
construction or planned (25). A number of utility companies are
constructing or planning FGD systems with no previous operating
experience (28).
The significant growth in utility FGD experience can also be
analyzed by quantifying the number of FGD hours accumulated by
each utility for its respective system(s). The utility companies
with significant experience levels are listed in Figure 3. The
hours listed represent aggregate elapsed calendar time, from
system(s) start-up to the present (mid-1983). Actual service or
dependability considerations are not factored into this analysis
for a variety of reasons. First, we are presenting a simplistic
means by which to compare relative utility experience levels in
FGD. Second, regardless of the dependability or service require-
ments associated with a given FGD system, the utility will con-
tinue to accumulate experience with it.
A similar analysis can also be performed for the system sup-
pliers. The results of this analysis are presented in Figure 4.
The suppliers listed in Figure 4 represent those companies who
have provided commercial FGD systems and are currently active in
the commercial market. Again, the experience hours listed repre-
sent aggregate elapsed calendar time for their respective sys-
tems, irrespective of actual service or system dependability
considerations.
1-38
-------
no
100
90
80
10
i/)
0
D-
(/)
60
50
40
30
20
10
PLANNED
UNDER CONSTRUCTION
OPERATIONAL
'j888f
N*A*X*v
&S
mm
•X&&
•••:
m
•:•:•:•:•:•»•:•:•:•:•:•:
•X'X-Xl'X-X'X
SS*
JUNE
w 70 71 72 73 74 75 76 77 78 79 80 81 82 83
YEAR
Figure 2. Historic review of the annual status of utility FGD.
1-39.
-------
UTILITY NAME
ALABAMA ELECTRIC
ARIZONA ELECTRIC POWER
ARIZONA PUBLIC SERVICE
BASIN ELECTRIC POWER
BIG RIVERS ELECTRIC
CENTRAL ILLINOIS LIGHT
CENTRAL ILLINOIS PUBLIC SERVICE
CINCINNATI GAS & ELECTRIC
COLORADO UTE ELECTRIC
COLUMBUS & SOUTHERN OHIO ELECTRIC
COOPERATIVE POWER ASSOCIATION
DELMARVA POWER & LIGHT
DUQUESNE LIGHT
INDIANAPOLIS POWER & LIGHT
KANSAS CITY POWER & LIGHT
KANSAS POWER & LIGHT
LOUISVILLE GAS & ELECTRIC
MINNKOTA POWER
MONONGAHELA POWER
MONTANA POWER
NEVADA POWER
NORTHERN INDIANA PUBLIC SERVICE
NORTHERN STATES POWER
PACIFIC POWER & LIGHT
PENNSYLVANIA POWER
PHILADELPHIA ELECTRIC
PUBLIC SERVICE OF NEW MEXICO
SALT RIVER PROJECT
SOUTH CAROLINA PUBLIC SERVICE
SOUTH MISSISSIPPI ELECTRIC POWER
SOUTHERN ILLINOIS POWER
SOUTHERN INDIANA GAS & ELECTRIC
SPRINGFIELD CITY UTILITIES
TENNESSEE VALLEY AUTHORITY
TEXAS UTILITIES
UTAH POWER & LIGHT
OPERATING EXPERIENCE, HOURS x 103
0 100 200 300 400
m^mm^m
^•^•^
—
mmma^fm
•— —
I^Bi>M^H
.
—
_
•••
•
-
Figure 3. Utility company FGD operating experience.
1-40
-------
SYSTEM SUPPLIER
AIR CORRECTION DIVISION, UOP
AMERICAN AIR FILTER
BABCOCK & WILCOX
COMBUSTION ENGINEERING
DAVY McKEE
ENVIRONEERING, RILEY STOKER
FMC
GE ENVIRONMENTAL SERVICES
JOY/NIRO
MITSUBISHI HEAVY INDUSTRIES
M.W. KELLOGG
PEABODY PROCESS SYSTEMS
RESEARCH-COTTRELL
ROCKWELL INTERNATIONAL
THYSSEN/CEA
UNITED ENGINEERS
OPERATING EXPERIENCE, HOURS x 103
0 200 400 600 800
^^^••M
=-
•
•
m—^m
-
-
•
••
Figure 4. FGD operating experience associated with system suppliers.
1-41
-------
DEVELOPMENTS IN SYSTEM DESIGN AND APPLICATION
A number of significant developments have affected the
electric utility power industry in the United States with
respect to the application and design of FGD systems for coal-
fired power generation. These developments include revised (and
more stringent) air emission control standards, decline in power
demand, increases in new plant lead time and costs, and the
supply and costs of fossil fuels. These developments have
affected the selection of emission control strategies, FGD
processes, and FGD system design configuration.
EMISSION CONTROL STANDARDS
The current environmental driving force for S02 emission
reduction is the Clean Air Act, implemented in 1970 and amended
in 1977, which created a schedule to decrease S02 emissions from
power plants and other major sources for both new and existing
facilities. New sources are regulated by the New Source Perform-
ance Standard (NSPS) of December 1971 and the current or revised
NSPS of June 1979. In addition, the Clean Air Act amendments
indirectly provide for regulation of existing power plants
because State Implementation Plans (SIPs), required for each
state, include a control strategy for ensuring maintenance of
the promulgated ambient air quality standards for S02 and other
regulated pollutants.
The FGD-equipped units identified in FGDIS are classified
according to one of five basic regulatory classification cate-
gories. These categories are identified in Table 4 complete
with a summary of the number of units and installed capacities
by status.
EMISSION CONTROL STRATEGIES
The control of particulate matter must be considered along
with the installation of an FGD system. The three basic methods
currently considered are electrostatic precipitation, fabric
filtration, and wet scrubbing. There are a number of site and
design factors involved in such a selection process as they
relate to FGD:
• Dry collection of fly ash upstream may be desirable
for use as an additive to fix or stabilize the waste.
1-42
-------
TABLE 4. FGD SYSTEMS BY STATUS AND REGULATORY CLASSIFICATION
Regulatory
classification
More stringent than
6/79 NSPS
6/79 NSPS
Less stringent than
6/79 NSPS and more
stringent than
12/71 NSPS
12/71 NSPS
Less stringent
than 12/71 NSPS
Total
Operational
No.
12
9
40
41
12
114
MW
5,394
2,746
17,967
16,912
2,731
45,750
Under construction
No.
14
7
2
4
-
27
MW
7,899
3,809
1,151
1,853
-
14,712
Planned
No.
27
42
3
1
-
73
MW
15,706
26,886
1,650
370
-
44,612
Total
No.
53
58
45
46
12
214
MW
28,999
33,441
20,768
19,135
2,731
105,074
1-43
-------
• Dry collection upstream decreases the solids loading
on the FGD system which can create or promote plugging,
scaling, or erosion of internals.
• The fly ash may contain alkaline compounds that assist
in S02 removal in the combined scrubbing of fly ash
and S02.
• Plants faced with the necessity of retrofitting an FGD
system may contain existing fly ash collection equip-
ment of marginal performance capability, thus making
fly ash and S02 scrubbing an economically attractive
option.
• Supplemental fly ash collection is an inherent design
feature of some regenerable/product-recovery technol-
ogies for process considerations. Supplemental fly
ash removal (as well as absorption of S03 and HC1) may
occur via a separate wet scrubbing step or, less
frequently, use of a wet ESP.
• The selection of a dry scrubbing system involves the
collection of particulate matter containing S02 reac-
tion products and fly ash.
• New plants faced with stringent particulate emission
regulations may use the FGD wet scrubber to provide
secondary particulate removal and thus save on the
costs associated with an extremely high efficiency
upstream particulate matter collection device.
The methods used and planned for future use on FGD-equipped
units are tabulated in Table 5. A clear preference for ESP's is
shown. The emergence of fabric filters is driven by the increas-
ingly popular application of dry scrubbing technology. Wet
scrubbing for particulate matter control, although significant
in currently operational units, shows a clear decline in future
FGD-equipped units.
FGD PROCESS APPLICATION
Application of FGD technology requires the consideration of
design characteristics, including the use of waste producing or
product recovery technologies, regenerable versus nonregenerable
technologies, chemical reagent and additive selection, and coal
sulfur content. A brief overview of FGD technology with respect
to these factors for systems presently in service and planned
for imminent future service is discussed below.
FGD systems currently in service, under construction, and
contract awarded are shown in Table 6 as a function of throwaway
1-44
-------
TABLE 5. FGD-EQUIPPED PARTICULATE MATTER EMISSION CONTROLS
Electrostatic
precipitator
Fabric filter
Wet scrubbing
ESP/wet scrubbing
(process)
ESP/wet scrubbing
(secondary
particulate)
Total
Operational
No.
68
7
22
8
9
114
MW
28,266
1,989
8,220
2,576
4,699
45,750
Under
construction
No.
16
8
2
-
1
27
MW
9,806
2,802
1,400
-
704
14,712
Contract
awarded
No.
15
3
-
-
-
18
MW
9,544
1,950
-
-
-
11,494
Total
No.
99
18
24
8
10
159
MW
47,616
6,741
9,620
2,576
5,403
71,956
1-45
-------
TABLE 6. SUMMARY OF FGD PROCESS DESIGN
Throwaway product
Wet
Nonregenerable
Limestone
Lime
Sodium carbonate
Regenerate
Dual alkali
Dry (nonregenerable)
L i me
Sodium carbonate
Salable product
Wet
Nonregenerable
Limestone
Lime
Regenerate
Well man Lord
Magnesium oxide
Dry (regenerable)
Aqueous carbonate
TOTAL THROWAWAY PRODUCT
TOTAL SALABLE PRODUCT
TOTAL WET
TOTAL DRY
TOTAL NONREGENERABLE
TOTAL REGENERABLE
Operational
No.
51
34
' 6
4
5
1
1
1
7
3
1
101
13
107
7
99
15
MW*
20,756
13,634
1,505
1,572
1,214
440
166
65
1,959
724
100
39,121
3,014
40,381
1,754
37,780
4,355
Under
construction
No.
13
5
1
7
1
-
—
26
1
20
7
26
1
MW*
7,546
3,350
391
2,679
475
-
—
13,966
475
11,762
2,679
14,050
391
Contract
awarded
No.
11
2
1
4
-
_
-
18
-
14
4
17
1
MW*
7,591
1,090
265
2,440
-
_
-
11,386
-
8,946
2,440
11,121
265
Total
No.
75
41
6
6
16
1
2
1
7
3
1
145
14
141
18
142
17
MW*
\
35,893
18,074
1,505
2,228
6,333
440
641
65
1,959
724
100
64,473
3,489
61,089
6,873
62,951
5,011
*Equivalent scrubbed capacity
1-46
-------
product versus salable product, regenerable versus nonregenerable,
wet versus dry, and chemical reagent.
Table 6 reflects a strong industry preference for nonregen-
erable, calcium-based, wet slurry processes that produce a waste
product for disposal. Examining the lime and limestone categories
reveals a continuing preference for limestone. The emergence of
dry scrubbing technology is reflected in the under construction
and planned categories.
The application of FGD systems with respect to coal sulfur
content can be analyzed by distinguishing three sulfur ranges:
Low sulfur coal:
less than or equal to 1 percent
sulfur in the coal
• Medium sulfur coal: greater than 1 percent but less
than or equal to 3 percent sulfur
in the coal
• High sulfur coal: greater than 3 percent sulfur in
the coal
In accordance with these categories, Table 7 provides a break-
down of FGD systems by coal sulfur content. It is evident from
this table that about 50 percent of the FGD systems in each
status category are low-sulfur coal applications.
TABLE 7. FGD SYSTEMS BY COAL SULFUR CONTENT
Coal sulfur
Low
Medium
High
Total
Operational
No.
53
26
35
114
MW
22,557
8,663
14,530
45,750
Under
construction
No.
15
7
5
27
MW
7,727
4,445
2,540
14,712
Contract
awarded
No.
10
5
3
18
MW
7,100
2,981
1,413
11,494
Total
No.
78
38
43
159
MW
37,384
16,089
18,483
71,956
1-47
-------
PERFORMANCE OVERVIEW
During the past several years, FGD has become the most
commercially developed means of controlling S02 emissions from
coal-fired boilers. FGD operating experience increases each year
as more and more systems are placed in service. The FGD survey
program has been monitoring the performance of the operating FGD
systems and logging monthly operating parameters (e.g., boiler
and FGD system operating hours, forced outage times, and scheduled
outage times). With this operational information, system per-
formance trends over the years can be analyzed and reviewed.
One important criterion for the evaluation of FGD perform-
ance is system dependability. We have developed a number of
dependability parameters to measure this index for various situ-
ations, conditions, and information-reporting formats. Of these
parameters, "availability" seems to be most universally accepted
and understood. In context of FGD, system availability is defined
as the hours the FGD system is available for operation (whether
operated or not) divided by hours in the period. Period hours
usually equal hours in a month.
Availability data for operating FGD systems have been col-
lected through the Utility FGD Survey program on a monthly basis.
The monthly data were used to calculate the average annual availa-
bilities for each system.
Figures 5 and 6 are the median analyses of the annual average
availabilities for all FGD systems (for which data are available);
broken down for low-, medium-, and high-sulfur coals.* Median
analyses were based on data from the period of 1978 through 1982
(5-year period). In each case, the availability points were
plotted and the median was determined.
As can be seen from these figures, low sulfur systems show
the steadiest median of the three, with the median changing only
a few percentage points. The high sulfur systems exhibit the
most dramatic improvement over the period, as well as exhibiting
a relatively constant trend for the past 3 years. The medium
sulfur coal systems also exhibit significant improvement; however,
two fluctuations are observed for 1980 and 1982.
*
The coal sulfur content ranges are defined as less than 1
percent, 1 through 3 percent, and greater than 3 percent for
low, medium, and high, respectively.
1-48
-------
s-s
CD
—I
*—t
<:
100
90
80
70
60
50
40
30
20
10
0
ALL PROCESSES
\\
UJ
o
oo
1978
1979
1980
1981
1982
Figure 5. Median analysis of the annual availability for
coal-fired FGD systems.?
1-49
-------
I
Ln
O
100
BO
- '0
P
i»
i
so
LOW SULFUR
1978 1979 1980 1981 1982
100
80
70
3*.
I
so
30
20
1978 I97» I960
MEDIUM SULFUR
70
i"
so
40
30
1978
\t»\ 1982
HIGH SULFUR
Figure 6. Median availability analysis by coal sulfur content.
-------
Overall, the performance of FGD with respect to system availability
tends to show steady improvement. Lime and dual alkali processes have
significantly improved the technology's performance on high-sulfur coal.
COSTS OVERVIEW
Capital and annual cost data for operational FGD systems
have been obtained continuously under the utility survey program
since March 1978. Costs for each system are obtained directly
from the utilities and then itemized by individual FGD cost
element. The itemized costs are then adjusted to a common basis
to enhance comparability. This adjustment is made by using
factors for estimating costs not given by the utilities and by
using cost index factors to express all costs in constant year
dollars (mid-1981). All adjusted cost data and computations are
reviewed and verified with the appropriate utility before publi-
cation.
The cost adjustment procedure is summarized below.
Capital Costs
• All costs associated with control of particulate matter
emissions are excluded.
• Capital costs for modifications necessitated by installa-
tion of an FGD system are added if they were not included
in the reported costs.
• Sludge disposal costs are adjusted to reflect a 20-year
life span for retrofit systems and a 30-year life span
for new systems.
• Any unreported direct and indirect costs incurred are
estimated and included.
• All capital costs are expressed in mid-1981 dollars.
• All capital costs reflect the gross generating capacity
of the unit.
Annual Costs
• Direct costs that were not reported are estimated and
added.
• Overhead and fixed costs that were not reported are
estimated and added.
1-51
-------
• All annual costs are expressed in mid-1981 dollars.
• All annual costs are based on a 65 percent capacity
factor and the net generating capacity of the unit.
Table 8 summarizes both reported and adjusted costs for all
operational FGD systems on which cost data were obtained. This
table also summarizes the results, by application and by process
category. Table 9 lists the results by process type. Table 10
provides a plant-by-plant listing of the reported and adjusted
costs for the operational FGD systems addressed in this study.
The capital and annual costs for a significant portion of
the commercial population reflect an overall cost of approxi-
mately $119/kW and 8 mills/kWh, respectively. These figures
apply to a wide variety of systems that represent different
processes, dates of installation, applications, and severity of
duty for units in commercial service from 1972 to 1981.
1-52
-------
TABLE 8. CATEGORICAL RESULTS OF THE REPORTED AND ADJUSTED
CAPITAL AND ANNUAL COSTS FOR OPERATIONAL FGD SYSTEMS
I
Ul
u>
All
New
Retrofit
Salable
Throwaway
Capita 1
Range, $/kW
23.7-213.6
23.7-213.6
29.4-157.4
132.8-185.0
23.7-313.6
Reportec
Average,
S/kW
80.2
80.4
79.7
153.1
75.8
Annual
Range,
mills/kWh
0.1-13.0
0.1-5.5
0.5-13.0
13.0-13.0
0.1-11.3
Average ,
mills/kWh
2.3
1.7
4.5
13.0
2.1
Capi tal
Range, $/kW
38.3-282.2
38.3-263.9
60.4-282.2
254.6-282.2
38.3-263.9
Adjustec
Average,
SAW
118.8
110.8
139.3
271.6
110.9
Annual
Range,
mills/kWh
1.6-20.8
1.6-14.6
4.3-20.8
16.7-20.8
1.6-17.6
Average,
mil Is/kWh
7.6
6.8
9.7
18.1
7.0
-------
TABLE 9. ADJUSTED CAPITAL AND ANNUAL COSTS FOR OPERATIONAL
FGD SYSTEMS BY PROCESS TYPE
Limestone
Lime
Dual alkali
Lime/alka-
line fly
ash
Sodium
carbonate
Wellman
Lord
Limestone/
alkaline
fly ash
Reported
Capital
Range, $/kW
2.37-170.4
29.4-213.6
47.2-174.8
43.4-173.8
42.9-100.8
132.8-185.0
49.3-49.3
Average,
$/kW
67.9
81.8
97.8
93.9
69.2
153.1
49.3
Annual
Range,
mills/kWf,
0.1-7.8
0.3-11.3
1.3-1.3
0.4-5.4
0.2-0.5
13.0-13.0
0.8-0.8
Average,
mills/kWh
1.6
3.2
1.3
2.1
0.4
13.0
0.8
Adjusted
Capital
Range, $/kW
38.3-194.3
60.4-210.0
87.8-263.9
52.5-184.4
87.1-150.9
254.6-282.2
102.6-102.6
Average,
S/kW
98.9
116.5
146.7
122.8
110.9
271.6
102.6
Annual
Range ,
mills/kWh
1.6-14.6
4.0-17.6
5.0-13.9
3.0-14.1
5.8-7.4
16.7-20.8
5.4-5.4
Average,
mills/kWh
6.1
8.1
8.7
7.2
6.4
18.1
5.4
-------
TABLE 10 REPORTED AND ADJUSTED CAPITAL AND ANNUAL COSTS
FOR OPERATIONAL FGD SYSTEMS
Alabama Electric
Tombiobpp ?
Tombtgtifp 3
Ari:ond Electric Power
Apache ?
Apache 3
Choi la 1
Cholla 2
Central Illinois Ltgnt
Duck Creek 1
Central Illinois Public Service
Newton 1
Colorado Ute Electric
Craig 1
Craig 2
Columbus & Southern Ohio Electric
Conesville 5
Conesville 6
Coal Creek 1
Coal Creek 2
Ouquesne Light
Elrama 1-4
Phillips 1-6
Indianapolis Power & Light
Petersburg 3
Kansas City Power ^ Light
Hawthorn 3
Hawthorn 4
LaCygne 1
Kansas Power & Light
Jeffrey 1
Jeffrey 2
Kentucky Ut i 1 ities
Green River ]-3
Louisville Gas i Electric
Cane Run t
Cane Run 5
Cane Run 6
Mill Creek 1
Hill Creek 3
Paddy.' s Run 6
Minnesota Power 1 Light
Clay Boswell 4
Mlnflkota Power
Milton R. »oung 2
Honongahela Power
Plef-.ants 1
Pleesants 2
Montana Power
Colstrip 1
Colstrip 2
Nevada Power
Reid Gardner 1
Reid Gardner 2
Reid Gardner }
Northern Indiana Public Service
Dean H. Mitchell 11
Northern States Power
Shertoume 1
Sherburne 2
Pacific Power t LigM
Jim Bridge*- 4
Pennsylvania Power
Bruce rtensfield 1
Bruce Kansfield 2
Bruce Mansfield 3
Public Service of New Mexico
San Juan 1
San Juan 7
Salt River Project
Coronado 1
Coronado 2
South Carolina Public Service
Ninyah 2
hinyah 3
RepO'fd
Capital
6,921,100
6. 99? , 100
8, 148,069
7.214,052
6.550,000
44,352,000
31.125,000
107.631,000
36.613.500
32,810. 500
20,663,500
20,663,500
23,650.000
23.650,000
61,320.000
52.055.000
55.620.000
3.220.000
3.220,000
46.900.000
27.306.900
26, 458.800
3.966,155
12,647,000
13.759,000
20.596,900
17.661,500
16.846,880
4.320.000
84.900.000
44.119.600
61.452.400
81,452.400
36.500.000
36.500.000
5,363.375
5,363,375
12.599.160
IB. 192, 040
34.892.020
34,962.020
49,643,000
110,639,000
110,639,000
195.857.200
47,944,400
47,965,000
32.624,500
32.624.500
6.646.096
10.742.500
>/kk
27.4
27.4
41.6
37.0
52.0
166.0
74.8
174.8
66.5
73.5
50.3
50.3
43.4
43.4
120.'2
127.0
104.9
29.4
29.4
53.7
37.9
37.8
62.0
66.6
66.6
71.5
55.0
42.6
61.7
173.8
100.3
119.1
119.1
101.4
101.4
42.9
42.9
100.6
157.4
49.3
49.3
90.3
120.7
120.6
213.6
132.6
137.1
79.4
79.4
23.7
38.4
Annual
434,929
434,929
51.784
51.764
KA
J. 003. 568
10,851,000
NA
462.600
1.128,400
6,670,530
6.670.530
1,166.200
1.166.200
21,027.451
16.301.000
NA
346,441
346.441
8.060.047
321. OOC
949.900
364.005
960,301
763,443
NA
NA
321.463
NA
16.644,512
754,445
5.532.600
5.532,800
6,126,000
6,126.000
251.514
251,514
131.624
2.414,569
2.716,759
2.716,659
NA
9,979,900
9,979.900
11.575.300
NA
NA
1.788,000
1.768,000
527.000
NA
•Illl/
k»
0.33
0.33
0.10
0.10
NA
0.76
5.04
NA
0.53
0.46
5.52
5.52
0.41
0.41
7.18
11.00
NA
1.15
1.15
5.38
0.11
0.2S
'5.20
1.29
0.92
NA
NA
1.25
NA
5.38
0.53
1.46
1.46
2.97
2.97
0.46
0.46
0.23
13.02
0.75
0.75
NA
3.28
3.26
3.11
NA
NA
0.85
0.65
0.29
NA
Adjusted
Capital
9,756.200
9,756,200
13,171,700
11,661.900
10.246.100
42,767,600
55.001.000
162,647.600
46.179.000
41.070.100
40.857.200
40.857,200
28.633.100
26,633.100
95.755.400
86.097,500
66,227,500
6,909,650
6,909.650
67,496,400
39,710,500
26,840.100
7,510.600
21,882.500
20.476,600
25,290,600
21.627.800
29,163.000
9,307.300
102.165.300
68.467.500
1/kU
M.3
36.3
67.5
59.6
81.3
162.1
132.2
263.9
105.9
90.3
99.4
99.1
52.5
52.5
187.6
210.0
162.1
62.8
62.8
100.1
55.9
41.2
117.8
115.2
Annual
4.249.650
4,249,650
3,739.900
3.739.900
4.606.800
10,998.300
12.433.100
45,624,900
12.148.30C
10.382.700
14,475,100
14.475,100
6.414.350
B. 414. 350
35,731.700
35.731.700
28.565.000
2.9K.100
2.912,100
52,735,300
•till/
ttti
3.16
3.18
3.75
3.75
4.79
8.22
5.78
13.94
5.33
4.56
6.76
6.78
2.99
2.59
12.89
17.65
9.74
5.25
5.25
1J.29
9,353.206
6.396.400
3.754,600
' 6.422.400
102.4 5.764,300
87.6 11.073.000
2.42
1.65
10.99
6.20
5.30
7.23
60.41 8.107,100 4.26
66.0
133.0
9.636.200 4.03
4. 430, 300 ! 12.25
1
!
184.4 40,437.300 14.09
i
155.7 H, 966. OOC
1
S.4«
104,562,000! 152.9! 27, 951, 300 i 7.9«
104.562.000
152.9' 27.951.300
7.94
52.519.850; 145.9 15.642,050 6.27
52. 519,6501 145.9! 15.642,050
1
10. 891.3501 87.1
10.891.350
16.666.173
29.431,806
73.654.486
73.654.486
65,113,736
132,234.039
132.234,039
166,043.400
'100.333,473
96.777.569
35.560.700
35.560,700
13.148,672
67.1
150.9
254.6
102.6
102.6
116.4
3,617.450
3.617.45C
4.637.163
11.129.239
21.018,195
21.016,195
19,661.160
144.2] 53.122.918
144.2 53,122.916
161.1
277.9
:282.2
66.5
66.5
47. C
16.646.700 66. (
46,255.200
29,836.602
29,423,764
15,339.100
15,339,100
2,697,69f
6.27
5.76
5.76
7.40
20.79
5.43
5.43
6.61
11.31
11.31
• 10.27
16.69
16.89
7.70
7.70
1.64
6,926,700 4.63
1-55
-------
TABLE 10. REPORTED AND ADJUSTED CAPITAL AND ANNUAL COSTS
FOR OPERATIONAL FGD SYSTEMS (CON'T)
South Mississippi Electric Power
R. D. Morrow, Sr . 1
R. 0. Horrow, Sr. 2
Southern Illinois Power
Marion 4
Southern Indiana &as & Electric
A. B. Brown 1
Springfield City Utilities
Southwest 1
Springfield Uater, Light t Power
Oallman 3
Tennessee Valley Authority
ttidows Creel 8
Utah Power 1 Light
Hunter 1
Hunter 2
Huntington 1
10,896,000
10,896.000
15,200,930
12,495,000
16,744,500
34,923,800
47,900,000
24.400,000
25,880,000
2,709.000
53.7
53.7
87.9
47.2
B6.3
170.4
87.1
56.7
60.2
63.0
NA
NA
859,453
1,850.565
778.749
4.381,500
14,576,400
NA
1,072,600
2,946.400
NA
NA
0.98
1.30
1.20
4.68
7.76
NA
0.32
1.27
24,047,200
24,047.200
20,906.447
23.414,388
27.822.607
39.829.200
86.968.826
32.294.620
31.225,000
38.698.607
118.5
118.5
120.8
88.4
143.4
194.3
158.1
75.1
72.6
9C.O
7,115,950
7,115.990
7,107,923
7,094,130
8.092.238
15,913.700
21,379,073
10.614.532
1C. 355. 700
13.4I1.6S2
6.94
6.94
7.75
4.98
8.21
14.56
7.28
4.66
4.55
5.69
REFERENCES
1. U.S. Department of Energy. Inventory of Power Plants in the
United States - 1981 Annual. Energy Information Administra-
tion. Office of Coal, Nuclear, Electric, and Alternate
Fuels. DOE/EIA-0095(81), September 1982.
2. North American Electric Reliability Council. Electric Power
Supply and Demand 1983-1982. Annual Data Summary Report for
the Regional Reliability Concils of NERC. July, 1983.
3. U.S. Department-of Energy. Inventory of Power Plants in the
United States, DOE/RA-001, December 1977. Office of Utility
Project Operations.
4. U.S. Department of Energy. Inventory of Power Plants in the
United States - December 1979. Energy Information Administra-
tion. DOE/EIA-0095(79), June 1980.
5. U.S. Department of Energy. Inventory of Power Plants in the
United States - 1980 Annual. Energy Information Administra-
tion. DOE/EIA-0095(80), June 1981.
6. U.S. Department of Energy. Energy Information Administra-
tion. Office of Coal, Nuclear, Electric, and Alternate
Fuels. Electric Power Monthly, DOE/EIA-0226(83/01), January
1983.
7. Flue Gas Desulfurization Information System (FGDIS). Com-
puterized data base. U.S. Environmental Protection Agency,
Industrial Environmental Research Laboratory, Research
Triangle Park, NC. Access to FGDIS can be obtained through
Walter Finch, NTIS, 5285 Port Royal Road, Springfield, VA
22161.
1-56
-------
SESSION 2: ECONOMICS
Chairman: Thomas M. Morasky
Reliability and Nonrecovery Systems
Electric Power Research Institute
Palo Alto, CA
-------
COMPUTER ECONOMICS OF PHYSICAL COAL CLEANING
AND FLUE GAS DESULFURIZATION
C. R. Wright, T. W. Tarkington,
J. D. Kilgroe
-------
COMPUTER ECONOMICS OF PHYSICAL COAL CLEANING
AND FLUE GAS DESULFURIZATION
by: Charles R. Wright and Terry W. Tarkington
Division of Energy Demonstrations and Technology
Tennessee Valley Authority
Muscle Shoals, Alabama 35660
James D. Kilgroe
Industrial Environmental Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
ABSTRACT
A computer model that simulates the performance and calculates the eco-
nomics of a coal-cleaning process, a flue gas desulfurization (FGD) process,
and a combination of the two processes for electric utility boilers has been
developed by the Tennessee Valley Authority (TVA) under sponsorship of the
U.S. Environmental Protection Agency (EPA). The model also determines other
economic benefits and penalties to overall power plant operations that are
associated with the use of cleaned coal. The model is described and its use
is illustrated for selected design and economic premises. The present status
of the model is assessed and possible future studies are discussed. The
illustrative examples show that in some instances a combination of coal
cleaning and FGD for SC>2 emission control can be more economical than FGD
alone.
For the cases studied, it was found that the use of coal cleaning in
combination with FGD can have a varied effect on the levelized annual cost of
power production. This effect, which is dependent on the coal and the spe-
cific operating conditions, ranges from a 0.5 percent increase to a 22.6
percent decrease in those costs.
The methods used to determine many of the other economic benefits and
penalties of coal cleaning are necessarily general in nature because detailed
data relating specific coal properties to boiler performance and operating
costs are scarce. Thus, development and incorporation of more detailed and
quantitative data in the model would greatly increase the usefulness of the
model in assessing overall economic effects of coal cleaning.
2-1
-------
INTRODUCTION
In recent years, coal cleaning has come to be regarded as a practical
alternative in many cases for the control of S02 emissions from coal-fired
power plants. TVA recently completed an economic evaluation for EPA that
demonstrated the potential of coal cleaning in S02 emission control (1). As
a continuation of this program, a computer model was developed to simulate the
performance and economics of full-scale coal cleaning alone and in combination
with limestone and lime FGD (2,3). The model determines the process design
and economics for a coal-cleaning process, an FGD process, or a coal-cleaning
process combined with FGD. It also determines other economic benefits and
penalties attributable to the use of cleaned coal.
The model should be useful in comparisons of the design and costs of
various combinations of coal-cleaning and lime or limestone FGD processes.
This should be particularly useful when evaluating processes for
specific applications because the model will illustrate the effect of changes
in process variables on the design and cost of the component processes or the
entire system as a whole.
MODEL DESCRIPTION
The coal-cleaning/FGD computer model consists of four separate computer
programs, which are linked by common input and output data. The first program
determines the design and economics of the coal-cleaning process; the second
and third programs determine the FGD process design and economics. The fourth
program quantifies selected economic benefits and penalties of using cleaned
coal. Unlike the first three programs, which are modified versions of
previously developed programs, the fourth program has been developed
specifically for this project. The model is written in FORTRAN IV and was run
on an IBM model 3083 computer.
The cost elements that the model calculates include the capital invest-
ment, first-year annual revenue requirements, and levelized annual revenue
requirements; each element can be generated for a coal-cleaning process, an
FGD process, or a combined coal-cleaning/FGD process. In the case of the
combined coal-cleaning/FGD process, an incremental cost is also calculated
which is the difference between the cost for the coal-cleaning process and any
net benefit realized from the use of cleaned coal (including any reduction in
the FGD costs).
2-2
-------
COAL-CLEANING PROGRAM
The coal-cleaning program is the primary control program (see Figure 1).
It was adapted from a program developed by the University of Pittsburgh and
the Bureau of Mines, and modified by Battelle Columbus Laboratories. The
Battelle version, known as the Coal Preparation Simulation Model Version 4
(CPSMH) (4), was obtained from Versar, Inc., who also provided valuable
assistance in its adaptation to the TVA computer system. The program required
many modifications to conform to the TVA process design criteria and to form
the combined coal-cleaning/FGD model. Although many of these changes were
relatively minor and did not alter the program processing sequence, some areas
required major revisions, necessitating the addition of five new subroutines
and the associated rearrangement of the program.
This program is capable of simulating many coal-cleaning processes pro-
vided that equipment performance, process design, and economic data are avail-
able. The process design used in the illustrative runs for this paper is felt
to be a typical application. A flow diagram for this process design is shown
in Figure 2. The process, as illustrated, separates the raw coal into three
size fractions, each of which is processed in a different type of separation
equipment. The coarse coal is cleaned in a dense-medium vessel, the interme-
diate coal in a dense-medium cyclone, and the fine coal in froth flotation
cells.
FGD MODEL
The FGD model is the Shawnee lime/limestone computer model developed by
Bechtel National, Inc., and TVA using data obtained from the test facility at
the TVA Shawnee Steam Plant near Paducah, Kentucky (5). This model consists
of two programs that produce full-scale design and economic data for a variety
of limestone or lime FGD systems.
The current model includes options for limestone or lime scrubbing: a
mobile-bed absorber (TCA), a spray tower, or a venturi/spray tower absorber,
and various waste disposal methods. The FGD process design used to illustrate
the capabilities of the combined model is illustrated in Figure 3. The
process employs limestone slurry scrubbing in a spray tower absorber, and the
untreated waste is disposed of in a pond.
ECONOMIC BENEFITS AND PENALTIES PROGRAM
There are many differences between cleaned coal and raw coal that result
in economic benefits and penalties for a power plant that uses cleaned coal.
This program utilizes appropriate equations to approximate many of the bene-
fits and penalties that accrue because of such differences as the lower ash
content, the higher moisture content, and the generally higher heating value
of the cleaned coal. Each of the benefits and penalties calculated by this
program is discussed below.
2-3
-------
Coal Cleaning
Process Printout
All Input Data
I
Coal Cleaning
Program
1
Benefits/
Penalties
Input Data
I
Benefits/
Penalties Program
1
Benefits/Penalties
Program Printout
-MFGD Input Data)
I
FGD Programs
1
FGD Process
Printout
Figure 1. Flow of data through the combined model.
2-4
-------
COAL RECEIVING
AND STORAGE
RAW COAL
SIZING
3/8 IN X 28M
COARSE COAL
CLEANING
INTERMEDIATE
COAL
CLEANING
FINE COAL
CLEANING
REFUSE
DISPOSAL
CLEAN COAL
STORAGE
CLEAN COAL
SHIPMENT
Figure 2. Flow diagram for the coal-cleaning process.
2-5
-------
EMERGENCY BYPASS
COAL .
t-0
I
Figure 3. Flow diagram for the limestone-scrubbing FGD process.
-------
FGD Costs
The most important effect of coal cleaning on power plant costs is in the
area of FGD costs. In some cases coal cleaning or coal cleaning with FGD can
be more economical than FGD alone in controlling SC>2 emissions. The lower
sulfur content and increased heating value of cleaned coal reduce the quantity
of S02 produced in the combustion process, thus reducing the SC^ removal
requirement of the FGD process. There are some cases where the use of cleaned
coal can even eliminate the need for FGD. FGD cost benefits which accrue from
coal cleaning are determined by comparing costs required when FGD is used to
control S02 emissions from raw versus clean coal.
Raw Coal Loss
No presently available coal-cleaning process is capable of achieving 100
percent separation of pure coal from its impurities. There is some coal lost
in the cleaning plant refuse and some impurities remain in the cleaned coal.
The coal that is lost is an economic liability for the coal-cleaning plant.
This liability or penalty can be estimated by the fraction or percentage of
the feed coal heat content that is recovered in the product coal (Btu
recovery).
Transportation Costs
Coal cleaning results in a product that has a higher heating value than
the raw coal, thereby reducing the weight of coal necessary to satisfy the
heat requirements of a specific power plant. A transporation cost reduction or
benefit occurs when coal cleaning is performed at the mine.
United Mine Workers' (UMW) Pension Fund Benefits
The provisions of the 1978 UMW contract require mine operators to pay
$1.385 to the Pensions and Benefit Trust Fund for every ton of coal shipped to
a consumer (6). Since coal cleaning generally reduces the quantity of coal
required by a power plant, the contribution to the trust fund is also reduced,
thereby resulting in a cost benefit for the utility.
State Taxes
State-imposed coal taxes, often called severance taxes, are presently
levied by the legislatures of 13 states (7). Depending upon the method of
taxation, coal cleaning can reduce or increase these taxes, thus creating a
cost benefit or penalty. There are currently four different methods which are
based upon (1) the tons of coal shipped, (2) the value of coal shipped, (3)
the tons of coal mined, and (4) the value of coal mined. The model is capable
of simulating any of the four; however, taxation based on the tonnage of coal
shipped is the most common and is used in the example runs for this paper.
This method will generally provide an economic benefit because cleaned coal as
compared with ROM coal will result in a reduction in the tonnage of coal
shipped.
2-7
-------
Size Reduction
In general, at least two and sometimes three size reduction operations
are performed on coal as it moves from the coal seam to the power plant
boiler. For the development of this program, it was assumed that the power
plant was equipped with pulverized coal boilers in which case three separate
size reduction operations are necessary. The first operation is performed at
the mine where the run-of-mine (ROM) coal is crushed to an appropriate top-
size. The next two operations are performed at the power plant where the coal
is crushed before being fed to the pulverizers and finely ground. When coal
cleaning is used, an additional size reduction operation is usually included
at the cleaning plant, resulting in a cost benefit to the utility because the
size reduction requirements at the power plant are reduced. Coal cleaning can
also produce a benefit in the size reduction operations at the power plant by
reducing the coal mineral matter, making the coal easier to grind. A third
consideration is the increased heating value of cleaned coal, which decreases
the amount to be ground. One effect of coal cleaning, the increase in surface
moisture, can increase pulverizer plugging, but it is not considered in the
program because of lack of data on the effects of moisture on pulverizer
operation.
Maintenance Costs
The quality and quantity of the coal used in a power plant have long been
recognized as having a direct effect on the maintenance costs of the plant
equipment. Some of the equipment areas have been mentioned in the preceding
paragraphs, but other areas which must be considered include the boilers and
accessories, coal conveying and storage equipment, and the air heaters. Coal
cleaning yields an economic benefit in maintenance costs for these and other
areas of power plant operation by raising the quality of coal burned in the
plant through the reduction of the ash and sulfur content. To quantify this
benefit the program uses a relationship based on the ash and sulfur in the
coal, which is derived from actual power plant maintenance data (8).
Power Plant Availability
The availability of a power plant for the production of electricity is
another area that is affected by the use of cleaned coal. Many of the factors
relating to availability, such as tube corrosion, soot blower failure, slag-
ging, and fouling, are influenced by the mineral content of the coal burned;
therefore, they are also influenced by the use of cleaned coal. It is dif-
ficult to quantify the effects of coal cleaning on availability because
factors unrelated to fuel quality tend to mask them. In the program, these
effects (which are benefits in all cases) are quantified through the use of a
logarithmic relationship based on the ash and sulfur in the coal (9).
Boiler Efficiency
The efficiency of a power plant boiler is often expressed in terms of the
amount of heat required to generate 1 kWh of power. This characteristic of
the power plant is affected by coal cleaning in much the same manner as avail-
ability, with one notable exception: the increased surface moisture, inherent
2-8
-------
in cleaned coal, reduces boiler efficiency. Conversely, for a given coal the
boiler efficiency generally increases as the amount of ash in the coal
decreases. Less heat is lost to the boiler ash and the fouling and slagging
of boiler heat transfer surfaces generally decrease. One additional factor
affecting efficiency, but not related to coal cleaning, is the age of the
boiler. Factors included in the equation used to quantify the effect of coal
cleaning include coal moisture content, coal ash content, and boiler age (9).
The ash content term accounts only for boiler ash heat loss.
Electrostatic Precipitation
The major factor controlling the size and cost of an electrostatic pre-
cipitator (ESP) is the specific collection area, which is determined primarily
by the dust loading and fly ash resistivity. Fly ash resistivity is related
to such factors as ash composition and sulfur content in the coal. Dust
loading is affected by coal ash content and boiler type. Coal cleaning, by
altering these coal properties, also affects the ESP cost. In some cases, the
reduction in dust loading associated with lower ash content does not compen-
sate for the increase in resistivity associated with reduced sulfur content.
Ash Disposal
The costs of transporting and disposing of ash produced by coal combus-
tion are decreased because coal cleaning reduces the ash content and increases
the heating value of the coal. Both of these factors decrease the quantity of
ash to be disposed of at the power plant.
DESIGN AND ECONOMIC PREMISES
The base case used in illustrating this model assumes a new, midwestern,
1,000-MW, pulverized-coal-fired power plant supplied by a coal-cleaning plant
at a mine 500 miles away, with transportation by unit train. The raw coal
design heat rate of the boiler is 9,200 Btu/kWh and it operates at full load
for 5,500 hours a year over a 30-year life.* The coal-cleaning process
incorporates three cleaning streams: dense-medium vessels for coarse coal (3
in. x 3/8 in.), dense-medium cyclones for medium-sized coal (3/8 in. x 28
mesh), and froth flotation for fine coal (less than 28 mesh).
The economic premises are based on regulated utility operations. Capital
investments, first-year annual revenue requirements, and levelized annual
revenue requirements are determined. The costs are based on 1982 for capital
investment and 198*1 for annual revenue requirements.
ILLUSTRATIVE EXAMPLES OF MODEL USE
The results obtained from this model can be used in many different ways,
some of which are listed on the following page:
•English to metric conversion factors are given at the end of this text.
2-9
-------
• To illustrate the effects of changes in process conditions on the
performance and economics of a coal-cleaning, an FGD, or a
combined coal-cleaning/FGD process.
• To determine the most economical method of S02 control using
coal cleaning, FGD, or a combination of both.
• To analyze the probable effects of coal cleaning on power plant
operations and economics by considering various benefits and
penalties attributable to coal cleaning.
• To compare the performance and economics for different coals, coal-
cleaning process variables, and FGD process variables.
The main objectives of the report, from which this paper is derived, are
to make potential users of this model aware of its availability and to
illustrate some possible uses. To accomplish the latter, the model was used
to simulate a range of coal properties, cleaning conditions, power plant
variables, and emission requirements. Eighty runs were made using four coals,
each cleaned at four specific gravities of separation, with the flue gas
restricted to five S02 emission limits. Other cases were run by varying
conditions such as cleaned coal topsize, power plant megawatt rating, and
distance from the cleaning plant to the power plant.
COALS AND EMISSION LIMITS
The illustrative cases are based on four coals, selected characteristics
of which are presented in Table 1. Each of these coals was cleaned by
controlling the specific gravity of the separating media at 1.40, 1.50, 1.60,
and 1.80. The emission standards evaluated consist of 0.6, 1.2, 2.0, and 4.0
Ib S02/MBtu emission limits and the 1979 New Source Performance Standards
(NSPS).
COAL-CLEANING OPERATING PERFORMANCE
In a coal-cleaning plant the specific gravity of separation is the
operating condition usually controlled to regulate the properties of the
cleaned coal. It affects the Btu recovery of the plant and the reduction in
potential S02 emissions.
The Btu recovery is a useful parameter for describing the effects of
specific gravity of separation on coal cleaning costs. It is a measure of the
raw and clean coal properties which incorporate the dual effects of yield and
Btu enhancement. For a given plant design and plant capacity the two
variables which most strongly influence the cost of cleaned coal are the plant
yield (weight recovery of clean coal from ROM coal) and the ash reduction or
Btu enhancement of the product. The capital and operating expenses of the
coal preparation plant are related to the amount of material which must be
processed. The costs added to a ton of clean coal increase as the plant yield
decreases; i.e., as more material must be mined, processed, and discarded as
refuse. Although a decrease in the specific gravity of separation reduces the
amount of material recovered, it increases the calorific value of the product.
2-10
-------
TABLE 1. CHARACTERISTICS OF SELECTED COALS
Coal Illinois No. 6
characteristic coal
Sulfur, %
Ash, %
Moisture, %
Heat content,
Btu/lb
Equivalent S02
content, Ib
S02/MBtu
4.34
29.39
1.36
9,667
8.97
Pittsburgh
coal
3.67
13-81
3.42
12,121
6.05
Upper Freeport
coal
2.32
16.80
5.60
11,668
3.97
Cedar Grove
coal
0.85
16.04
6.60 '
11,680
1.45
Basis: All values are for coal, as received
Source: Reference (10, 11).
2-11
-------
With increased heating value, a customer needs to buy a smaller weight of coal
to meet boiler heat requirements.
An important consideration for emission control applications is the
reduction of potential S02 emission which can be achieved by coal cleaning.
The emission reduction achievable by coal preparation depends on the raw coal
properties (washability), degree of size reduction (crushing), the type of
cleaning equipment, and the process conditions (specific gravity of
separation). As the specific gravity of separation decreases, the amount of
mineral contaminants (including pyritic sulfur) which is contained in the
cleaned coal product decreases. This improved coal quality is, however,
achieved at the expense of decreased Btu recovery.
The relationship between specific gravity of separation, the result-
ing Btu recovery, and reduction in potential S02 emission is shown in Table 2.
Both the Btu recovery and the potential S02 emissions decrease as the specific
gravity of separation is reduced. This is to be expected because the quantity
of coal lost in the cleaning plant refuse increases as the specific gravity
decreases. As the specific gravity of separation decreases, the sulfur reduc-
tion increases as a result of the increased pyritic sulfur removed. There is
thus a tradeoff, illustrated in Figure 4, between high Btu recovery and high
sulfur removal. Note that the raw and clean coal properties vary sub-
stantially from coal to coal. The clean coal properties are highly dependent
on the coal washability (an inherent property) and the conditions of cleaning.
A further illustration of the effect of the operating specific gravity in
the cleaning plant on the Btu recovery is provided in Figure 5. From this
graph, one can see that the relationship between Btu recovery and operating
specific gravity varies according to the type of coal being used. This
variation can be more specifically attributed to the differences in
washability of the coals. The values for coal washability are usually
presented as tabulations of selected physical and chemical characteristics for
a particular size fraction of the coal that sinks at one specific gravity and
floats at a second higher specific gravity. These characteristics must
include the weight fraction of the coal that floats at the particular specific
gravity interval, expressed as a percentage of all the coal in the size
fraction. Also required are the values for the percent ash, percent total
sulfur, and the heating value of the floating coal. Percent pyritic sulfur is
often included in the data, but it is not essential. A simplified version of
the washability table for the Pittsburgh coal used in the illustrative runs is
shown as an example in Table 3.
FGD OPERATING PERFORMANCE
The FGD process design used to illustrate the model removes 90 percent of
the S02 in the flue gas that is scrubbed. If the overall S02 removal
requirement of the FGD system is less than 90 percent, a portion of the gas is
bypassed to allow the scrubber to operate at the 90 percent removal level.
The bypassed flue gas reduces the steam requirements for reheating because it
is recombined with the scrubbed flue gas. For this reason, the amount of
bypass is the most important FGD operating condition affecting the FGD design
and cost. The portion of flue gas which can be bypassed is determined by the
2-12
-------
TABLE 2. SULFUR REDUCTION AND BTU
RECOVERY BY COAL CLEANING
Coal-cleaning operating,
Raw specific gravity
coal 1.4 1.5 1.6 1.8
Illinois No. 6 Coal
Ib S02/MBtu* 8.97 3.86 4.18 4.38 5.08
Sulfur reduction, % 57 53 51 43
Btu recovery, % - 78 88 92 96
Pittsburgh Coal
Ib S02/MBtu 6.05 3-79 4.18 4.37 4.68
Sulfur reduction, % 37 31 28 23
Btu recovery, % 82 92 95 97
Upper Freeoort Coal
Ib SOa/MBtu 3-97
Sulfur reduction, %
Btu recovery, %
Cedar Grove Coal
Ib S02/MBtu 1.45
Sulfur reduction, %
Btu recovery, %
1.76
56
78
1.17
19
91
1.93
51
89
1.20
17
93
2.03
49
93
1.23
15
95
2.35
41
97
1.25
14
96
•Assuming 100? conversion of sulfur to S02-
2-13
-------
8 -
6 -
3
4J
PQ
o
to
o>
5 *-
to
en
cfl
Illinois
No. 6 Coal
Pittsburgh / /
Coal /
Upper Freeport /
Coal /
Cedar Grove
Coal
70
I
75
I
80
85
90
I
95
100
Figure 4. Effect of Btu recovery on the actual SO- emission levels.
2-14
-------
100 -
95 _
90
o
a
a)
3
4J
CQ
85
80 -
75 -
70
Pittsburgh
Coal
Cedar Grove
Coal
Illinois No. 6
Coal
•Upper Freeport
;Coal
1.3
I I I I
1.4 1.5 1.6 1.7
Specific Gravity of Separation
Figure 5. The effect of specific gravity of separation on Btu recovery.
2-15
-------
TABLE 3. WA3MBJLHT DATA FOR THE PIITSBUBGH GOAL (4-INCH TOPSEE)
Direct, oercent
Sjpedfic
aravitv
FLQAT-1 .30
1.30-1.35
1 .35-1 .40
1.40-1.50
1 .50-1 .60
1 .60-1 .90
1.90-2.20
2.20-2.50
2.50-2.80
2.80-SINK
CumiLativeAJ3ercent
Pyritic Total
Weight
19.67
46.05
9.64
8.04
3.62
4.86
2.64
2.77
1.97
0.76
Ash
3.53
6.08
10.44
17.76
26.48
44.72
65.14
79.78
83.51
63.58
sulflir
0.61
1.42
2.96
3.99
4.56
5.62
6.59
6.47
6.12
32.81
?ilAr
1.96
2.72
4.27
5.25
5.90
6.47
7.16
6.92
6.50
34.59
Btu/lb
14,806
14,350
13,555
12,303
10,833
7,815
4,428
2,032
1,441
3,195
Weight
19.67
65.72
75.35
83.39
87.01
91.87
94.50
97.27
99.24
100.00
Ash
3-53
5.32
5.97
7.11
7.92
9.86
11.40
13.35
14.74
15.11
Pyritic
•"^l^Mr
0.61
1.18
1.40
1.65
1.77
1.98
2.11
2.23
2.31
2.54
Total
t^-fl fur
1.96
2.49
2.72
2.96
3.09
3.27
3.37
3.48
3.54
3.77
Btu/lb
14,806
14,486
14,367
14,168
14,030
13,701
13,442
13,117
12,886
12,812
Souroe: Reference (10)
2-16
-------
emission limit, the heating value, the sulfur content of the fired coal, and
the sulfur content of the ash. Figure 6 illustrates the effect that changes
in operating specific gravity have on the amount of flue gas that can be
bypassed. This set of curves (for the Pittsburgh coal) shows that the amount
of flue gas bypassing the FGD scrubbers decreases as the operating specific
gravity of the coal cleaning plant increases. This relationship is to be
expected because of the comparatively higher sulfur content and lower heating
value of coal cleaned at the higher specific gravities.
ECONOMICS
The combined computer model is a useful aid in determining the combina-
tion of coal cleaning and FGD that will meet a specified emission limit at the
minimum overall cost. The operating conditions for this minimum cost are not
the same for every case because they can change with variations in different
process characteristics (e.g., washing equipment efficiencies) and with the
particular cost element chosen for the analysis.
Capital Investment
The effects of changes in the degree of coal cleaning (in terms of Btu
recovery) on (1) the capital investment required for the cleaning plant, (2)
the change in FGD capital investment required, and (3) the capital investment
benefits and penalties, are illustrated in Figure 7. The cases are for the
Pittsburgh coal with a 1.20 Ib S02/MBtu emission limit. In terms of change
in capital investment, the degree of coal cleaning has a considerable effect
on FGD capital investment and the net capital investment. Considering all
three components, the capital investment is lower for the combined ,coal-
cleaning/FGD process than for FGD alone over much of the range of Btu
recoveries. This point is better understood if the net capital investment
curve is considered to be the incremental capital investment required to
implement a combined coal-cleaning/FGD process instead of using FGD alone.
With this in mind the reader can consider any point on the net investment
curve with a value less than zero to be a situation where combined coal-
cleaning/FGD is less costly. The minimum in the net capital investment curve
indicates the optimum level of coal cleaning with respect to capital invest-
ment. For the conditions used in Figure 7, this minimum occurs at a Btu
recovery of about 75 percent. The specific gravity of separation equivalent
to this recovery (approximately 1.37 from Figure 5) represents the conditions
at which the system should be designed to minimize the capital investment.
Annual Revenue Requirements
The relationships between the degree of coal cleaning and the various
components of net levelized annual revenue requirements are presented in
Figure 8 for the same cases that were used in Figure 7 for the total capital
investment. A comparison of these figures shows that coal cleaning has the
same general effect on FGD annual revenue requirements as it has on FGD
capital investment, which is a continuing decline as the degree of coal
cleaning increases (Btu recovery decreases). In the case of the other
benefits and penalties, however, there is an initial decline as the degree of
cleaning rises, followed by a rapid increase. This rise in costs can be
2-17
-------
100-
80_
OJ
en
P- 60'
pa
cfl
O
e
o
I
40-
20
1.40
Figure 6.
1.2
1979 NSPS
0.6
I
1.50
4.0 Ib S02/MBtu
2.0
I
1.60
Operating Specific Gravity
I
1.70
1.80
Effect of operating specific gravity at the cleaning plant on
the flue gas bypassing the FGD process (Pittsburgh coal).
2-18
-------
c
o
•H
CM
00
c
o
n
CD
•H
4J
Cfl
rH
(FGD o-
§ only)
B
CO
cu
60-
40-
20-
H
n)
-20-
-40-
o
H
-60
Coal Cleaning
Net Cost
Other Net Benefits and
Penalties
Change in FGD
costs
70
I I
80 90
% Btu Recovery
100
Figure 7.
Effect of Btu recovery on the net total
capital investment, Pittsburgh coal at
1.20 emission limit.
2-19
-------
C
O
Coal Cleaning
Other Net Bene-
fits and Penalties
Change in FGD
costs
70
Figure 8.
80 90
% Btu Recovery
100
Effect of Btu recovery on the net
levelized annual revenue requirements,
Pittsburgh coal at 1.20 emission limit,
2-20
-------
attributed to the increasing effect of coal lost to the refuse as the Btu
recovery decreases. From a further examination of Figure 8, one sees that the
combined coal-cleaning/FGD process does have a lower revenue requirement than
FGD alone for a certain range of Btu recovery. This is signified by the
points along the net cost curve that are less than the zero level, which in
this case are from about 80 to 98 percent Btu recovery.
Further comparisons of the curves in Figures 7 and 8 show that each curve
has a minimum cost point, but these minima are not located at the same Btu
recovery. The curve for the net capital investment has a minimum at about 75
percent Btu recovery, while the net annual revenue requirements curve has a
minimum at 93 percent Btu recovery for the Pittsburgh coal.
Figure 9 shows the effect of the value assigned to the coal lost to the
coal-cleaning plant refuse. One school of thought is that the coal lost
during coal cleaning should have a zero value assigned to it since the raw
coal may not have a market unless it is cleaned. This is particularly true
for coals with high ash contents. The other extreme is to assign full market
value to the coal lost. The cost of $20.81 per ton used represents the incre-
mental mining costs associated with mining the increased tonnage of coal, but
does not include transportation, UMW pension payments, royalties, sales
expenses, and profits. The correct cost to be used in the program is specific
to each case. As the value assigned to the coal lost in the refuse increases,
this becomes a very significant economic effect.
Figure 10 illustrates the reduction in levelized annual revenue require-
ments for the four coals at different 862 emission limits as compared to
sulfur removal with FGD alone. Significant differences in levelized annual
revenue requirements result from differences in coal washability. Also, the
economics generally become more favorable as the SC>2 emission limits become
less stringent; however, the curves for the Cedar Grove and Upper Freeport
coals do have a maximum point beyond which the economics become less favor-
able. These maxima are located at the points where the optimum pollution
control process changes from a combined coal-cleaning/FGD process to a coal-
cleaning process, alone. In the cases of the curves for the Illinois No. 6
and Pittsburgh coals, these curves did not reach the point at which coal
cleaning could provide the most economical pollution control process without
some degree of FGD being necessary.
It should be emphasized that the values assigned to some of the economic
benefits and penalties (e.g., boiler availability and maintenance due to
burning coal with lower sulfur and ash levels) make significant differences in
the economics. While these are based on the best information currently avail-
able, some of these relationships have not been fully defined and quantified.
Continued work in this area would lead to a better understanding of these
relationships.
PROJECT CONCLUSIONS AND RECOMMENDATIONS
• A computer model capable of calculating the economics of coal-cleaning
processes alone and combined with FGD in utility applications has been
2-21
-------
-20
Btu Recovery, %
Figure 9. Effect of Btu recovery and unit price charged to coal lost in the
refuse on the levelized annual revenue requirements, Pittsburgh
coal and 1.2 Ib SO /MBtu emission limit.
2-22
-------
30 H
en
4-1
CO
O
C
o
•H
4-J
cfl
k B-S
OJ
c »
0) M
U C
•H
(-1 C
-------
developed. Selected values from the model agreed closely with values
obtained from manual calculations.
• The illustrative runs using four premise coals had coal-cleaning sulfur
removals from 14 percent to 57 percent and met some less stringent emission
limits without FGD. The combined coal-cleaning/FGD process for each coal
had a specific gravity of separation at which costs were at a minimum.
Coal cleaning reduced the FGD costs in all cases, primarily by allowing
more flue gas to be bypassed. In some cases this reduction in FGD costs
offset the costs of coal cleaning. Case-by-case comparisons must be made
to determine the most economical approach.
• For the cases studied, it was found that the use of coal cleaning in combi-
nation with FGD can have a varied effect on the levelized annual cost of
power production. This effect, which is dependent on the coal and the
specific operating conditions, ranges from a 0.5 percent increase to a 22.6
percent decrease in those costs.
• The methods used to determine many of the other economic benefits and
penalties of coal cleaning are necessarily general in nature because de-
tailed data relating specific coal properties to boiler performance and
operating costs are scarce. Thus development and incorporation of more
detailed and quantitative data in the model would greatly increase the
usefulness of the model in assessing overall economic effects of coal-
cleaning. Similarly, the scope of the model could be increased by incor-
porating other coal-cleaning processes and alternate emission control
processes such as fabric filter baghouses and spray dryer FGD and by modi-
fying the functions to allow multiple simultaneous simulations. The eco-
nomics of retrofit situations should also be considered.
DISCLAIMER
This paper was prepared jointly by the Tennessee Valley Authority (TVA)
and the U.S. Environmental Protection Agency (EPA). TVA, EPA, or any person
acting on its behalf does not:
a. make any warranty or representation, express or implied, with
respect to the use of any information contained in this paper, or
that the use of any information, apparatus, method, or process
disclosed in this paper may not infringe privately owned rights; or
b. assume any liabilities with respect to the use of, or for damages
resulting from the use of, any information, apparatus, method, or
process disclosed in this paper.
This paper does not necessarily reflect the views and policies of TVA or
EPA.
2-24
-------
CONVERSION FACTORS
Multiply
English Unit
pound (Ib)
ton (2000 Ib)
inch (in.)
foot (ft)
British Thermal Unit (Btu)
Btu/lb
lb/106 Btu
$/ton
Btu
Ib/ft3
mile
Tyler Screen Size Mesh Openings
Mesh Size
14
28
48
100
ly.
453.59
0.907
0.0254
3.048
1054.88
2.326
429.907
1.1025
0.9480
0.0160
1.609
itta.
0.0469
0.0234
0.0117
0.0059
Sieve Opening
To Obtain
SI Unit
gram (g)
megagram (Mg)
= metric ton
meter (m)
meter (m)
Joule (J)
J/g
ng/J
$/Mg
mill/MJ =
10-3$/MJ
g/cm3
km
1.18
0.60
0.30
0.15
2-25
-------
REFERENCES
1. T. W. Tarkington, F. M. Kennedy, and J. G. Patterson, 1979, Evaluation of
Physical/Chemical Coal Cleaning and Flue Gas Desulfurization, EPA-600/7-
79-250, (NTIS PB 80-U7622), U.S. Environmental Protection Agency,
Washington, D.C.
2. C. R. Wright, 1983, Coal-Cleaning/Flue gas Desulfurization Computer Model
Users Manual. Draft report prepared by TVA for the U.S. EPA.
3. C. R. Wright, L. Larkin, F. M. Kennedy, and T. W. Tarkington, 1983,
Computer Economics of Physical Coal Cleaning and Flue Gas Desulfuriza-
tiofl. Draft report prepared by TVA for the U.S. EPA.
1. F. K. Goodman and J. H. McCreery (Battelle Columbus Laboratories), 1980,
Coal Preparation Plant Computer Model; Volume I. User Documentation,
EPA-600/7-80-010a, (NTIS PB80-177116), U.S. EPA.
5. W. L. Anders and R. L. Torstrick, 1981, Computerized Shawnee Lime/ Lime-
stone Scrubbing Model Users Manual. EPA-600/8-81-008 (NTIS PB 82-178963),
U.S. EPA.
6. Gibbs & Hill, Inc., 1978, Coal Preparation for Combustion and Conver-
sion f EPRI AF-791 , Electric Power Research Institute, Palo Alto,
California.
7. State-Imposed Coal Taxes as of September 30 f 1Q8Q. The National Coal
Association, Washington, D.C.
8. E. C. Holt (The Hoffman-Muntner Corporation), 1980, Effect of Coal
Quality on Maintenance Costs at Utility Plants, Draft report prepared for
the U.S. Department of Energy, Contract No. DEAC01-75ET12512
9. E. C. Holt, 1983, Cost/Benefit Analysis Applied to Coal and Utility Site
Specific Situations, Draft report prepared for the U.S. Environmental
Protection Agency, Contract No. 68-02-3136.
10. J. A. Cavallaro, Detailed Washability Data U.S. Bureau of Mines, Personal
communication to James D. Kilgroe, April 1977-
11. M. K. Buder, et al., Impact of Coal Cleaning oq the Cost of New Coal-
Fired Power Generation, (Bechtel National, Inc.), EPRI CS-1622, Electric
Power Research Institute, Palo Alto, California, March 1981.
2-26
-------
ECONOMIC EVALUATION OF FGD SYSTEMS
J. B. Reisdorf, R. J. Keeth,
C. P. Robie, R. W. Scheck,
T. M. Morasky
-------
ECONOMIC EVALUATION OF FGD SYSTEMS
By: J.B. Reisdorf, R.J. Keeth, C.P Robie, R.W. Scheck
Steams-Roger Engineering Corporation, Denver, Colorado
T.M. Morasky
Electric Power Research Institute, Palo Alto, California
ABSTRACT
This paper presents the estimated cost for 17 throwaway and
regenerable FGD systems based on December 1982 cost and technology. These
systems were also evaluated for operability, technical merit, and
commercial availability. The FGD systems were evaluated for high sulfur
coal applications at a hypothetical 1000 MW (two 500 MW units) power plant
in Kenosha, Wisconsin. This arbitrary reference plant was selected to
ensure consistent comparisons, and to increase the relative accuracy of
the costs presented.
A flow sheet, material balance, equipment list, system description,
and utility consumption list form the basis of each FGD evaluation. Cost
information was obtained from process vendors, Stearns-Roger information,
and published reports. Capital costs were estimated by factoring costs of
process equipment (i.e., an EPRI Class II estimate). Operating costs were
estimated from reagent and utility consumption. The levelized capital and
operating costs were developed using EPRI's standard economic premises.
The costs reported in this study are estimated within an absolute accuracy
of +30 percent. However, since methodology, scope and unit costs are
consistent, the relative accuracy between processes is approximately +15
percent.
INTRODUCTION
The cost of flue gas desulfurization (FGD) has changed in the last
few years due to technological developments, new regulatory requirements,
and advances in commercial experience. This paper presents estimated costs
for 17 throwaway and regenerable FGD systems based on 1982 costs and
technology. These systems were evaluated for operability, technical
merit, and commercial availability, and these factors are reflected in the
costs presented.
2-27
-------
The 17 FGD systems range widely in level of development, from those
processes which have received only bench-scale testing to those which have
been commercially available for many years. The estimates and evaluations
of the commercially available systems (e.g., conventional limestone and
Wellman-Lord) provide a reference to which the various processes are
compared.
The FGD systems were evaluated for both high and low sulfur coal
applications at a hypothetical 1000 MW (two 500 MW units) power plant
located in Kenosha, Wisconsin. A specific reference plant was selected to
ensure consistent comparisons, and to increase the relative accuracy of
the costs presented.
Process vendors, Steams-Roger, and published reports provide the
basis for the estimates and evaluations. Capital costs were estimated by
factoring costs of process equipment (i.e., an EPRI Class II factored
estimate with an accuracy of +_ 15 to 30 percent). Process equipment
sizing and costs were developed by Stearns-Roger. Operating costs were
calculated based on reagent and utility consumption, operating labor and
maintenance requirements. Levelized costs, i.e., the additional cost to
produce power as a result of the FGD system, were developed using EPRI's
standard economic premises. Costs have an absolute accuracy of +30
percent. However, since methodology, scope, and unit costs are
consistent, the relative accuracy of the estimates is approximately +15
percent.
Cost and technical evaluations are summarized for each of the 17 FGD
processes and are arranged in the following three categories:
High-sulfur coal—throwaway, where the FGD system is applied to a
high-sulfur coal, and the waste product is gypsum or a sludge.
Processes in this category include conventional limestone,
Chiyoda Thoroughbred 121, Dowa, limestone with forced oxidation,
Saarberg-Holter, limestone dual alkali, lime dual alkali, and wet
lime.
High-sulfur coal—regenerable, in which the FGD system is applied
to a high sulfur coal and produces a by-product while
regenerating the absorbing reagent. Processes in this category
include Wellman-Lord, MgO, SULF-X, Flakt-Boliden, aqueous
carbonate, and Conosox.
Low-sulfur coal—throwaway, where the FGD system is applied to a
low sulfur coal, and the waste product is a sludge or dry solid.
Processes in this category include conventional limestone, lime
spray dryer, and nahcolite/trona injection.
2-28
-------
HIGH SULFUR COAL--THROWAWAY
Processes in this category include first generation FGD technologies
(conventional limestone and wet lime), and second generation FGD
technologies which have the potential for lower operating and capital
costs. The conventional limestone process has experienced the broadest
application and is considered to be the reference or base case for this
category.
Capital and levelized busbar costs are shown in Figures 1 and 2,
respectively, with the base case (conventional limestone) process shown at
the left. The level of development is also indicated for each process.
Figure 2 shows capital, and fixed and variable operating costs which
comprise the total levelized busbar cost. The costs exclude particulate
removal equipment. It has been assumed that flue gas reheat will be
required for all high sulfur, wet FGD systems. This assumption is site
specific and should be treated accordingly. The following conclusions can
be drawn from Figures 1 and 2:
The Chiyoda Thoroughbred 121 process has the lowest levelized
busbar cost. This is largely due to its low operating cost which
results from high limestone utilization, low reagent cost, and
low solid waste handling costs.
The capital cost of the lime dual alkali process is a little less
than the other dual alkali processes because of its less costly
lime handling and preparation area (versus limestone) and a lower
process contingency due to a higher level of development.
The lime dual alkali process has one of the highest variable
operating costs primarily because it uses lime for regeneration.
The Dowa process has the lowest variable operating cost because
it produces a gypsum waste which is less expensive to dispose of
than the sludge produced by other processes which do not oxidize
« the waste solids.
Cost sensitivities to various parameters were developed for each of
the high sulfur coal-throwaway processes by adjusting capital and
operating costs. The parameters investigated are shown in Table 1 along
with the results of the sensitivity analyses.
Cost sensitivities to both zero-inflation and 10-year plant life were
also investigated for each process. The zero-inflation parameter reduced
system costs to approximately 40 percent of the base case levelized cost.
A 10-year plant life was found to reduce process levelized costs to
approximately 80 percent of the original base case cost. However, the
cost of each process relative to the other processes did not change.
2-29
-------
to
I
u>
o
180-
160-
140-
g 120-
82 100-
_IV
cC\
L* 801
a.
a:
<-> 60-
40-
20-
0-
(t
r
CONVEN
LIMES
D
75
d
:: I/
TIONAL CHIY
TONE THOR(
121
a
r
D :
\Q m
ODfl DO
3UGHBRED
D Q
75 1"
WR FORCE
OX ID/
OF LI
9
75
(
rm i:
ID SflflF
^TION HOL1
MESTONE
a
K
5 1
30 ttti
?8ERG- LIME
PER DUfll
ALK/
30 ^
IE
ESTONE LIM
DUfl
^LI ALK
a
D 1E
'° ITTT
: :
E WE
L LI
ALI
&
55
r
^E
(A) COMMERCIAL DEVELOPMENT
(B) FULL-SIZE (20-100 MW) DEVELOPMENT
© PILOT PLANT DEVELOPMENT
© BENCH-SCALE DEVELOPMENT
NOTE: COST «/KW IN DEC. 1982 DOLLARS
FIGURE 1. High Sulfur Coal - Throwaway Processes Capital Cost
-------
\/%/X\ VARIABLE COST
fcggfl FIXED COST
B88883 CAPITAL COST
o
a:
UJ
Ld
20 n
15-
10-
5-
18
14
14
16
16
16
17
20
CONVENTIONAL CHIYODA DOHA
LIMESTONE THOROUGHBRED
121
FORCED SAARBERG- LIMESTONE LIME WET
OXIDATION HOLTER DUAL DUAL LIME
OF LIMESTONE ALKALI ALKALI
® COMMERCIAL DEVELOPMENT
© FULL-SIZE (20-100 MW) DEVELOPMENT
© PILOT PLANT DEVELOPMENT
© BENCH-SCALE DEVELOPMENT
NOTE: COST LEVELI7ED MILL/KWH IN DEC. 1982 DOLLARS
FIGURE 2. High Sulfur Coal - Throwaway 3rocesses Levelized Busbar Cost
-------
FABLE 1. HIGH SULFUR COAL - THRGWAWAY PROCESSES COST SENSITIVITY
N3
I
SENSITIVITY PARAMETER
% SULFUR IN COAL
REAGENT CONSUMPTION/COST
DISPOSAL COST
POWER COST
MAINTENANCE REQUIREMENT
ALTITUDE (GAS FLOW)
CONVENTIONAL
LIMESTONE
...
..
* *
* *
* *
* *
CHIYODA
THOROUGHBRED
121
* * *
* *
*
* *
* w
* #
§
Q
# » »
W *
»
W *
* *
» *
FORCED
OXIDATION OF
LIMESTONE
* » *
*
*
* *
» *
* *
SAARBERG-
HOLTER
* * #
•
*
« *
..
* *
LIMESTONE
DUAL ALKALI
* * *
» *
..
* *
* *
* *
LIME DUAL
ALKALI
* * *
* *
* *
* *
« *
* *
UJ
t-
UJ
5
# » #
* *
..
» »
* *
* *
•LITTLE OR NO COST SENSITIVITY
"MODERATE COST SENSITIVITY
***HIGH COST SENSITIVITY
•••EXTREME COST SENSITIVITY
-------
Sensitivity analyses allow utility personnel to estimate the cost of
an FGD system given the site specific conditions. Base case values are
defined by General and Process Specific Design Criteria assumed for this
study.
A brief summary of each process follows.
Conventional Limestone (Reference Process)
A limestone slurry solution absorbs 50-2 in a spray tower and forms
a calcium sulfite/sulfate sludge. The major advantage of this process is
its wide range of commercial development and its abundant and low cost
absorbent. Also, this process usually will meet the 50-2 reduction
requirement for all types of coals. This system has the potential
disadvantages of equipment scaling, plugging, corrosion, and erosion.
Chiyoda Thoroughbred 121
This process uses a limestone slurry in a single vessel jet bubbling
reactor to remove S02, oxidize the reaction products, and form calcium
sulfate. The major advantages of this second generation process relative
to the conventional limestone process are the better settling, filtering,
and structural properties of the waste, and better utilization of the
limestone. Slurry pumps are not required, reducing capital and
maintenance costs. Scaling and plugging problems are reduced because
precipitation occurs on recirculated crystals rather than on equipment
surfaces. A disadvantage of this process is its low level of development
in this country.
Dowa
An aluminum sulfate solution absorbs 502 in a packed tower. The
spent reagent is oxidized and then regenerated with limestone to form a
gypsum product. The major advantages of the Dowa process relative to the
conventional limestone process are better settling, filtering, and
structural properties of the waste, more efficient use of limestone, and
better turndown and load following capabilities. A possible disadvantage
is that waste disposal regulations may require a lined waste disposal site
because of the soluble aluminum sulfate occluded in the waste.
Additionally, this process has a low level of development in this country.
Forced Oxidation-Limestone
A limestone slurry sprayed into a tower removes 502 an^ forms a
precipitate of calcium sulfite/sulfate. This waste is subsequently
oxidized to form a gypsum product. The major advantages of this process
relative to conventional limestone are the better settling, filtering, and
structural properties of the waste.
2-33
-------
Saarberg-Holter
A clear alkaline (lime) solution is used in a "Rotopart" scrubber to
remove SO? and form a precipitate of calcium sulfite/sulfate. The
calcium sulfite is oxidized within the scrubber vessel to form a gypsum
product. The major advantages of this process relative to conventional
limestone are virtually 100 percent utilization of the reagent, better
settling, filtering, and structural properties of the waste, and better
turndown and load following capabilities. This process results in a lower
levelized cost than the conventional limestone process even though it uses
a more costly reagent. A throwaway FGD process using lime as a reagent
requires more potable make-up water than does a throwaway process
utilizing limestone; lime slaking requires uncontaminated water. This is
a possible disadvantage to the water and wastewater management system.
Some additional observations can be made when comparing oxidized
by-product processes to one another. The Saarberg-Holter, Chiyoda
Thoroughbred 121, and Forced Oxidation-Limestone processes are similar
because they utilize a single loop system to produce a gypsum waste.
Oxidation of the calcium sulfite to gypsum is internal to the absorber
module in the Saarberg-Holter and Chiyoda Thoroughbred 121 processes and
external in the Forced Oxidation-Limestone process. External oxidation
increases pumping requirements and results in higher maintenance and fixed
costs than the Saarberg-Holter and Chiyoda Thoroughbred 121 processes.
The Saarberg°Holter uses lime, which is more expensive than the limestone
reagent used in the Chiyoda Thoroughbred 121 and Forced Oxidation-
Limestone processes. The use of lime causes the Saarberg-Holter to have a
higher variable operating cost than the other two processes. The Chiyoda
Thoroughbred 121 is attractive because it has the lowest levelized busbar
cost compared to the other two processes. However, it's level of
development is less than the commercial Saarberg-Holter and Forced
Oxidation-Limestone processes.
Limestone Dual Alkali
A sodium sulfite solution is used in a spray tower to remove S02»
Limestone is added to the spent solution in a separate loop to
simultaneously regenerate the spent solution and form a calcium
sulfite/sulfate sludge. The major advantage of this process relative to
conventional limestone is lower scaling, plugging, corrosion, and erosion
potential which results in higher FGD system availability and lower
maintenance requirements. The high reactivity of the sodium reagent
allows a lower liquid/gas (L/G) ratio, lower pumping power, and better
turndown and load following capabilities than the conventional limestone
process. Disadvantages include the possibility that local regulations may
require the use of a lined waste disposal site because of the soluble
sodium salts occluded in the waste, and low level of development.
2-34
-------
Lime Dual Alkali
Similar to the Limestone Dual Alkali system, a sodium sulfite
solution is used in a spray tower to remove SOo. However, lime, rather
than limestone, is added to the spent absorbent solution to simultaneously
regenerate the sodium sulfite solution and form a calcium sulfite
sludge. The major advantages of this process relative to conventional
limestone are once again lower scaling, plugging, corrosion, and erosion
potential which result in higher availability and lower maintenance
requirements. As previously discussed, the sodium reagent allows a lower
L/G ratio, lower pumping power consumption, and better turndown and load
following capabilities than the conventional limestone process. Local
regulations may require the use of a lined waste disposal site because of
the soluble sodium salts also occluded in this waste.
The Lime Dual Alkali, Limestone Dual Alkali, and Dowa processes are
similar in that they are dual-loop processes. These processes use a clear
absorbing solution in the first loop and regenerate the spent absorbent in
the second loop (external to the absorber) where the solid wastes are
generated. The Dowa process has the lowest levelized busbar cost;
however, it has only been tested at the pilot plant level (10 MW). The
higher-cost limestone dual alkali process has been demonstrated at a 23 MW
facility. The lime dual alkali process has the highest cost; however, it
has been operated successfully on a commercial level.
Wet Lime
A lime slurry solution absorbs S02 in a spray tower and forms a
calcium sulfite/sulfate sludge. The advantages of the Wet Lime process
relative to conventional limestone are less erosion, approximately 100
percent use of the reagent, and better turndown and load following
capabilities. The major disadvantage is the high cost of lime versus
limestone. The lime slurry also requires service water for the slaking
process instead of the less expensive cooling tower blowdown water used
for limestone slurry preparation.
HIGH-SULFUR COAL--REGENERABLE WITH BY-PRODUCT
Although processes in this category exhibit higher costs than
throwaway processes, by-product regenerable processes are beneficial when
space or disposal requirements are restricted and/or when there is an
available market for process by-products. The Wellman-Lord process has
experienced the broadest application and is considered the basis for
comparison with other processes in this category.
The capital and levelized busbar costs are shown in Figures 3 and 4,
respectively, with the base case Wellman-Lord process shown on the left.
The level of development is also indicated for each process. Figure 4
shows the capital, fixed, and variable operating costs which comprise the
2-35
-------
o
u
o.
cC
U
600 -
500-
400-
300-
100-
0
©
390
400
©
430
270
295
WELLMflN- MGO SULF-X FLAKT- flQUEOUS CONOSOX
LORD BOLIDEN CARBONATE
® COMMERCIAL DEVELOPMENT
(B) FULL-SIZE (20-100 MW) DEVELOPMENT
© PILOT PLANT DEVELOPMENT
© BENCH-SCALE DEVELOPMENT
NOTE: »/KW IN DEC. 1982 DOLLARS
FIGURE 3. High Sulfur Coal - Regenerate Processes Capital Cost
2-36
-------
VARIABLE COST
FIXED COST
CAPITAL COST
en
o
u
501
45-
40
35-
30-
25-
20
15
10-
2B
19
29
30
45
WELLMAN- MGO
LORD
SULF-X FLAKT- AQUEOUS CONOSOX
BOLIOEN CARBONATE
® COMMERCIAL DEVELOPMENT
(D FULL-SIZE (20-100 MW) DEVELOPMENT
© PILOT PLANT DEVELOPMENT
© BENCH-SCALE DEVELOPMENT
NOTE: LEVELIZED MILL/KWH IN DEC. 1982 DOLLARS
FIGURE 4. High Sulfur Coal - Regenerable Processes Levelized
Busbar Cost
2-37
-------
total levelized busbar cost. It has been assumed that flue gas reheat
will be required for all high sulfur, wet FGD systems (this does not
include the spray dryer based Aqueous Carbonate process). This assumption
is site specific and should be treated accordingly.
An economic evaluation of these regenerable FGD processes resulted in
the following major conclusions:
The MgO process has the lowest levelized cost but is very
sensitive to sulfuric acid by-product credit. This recovery
process carries the lowest capital cost of all the recovery
processes.
The Sulf-X process has a slightly higher levelized cost than has
the MgO process but has the advantage of possibly reducing NOX
emissions simultaneously. This process does not have a high
level of development.
• • Some of the other regenerable processes, notably the Wellman-Lord
and Flakt FGD systems, produce a concentrated 50-2 stream which
could be used for sulfuric acid production. However, for the
purposes of this evaluation, these processes were combined with
an Allied Chemical/Claus sulfur plant to produce elemental
sulfur. The levelized costs for both these processes would be
lower if they were combined with a sulfuric acid plant.
Producing sulfuric acid product avoids the need to purchase
high-cost methane and provides (presently) for a more valuable
by-product.
Cost sensitivities to various parameters were developed for each
process by adjusting the capital and operating cost. These parameters are
shown in Table 2 along with the relative results. The cost sensitivity
analyses to both zero-inflation and 10-year plant life were found to have
no significant impact on the relative costs between the regenerable
processes. This result was discussed previously in the High Sulfur-
Throwaway section. Once again, sensitivity analyses allow utility
personnel to estimate the cost of an FGD system given the site specific
conditions.
Wellman-Lord (Reference system to which other regenerable processes are
compared)
A concentrated solution of sodium sulfite is used in a tray tower to
produce a sodium bisulfite-rich solution. This solution is thermally
regenerated to sodium sulfite in a steam evaporator and a concentrated
S02 gas stream is produced. The 502 is subsequently converted to
sulfur in an Allied Chemical plant. The Wellman-Lord process has the
advantages of producing a salable by-product and does not experience
scaling or plugging. Disadvantages include high operating costs because
of the large quantities of low pressure steam and methane required and
2-38
-------
high capital costs. Additionally, the Wellman-Lord process is sensitive
to flue gas participates, HC1 and 503 and therefore requires a
prescrubber. A prescrubber results in higher operating costs due to
higher pressure drop and prescrubber loop blowdown treatment.
MgO
A magnesium oxide solution is sprayed into a grid-packed tower to
remove 50-2 and form magnesium sulfite/sulfate. A calcining operation
regenerates the waste solids to magnesium oxide and forms an 50-2 gas.
The 5.0-2 gas is subsequently processed to produce sulfuric acid. The
advantage of this process relative to the Wellman-Lord process is much
lower operating cost. This is due to steam production during the
calcination step, whereas the Wellman-Lord process consumes large
quantities of steam. The disadvantages of the MgO process include a
higher reagent makeup requirement due to a high oxidation rate, use of a
high temperature (1800°F) calciner, potential pluggage of the gas/solid
separation equipment at the calciner outlet, and corrosion/erosion
problems in the slurry handling equipment.
TABLE 2. HIGH SULFUR COAL - REGENERABLE PROCESSES COST SENSITIVITY
SENSITIVITY
PARAMETER
% SULFUR IN COAL
REAGENT CONSUMPTION/COST
BY PRODUCT VALUE
POWER COST
MAINTENANCE REQUIREMENT
ALTITUDE (GAS FLOW)
WELLMAN-
LORD
» » »
•
* »
* »
* *
*
0
C3
5
* *
* *
» » •
» »
• *
•
X
uL
_i
CO
* »
•
* ft
»• *
« •
»
FLAKT
BOLIOEN
* *
* *
* c
* *
• »
'•
AQUEOUS
CARBONATE
* *
* »
» »
* *
* *
•
CONSOX
••
* *
' •
••
•
•LITTLE OR NO COST SENSITIVITY
"MODERATE COST SENSITIVITY
•••HIGH COST SENSITIVITY
'••EXTREME COST SENSITIVITY
2-39
-------
Sulf-X
An iron sulfide slurry passes through a packed tower removing
SO? from the flue gas. The spent absorbent is regenerated in a
calciner, producing an elemental sulfur off-gas which is condensed to form
liquid sulfur. The advantages of the Sulf-X process relative to the
Wellman-Lord process are a low oxidation rate which results in a lower
reagent make-up requirement, potential for simultaneous NOX removal, and
low fuel consumption. Disadvantages of the Sulf-X process include:
possible release of H^S gas in the absorber during an upset; handling of
three reagents (pyrites, coke, and sodium sulfide) which increases
material handling; slurry absorption which increases the potential for
equipment erosion; high temperatures required in the kiln increase the
potential for process control problems and equipment failures; and low
level of development.
Flakt-Boliden
A sodium citrate solution flows downward (countercurrent to the flue
gas) through a packed tower to remove S02. The spent absorbent solution
is regenerated in a stripper column while producing an 50-2 rich gas
stream, which is converted to sulfur in an Allied Chemical/Claus plant.
The advantage of the Flakt-Boliden process relative to the Wellman-Lord
process is its lower oxidation rate, which results in a lower reagent
consumption rate. The vacuum crystallization process used in the
sulfate-purge system is a disadvantage in the Flakt-Boliden process
because it is more difficult to control than the thermal crystallization
system in the Wellman-Lord process.
Aqueous Carbonate
A solution of sodium carbonate absorbs 50-2 in a spray dryer. The
waste product from the spray dryer is treated using a coke reductant in a
molten-salt bed to regenerate the scrubbing reagent while producing a
hydrogen sulfide-rich gas. The hydrogen sulfide is converted to sulfur in
a Claus plant. The major advantages of the aqueous carbonate process
relative to the Wellman-Lord process are a lower steam consumption rate
and insensitivity to oxidation in the absorber. Disadvantages result from
the complexity of the operation, handling two phase flow and high
equipment operating temperatures. Additionally, the sodium reagent and
fly ash may tend to accumulate on the spray dryer walls, reducing
absorption efficiency; also the crystallizer and carbonator are
susceptible to plugging due to the formation of bicarbonate crystals.
This process is at the demonstration stage of development.
Conosox
A solution of potassium salts is used to remove 502 in a packed
tower. The spent potassium bisulfite absorbent is converted to a
2-40
-------
potassium thiosulfate solution, which is in turn reduced with carbon
monoxide to yield regenerated potassium carbonate and hydrogen sulfide
gas. The hydrogen sulfide gas is subsequently converted to sulfur in a
Claus plant. The major advantages of the Conosox process relative to the
Wellman-Lord process are the use of a more reactive reagent (potassium vs.
sodium), lower oxidation rate (i.e., lower reagent consumption), lower
steam consumption, and a low pressure drop across the absorber.
Disadvantages of the Conosox process are the use of high temperatures,
high pressure vessels, high consumption of fuel and liquid oxygen, and the
potential for release of hydrogen sulfide gas in the absorber.
LOW-SULFUR COAL--THROWAWAY
Included in this category are lime spray drying, dry injection of
nahcolite/trona, and the conventional limestone slurry scrubbers. The
conventional limestone process has experienced the broadest application
and is considered as the basis for comparison in this category.
The capital and levelized busbar costs are shown in Figures 5 and 6,
respectively, with the base case (conventional limestone) process shown at
the left. The level of development is also indicated for each process.
Figure 6 shows the capital, and fixed and variable operating costs which
comprise the total levelized busbar cost. The costs exclude particulate
removal equipment.
The following conclusions can be drawn from Figures 5 and 6:
The Trona Injection process has the lowest levelized cost
resulting from very low capital and fixed operating costs. The
high variable operating cost due to consumption of expensive
reagents offset the former cost advantages substantially.
The Lime Spray Dryer has a low operating cost due to a
combination of factors: low consumption of relatively expensive
lime reagent (compared to limestone) due to the low sulfur coal,
low waste disposal costs due to the dry waste product, and low
energy requirements.
Nahcolite operating cost exceeds both spray drying and trona
injection. However, the operating cost is very sensitive to the
assumed nahcolite cost which is 140 $/ton in this evaluation.
Other EPRI reports have based their evaluations on lower
nahcolite costs ranging between $50-100/ton delivered which
reduced the operating cost significantly. This range was
selected based on more recent reagent cost data from suppliers
using solution mining methods and on transportation costs for
sodium-based compounds shipped throughout the U.S.
The cost sensitivities for each of the processes were developed by
adjusting the capital and operating cost in accordance with various
parameters. The parameters investigated for their effect on cost are
shown in Table 3 along with the results.
2-41
-------
V)
o
u
(-«=
i—I
Q.
cc
O
175-
150-
125-
100-
75-
50-
25-
0
110
25
25
CONVENTIONAL LIME
LIMESTONE SPRAY
PROCESS DRYER
NflCHOLITE TRONfl
INJECTION INJECTION
® COMMERCIAL DEVELOPMENT
® FULL-SIZE (20-100 MW) DEVELOPMENT
© PILOT PLANT DEVELOPMENT
© BENCH-SCALE DEVELOPMENT
NOTE: COST «/KW IN DEC. 1982 DOLLARS
FIGURE 5. LGV* Sulf'jr Coal - Throwaway Processes Capital Cost
2-42
-------
VARIABLE COST
FIXED COST
CAPITAL COST
o
QL
tC-
00 =
00
UJ
20 n
15-
10-
5-
0
(A)
8
7.5
8
CONVENTIONAL LIME
LIMESTONE SPRAY
PROCESS DRYER
NACHOLITE TRONA
INJECTION INJECTION
® COMMERCIAL DEVELOPMENT
© FULL-SIZE (20-100 MW) DEVELOPMENT
© PILOT PLANT DEVELOPMENT
© BENCH-SCALE DEVELOPMENT
NOTE: 30 YR. LEVELI2ED COST MILL/KHH IN DEC. 1982 DOLLARS
FIGURE 6. Low Sulfur Coal - Throwaway Processes Levelized Busbar Cost
2-43
-------
TABLt 3. LOW SULFUR COAL - THROWAWAY PROCESSES COST SENSITIVITY
SENSITIVITY
PARAMETER
% SULFUR IN COAL
REAGENT CONSUMPTION/COST
DISPOSAL COST
POWER COST
MAINTENANCE REQUIREMENT
ALTITUDE (GAS FLOW)
LIME
SPRAY
DRYER
«*
..
*
+ *
*
* «
NACHOLITE
INJECTION
# * * *
* # # »'
«
*
..
•
'LITTLE OR NO COST SENSITIVITY
"MODERATE COST SENSITIVITY
"HIGH COST SENSITIVITY
"EXTREME COST SENSITIVITY
2-44
-------
Conventional Limestone (Reference System)
This process is identical to that described in the conventional
limestone discussion under high-sulfur—throwaway processes, but differs
in that it has been evaluated for a low sulfur coal application requiring
only 70 percent 50-2 removal.
Lime Spray Dryer
The lime spray dryer process contacts flue gas with lime slurry in a
spray dryer to produce a dry waste. The slurry reacts with SO? to form
a solid which is collected in a baghouse with the fly ash. Relative to
the conventional limestone process, the advantages of this process are dry
waste handling, lower maintenance requirements, lower energy requirements,
and lower capital and operating costs. Disadvantages include the
potential to "blind" the baghouse bags during process upsets, and the
potential for scale formation in the spray dryer.
Nahcolite Injection
In this process nahcolite (sodium bicarbonate) is pulverized and
injected into the flue gas duct to react with SOo. The dry waste is
removed in a baghouse. The advantages of nahcolite injection relative to
the conventional limestone process are dry waste handling, lower
maintenance requirements, reduced energy requirements, and lower capital
and operating costs. Disadvantages include increased reagent use below
275°F flue gas temperature, lack of a reagent supplier, the possibility of
local regulations requiring the use of a lined waste disposal site because
of the soluble sodium salts in the waste, and high sensitivity to largely
unknown reagent cost. The cost of nahcolite was estimated at $l40/t which
was obtained from the trona mining industry. However, nahcolite has never
been mined commercially and the cost is very uncertain (see the
conclusions summarized previously). Nahcolite injection is best suited to
locations near the reagent source in Colorado and in dry regions where
waste disposal is less of a problem.
Trona Injection
The pulverization and injection of trona (sodium bicarbonate plus
sodium carbonate) into the flue gas was investigated as an alternative to
nahcolite injection. Trona is currently less expensive than nahcolite and
is commercially available. The lower cost of trona results in a variable
operating cost of 6 mills/kWh versus 7 mills/kWh with nahcolite. The
capital and fixed cost of the trona injection system are the same as that
of nahcolite. The levelized busbar cost of the trona injection system is
the lowest in this category; however, its level of development is lower
than the commercial lime spray dryer and the conventional limestone
process.
2-45
-------
ESTIMATING PROCEDURE FOR RETROFIT FGD COSTS
R. R. Mora, P. A. Ireland, R. J. Keeth,
T. M. Morasky
-------
ESTIMATING PROCEDURE FOR RETROFIT FGD COSTS
By: R. R. Mora, P. A. Ireland, and R. J. Keeth
Steams-Roger Engineering Corporation
Denver, Colorado
and
T. M. Morasky
Electric Power Research Institute
Palo Alto, California
ABSTRACT
A procedure has been developed for utility engineers to estimate
site-specific flue gas desulfurization (FGD) retrofit costs on existing power
plants. This procedure has been developed for six FGD processes: limestone,
lime, dual alkali, lime spray dryer, wallboard gypsum, Wellman-Lord, and
Chiyoda. Economic results from the calculation procedure include capital
cost, operating and maintenance (O&M) costs, levelized cost, and cost per ton
of sulfur dioxide removed.
INTRODUCTION
Previous studies have estimated the incremental cost of an FGD system on
a new power plant; however, limited information is available concerning the
cost of an FGD system retrofitted to an existing plant. Many utility
engineers estimate retrofit FGD capital cost for a specific existing plant
using an average "retrofit factor" of 1.3 as a guide. Stearns-Roger, under
EPRI sponsorship, has developed a method to more accurately estimate FGD
retrofit costs. A report, entitled "Retrofit FGD Estimating Guidelines," will
be published in mid-1984.
The procedures outlined in this study consider major factors which affect
FGD capital costs at a particular site such as chimney modifications, space
availability, demolition and relocation of existing equipment, underground
obstructions, soil conditions, local labor rates, degree of site congestion,
and accessibility of the site for construction. Several process-related items
also are considered, including percent sulfur in coal, percent S02 removal,
plant size, flue gas flow rate, and gas reheat requirements.
2-47
-------
The calculation methods described in this study should apply to over 90%
of FGD retrofit situations. When all the retrofit items previously mentioned
are totalled, retrofit FGD costs can range from 10 percent to 100 percent
greater than an FGD system on a new plant. The FGD retrofit costs developed
through use of this method are sufficient for budget and preliminary planning
purposes, but they do not replace the need for more detailed engineering.
The method outlined in this paper establishes a procedure for estimating
FGD retrofit costs, but does not provide guidance for selecting the best
option or combination of options (i.e., coal cleaning vs. FGD). FGD, one of
the more effective S02 removal techniques, is also one of the more costly
compliance options. Therefore, accurate FGD economics are a key issue
affecting compliance strategies. FGD also is one of the more difficult
options to estimate due to specifics of retrofit constraints, site
characteristics, space availability, and type of FGD process selected.
DISCUSSION OF ESTIMATING PROCEDURE
The basis for economic evaluations in this paper is the capital cost of a
new plant FGD system developed in a previous EPRI study, "Economic Evaluation
of FGD Systems," CS-3342. This report evaluates FGD costs for a new plant
having two (2) 500 MW units, located in Kenosha, WI, burning a A percent
sulfur coal, designed for 90 percent SO? removal efficiency, zero bypass,
and 50°F indirect steam reheat. The detailed retrofit estimating method is
intended specifically for application to the costs developed in CS-3342;
however, the estimating method can be applied to new plant costs from any
source.
Once the new plant FGD costs are determined, adjustments are made for
scope, process, location, and retrofit differences from the base case new
plant. Figure 1 outlines the procedure used to adjust the base new plant FGD
cost and to determine retrofit cost. The resultant retrofit factors can be
applied to cost data from other sources, but this alternate cost data must be
based on the same system scope.
SCOPE ADJUSTMENTS
FGD retrofit construction may require work not necessary for a new plant
FGD system. An example would be the addition of an acid resistant chimney
downstream of the scrubber or reinforcing a boiler to withstand the higher
draft potential. Inclusion of these additional costs in the new plant FGD
cost provide a more accurate assessment of the total retrofit impact.
2-48
-------
NEW PLANT
FGD COST
(Capital Cost)
($/kW)
T
SCOPE ADJUSTMENTS
($/kW)
PROCESS FACTORS |
I
LOCATION FACTORS
RETROFIT FACTORS
TOTAL PROCESS
CAPITAL
ADDITIONAL CAPITAL
REQUIREMENTS
TOTAL RETROFIT
CAPITAL REQUIREMENT
Figure 1. Logic Diagram - Total Retrofit FGD Capital Requirement
2-49
-------
Scope adjustments are defined as additional equipment required for a
particular FGD retrofit, which would not normally be required for an FGD
system on a new power plant. New plant FGD costs should include the following
equipment:
• 50-2 removal absorber tower
• Duct work and structural support steel from the Induced Draft
(I.D.) Fan outlets to the scrubber and from the scrubber outlet
to the chimney
• Incremental I.D. fan capacity which provide sufficient draft to
overcome the pressure loss in the scrubber
• Reagent feed equipment, including unloading and storage,
conveying, and reagent preparation equipment
• Waste handling, including the primary and secondary dewatering
equipment as well as sludge fixation equipment, if required
• General support equipment, which includes makeup and seal water
tanks and pumps, and an instrument air system
Scope adjustments include any additional equipment costs due to the
addition of a scrubber and not included in the equipment areas defined in the
previous paragraph. The cost adjustments can contribute an additional
10 percent to a retrofit FGD system's capital cost. The following areas are
potential scope adjustments:
e A new chimney, or protection of the existing chimney against
the acidic effects of a wet scrubber, including application of
linings or pressurization of the annular space
• Structural reinforcement of the boiler to provide negative
pressure surge protection, or a draft control system to prevent
surges in the boiler
• Any demolition and/or equipment relocation necessary to provide
space for the new equipment, such as building removal or cable
tray rerouting to allow construction of the FGD system
2-50
-------
PROCESS ADJUSTMENTS
The second adjustment in Figure 1 is the modification of the new plant
cost for process differences from the base case. These might include the
following:
-------
RETROFIT ADJUSTMENTS
Retrofit adjustments consider the retrofit difficulty which is defined by
the following additional site-specific items:
• Access and congestion
e Underground obstructions
e Ductwork tie-in distances
• Distance between the scrubber and waste handling
Site accessibility, area availability and congestion are some of the most
serious problems facing any retrofit installation. A severely limited plant
site may require location of FGD equipment at some distance from the boiler,
which will necessitate long runs of ductwork and piping, increased pumping
costs, etc. Elevated construction will also increase the cost of engineering
and installation. Poor access to a construction site limits the ability of
large cranes to move materials, increasing costs. Finally, congestion may
limit crew work space and affect the design and layout of piping, ductwork,
etc. The resulting lost productivity and unusual design requirements can
significantly increase the total capital requirement for the FGD retrofit.
Underground obstructions are defined as those items which cannot be
removed because of logistics, economic considerations, or service
requirements. They may include circulating water pipes, gas mains, and duct
banks. The cost impacts of these interferences are calculated.for each
equipment area, based on the degree of difficulty these obstructions would add
to the equipment installations. The resulting retrofit factors can then be
tabulated for each FGD process.
The distance between the scrubber and the plant tie-in point, and then
from the scrubber to the chimney can significantly affect the total retrofit
capital cost. Every linear foot of additional ductwork will increase the
installed cost of the FGD retrofit above the new plant FGD ductwork design.
The final site-specific retrofit adjustment considers the distance
required between the scrubber and the waste disposal and handling system.
This increase in piping distance to the primary and secondary dewatering
systems can add significantly to the installed FGD retrofit capital
requirement.
Access, congestion, and the ductwork distance are the major contributors
to the retrofit adjustment. Accessibility and congestion can increase the
retrofit cost of any particular system by 50 percent. The retrofit
adjustments have the largest effect on the FGD capital cost. These factors
can increase capital cost by 70 percent, depending on the degree of retrofit
difficulty.
2-52
-------
TOTAL RETROFIT CAPITAL REQUIREMENT
The four adjustment factors applied to the new FGD system cost yield the
total process cost (TPC). The total process capital is the constructed cost
of on-site FGD and related equipment, plus direct and indirect construction
costs. These additional costs must be calculated by factoring the TPC to
derive the total retrofit capital requirement. These costs include the
following:
o Engineering - Fees for hours spent by equipment supplier,
architect-engineer and the utility to finish the installation
and FGD system startup.
a Escalation to the date of construction - Cost increases due to
equipment price increases after the base case construction date
of December, 1982.
• Cost of general support facilities - Including roads, offices,
laboratories, fences, etc. which need to be built or
reconstructed.
a Allowance for funds during construction - Those charges which
accumulate annually as equipment is purchased and installed,
prior to plant startup.
a Royalty allowance - The additional fee paid to the process
vendor for the use of a proprietary system design.
o Preproduction costs - Intended to cover the operator training,
equipment checkout, extra maintenance, and inefficient use of
materials during startup.
a Reagent inventory capital - Covers the value of raw materials
and other consumables as an initial capital expenditure.
The summation of these costs and the TPC provides the total retrofit
capital requirement for the FGD system. If only capital costs are required,
the evaluation is complete; however, if the levelized busbar cost is desired,
additional calculations are required. These steps are shown in Figure 2.
2-53
-------
TOTAL RETROFIT
CAPITAL REQUIREMENT
Fixed Charge Rate
FIRST-YEAR O&M
COST
(5/kW-yr)
Levelizing Factors
I
TOTAL LEVELIZED
RETROFIT FGD
COST
(.$/kWh or $/ton S02)
Figure 2. Logic Diagram - Total Levelized Retrofit FGD Cost
2-5^
-------
OPERATING AND MAINTENANCE COSTS
The levelized busbar cost, in mills/kWh or $/ton of SOo removed, is
computed by applying a two-step levelizing procedure. The fixed charge rate
when multiplied by the capital cost produces a levelized capital cost over the
projected FGD plant life. Levelization factors are applied to the first-year
O&M costs to yield a levelized operating cost. These levelization factors are
based on the economic premises defined by one's own company. They assume a
specific expected inflation rate and discount rate over the life of the
plant. O&M costs for the wet limestone system include the following:
e Fixed costs:
—O&M personnel
—Maintenance material
—Administration and support personnel
9 Variable costs:
—Power
—Reagent
—Steam usage
—Waste disposal cost
—Water consumption
Adding the levelized capital cost to the levelized operating cost results
in the total levelized busbar cost for the FGD system. <
LEVELIZED COST
As noted, the retrofit capital and O&M costs are levelized by applying
the proper fixed charge rate and levelization factor. These economic factors
consider the interest rate, remaining plant life, capacity factors, reagent
costs, utility cost, and sludge disposal cost. As the remaining plant life
decreases, the fixed charge rate increases and the levelization factor
decreases. Therefore, the fixed charge rate will cause the levelized capital
cost to increase as the remaining plant life decreases. Levelization factors
applied to the O&M cost cause the levelized costs to decrease as the remaining
plant life decreases.
SAMPLE RETROFIT ESTIMATING WORKSHEETS
The retrofit estimating procedure is illustrated in Tables 1 through 3.
Only the detailed retrofit cost estimating worksheets are presented. The
tables and graphs needed to complete these forms will be published in the
final report.
2-55
-------
Table 1, the Total Retrofit Capital Estimate Summary, is an example of
the procedure used to estimate the total capital requirement for a wet
limestone FGD system. In this table, the FGD system capital costs are divided
into process areas. Factors are developed for each process area and each cost
correction required. Multiplying these factors and summing the retrofit cost
yields the total process capital. Economic factors are then calculated from
and added to the Total Process Capital to obtain the total retrofit capital
requirement. The factors developed for Table 1 are based on Steams-Roger
historical plant cost data and an assessment of retrofit impacts.
Following Table 1 are examples of figures and tables (provided in the
final report) which will allow the user to determine the adjustment factors
previously discussed in this paper. Table 2 is an example of the location
dependent adjustments, in this case the effect of soil on plant cost. For
example, a soil with poor bearing capacity, shallow bedrock, and ground water
would require five times the foundation cost that dense insitu soils, having a
bearing capacity of 4 ksf and over, would require. The foundation cost may
represent only 1 percent of the installed equipment cost. Therefore the
overall effect of a 500 percent increase in the foundation cost is a 5 percent
increase in the total capital requirement.
Figure 3 provides an example of the graphical presentation of unit
specific scope adjustments. This example allows calculation of the cost of a
chimney addition based on the flue gas flow rate to the absorber.
Once these cost factors are determined and the total capital requirement
calculated, the first-year O&M cost must be derived. The first-year O&M costs
are divided into fixed and variable costs. Fixed costs calculated for the
retrofit estimating procedures are based on EPRI standards and are determined
by factoring the equipment cost. The factors and equations necessary to
determine these costs will be included in the final report.
The variable O&M cost is calculated for each retrofit situation. The
user can calculate variable costs for power, reagent usage, steam, waste
disposal, and water within two hours using the questions, tables, and graphs
presented in the soon-to-be published study.
Table 3, the First-Year O&M Summation Sheet, is used to tabulate and sum
the individual costs; this yields the total fixed and variable first-year O&M
cost. The results of Table 3 serve as the basis for determination of the
levelized busbar cost. The levelized capital cost and first-year O&M cost are
calculated by applying the fixed charge rate and levelization factors
respectively.
The final report will provide the figures which graphically illustrate
the fixed charge rate (FCR) and levelization factors as a function of
remaining life of the plant. By applying these factors and the appropriate
conversion constraints, the user can determine the levelized busbar cost in
mills/kWh or $/ton S02 removed.
2-56
-------
TABLE 1. TOTAL RETROFIT CAPITAL COST ESTIMATE SUMMARY
Base
Process* Process Location Retrofit Retrofit
Process Area Description Cost $/kW Factors Factors Factors Cost $/kW
Reagent Feed
S02 Removal
Flue Gas Handling
Waste Handling
Gen. Support Equip.
Scope Adjustments
Chimney 1
Boiler 1
Draft Controls (
Demolition 4 Relocation I
Other i
23.0
78.0
30.0
23.6
2.6
( )
C )
( )
( )
x ( :
x (
x ( ;
x ( ;
x ( ;
x ( :
x c ;
x (
x 1.0
x ( ;
> x (
» x (
> x (
1 x (
) x (
> x (
> x (
> x (
x (
) x (
) x ( )
) x ( )
) x ( )
) x ( )
) x ( )
) x ( )
) x ( )
) x ( )
) x ( )
) x ( )
(A) Total Process Capital (Sum of Retrofit Cost)
(B) Escalated Total Process Capital, A x ( ) =
(C) General Facilities (0 to 15% of B)B x ( )
(D) Engineering and Home Office Fees =
(10 to 15% of B) 10% + ( ) + ( )
(E) Total Plant Cost TPC, (B + C + D)
(F) AFDC
1 yr = 0, 2 yr = .018, 3 yr = .037, ( , AFDC) x ( , TPC) =
(G) Total Plant Investment TPI, (E + F)
(H) Royalty Allowance =
(I) Preproduction Costs (2% of G)
(J) Inventory Capital =
(K) Initial Catalyst & Chemicals
(L) Total Capital Retrofit Requirement
(G + H + I+J + K)
* Base Process Cost shown includes EPRI standard project and process
contingencies. Costs shown are for a 500 MW conventional limestone FGD
process.
2-57
-------
TABLE 2. SOIL-GEOTECHNICAL FACTORS
NJ
I
00
DESCRIPTION
Insitu soils are dense, having
a bearing capacity of 4 ksf and
over, or existing structural
fill, footings can be placed
after excavation.
Loose insitu soils, excavation
& replacement with compacted
structural fill is required.
A. Sandy clay
B. Sand
C. Loose sand
D. Very loose silt
Poor bearing capacity soils
with deep bedrock formation.
Bearing or friction type pile
is the most feasible solution.
Bedrock is:
A. 25' below surface(base case)
B. 35' below surface
C. 50' below surface
D. 75' below surface
E. 100 'below surface
F. 125'below surface
*Limited headroom requiring
special drilling or driving
equipment; add to above.
REAGENT
FEED
0.93
0.96
0.97
0.98
1.0
1.0
1.01
1.03
1.04
1.06
1.07
.08
SULFUR
DIOXIDE
REMOVAL
0.97
.98
.99
.99
1.0
1.0
1.0
1.01
1.01
1.02
1.02
.025
FLUE
GAS
0.93
.96
.97
.99
1.0
1.0
1.01
1.03
1.04
1.06
1.07
.075
WASTE
HANDLING
0.95
.97
.98
.99
1.0
1.0
1.01
1.02
1.03
1.04
1.05
.05
GEN. SUPP
EQUIPMENT
0.93
.96
.97
.99
1.0
1.0
1.01
1.03
1.04
1.06
1.07
.075
CHIMNEY
0.95
.97
.98
.99
1.0
1.0
1.01
1.02
1.03
1.04
1.05
.05
-------
TABLE 2. SOIL-GEOTECHNICAL FACTORS
Ln
vo
DESCRIPTION
Poor bearing capacity soils
with presence of ground water,
shallow bedrock, drilled piers
are necessary.
A. Ground water at 10'
B. Ground water at 5'
C. Ground water at 3'
Flood plain or by river bank,
drilled piers are suitable.
Caisson liner & wet drilling
are necessary.
*limited headroom requiring
special drilling or driving
equipment. Add to above.
REAGENT
FEED
1.0
1.01
1.03
1.07
^
.08
SULFUR
DIOXIDE
REMOVAL
1.0
1.0
1.01
1.02
.025
FLUE
GAS
s.
1.0
1.01
1.03
1.07
.075
WASTE
HANDLING
1.0
1.01
1.02
1.05
.05
GEN. SUPP
EQUIPMENT
1.0
1.01
1.03
1.07
.075
CHIMNEY
1.0
1.01
1.02
1.05
.05
-------
c
0
s
T
F
A
C
T
0
R
2.3-
2.2-
2.1-
2.0-
1.9-
1.8-
1.7-
1.6-
1.5-
1.4-
1.3-
1.2-
1.1-
1.0-
0.9-
0
Chimney
T
Base Case Chimney Cost
Cost for a Net 500 MW
Unit Assumed to be $7/kW -
(Erected)
T
T
0.5 1.0 1.5 2.0 2.5 3.0
Flue Gas Flow Rate to Scrubber, ACFM x 1.0 Million
Figure 3. Process Flow Rate Scope Adjustments
2-60
-------
TABLE 3. FIRST-YEAR O&M COSTS SUMMATION SHEET
Fixed O&M Cost
A) Operating Labor
B) Maintenance Labor
C) Maintenance Material
D) Administrative and Support Labor
Total Fixed O&M Cost (A+B+C+D)
Variable O&M Cost
E) Reagent Consumption
1)
2)
Lime
Limestone
$/kW-Yr.
$/kW-Yr.
Total Reagent Consumption (1+2)
F) Steam Usage Cost
G) Power Usage Cost
H) Water Usage Cost
I) Waste Disposal Cost
Total Variable O&M Cost (E+F+G+H+I)
$/kW-Yr.
$/kW-Yr.
$/kW-Yr.
$/kW-Yr.
$/kW-Yr.
$/kW-Yr.
$/kW-Yr.
$/kW-Yr.
$/kW-Yr.
$/kW-Yr.
$/kW-Yr.
CONCLUSION
The estimating guidelines presented in this study provide a consistent
methodology that allows the user to estimate site specific retrofit cost. The
procedure should save the user considerable time in assessing the cost of FGD
options available for retrofit. FGD processes to which this procedure is
directly applicable include conventional wet limestone, lime dual alkali,
forced oxidation of limestone, lime spray drying, Chiyoda Thoroughbred 121,
and Wellman-Lord. Simplified selection logic provided in the final report
should aid the utility engineer in his selection of a process design for
retrofit evaluation.
2-61
-------
COMPARATIVE COSTS OF S02 REMOVAL TECHNOLOGIES
J. 0. Milliken
-------
COMPARATIVE COSTS OF
S02 REMOVAL TECHNOLOGIES
Presented by:
John 0. Milliken
Industrial Environmental Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
ABSTRACT
Comparative Costs of S02 Removal Technologies
The costs of retrofitting four types of in-plant S02 control technologies
are compared for a 500 MW pulverized coal-fired utility boiler. Technologies
examined are at various stages of commercialization and market penetration,
and include: (1) fully commercial - limestone FGD, (2) early commercial -
adipic acid enhanced FGD, (3) early commercial - lime spray drying, and
(4) developmental - limestone injection multistage burners. Cost comparisons
are made by process subarea, by operating cost categories, by capital to
operating cost ratio, and by overall sulfur removal effectiveness.
2-63
-------
COMPARATIVE COSTS OF S02 REMOVAL TECHNOLOGIES
I. INTRODUCTION
As a single emission source category, coal-fired utility boilers represent
the largest source of S02 in the eastern United States. The options for
reducing S02 emissions from this source category include alternative
generating technology (e.g., nuclear and hydroelectric), fuel switching
(e.g., to low sulfur coal, oil, or gas), coal cleaning, and retrofit of
in-plant S02 control technology. The objective of this paper is to
compare the cost and performance characteristics of four retrofittable
in-plant S02 control technologies: (1) limestone flue gas desulfurization
(FGD), (2) adipic acid enhanced limestone FGD, (3) lime spray drying, and
(4) limestone injection multistage burners (LIMB). Note that these tech-
nologies are at various stages of commercialization or market penetration,
and that they represent different degrees of S02 control in terms of
percent removal. For example, conventional limestone FGD is a fully
commercial process; adipic acid enhanced limestone FGD has been demonstrated
on a commercial scale and is beginning to penetrate the FGD market; lime
spray drying is in the early commercial phase for western U.S. coals, and
in the demonstration phase for eastern U.S. coal applications; and
LIMB technology is clearly in the developmental phase.
It is important to note that the cost comparisons made here are of a
general nature only and cannot be used to distinguish the benefits or
disadvantages of different technologies for a site-specific application.
Selection of a preferred S02 control technology for any given site must
be made on the basis of site-specific conditions and performance require-
ments (e.g., percent S02 removal required, existing ductwork arrangement,
coal type, capacity). However, developing costs on a consistent basis
(e.g., for retrofit on a 500 MW boiler firing 3.5 percent sulfur bituminous
coal) and with a uniform cost methodology will be useful to environmental
policy analysts who need general cost and performance characteristics for
in-plant retrofittable S02 control technologies. This analysis is also
useful in quantifying the range of potential cost benefits for emerging
technologies such as LIMB, as well as the benefits of enhancements to
existing conventional technology such as the adipic acid enhancement to
limestone FGD.
2-64
-------
II. METHOD OF ANALYSIS AND DESIGN BASES
In comparing the costs of the four technologies, it was desired to
illustrate clearly where differences between the technologies exist, as
well as to make overall cost and performance comparisons of the four
technologies. To accomplish this, the method of analysis chosen was to
compare direct investment costs for various process subareas and operating
expense categories. After examining direct investment or process subareas
and operating expense categories, the usual parameters of total capital
investment ($/kW) and net annual operating expense (mills/kWh) were
accumulated for each of the four technologies. Then, using a uniform
fixed charge rate (16 percent), the annual revenue requirement attributable
to the control technology was computed. To examine the differences in
capital intensity for the four technologies, the capital investment to
operating expense ratio was computed. Lastly, a composite cost and
performance parameter of $/ton* S02 removed was computed for each technology.
The methods of comparing costs are summarized in Table 1.
Key design bases for this study were: retrofit to a 500 MW boiler firing
3.5 percent eastern U.S. bituminous coal. To account for retrofit, a 30
percent penalty was added to all estimates of process capital direct
investment. It was assumed that the boiler would have 20 years remaining
life at 5500 hr/yr subsequent to the retrofit. Although it is acknowledged
that these assumptions are optimistic relative to current expectations of
boiler life and capacity utilization, there is a trend in the utility
industry toward extending useful plant lifetime. All costs are presented
in 1983 dollars. These design assumptions are summarized in Table 2.
The cost estimates used in this study are based on previous estimates
prepared for EPA by the Tennessee Valley Authority's Office of Power (1,2).
The limestone FGD and limestone adipic acid FGD costs are based on estimates
provided in Reference 1, and the lime spray dryer costs are based on
estimates provided in Reference 2. The basis for the LIMB cost estimate
is discussed below.
Note that different S02 removal efficiencies are assumed for the four
technologies: limestone FGD - 90 percent, limestone adipic acid FGD -
90 percent, lime spray drying - 70 percent, and LIMB - 55 percent.
Clearly, these differences must be considered when comparing the capital
investments and operating expenses. However, a major objective of this
analysis is to compare the cost-effectiveness of the technologies in
terms of dollars per ton S02 removed. For this reason, the design basis
removal efficiencies were assumed at levels characteristic of the
technology capabilities and expected typical applications.
i
*Short tons are used here for the convenience of the reader. This may be
converted to metric units by noting 1 short ton equals 907.2 kg. See
Table of Conversion Factors at end of paper.
2-65
-------
TABLE 1. METHODS OF COST COMPARISONS
• Process subarea direct investments
• Operating expense categories
• Total capital investment ($/kW)
• Net annual operating expense (mills/kWh)
• First year annual revenue requirement (mills/kWh)
e Capital investment to operating expense ratio
• Cost and performance ($/ton S02 removed)
TABLE 2. GENERAL DESIGN BASES FOR FOUR TECHNOLOGIES COMPARED
• Retrofit to 500 MW
• Eastern U.S. coal-fired boiler
• 3.5% Sulfur bituminous coal
• 20 years remaining life at 5500 hr/year
• All costs presented in $1983
2-66
-------
For the limestone FGD and limestone adipic acid FGD cases, a design
employing spray tower absorbers, and utilizing forced oxidation with
landfill disposal of solid waste was used (1). The design also included
one spare absorber train, emergency bypass of 50 percent of the scrubbed
gas, and reheat to 79.4°C (175°F). The battery limits for the capital
cost estimates began with the inlet plenum downstream of the ESP and
extended to (but did not include) the stack plenum. The calcium to
sulfur ratio for limestone FGD was 1.4, and for adipic acid limestone
FGD, 1.07. All estimates were adjusted to reflect mid-1983 dollars. A
30 percent retrofit penalty was applied to all process capital direct
investment items, and for total indirect capital items, a uniform factor
of 45 percent of total capital investment was assumed.
Adjustments to the TVA estimates for lime spray drying (2) on high sulfur
eastern coal were more substantial than adjustments to the TVA limestone
FGD cases (1). Material handling and feed preparation estimates for lime
spray drying were adjusted downward to reflect a calcium to sulfur ratio
assumption of 1.3 as opposed to the TVA assumption of 1.5. Note that the
lime spray drying retrofit system examined here assumes retrofit to an
existing 500 MW unit with an existing electrostatic precipitator. No
fabric filter was included in the retrofit, in contrast to the TVA estimate
for a new lime spray drying system with a fabric filter (2). However,
provision for 0-10 $/kW increment to the direct investment for the lime
spray dryer was made for possible upgrading of the particulate control
system. This translates into an additional total capital investment cost
of about 0-20 $/kW, when the indirect capital is added. Also for the
lime spray drying estimates, the solid waste disposal estimates were
reduced so that this cost reflected the incremental cost due to the
additional solid waste from the sorbent (reacted and unreacted). The
original TVA estimate appears to provide for solid waste of both coal fly
ash and sorbent. Other adjustments made here are the same as for the
other technologies and include the 30 percent retrofit penalty to process
capital direct investment as well as adjustments for 1983 dollar basis.
Capital investment costs were compared for each process subarea noted in
Table 3. Operating expense categories examined are noted in Table 4.
Note that these do not include capital related charges, which are treated
later as a component of first year revenue required.
In the estimation of raw material expenses, sorbent costs were estimated
according to stoichiometric ratios of 1.4 for limestone FGD, 1.07 for
adipic acid limestone FGD, 1.3 for lime spray drying, and 3.6 for LIMB.
Stoichiometric ratio is defined here as moles Ca per mole S02 absorbed.
Note that for LIMB, the calcium to sulfur stoichiometric ratio is
usually expressed as moles of calcium per mole of sulfur in coal. At
55 percent sulfur capture, the stoichiometric ratio of 3.6 based on
sulfur absorbed is equivalent to a stoichiometric ratio of 2 based on
inlet sulfur. The cost of limestone delivered was assumed to be $15/ton,
and for lime, $55/ton.
2-67
-------
TABLE 3. PROCESS SUBAREAS COMPARED
« Materials handling and preparation
e Gas handling and S02 absorption
9 Boiler and burner modifications
a Oxidation, reheat, solids thickening
• Particulate control system upgrade
• Solid waste disposal (landfill)
TABLE 4. OPERATING EXPENSE CATEGORIES
a Raw materials
e Operating labor and supervision
e Utilities
« Maintenance, labor, and materials
• Analysis
« Plant and administrative overhead
2-68
-------
III. RESULTS
The estimated direct investment for each subarea is detailed in Table 5.
In the materials handling and preparation area, notice that the estimated
direct investment costs are very similar for limestone FGD, limestone
adipic acid, and lime spray drying with all of these at approximately
$20/kW. The lower direct investment for the LIMB materials handling and
preparation area is primarily due to the much simpler sorbent preparation
process which consists of crushing and grinding. One reason for the
lower LIMB limestone preparation equipment is that the other systems
examined require make-down, storage, and pumping of sorbent slurries.
The materials handling direct investment used for LIMB was estimated to
be the same as for limestone FGD. The estimate of the sorbent preparation
(i.e., crushing and grinding) for LIMB, was obtained by scaling-up an
estimate for LIMB limestone preparation on a 200 MW system (3).
From Table 5, it is also seen that the gas handling and SC>2 absorption
areas are the single largest components of the capital investment for the
first three processes. For purposes of comparison, the analagous LIMB
process area can be thought of as the boiler and burner modifications
required to achieve S02 capture. This LIMB item was estimated on the basis
of low-NOx B&W dual-register burners with lighters equipped for limestone
injection, modification to boiler pressure parts, new control system, and
new coal and limestone feeder (3). Due to the uncertainty and site-specific
variability expected for this item, a range of 5-15 $/kW has been given.
Retrofit of a dual register or distributed mixing burner to an existing
wall-fired boiler that does not have outboard or tertiary air ports would
indeed require modification to pressure parts for retrofit installation.
However, if an internally staged low-NOx burner were employed, the retrofit
would not require the costly modification to the boiler tubes proximate
to the burner.
Miscellaneous process areas contributing to the capital costs of limestone
FGD and limestone adtpic acid FGD include oxidation, reheat, and solids
thickening. That these process areas are not applicable to either lime
spray drying or LIMB explains some of the cost benefits of spray drying
and LIMB relative to wet FGD. The contribution of oxidation, reheat, and
solids thickening to direct investment is given in Table 5.
For both the lime spray drying and LIMB technology, we assume that:
0-10 $/kW in direct investment will be required for potential particulate
control system upgrade. This is to cover, if needed, gas conditioning
(e.g., humidification, 863 conditioning), increased ESP plate area for
higher grain loadings, and upgraded solids handling equipment.
In this analysis, solid waste disposal costs are treated by assigning
direct investment estimates to landfill construction and landfill equip-
ment. This, of course, has the effect of increasing the estimate of
capital relative to the case where estimates of solid waste disposal
costs are handled strictly as an operating expense. This item adds
2-69
-------
TABLE 5. ESTIMATES OF DIRECT INVESTMENT FOR
DIFFERENT PROCESS SUBAREAS
Estimated Direct Investment, $/kWa>b
Limestone Lime Spray
Limestone Adlplc Acid Drying LIMB
Materials handling
and preparation
Gas handling and
S02 absorption
Boiler and burner
modifications
Oxidation, reheat,
solids thickening
Particulate control
system upgrade
Solid waste
disposal (landfill)
20.4
88.6
19.2
79.1
28.0
25.3
10.0
8.7
19.2
57.4
0-10
8.5
8.9
5-15
0-10
9.2
Total direct investment
147.0
132.3
85.1-
95.1
23.1-
43.1
aAll cost estimates for direct investment were adjusted to 1983$ using
Marshall and Stevens Installed equipment cost indices, assuming a
1983 M&S* equipment index of 759.
^Direct investment estimates Include a 6 percent add-on for services
and utilities, and a 30 percent retrofit penalty for all process sub-
areas except solid waste.
2-70
-------
approximately $10/kW of direct investment (about $20/kW total capital)
to each retrofit control technology. The variation noted in Table 5 is
ascribed to differences in the quantity and characteristics of solid
waste generated by the four processes.
Indirect capital was estimated at 45 percent of total capital investment
for all technologies examined. This includes indirect fixed investment
(e.g., engineering design and supervision, architect and engineering
contractor, construction expense, contractor fees, and contingencies),
and non-depreciable capital investment items (e.g., allowance for start-up
and modifications, interest during construction, royalties, land, and
working capital). Although this fraction of total capital for indirect
investment is somewhat greater than usually recommended for either solid
processing or solid/fluid processing, it is consistent with the ratio
used in the more detailed estimates provided in References 1 and 2. The
contingency component of indirect investment would normally be increased
for a developmental process such as LIMB. This adjustment is not made
here since generous contingencies for the developmental process subareas
have already been accounted for in the expressed range of direct invest-
ment for boiler and burner modification, and upgrade to particulate
matter control. Direct investment, indirect capital costs, and total
capital investment are summarized in Table 6.
Operating expense categories examined for the four technologies are listed
in Table 4, and the results for operating expenses are summarized in
Table 7. In the case of raw materials, the higher cost for lime spray
drying relative to the other three technologies clearly reflects the
higher cost per ton of lime ($55/ton) versus limestone ($15/ton). For
the three technologies employing limestone sorbent, the raw materials
cost estimates in general are proportional to the calcium to sulfur ratio
with a slight exception for the limestone adipic acid case which includes
the incremental costs of the adipic acid. The lower operating labor and
supervision costs, also noted in Table 7, for lime spray drying and LIMB
relative to limestone and limestone adipic acid reflect the simpler
nature of the former two processes.
The operating expense category of utilities includes direct operating
costs for steam, process water, electricity, and diesel fuel. Electricity
consumption represents the major cost element in this category and is due
primarily to providing motive power for crushing and grinding, materials
handling, and pumping. The estimates for electricity and other utilities
are also presented in Table 7. Note that the electricity consumption is
greatest for the limestone process and least for lime spray drying.
Although one might expect the electricity assumption for LIMB to be
similar to that for the lime spray drying system, a higher value was used
to reflect the increased requirement for grinding and crushing due to the
larger amount of sorbent. Steam requirements for limestone adipic acid
are about 20 percent less than for conventional limestone due primarily
to lower reheat requirements. No steam requirements were assumed for
either lime spray drying or LIMB. However, an energy penalty for calcina-
tion of limestone was introduced into this category for the LIMB process.
2-71
-------
TABLE 6. TOTAL CAPITAL INVESTMENT SUMMARY
Limestone
Limestone
adipic acid
Lime spray
drying
LIMB
Direct
Investment
$/kW
147
132
85-95
23-43
Indirect
Capital Costs3
$/kW
120
108
70-78
19-35
Total Capital
Investments
$/kW
267
240
155-173
42-78
alndirect capital estimated at 45% of total capital investment.
2-72
-------
TABLE 7. ANNUAL OPERATING EXPENSES BY CATEGORIES
Annual Operating Expense, mllls/kWh
Limestone Lime Spray
Limestone Adipic Acid Drying LIMB
Raw materials3 0.78 0.67 1.95 1.23
Operating labor 0.44 0.40 0.29 0.23
and supervision
Utilities
Steamb 0.46 0.36 — 0.28
Water . 0.01 0.01 0.01
Electricity 0.73 0.57 0.47 0.62
Diesel Fuel 0.05 0.03 0.01 0.02
Maintenance, labor, 2.05 1.84 1.36 0.53
and materials0
Analysis 0.04 0.02 0.03 0.03
Plant and adminis- 1.52 1.36 1.01 0.47
trative overhead^
Net annual 6.08 5.26 5.13 3.41
operating expenses
3Limestone and lime costs were assumed at $15/ton and $55/ton,
respectively, on a delivered basis.
bsteam for limestone and limestone adipic acid was for reheat;
although no steam was required for LIMB process, a calcination
energy penalty, of 1400 kJ/kg (600 Btu/lb) limestone was added here.
°Maintenance, labor, and materials were estimated at 8 percent of
direct process investment plus 3 percent of waste disposal investment.
dpiant and administrative overhead was estimated at 60 percent of
the sum of operating labor, maintenance, and analysis.
2-73
-------
For this purpose it was assumed that 600 Btu/lb* limestone was required
for calcination. Process water costs are a minor component of utilities
and contribute on the order of 1 percent to the utility cost category.
Diesel fuel expenses relate primarily to the solid waste disposal area,
and in general vary with the amount of solid waste generated per process.
Maintenance, labor, and materials represent an important component of the
net annual operating expenses as indicated in Table 7. As in the EPA/
TVA studies (1,2) these expenses were estimated at 8 percent of direct
process investment plus 3 percent of the waste disposal investment.
Again, the higher maintenance labor and materials expenses for limestone
and limestone adipic acid reflect the more complex nature and more capital
equipment intensive feature of these processes relative to lime spray
drying and LIMB.
Plant and administrative overhead expenses are indirect operating expenses.
They are estimated at 60 percent of process conversion expenses minus
utilities expense. This is equivalent to 60 percent of the sum of operating
labor and supervision; maintenance, labor, and materials; and laboratory
analysis. Plant and administrative overhead expenses are noted in Table 7.
Net annual operating expenses are defined as the sum of the annual direct
operating expenses plus the indirect plant and administrative overhead.
Note that net annual operating expenses is analogous to the frequently
used term "total operating and maintenance (O&M)" expense, and does not
include capital related charges. The net annual operating expenses for
the four processes examined are given at the bottom of Table 7. Note
that the net annual operating expenses for LIMB are substantially lower
than for the other three processes. It is expected that the lime spray
drying net annual operating expenses would be more comparable to LIMB
except for the high cost per ton of lime relative to limestone.
Revenue requirement estimates are presented as first year annual revenue
requirements. For each process the revenue requirement is equal to the
sum of the capital charge and the net annual operating expense. The
capital charge is computed by multiplying a fixed charge rate of 16
percent times the total capital investment. The 16 percent fixed charge
rate results from the economic assumptions noted in Table 8. It is
expected that shortening the anticipated remaining life of the process
from 20 to 15 or 10 years would have a substantial impact on the fixed
charge rate, resulting in higher annual capital charges. Table 9 details
the capital charges, operating expenses, and revenue requirements for the
four processes.
A measure of the capital intensity of each process is taken as the ratio
of the annual capital charge to the net annual operating expense. These
ratios are listed in Table 10, and indicate the varying degrees of capital
intensity for the four processes examined. The most notable difference
is the low capital intensity feature of the LIMB process relative to
*To convert to metric, see Table of Conversion Factors at end.
2-74
-------
TABLE 8. ECONOMIC PREMISES FOR FIXED CHARGE RATE
System lifetime
Accounting lifetime
Method of depreciation
Income tax rate
Investment tax credit
Property tax and insurance
Weighted cost of capital, after tax
20 years
10 years
straight line
50 percent
10 percent
2.25 percent
10 percent
Fixed charge rate (resulting from above) 16 percent
TABLE 9. FIRST YEAR ANNUAL REVENUE REQUIREMENT
Limestone
Limestone
adipic acid
Lime spray
drying
Capital Charge
mills/kWh
7.8
7.0
4.5-5.0
Net Annual
Operating Expense
mills/kWh
6.1
5.3
5.1
First Year Annual
Revenue Requirement
mills/kWh
13.9
12.3
9.6-10.1
LIMB
1.2-2.3
3.4
4.6-5.7
2-75
-------
TABLE 10. CAPITAL TO OPERATING EXPENSE RATIOS
Limestone
Limestone adlplc acid
Lime spray drying
LIMB
1.3
1.3
0.9
0.5
aRatio presented here is defined as the
annual capital charge divided by the
net annual operating expense.
TABLE 11. COST AND PERFORMANCE SUMMARY
Limestone
Limestone adipic acid
Lime spray drying
LIMB
Revenue
Requirement
mills/kWh
13.9
12.3
9.6-10.1
4.5-5.6
SO 2 Removal
Efficiency
%
90
90
70
55
Cost Effectiveness
$/ton SO? Removed
620
550
550-580
330-410
2-76
-------
limestone and limestone adipic acid, with lime spray drying having an
intermediate capital intensity ratio. The low capital intensity feature
of LIMB will tend to improve its already favorable cost-effectiveness
relative to more capital intensive processes, for shorter lifetime,
lower capacity applications. For this reason, the economic preference
for LIMB in a retrofit application should be even stronger than that
shown here.
IV. SUMMARY
In summary, the costs of limestone and limestone adipic acid FGD are
substantially higher than those for either lime spray drying or LIMB,
although the former two processes achieve a 90 percent S02 removal
efficiency while lime spray drying and LIMB achieve 70 and 55 percent S(>2
removal, respectively. The lime spray drying process estimated here was
assumed to achieve an SC>2 removal efficiency of 70 percent, and the LIMB
process, an S02 removal efficiency of 55 percent. To compare the cost per
unit of SC>2 removed, the $/ton of S02 removed was computed for each of
the four processes. These results are listed in Table 11. Note that the
effectiveness of lime spray drying retrofit to an eastern U.S. high-
sulfur coal-fired boiler is very similar to limestone adipic acid FGD,
certainly within the range of uncertainties in these estimates. Both the
limestone adipic acid and lime spray drying appear to be slightly more
cost-effective than conventional limestone FGD. Even within the range of
uncertainty in the LIMB cost estimate, the cost-effectiveness parameter for
LIMB is 50 to 90 percent more favorable than the cost-effectiveness
parameter for conventional limestone FGD. This economic preference, coupled
with the simplicity of the LIMB process relative to conventional FGD,
suggests that LIMB has a strong potential for medium performance applications
on retrofit S02 control technology for eastern U.S. high-sulfur coal-fired
boilers.
V. TABLE OF CONVERSION FACTORS
Multiply By To Obtain
ton (short) 907.2 kg
$/ton 1.102 x ID"3 $/kg
Btu/lb 2.326 kJ/kg
2-77
-------
VI. REFERENCES
1. Burnett, T.A., et al. (Tennessee Valley Authority). Economic
Evaluation of Limestone and Lime Flue Gas Desulfurization Processes,
EPA-600/7-83-029 (NTIS PB84-133644), May 1983.
2. Burnett, T.A., and K.D. Anderson (Tennessee Valley Authority).
Technical Review of Dry FGD Systems and Economic Evaluation of Spray
Dryer FGD Systems, EPA-600/7-81-014 (NTIS PB81-206476), February 1981
3. Private communication with R. Johnson, Tennessee Valley Authority,
September 1983.
2-78
-------
SESSION 3: MATERIALS OF CONSTRUCTION
Chairman: Charles E. Dene
Electric Power Research Institute
Palo Alto, CA
-------
EPRI RESEARCH ON CORROSION AND DEGRADATION
OF MATERIALS FOR FGD SYSTEMS
B. C. Syrett
-------
EPRI RESEARCH ON CORROSION AND DEGRADATION
OF MATERIALS FOR FGD SYSTEMS
by: B. C. Syrett
Materials Support Group
Electric Power Research Institute
P.O. Box 10412, Palo Alto, CA 94303
ABSTRACT
An EPRI survey of materials problems in full-scale lime/limestone wet
scrubbers in the USA revealed that the most frequent and most critical
failures occurred in the outlet ducts and the stack, but that a significant
number of failures also occurred in the prescrubbers, absorbers, reheaters,
dampers, pumps, piping and valves. The root cause of most of these failures
was corrosion or degradation of the materials of construction. Over the last
few years, EPRI has initiated a multitude of projects aimed at understanding
and eliminating corrosion induced failures in scrubbers. Primary emphasis has
been on metallic materials, but an increasing amount of effort is being
directed towards coatings, nonmetallic materials of construction, and
corrosion control techniques. The range of EPRI's research on corrosion and
degradation of materials for FGD systems is described in this paper.
BACKGROUND
Some of the more critical corrosion problems faced by the U.S. utility
industry are in the flue gas path between the coal-fired boiler and the top of
the stack. High-sulfur and medium-sulfur coals are being used to an
increasing extent and, with the introduction of ever more stringent
environmental control laws, utilities have been required to remove a large
percentage of the sulfur dioxide (802) from the flue gas before it is emitted
from the stack. Among the flue gas desulfurization (FGD) systems that have
been devised, lime and limestone scrubbers are the most developed. Typically,
flue gas at about 150°C (300°F) will enter the quench section of a scrubber
(the "prescrubber") then flow through a spray of lime or limestone slurry
delivered from nozzles at the top of the scrubber. As the gas passes through
this spray zone (the "absorber"), SO2 and other acid gases are absorbed and
neutralized by the slurry. The scrubbed gas leaves the absorber at the
adiabatic saturation temperature, typically 50-55°C (120-130°F); it then
passes through mist eliminators and on to the stack. Unless some type of
reheat system is installed after the mist eliminators, the water-saturated gas
condenses on the walls of the duct leading to the stack and on the stack
lining. These condensates can be very corrosive. Reheaters are sometimes
used to raise the temperature of the outlet gas to 70-95°C (160-200°F) so that
3-1
-------
corrosion in the outlet duct and stack is minimized and so that the gas is
more buoyant and the plume is less visible.
A survey of all full-scale lime/limestone wet scrubbers in the.U.S.A. was
performed for EPRI in 1980 to establish the type and frequency of materials
related problems being experienced in FGD systems (_!_). The survey revealed
that all manner of construction materials have been used, including metals,
organic coatings and linings, and ceramic and other inorganic materials. The
most frequently cited problem areas were the outlet ducts and the stack, both
of which are particularly critical components in that failures may require
boiler shutdown and loss of generating capacity. Less frequent and less
critical failures were noted in the prescrubbers, absorbers, reheaters,
dampers, pumps, piping, and valves. This survey made it clear that corrosion
and, to a lesser extent, erosion were causing extensive and very costly
problems in power plant FGD systems. There appeared to be no consensus on the
choice of construction materials, and this uncertainty was not improved by
reports that materials that worked well in one scrubber failed miserably in
another. As a consequence, EPRI initiated a number of projects to identify
materials degradation problems in FGD systems and to develop solutions to
those problems.
MATERIALS RELATED RESEARCH
Table 1 lists by research project (RP) number the past and present EPRI
contracts related to corrosion and degradation of materials in FGD
environments. Also listed are the relevant periods of performance, project
titles, contractor's names, and resulting EPRI reports. Completion dates are
missing for those projects still in progress.
In this paper, the objectives of each of the projects listed in Table 1
are summarized and a few of the more interesting results are discussed. Also
included is the history of the development of simulated scrubber environments
for laboratory testing purposes.
OBJECTIVES OF RESEARCH PROJECTS
RP982-14: Construction Materials for Wet Scrubbers
As mentioned above, one of the objectives of this project was to collect
and summarize data on materials of construction used in full-scale utility
lime and limestone FGD systems. This was achieved by visiting all existing
lime and limestone sites, most of the FGD system vendors, and several coating
suppliers. The data collected were quite difficult to interpret: materials
that had enjoyed success in one scrubber had failed in another and, because
maintenance records were sparse or absent, the cause of the failures could
seldom be determined. One of the most controversial subjects was the cause of
coating failures. The suppliers typically blamed faulty application (too
humid, rust bloom, unsuitable surface profile, too cold etc.) whereas, the
users were left wondering whether the coating itself had had inadequate
chemical resistance to the harsh environments inside the FGD system.
3-2
-------
TABLE 1. EPRI RESEARCH PROJECTS ON CORROSION AND
DEGRADATION OF MATERIALS IN FGD ENVIRONMENTS
Project Period of
No. Performance
RP982-14 10/78-2/81
RP982-19 2/80-11/80
RP1871-1 6/80-6/81
RP1871-2 3/81-
RP1871-3 5/81-
RP1871-4 10/81-12/82
RP1652-2 11/81-
RP2248-2 10/82-
RP1871-5 10/82-
RP1871-6 11/82-
RP2248-3 12/82-
RP1871-7 2/83-
RP1871-11
RP1871-10
Project Title and Contractor
Construction Materials for Wet
Scrubbers. (Battelle Columbus Labs.)
Screening Study for Corrosion
Inhibitors in S02 Removal Systems
(SumX Corporation)
Corrosion-Resistant Alloys for Flue
Gas Desulfurization Systems (Battelle
Columbus Labs.)
Corrosion/Erosion Laboratory and Field
Testing (Battelle Columbus Labs.)
EPRI
Reports
CS1736
(Vols. 1 & 2)
CS2533
CS2537
(Report
pending)
Construction Material for Wet Scrubbers (Report
Update (Battelle Columbus Labs.) pending)
Corrosion Field Testing Support Effort
at R. D. Morrow (Burns and McDonnell)
Cyclic Reheat Materials Evaluation
(Stearns Roger/Battelle Columbus Labs.)
Materials Failure Causes in FGD
Systems I (Radian Corp./Stearns Roger)
Mechanisms of Failure of Coating Used
in FGD Systems (Lehigh University)
Corrosion Chemistry of SO2 Scrubbers
(Battelle Columbus Labs.)
Materials Failure Causes in FGD
Systems II (Battelle Columbus Labs./
Black and Veatch)
Effect of SO2 Scrubber Chemistry on
Corrosion (Rockwell International)
Characterization of Ceramic Materials
for Electric Power Plant Applications
(Ohio State University)
Corrosion Inhibitors for FGD Systems
(Contractor not yet selected)
CS3240
3-3
-------
RP1871-3: Construction Materials for Wet Scrubbers Update
As the project title suggests, this is an \ipdate of the RP982-14
survey. Now almost complete, the project included visits to all the lime and
limestone wet scrubbers visited in the course of RP982-14 and a reevaluation
of the materials of construction data. This information was supplemented with
data on materials used in the construction of all new lime and limestone
scrubbers not included in the initial survey and with data on materials used
in other types of FGD systems, including spray dryers, dual alkali systems,
Wellman-Lord systems, and non-recovery Na2CO-j wet scrubbers. It is
anticipated that the results of this work will be published in an EPRI report
before the end of 1983.
RP1871-1: Corrosion-Resistant Alloys For FGD Systems
This project was one of the first aimed at improving our understanding of
the nature of corrosion and degradation of materials in SC>2 scrubbers. A
literature review was performed in which field data were combined with releant
laboratory test data to create a state of the art report on corrosion of
alloys and linings in FGD systems (2). Recognizing that outlet ducts were
particularly prone to failure, it was decided that candidate alloys and
linings would be evaluated in the laboratory in simulated duct environments to
determine their resistance to the various forms of corrosion and
degradation. Two simulated duct environments were employed: both were acid
mists generated in a fog cabinet at 48°C by atomizing solutions containing
sulfurous acid and calcium chloride. In one case, the pH was adjusted to 1.0
with hydrochloric acid while, in the other case, the pH was adjusted to the
same value with sulfuric acid (Table 2). Thus, the first mist contained only
TABLE 2. SIMULATED DUCT ENVIRONMENTS AT TWO CHLORIDE LEVELS
200 g/mj C1"
17 ml/1 of 7.4% H2SO3
0.3 g/1 CaC12
3.7 ml/1 of 95% H2SO4
5000 g/m3 C1~
17 ml/1 of 7.4% H2S03
0.3 g/1 CaC12
11 ml/1 of 37% HC1
200 g/m chloride (from the calcium chloride) whereas the second mist
contained 5000 g/m chloride (from both the calcium chloride and the
hydrochloric acid). To simulate exposure to the hot dry flue gas under bypass
conditions, some specimens were taken from the fog cabinet 5 times per week
and placed in a dry oven at 120°C for 5 hours before being returned to the fog
cabinet. The oxygen concentration in both.the dry oven and the fog cabinet
was maintained at 8% by volume, a value typically found in scrubber ducts. At
Concentrations quoted in g/m are numerically equal to concentrations in
parts per million by weight (ppm).
3-4
-------
the time this work was performed (1980-81), the test environment was
considered to simulate closely the environment actually found in a scrubber
duct: proper consideration had been given to the form of the environment (a
calcium-based condensate), the temperature (cycling between 48°C and 120°C),
the pH (~1), and the chloride level (200-5000 g/m3).
The results of this work (_3, _4_) contained no surprises: in the
low-chloride environment, all alloys tested were resistant to uniform
corrosion and localized attack whereas, at the higher chloride level, several
of the stainless steels tested suffered a considerable amount of pitting,
crevice corrosion, uniform corrosion and, in some instances, intergranular
attack. The nickel alloys and titanium tested, on the other hand, appeared to
be resistant to these forms of corrosion. Most of the linings tested were
chemically resistant to both the low- and high-chloride acidic mists but none
could withstand the temperature cycling between 48°C and 120°C: the coatings
either deteriorated at the higher temperature or delaminated from the C-steel
substrate.
RP1871-2; Corrosion/Erosion Laboratory and Field Testing.
This project has been in progress for over 2 years and consists of both
laboratory and field testing of a wide variety of materials of construction,
including alloys and linings. The primary objective of this project is to
obtain sufficient field and laboratory test data so that our understanding of
the mechanisms of corrosion and degradation is greatly improved. It is hoped
that, with this improved understanding, cost-effective materials of
construction can be selected with much greater confidence than is currently
possible.
Field tests are currently being performed at three stations burning
medium-, low-, and high-sulfur coals: these are, respectively, the R. D.
Morrow, Sr. Station of the South Mississippi Electric Power Association, the
Clay Boswell Station of Minnesota Power Company, and Duck Creek Station of
Central Illinois Light Company. Although the field test environments cannot
be controlled with the precision possible in a laboratory test, a substantial
effort is going into their characterization (see later, RP1871-4); and one of
the tasks in RP1871-2 has been to install special condensate collectors in the
outlet duct of two of the host scrubbers so that samples of the corrosive
condensates can be collected for pH measurement and chemical analysis. It is
hoped that the field test conditions will-be sufficiently well characterized
that results can be correlated with the results of the concurrent laboratory
tests. However, such correlations are only possible if the corrosion
processes are well understood and all relevant variables have been given
proper consideration.
By the time the earlier project, RP1871-1, was drawing to a close, there
was a growing concern that the aggressiveness of scrubber environments was not
simply controlled by pH, temperature, and chloride content. It was recognized
that scrubber environments were chemically quite complex and that some of the
species present in trace or minor amounts could play a critical role in the
corrosion and degradation of materials.
3-5
-------
Thus/ when the RP1871-2 laboratory studies were initiated an effort was
made to more closely simulate the scrubber environments of interest, namely
those found in prescrubbers, absorbers, outlet ducts, and reheaters (see
Table 3).
TABLE 3. RANGE OF ENVIRONMENTAL CONDITIONS OCCURRING
IN PRESCRUBBERS, ABSORBERS, OULET DUCTS, AND REHEATERS
Component
Prescrubber
Absorber (after prescrubber)
Absorber (no prescrubber)
Outlet duct
Reheater tubes
pH range
1-4
4.5-7.5
4.5-7.5
1-4
1-4
_ *
C1 range
kg/m3
5-50
1-5
1-20
1-20
1-20
Temp.
°P i
120-300
120-130
120-300
120-130
250-400
range
(°C)
(50-150)
(50-55)
(50-150)
(50-55)
(120-205)
1 kg/m = 1000 parts per million by weight. Chloride levels up to 100 kg/m"
predicted for future closed loop scrubber systems.
Calcium-based solutions have been used throughout this program and the
chloride levels, pH values, and temperatures have been chosen to fall within
the ranges identified in Table 3. The major difference between these
environments and those used in RP1871-1 is that three other species known to
be present in scrubber liquids (fluoride, sodium, and magnesium ions) have
been added to each simulated scrubber environment. After examining the
available chemical analyses of absorber liquids, prescrubber liquids, and
typical coals, the decision was made (a) to add chloride partly as CaCl, and
partly as MgCl2 so that a Ca:Mg ratio of 2:1 is achieved (b) to add enough
fluoride to give a C1:F ratio of 10:1, achieved primarily by adding CaF2/ but
also by adding sufficient NaF to give a background Na content of 500 g/m . It
should be noted that, quite often, not all of the CaF2 addition dissolves in
the test solution so that the C1:F ratio in solution can be higher than 10:1
(_£). The pH of each solution is adjusted to the desired value by adding
sulfuric acid and, as in the RP1871-1 tests, the solutions are maintained in
equilibrium with a nitrogen/8% oxygen gas mixture.
The fog cabinet, used previously for the simulated duct tests, is also
being used at the same temperature (48°C) for the simulated reheater tests.
However, the cabinet internals have been modified to allow the specimens to be
locally heated to a typical reheater temperature (121°C). Prescrubber and
absorber conditions are simulated in large tanks filled with the test
solutions described above; specimens are immersed in the liquid and also
suspended in the vapor phase above the liquid.
Some of the preliminary results of RP1871-2 were quite surprising (see
Table-4).
3-6
-------
For instance, in one series of tests in which prescrubber conditions were
being simulated, the corrosion rate of all alloys increased steadily as the
chloride level of a pH 1 solution was increased to 30 kg/m (fluoride addition
increased correspondingly to 3 kg/m ); however, at higher halide
concentrations, the corrosion rate decreased sharply (Table 4). For instance,
Hastelloy Alloy C-276 had an average corrosion rate of 104 ym/y in the
TABLE 4. AVERAGE CORROSION RATES (pm/y) DURING
A SIX-WEEK EXPOSURE TO SIMULATED PRESCRUBBER
ENVIRONMENTS AT 93°C.
Alloy
Hastelloy C-276
Inconel 625
317 LM stainless
31 6L stainless
Ferralium 255
Titanium
TiCode 12
30 kg/m chloride
pH 1
104
529
3205
4206
685
»25,000*
1409
100 kg/m3
pH 1
1.4
19.8
27.1
138
18.3
8.4
11.3
chloride
pH 4
0.8
0.4
5.1
3.6
1.2
0
2.0
Specimen completely dissolved in 1 week.
solution containing 30 kg/m chloride, whereas in the 100 kg/m chloride
solution, the corrosion rate dropped to 1.4 pm/y. Even more surprizing was
the observation that samples of titanium, a very corrosion resistant alloy in
a wide variety of aggressive environments, disintegrated within a week in the
30 kg/m chloride solution. Such high corrosion rates were not seen when
halide concentrations were somewhat higher (Table 4) or lower; and other work
(6_) demonstrated that corrosion of titanium was much reduced in the absence of
fluoride. Although these experiments demonstrate that fluoride can greatly
accelerate corrosion in these laboratory test environments, their importance
in scrubber environments has been challenged by some workers (_6) who found
that titanium is not susceptible to accelerated attack in similar low-pH
halide solutions if flyash is also present. These workers suggest that some
water-soluble ingredient in the flyash (perhaps iron or silica) is responsible
for inhibiting corrosion. This observation is of great practical significance
because flyash is usually carried into the scrubber with the flue gas.
Results like these prompted EPRI to initiate two additional projects to
investigate much more thoroughly the effects and interactions of many of the
other trace elements in scrubbers (see later, RP1871-6 and RP1871-7).
3-7
-------
RP1871-4; Corrosion Field Testing Support Effort at R. D. Morrow
This project supports, and is closely coordinated with the field tests of
RP1871-2 at the R. D. Morrow, Sr. Station of South Mississippi Electric Power
Association. This plant has two 230 MW coal-fired boilers each equipped with
a wet limestone scrubber representing the state of the art in closed loop,
limestone FGD design. The objectives of this project are (a) to document the
environment to which the RP1871-2 alloys and coatings are exposed and (b) to
assist in identifying operating conditions and chemical species that lead to
particularly aggressive environmental conditions. Samples of the coal,
make-up water, absorber recycle slurry, demister wash water, flyash and
limestone are being analyzed for trace metals, chlorides, fluorides, bromides,
etc. The aggressiveness of the flue gas itself is being assessed by taking
controlled condensation samples for chemical analysis and pH measurement.
RP1871-6; Corrosion Chemistry of
Scrubbers.
As indicated above, this project is one of two designed to investigate
more thoroughly the effects of trace and minor environmental constituents on
the corrosion of alloys. Work has been underway for only a few months, but a
review of the literature and all other currently available data has been
completed. It was concluded that all the species listed in Table 5 may be
present in FGD system environments and that their presence could affect the
corrosion rates of nickel-, iron-, or titanium-based alloys.
TABLE 5. MAXIMUM CONCENTRATION OF ELEMENTS
USED IN RP1871-6 LABORATORY TEST SOLUTIONS
Species
Mg2+
Ca2+
A13+
Si4+
Cu2+
Cr3"1"
Fe3+
Na+
Max. Concn.
g/m3
5000
1000
2000
100
1000
500
2000
*
See note
Max.
Species
NO3~
Mo042~
(i.
4
(i.
Cl~ 100
F~ 10
Br~
T"
Concn.
g/m3
200
333
e. 200 Mo)
1533
e. 500 P)
,000
,000
500
100
Na to be added as necessary for ionic charge balance.
Table 5 also lists the maximum concentrations considered likely in FGD system
environments. These species will be investigated in the RP1871-6 laboratory
test program at levels up to the quoted maxima.
3-8
-------
Table 5 represents the distillation of information from several
sources. First the chemical analyses of coals, limes, limestones, bottom ash,
and flyash were reviewed. Second, the concentration of trace and minor
elements in the flue gases at the inlets and outlets of two scrubbers were
evaluated along with the chemical analyses of the corresponding liquid
streams. Third, the concentration of elements in the condensate at the wall
of the outlet ducts of two stations (collected and analyzed in RP1871-2) was
assessed. Finally, available literature was reviewed to determine which of
the many species present could conceivably have an effect on alloy
corrosion. In addition to the dissolved solids listed in Table 5, the test
solutions will contain dissolved gases, introduced by sparging the solutions
with an argon carrier gas containing one or more of the following gases: 09
(8% max.), S02 (0.3% max.), and f^S (5ppm by volume max.).
The great variety of species considered in the test matrix represents a
formidable task, but with the help of a carefully planned statistical
experimental approach, it is hoped that the effects of these species, both
direct and synergistic, will be identified. Hopefully, such information will
allow correlations between laboratory and field test data to be made with much
greater confidence than is currently possible.
RP1871-7; Effect of SO2 Scrubber Chemistry on Corrosion
The objectives of this project are almost identical to those of
RP1871-6. Indeed, the environments evaluated will be very similar, and they
will also be based on the list of species shown in Table 5. However, the
unique feature of this project is that primary emphasis will be on vapor phase
corrosion whereas bulk liquid phase corrosion is emphasized in RP1871-6.
There is evidence that the corrosion that occurs in thin condensate films is
more severe than would occur in a bulk liquid of the same composition. Like
RP1871-6, this project has been underway for only a few months.
RP1871-5; Mechanisms of Failure of Coatings Used in FGD Systems
In an attempt to assess the relative importance of chemical resistance
and application technique on the failure of coatings in FGD systems, RP1871-5
was initiated a little less than a year ago. The mechanisms of coating
failure are being investigated in the laboratory in simulated scrubber
environments. About a dozen types of coating will be evaluated ranging from
vinyl esters to fluorocarbons and epoxies.
Perhaps the most critical variable being studied is surface preparation
technique. Early results indicate that the adhesive strength of the coating
is strongly dependent on the preparation technique used. For instance, the
adherence of one fluoroelastomer to a C-steel substrate was initally much
higher after chemically etching in a HF/HNO-j solution than after abrading with
600-grit silicon carbide paper which, in turn, provided slightly better
adherence than after a commercial alkaline washing treatment. However, after
about 5 days in an acidic environment, the ranking of the surface preparation
techniques was competely reversed, with the acid etch giving the poorest
adherence and the alkaline wash the best.
3-S
-------
It is hoped that this research will provide the electric utilities with
the information needed to properly specify the coating application methods.
RP1652-2; Cyclic Reheat Materials Evaluation
When the scrubbed flue gas enters the outlet duct, it is at the adiabatic
saturation temperature and, in the absence of any reheat, it condenses on the
walls of the duct and the stack. This condensate can be very corrosive and it
has caused numerous corrosion failures in the past (see RP1871-1 and
RP1871-3). Furthermore, the wet gas is not very buoyant and produces a
visible plume. To circumvent these problems, some type of flue gas reheat
system is often employed. Most systems are expensive to operate in that they
require some external energy source to provide the heat. However, one type of
reheat system, called cyclic reheat, derives the necessary heat from the hot
flue gas upstream of the scrubber, so operating.costs are potentially low.
The cyclic reheat system most developed in the USA makes use of two in-line
tube-type heat exchangers: one in the ductwork upstream of the scrubber and
one in the outlet duct. Water or some other appropriate working fluid is
circulated from one heat exhanger to the other so that heat, extracted from
the hot (~150°C) flue gas entering the scrubber, is released to the cooler
(~55°C) wet flue gas leaving the scrubber. While the concept of using "free"
heat is appealing, its success depends to a large extent on the resistance of
the upstream heat exchanger to corrosion and fouling. The tube wall
temperature of this heat exchanger must be below that of the surrounding
unscrubbed flue gas if heat is to be extracted; and, if this tube wall
temperature is below the acid dew point, the resulting condensate may
stimulate corrosion and encourage fouling.
EPRI initiated RP1652-2 to evaluate tube materials for the inlet gas heat
extractor of a cyclic reheat system. In this work, a model heat exchanger has
been designed, fabricated, and installed in a test loop at the Scholz Electric
Generating Plant of Gulf Power Company. Over 20 materials are being
evaluated, including some coatings. Test variables of interest are tube wall
temperature and the type and frequency of tube cleaning. Many parameters are
being maintained and recorded as a function of time, including heat transfer
rate, acid dew point, temperature, and, on some specially instrumented tubes,
the tube wall thickness. The rate at which the last parameter changes with
time provides a measure.of the corrosion rate. Results of this project are
expected to be available in the latter half of 1984.
RP982-19; Screening Study for Corrosion Inhibitors in SO2 Removal Systems.
It is well established that corrosion of metals can often be inhibited by
the addition of quite small amounts of a chemical species that would not
naturally be present. The feasibility of using inhibitors to control
corrosion in SC>2 scrubbers was addressed in this small laboratory study. The
results indicated that none of the 26 compounds evaluated (see Table 6)
inhibited corrosion of C-steel well enough to consider its use in scrubbers.
However, one compound (N-lauroylsarcosine) and several of its analogs were
found to inhibit general corrosion of Type 304 and Type 316 stainless
steels. More important, these compounds greatly increased the resistance of
the stainless steels to pitting attack. Since pitting and other forms of
3-10
-------
localized corrosion are responsible for the failure of many scrubber
components/ the identification of this pitting inhibitor was considered most
significant.
TABLE 6. COMPOUNDS TESTED AS POTENTIAL CORROSION INHIBITORS IN RP982-19
3,3',4,4'-Benzophenone tetracarboxylic dianhydride
benzotriazole
S-(t-butyl)-thioglycolic acid
a-cyclohexylaminobutyric acid
N-cyclohexyl-2-pyrrolidone
N-cocoaIky1-2-pyrrolidone
1-ethylquinolinium iodide
gelatine
potassium thiocyanate
linear (C-12) alkyl sulfonate
S-lauryl thioglycolic acid
sodium benzoate
•sodium diethyldithiocarbamate
sodium dimercaptothiadiazole
N-lauroylalanine
N-lauroylglycine
N-lauroylsarcosine (sodium salt)
N-octanoylsarcosine
N-stearolysarcosine
octylamine
di-triethanolammonium dimercaptothiadiazole
N-formylpiperidine
N,N,N',N',2-pentamethy1-1,2-propanediamine
phenylthiourea
sarcosine
tetraethylammonium toluenesulfonate
EPRI plans to fund additional work to determine the feasibility of using
corrosion inhibitors in SO2 scrubbers. This work will confirm and extend the
results obtained in RP982-19
RP1871-10; Corrosion Inhibitors for FGD Systems
This project is due to commence early in 1984 and, as mentioned above,
its objective is to evaluate further, the feasibility of using corrosion
inhibitors in SO2 scrubbers. Candidate inhibitors will also be assessed for
their effect on the scrubber process (SO2 removal).
3-11
-------
RP1871-11; Characterization of Ceramic Materials for Electric Power Plant
Applications
Ceramic materials have been used quite often in FGD systems as liner
materials on C-steel support structures. They are particularly useful in
areas where erosion is a potential problem, but some ceramics (such as acid
resistant bricks) have been used to line stacks and other scrubber components
where corrosion, not erosion, has been of prime concern. There are
disadvantages of using conventional ceramic liner materials. For instance,
they are normally quite bulky and, as a result, cannot be retrofitted in some
narrow section locations, such as venturi throats; and, rapid changes in
temperature can generate sufficiently high stresess in ceramics that they will
crack and spall.
Some of these disadvantages can be avoided by using relatively thin
prestressed ceramic tiles. Here, the tiles are carefully designed and
assembled to support an initial compressive stress which counterbalances the
stresses induced in service by rapid temperature changes. This concept has so
far seen only very limited application: prestressed SiC tiles have been
successfully retrofitted in one venturi prescrubber that had experienced
serious erosion problems, but more widespread application is limited by our
lack of knowledge of the physical and mechanical properties of the ceramic
materials at the temperatures of interest. The objective of RP1871-11 is to
generate such information for selected ceramic materials so that prestressed
ceramic components can be incorporated into critical areas of FGD systems as
needed. This project is due to start within a few months.
RP2248-2 and RP2248-3; Materials Failure Causes In FGD Systems
There is no doubt that materials used in FGD systems have failed to
perform their intended function on more than a few occasions. There is ample
documented evidence (see RP982-14 and RP1871-3, for instance) that metals and
nometallic materials can corrode, degrade, crack, and erode, and that coatings
can disbond or degenerate. However, it is also apparent that much of the
information available is inconsistent: a material that has performed well in
one scrubber is reported to fail miserably in another scrubber operating under
supposedly identical conditions. Furthermore, while the type of failure has
often been documented (stress corrosion cracking, pitting, disbonding, etc.),
there has seldom been any attempt to determine the cause of failure. For
instance, Type 316 stainless steel may have given excellent service for
several years in a scrubber environment containing 500 g/m chloride, but
because of an unusual upset condition (e.g. the temperature jumps from the
normal 50°C to 100°C), the stainless steel suffers extensive pitting. The
root cause of this failure is the upset condition but, unless such information
is reported, a subsequent review of the available information may suggest that
Type 316 stainless steel cannot be used safely in scrubber environments
containing 500 g/m chloride.
Consequently, EPRI initiated two projects recently (RP2248-2 and
RP2248-3) which are designed to investigate and report the true causes of the
failures of metals, coatings, and linings in FGD systems. It is hoped that
3-12
-------
these investigations will help avoid the apparent conflicts and
inconsistencies in the literature that have developed in the past.
SUMMARY
Over the last four years, EPRI has initiated many research projects to
identify materials degradation problems in FGD systems and to develop
solutions to those problems. During this period, our understanding of the
mechanisms of degradation has improved and, with this understanding has come
the ability to simulate scrubber conditions more accurately in the
laboratory. However, the scrubber industry is still young and its problems
plentiful. Materials scientists and corrosion engineers will play an
important role in the future development and success of FGD systems.
ACKNOWLEDGMENTS
The author wishes to thank S. Dalton, C. Dene, R. Rhudy, and D. Stewart
of EPRI for reviewing this paper and for making many useful suggestions.
REFERENCES
1. H. S. Rosenburg, H. H. Krause, L. J. Nowacki, C. W. Kistler, J. A.
Beavers, R. B. Engdahl, R. J. Dick, and J. H. Oxley, "Construction
Materials for Wet Scrubbers (Volumes 1 & 2)," EPRI Report No. CS-1736,
Electric Power Research Institute (March 1981).
2. G. H. Koch, J. A. Beavers, N. G. Thompson and W. E. Berry, "Literature
Review of FDG Construction Materials," EPRI Report No. CS-2533, Electric
Power Research Institute (Aug. 1982).
3. G. H. Koch and J. A. Beavers, "Materials Testing in Simulated Flue Gas
Desulfurization Duct Environments, "EPRI Report No. CS-2537, Electric
Power Research Institute (Aug. 1982).
4. G. H. Koch, J. A. Beavers, and B. C. Syrett, "Experimental Evaluation of
Alloys and Linings of Simulated Duct Environments for a Lime/Limestone
Scrubber," Paper No. 197, NACE Annual Conference, Houston, Texas; National
Assoc. of Corrosion Engineers, Houston, Texas (1982).
5. G. H. Koch and J. A. Beavers, "Laboratory and Field Evaluation of
Materials for Flue Gas Desulfurization Systems," Paper No. 2D, EPA and
EPRI Symp. on Flue Gas Desulfurization, Hollywood, Florida (May 1982).
6. D. E. Thomas and H. B. Bomberger, "Effect of Chlorides and Fluorides on
Titanium Alloys in Simulated Scrubber Environments," Paper No. 189, NACE
Annual Conference, Anaheim, California, U.S.A.; National Assoc. of
Corrosion Engineers, Houston, Texas (1983).
3-13
-------
SIMULTANEOUS DESIGN, PLANNING, AND MATERIALS
OF CONSTRUCTION SELECTION FOR FGD SYSTEMS
A. Kirschner, N. Ostroff, R. F. Miller,
W. L. Silence
-------
SIMULTANEOUS DESIGN, PLANNING, AND MATERIALS
OF CONSTRUCTION SELECTION FOR FGD SYSTEMS
A. Kirschner and Dr. N. Ostroff
Peabody Process Systems, Inc.
Stamford, CT
R. F. Miller and W. L. Silence
Cabot Wrought Products Division
Cabot Corporation
Kokomo, IN
ABSTRACT
Processes and systems designed for emissions control from
fossil fuel fired boilers for electric generating stations
are many and varied as are the choices of materials of
construction. Good performance and efficiency of operation
and maintenance of flue gas desulfurization systems can be
enhanced by the simultaneous selection of the process, the
design, and the materials of construction of the system. The
wet lime and limestone systems are the dominant processes for
removal of sulfur dioxide from coal fired boiler flue gases
and will remain so for the next several years, based on EPA
surveys. The choices of material include carbon steel,
coatings and linings that can be applied over carbon steel, a
wide range of alloys that can be used in place of lined
carbon steel, and various nonmetallic materials. It is the
purpose of this paper to discuss the corrosion and abrasion
resistance of these materials and to show how each may be
most advantageously used in the design of an FGD system.
INTRODUCTION
The composition of flue gases produced during the
combustion of coal vary widely. The major polluting
impurities in coal are sulfur, chlorine, and non-combustible
inorganic matter which appear in flue gas as sulfur dioxide,
hydrogen chloride, and flyash respectively. Federal and
state regulatory agencies have enacted stringent legislation
requiring utilities to remove the major portion of flyash and
the oxides of sulfur from flue gas before it can be returned
to the atmosphere. The use of large scale desulfurizing
equipment became important in the early 1970's.
3-15
-------
The early operating problems were solved and FGD
technology gradually improved due in part to:
- fundamental research done at Research Triangle Park,
Battelle Institute, the University of Texas at Austin, and
the University of North Dakota,
- process development work done at the Shawnee Test Facility,
and at various pilot plants,
- engineering during the design of new FGD systems,
- tests performed at various full scale installations.
Good overall system performance has been achieved by
properly selecting the process, the design, and the materials
of construction. A utility must often invest many millions
of dollars to design, install and operate an FGD system, so
economic optimization of both process parameters and
materials of construction are very significant.
The primary chemical pollutants produced during coal
combustion are sulfur dioxide and sulfur trioxide; one or two
percent of the total coal sulfur is usually converted to
sulfur trioxide. The required degree of sulfur dioxide
removal is determined by NSPS regulations, summarized by
Stevenson(1). At the present time, the removal of sulfur
dioxide is most commonly accomplished by a wet process in a
multilevel spray tower, using lime or limestone as the
alkali.
PHILOSOPHY OF SYSTEM DESIGN
An FGD system is a chemical plant, and, as such, the .
materials of construction must be carefully selected to
provide the maximum resistance to the system environment at
the minimum cost. Site specific considerations such as the
generating capacity of the power plant, and its geographic
location, determine the type and quantity of coal burned, and
therefore the amount of gas to be treated and the amount of
sulfur to be removed. These parameters determine the
physical size of the process equipment. Other regulations
pertaining to the nature of the final disposal product, and
the requirement for closed loop water balance, in conjunction
with the chlorine content of the coal determine the
concentration of dissolved solids in the scrubbing loop.
Laslo and Chang(2,3) have discussed the effects of
dissolved c'hlorides on FGD chemistry.
Sulfur trioxide condenses as very small droplets at the
temperatures of the system inlet ductwork. These extremely
fine sulfur trioxide droplets cannot be efficiently trapped
in spray towers; they pass through the absorbers in the gas
stream, and are deposited on the walls of the downstream
equipment and ductwork. Sulfur trioxide particles which do
3-16
-------
not impinge on any surfaces leave the system with the stack
gas. Sulfur trioxide condenses in the form of 40 - 60
percent sulfuric acid solutions at about 100 to 120 F to
create extremely corrosive media. Pierce^) has correlated
much of the available data on sulfuric acid condensation (dew
point). The results of his work are shown graphically in
Figure 1.
Chlorine is a second chemical that causes corrosion
problems in FGD systems. It is present in almost every coal;
its concentration may be negligible or as high as 0.5 weight
percent. Many FGD systems are designed based upon coals
containing approximately 0.1 percent of chlorine. In colder
climates, coal piles are often sprayed with dilute calcium
chloride solutions to prevent freezing. Chlorine introduced
in this fashion is identical to that introduced by naturally
occurring chlorine. During and immediately after combustion,
the contained chlorine is converted to hydrogen chloride gas
which is removed in the wet scrubbing system. All chlorine
compounds of interest are highly water soluble, and all of
the absorbed chlorine must leave in the free-water fraction
of the final product.- Since most regulatory agencies require
FGD systems to operate under closed loop water balance
conditions, and many systems operate with forced oxidation to
improve the disposal properties of the sludge, high
concentrations of chlorides can occur in the circulating
liquids .
The final process material that may create operating
problems is the scrubbing slurry itself. In lime or
limestone systems the slurry is a mixture of calcium s:ulfite,
calcium sulfate, fresh alkali and flyash. In at least one
successful operation, flyash is removed from the flue gas in
a high efficiency venturi scrubber and the resulting flyash
slurry is used to remove sulfur dioxide in a spray
tower' ). The major material problem associated with the
handling and pumping of these slurries is erosion in such
system components as spray nozzles, pumps, agitators, and the
absorption zone. In fact, the slurry exiting the spray
nozzles may be likened to a wet sandblast.
STATEMENT OF THE PROBLEM
It is the object of this paper to review the FGD process
from the standpoint of materials of construction and to
demonstrate how the proper material selection enhances FGD
plant operation and economics. The major material problems
that must be addressed are corrosion, erosion, thermal
effects, and chemical imbalances such as plugging and scaling
which can occur at various points in the system. A summary
of how these problems affect the design and operation of
3-17
-------
0.01
Humidity (Ibs. HjO/lbs. dry gas)
0.02
0.03
0.04
0.05
0.06
0.07
0.08
300
OJ
0.
e
a
E-
250
200
10
12
Volume % H-0
Figure 1: Sulfuric acid dew point in flue gas.
3-18
-------
various system components is given in Table 1. A schematic
flov; diagram highlighting these problem areas is shown in
Figure 2.
In choosing materials of construction for an FGD system,
one must consider the corrosive and erosive nature of the
process stream and the temperatures and temperature gradients
to which the system components will be exposed. Chemical
imbalances, although important in the overall system design,
will only receive minor attention here.
MATERIALS OF CONSTRUCTION
The choices of materials of construction for system
components include base metal, such as carbon steel, carbon
steel coated with various organic and inorganic coatings, and
alloys. These are summarized in Table 2. Alloy compositions
are listed in Table 3.
The following sections describe these materials.
CARBON STEEL
The most extensively used material in FGD systems is
carbon steel. It is used in hot, dry atmospheres, and in
alkali environments where it needs no protection against
corrosion. Carbon steel corrodes rapidly in wet or acidic
environments.
LINED CARBON STEEL
Lined, or coated, carbon steel economically overcomes
many of the shortcomings of carbon steel.
Organic Linings
Flakeglass
Flakeglass filled vinylester and polyester resins are
used in wet sections of FGD systems, such as the absorber
vessel, the outlet ductwork, and various tanks. They provide
economical corrosion protection with moderate erosion
resistance. These liners cannot withstand high temperatures,
and a reasonable amount of temperature control is necessary.
These materials are usually specified because of their
generally high reliability and low initial cost.
Rubber
Rubber linings are used where abrasion is the primary
problem. They are most frequently used in scrubber
3-19
-------
TABLE 1
SERVICE CLASSIFICATION OF
FGD SYSTEM COMPONENTS
THERMAL CHEMICAL
CORROSION EROSION EFFECTS IMBALANCES
1 . Inlet Ductwork M N M N
2. Absorber Inlet S M M S
3. Absorber Body S S M S
4. Mist Eliminator S M M S
5. Absorber Outlet S M M M
6. Outlet Ductwork S M M S
7. Reheaters & Assoc. S M S S
Ductwork
8. Stack Breeching S M S S
9. Alkali Tanks N M N N
10. Recycle Tanks S M M N
11. Slurry Piping S S M M
12. Slurry Nozzles M S M M
13. Dampers S M S S
S =Severe problem; M =Minor problem; N =not usually a problem
3-20
-------
GAS
FLOW
Figure 2: Schematic Flow Diagram of a Wet FGD System
Numbers in circles, refer to Table 1.
3-21
-------
TABLE 2
MATERIALS OF CONSTRUCTION
I CARBON STEEL 1.0*
II LINED CARBON STEEL
A. ORGANICS
a. Flakeglass 2.0
b. Vinylester 2.0
c. Natural Rubber 2.5
d. Neoprene 2.8
e. Chlorobutyl Rubber 2.6
f. Fluoroelastomer 5.5
B. INORGANICS
a. Monolithics 2.5-3.5
b. Acid Brick 3.1
c. Borosilicate Block 3.0
III ALLOYS
A. LOW ALLOYS
a. 316L 3.0
b. 317L 4.0
c. 317LM 4.5
d. 255 4.0
B. MEDIUM ALLOYS
a. 20Cb-3 3.8
b. 825 5.5
c. 904L 5.0
d. JS-700 5.0
C. HIGH ALLOYS
a. G 5.5
b. G3 5.5
c. 625 8.5
d. C-276 9.0
See Table 3 for alloy chemical compositions
Approximate Relative Installed Costs
3-22
-------
TABLE 3
NOMINAL CHEMICAL COMPOSITIONS
N3
U>
Carbon Steel
LOW ALLOYS
316L
317L
317LM1
FERRALIUM alloy 255
MEDIUM ALLOYS
Alloy 20CB-3
UHB 901L2
JS-7003
INCOLOY alloy 825
HIGH ALLOYS
HASTELLOY alloy G
HASTELLOY alloy G-3
INCONEL alloy 6254
HASTELLOY alloy C-276
Fe
Bal
Bal
Bal
Bal
Bal
Bal
Bal
30
20
20
55
5
Cr
17
19
19
25.5
20
20
21
22
22
22
21 .5
16
Mo
2.5
3.5
1.35
3.5
2.5
1.5'
1.5
3.0
6.5
9
16
Nl Cu
Other
Approx.
Relative Cost6
12
13
11
5
35
25
25
Bal
Bal
Bal
Bal
Bal
0.035
0.035
0.035
1.7 0.01
3.5 0.075
1.5 0.02
0.05
2 0.03
2 0.055
1.5 0.007
0.10
0.005
N-0.17
Cb-0.3
W-1 .O5
Cb+Ta-2.0
Co-5.o5,
W-1 .55,
Cb+Ta-0.55
Cb+Ta-3.5
V-0.355,
0.1
1 .0
1-3
1.1
1 .1
2.8
2. 1
2. 1
2.9
3.9
3.9
5.9
6.6
Eastern Stainless Steel Corporation
Uddeholra AB
Jessop Steel
Also CABOT alloy No. 625
Maximum
Equal weight basis
-------
absorption zones, slurry piping, and slurry pumps. During
construction, care must be taken when the steel substrate is
prepared and the rubber lining is applied. The steel surface
must be freshly cleaned and sandblasted.
Natural rubber is lowest in cost, easiest to apply, and
has the highest degree of abrasion resistance. The drawbacks
to using natural rubber are its higher flammability, and its
lower resistance to oil. Oil, used in boiler startup and in
flame stabilization, has not proven to be harmful, but if oil
alone is used for fuel for extended period of time it may not
be'completely burned and may enter an FGD system via the flue
gas and soften the rubber.
Neoprene rubber has improved fire and oil resistance but
is more expensive to purchase and more difficult to apply.
Neoprene has higher permeability and lower resistance to
abrasion than does natural rubber. The applied cost of
neoprene linings (per square foot) is approximately 20
percent higher than natural rubber.
Chlorobutyl linings are becoming increasingly popular in
new construction and in replacements. Chlorobutyl rubber has
lower permeability than does natural rubber. Its abrasion
resistance is somewhat lower than natural rubber; its oil
resistance is higher than that of natural rubber and lower
than that of neoprene. The applied cost of Chlorobutyl
linings (per square foot) is approximately 10 percent higher
than natural rubber.
Rubber lined carbon steel is the most commonly used
material of construction where abrasion is a problem.
Fluoroelastomer
Fluoroelastomer linings are not usually used in
absorbers. They have been successfully used in outlet
ductwork where temperatures may be high due to reheat
requirements. Fluoroelastomers are usually applied by
spraying. The applied cost of fluoroelastomer linings (per
square foot) is approximately 300 percent higher than natural
rubber. The thickness of fluoroelastomer linings is only
about 20 percent of rubber linings.
Inorganic Linings
Inorganic linings are more expensive than organic
linings, with the exception of the fluoroelastomers. They
are used in parts of the system where corrosion protection is
required and where the temperatures are too high for the
organics. An important application of inorganic linings is
3-24
-------
in outlet ductwork, downstream of the reheater. They are
applied to the base steel as either monolithic or as
brick/mortar linings. Monolithic linings are gunned onto the
steel. A steel wire mesh, anchored to the base steel, is
required for support. In areas where these linings become
wet, an impervious membrane is required at the steel
interface. A relatively large number of proprietary
monolithic lining materials are available for FGD
applications. These materials are quite heavy; many are not
resistant to strong acids that may exist in the systems.
Acid brick/mortar has good corrosion and abrasion
resistance and is used in some prescrubbers, where erosive
slurries wear away most coatings. A membrane is necessary to
protect the substrate. Many systems are designed with a
layer of acid proof brick to protect the bottoms of various
tanks against physical damage while being.cleaned . Acid
brick is resistant to the acidic environments in wet stacks
(no reheat) .
Borosilicate glass blocks are being used to line outlet
ductwork in several installations. This material has the
added advantage of being a good thermal insulator and
external insulation is unnecessary. The mortar must be
selected to withstand the environment. A membrane is
necessary to protect the substrate.
The major problems associated with inorganic linings are:
- Cracks often develop during the curing of linings and
mortars.
- Corrosive gases (S02, SO^, HC1) can diffuse through
lining and mortars requiring impervious membranes to
protect carbon steel substrates.
PLASTICS
Fiberglass Reinforced Polyester (FRP) is noted for its
very good corrosion resistance and moderate temperature
resistance. Solid FRP construction is used in piping
systems, for small tanks, mist eliminators, and stack
liners. FRP flues are used in several stacks.
CERAMICS
The use of ceramic materials has been limited to spray
nozzles. They are highly successful there, and the success
justifies the high initial cost.
ALLOYS
Alloys are chosen because their physical or chemical
3-25
-------
properties are superior to those of the component metals in
such areas as strength, and corrosion resistance. Alloys
have been classified "low", medium", and "high" according to
their resistance to corrosion and approximate cost range.
These are listed in Table 3; registered trademarks are listed
in Table 4.
The low alloys include the austenitic (or 300 series)
stainless steels and FERRALIUM ® Alloy 255, a duplex
(austenitic-ferritic) stainless steel that contains less
nickel, about the same amount of molybdenum, and more
chromium than austenitic stainless steels. Alloy 255 has
seen wider application in Europe and is becoming increasingly
popular in FGD applications in the USA.
The baseline corrosion resistant alloy for FGD systems is
316L, for the following reasons:
- generally good corrosion resistance,
- easy to clean,
- low carbon levels minimize welding problems,
- it is readily available at moderate cost.
316L has been successfully used in scrubber vessels,
internal components and reheater components, where
temperatures are below 140 F, pH's are above 5.5, and
chloride levels are below 1,000 ppm. In lower pH or higher
chloride environments, 316L is vulnerable to corrosion,
particularly when it is used as shell material or structural
members. Severe corrosion h.as been observed between any
gypsum scale that might form, and the metal. 317LM (4.0/5 Mo
minimum) is used for slightly more severe conditions. A few
utilities have selected Alloy 255 to replace 316L and 317LM
parts that have undergone pitting corrosion.
Medium alloys are used in areas where the corrosive
environments are more aggressive. They are characterized by
generally higher chromium, molybdenum, and nickel contents
than are contained in the low alloys. Some alloys in this
class have small amounts of copper added for higher
resistance to sulfuric acid.
High alloys are used when chloride and/or sulfuric acid
concentrations are high, pH levels are low, or when
temperatures are high. High alloys differ from low and
medium alloys in that they are nickel, rather than iron-base
alloys. They contain higher molybdenum and generally higher
chromium levels than do the medium alloys.
The Limestone Users Handbook^) states that:
"316L SS is finding widespread use as a material of
3-26
-------
TABLE 4
REGISTERED TRADEMARKS
FERRALIUM is
20-Cb-3 is a
INCOLOY is a
INCONEL is a
HASTELLOY is
CABOT is
STEBBINS
a registered trademark of Bonar Langley Alloys, LTD.
registered trademark of Carpenter Technology Corp.
registered trademark of International Nickel Co.
registered trademark of International Nickel Co.
a registered trademark of Cabot Corporation
a registered trademark of Cabot Corporation
is a registered trademark of Stebbins Engineering and
Manufacturing Company
3-27
-------
construction, however, under certain conditions of
scrubber liquor pH, temperature, and chloride
content, this alloy can undergo localized attack.
Under these more stringent conditions, nickel-base
alloys with higher molybdenum and chromium content
are superior to 316L stainless steel. It has been
shown that resistance to stress-corrosion cracking
is achieved with higher nickel contents because
nickel accelerates repassivation of the metallic
surface^?). Although more expensive initially,
these high-grade alloys may be economically
justified for use in certain severe scrubber
environments.
The beneficial effect of molybdenum content is
shown in Figure 3 which indicates that the
resistance of alloys to pitting and crevice
corrosion generally increases as the molybdenum
content increases (8). Moreover, molybdenum
content alone does not ensure resistance to
localized attack."
Chromium content is also important as shown by the Shawnee
test results plotted in Figure H (°) an
-------
c
o
CO
O
O
u
a)
o
u
T3
C
n)
60
C
3
O
•H
3
03
a.
efl
100
90 —
80
10'
60
50
40
30
20.
I
(2.3%
(2.8% Mo
316L Stainless Steel
316L Stainless Steel
317L Stainless Steel (3.2% Mo
HASTELLOY alloy G _
INCONEL 625
HASTELLOY alloy C-276
I
4 6 8 10 12
Molybdenum Content, Wt. %
14
16
Figure 3: Effect of molybdenum content on resistance
to pitting and crevice corrosion.
3-29
-------
1 i i i i i I r
u
1 316L Stainless Steel
(2.3% Mo)
2 316L Stainless Steel
(2.8% Mo)
3 317L Stainless Steel
4 HASTELLOY alloy G
5 INCONEL 625
6 HASTELLOY alloy C-276
T
I i I I I I i I I
0 10 20 30' 40 50 60 70 80 90 100
Samples with Pitting and Crevice Corrosion, %
Figure 4: Effect of molybdenum and chromium
content on corrosion resistance
(Shawnee tests).
3-30
-------
TABLE 5
SERVICE APPLICATION GUIDELINES BASED ON CABOT TEST RESULTS
Service Condition
Maximum Temperature, °V
pH
Chloride Content, ppm
316L
Stainless Steel
150
Over 5.5
Under 1500
FERRALIUM
alloy 255
150
Over 4.5
Under 3000
HASTELLOY
alloy G/G-3
160
Over 3.5
Under 5000
CABOT
alloy No. 625
160
Over 3.5
Under 8000
HASTELLOY
alloy C-276*
225
Over 2.0
30,000
* Alloy C-276 has been used successfully up to at least these levels.
OJ
i
LJ
-------
Absorber Inlet
This is the interface between the inlet duct and the body
of the absorber. It is subjected to thermal stresses and the
most aggressive corrosive environment in an FGD system.
Gases enter this section at temperatures varying between
their normal values of 250 to 350 F, and upsets which may be
as high as 700 or 800 F; and leave at or near their
saturation temperatures which vary between about 110 and 150
F, depending on gas inlet temperature, humidity, and
altitude. In several systems, the absorber inlet sections
serve as pre.saturators, where large volumes of FGD liquids or
slurries, sprayed into the gas, quench it to its saturation
temperature. Other systems provide liquid sprays to flush
the walls and prevent solids from building up at the wet/dry
interface; the gases are only partially saturated here.
Other systems use carefully constructed inlet vanes to
maintain stable gas velocity profiles, and still others
provide no protection at all.
The high temperatures and temperature gradients at the
entry section of the absorber inlet preclude the use of most
organic linings. Inorganic linings can, and have been, used
here but the use of high and medium alloys is preferred.
This region of the ductwork where the most severe temperature
problems and the most agressive corrosion environments exist
is usually quite small, and the additional cost of the alloy
parts is justified by the trouble free operation that will
ensue. Alloys are not subject to the major defficiencies of
inorganic linings which are delamination from the base steel
and leakage of corrosives to the base metal. The absorber
inlet should be designed to prevent accumulation of solid
materials at the wet/dry interface between the ductwork and
the absorber, and to avoid stagnation^ zones, where pools of
absorber liquids can accumulate and cause corrosion.
Absorber Body
The absorber body is the largest and costliest component
of the FGD system and must be protected from corrosion,
erosion, and thermal and chemical effects. Of these, the
thermal effects are usually the smallest.
The majority of FGD absorber bodies are constructed of
either organic lined carbon steel or low alloys. The "EPA
Utility FGD Survey" (C1110) reports that organic lined carbon
steel is used in approximately 75% of all FGD absorbers. The
alloy absorbers, almost exclusively- 316 or 316L, have only
been successful when exposed to low concentrations of
corrosive materials -- notably chlorides. Alloy absorbers
3-32
-------
must be designed with a minimum of sites for corrosive
materials to accumulate. Alloys have the advantage of ease
of repair and modification; penetrations and welds can be
made with a minimum of difficulty. Organic lined steel
absorbers do not have the same limitations on corrosive
materials or physical design but linings must be periodically
inspected and repaired as necessary.
It is also recognized that different locations within
lined FGD absorbers are exposed to different corrosive and
erosive environments. Absorbers are therefore designed with
different materials at different points. The spray zones,
subjected to the severest abrasive conditions, are usually
rubber lined. The direction of the recycle sprays must be
carefully adjusted to minimize impingement on, and erosion
of, other internal components, such as walls, beams and
pipes.
The components of the absorber not directly exposed to
the abrasive action of the recycle sprays are often coated
with glass filled polyester resin. This material has
equivalent corrosion resistance and is easier to apply and is
less expensive than rubber. Alumina filled resins may be
used in applications involving high fluoride concentrations.
A well designed steel absorber shell is usually covered
with at least two different linings, typically rubber and
flake filled polyester. Direct chemical bonding between such
materials is not usually successful, and the absorber shell
should be made of flanged sections, one covered with rubber,
bolted to another covered with polyester. The linings are
extended oustide of the absorber body and completely cover
the flange faces to assure leak tight joints. This is shown
schematically in Figure 5.
A small number of ceramic tile absorbers are being
successfully operated.
Chemical phenomena include SC>2 removal, limestone
dissolution, gypsum desupersaturation, and system pH, and
have been studied at length and reported in various symposia
and similar learned publications.
Mist Eliminators
Mist eliminators are generally constructed in two
stages. In the Peabody system, the first, or lower stage, is
a perforated interface tray that acts as a bulk entrainment
separator. It is continuously washed with a combination of
process slurry and makeup water. Despite the potentially
3-33
-------
STEEL ABSORBER SHELLr—
GASKET
FLAKEGLASS
A
GAS FLOV7
RUBBER
STEEL ABSORBER SHELL
Figure 5: Mechanical Joint for Dissimilar Linings
3-34
-------
corrosive environment, successful operation with 316L
stainless steel interface trays has been achieved since it is
not a structural member and is continuously flushed. In
applications where very high chlorides are expected, higher
alloys such as Alloy G have been specified.
The second, or upper mist eliminator stage, is usually a
set of chevrons. The major problems here are thermal
deformation of the blades, and slurry deposition and scaling
within the chevron passages. The thermal deformation has been
attributed to the diffusion of hot flue gas from an outlet or
bypass duct into the top of an absorber module that is being
taken off line, and where the inlet damper is closed before
the outlet damper. The problem has been successfully solved
by altering the damper startup/shutdown sequences and by
replacing the somewhat fragile thermoplastics with more
rugged FRP or Alloy parts. Solids deposition and scaling are
potentially serious but these have been overcome by the
introduction of a sufficient flow of water to remove any
deposits before they can be converted to gypsum and adhere to
the surface. The absorber shell in this area is carbon steel
with a flakeglass filled polyester resin lining.
Absorber Outlet
The absorber outlet may experience a more highly
corrosive environment than the body of the absorber but
abrasion is not a problem. The corrosive environment is
being created when residual sulfur dioxide or sulfur trioxide
is absorbed by condensed water and not neutralized by the FGD
alkali. This condition may be exascerbated by the presence
of chlorides .
Despite the corrosive nature of the liquids here,
corrosion is usually minimal because of the rather low
(saturation) temperatures, the suitability of the linings,
and the vertical arrangement of the equipment which minimizes
accumulations (especially important in alloy vessels). The
same materials of construction used, in the mist eliminator
sections are used here: the absorber shell is carbon steel
with a flakeglass filled polyester resin lining.
Outlet Ductwork
The outlet ductwork is subjected to the same conditions
as the upper sections of the absorber but corrosion is
potentially more severe because of the horizontal position of
the ductwork. Lined steel is the most frequently used
material of construction here.
Reheaters and Associated Ductwork
3-35
-------
Many FGD systems are designed with reheaters to increase
gas buoyancy and velocity within the stack. Outlet ducts,
prior to reheat or bypass zones are carbon steel with
flakeglass filled polyester resin linings. For the hotter
zones, monolithic inorganic linings, borosilicate glass
blocks, and fluoroelastomers are used to protect carbon steel
ducts. Alloys are used as both.corrosion resistant linings
for carbon steel, and as free standing structural materials
in the ductwork at several installations. The 300 series
stainless steels often undergo pitting corrosive and, in some
systems even meduim alloy are inadequate.
Reheat is accomplished either directly or indirectly. In
direct heating a row of heated tubes is placed in the path of
the scrubbed flue gas. The first few rows of tubes are prone
to scaling due to the evaporation of entrained water droplets
and deposition of the solid material on the tube surface.
The tube surfaces may also be moist due to the presence of
residual liquid, and thus an area of potentially high
corrosion can be established. Unfortunately good heat
transfer requires bare metal surfaces so the tubes cannot be
coated with a protective material. Solid deposits are
removed from the surfaces with soot blowers. Tubes are
generally constructed of high alloys at the reheater inlet
and lower alloys, or even carbon steel for the remaining
tubes.
In indirect heating, a gas, usually air, is externally
heated and then mixed with the scrubbed flue gas to bring the
mixed temperature up to the desired point. A mixing device
is required to thoroughly mix the hot reheat gas with the
cooler scrubbed gas. The gas mixing device has many of the
same corrosion problems as the absorber inlet section where
the most aggressive corrosive environments and largest
temperature gradients exist. The mixing baffle is usually
constructed of medium or high alloys. Indirect reheating is
more commonly used than is direct reheating.
A variation of indirect reheating is "bypass reheating".
This is usually economically attractive in systems burning
low sulfur coal and requiring low (approximately 70%} overall
862 removal. The absorbers are operated at high efficiencies
and a portion of the flue gas is bypassed and mixed with the
scrubbed gas to increase the temperature. The reheat gas is
not scrubbed and carries with it the sulfur dioxide, sulfur
trioxide, and hydrogen chloride present in the original flue
gas. This creates more serious corrosion problems for the
ductwork and the gas mixers. The ductwork is usually lined
as mentioned above and the mixing baffle is made of a higher
alloy. Another application of bypass reheat involves using a
3-36
-------
very small amount of air preheater inlet gas (at about 800 F)
to heat scrubbed gas. A small electrostatic precipitator
removes particulate from the reheat gas prior to mixing.
Fluoroelastomers appear to be successful in outlet and
reheat ductwork but experience is limited.
Reheaters generally employ inorganic linings over carbon
steel for the ductwork, and medium and high alloys for the
gas mixer.
Stack Breeching
If neither reheat or bypass gas is introduced, flakeglass
linings are acceptable. If the design requires reheat,
higher temperatures preclude the use of organic linings.
Inorganic linings have been used with some success, and
considerable attention is being directed towards the use of
alloys. Solid FRP flues are used in a number of
installations. An alloy flue is used in at least one
installation.
Alkali Tanks
Alkali storage tanks are usually made of unlined carbon
steel.
Recycle Tanks
Recycle tanks, located at the base of absorber modules,
provide the necessary residence time for the FGD chemical
reactions. The preferred material of construction here is
carbon steel with linings of either glass filled polyester or
vinylester resin and an acid brick covered base. When
agitation is required, rubber covered agitators are used.
3-37
-------
Slurry Piping
Both rubber lined steel and FRP pipe are used for slurry
service. Rubber lined and rubber covered pipe is used inside
the absorber vessels. Rubber lined pipes are expensive but
are extensively used because they combine abrasion and
corrosion resistance, and are very reliable.
Slurry Nozzles
Nitride bonded silicon carbide is almost exclusively used
for slurry nozzles. There have been no reports of failures
due to either abrasion or corrosion with this material.
Applications that require water sprays, such as in mist
eliminators and certain presaturators, employ cobalt-base
wear resistant alloy nozzles. This minimizes abrasion due to
small amounts of suspended solids.
Dampers
Dampers are subjected to the same environments as the
ducts in which they reside. Blades are carbon steel or low
alloy in the inlet ductwork, and medium or high alloy in the
outlet ductwork. The choice of materials is site specific
and depends upon such considerations as temperature and
sulfur trioxide level of the incoming gas, and temperature
and sulfur dioxide, sulfur trioxide, and chloride contents of
the absorber outlet gas. Damper seals are usually fabricated
of high alloys, such as 625 and C-276 for both inlet and
outlet dampers.
EXPERIENCE
This section describes several case histories where
materials of construction listed in Table 2 have been used.
ORGANIC COATING APPLICATIONS
Minnesota Power and Light Company, Clay Boswell No. 4,
Cohasset, Minnesota is a 550 MW coal-fired plant. The plant
burns 1.0 % sulfur Montana coal. A lime/alkali flyash FGD
system that operates in a closed water loop mode removes 90
percent of the S02 from the flue gas. Reheat is supplied by
bypass gas. Operations commenced in April 1980. Extensive
use of non-metallic coatings over carbon steel was made in
this system. The venturi is lined with natural rubber in
wetted areas. The inlet section of the venturi which
receives the hot flue gas and directs it to the orifice area
is carbon steel and inorganic lining. The absorber vessel is
fabricated in two sections. The absorption zone is lined
3-38
-------
with natural rubber. The mist elimination section above the
absorption zone is lined with polyester flakeglass. The
internal spray headers are carbon steel rubber lined, and
rubber covered. The outlet ductwork downstream of the
absorber has a vinyl ester flakeglass lining. The reheater
mix zone is 316L. The materials of construction selected for
this unit have performed satisfactorily. A problem was
initially experienced with the rubber lined venturi vessel
wall. The very high swirling velocity of the gas/slurry
mixture leaving the venturi orifice eroded the rubber
lining. The problem was solved by the installation of
straightening vanes in the inlet duct.
RUBBER COATINGS AND ALLOY APPLICATIONS
Alabama Electric Cooperative, Tombigbee Station, Units
No. 2 and No. 3> Leroy, Alabama, are each 255 MW units that
burn 1.8 percent sulfur (maximum) coal. The limestone FGD
system scrubbing liquor contains about 800 ppm of chlorine
and the pH is normally maintained at 5.8. Organic coatings
over carbon steel are used extensively in this installation.
Natural rubber and flakeglass filled polyester resin linings
are used in the absorber and recycle tanks; and a
fluoroelastomer lining is used in the outlet ductwork. The
absorber inlet is INCOLOY (g) Alloy 825. The system has been
in service for more than four years. Only minor repairs have
been necessary.
Solids deposition in the absorber inlet ducts and the
resulting corrosion is avoided by the installation of
deflector baffles around the perimeter of the inlet duct to
divert the flow of gas away from the wall. Fan shaped sprays
located behind the deflector baffle deluge the duct wall.
The duct is inclined so that the liquids drain into the body
of the absorber.
South Mississippi Electric Power Association, R. D.
Morrow, Sr. Plants (2 units, 190 MW each), Purvis,
Mississippi uses a limestone process with a closed water
loop. The plant burns 1.1 percent sulfur coal, and uses
bypass gas to reheat the scrubbed gas. The gas temperature
in the outlet duct and the stack breeching is above 200 F,
which is higher than that in most other plants. A
scale/sludge buildup, a foot or more thick, occurred on the
inlet duct floor. Samples of this scale have been analyzed
and found to contain high levels of chlorides and fluorides.
A history of the severe corrosion problems and the materials
used in an effort to combat these corrosion problems at this
plant has been reported by Froelich and Ware(13). Rubber
lining in the upper section of the absorber and the breeching
between the top of the absorber and the outlet duct, and
3-39
-------
HASTELLOY © alloy G in the lower section of the absorber have
proven to be satisfactory materials for those areas. Areas
with the most stubborn corrosion problems have been the floor
of the inlet duct and portions of the outlet duct. Based on
corrosion tests conducted in the outlet duct for one year,
the results of which are summarized in Table 6 and detailed
elsewhere( 11*) , HASTELLOY Alloy C-276 was selected for the
inlet duct floor material and for use in the most severely
corroded areas in the reheater section of the outlet duct.
The alloy C-276 was installed nearly three years ago. To
date, only minor corrosion has been observed on the inlet
duct floor and no corrosion is evident in the outlet duct.
ALLOY APPLICATIONS
Southern Indiana Gas and Electric, A. B. Brown No. 1,
West Franklin, Indiana is a dual alkali process with a closed
loop water system. The fuel is a bituminous coal with 2.8 to
3.9 percent sulfur and 100 to 800 ppm chlorides. HASTELLOY
alloy G was selected for two absorber tower disc components
and two bottom spargers. Each contactor consists of a disc
and doughnut, approximately 30 ft. in diameter. The initial
startup was in April 1979. An inspection 4 years later
revealed no corrosion.
The Northern Indiana Public Service Company, Dean H.
Mitchell Station No. 11, Gary, Indiana, used a Davy
McKee/Wellman Lord S02 recovery process with an Allied
Chemical Corporation reduction process to reduce the S02 to
99-9 percent pure sulfur. These combined processes
eliminated the need for sludge disposal. The plant burned
coal with 3«5 percent sulfur. The absorber was constructed
of STEBBINS ® tile. The orifice contactor and bottom
supports of the absorber trough were made of HASTELLOY alloy
C-276 and the outlet duct of HASTELLOY alloy G. This
demonstration plant discontinued operation after running
seven years, and was placed in the standby mode. Examination
of the tile and the HASTELLOY alloy components after shutdown
showed that they were essentially free of corrosion. The
tile components were trouble free for the duration of the
test.
Montana Power Company, Colstrip No. 1 and 2 (360 MW
each), Colstrip, Montana has plate type stack gas heaters
fabricated of HASTELLOY alloy G and CABOT ® alloy No. 625
plate type stack gas reheaters. These reheaters have
performed satisfactorily with no corrosion problems since
startup (No. 1, Nov. 1975; No. 2 Oct. 1976). The same
combination of alloys has been chosen for the new power
plants, Colstrip No's. 3 and *4, scheduled for startup in 1983
3-40
-------
TABLE 6
RESULTS OF CORROSION TESTS AT SOUTH MISSISSIPPI
ALLOY
Alloy C-276
Alloy No. 625
Alloy G-3
Alloy G
JS-700
Alloy 255
904L Alloy
317L Stainless Steel
316L Stainless Steel
REMARKS
No localized corrosion
Weld metal pitting
Moderate pitting in weld,
light in base metal
Moderate pitting in weld,
light in base metal
Moderate pitting
Moderate pitting
Heavy pitting
Heavy pitting
Heavy pitting
3-41
-------
.and 1984, respectively.
Central Illinois Public Service Company, Newton No. 1,
Newton, Illinois has a dual alkali closed water loop FGD
system. The plant burns Southern Illinois coal with 2.8 -
3.2 percent sulfur (4.0 percent maximum) and 0.2 percent
chlorine. The scrubbing liquor contains 15,000 to 18,000 ppm
chlorides and the pH is maintained at 6.5. The system has
four absorber towers each with a 22.5 foot diameter by 2 ft.
high precooler ring at the top where the hot (approximately
325 F) unscrubbed gases enter and are quenched. The ring and
spray quench system and other internals are made of HASTELLOY
alloy C-276. Corrosion of these components has not been a
problem in the four years of operation of this system.
FUTURE TEST WORK
Cabot Wrought Products Division and Peabody Process
Systems, Inc. have joined forces with three major utilities;
NYSEG, Seminole, and Santee Cooper; for an extensive
experimental survey of the corrosion resistance of various
materials under actual field conditions. The three utilities
are burning different coals which will provide three
distinctively different corrosion environments.
Coupon racks have been designed by Peabody and built by
Cabot. They will be installed prior to plant startup, at
selected locations within the FGD system. Operating data,
such as temperature, pH, and chloride levels, will be
monitored by Peabody and by the utilities; and Cabot will
analyze the samples in their research laboratories. The test
racks will be examined after 12 and 24 months of exposure.
Cabot and Peabody will jointly publish the results of this
investigation .
SUMMARY
This section highlights the preferred materials of
construction from the large number of choices available.
The recommended material of construction for inlet
ductwork is carbon steel. The recommended material of
construction for the absorber inlet is a medium alloy because
it can withstand both the thermal and corrosive
environments. The physical design must prevent solids from
building up. Carbon steel with one or more organic linings
is the preferred material of construction for the absorber.
Natural rubber is recommended for the absorption zone, the
pump linings, and the internal slurry piping, which should be
lined and covered. If oil firing is a major factor then
neoprene rubber is preferred. Flakeglass filled polyester
3-42
-------
has proven satisfactory for the upper and lower sections of
the absorber and the outlet ductwork, and is recommended
there. The mixing sections of reheaters present potentially
corrosive environments and medium and high alloys are
recommended. This is especially important in applications
involving bypass reheat where there may be excessive amounts
of chlorine and sulfur trioxide. The tubes of direct
reheaters should be high or medium alloy at the inlet and low
alloy or carbon steel for the remaining tubes.
Damper selection is a complex matter and can itself
provide the basis for a detailed study. The problems in
damper selection are well known and need not be repeated
here. The recommended materials of construction for the
blades of inlet dampers are either carbon steel or low
alloys. The blades of outlet dampers should be low alloys.
Damper seals are continuously exposed to corrosive
environments, and high alloys such as C-276 and 625 are
recommended.
An FGD system is comprised of many subsystems, which when
joined together, form a functioning unit. This paper has
stressed the materials aspect of system design; it has
avoided the chemical phenomena, the mechanical considerations
and the other technologies that are incorporated into the
design of a complete system. The point that must be stressed
during the design of a system is that there is no best single
material of construction for all components of all systems.
There are many excellent products, made by established and
reliable manufacturers. The choices are not always simple
nor obvious. There are site specific situations that make
selections that were optimal for one FGD system practically
useless in another. In preparing this paper the authors have
surveyed a broad spectrum of available materials and related
these to the design, construction, and operation of several
operating installations. They are intended to provide a
place for the reader to begin investigation rather than a
solution to all problems. Much progress has been made during
the rather brief lifetime of the FGD industry. Much is
continuously being learned, and mu'ch remains to be learned.
ACKNOWLEDGEMENT
The authors wish to thank Mr. T. Newhams for his
assistance and support, and Miss J. Comeau and Mrs. D. Green
for the preparation of the manuscripts.
3-43
-------
REFERENCES
1. Stevenson, W. L., "Federal S02 Emission Standards: What do They
Mean", POWER, May 1980, p. 130-131.
2. Chang, J. E. S. and Laslo, D. J., "Chloride Ion Effects on
Limestone FGD System Performance", Proceedings of the EPA/EPRI FGD
Symposium, Hollywood, Florida, May 1982.
3. Laslo, D. J. and Chang, J. E. S., "Pilot Plant Tests on the
Effects of Dissolved Salts on Lime/Limestone FGD Chemistry",
Proceedings of the EPA/EPRI FGD Symposium, New Orleans, Louisiana,
November, 1983.
4. Pierce, R. R., "Estimating Acid Dew Points in Stack Gases",
Chemical Engineering, April 11, 1977.
5. Johnson, C. A., "Flyash Alkali Technology - Low-Cost Flue Gas
Desulfurization", Proceedings of Coal Technology '80, Houston,
Texas, November 1980.
6. Henzel, D. S. et al, "Limestone FGD Scrubbers: Users Handbook",
EPA-600/8-81-017 (1981).
7. Rhodin, T. N. and Carson, H. R., "Physical Metallurgy of Stress
Corrosion Fracture", Interscience Publishers, 1959, p. 451 - 456.
i
8. Crow, G. L., "Corrosion Tests Conducted in Prototype Scrubber
System", Proceedings of the Corrosion Problems in Air Pollution
Control Equipment Symposium, January, 1978, Atlanta, Georgia.
9. Hoxie, E. C. and Michaels, A. T., "How to Rate Alloys S02
Scrubbers", Chemical Engineering, June 5, 1978, p. 161-165.
10. Leonard, R. B., "Application of Nickel-Chromium Alloys in Air
Pollution Control Equipment", presented at NACE/APCA Air Pollution
Control Seminar, Atlanta, Georgia, January 17-19, 1978.
11. Koch, G. H. and Beavers, J. A., "Materials Testing in Simulated
Flue Gas Desulfurization Duct Environments", EPRI Report CS-2537,
August 1982.
12. Paul, G. T., "Corrosion in Flue Gas Scrubbers and Some Methods of
Mitigation", Proceeding of the Solving Corrosion Problems in- the
Air Pollution Control Equipment Seminar, August, 1981, Denver,
Colorado.
13. Froelich, D. A. and Ware, M., "Corrosion Problems with a Closed-
Loop Limestone FGD System", Paper No. 203, NACE CORROSION/82,
March, 1982, Houston, Texas.
3-44
-------
Silence, W. L. and Manning, P. E., "Laboratory and Field Corrosion
Test Results Related to FGD Systems", Paper No. 185, NACE
CORROSION/83, April, 1983, Anaheim, California.
3-45
-------
ACID DEPOSITION IN FGD DUCTWORK
D.A. Froelich, C. V. Weilert,
P. N. Dyer
-------
ACID DEPOSITION IN FGD DUCTWORK
by
Daniel A. Froelich, P.E.
Carl V. Weilert, P.E.
Paul N. Dyer
Burns & McDonnell Engineering Company, Inc.
Kansas City, Missouri
BACKGROUND
In the mid 1970s, first generation wet lime/limestone flue gas
desulfurization (FGD) systems were included in the design of many new coal-
fired power plants across the United States in response to federal, state,
and local sulfur dioxide (S02) emission control regulations. At the time,
the primary concern within the electric power industry was whether the new
systems would work at all. In retrospect, the overall performance of these
FGD systems has been good. Based upon data from the best of these first
generation systems, Congress and EPA have effectively mandated the use of FGD
systems on all future coal-fired power plants. Continuous S02 reduction is
the law of the land. Reliability is the new watchword for the utility
industry.
For owners and operators of existing FGD systems and for designers and
suppliers of new systems, problems with ductwork corrosion represent a major
obstacle to the goal of 100 percent reliability. Corrosive failure of
materials of construction in FGD system ductwork is a leading contributor to
low reliability in existing FGD systems.
The corrosion experiences at the R. D. Morrow Station are a prime
example of this industry-wide problem. The corrosion rates of the alloys and
coating materials used at R. D. Morrow have been much higher than anticipated
in the design. Some alloy materials chosen for repair of the initial
corrosion on the basis of the results of laboratory corrosion tests in
simulated FGD environments have also experienced rapid corrosion. In most
cases, even coupon spool tests conducted at R. D. Morrow have been unable to
predict the severe corrosion that has occurred when the test material was
installed full-scale in the ductwork.
Based upon this experience it was concluded that: (1) the corrosive
environment in the ductwork is more aggressive than that to which the spool
coupons are exposed; and (2) the corrosive environment in the ductwork is
more aggressive than that which had previously been simulated in the
3-47
-------
laboratory. It was clear that more and better data was needed to define the
conditions that existed in the ductwork.
There was not sufficient information available regarding the chemical
composition of the material that collects on the FGD ductwork surface.
Collection, preservation and analysis of duct surface deposits obtained
during periodic inspections at the time of system outages had proven to be
difficult. Data from this type of sample was inconsistent and misleading.
It became clear that there was a critical need for a method by which ductwork
deposition samples could be obtained while the FGD system was operating.
Burns & McDonnell has identified and developed two innovative methods
for characterization of the corrosive environment in FGD system ductwork. A
predictive technique based on vapor-liquid equilibrium data for acid
solutions is used to estimate acid concentrations in ductwork condensate. An
extractive test method utilizing a controlled-temperature condenser is used
for on-line collection of duct deposits from FGD system ductwork. Working
under contract with EPRI (Research Project 1871-4), Burns & McDonnell
conducted a test program at R. D. Morrow to evaluate the two methods. This
paper describes these two methods and their applicability to the solution of
corrosion problems in FGD system ductwork.
OCCURRENCE OF ACID CONDENSATION IN FLUE GAS DUCTWORK
Flue gas from the combustion of coal contains a number of different
acids in the form of condensable vapors. These acid vapors are present in
trace concentrations in the gas stream. Sulfuric acid (F^SOij) is the most
common, but hydrochloric acid (HCL), hydrofluoric acid (HF), hydrobromic acid
(HBr) and nitric acid (HNO^) may also be present.
If the temperature of the flue gas stream, or of the surfaces to which
the gas stream is exposed, drops below the acid dew point temperature, acid
condensate will form. The condensate may appear either as a thin film on the
duct surfaces, or as aerosol droplets entrained in the gas stream, or a
combination of the two. The acid dew point temperature for each of the
condensable acid vapors listed above can be calculated from thermodynamic
relationships. These relationships have been reduced to relatively simple
equations that express the acid dew point temperature for each acid as a
function of the acid vapor concentration and the water vapor concentration in
the flue gas stream.
The relationship between f^SOij vapor concentration and sulfuric acid dew
point, at some constant H20 vapor content, is commonly displayed as the
familiar "acid dew point curve." The family of dew point curves shown in
Figure 1 illustrates the effect of flue gas moisture content on the sulfuric
acid dew point.
Considering that H2SOi| vapor concentration in the flue gas from coal-
fired boilers may be as much as 1 to 3 percent of the S02 concentration, it
is apparent from Figure 1 that sulfuric acid condensate will form in ductwork
downstream of the air heaters, where flue gas temperatures are typically less
than 300 degrees F. For example, consider a flue gas containing 2,000 ppm
3-48
-------
330-
300-
270-
0)
0)
Q
O
Q.
5
0)
Q
240-
210-
180-
150«
0.01
0.1
1.0
I
10
100
H2 SO Jn Flue Gas (PPM)
Figure 1. Effect of Flue Gas Hz S04 Vapor Concentration and
Moisture Content on Acid Dewpoint
3-49
-------
S02. At 1 percent conversion to ^SOq this represents 20 ppm t^SOij vapor in
the flue gas. If the moisture content is 7.5 percent by volume, the sulfuric
acid dew point temperature, from Figure 1, is approximately 285 degrees F.
When this flue gas enters a wet FGD system the temperature drops rapidly
and the moisture content increases. The sudden drop to temperatures well
below the acid dew point results in the condensation of most of the ^504
vapor as aerosol droplets (acid mist). This acid aerosol remains entrained
in the gas stream. Due to the small diameter of the droplets a large
percentage of the total mass of the F^SOij can pass through the absorption
system without being removed from the gas stream.
The high-moisture, low-temperature conditions of the absorber outlet gas
stream are also conducive to the condensation of the other acid vapors, which
typically have dew points much lower than that for sulfuric acid.
CHARACTERIZATION OF FGD DUCTWORK ENVIRONMENT
Acid condensation in flue gas ductwork is common. Yet acknowledgement
of this fact does not explain why severe corrosion has occurred at one FGD
installation and not another, or why corrosion rates observed in full scale
installations have been often much greater than those predicted from
laboratory corrosion tests. As an important step toward the answer to these
questions, two methods have been used to characterize the corrosive
environment which exists in the ductwork of FGD systems. One method is a
predictive technique based on empirical vapor-liquid equilibrium data for
acid solutions. The second is an extractive test method which collects
simulated duct surface deposits for chemical analysis.
PREDICTIVE TECHNIQUE
The first method for characterization of acid condensate in FGD system
ductwork involves the use of vapor-liquid equilibrium data for acid solutions
as a function of temperature and acid solution concentration. As noted
above, flue gas temperatures or duct surface temperatures below the acid dew
point will result in formation of an acid condensate. The condensate will be
a solution of the acid in water. The concentration of the acid solution that
forms in this manner can be predicted from the analysis of vapor pressure
data for the acid in question.
Physical Principle
The flue gas contains a certain percentage of water vapor, depending
upon the moisture content of the coal, the ambient humidity, the amount of
excess air and the flue gas saturation condition. The amount of water vapor
in the flue gas varies from as little as six percent in the unscrubbed gas to
15 percent or more at the saturated conditions in the absorber and the
absorber outlet ductwork.
In any case, the amount of water vapor in the flue gas is many times
greater than the amount of acid vapor present. The total vapor pressure is
therefore essentially a function of the moisture content only. When
3-50
-------
conditions are such that an acid condensate forms, the vapor above the
condensate is in equilibrium with the liquid.
The equilibrium relationship between the composition of the vapor and
the composition of the condensate is a function of temperature only.
Consequently, if the temperature at which the condensate has formed is known,
and if the moisture content of the vapor is known, the concentration of the
acid in the condensate can be determined. Figure 2 displays this
relationship in graphical form for the case involving sulfuric acid
condensate.
Examples
To illustrate the use of Figure 2, consider the same examples that were
used previously. A flue gas with 7.5 percent moisture and 20 ppm H2SOjj (285
degrees F acid dew point) leaves the air heater at 300 degrees F. As this
gas moves through the ductwork it encounters duct surface temperatures of 280
degrees F. A condensate film forms on the duct surface. From Figure 2, it
is predicted that the condensate is between 80 and 85 percent H2SOij by
weight.
Now, the same flue gas passes through an electrostatic precipitator and
into a wet limestone FGD system. The gas is cooled to the adiabatic
saturation temperature of approximately 130 degrees F and humidified to 15
percent moisture by volume. The I^SOij is condensed as an aerosol mist
entrained in the flue gas. From Figure 2, the concentration of the acid in
the aerosol droplets is about 10 percent H^SOij by weight. Any remaining
f^SOij vapor that condenses on the surfaces of the outlet ductwork will
also form condensate that is 10 percent or less H2S04 by weight.
This type of graphical analysis using Figure 2 can be used for any case
in which acid vapor and acid condensate are both present and equilibrium has
been established. The data used to draw the curves of Figure 2 are derived
from experimental measurements of vapor pressures over acid solutions of
various concentrations over a wide range of temperatures. However,
calculation of -the data from thermodynamic relationships is also possible.
Implications for FGD System Design
There are several important trends evident in the curves of Figure 2.
It will be noted that acid condensate concentration is lowest when moisture
content is high. Also, condensate concentration decreases with temperature.
These phenomena are relevant to the design of FGD systems.
For example, consider the implication of the use of reheat. Flue gas
reheat is accomplished by several methods. One method involves the bypass of
an unscrubbed gas stream around the absorber for mixing with the scrubbed gas
in the outlet ductwork or in the stack. The result is a gas which is warmer
and dryer than the saturated gas leaving the absorber. From Figure 2 we see
that both these characteristics will tend to increase the acid concentration
in the aerosol droplets and/or the ductwork surface condensate film.
3-51
-------
O
C
O
*^
co
0)
O
C
O
O
90-
80-
70-
60-
50-
40-
30-
20-
10-
293
158
113
4
6 8 10 12 14 16
Volume % H2O (Based On 760 mm Hg Total Pressure)
18
2
) 40
1
60
1
80
1
100
1
120
I
140
H2O Partial Pressure (mm Hg)
Figure 2. Effect of Flue Gas Moisture Content and Temperature
on H2 SO4 Concentration in Condensate
3-52
-------
Another reheat method increases the temperature of the scrubbed gas
without changing the moisture content. The effect of this type of reheat
system is easily displayed on Figure 2. Consider the example case in which
15 percent moisture and 130 degrees F at the absorber outlet result in acid
condensate that is 10 percent H2SO|| by weight. If a reheat system is used to
provide a 30 degree F increase in flue gas temperature, the acid
concentration in the aerosol droplets and/or the ductwork condensate film
will increase to nearly 50 percent f^SOij by weight. On this basis it is easy
to understand why corrosion of reheater tubes is such a common problem.
SELECTIVE ACID DEPOSITION TEST METHOD
The second method for characterization of the FGD ductwork environment
is a flue gas testing procedure that Burns & McDonnell calls the Selective
Acid Deposition (SAD) Test Method. This test procedure is designed to
provide positive information on the composition of the material deposited on
the surfaces of ductwork during operation of FGD systems. Although it is
much more complex and difficult to use than the predictive technique method
it has several significant advantages over that method.
• SAD testing can identify the effects of fly ash on the composition of
the duct surface deposits. (The predictive technique is based on
data from pure two-component mixtures of acid and water under
laboratory conditions.)
• SAD testing can detect the presence of an acid condensate that was
not known or expected to occur in the duct deposit. (The predictive
technique method requires an assumption that the acid in question is
present and that the temperature under consideration is below the dew
point for that acid.)
• SAD testing can provide a complete analysis of the chemical
composition of the duct surface deposit. (The predictive technique
is limited to use for those acid solutions for which laboratory vapor
pressure data are available.)
Physical Principle
FGD system ductwork deposits can form by a number of different
mechanisms. Acids, most commonly sulfuric acid, can condense from the flue
gas onto the duct surface. Particulate can adhere directly to the duct
surface or be trapped by the acid condensate film. Gas molecules and liquid
aerosol droplets can be adsorbed onto the duct surface or onto the surfaces
of the particulate that collects there. Gases may also be absorbed into the
condensate film. The SAD sampling apparatus is designed so that all of these
deposition mechanisms can occur in the sampling train just as they do in the
ductwork.
*
In an operating FGD system, the interior surface of the ductwork is not
at the same temperature as the flue gas stream. Temperature gradients result
from convective and radiative heat transfer from the gas to the duct wall,
through the insulation (if any), and ultimately to the outside air. The
3-53
-------
interior surface of the duct thus may be considerably cooler than the flue
gas. Additional "cold spots" result from conductive heat transfer at
locations where structural support members penetrate the insulation.
By allowing simulation of these heat transfer phenomena the SAD test
method avoids one of the primary faults of the spool rack coupon corrosion
test method. Because the coupons on a spool rack are thermally isolated from
the duct surface, they are effectively maintained at the same temperature as
the gas stream. This inability of the spool rack to simulate the actual
conditions at the duct wall is believed to be the major cause of the
discrepancy between performance of test materials and duct construction
materials at R. D. Morrow with respect to corrosion resistance.
Description of Test Method
The sampling train used for SAD flue gas testing is displayed in Figure
3. The sample is extracted isokinetically from the gas stream using a probe
with a nozzle of appropriate diameter. Simulated duct deposits are collected
by condensation and other mechanisms in the condensate trap. The
thermostatically controlled oil bath maintains the condensate trap at the
desired temperature. The impinger train is used to cool and dry the sample
gas stream to protect the vacuum pump and the control module, which provides
monitoring of sampling flows, pressures and temperatures.
After collection, samples are recovered from the condensate trap by
washing with deionized water. The samples are tested for a total of 37
chemical species by ion chromatography, inductively coupled argon plasma
emission spectroscopy, and pH electrode. Constituent concentrations are
reduced to terms of weight percent on a dry basis for comparison and
analysis.
SAD TESTING AT R. D. MORROW
TEST PROGRAM
To provide an evaluation of the SAD test method for characterization of
FGD ductwork acid deposition, two separate test programs were conducted at
the R. D. Morrow Station during 1982. These test programs were conducted as
part of EPRI Research Project RP 1871-4. The objective of the test programs
also included investigation of the effects of FGD system operating conditions
and duct surface temperatures on the composition of the ductwork deposits.
FGD system bypass (as a percentage of total gas flow) was varied from
zero percent (full scrubbing) to 100 percent (full bypass) in increments of
20 percent. Simulated duct surface temperatures were varied from 20 degrees
F below the flue gas temperature in the outlet ductwork mixing zone to as low
as 100 degrees F. Test conditions are summarized in Table 1.
3-54
-------
Tygon
Sampling
Hose
Vacuum Pump
Temperature Sensor
S-Type Pilot Tube
Nozzle
Figure 3. Sampling Train for SAD Test Method
3-55
-------
TABLE 1. SAD TEST CONDITIONS AT R. D. MORROW
Percentage of
Flue Gas Scrubbed
100
80
60
40
20
0
Mixing Zone
Gas Temp. (°F)
130*
160
200-240
245
280
300-320
Condensate Trap
Temperatures (°F)
130, 120, 100
140, 120, 100
220, 200, 180, 160, 140, 120, 100
225, 205, 185, 165
260, 240, 220, 200
300, 280, 260, 240, 220, 180, 140, 100
*Samples collected at absorber outlet. All other samples collected in stack.
Simulated duct surface temperatures in the range from 20 degrees F to 60
degrees F below the flue gas temperature represent conditions that could
occur during steady-state FGD system operation, depending upon the gas
temperature, the ambient temperature, and the thickness and condition of the
ductwork insulation. The other duct surface temperatures simulated represent
transient conditions that could occur only during start-up.
TEST RESULTS
The SAD testing programs at R. D. Morrow generally confirmed that the
test method can produce repeatable results with a reasonable degree of
accuracy. The test method succeeded in collecting and identifying the
constituents of the material deposited on a simulated duct surface during FGD
system operation.
Simulated duct deposit samples from the R. D. Morrow FGD system were
found to consist primarily of six constituents (Table 2).
TABLE 2. SAD TESTING AT R. D. MORROW - SAMPLE COMPOSITION
Sample
Constituent
Sulfate
Chloride
Fluoride
Aluminum
Iron
Silicon
Weight Percent in Sample (Dry Basis)
Test 1 Test 2 All Samples
Mean S.D.*
(N=19)~
62,
13-
2.
6,
4.
1.
2
5
50
10
41
44
90.2
25.0
22.1
3.20
2.97
3-18
1.10
54.6
13.6
2.85
10.9
4.64
3-90
90.5
Mean S.D.
"(N=15)
19.6
21.7
3-77
4.02
2.46
2.26
Mean
58.9
13.5
2.65
8.23
4.51
S.D.
22.
21.
3.
4.
2.
.7
.6
.41
.20
.84
90.3
2.09
^Standard Deviation.
3-56
-------
The data presented In Table 2 support the following conclusions
regarding the composition of the duct surface deposits.
• Sulfate is the dominant constituent of the deposits. Based upon its
predominance compared to the metallic constituents it is clear that
the primary source of the sulfate in the samples is sulfuric acid.
• The chloride and fluoride contents of the duct surface deposits are
highly variable. (Reasons for this variability are addressed later.)
• The ash contents of the duct deposit samples from Test 2 are
significantly higher than those from Test 1. This result can be
explained from stack opacity levels, which were consistently much
higher during Test 2 than during Test 1.
Other sample constituents observed frequently at levels of 1.0 percent
or greater by weight are potassium, calcium, phosphorous, sodium, selenium
and magnesium.
Effects of Mixing Scrubbed and Unscrubbed Flue Gas
The R. D. Morrow FGD systems were designed for partial bypass of the
flue gas around the absorber. This design reduced the size of the absorber
required and provided for reheat of the scrubbed gas. The mixing of scrubbed
and unscrubbed flue gas in the outlet ductwork had been suspected as one
cause of the severe corrosion of the ductwork construction materials in this
zone.
SAD test results from R. D. Morrow confirm that mixing of scrubbed and
unscrubbed flue gas has a dramatic effect on duct surface deposit
characteristics (Table 3)-
TABLE 3. EFFECT OF MIXED GAS RATIO
Mixed' Gas Ratio
Scrubbed/Unscrubbed
100/0
80/20
Sample
Constituent
Chloride
Fluoride
Chloride
Fluoride
Weight Percent in Sample
Dry Basis
5.70
0.26
66.0
5.43
Based upon these test results it can be concluded that:
• Absorber removal efficiencies for HC1 and HF are very high, possibly
98 percent or above.
• Even very small portions of bypassed gas mixed with scrubbed gas can
result in order-of-magnitude increases in the halide content of the
duct deposits. This could explain the occurrence of severe pitting
corrosion of alloy materials in the mixing zone of the R. D. Morrow
FGD outlet ductwork.
3-57
-------
Effect of Condenser Temperature
Analysis of vapor-liquid equilibrium data for sulfuric acid solutions
indicates that variations in the SAD condensate trap temperature should
produce corresponding changes in the concentration of f^SOij in the
condensate. However, the SAD test method as used at R. D. Morrow cannot
detect changes in acid condensate concentration.
The condensate forms as a thin film on the surface of the condenser
tubing. It was not possible at R. D. Morrow, in most cases, to determine the
mass or volume of the condensate. Samples were recovered by washing the
condenser with deionized water. Because the resultant constituent
concentrations could not be related to "as condensed" acid solution
concentrations, all data was reduced to a dry basis for analysis. Thus, for
example, it was not possible to differentiate between a sample that condensed
as a 50 percent F^SOij solution and one that condensed as a 75 percent H2SOij
solution.
Despite this limitation, SAD test results from R. D. Morrow did provide
information on the effects of condenser temperature on sample composition.
Chloride and fluoride compositions of the samples were found to be a strong
function of condenser temperature. Chloride composition of the samples was
observed to increase sharply below 140 degrees F. Fluoride composition
followed a similar, although more moderate, trend at temperatures below 165
degrees F. The data supporting these conclusions are summarized in Table 4.
Data for the full scrubbing case have been excluded from this analysis to
isolate the temperature effect.
TABLE 4. EFFECT OF SIMULATED DUCT SURFACE TEMPERATURE
Sample Temperature Weight Percent in Sample
Constituent Range (°F) Dry Basis
Chloride Less than 140 51.6
140 and above 3.41
Fluoride Less than 165 6.38
165 and above 0.532
The fact that both chloride and fluoride exhibit such a marked
dependence on condenser temperature indicates that the primary source of each
constituent is a vapor that condenses in the sampling train. The occurrence
of nonzero values for the compositions of these constituents at the higher
temperatures reflects the fact that they also exist in the flue gas in solid
form as components of the fly ash.
CONCLUSIONS AND RECOMMENDATIONS
Two methods for characterization of the corrosive environment in FGD
system ductwork have been developed by Burns & McDonnell. Each method can
3-58
-------
provide important information on the composition of the acidic material that
collects on duct surfaces by condensation and other mechanisms.
The predictive technique can be used in the planning and design of new
FGD systems to identify the concentrations of acid solutions that will
condense on ductwork surfaces. This information should be used to evaluate
the acceptability of alternative corrosion-resistant materials for duct
construction and/or lining.
For operating FGD systems that are experiencing corrosion problems in
ductwork, the SAD test method can be used to provide complete chemical
characterization of the material that is deposited on the duct surfaces
during operation. The test results should be used to define the corrosive
environment to which ductwork construction materials and/or linings are being
exposed. This environment can then be duplicated in laboratory corrosion
tests to identify the proper corrosion-resistant material required for
solution of the problem.
*****
3-59
-------
IN SITU EVALUATION OF HIGH PERFORMANCE ALLOYS IN
POWER PLANT FLUE GAS DESULFURIZATION SCRUBBERS
R. W. Schutz, C. S. Young
-------
IN SITU EVALUATION OF HIGH PERFORMANCE ALLOYS
IN POWER PLANT FLUE GAS DESULFURIZATION SCRUBBERS
R. W. Schutz* and C. S. Young**
*TIMET Division, Titanium Metals
Corp. of America, Henderson, NV 89015
**Astro Metallurgical, Division of
HARSCO Corporation, Wooster, OH 44691
ABSTRACT
Preliminary data from ongoing in situ FGD scrubber exposure tests,
involving specific high performance stainless steels, nickel-base, and
titanium alloys are presented and discussed. Spool rack exposures of
4-9 month duration involving candidate alloys, Titanium Grades 2 and
12, 904L stainless steel and nickel-base Alloys 625 and C-276, were
conducted in the inlet quench and outlet duct zones of several operating
power plant FGD scrubber systems. In general, the titanium alloys
exhibited performance equal to or superior to the steel and nickel
alloys, particularly in outlet ducts. The 904L alloy consistently
exhibited poor resistance to localized attack, while 625 and C-276
alloy performance was varied. These findings are discussed relative
to specific environmental considerations, and parallel laboratory and
field studies reported in the literature.
INTRODUCTION
The flue gas desulfurization (FGD) scrubber industry is in a state
of flux, not only in terms of changes in process technology, but also
relative to the materials of construction used in the systems. In
the early 1970's when the FGD industry was beginning, carbon steel
and low alloy austenitic stainless steels were used as materials of
construction in what was considered to be a mild environment. As time
progressed, these materials were found to be unsatisfactory, and highly
alloyed austenitic stainless steels and some nickel-base alloys became
the accepted construction materials. However, these alloys still have
their limits, and some FGD systems have experienced such severe cor-
rosion in recent years that more corrosion resistant alloys are now
being considered(0 to limit costly downtime and maintenance.
FGD scrubber systems are composed of a variety of equipment^2)
from prescrubbers and absorber vessels to dampers, ducts, stacks, and
3-61
-------
piping systems. However, the major material problem areas are the wet/dry
interfaces at the inlet quench (pre-scrubber) and the mixing/reheat
zone.(3) Additionally, the entire outlet duct, from the mist elimina-
tors to the stack, can be subjected to severely corrosive conditions
of low pH, high chloride levels, and relatively high (or fluctuating)
temperatures. Outlet duct problems are also associated with acidic
gases near or below the dew point. The test program reported on in
this paper is concentrated in these areas.
Several laboratory and in situ programs^1*"13»19 >22 >23' for evalu-
ating resistant scrubber materials, in addition to actual service ex-
perience, (14~18) have been reported in the literature. Limitations
in performance of highly alloyed stainless steel and nickel alloys
are noted, particularly in closed-loop systems. Laboratory studies
suggest that acidic chlorides, in conjunction with possible fluoride
interactions,(18'19) are primarily responsible for limitations in pit-
ting and crevice corrosion resistance of the high performance alloys
evaluated to date.
The relatively few evaluations conducted in the past to identify
severe inlet and outlet duct corrosion problems have not included all
candidate metals and alloys (e.g. Alloys 904L, 625, C-276, and Titanium
Grades 2 and 12), or have involved conditions not directly representa-
tive of actual scrubber environments. In particular, titanium alloys,
noted for their resistance to hot chloride environments, have been
evaluated only to a very limited extent.
Early in situ tests in an FGD scrubber and a copper smelting lime-
stone scrubbed20) indicated that unalloyed titanium, C-276 and 625
alloys, were the most resistant materials, exhibiting no measurable
attack. Similar results were reported in several municipal refuse
incinerator flue gas scrubbers(2:) and for simulated outlet duct ex-
posures. (22) Recent laboratory studies, in fact, suggest that titanium's
potential suitability in FGD scrubber environments may be due to in-
hibiting effects of flyash.(23'
The major factor limiting consideration of titanium and other
resistant alloys has been material cost. Table 1 summarizes relative
material costs, normalized per unit surface area, of candidate corro-
sion-resistant alloys for problem areas in FGD scrubbers. Clearly,
titanium represents a cost-effective alternative to highly alloyed
ferrous and nickel alloys and,thus, should be considered as an alter-
nate material of construction.
The spool rack test program discussed in this report was performed
to provide a comprehensive in situ evaluation of the corrosion resistance
of potentially viable high-performance alloys, including titanium.
Because the alloys were tested in the same FGD scrubber locations,
direct comparison of their corrosion resistance is possible, allowing
for possible ranking of materials. Additionally, the range of alloys
and the variety of environments tested should allow for a more complete
comparison with previous in situ scrubber experience and laboratory
3-62
-------
TABLE 1. COST COMPARISONS (PER UNIT SURFACE AREA)
OF CANDIDATE ALLOYS FOR CORROSIVE FGD SCRUBBER SERVICE
Alloy
904L"
825
Ti-Grade 2
Ti-Grade 12
G
625
Ti-Grade 7
C-276
Density
(Ib/cu in)
0.
0.
0.
0.
0.
0.
0.
0.
289
294
163
163
300
305
163
321_
Weight
(Ib/sq ft)
10.
10.
5.
5.
10.
11.
5.
11.
5
6
9
9
8
0
9
6
Relative Alloy Cost*
(Per Unit Surface Area)
0
0
1
1
1
1
1
2
.52
.99
.00
.20
.23
.86
.89
.11
Relative Total Cost**
Including Fabrication
(Per Unit Surface Area)
0
0
1
1
1
1
1
1
.60
.94
.00
.13
.09
.52
.63
.69
*Based on mill quantity quotes for 6.35-mm plate product as of May 1983.
**Assumes a 2:1 contribution of material to fabrication cost for overall cost.
3-63
-------
results.
EXPERIMENTAL
ALLOYS EXPOSED
High-performance alloys evaluated in this investigation included
C.P. Grade 2 titanium, Grade 12 titanium (Ti-Code 12), Alloy C-276,
Alloy 625, and Alloy 904L. Stainless steel 20Cb-3 and 316L were also
evaluated in one scrubber. Nominal compositions of these alloys are
presented in Table 2. The Grade 2 titanium represents common unalloyed
titanium, whereas Grade 12 titanium includes alloy additions of 0.3%
Mo and 0.8% Ni, which provide improved crevice corrosion resistance
compared to unalloyed titanium grades. Alloys C-276 and 625 are highly
alloyed nickel alloys containing molybdenum contents of 16 and 9 weight
percent, respectively, that possess relatively good resistance to local-
ized chloride attack. The 904L and 20Cb-3 alloys are super-austenitic
stainless steels.
All alloys were tested as 3.18-mm (0.125-in) sheet product in
the mill-annealed condition.
SPECIMEN PREPARATION/EVALUATION
Specimens consisted of 50.8-mm (2-in) square 3.18-mm (0.125-in)
sheet, containing a 7.9-mm (,31-in) diameter center hole for mounting.
The two titanium alloy specimens were prepared by a light pickle in
35 vol%HN03~5 vol%HF solution, rinsing in distilled water, and drying
before weighing. The nickel and stainless steel alloys were abraded
dry with 120 grit alumina paper, followed by a distilled water rinse.
Specimen weights were measured to one-tenth of a milligram.
Weldments of each alloy were prepared by standard (argon-shielded)
autogenous TIG weld procedures. For each welded specimen, one weld
pass was performed on a single specimen face, with full specimen thick-
ness penetration.
Corrosion coupons were mounted onto an unalloyed titanium spool
rack, such that three specimens of each titanium alloy and two specimens
of each of the C-276, 625, and 904L alloys were included. One coupon
of each alloy consisted of a welded specimen. Figure 1 illustrates
a fully assembled spool rack ready for exposure.
Flat Teflon® washers, 9.53-mm (,375-in) thick, provided coupon
separation in addition to a 14.8-mm2 (0.23-in2) effective crevice surface
area on every coupon face. TeflonP insulating sleeves were used to
prevent spool rack or intercoupon galvanic interactions.
Post-exposure evaluation included specimen weighing, microscopic
examination at 10-30x, and maximum pit depth measurement after thorough
specimen cleaning. Removal of tenacious solid surface deposits was
difficult via conventional inhibited-acid cleaning techniques.
3-64
-------
TABLE 2. NOMINAL COMPOSITIONS OF HIGH
PERFORMANCE ALLOYS EVALUATED IN IN SITU SCRUBBER EXPOSURES
Weight %
Grade
Grade
Alloy
Alloy
Alloy
Alloy
Alloy
Allov
2 Titanium
12 Titanium
C-276
625
904L
20Cb-3
316L
Ti Ni
Bal
Bal 0.80
Bar
Bal
— 25.0
— 34.0
— 12.0
Fe
0.12
0.12
5.0
2.5'
Bal
Bal
Bal
Cr
-
-
16
21
19
20
17
-
-
.0
.5
.5
.0
.0
Mo
0
16
9
4
2
2
—
.30
.0
.0
.5
.5
.0
W Cu C
— .02
— — .02
4.0 — .01
— .05
— 1.5 .02
— 3.5 .02
— — .03
Others
.01N, 0.1202
.01N, 0.1202
0.3V
3.5 Cb+Ta
--
IMn, .5Cb, .5Si
2Mn, ISi
Figure 1. Fully assembled corrosion test spool rack for in situ scrubber
exposure evaluation of alloys.
3-65
-------
Specimen cleaning utilizing a brass wire brush and a warm detergent
solution proved successful in this regard. Pit depths were measured
to the nearest 0.025-mm (0.001-in), using a depth-measuring microscope
by focusing on the edge and bottom of each pit.
IN SITU SCRUBBER EXPOSURE SITES
Spool racks were installed in FGD scrubbers of the following four
power plants with the cooperation of power plant personnel: South
Mississippi Electric Power-R.D. Morrow Sr. Power Plant, Indianapolis
Power and Light - Petersburg Station, Allegheny Power Service Corpora-
tion - Pleasants Station, and Potomac Electric Power - Dickerson Station.
FGD scrubbers associated with the first three power plants are lime-
stone slurry absorption systems for flue gas S02 removal. The R. D.
Morrow scrubber represents the only closed-loop system evaluated, gen-
erally providing the most aggressive conditions, resulting from high
chloride and fluoride levels in the recycled slurry. Outlet duct gas
reheat is achieved with bypassed flue gas for all three limestone scrub-
ber systems.
The PEPCO-Dickerson Station utilizes a variable-throat venturi
particulate scrubber for direct flyash removal and partial S02 absorp-
tion. The scrubbing media is partially recycled water, resulting in
low media pH's relative to the buffered limestone slurry systems.
The most corrosive areas of the inlet and outlet scrubber ducting,
based on previous individual scrubber experience, were chosen for spool
rack exposure. In this regard, the inlet quench duct (wet/dry gas
interface zone), mixing zone, and breech duct were selected at the
R. D. Morrow Sr. Plant. At the Petersburg Station, the presaturator
(water spray quench) and mixing zones provided potentially aggressive
environments for evaluation. Spool racks were attached to a duct expan-
sion joint shield between the mixing and breech duct zones at the Pleas-
ants Station. Among the most aggressive areas of the Dickerson Station's
particulate scrubber was the vapor zone at the base of the stack (above
the sump). Conditions associated with each of these specific exposure
sites are outlined in Table 3.
A pair of (identical) spool racks were installed in each desig-
nated scrubber location. This allowed four to seven month, and eight
to twelve month exposure data to be collected. With the exception of
R. D. Morrow Sr. Plant exposures, to date, only four to seven month
data has been obtained for discussion in this report.
RESULTS OF IN SITU SCRUBBER EXPOSURES
R. D. MORROW SR. PLANT CORROSION DATA
Corrosion data generated from R. D. Morrow's Scrubber No. 1 mixing
zone after five and nine-month exposures are presented in Table 4.
Weight loss values clearly indicate no measurable corrosion on Titanium
Grades 2 and 12, whereas significant attack is evident for Alloys 625,
3-66
-------
TABLE 3. EXPOSURE CONDITIONS FOR
INDIVIDUAL SCRUBBER DUCT ZONES EVALUATED
Power Plant
Scrubber Zone
R. D. Morrow Sr. Plant Inlet quench duct
(Scrubber //I)
Mixing Zone
(Scrubber #1)
Breech Duct
(Scrubber ill)
Petersburg Station Inlet Presaturator
(D-Module)
Mixing Zone
(3-2 Scrubber)
Temp
Chlorides
Other Factors
(°C) (ppm)
52-177 >15,000 1.0-5.A Wet/dry gas interface zone
Pleasants Station
Dickerson Station
52-177 3,100-10,000 1.3-2.4 "
106-115 3,000-10,000 1-3 — —
54-149 >3,000 1.0-3.5 Water spray quench
66-149 — — Wet gas 70% of exposure
Mixing/Breech Duct Zone 52-149 -- — — —
Base of Stack
82
1,270
1.6-2.2 Wet gas above stack sump
TABLE 4. R. D. MORROW SR. POWER PLANT
FGD SCRUBBER NO. 1 - MIXING ZONE CORROSION DATA
Alloy Exposed
Grade 2 Titanium
Grade 12 Titanium
Alloy C-276
Alloy 625
Alloy 904L
Specimen Type Weight Loss (A mg) Max Pit Depth* (Mils)
Base
Base
Welded
Base
Base
Welded
Base
Welded
Base
Welded
Base
Welded
5 Mo.
Nil
Nil
Nil
Nil
Nil
Nil
32
61
186
357
1724
1940
(9Mo.)
( Nil)
( Nil)
( Nil)
( Nil)
( Nil)
( Nil)
( 76)
( 86)
( 374)
( 557)
(2708)
(2844)
Avg 5 Mo-
Nil (Nil) j- j
0, 0
0, 0
0, 0
0, 0
4, 2
19**, 5**, 3
1,3**,1
29,21
— Perforated
54,93
(9 Mo.)
(0, 0)
(0, 0)
(0, 0)
(0, 0)
(0, 0)
(0, 0)
( 16**, 0)
(35**, 0)
(70,4**)
(38**,43**,1)
(Perforated)
(Perforated)
*Maximum pit depth data given for both specimen faces.
**Pit formed in Teflon®-metal crevice.
3-67
-------
C-276, and 904L. The extent of attack on the nickel and 904L alloys
appears to increase non-linearly with exposure period.
Correspondingly, the pit depth data reveals no localized attack
on any titanium coupons (Figures 2 and 3), with severe pitting occur-
ring on Alloys C-276, 625 and 904L specimens. In fact, Alloy 904L
samples exhibited extensive perforation from pitting, in addition to
shallow surface craters resulting from deposit crevice corrosion (Fig-
ure 4). The majority of the weight loss observed on C-276 specimens
is attributable to Teflon@-metal crevice corrosion. Crevice attack
on the two nickel alloys generated pit depths as large as 30-40 mils
(Figures 5 and 6). The 625 alloy also displayed 20-30 mil pits of
small diameter-to-depth ratios, independent of Teflon^-to-metal crevices.
Breech duct corrosion data is outlined in Table 5. Again, titanium
alloys experienced near nil corrosion, demonstrated by very low weight
loss and no pitting. The C-276 alloy coupons similarly displayed in-
significant weight loss, but did indicate incipient pitting and Teflon®-
metal crevice attack, penetrating 1-3 mils deep. The 904L and 625
alloys revealed pit depths of up to 8 mils, with associated increases
in net weight loss. Incipient crevice attack of Alloy 625 in TeflorP-
metal crevices was also noted.
Compared to the mixing zone, the breech duct proved to be signi-
ficantly less aggressive to nickel and stainless steel alloys. The
trend in relative overall corrosion resistance, however, is consistent
in both outlet duct exposures after five and nine months:
Titanium Grade 2, 12 > C-276 > 625 > 904L
None of the high-performance alloys exposed to the wet/dry gas
zone of the inlet quench duct were immune to attack. Table 6 shows
that, contrary to outlet duct performance, both Titanium Grades 2 and
12 displayed extensive localized attack considered to be deposit crevice
corrosion (Figure 7). The C-276 and 625 alloys exhibited very shallow
surface pits ranging in depths to 6 mils. Enlarged views of this at-
tack are illustrated in Figures 8 and 9. In this zone, the nickel-
base alloys proved to be more resistant to localized attack. Suscep-
tibility to Teflon®-metal crevice attack for all alloys was somewhat
diminished in this zone compared to the mixing zone.
PETERSBURG STATION CORROSION DATA
At the Petersburg Station, both outlet and duct mixing zone and
inlet presaturator environments were evaluated. The mixing zone (Table
7) proved to be very benign to all alloys tested including 316L, 904L,
and 20Cb-3 stainless steels. This appears to reflect low chlorides
in the scrubber outlet gas, and spool rack positioning relative to
the mixing zone wet/dry gas interface.
Conditions in the water spray D-Module presaturator provided a
more discriminating test of alloys (Table 8). After seven months,
3-68
-------
Figure 2. Grade 2 titanium welded coupon (as-cleaned) after 9-month
exposure in the mixing zone of R. D. Morrow Sr. Scrubber
No. 1.
3-69
-------
Figure 3. Grade 12 titanium base metal coupon (as-cleaned) after 9-
month exposure in the mixing zone of R. D. Morrow Sr. Scrubber
No. 1.
Figure 4. Alloy 904L base metal coupon (as-cleaned) after 9-month ex-
posure in the mixing zone of R. D. Morrow Sr. Scrubber No. 1
3-70
-------
Figure 5. Crevice corrosion and pitting on Alloy 625 (at 8x) after
9-month exposure in the scrubber mixing zone at the R. D.
Morrow Sr. Station.
Figure 6. Crevice corrosion on Alloy C-276 weld coupon (at 8x) after
9-month exposure to the scrubber mixing zone at the R. D.
Morrow Sr. Station.
3-71
-------
TABLE 5. R. D. MORROW SR. POWER PLANT
FCD SCRUBBER NO. 1 - BREECH DUCT CORROSION DATA
Alloy Exposed
Grade 2 Titanium
Grade 12 Titanium
Alloy C-276
Alloy 625
Alloy 904L
Specimen Type Weight Loss (A rag)
Base
Base
Welded
Base
Base
Welded
Base
Welded
Base
Welded
Base
Welded
5 Mo.
4
3
14
5
1
2
12
17
76
87
140
140
(9 Mo. )
Avg
( 18)
( 14)
( 18)
( 14) , ,2(n
( 25) 3 (20)
( 15)
( 27)
( 35)
( 66)
( 68)
(122)
(120)
Max Pit Depth* (Mils)
5 Mo.
0, 0
0, 0
0, 0
0, 0
0, 0
0, 0
0, 2**
0, 3
1, 2
5,2**,1
5, 3
8, 5
(9
(0,
(0.
(0,
(0,
(0,
(0,
(0,
(3,
(1**
(1**
(1,
(3,
Mo.)
-------
Figure 7. Localized attack on Grade 12 titanium base metal coupon after
9-month exposure to the wet/dry zone of the inlet quench duct
at the R. D. Morrow Sr. Plant.
Figure 8. Shallow surface pitting of Alloy C-276 (at 8x) after 9-month
exposure to the wet/dry zone of the inlet quench duct at
the R. D. Morrow Sr. Plant.
3-73
-------
Figure 9. Shallow surface pitting of Alloy 625 (at 8x) after 9-month
exposure to the wet/dry zone of inlet quench duct at the
R. D. Morrow Sr. Plant.
3-74
-------
TABLE 7. PETERSBURG STATION CORROSION DATA
FGD Scrubber - 3/2 Outlet Mixing Zone
(4.5-Month Exposure)
Alloy Exposed
Grade 2 Titanium
Grade 2 Titanium
Grade 12 Titanium
Grade 12 Titanium
Alloy 625
Alloy 625
Alloy 904L
Alloy 904L
Alloy 20Cb-3
Alloy 20Cb-3
316L Stainless
316L Stainless
Specimen Type
Base
Welded
Base
Welded
Base
Welded
Base
Welded
Base
Welded
Base
Welded
Weight Loss
(mg)
41
40
135
143
170
176
31
31
20
23
46
47
Max Pit Depth*
(mils)
0, 1
1, 1
0, 0
1, 1
1, 2
1, 1
0, 0
0, 0
0, 0
0, 0
0, 0
0, 0
*Maximum pit depth values given for both specimen faces.
TABLE 8. PETERSBURG STATION CORROSION DATA
FGD Scrubber - D Module Presaturator
(7-Month Exposure)
Alloy Exposed
Grade 2 Titanium
Grade 2 Titanium
Grade 12 Titanium
Grade 12 Titanium
Alloy 625
Alloy 625
Alloy 904L
Alloy 904L
Alloy 20Cb-3
Alloy 20Cb-3
316L Stainless
316L Stainless
Specimen Type
Base
Welded
Base
Welded
Base
Welded
Base
Welded
Base
Welded
Base
Welded
Weight Loss
(mg)
0.0
0.0
0.0
0.0
102
189
241
293
363
283
548
345
Max Pit Depth*
(mils)
0, 0
0, 0
1, 1
1, 1
1, 2
4, 1
7, 4
6, 4
3, 4
7, 3
7, 5
5, 7
*Maximum pit depth values given for both specimen faces.
3-75
-------
all stainless steel and nickel-base alloys tested experienced measurable
weight loss, which was generally the result of pitting and some Teflon©-
metal crevice attack. Pit depths ranged from 1-7 mils, with the 625
alloy exhibiting the least weight loss. In contrast, both titanium
alloys demonstrated excellent corrosion resistance with negligible
localized attack.
PLEASANTS STATION CORROSION DATA
After 6.25 months, nickel-base and titanium alloy coupons exposed
to the mixing/breech zone at Pleasants Station FGD scrubber revealed
very mild surface pitting (^ 1-4 mils in depth). As presented in Table
9, pit penetration amounted to almost 15 mils for 904L stainless steel.
The extent of pitting, judged with respect to weight loss, indicates
that both titanium alloys suffered the least degree of overall attack,
followed closely by the 625 alloy. Surprisingly, C-276 coupons suf-
fered similar metal losses as the 904L coupons, apparently caused by
additional general corrosion of the nickel alloy.
DICKERSON STATION CORROSION DATA
Table 10 outlines data collected after a 7-month exposure at the
base of the particulate scrubber stack. Clearly, Titanium Grades 2
and 12 proved to be totally resistant. The 625 and 904L alloys, how-
ever, experienced severe pitting and some Teflon®-metal crevice attack.
Pits penetrated almost one-third of the thickness of sheet samples.
The C-276 alloy exhibited significant weight loss in the form of shal-
low surface pits associated with apparent deposit and gasket-metal
crevice corrosion. Visual examples of this diverse corrosion behavior
of high-performance alloys are presented in Figure 10.
SUMMARY OF HIGH-PERFORMANCE ALLOY CORROSION RESISTANCE
IN SCRUBBER OUTLET DUCTS
A consistent trend in relative alloy corrosion resistance is appar-
ent among all power plant scrubber outlet duct exposures evaluated
in this study. Based on coupon weight loss, maximum pit depth, and
Teflon®-metal crevice attack, the ranking of alloy resistance is:
Titanium Grades 2 and 12 > C-276 > 625 > 904L
Both titanium alloys were immune to Teflon®-metal crevice attack,
while resisting pitting in all mixing zone and breech duct exposures.
The C-276 alloy was subject to finite crevice corrosion and pitting
attack in several outlet duct areas, and was generally much more re-
sistant than the 625 alloy. This was particularly evident in the closed-
loop scrubber mixing zone at the R. D. Morrow Sr. Station. Heavy pit-
ting was characteristic of the 904L alloy in most outlet duct areas
tested. The venturi particulate scrubber stack exposure at the Dickerson
Plant produced exceptionally high rates of localized attack on these
nickel and stainless alloys. The two open loop FGD scrubber outlet
duct exposures at the Petersburg and Pleasants Stations, produced less
3-76
-------
TABLE 9. PLEASANTS STATION CORROSION DATA
Alloy Exposed
FGD Scrubber - Mixing Zone
(6.25-Month Exposure)
Specimen Type Weight Loss
Avg
Max Pit Depth*
(mils)
Grade 2 Titanium
Grade 2 Titanium
Grade 2 Titanium
Grade 12 Titanium
Grade 12 Titanium
Grade 12 Titanium
Alloy C-276
Alloy C-276
Alloy 625
Alloy 625
Alloy 904L
Alloy 904L
Base
Base
Welded
Base
Base
Welded
Base
Welded
Base
Welded
Base
Welded
554 557
559
567
549 568
587
528
1,424
1,351
613
619
1,243
1,145
1.9,
1.9,
4.1,
1.9,
3.5,
4.1,
1.3,
4.8,
1.7,
1.3,
9.3,
14.6,
1.5
3.0
2.2
2.0
3.0
2.0
1.3
2.8
1.7
1.9
11.7
5.2
*Maximum pit depth values given for both specimen faces
TABLE 10. DICKERSON STATION
PARTICULATE FLUE GAS SCRUBBER CORROSION DATA
Base of Stack Attached to Drain
(7-Month Exposure)
Alloy Exposed
Grade
Grade
Grade
Grade
Grade
Grade
Alloy
Alloy
Alloy
Alloy
Alloy
Alloy
2 Titanium
2 Titanium
2 Titanium
12 Titanium
12 Titanium
12 Titanium
C-276
C-276
625
625
904-L
904-L
Specimen Type
Base
Base
Welded
Base
Base
Welded
Base
Welded
Base
Welded
Base
Welded
Weight Loss
(mg)
0.
0.
0.
0.
0.
1.
685
461
2,508
2,611
3,827
4,802
Max Pit
Depth*
(mils)
0
0
0
0
0
0
0,
0,
0,
0,
0,
o,
8,
14,
39,
33,
31,
29,
0
0
0
0
0
0
3
5
34
22
32
34
*Maximum pit depth data given for both specimen faces.
3-77
-------
Figure 10. Comparison of various high-performance alloy coupons exposed
at the base of the scrubber stack at the Dickerson Station.
3-78
-------
distinction among high-performance alloys tested, probably the result
of lower operating chloride levels.
DISCUSSION
TITANIUM ALLOY PERFORMANCE IN SCRUBBER OUTLET DUCTS
Titanium's apparent excellent resistance to FGD scrubber environ-
ments has been reported in a few previous field studies, and is, thus,
not totally unexpected. Tice^20' included C.P. titanium in a spool
rack exposure program in three types of FGD limestone slurry scrubbers.
These exposures included absorber and outlet duct zones in flooded disc,
venturi-absorber, and smelter gas scrubbers. In all exposures, including
reheat zones, titanium exhibited no evidence of general or localized
attack.
Rice and Burford^21' evaluated titanium in various municipal refuse
incinerator flue gas scrubbers, which employ recycled water as the
absorbing media. Exposures included zones before and after the absorber
mist eliminator, including outlet duct and stack. Titanium proved to
be totally immune in all tests and was judged to be the most reliable
material evaluated for outlet duct/stack service. C-276 and 625 alloys
both experienced pitting and crevice attack in several outlet duct
exposures.
Consideration of the acidic oxidizing chloride nature of scrubber
duct condensate as the basis for corrosive attack may well explain the
observed behavior of titanium in these tests. Titanium is known to
be extremely resistant to general corrosion and pitting in aqueous
chlorides over a wide pH range and to very high temperatures.(24>25)
In chloride brines, for example, titanium has anodic pitting potentials
in excess of +1 volt (vs. Ag/AgCl reference) to temperatures greater
than 200°c(26>27>28) and, thus, would not be expected to spontaneously
pit under anticipated scrubber conditions. Exposures of titanium alloys
to boiling ferric chloride, cupric chloride, Oa or Cla-saturated brine,
and other oxidizing chloride solutions similarly reveals excellent
resistance to general and pitting attack.(25>29) Dissolved oxidizing
species (cathodic depolarizers) are well known to provide passivation
of titanium in normally aggressive reducing acid chloride and sulfate
environments.(3°)
Laboratory studies performed by Koch and Beavers(22) in a simulated
outlet duct environment correlates with in situ results with titanium
alloys. In this aerated "wet fog" test containing 5,000 ppm chlorides,
Titanium Grades 2 and 12 experienced no general, crevice or pitting
corrosion after the 1,000-hour test period. Thomas and Bomberger(23)
similarly investigated resistance of several titanium alloys to boiling
simulated scrubber solutions containing a wide range of chlorides.
Corrosion resistance of these alloys proved to be excellent to fairly
low pH's, often exceeding resistance of the highly alloyed nickel-base
alloys studied.
3-79
-------
Attempts by Battelle researchers to simulate scrubber outlet duct
media in laboratory tests(12) strongly suggest that the actual complex
scrubber duct environment cannot be adequately represented as a basic
mixture of HC1, CaCla, H2SOit, and fluorides. This is indicated by
the very high general corrosion rates reported for stainless, nickel
and titanium alloys in the referenced study, which are not representa-
tive of field experience with scrubber duct materials. Since significant
levels of chlorides and fluorides are present in both outlet duct con-
densate and ducting deposits,(18) additional factors are obviously
involved that prevent corrosion of titanium under these acidic condi-
tions. This is especially apparent considering titanium's lack of
resistance to pure, hot HF and/or HC1 solutions.
The presence of flyash in simulated scrubber liquors has been
demonstrated to be a significant factor in explaining titanium's re-
sistance in aggressive, hot, acidic chloride/fluoride environments.
In this regard, Thomas and BombergerC23) reported that small additions
of flyash inhibit attack of various titanium alloys to much lower pH,
particularly when fluorides are present. Explanations of this behavior
are based on the silicon, aluminum, and calcium content of the flyash.
It is theorized that these species reduce free fluoride ion activity
by combination to form insoluble compounds or soluble complexes.
Titanium's resistance to deposit crevice corrosion, evident in
field outlet duct exposures, also requires explanation. Susceptibility
of titanium to crevice attack in aqueous chloride media increases with
decreasing pH and increasing temperature, as shown in Figure 11. (25)
From this figure, one would infer the probability of occurrence of
C.P. titanium crevice corrosion under deposits in these low pH chloride
exposures. However, the major presence of iron in both flyashA23'
and ducting depositsO8) may be a key factor.
The high oxygen partial pressure in the flue gas would insure oxida-
tion of any ferrous species to the ferric (Fe+3) form. This ferric
ion, as a strong cathodic depolarizer in acidic solutions,'25'' stabi-
lizes the oxide film on titanium, rendering it immune to attack from
both acidic condensate and surface deposits. Since the ferric compound
is an integral part of ducting deposits, under-deposit attackC29) can-
not initiate. Further laboratory studies are necessary to substantiate
these explanations.
In summary, the excellent performance of Titanium Grades 2 and
12 in the scrubber outlet ducts evaluated appears to be consistent
with the oxidizing nature of ducting liquors and solids. The signifi-
cant presence of oxygen and ferric species may explain passivation
of titanium in hot acid ducting condensates, with flyash components
possibly providing inhibition to fluoride attack.
PERFORMANCE OF HIGHLY ALLOYED STAINLESS AND
NICKEL ALLOYS IN SCRUBBER OUTLET DUCTS
The nature of localized attack observed on the 904L, 625, and C-276
3-80
-------
alloys evaluated in these outlet duct field tests correlates very well
with parallel in situ studies and actual service experience. This
is, perhaps, most apparent in reported R. D. Morrow Sr. closed-loop
scrubber exposures where chlorides, at levels an order of magnitude
greater than in single pass systems, accentuate marginal differences
between high performance alloys.
At the R. D. Morrow station, reported operating experience with
Hastelloy G scrubber ducting revealed extensive pitting throughout
mixing and breech duct zones.(3>9>14>16»1 a ) Test sheet panels of
625 and C-276 alloys in the mixing zone were reported to have incurred
pitting(9) and crevice corrosion,^14) respectively. Spool rack data
generated in this zone indicate weld metal pitting of Alloy 625, with
heavy pitting of the 904L alloy.(9) Maximum pit depths of 5-7 mils
are similarly reported for 625 and 904L alloy coupons in an outlet duct
exposure near the stack, d8) These findings, similar to the data pre-
sented herein, suggest a parallel ranking of stainless/nickel alloy
corrosion resistance based on the (molybdenum + chromium) alloy equiva-
lence.
Significant localized attack of high performance nickel/stainless
alloys in scrubber outlet ducts has not been limited to closed-loop
scrubber systems. At the Dallman and Laramie River Stations, Paul
and RossC17) have reviewed the extensive pitting and crevice attack
experienced on Alloy 904L outlet ducting. Subsequent spool rack tests
in the Dallman scrubber have shown deep localized attack with 904L and
G alloys, incipient pitting of Alloy 625 weldments, and general immunity
of Alloy C-276. Field coupon exposures at CILCO's Duck Creek No. 1
system have similarly indicated that all highly alloyed materials tested,
with the exception of Alloy C-276, exhibit significant outlet duct
pitting corrosion.(9)
Anderson'11' concludes that nickel-base alloys containing 9-16%
molybdenum are required for the severe corrosive outlet duct condensate
zones associated with limestone FGD scrubbers. In his survey of field
outlet duct spool rack data, however, several cases of severe localized
attack (£7 mils in depth) of Alloy 625 and C-276 coupons in FGD scrub-
ber mixing zones suggest that even the most resistant nickel alloys
are not always adequate. As expected, Alloys 904L and G exhibited
heavy pitting.
Laboratory screening tests of highly alloyed stainless steel and
nickel alloys in simulated SOa-scrubber environments support the ranking
of alloys observed, in addition to implicating species primarily respon-
sible for the localized attack. Several oxidizing acid chloride screen-
ing tests conducted by Silence and Manning^8'9) predict critical pitting
and crevice corrosion temperatures for various nickel-containing alloys.
The oxidizing NaCl-HCl test exposures (involving 4%NaCl + 0.l%Fe2(S04)3
+ 0.01MHC1), for example, reveal that pitting and crevice corrosion
will occur respectively at approximately 100°C and 25°C for Alloy 625,
and at 45°C and 20°C for Alloy 904L. The C-276 alloy appears to suffer
crevice attack at temperatures as low as 80°C under these acidic oxidizing
3-81
-------
conditions. In ambient 10% ferric chloride exposures, light pitting
is noted for the 625 alloy, while very heavy pitting is observed on
the 904L and G alloys.(8)
Hibner and Ross^1* '6 >19) compared extensive laboratory and field
test corrosion data, demonstrating that localized attack occurs on
Alloys 625, G and 904L at chloride levels as low as 1,000 ppm in lower
pH S02~scrubber environments. The additional presence of dissolved
fluorides in FGD scrubber duct liquors was implicated in synergistically
accelerating localized chloride corrosion. (*8>19)
In summary, previous laboratory and field exposure tests predict
the pitting and crevice corrosion observed on the highly alloyed nickel/
stainless steel alloys evaluated in this spool rack study. The hot,
oxidizing acid chloride character of the FGD outlet duct environment
appears to adequately explain the extensive localized attack observed
on these alloys, particularly at higher chloride levels found at the
R. D. Morrow Sr. Station (closed loop) scrubber. Interestingly, the
same species responsible for maintaining titanium alloy passivity in
acidic outlet duct environments (i.e., (h and FeCla), appear to be
very deleterious to corrosion resistance of nickel and ferrous alloys
in the presence of chlorides.
ALLOY PERFORMANCE IN FGD SCRUBBER INLET QUENCH DUCTS
The two scrubber inlet quench exposures evaluated in this study
revealed significant variations in alloy performance. The open loop
(low chloride) inlet presaturator at the Petersburg Station produced
an overall ranking of high performance alloy corrosion resistance simi-
lar to that summarized for outlet ducts:
Titanium 2, 12 > 625 > 904L (>20Cb-3 > 316L)
The inlet quench exposure at the closed loop (high chloride) R. D.
Morrow Sr. Scrubber No. 1, however, resulted in severe localized attack
on both titanium alloys, with significantly less attack on Alloys C-
276, 625 and 904L.
The wet/dry interface of the inlet quench duct potentially repre-
sents the most corrosive zone in the scrubber, resulting from very
high temperatures (143-177°C), very low pH, and oxidizing species asso-
ciated with incoming hot, acidic flue gases and flyash. Very few ma-
terials can resist localized attack under these conditions, especially
when scrubber liquor chloride levels become significant.
The localized attack observed on these titanium alloys is not
unexpected, considering inlet quench conditions at the R. D. Morrow
Sr. Plant. Specifically, deposit crevice corrosion is predicted in
low pH chloride environments when temperatures lie in the 121-177°C
range, as shown in Figure 11. Calcium sulfate (or sulfite) deposits
which form at the wet/dry inlet interface can provide viable crevice
sites for attack. The calcium -carbonate and ferric chloride containing
3-82
-------
HYDROGEN PCKUP
NO HYDROGEN PICKUP
NO CORROSION
Grade 2 Ti
CREVICE CORROSION
7 Ti M-
IOO°F
(37.8'C)
200°F
(93°C)
300° F 400° F
(149-C) (204°C)
TEMPERATURE
500'F
(260»C)
600 "F
(315 »C)
Figure 11. Crevice corrosion of titanium in saturated NaCl brines.
3-83
-------
components of inlet quench deposits, however, would not be expected
to create crevice problems for titanium due to the alkaline and oxidizing
character of these components, respectively.
Assuming that this chloride crevice attack mechanism is in effect,
one would predict that the Ti-Pd (Grade 7 Titanium) alloy could resist
these low pH, high-temperature conditions (Figure 11). If, however,
fluorides are integral to this corrosion mechanism, the Ti-Pd alloy
probably would suffer similar attack. Current in situ inlet quench
exposures at R. D. Morrow Sr. Scrubber No. 1 may soon shed some light
on this mechanism.
The localized attack exhibited by nickel-base and stainless steel
alloys in the R. D. Morrow Sr. inlet quench duct correlates with that
reported for previous field and service experience. Silence and Manning'9^
describe "saucer-shaped" depressions, up to 10 mils deep, on Alloy
C-276 sheet liners in the wet/dry zone of the R. D. Morrow Sr. inlet
quench duct, similar to the shallow pits noted in this study (Figure
8). At this location, Hastelloy G previously suffered severe localized
attack in service.(8>18)
Severe pitting and crevice attack of Alloys 904L and 625 were
reported at the Dallman and Laramie River Station quencher zones.^l6»17'
Similar performance with 904L occurred at Springfield City Utilities'
Southwest 1 Station.^16' Anderson's reviewO1) of field corrosion
studies in FGD limestone scrubbers further reveals examples of signi-
ficant inlet quench pitting of Alloys 904L and G. These findings can
also be explained by the low pH, oxidizing chloride conditions of the
inlet duct, similar to effects discussed in the previous outlet duct
review.
Curiously, for both inlet quench zones included in this study,
net differences in maximum pit depth experienced by C-276, 625 and
904L alloys were quite small, in contrast to. distinctive alloy differ-
ences produced during outlet duct exposures. This may be related to
the alkaline CaCOa component of inlet quench duct deposits and also
be influenced by positioning within the wet/dry interface zone.
In summary, C-276, 625 and 904L alloys experienced finite local-
ized attack in the two scrubber inlet quench zones evaluated in this
study. The titanium alloys, however, exhibited diverse behavior, appar-
ently related to deposit crevice corrosion in the high chloride, closed-
loop scrubber system. The variable behavior observed may be attributable
to spool rack positioning in the wet/dry zone of the inlet quench duct,
the chemical nature of inlet duct deposits formed, and chloride (and/or
fluoride) scrubber levels.
CONCLUSIONS
(1) Field corrosion tests, of up to nine months duration, reveal the
following ranking of high-performance alloys in outlet duct ex-
posures:
3-84
-------
Titanium Grades 2 and 12 > C-276 > 625 > 904L
This was particularly evident in the corrosive closed-loop scrubber
system at the R. D. Morrow Sr. Plant, and the venturi particulate
scrubber at the Dickerson Station.
(2) The superior resistance of titanium alloys compared to the nickel/
stainless steel alloys tested in outlet ducts is very well explained
by the acidic oxidizing chloride character of outlet duct conden-
sate and solids. Ferric compounds found in flyash, in conjunction
with oxygen from flue gas, may be responsible for passivation of
titanium under these acidic conditions. More laboratory testing
is necessary to substantiate these effects.
(3) Wet/dry interface zones in two inlet quench duct exposures reveal
variable performance of titanium alloys. High temperature deposit
crevice corrosion is indicated, and may require Ti-Pd (Grade 7
Titanium) alloy when scrubber chlorides are high. C-276, 625 and
904L alloys consistently experienced localized attack involving
pit depths of 1 to 10 mils.
(4) Based on this preliminary field test data, titanium is recommended
for serious future consideration as a cost-effective candidate
material for problem areas of FGD scrubber systems.
REFERENCES
(1) R. W. Ross, Jr.; "Nickel Alloys Keep FGD Systems Performing,"
Pollution Engineering, June 1982, pages 31-33.
(2) H. S. Rosenberg, et al; "Operating Experience With Construction
Materials for Wet Flue Gas Scrubbers," Presented at the 7th Energy
Technology Conference and Expo, March 1980, Washington, B.C.
(3) D. A. Froelich and M. Ware; "Corrosion Problems With a Closed
Loop Limestone FGD System," NACE Materials Performance, October
1982, pages 40-43.
(4) E. L. Hibner and R. W. Ross, Jr.; "Effect of pH, Chloride Concentra-
tion, and Temperature on Localized Corrosion of Nickel Alloys
in S02 - Scrubber Environments," Paper No. 196, Presented at NACE
Corrosion '82, March 22-26, 1982, Houston, Texas.
(5) E. C. Lewis, G. W. Driggers, and K. W. Malki; "Laboratory Evalua-
tion of Several Alloys in High Chloride FGD Environment - Progress
Report," Preprint No. 17, Presented at the Solving Corrosion Prob-
lems in Air Pollution Control Equipment Seminar, August 11-13,
1981, Denver, Colorado.
(6) J. R. Crum, et al; "Corrosion Resistance of High Nickel Alloys
in Simulated S02 - Scrubber Environments," Paper No. 276, Pre-
sented at NACE Corrosion '80, Chicago, Illinois.
3-85
-------
(7) G. L. Crow and H. R. Horsman; "Corrosion in Lime/Limestone Slurry
Scrubbers for Coal-Fired Boiler Flue Gases," Materials Performance,
July 1981, pages 35-45.
(8) W. L. Silence, P. E. Manning, and A. I. Asphahani; "Alloy Selection
For Flue Gas Desulfurization Systems," Preprint No. 18, Presented
at the Solving Corrosion Problems in Air Pollution Control Equip-
ment Seminar, August 11-13, 1981, Denver, Colorado.
(9) W. L. Silence and P. E. Manning; "Laboratory and Field Corrosion
Test Results Related to Flue Gas Desulfurization Systems," Paper
No. 185, Presented at NACE Corrosion '83, April 18-22, 1983,
Anaheim, California.
(10) D. B. Anderson; "Corrosion in Wet Scrubber Systems - A Case History
Review," Paper No. 127, Presented at NACE Corrosion '81, April
6-10, 1981, Toronto, Ontario.
(11) D. B. Anderson; "Corrosion in Wet Scrubber Systems - A Case History
Review: Part 2," Paper No. 195, Presented at NACE Corrosion '83,
April 18-22, 1983, Anaheim, California.
(12) G. H. Koch and J. A. Beavers; "The Influence of Scrubbers Slurry
Chemistry on the Corrosion of Alloys in Lime/Limestone Flue Gas
Desulfurization Systems," Paper No. 186, Presented at NACE Corro-
sion '83, April 18-22, 1983, Anaheim, California.
(13) G. W. Driggers, P. C. Rader, and J. E. Johnson; "Applying Field
and Laboratory Experience to Alloy and Coating Selection in Com-
bustion Engineering FGD Systems," Paper No. 184, Presented at
NACE Corrosion '83, April 18-22, 1983, Anaheim, California.
(14) J. A. Beavers and G. H. Koch; "Review of Corrosion Related Fail-
ures in Flue Gas Desulfurization Systems," Materials Performance,
October 1982, pages 13-25.
(15) W. L. Silence, P. E. Manning, and A. I. Asphahani, "Service Per-
formance of Highly Alloyed Materials in Air Pollution Control
Systems," Paper No. 198, Presented at Corrosion '82, March 22-
26, 1982, Houston, Texas.
(16) H. S. Rosenberg, C. W. Kistler, Jr., L. 0. Nilsson, L. J. Norwacki,
J. A. Beavers, G. H. Koch, and H. H. Krause, Jr.; "Recent Operating
Experience With Construction Materials for Wet Flue Gas Scrubbers,"
Paper No. 182, Presented at NACE Corrosion '83, April 18-22, 1983,
Anaheim, California.
(17) G. T. Paul and R. W. Ross, Jr.; "Corrosion Performance in FGD
Systems at Laramie River and Dallman Stations," Paper No. 194,
Presented at Corrosion '83, April 18-22, 1983, Anaheim, California.
(18) E. R. Dille, T. F. Lemke, and R. E. Smith, "FGD Corrosion at the
3-86
-------
R. D. Morrow Sr. Generating Plant," Preprint No. 15, Presented
at the Solving Corrosion Problems in Air Pollution Control Equip-
ment Seminar, August 11-13, 1981, Denver, Colorado.
(19) E. L. Hibner and R. W. Ross, Jr.; "Laboratory Investigation of
the Localized Corrosion Resistance of Nickel Alloys in SOa Scrubber
Environments," Paper No. 192, Presented at NACE Corrosion '83,
April 18-22, 1983, Anaheim, California.
(20) E. A.Tice; "Corrosion Tests in Flue Gas Desulfurization Processes,"
Materials Performance, April 1974, pages 26-32.
(21) H. D. Rice, Jr. and R. A. Burford; "Corrosion of Gas-Scrubbing
Equipment in Municipal Refuse Incinerators," Paper No. 106, Pre-
sented at Corrosion '73, March 19-23, 1973, Anaheim, California.
(22) G. H. Koch and J. A. Beavers; "Experimental Evaluation of Alloys
and Linings in Simulated Duct Environments for a Lime/Limestone
Scrubber," Paper No. 197, Presented at NACE Corrosion '82, March
22-26, 1982, Houston, Texas.
(23) D. E. Thomas and H. B. Bomberger; "The Effect of Chlorides and
Fluorides on Titanium Alloys in Simulated-Scrubber Environments,"
Paper No. 189, Presented at NACE Corrosion '83, April 18-22, 1983,
Anaheim, California.
(24) F. W. Fink and W. K. Boyd; "The Corrosion of Metals in Marine
Environments," DMIC Report 245, May 1970.
(25) L. C. Covington and R. W. Schutz; Corrosion Resistance of Titanium,
TIMET Brochure, 1982.
(26) T. Koizumi and S. Furuya; "Pitting Corrosion of Titanium in High
Temperature Halide Solution," Titanium Science and Technology,
Vol. 4, pages 2383-2393, Plenum Press, 1973.
(27) F. A. Posey and E. G. Bohlmann; "Pitting of Titanium Alloys in
Saline Waters," Paper Presented at the Second European Symposium
on Fresh Water From the Sea, Athens, Greece, May 9-12, 1967.
(Submitted for Publication in Desalination).
(28) J. W. Braithwaite, N. J. Magnani, and J. W. Munford; "Titanium
Alloy Corrosion in Nuclear Waste Environments," SAND79-2023C (De-
cember 1979), Paper No. 213 Presented at NACE Corrosion '80, Chicago,
Illinois, March 1980.
(29) L. C. Covington; "The Role of Multi-Valent Metal Ions in Suppres-
sing Crevice Corrosion of Titanium," Titanium Science and Technology,
Vol. 4, Ed. by R. I. Jaffee and H. M. Buite, Plenum Press, New
York (1973).
(30) R. W. Schutz and L. C. Covington; "Hydrometallurgical Applications
3-87
-------
of Titanium," Paper Presented at the 3rd ASTM Conference on Titanium
and Zirconium in Industrial Applications, New Orleans, Louisiana,
September 21-23, 1982.
CHAIRMAN'S NOTE
Following the presentation of this paper by Mr. Charles Young, the
possibility of a sample mix-up was raised. It was the opinion of
Mr. Ross of Huntington Alloys, Inc. that the corrosion behavior
attributed to Alloy 625 was more indicative of a lower molybdenum
(Mo) containing alloy, such as Alloy G.
The samples were subsequently analyzed and were confirmed to be Alloy
625. However, with Mo contents of 8.05% and 8.15%, these samples are
barely above the minimum Mo content for Alloy 625 and significantly
less than samples previously tested by Mr. Ross. A 1.0% difference
in Mo could significantly change the corrosion behavior of the alloy
samples.
3-88
-------
SESSION 4: DRY FURNACE ABSORBENT INJECTION
Chairman: Randall E. Rush
Flue Gas Treatment and Water
Quality Section
Southern Company Services
Birmingham, AL
-------
RESULTS FROM EPA'S DEVELOPMENT OF LIMESTONE
INJECTION INTO A LOW NOX FURNACE
D. C. Drehmel, G. B. Martin, J. H. Abbott
-------
RESULTS FROM EPA's DEVELOPMENT OF LIMESTONE INJECTION
INTO A LOW NOV FURNACE
by: D.C. Drehmel, G.B. Martin, and J.H. Abbott
Industrial Environmental Research Laboratory-RTF
U.S. Environmental Protection Agency
Research Triangle Park, N.C. 27711
INTRODUCTION
As discussed in previous papers (1,2), the Environmental Protection
Agency (EPA) has been developing control technologies for SOX and NOX
emissions. Because of the magnitude of, and the possible link between,
such emissions from power plants and environmental problems such as acid
rain, one focus of SOX/NOX control has been combustion processes.
Previously developed low NOX systems for coal combustion are being
demonstrated, and methods for removal of SOX in these systems are under
development. The primary method for SOX control is the use of Limestone
Injection into a Multistage Burner (i.e., low NOX burner) which is
abbreviated as LIMB. The objective of the EPA program is to develop LIMB
and low NOX burner technology for both retrofit and new applications so
that industry can demonstrate and commercialize this technology in the
post-1986 time frame for both wall-fired and tangentially-fired boilers.
The LIMB program is an EPA effort to develop effective and inexpensive
emission control technology for coal-fired boilers that will reduce SOX
and NOX. LIMB technology represents a low-cost alternative to currently
available SOX control approaches; e.g., flue gas desulfurization, coal
cleaning, and coal switching. LIMB technology is attractive if coal
combustors must be controlled to minimize emissions acid rain precursors
because LIMB is easily retrofitted to large and small coal-fired boilers,
is lower in cost than any available alternative, and can control both SOX
and NOX — the two major acid rain precursors. The technical goals of the
program are: 1) for retrofit applications, achieve 50-60% reduction of
both SOX and NOX from uncontrolled levels; 2) for new systems, achieve 70-
80% NOX and 70-90% SOX reduction from uncontrolled levels; and 3) for both
retrofit and new systems, achieve the above goals at costs at least $100/kW
less than the major technology alternative, flue gas desulfurization.
Given these goals, the program was structured to address major
aspects of LIMB development and application with consideration to three
issues. First, that the existing boiler population is approximately 45%
wall-fired and 45% tangentially-fired. Second, that the program must
define variables governing sulfur capture, impact on particulate control
equipment, and impact on boiler operation (including potential problems
4-1
-------
with slagging and fouling). There is also the question of ultimate
disposal or reuse of the limestone modified fly ash. Lastly, that there is
the need to be able to extend development results to boiler design which
can only be verified by a demonstration.
To achieve the program goals, the EPA has organized a program
consisting of:
1. Basic understanding of mechanisms and kinetics.
2. Systematic small bench and pilot scale development.
3. Large scale pilot testing.
4. Process and systems analysis.
5. Field application to representative boilers.
Results from item 1 have been discussed in detail in other papers
(3,4) and a summary of conclusions is provided under Background, below.
Results from items 3, 4, and 5 are at a very early stage which do not
provide a basis for significant comment. Consequently, the subject of this
paper is item 2; namely, results from bench and pilot scale testing.
BACKGROUND
The purpose of this paper is to update the status of LIMB development
from the time of the paper presented at the previous FGD meeting (May 1982
in Florida).
A quick review of the history of the limestone injection/LIMB process
follows. From 1968 to 1971, dry sorbent injection was evaluated. General
conclusions were that there was insufficient SOX removal (15-50%), that
thermal deactivation (deadburning) of the sorbent was a major problem, and
that power plant operational problems were possible including fouling in
the convective passes and loss in the efficiency of the electrostatic
precipitator. In the summer of 1979, EPA pilot plant tests indicated the
possibility of 70% removal of SOX with limestone injection through low-NOx
burners at reasonable alkali to sulfur ratios. Subsequently, the LIMB
program was initiated (October 1980). In-house experiments started with a
program to define mechanisms and kinetics. Conclusions from that work
include: 1) surface areas formed during rapid limestone calcination in the
dispersed phase are in the range of 50-60 m^/g, 2) there is a strong
relationship between surface area and lime reactivity, and 3) calcination
kinetics can be expected to have a significant effect on SOX capture using
limestone injection.
During recent years EPA has proceeded with development on a wide range
of furnaces (30 kWt to 30 MWt) at five locations using primarily wall-fired
pulverized coal combustion technology. However, other combustion systems
are being studied; e.g., tangential pulverized coal combustion and
spreader stoker coal combustion. Results presented below are from wall-
and tangentially fired furnaces of a small pilot size (0.3 to 0.5 MWt). The
wall-fired unit is the boiler simulator furnace (BSF), at Energy and
Environmental Research Corporation's plant in Irvine, California. The
tangentially fired furnace is at Acurex Corporation's plant in Mountain
4-2
-------
View, California. A detailed description of these furnaces has been
published (1). An extremely important feature of these furnaces is that
the gases leaving the combustion zone pass through a temperature profile
that was designed to simulate the range of temperature profiles in a coal-
fired, steam-electric generating plant. It was anticipated that the
temperature profile would be important in the reaction between limestone
and SOX in the flue gas.
A presentation at the May 1982 EPA symposium noted that very high SOX
captures (70-80% at 2X stoichiometric) were possible using the BSF fired on
natural gas. At that time it was suspected that there was an interaction
between the sorbent and some component of coal which inhibited good
performance of the sorbent during coal firing. In any case, the historical
explanation of deadburning"s causing poor performance was refuted because
the natural gas firing conditions could be made to produce temperatures
even higher than those in coal firing without loss of sorbent reactivity.
RESULTS
To put effects of furnace and limestone injection options in context,
the results will be discussed in three sections, corresponding to the
critical steps for SOX control with limestone:
1. Limestone conversion to lime (with good reactivity).
2. Reaction of lime with sulfur species.
3. Thermal and chemical deactivation of lime.
LIMESTONE CONVERSION TO LIME
Results for surface areas of three limes are shown in Figure 1. Data
for most of these results were developed under EPA contract 68-02-2566 with
Northrup Services, Inc. The limestones from which these limes were derived
are all calcites (rhombohedral form of CaC03). The calcites called Vicron
and El Dorado are both coarse grained; that is, the fundamental size of an
individual crystallite is large (>50 ym). The marl by formation is a
reprecipitated deposit and is fine grained (the grains are 1-3 y m).
Surface areas have been determined by two methods: mercury porosimetry and
the BET method. Most of the data shown in the Figure 1 are for batch
calcination in a platinum crucible. Whether measured by BET or mercury
porosimetry, the surface areas of limes prepared under soft-burning
conditions (980°C for 2 hours) are in a range of 1 to 4 m^/g. By searching
combinations of calcination conditions and allowing as much as 30% to be
uncalcined, it is possible to achieve surface areas an order of magnitude
higher. However, the best surface area for the fine grained marl does not
ever come close to the best surface areas for the coarse grained calcites.
For the Vicron limestone tested under EPA contract 68-02-3633 with EERC,
results under shock calcination conditions are also shown. For this
calcination, the limestone was exposed to 1400°C for only a few milli-
seconds. This illustrates the high surface areas that can be achieved
during very short calcination times although even higher surface areas
have been reported elsewhere (3).
4-3
-------
25
H
(A 20
UJ
CC
UJ
O
£ 15
cc
D
(A
CC
UJ
LU
5 10
O
cc
O
0.
Ob
t 6
10
0
>-
—
^
MM
_
—
^
-
m*m
n
'« — 1 twifl.1
':M:
•'•'!'•!•
li
L
p—
p^BM
•
nF^H
HQ/15M/.6 HOUR IM BCT/2550/M8EC ~
g H BET/2550
[=3 HO/1800/2 HOUR • 40% KAOLIN
^ BETM800/2 HOUR Illlill BET/3000/KAOLIN_
.
LM^J 1
—
-
-
—
-
-
1* 'E^ •'
VICRON
(COARSE GRAINED)
ELDORADO
(COARSE GRAINED)
MARL
(FINE GRAINED)
FIGURE 1. SURFACE AREAS FOR THREE CALCITES.
4-4
-------
Since an important impurity in coal is kaolinite, it is interesting to
note the effect of kaolin addition to limestone on the surface area of
shock calcined Vicron. At 1400°C, the surface area is reduced by almost
two thirds, and at 1650°C, the surface area is reduced to less than a fourth
of that for the pure lime.
A comparison of the utilization of different SOX sorbents is given in
Figure 2 for wall-fired propane (doped with t^S) or pulverized coal
combustion as derived from EPA contract 68-02-3921 with EERC. These data
were taken from experiments on the BSF. In Figure 2, dolomite and
hydroxide are compared as well as three calcites. With the knowledge that
the hydroxide is inherently fine grained, the results for propane firing
show that the calcium utilization for SOX capture with Vicron (coarse
grained) is a third higher than that for fine grained sorbents. Moreover,
the coarse particle Vicron (diameter about 11ym) appears to perform better
than the fine particle Vicron (diameter about 3 ym). The performance of
dolomite is within the range of the other sorbents during propane firing.
The interaction between fuel type and sorbent type is shown in Figure
2. Note the changes in sorbent utilization in comparing propane firing and
coal firing. The dolomite loses one third of its reactivity, and the
calcites lose one half to two thirds of their reactivity.
Derived from EPA contract 68-02-3684 with Acurex, data for tangential
firing with natural gas (doped with H2S) are given in Figure 3. As before,
the very finest particles do not give the best results, and one set of data
implies that there is an optimum particle size.
REACTION OF LIME WITH SULFUR SPECIES
As noted above, the capture of SOX should be favorable in a certain
temperature range which is bounded above by decomposition of the product
(CaS04) and below by slow kinetics. Also derived from EPA contract 68-02-
3684 with Acurex, Figures 4 and 5 give some additional verification of the
important effect of temperature profile. Figure 4 displays three
temperature profiles that have been used in the tangentially fired
furnace. These profiles start from a point approximately 2 m above the
level of the burners so that the beginning temperatures in Figure 4
correspond to a residence time of 2.4 seconds at normal load. Changes in
temperature profile were accomplished by modification to the placement and
size of heat exchangers. For the overall temperature range displayed, note
that the first profile has a precipitous drop in the middle of the range,
while the third profile shows the most time in the middle of the range.
Figure 5 gives the SOX capture during coal firing for these different
temperature profiles. The first profile has less than half the SOX capture
of the third profile, and the second profile has an intermediate per-
formance .
Limestone can be injected at several locations for each type of furnace.
For the tangential furnace, air is provided for combustion either with the
coal, near the coal, or some distance above the level of coal injection,
called primary air, lower auxiliary air, and overfire air, respectively.
4-5
-------
40
30
#
*
z
o
N
P 20
V)
LU
2
3 10
I
I VICRON/FINE
I FREOONIA
I HYDROXIDE
D
I VICRON/COARSE
DOLOMITE
MARL
PROPANE-FIRED INDIANA COAL-FIRED UTAH COAL-FIRED
FIGURE 2. WALL-FIRED: SORBENT AND FUEL EFFECTS.
4-6
-------
30
CD FINE
[Ml MEDIUM
B COARSE
N
Ol
Z
o
§ 10
BURNER INJECTION
HIGH LOAD
CENTER INJECTION
HIGH LOAD
CENTER INJECTION
LOW LOAD
FIGURE 3. T-FIRING: :PARTICLE SIZE EFFECT.
4-7
-------
FURNACE
TEMPERATURE, °C
1400
1100
800
500
PROFILE!!
^I^^^^^^^A— '
PROFILE; 2
B_
2.5
1.0 1.5 2.0
RESIDENCE TIME, sec
FIGURE 4, TEMPERATURE PROFILES IN T-FIRING.
3.0
4-8
-------
50
40
I I I
I I I I I I I I
.PROFILE 1
. PROFILE 2
.PROFILES
Z
O
00
LU
z
o
Q
LU
oc
30
20
10
1 1 1
1 2 3
CALCIUM TO SULFUR MOLAR RATIO
FIGURE 5. T-FIRING: TEMPERATURE PROFILE EFFECT.
4-9
-------
All of these air injection locations have been used for limestone
injection. However, a tangentially fired furnace introduces all of these
air streams from the side (specifically, the corner) of the furnace.
Because the tangentially swirling motion of the combustion gases could
prevent even distribution of limestone injected from the side, an
additional location (using a probe) was devised to inject limestone into
the center of the furnace. A comparison of the SOX capture in the Acurex
tangential furnace using these different injection locations is plotted in
Figure 6. Injection with the coal gives the poorest result, and injection
into the center, the best.
THERMAL AND CHEMICAL DEACTIVATION OF LIME
Figure 7 demonstrates the effect of fly ash on lime reactivity in the
BSF under EPA contract 68-02-2667 with EERC. Using natural gas as the
fuel, the increase in fly ash addition results in a decrease in limestone
utilization. This effect is also related to the limestone concentration as
shown by the family of lines for utilization at different calcium to sulfur
molar ratios. The decrease in utilization for a given fly ash addition is
not as bad when the calcium to sulfur ratio is higher, but the absolute
value of utilization is worse at higher calcium to sulfur molar ratios.
By firing the Acurex tangential furnace (EPA contract 68-02-3684) on
natural gas and injecting fly ash, it was verified that this system also
saw an effect of limestone and fly ash interaction on the SOX capture.
Figure 8 shows the change in limestone utilization in collecting SOX when
fly ash is added. The full range of fly ash addition spans from clean
conditions to those similar to coal combustion. For center limestone
injection, the limestone utilization is reduced by half. Data were not
available for side injection over the whole range, but the rate of
reactivity loss with fly ash addition is the same.
Besides adding fly ash to natural gas firing in the BSF, other
compounds have been tried to narrow the range'of mineral species which may
cause a loss of limestone reactivity under EPA contract 68-02-3921 with
EERC. In particular, bentonite and kaolin were used to determine the
effect of silica or alumina on the performance of the limestone Vicron in
capturing SOX. An important result of that experiment is plotted in Figure
9 which shows a good inverse correlation between SOX capture and the
alumina content of the inlet solids. This correlation draws together data
from simple coal firing (Indiana coal), from fly ash addition (ash from a
furnace called the Small Watertube Simulator or SWS), and from mineral
matter addition (bentonite and kaolin).
DISCUSSION OF RESULTS
LIMESTONE CONVERSION TO LIME
Data shown under Results, above, suggest the possibility of very high
surface area limes (>20 m^/g) being present in the furnace for very short
periods of time—certainly less than 1 second. Because SOX capture is a
4-10
-------
50
40
(A
Z
o
30
2 20
o
o
LU
cc
10
i r
PRIMARY AIR
r~ipi i i i p~~i i i n
LOWER AUX AIR .
OVERFIRE AIR .
CENTER .
0 1 I I I
I I I
I I I I
CALCIUM TO SULFUR MOLAR RATIO
FIGURE 6. T-FIRING: INLJECTION LOCATION EFFECT,
4-11
-------
20
15
Z
o
N
13
P
UJ
to
UJ
10
I I IITilIIjTilI I I I I I|T1TilTilT
—— Ca/S = 1
Ca/S=2
Ca/S=3
ft I I I I I I I I I I I I I I I I I
0 1
I I I I I I I I I I
2 3
WEIGHT RATIO OF ASH TO LIMESTONE
FIGURE 7. WALL-FIRING: FLY ASH EFFECT (FUEL = GAS)
4-12
-------
10
I I I I i I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I
CENTER INJECTION
___. SIDE INJECTION
z
o
05
z
o
I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I
10 20 30
AMOUNT OF FLY ASH. %
FIGURE 8. T-FIRING: FLY ASH EFFECT (FUEL = GAS).
40
4-13
-------
• SWS ASH + VICRON (1:1)
• BENTONITE + VICRON (2:11
A INDIANA ASH + VICRON
• SWS ASH 4 VICRON (3:11
T KAOLIN 4 VICRON (1:1)
I
g
SO
40
30
20
10
I
I
30
FIGURE 9.
10 20
% OF INLET SOLIDS
IMPORTANCE OF COAL MINERAL MATTER.
4-14
-------
direct function of surface area, optimizing the presence of high surface
area lime is vital. That this critical path can be controlling is
demonstrated by the relative performance of the coarse and fine grained
calcites. The coarse grained calcites achieve higher surface areas and
show better SOX capture in furnace injection during propane firing.
Similarly, the effect of particle size should be noted. As the original
particle size is reduced, it is expected that the reactivity will be
enhanced because of the greater surface area of the limestone. However,
very small limestone particles may calcine too rapidly and may actually
give smaller surface areas in their calcined forms than those of slightly
bigger particles. Certainly, the data show that continued reduction of the
limestone particle size may be detrimental, and an optimum limestone grind
should be determined.
REACTION OF LIME WITH SULFUR SPECIES
The presence and critical importance of an optimum temperature range
has been discussed in the past. Current data show the strong effect of
temperature profile, although the exact definition of the temperature
range is not possible. If the tangential furnace temperature values are
for bare thermocouples, the actual temperatures may be 100°C higher than
shown in Figure 4. It would appear that the temperature range of interest
is approximately 850 to 1300°C. The effect of different limestone
injection locations provides different mixing conditions and different
temperature profiles for calcination. The fact that center injection was
superior to any of the side injection locations in the tangential furnace
verifies that mixing is critical in this case. The variation in
performance of the other injection locations suggests a shifting of the
match between the availability of the very high surface area lime and the
temperature range for the sulfation reaction. Because calcination is very
rapid and the very high surface area would be available early in the
calcination process, it is reasonable that the best performance for side
injection should come from the overfire air location where the limestone
quickly passes into the temperature range for sulfation.
THERMAL AND CHEMICAL DEACTIVATION OF LIME
For both wall- and tangentially-furnaces, fly ash has a negative
influence on limestone utilization. Because of a correlation of this
negative influence with the alumina content, it may be supposed that the
fly ash collides with the limestone, and low melting compounds of alumina
and lime are formed which may flux the surface of the lime particle to a
glassy nonporous sphere. The surface area data with and without kaolin
addition demonstrate the negative effect of kaolin on the open structure of
the sorbent. Questions that arise are how do the fly ash and lime particles
collide, do the fly ash particles have to be sticky, and does the chemical
nature of the sorbent (calcite vs dolomite) influence the reaction between
alumina and the sorbent. The information in Figure 2 indicates some
answers to these questions. Comparing the results for propane firing and
those for Indiana coal firing, indicates the possibility of the influence
of particle size. Specifically, the reduction in utilization for the
coarse Vicron is worse than that for the fine Vicron. However, this is a
4-15
-------
small effect and must be considered because Vicron shatters during shock
calcination to sizes an order of magnitude lower. The importance of a
particle size effect would be to suggest that collision of the fly ash and
lime depends on inertia (and thus size) differences. Again comparing the
same two data sets, the marl and hydroxide—both fine in size—lose
reactivity dramatically in switching from propane to coal firing, and the
dolomite—coarse in size—retains much of its reactivity. Dolomite may
also retain reactivity because the structure of its calcine may not be as
easily blinded by surface reactions. Comparing the two coal data sets,
indicates that the marl is much more reactive with Utah coal firing than
with Indiana coal firing. This may be attributed to chemical deactivation
because a change in fly ash size gives less interaction between fly ash and
sorbent, or it may be attributed to limestone's conversion to lime because
of a change in the peak temperature which affects calcination. Obviously
at this point, more experiments are required to-define the interaction of
different sorbents with fly ash.
Future work will also address the general problem of minimizing the
interaction of fly ash and sorbent. It is possible that some coals are less
likely to produce troublesome fly ash. Consequently, several coals with
different rank or mineral composition will be tested. Fly ash can also be
minimized by use of a cleaned coal or a precombustor which removes slag
before gases enter the main furnace. If aluminum oxide is the only
problem, then coal cleaning could be optimized for its removal. Other
approaches to the fly ash interaction problem involve the sorbent. The
sorbent could be modified to allow its injection at a point where fly ash
is no longer active. It is also possible that the sorbent could be modified
to be protected from the fly ash.
CONCLUSIONS
The success of SOX control with limestone injection will depend on the
success of three critical steps:
1. Production of a highly reactive sorbent.
2. Favorable conditions for the sulfation reaction.
3. Avoidance of sorbent deactivation by interaction with fly ash.
For a given system, any one or more of these critical steps could be
controlling; therefore, careful analysis of the data is required to
identify the appropriate problem(s). The small pilot scale data for both
wall- fired and tangentially fired furnaces have shown that failure in any
of the above steps can reduce SOX capture by a factor of 2 or more. Without
fly ash (firing gaseous fuel) and operating under reasonable furnace
conditions, limestone utilizations as high as 40% have been achieved. At
a calcium to sulfur molar ratio of 2, this means that 80% control, is
possible. Problems in any one of the critical steps will reduce this
accordingly. For the last critical step-—fly ash interaction— specific
approaches to a solution are under investigation, and minimization of this
problem should be possible for most furnaces.
4-16
-------
REFERENCES
1. Drehmel, B.C., Martin, G.B. and J.H. Abbott, "S02 Control with
Limestone in Low NOX Systems: Development Status," in Proceedings:
Symposium on Flue Gas Desulfurization, Volume 2, EPRI CS-2897, March
1983.
2. Drehmel, D.C., Martin, G.B., Milliken, J.O., and J.H. Abbott, "Low
NOX Combustion Systems with SC>2 Control Using Limestone," paper No.
83-38.7, presented at the annual APCA meeting, Atlanta, Georgia, June
1983.
3. Borgwardt, R.H., Gillis, G.R., and N. Roach, "IERL-RTP Research on
Sulfur Capture in LIMB," presented at the EPA/EPRI Symposium on
Stationary Combustion NOX Control, Dallas, Texas, November 1982.
4. Borgwardt, R.H., "Calcination Kinetics and Surface Area of Dispersed
Limestone Particles," in press.
ACKNOWLEDGEMENTS
The authors wish to thank the many investigators, both inside and
outside EPA, who were part of the research and development work to produce
these results. In particular, we wish to acknowledge the dedicated efforts
of the project manager and principal investigator with each of the
contractor organizations: at Acurex, John Kelly and Shigeto Ohmine; at
EERC, Roy Payne "and Patti Case; and at Northrup, Hunter Daughtrey and
Connie Turlington.
4-17
-------
REVIEW OF EPRI RESEARCH ON FURNACE SORBENT
INJECTION OF S02 CONTROL
M. W. McElroy
-------
REVIEW OF EPRI RESEARCH ON FURNACE
SORBENT INJECTION S02 CONTROL
by: M. W. McElroy
Coal Combustion Systems Division
Electric Power Research Institute
Palo Alto, California 94303
ABSTRACT
Direct furnace injection of alkaline material is currently under deve-
lopment as a potential low cost S02 control approach for new and existing
coal-fired boilers. Major developmental efforts now focus on 1) under-
standing process fundamentals, 2) process optimization for both new and
retrofit applications and 3) the resolution of power plant impact issues
and costs. Current and planned EPRI research projects in these areas are
reviewed.
INTRODUCTION
In the early 1960's the concept of direct sorbent injection into the
furnace of a utility boiler originated as a means for reducing S02 emis-
sions without involving utility companies in sophisticated chemical flue
gas treatment systems. It was felt at the time that direct injection of
limestone followed, for example, by wet particulate scrubbing was the least
complicated and most economical procedure for meeting anticipated S02 as
well as particulate removal requirements. However, various trials of this
concept at laboratory furnaces and full scale utility boilers in the U.S.,
Europe and Japan, generally failed to demonstrate sufficient in-furnace
S02 removal at practical sorbent to sulfur ratios. S02 removals during
tests at utility boilers typically ranged from 15% to 40%, well below the
target values of 70% to 90%. S02 removals were also highly dependent on
the design and operation of the boiler as well as the type of sorbent and
injection system used. The potential for adverse effects on boiler per-
formance also surfaced during these early test programs. Most notably,
increases in slagging and fouling of boiler heat transfer surfaces and de-
graded performance of electrostatic precipitator controls were reported.
Because of the low and variable S02 removal efficiencies combined with
boiler operational concerns, further process testing was abandoned by the
early 1970's and the development of wet S02 scrubber systems became the
primary focus of SO2 control efforts. (1)
A variety of factors have contributed to the renewed interest in
furnace sorbent injection. Foremost among these are the recent experimental
data from the U.S. and Europe that indicate the potential to achieve higher
4-19
-------
SC>2 removal efficiencies at practical sorbent injection rates, if combustion
and sorbent conditions can be properly controlled. New combustion systems
being developed to control emissions of nitrogen oxides (NOX) fortuitously
may help achieve these necessary conditions. Specifically, controlled
mixing conditions associated with low-NOx combustion may provide new oppor-
tunities to optimize the sorbent injection process and also provide the
added benefit of simultaneous in-furnace NOX and -SC>2 control. A final
factor contributing to the resurgence of interest in dry sorbent injection
is the growing incentive for low-cost incremental SC>2 controls applicable
to existing as well as new power plants. Also, the potential exists for com-
bining furnace sorbent injection with other SC>2 control technologies. For
instance, its use along with coal cleaning or coal blending may provide
SC>2 emission compliance flexibility or allow purchase of cheaper, higher
sulfur coals for existing units. Other opportunities may result from inte-
grating this process with other SC>2 flue gas treatment systems, with the
possibility of providing overall SC>2 control capability adequate to meet
higher SC>2 control requirements for new plants.
EPRI RESEARCH RESULTS
To provide a perspective on EPRI's R&D program, Table 1 lists a number
of the development issues that ultimately need to be resolved before wide-
spread application of this process can be considered. Regarding the SC>2
capture process itself, it is a complex physical/chemical process which is
by no means completely understood. There are a multitude of parameters
affecting SC>2 capture: sorbent properties, coal properties, injection sys-
tem design, and boiler temperature characteristics just to name a few. A
big issue in this regard is how to improve calcium utilization which, in
much of the work to date, has been in the 20 to 25 percent range.
Another area requiring further R&D is the impact of the process on the
rest of the plant. The introduction of limestone or other alkaline
materials into the furnace cavity not only adds to the total mass of solids
in the boiler but changes the physical and chemical characteristics of the
ash. This in turn can affect: 1) slagging and fouling deposits on the
furnace heat transfer surface, 2) convective pass tube erosion, 3) and back
end ash deposition patterns. These potential problems can conceivably be
designed out of a new unit, but they could be a serious issue especially
for retrofits. Another major question concerns the effect on the electro-
static precipitators. Increased ash loading and higher ash resistivity
will likely result in degraded ESP performance. This again would be an
especially relevant issue for retrofits.
Finally, these development issues and their likely site-specific solu-
tion result in a level of uncertainty regarding the cost of commercial appli-
cation. In addition to the technical development programs at EPRI, an im-
portant objective is to provide realistic cost estimates for utility boiler
applications and refine these estimates as the development efforts proceed.
The following discussion focuses on the results from EPRI projects in the
areas of: 1) sulfur removal capability and process development, 2) boiler
slagging and fouling consequences and 3) current process cost estimates.
4-20
-------
Table 1. Furnace Sorbent Injection Unresolved Commercialization
Issues
• INADEQUATE PROCESS UNDERSTANDING FOR
OPTIMUM S02 REMOVAL
• BOILER SLAGGING AND FOULING CONSEQUENCES
• CONVECTIVE PASS EROSION
• AIR HEATER DEPOSITION
• PARTICULATE COLLECTOR PERFORMANCE
0 ASH HANDLING AND DISPOSAL REQUIREMENTS
• RETROFIT Low-N0x BURNER AVAILABILITY AND NEED
• RETROFIT CAPITAL AND OPERATING COST
4-21
-------
PROCESS DEVELOPMENT
To confirm the S02 removal potential of sorbent injection, pilot
testing was performed by Mitsubishi Heavy Industries (MHI) in a jointly
sponsored project with the Electric Power Development Company of Japan. (2)
These initial proof-of-concept tests were conducted on a 80 million Btu/hr
(24 MWt) test furnace equipped with MHl's low-NOx "PM" burner. An earlier
EPRI project with MHI and Combustion Engineering to verify the NOX control
performance of the PM burner (3) established that temperature and com-
bustion conditions in this furnace closely similated those of an actual
utility boiler.
The sorbent injection testing was intended to demonstrate S02 removal
without process optimization, and also provide a preliminary indication
of the influence of major process and combustion parameters. Two coals
selected for the tests were a Japanese high sulfur bituminous coal
(3.1% sulfur) and a low-sulfur western U.S. subbituminous coal (0.7%
sulfur). A high calcium limestone was used as the sorbent.
Initial screening tests evaluated a variety of sorbent injection
locations within the furnace. These included the burner's coal and
secondary air nozzles, and separate ports downstream of the burner zone
used to inject air (overfire air ports) or recirculated flue gas into the
furnace. Injecting the limestone through the overfire air ports was the
most effective approach. Typical results on both coals are shown in
Figure 1 for a range of limestone injection rates (expressed as Ca/S molar
ratio).
For the bituminous coal, S02 removal was over 40% at a Ca/S of 2 and
approached 60% at a Ca/S of 3. This corresponds to a calcium utilization
of about 20 percent. The S02 removals for subbituminous coal were con-
sistently lower at an equivalent Ca/S ratio. Other parameters, including
limestone fineness, furnace excess oxygen and coal firing rate influenced
S02 capture but to a lesser extent than sorbent injection location and
Ca/S.
To complement the pilot tests, MHI also conducted a series of bench
scale experiments to examine S02 capture under idealized laboratory condi-
tions. These tests used a small flow reactor in which pulverized lime-
stone and synthesized flue gas mixtures could be mixed and reacted. Fun-
damental parameters such as reaction temperature, sorbent/S02 contacting
time and initial S02 concentration could be independently\varied. An
example of the data from this work is shown in Figure 2. The results
indicated that there exist optimum combinations of temperatures and
reaction times which maximize S02 capture for a given sorbent material.
The highest S02 removals were observed at the longest reaction time of
1.5 seconds which may not be unlike the residence time typically available
in the furnace of a utility boiler. The effects of S02 concentration
observed in the pilot testing were also duplicated in the flow reactor.
4-22
-------
SO2 Removal Rate (%)
60
40
20
Bituminous (3.1% S)
Subituminous (0.7% S)
Test conditions
OFA limestone injection
80 MBtu/h
20% OFA
4o/o O,
Ca/S Molar Ratio
Figure 1. SO, Removal Efficiency Versus Ca/S — 80 MBtu/hr
Pilot Tests
SO2 Removal (%)
100
80
60
40
20
Residence time = 1.54s
Residence time = 0.54 s
234
Ca/S Molar Ratio
Figure 2. Effect of Residence Time on S02 Removal Efficiency
Bench Scale Flow Reactor Tests
4-23
-------
The MHI pilot and bench scale work verified the potential for S0£ con-
trol and provided some insights into the SC>2 removal process. However,
these programs along with other research efforts in the U.S. and abroad have
also identified major deficiencies in the current understanding of the mecha-
nisms and applicability of the process. For example, serious questions con-
cerning the most effective way to improve calcium utilization, and process
design criteria for the range of furnace conditions encountered throughout
the utility boiler population, still remain. Such questions need to be re-
solved before full scale application of the process can proceed.
To complete the process development, a jointly funded project with
Southern Company Services has been initiated. This project involves conti-
nuation of bench and pilot testing with the objective of developing design
guidelines for utility boiler applications. Furnace sorbent injection alone
and its integration with dry "back end" flue gas treatment processes, are
being considered. Southern Research Institute and KVB, Inc. are subcon-
tractors performing the pilot and bench scale testing. This work is sche-
duled for completion in early 1985.
POWER PLANT IMPACTS
As mentioned previously, the introduction of limestone or other sor-
bents into the furnace will alter the chemical and physical properties of
the fly ash as well as increase the quantity of solids passing through the
boiler. For example, as shown in Figure 3, a limestone injection rate cor-
responding to a calcium/sulfur ratio of 2.0 would increase total solids
loading 25% for a coal containing 0.5% sulfur and 10% ash, but would nearly
tripple the solids for a 4% sulfur, 10% ash coal. (4)
Based on a review of earlier utility experience with this technology,
a preliminary retrofit feasibility study by Combustion Engineering identi-
fied a number of possible related power plant impacts. (1) Foremost among
these are the likelihood for increased slagging and fouling deposits on
furnace heat transfer surfaces and degraded 'electrostatic precipitator per-
formance, in addition to increased ash handling and disposal requirements.
A preliminary indication of potential fouling and slagging consequences
was obtained during pilot combustion tests by Combustion Engineering (5).
Tests were conducted at CE's 4 MBtu/hr Fireside Performance Test Facility
(FPTF) to measure the slagging and fouling properties of fly ash produced
during limestone injection. The FPTF similates the: 1) geometry, tempe-
rature ranges, and velocities of a utility boiler's convective section,
and 2) thermal environment of the radiant furnace walls. Two coals with
differing ash characteristics, representing a range of slagging and fouling
behavior encountered in the utility industry, were examined. The first
coal, a high iron content Midwestern bituminous coal (3.4% sulfur) charac-
teristically exhibits high slagging and moderate fouling tendencies. The
second coal, a high sodium content Western subbituminous coal (0.5% sulfur)
exhibits high fouling and moderate slagging behavior.
4-24
-------
Percent Increase
400
300
200
100
1 2 3
Percent Sulfur in Coal
Figure 3. Increased Boiler Solids Loading During Furnace Lime-
stone Injection for 10% Ash Coal
Required soot blowing interval (hours)
40 _ m
. :»™_ Eastern bituminous (2000'F)
30
20
10
| | Western subbituminous (2100'F)
>20
1 2
Ca/S Molar Ratio
Figure 4. Impact of Furnace Limestone Injection on Convective
Pass Sootblowing Frequency (Ca/S = 2)
4-25
-------
The focus of the work was on measuring the chemical and physical pro-
perties of the deposits including chemical composition, deposit accumula-
tion rate, deposit bonding strength, heat transfer effects, and ease of
removal ("deposit cleanability"). These measurements were than utilized
by CE to predict the impact of furnace limestone injection on boilers
firing these two coal types. For both coals, the increased mass loading
during limestone injection resulted in a greater fouling deposit accumu-
lation rate on the convective tubes. However; increased deposit friability
and lower deposit-to-tube bonding strength enhanced the effectiveness of
conventional sootblowing by improving deposit cleanability. For the
Western subbituminous coal, the addition of limestone may serve to alle-
viate severe fouling problems commonly encountered with this type of coal.
Generally, it appears that the increased fouling deposit accumulation
rates with limestone addition can be managed with conventional sootblowing
equipment, although increased sootblower frequency (Figure 4) and/or addi-
tional sootblowers are likely to be required.
Limestone injection also increased the quantity of slagging deposits
in the radiant section. The physical characteristics of the deposits
were also affected. In general, the addition of limestone resulted in a
drier, more friable deposit which is more easily removed with conventional
sootblowing. To overcome the increased quantity of slagging deposits more
frequent sootblowing would be required.
One significant result of the FPTF work was to demonstrate a laboratory
procedure to determine the slagging and fouling consequences of furnace
limestone injection. While the results obtained pertaining to these two
coals were instructive in developing the general impact of limestone
injection on boiler heat transfer surfaces, utility companies considering
this technology for retrofit would be advised to consider similar tests
to screen particular coals for their slagging and fouling characteristics
due to the variability in coal properties. Engineering studies would also
be necessary to account for unit-specific factors such as furnace design,
convective tube spacing and geometries, ,sootblower coverage and ash hand-
ling capabilities.
ECONOMICS
The capital and operating costs of furnace limestone injection have
been estimated by Combustion Engineering (1), Steams-Roger (6) and KVB (4)
for new and retrofit applications. For new 500 MW boilers equipped with
fabric filter particulate controls, the incremental capital cost estimates
range from $15 to 30/kW depending on coal sulfur content and the Ca/S
ratio (1,6). The major portion of these costs are associated with lime-
stone delivery, storage, pulverization and injection (including combustion
system design modifications) which represent added plant components not
otherwise needed.
For retrofit applications (1,4) the capital cost can be considerably
higher due to the probable need for upgrading sootblowing systems, electro-
static precipitators and ash handling and disposal systems. Such costs
4-26
-------
would be very site specific but could result to total retrofit capital
costs 2 to 3 times the new unit costs. The range of capital cost estimates
associated with the major plant subsystems affected are shown in Figure 5.
The total plant incremental capital costs for two retrofit scenarios are
summarized in Figure 6.
The operating costs increase nearly proportionately to coal sulfur
content. This reflects the fact that higher coal sulfur requires greater
quantities of raw sorbent and produces more byproduct for disposal, both
of which represent major operating cost items. As shown in Figure 7 for
Ca/S ratios between 2 and 3, 30-year levelized operating costs range from
about 3 to 5 mills/kW-hr for a 0.5% sulfur coal and increase to 8 to 12
mills/kW-hr for a 4.0% sulfur coal. As additional research further defines
the process design and plant upgrading requirements, refinements in these
preliminary cost estimates will be possible. .
FUTURE RESEARCH
In addition to the research described above, results from other EPRI
projects are expected to be applied to this technology. Low-N0x burner
evaluations (7) with Riley Stoker, Babcock & Wilcox, Combustion Engi-
neering, MHI and KVB, Inc. are aimed at demonstrating retrofit NOX com-
bustion control systems and providing the guidelines for their application.
If such systems are shown to be an essential element of the sorbent in-
jection process, the results from these efforts would play a key role in
its commercial application. In addition, research by Southern Research
Institute on flue gas conditioning and pulsed energization (8) for enhan-
cing electrostatic precipitator performance could play an important part
where the retrofit application of furnace sorbent injection may require
upgrading of existing particulate controls.
These efforts together with the process development, boiler impact
studies and economic evaluations discussed above will provide the ground-
work for prototype testing of sorbent injection of coal utility boilers.
The retrofit of several boilers in the 40 to 100 MWe size range is consi-
dered the ne'xt logical step towards determining the commercial potential
of this process. . They will enable final optimization and verification of
the sorbent injection process design and also allow economical resolution
of remaining plant impact issues that may not be fully accomplished in
laboratory tests. Experimental work at the prototype scale will include
evaluation of 1) alternate electrostatic precipitator upgrades, 2) long
term effects on furnace slagging, fouling and erosion, 3) alternate sor-
bent injection systems, and 4) ash handling and disposal requirements.
4-27
-------
Capital Requirement ($/kW)
70
60
50
40
30
20
10 —
W////////A
1/////7/////A
Sorbent Economizer Low-NOx Particulate Ash Sootblowing
Storage & Modifications Combustion Collection Handling System
Handling System System
Retrofit Upgrade
Figure 5. Range of Power Plant Subsystem Capital Costs for
Retrofit Furnace Limestone Injection. Particulate
Control Costs Correspond to: Fly Ash Conditioning
($5/kW), Double SCA ($35/kW) and Baghouse Retrofit
($70/kW)
4-28
-------
$/kW
100
80
60
40
20
Cost Assumptions
Case A
• Minor Low-NOx modifications (2 $/kW)
• ESP fly ash conditioning system
• No change to ash handling or
sootblowing systems
Case B
• Moderate low-NOx modifications (7 $/kW)
• Increase ESP SCA from 300-600
• Modify ash handling and sootblowing
systems
• Modify economizer
1 2 3
Coal Sulfur Content (%)
Figure 6. Estimated Capital Requirement of Furnace Limestone
Injection
Mills/kWh
18
16
14
12
10
8
6
4
2
Band defined by Ca/S range
from 2 to 3
1234
Coal Sulfur Content (%)
Figure 7. Estimated 30-Year Levelized Busbar Costs of Furnace
Limestone Injection (Cases A and B same as Figure 6)
4-29
-------
REFERENCES
1. Kokkinos, A., et al, "Feasibility of Furnace Injection of Limestone
for SC>2 Control", in Proceedings of the 1982 Joint Symposium on Sta-
tionary Combustion NOX Control, Dallas, Texas, November 1-4, 1983.
EPRI CS-3182, Vol. 1, July 1983.
2. EPRI Research Project RP1836-1 in progress.
3. Takahashi, Y., et al, "Evaluation of Tangential Fired Low-N0x Burner"
and Kokkinos, A. et al, "Feasibility Study of a low-NOx Retrofittable
Firing System with U.S. Coals", in Proceedings of the 1982 Joint Sym-
posium on Stationary Combustion NOX Control, Dallas, Texas, November
1-4, 1983. EPRI CS-3182, Vol. 1, July 1983.
4. McElroy, M.W., Muzio, L.J., and Thompson, R.E., "Retrofit NOX and SC>2
Controls for Coal-Fired Utility Boilers", Coal Combustion Systems
Division of EPRI, May 1983.
5. EPRI Research Project RP899-2 in progress.
6. Naulty, D.J., et al, "Economics of Dry FGD by Sorbent Injection",
presented at the 6th International Coal and Lignite Utilization Exhi-
bition and Conference, Houston, Texas, November 15-17, 1983.
7. McElroy, M.W., "Status of Retrofit Low-N0x Combustion Control for
Coal-Fired Utility Boilers", presented at the 76th Annual Meeting
of the Air Pollution Control Association, Atlanta, Georgia, June 21-
23, 1983.
8. EPRI Research Projects RP724 and RP1868 in progress.
4-30
-------
DIRECT DESULFURIZATION THROUGH ADDITIVE INJECTION
IN THE VICINITY OF THE FLAME
M. Y. Chughtai, S. Michelfelder
-------
DIRECT DESULFURIZATION THROUGH ADDITIVE
INJECTION IN THE VICINITY OF THE FLAME
by: M. Y. Chughtai and S. Michelfelder
L. & C. Steinmiiller GmbH
Postfach 1949/1960
D-5270 Gummersbach 1
West Germany
ABSTRACT
After a short description of the staged mixing burner, the report
goes on to deal with the process of direct desulfurization in the furnace
of steam generator by injection of additives around the flame. Further,
the results from tests with a pilot burner and intermediate results from
tests in a full-size boiler are given. Finally, the possible applications
and economic aspects of direct desulfurization are discussed.
Direct desulfurization, is defined as binding of sulfur within the
boiler itself (Figure 1).
• Fluidized-bed firing: A well-known process is fluidized-bed firing,
which is not the subject of this paper.
• Flame firing: We want to take a look at direct desulfurization in
flame firing, which is basically not a new system at all. Beginning as
long ago as the early fifties, people have tried for various reasons to
bind sulfur by injection of calcium-containing additives. In the early
sixties, the idea behind this was to prevent corrosion (S03 binding)
but lately the emphasis has shifted to reduction of noxious emission
(S02 binding), which remains the main concern today.
These investigations have revealed that certain quite definite condi-
tions must be fulfilled for reaction temperature and residence time if
the desulfurization process is to proceed.
1. Additive injection above flame zone: The idea of injecting
the additive above the flame zone is to try to avoid the
sintering of the active surface area of the additive which
occurs at too high a temperature. The disadvantage of this
process is the greatly inferior mixing of the reactants just
at the point where~.the temperature is most favorable for the
desired reaction.''
4-31
-------
CO
S3
a) Fluidized - Bed Combustion
b) Flame Firing
- Additive injection above the flame zone
- Additive mixed with the fuel
- Additive injection around the flame
Steinmiiller-Direct Desulfurization Process "DEP
II
Fig. 1: Direct Desulfurization (Desulfurization in Boiler)
-------
2. Addition of additive to the fuel: If the additives are in-
troducedwith the fuel, they inevitably pass through the
region of highest flame temperatures. Processes of this kind
are therefore restricted to use with fuels of low calorific
value, i.e. with relatively low flame temperatures.
It can therefore be readily understood that in bituminous
coal-firing systems, desulfurization results are much less
satisfactory than with lignite coal-firing systems.
3. Additive injection around the flame: The rest of this paper
will deal with additive injection around the flame.
In this process, which is still in the development stage, all neces-
sary conditions for sulfur binding within the boiler in respect to tem-
perature, mixing and residence time can be more effectively fulfilled,
even for high calorific value coals and at a reasonable cost.
Before we go on to discuss the technical details of this process, we
should like to touch briefly on the physical fundamentals of the direct
desulfurization reaction.
When Ca(OH)2 or CaC03 are employed as additives to capture S02 and
other pollutants, the dehydration or decarbonation (calcination) of the
additives takes place according to reactions (1) and (2) (Figure 2).
Ca(OH)2 -» CaO + H20 (1)
CaC03 • • -» CaO + C02 (2)
The heterogeneous gas/solid reactions which occur during capture of
pollutants in the furnace and downstream of it take place under a constant
overall pressure which lies just below that of atmospheric pressure.
Pollutant reactions (3) to (6) can be expected to occur.
CaO + S02 + 1/202 -> CaS04 . ' (3)
CaO + 2HC1 -»• CaCl2 + H20 (4)
CaO + 2HF •*• CaF2 + H20 (5)
CaO + S03 -» CaS04 (6)
The enthalpies of reactions indicate that dehydration and decarbona-
tion are endothermic and all pollutant capture reactions exothermic. In
Figure 2 the development of the reaction enthalpies for reactions (1) to
(6) is shown as a function of temperature. If the thermal energy required
for the formation of CaO is now compared with that liberated as a result
of sulfation, we find that, for full recovery of the dehydration/decarbon-
ation energy, a calcium efficiency of 22% or 36%, respectively, is re-
quired. Calcium efficiency is the term used to describe the percentage of
injected calcium utilized for stoichiometric binding of the S02.
4-33
-------
I
u>
800 1000 1200 K 1400
Temperature
1600
Fig. 2: Reaction Enthalpy as a function of tenperature
-------
Figure 3 shows the equilibrium curves for the dehydration and de-
carbonation reactions. Since these two reactions do not tend of their own
volition, the reaction temperature must lie higher than the equilibrium
temperature. Assuming a portion of 5% H20 and 15% C02 in the flue gas,
dehydration will take place at temperatures above 385°C and decarbonation
at temperatures above 760°C.
In Figure 4 the S02 partial pressure is presented as a function of
temperature for 02 contents in the flue gas of 1% and 6%, respectively.
Using these equilibrium curves, it is possible, for a given S02 partial
pressure, to make an initial prediction on the possibility of stable
sulfur dioxide binding at the flue gas temperatures obtaining in the
furnace. Sulfur capture is possible when the temperature falls below the
equilibrium temperature predicted by the equilibrium curve for a given
partial pressure of sulfur dioxide. At an overall pressure of 1 bar and
6% excess oxygen in the flue gas, the equilibrium temperature for an S02
concentration of 800 ppm is approximately 1160°C.
These considerations based on chemical thermodynamics show that,
providing additive and flue gases are thoroughly mixed and no sintering of
the particle surface takes place, desulfurization of the flue gases in the
boiler is possible. The question whether a reaction which is theoretical-
ly possible on thermodynamic grounds will actually take place at the
required speed is one of reaction kinetics and cannot therefore be an-
swered on the basis of equilibrium considerations.
Very few data are available on the reactions here described (Figure
5). In the sixties, Wickert carried out laboratory experiments to deter-
mine the effect of temperature upon desulfurization efficiency and his
results are presented here. It can be seen that Ca(OH)2 as an additive is
significantly more effective than CaO and demonstrates maximum sulfur
capture efficiency in the temperature range from 800°C - 1000°C.
!
These theoretical considerations show that direct desulfurization
would also seem to be possible in the furnace of flame-firing systems when
the additive is injected around the burner. An integral component of this
process is the staged mixing burner which has been developed in the last
few years for the purposes of low NO combustion. Its working principles
are outlined in Figure 6.
The design of the staged mixing burner is based on the reliable
circular-type burner, which has enjoyed an excellent track record over
many years. In this burner the p.f./air mixture is introduced into the
furnace via an annular section.
In the staged mixing (SM) burner, the secondary air is divided into a
secondary air flow proper and a staging air flow. The secondary air
enters the furnace via an annular nozzle concentric with the p.f./air
nozzle. Interaction between the p.f./air mixture and the secondary air
flow produces a primary combustion zone in the area of the burner throat
where substoichiometric conditions reign.
4-35
-------
200 300 0)0 500 600 700 800 °
Temperature
Fig. 3: pt-Diagram
4-36
-------
xir
10
900
1000 1100
Temperature °C
Fig. 4: pt-Diagram
4-37
-------
-e-
I
u>
00
1000 1200
Temperature0 C
1400
Fig. 5: Influence of Temperature on Sulfure Capture
-------
Secondary air
Fuel Supply Staging air
Fig. 6: SM Burner Flow Pattern
-------
To ensure full combustion the staging air is nozzled in around the
primary flame into the burnout zone in a number of individual streams.
As has already been reported, this SM burner is capable of reducing
NO formation (and therefore pollutant emission) by more than 50%.
/\
Compared with earlier circular-type burners, the SM burner offers
improved possibilities .for injecting the additive into the furnace in such
a way that the three essential conditions for sulfur capture, viz, temper-
ature, mixing and residence time, can be realized (Figure 7). For the
most part, the additive does not enter that area of the flame where the
highest temperatures are found and thus sintering of the additive surface
can be avoided. The high turbulence around the flame ensures excellent
mixing of the additive with the flue gases. Practically the same maximum
residence time from burner to electrostatic precipitator is available for
sulfur capture as with additive injection mixed with the fuel, but the
disadvantages of the latter method are avoided.
To discover the attainable sulfur capture levels and optimize the
process, Steinmuller has carried out experiments at the International
Flame Research Foundation (IFRF). The most important results obtained are
listed below.
The following relevant parameters were examined (Figure 8):
• Fuel: Natural gas doped with S02 or H2S and three different
types of bituminous coal.
• Additive: Ca(OH)2, CaC03 and activated Ca(OH)2.
• Molar ratio Ca/S: From 1 to 4.
• Point of additive injection on burner: With staging air,
with fuel and via nozzles in the external recirculation zone
(ERZ).
• Partial air flow rates at burner: Ratios of air (n) and air
in the primary combustion zone (n from 0.6 to n).
• Flue gas temperature: From 900°C to 1100°C.'
• Swirl: From minimum to maximum.
• Load: From 60% to 150%.
• Velocity of staging air: Low and high.
Figure 9 shows the influence of the type of additive, the point of
additive injection and the sulfur content in the fuel on~ the degree of
desulfurization as a function of the molar ratio Ca/S.
4-40
-------
.p-
I
.p-
Staging Air with Additive
Internal Recirculation Zone
Primary Combustion Zone
\ ^ y x Mixing and Burnout Zone
'Outer Recirculation Zone
Fig. 7: Direct Desulfurization through Additive Injection around the SM Burner
-------
Fuel
Natural gas doped with SOz ,3different
bituminous coals,Heavy Fuel Oil
Additive
Ca(OH)2,CaC03,Ca(OH)2- active
Molar Ratio Ca/S
from 1 to A-
Additive injection location on burner
Staging air, fuel,tubes in ERZ
Airflow ratios at burner
n,np = 0,6 to n
Furnace Temperature
from 900°C to 1100°C
Swirl
from minimum to maximum
Load
from 60% to 150%
Velocity of staging air
low, high
Fig. 8: Parameters tested at IFRF/Ijmuiden
4-42
-------
a)
b)
IW
80
40
7ft
L\J
0,
Additive with staging
S- Content :3%
Activ
/
/
ited
CaP
s
^
Ca(O
\J
\\L/\
OJr
80
60
40
20
0
0123
Ca/S
0 1 2 3
Ca/S
Fig. 9: Sulfur capture nS02 as a function of molar ration Ca/S with natural gas
-------
100
80
Blumenthol
>
AugusteViktorio/
Natural
gas 3% S
Fuel
S02 Concentration in
flue gas mg/m3 (STPdry)
Blumenthal
Auguste Viktoria
Gottelborn
Natural gas(3%S)
7712
2385
1810
4693
Fig. 10: Sulfur capture nS02 as a function of the molar ration CA/S
4-44
-------
The type of additive has a very definite influence on sulfur capture.
CaC03 has a lower affinity for sulfur than Ca(OH)2. Activated Ca(OH)2 by
contrast only demonstrated better desulfurization potential when the S02
concentration in the flue gas was high. Activated calcium hydroxide is
manufactured by grinding quicklime with fatty acids and then slaking with
a calcium chloride solution, so that the finished product contains around
2% CaCl2.
This figure shows the influence of the point of additive injection on
the degree of desulfurization. Injection into the flame core produces
significantly worse desulfurization than when the additive is injected
into the outer recirculation zone of the flame, e.g., with the staging
air. This tendency is to lower sulfur capture when Ca(OH)2 is injected
with the transport air can be attributed to the longer residence time of
the additive particles in the hot flame zone. As a result, the activity
of the additive is considerably reduced by sintering of the surface and
the formation of thermodynamically stable CaS04 prevented.
Sulfur capture is more effective for high sulfur contents in the
fuel, at least up to a certain concentration of S02. This can largely be
explained, as can be seen from the pT diagram for sulfation, by the higher
dissociation temperature of CaS04.
«
The greater degree of S02 diffusion towards the additive surface also
plays a role.
It can be seen from these diagrams that the levels of desulfurization
achieved with doped natural gas were unsatisfactory, both from a technical
and an economic viewpoint. Continuation of the experiments with bitumi-
nous coals seemed nevertheless justified for two reasons: First of all,
earlier tests with coal showed better sulfur capture results even when the
additive was injected with the primary air; and, secondly, we achieved an
improvement of sulfur capture by injecting additive with the staging air.
The results in Figure 10 show that this conclusion was fully justi-
fied. Here the maximum degree of desulfurization measured with various
different fuels is presented as a function of the molar ratio Ca/S, using
Ca(OH)2 as additive.
Injective of the additive was carried out in the outer recirculation
zone of the flame. For Ca/S = 2, the following desulfurization levels
were measured:
• 50% for Ruhr coal from the Auguste Viktoria pit with a maxi-
mum S02 concentration in the raw gas of 2385 mg/m3 (SPT dry).
• 57% for Saar coal from the Gb'ttelborn pit with a maximum S02
concentration in the raw gas of 1810 mg/m3 (STP dry).
• 65% for Ruhr coal from the Blumenthal pit with a maximum S02
concentration in the raw gas of 7712 mg/m3 (STP dry).
4-45
-------
• As a comparison, 41% for natural gas with an .S02 concentra-
tion of 4693 mg/m3 (STP dry).
These results show that for coal firing systems, S02 capture achieves
levels at which it becomes a thoroughly realistic prospect in process
terms. One of the chief reasons for this shift is the different tempera-
ture/time history of the additives in coal-firing systems, as shown by the
detailed measurements. On the basis of the pilot experiments we carried
out an economic feasibility calculation to see if further development of
the process was justified.
The desulfurization costs for a 700 MW , power station were worked
out using the direct desulfurization process (Figure 11). These costs
were compared with those for a "wet" system flue gas desulfurization plant
(FGD) with gypsum as the end product.
The following design data were utilized for the projecting of each
plant:
Fuel: Bituminous coal
Sulphur content 1%
Ash 7%
Flue gas: S02 concentration 2200 mg/m3 (STP dry)
Clean gas: Desulfurization efficiency 72%
S02 concentration 616 mg/m3 (STP dry)
For FGD 80% of the gas flow is treated with a desul-
furization efficiency of 90%.
Additive: For DEP: Ca(OH)2
Molar ratio Ca/S = 4
For FGD: CaC03
Molar ratio Ca/S = 1.05
The total costs are made up of investment and running costs. Plant
costs for DEP are mainly incurred for the additive dosing system (bunker,
fan, dosing equipment, pipework, control system, etc.), the increased
number of sootblowers and the expansion of the electrostatic precipitator
and de-ashing systems, etc. FGD personnel costs are 3:1 by comparison
with DEP, although this naturally depends on the degree of automation of
the FGD system.
In calculating the investment costs per year, assuming a basic in-
terest rate of 8% per year for the investment capital and an amortization
period of 15 years, a redemption factor of 11.7% was calculated.
In addition to the redemption factor, which allows for the annual
depreciation and payment of interest on the plant capital, 10% of plant
4-46
-------
Fuel: Bituminous coal
S- content 1%
Ash 7%
Rue gas: SCh 2200 mg/m3 (STPdry)
Clean gas: ?so2 72%
SOz 616mg/m3
FGD; 80% of flue gas
with ? so2 = 90%
Additive: For DEP: Ca(OHh;Ca/S = 4
For FGD:CaCQ3;Ca/S = 1,
Total Costs: Investment + Operation
Investment Costs: Additive dosing system,
Increased number of sootblowers, Increase in
size of ESP and deashing system.
Interest rate: 8%, Amortisation period: 15 years
«* Redemption f actor: 11,7 %+10% for taxes
Administration and repairs .
Operating Costs : Additive. solid effluents
disposal ;sootblowing and
electric power
Fig. 11: Boundary Conditions for Economic Analysis
4-47
-------
capital was reckoned to be devoted to administration, repairs and taxes,
so that the investment costs per year amount to 21.7% of the plant capital
investment.
For DEP the operating costs make up the major proportion of the
expense, consisting chiefly of those for additive, disposal, cleaning the
heating surfaces and power. As far as disposal is concerned, it is as-
sumed that fly ash would have been dumped even without direct desulfuri-
zation.
Operating costs are some 50% higher than for FGD but the plant costs
are only about 10% of those for the latter.
Figure 12 shows the percentage reduction of desulfurization costs
when using DEP instead of FGD with gypsum as the end product, given the
boundary conditions which have been assumed.
The reduction in costs is entered for full load, depending on the
number of operating hours per year. Because of the high investment costs
associated with FGD, this example shows that the smaller the number of
operating hours per year, the more significant is the cost saving to be
obtained using DEP. For a power station operating 4000 h/annum, the
reduction in costs according to this calculation would be 30%.
The economic advantages with DEP increase as the sulfur content of
the coal and the size of the plant decrease. This process would seem to
be of particular interest for older power stations, where for reasons of
space, it is impossible to install FGD equipment but where, in most cases,
it is relatively easy to reequip the plants for DEP. Keeping even lower
pollutant limits would seem to make a coupling of the two systems---DEP and
FGD--desirable, but this problem must be the subject of a separate ap-
proach.
After the successful conclusion of the pilot experiments and investi-
gation of the economic viability, it was decided, as the next stage of
progress towards commercial application, to confirm the results obtained
from the pilot experiments on a full-size steam generator and demonstrate
the functional effectiveness of the process. See Figure 13.
This full-scale technical demonstration will concentrate primarily on
the investigation of the following points:
a. Fouling behavior of the boiler furnace.
b. Fouling of the convective banks.
c. Alteration in the ash and its utilization potential.
d. Fouling of the regenerative air heater.
4-48
-------
Qf\ -T- - - T
80
%
7fl
/v
Afl-
Ov
^fl
JV
c
g
1 40
"8
CfL
•ft 30
o
LJ
20
10
0
v
\
S
\
I
700
S-Cont
\
I
MW Uni
ent of i
'
v
\
t
:OQl:1%
\
X
3 2000 4000 6000
Operating hours per annum
Fig. 12: Cost reduction when applying Direct Desulfurization instead of wet FGD
4-49
-------
Further investigations using a complete
boiler will be carried out to confirm the
results so for obtained and explore the
following fields:
- Efficiency of desulfurization with
multi - burner combustion
-Slagging and fouling behaviour of the boiler
-Possibilities for ash utilization
-Efficiency of the Precipitator
•Scaling and fouling behaviour of
the air heater
Fig. 13: Further Work
4-50
-------
e. Dust-collecting efficiency of the electrostatic precipi-
tator.
f. Degree of desulfurization and calcium utilization.
For the investigation of the above points, the Saarbergwerke AG have
kindly placed the 700 MW ..power station Weiher III at our disposal.
Because of the already installed staged mixing burners, Weiher III is
extremely suitable for demonstrating direct desulfurization and NO re-
duction with comparatively little modification.
Figure 14 shows the schematic diagram of the dosing system for the
Steinmiiller Direct Desulfurization Process applied to Weiher III. The
furnace of this boiler is divided by a division wall into two combustion
chambers and has a total of 24 staged mixing burners, 12 each in the front
and rear walls. The 12 burners in these walls are arranged in 3 planes.
Groups of 4 burners are each served by one mill and are supplied with
additive by a dosing unit. The additive is transported by the compressed
air.
Since availability of the boiler is not to be put at risk in any way,
initially only two burners on the middle plane of the left combustion
chamber were modified to investigate the potential operating problems.
The object of this first phase was primarily to investigate fouling in the
immediate area of the burner, in the furnace and in the convective sec-
tion, as well as the alteration in the ash. This phase has meanwhile been
completed with thoroughly satisfactory results, although there are nat-
urally still points which remain open.
In the following sections the relevant results are presented.
Figure 15 shows the rear wall of the left combustion chamber. Lime-
stone was blown in round the two burners of the centre plane. The view
remained the same after the experiments as well, i.e., there were no signs
of an increase in fouling and slagging.
The furnace was completely inspected and showed no increase in
slagging. Figure 16 shows slagging on the furnace tubes. On the left is
the left-hand side wall of the combustion chamber with limestone injection
and on the right-hand side wall of the combustion chamber without lime-
stone injection. On both photographs light hard deposits of roughly
similar thickness on the side exposed to the flue gas flow can be seen.
Figure 17 shows the first convective heating surfaces after the
furnace. A comparison shows that the tubes on the left-hand side of the
furnace, where the additive was blown in, exhibit more fouling. But this
fouling only appeared in the initial bank and was of a consistency easily
dealt with by appropriate soot blowing measures. During the experiments,
a total of 300t of slaked lime and 60t of powdered limestone were in-
jected. The experiments continued for four weeks in all.
4-51
-------
I
Ln
1-0
Ambient air
v T
A A A
Y Y Y
^ A *>
Additiv Air
_J I
Additive-Sil
V= 1000m 3
O O O O O
.a .a
a a
n :a
fl
a a
the burners on rear wall
Fig. 14: Schematic of Additive Dosing system for Steirimuller Direct Desulfurization
process "DEP"
-------
Fig. 15: View on the combustion chamber wall
4-53
-------
With calcium injection
Boiler height: 49 m
Without calcium injection
Boiler height: 49 m
Fig. 16: Slagging in furnace
With calcium injection
Boiler height: 74 m
Without calcium injection
Boiler height: 74 m
Fig. 17: Fouling on convective heating surfaces
4-54
-------
Figure 18 shows the x-ray diffraction analysis of fly-ash at full
load without limestone, fly ash at full load with limestone in the propor-
tion Ca/S = 0.33, fly ash at 50% load with additive again in the propor-
tion Ca/S = 0.33 and superheater fouling deposits with limestone injec-
tion. The components are identified in the order shown.
In the uncontaminated fly ash there is a certain degree of sulfur
capture by calcium naturally occurring in the coal. A very small amount
of free CaO is also present.
Injection of limestone around the flame increases the degree of
sulfur capture. CaS04 is no longer the last component but one; instead it
is now the last component but 2 and no CaS03 can be found. This is im-
portant for further processing and possible dumping.
Mainly because of the lower furnace temperatures involved, sulfur
capture is significantly better at low load. CaS04 moves up one place in
comparison to its listing at full load.
The superheater fouling deposits on the boiler side where limestone
was injected contain the highest quantities of CaS04 and the level of free
CaO is very low.
Figure 19 shows the theoretically calculated ash composition for
direct desulfurization on all burners. The Gb'ttelborn coal from the Saar
region used at Weiher power station is reckoned to have an ash content of
DO/
o/%.
For a molar ratio Ca/S = 2, the original fly ash makes up 61% of the
total and calcium products are composed as follows: CaO = 21%,
CaS04 = 17%, CaCl2 = 0.5% and CaF2 = 0.1%. At a molar ratio of Ca/S = 4,
the original fly ash makes up only 45% of the total fly ash, the rest
being calcium products.
These residues are naturally not officially certifiable for use, like
a normal fly"ash. However, Steinmuller is investigating the possibility
of utilizing these residues in the building material industry.
CURRENT ACTIVITIES
(a) Weiher III Power Station
After the preliminary tests on two burners proved positive, we are
now equipping all 24 for direct desulfurization. The tests should
begin at the end of this year.
(b) 30 MW Burner at EER in the USA
A 30 MW burner is being tested this year in the United States.
Besides Gb'ttelborn coal, two American bituminous coals are being
used.
4-55
-------
Fly ash
at full load
Ca/S=0
Si 02
3A1203-2S102
Fe203
Fe30<,
CaSO*,
CaO
Amorphous
components
Fly ash
at full load
Ca/S=0,33
Si 02
CaO
3Al20s-2Si02
CaSO<,
Fe203
FesO^
Amorphous
components
Fly ash
at half load
Ca/S = 0,33
Si 02
CaO
CaSO*,
3A1203 2 Si 02
Fe?03
Fes04
Amorphous
components
Deposits on superheater
with lime injection
CaS04
Fe203
Si 02
3Ab032Si02
K2Mg2(S003
CaO
MgO
Fe30<>
Fig. 18: X-Ray Diffraction Analysis
-------
Gbttelborn Coal
Fly ash
CaO
CaSCk
CaCl2
CaF2
-------
(c) DeNO /DeSO Firing System in Canada
/\ J\
We have already completed a study on a DeNO /DeSO firing system for
CFB Gagetown in Canada and have received a contract for the hardware.
The commissioning should take place in the middle of next year.
(d) Combustion of Petroleum Coke at IFRF/Ijmuiden
Direct desulfurization in petroleum coke combustion has been experi-
mentally investigated on a 3 MW burner at IFRF/Ijmuiden, as part of a
larger R&D program.
(e) Direct Desulfurization on Travelling-Grate Stoker Firing
We have recently booked an order for applying in-furnace desulfuri-
zation to a steam generator of 100 MW thermal capacity equipped with
a travelling-grate stoker firing. The commissioning should take
place in the middle of next year.
Thus, by the middle of next year, we shall have gained a great deal
of additional experience in the area of direct desulfurization.
4-58
-------
SESSION 5: DUAL ALKALI
Chairman: Norman Kaplan
Industrial Environmental Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park,'NC
-------
UTILITY DOUBLE ALKALI OPERATING EXPERIENCE
D. L. Clancy, R. J. Grant, L. K. Legatsky,
J. H. Wilhelm, B. A. Wrobel
-------
UTILITY DOUBLE ALKALI OPERATING EXPERIENCE
by: Dennis L. Clancy
Southern Indiana Gas & Electric Company
Evansville, Indiana
Richard J. Grant
Central Illinois Public Service Company
Springfield, Illinois
L. Karl Legatski
FMC Corporation
Schaumburg, Illinois
James H. Wilhelm
Codan Associates
Sandy, Utah
Beth A. Wrobel
Northern Indiana Public Service Company
Wheatfield, Indiana
ABSTRACT
On April 6, 1983, Northern Indiana Public Service Company started up
its Schahfer Unit 17, the first high sulfur coal boiler required to be in
compliance with the 90% S02 removal provision of the 1979 revisions to
the New Source Performance Standards. This is the fourth utility double
alkali FGD system to come on line in the last four years. All have
consistently met their S02 performance criteria. This paper summarizes
cost and availability data for three of these systems. While capital
costs vary with site specific design criteria, operating costs exclusive
of capital charges are typically 4 mills/kWh. Availabilities for the
double alkali systems as a group are consistently higher than for direct
limestone. Unique features of the NIPSCO design and operation are
discussed as well as some important recent improvements implemented at
the Southern Indiana Gas and Electric and. Central Illinois Public Service
Company installations.
5-1
-------
INTRODUCTION
The double alkali process is a second generation throw-away FGD
technology that was first offered commercially to the utility industry in
1975. A fundamental characteristic of the process is its capability to
achieve very high efficiencies reliably even on very high sulfur dioxide
concentrations. This makes the process particularly appropriate for high
sulfur Eastern coals, especially in light of the 1979 revisions to the New
Source Performance Standards which mandate 90% removal on a continuous
basis.
Since 1975, six double alkali systems have been sold, all in the high
sulfur coal Midwest. Three of these systems were started up in 1979, and
the fourth, Northern Indiana Public Service Company's R. M. Schahfer Unit
17, was started up this spring. The two remaining units are scheduled to
start-up in 1986.
The purpose of this paper is to consolidate and bring up to date the
noteworthy developments of the past two years at these installations.
Special features of the NIPSCO system design are reviewed along with the
early performance and start-up results. In addition, two of the three
systems that have been in operation since 1979 (Southern Indiana Gas &
Electric Company's [SIGECO] A. B. Brown Unit 1 and Central Illinois Public
Service Company's [CIPS] Newton Unit 1) have undergone important
modifications in the last few years. These modifications are discussed and
the key operating results (emissions performance, reliability, and cost) are
presented for the three systems as a group.
SCHAHFER UNIT 17 DESIGN
DESIGN CRITERIA
NIPSCO's Rollin M. Schahfer Generating Station is located on the
Kankakee River near Wheatfield in northwestern Indiana. The two newest
units at the station, 17 and 18, have a gross generating capacity of 393
megawatts each. These units, for which construction began in 1980, are the
first ones built under the revised New Source Performance Standards
provision requiring 90% sulfur dioxide removal on a continuous 30-day
rolling average basis. Each unit has a design heat input of 3,967 x
Btu's per hour and a flue gas flow at the FGD system inlet of 1,591,500 acfm
at 327°F. The performance coal data are given in Table 1.
5-2
-------
TABLE 1. PERFORMANCE COAL DATA
(Percent by weight)
Typical Range
As Rec'd Dry Basis As Received
Moisture 11.25 - 9.70 - 13.10
Carbon 61.64 69.45 57.36 - 63.09
Hydrogen 4.53 5.10 4.33 - 4.90
Nitrogen 1.12 1.26 1.06 - 1.31
Chlorine 0.02 0.02 0.01 - 0.02
Sulfur 3.20 3.61 2.60 - 3.57
Oxygen (diff.) 7.90 8.90 6.99 - 8.43
Ash 10.34 11.66 7.80 - 13.90
Total 100.00 100.00
Heating Value (Btu/lb) 11,085 12,490 10,500 11,600
Key requirements of the specificatori include:
1. Four absorbers per unit (three operating and one spare).
2. All alloy construction in the absorbers and downstream ductwork
with no bypass.
3. Thirty degrees Fahrenheit of inline reheat.
4. Two complete regeneration trains per unit, each capable of
supporting the system at 67 percent of full load using the maximum
sulfur fuel.
5. Waste product stabilization designed to allow blending of all of
the fly ash from Units 17 and 18 plus Units 14 and 15 with the FGD
waste product.
PROCESS DESIGN
The NIPSCO system utilizes the conventional concentrated double alkali
process furnished by FMC Corporation. Figure 1 is a schematic
representation of the process. Sulfur dioxide is absorbed according to the
following reaction:
N32S03 + S02 + H20 —»- 2NaHS03 (1)
Sodium bisulfite formed in the absorber is removed from the absorber
loop as a bleed stream to a separate lime mixing tank where it is reacted
with hydrated lime to precipitate primarily calcium sulfite according to the
following reaction:
2NaHS03 + Ca(OH)2 —•- Ka^O^
5-3
-------
BV-MOOUCT
3TOHAOEPILE
STACKER
Figure 1. NIPSCO FGD PROCESS SCHEMATIC
The sodium sulfite that is regenerated by the above reaction is returned
to the scrubbing process for reuse. The small amount of sodium that is not
recoverable is replaced through the addition of soda ash to the scrubbing
system. The pH of the sodium scrubbing solution is maintained at a level
which buffers the scrubbing system and eliminates calcium scaling.
FACILITIES DESCRIPTION
The NIPSCO facility is depicted schematically in Figure 1 and in plan
view in Figure 2. The gas side control loops for Unit 17 and the future
Unit 18 are independent of each other and under the control of the boiler
control room operating personnel. The regeneration portion of the process,
however, is integrated for Units 17 and 18 and under the control of FGD area
operating personnel in an area several hundred feet away from the absorber
area. The regeneration facility will ultimately contain four independent
regeneration trains, each capable of accommodating two-thirds of the
capacity of each boiler when operating on its maximum sulfur fuel. Under
normal sulfur and load operating conditions, the regeneration area has
effectively 100% spare capacity.
5-4
-------
L/l
I
U1
FOD BATTERY LIM
-LIME SILO
y V
UNIT 17
Figure 2. NIPSCO FGD AREA PLAN
-------
The flue gas leaves the boiler and enters the rigid frame electrostatic
precipitator. Immediately downstream of the precipitator are the induced
draft fans which draw the flue gas through the boiler and pressurize the
FGD system. Flue gas enters the bottom of the 32-foot-diameter absorbers.
These absorbers are constructed of 317LM stainless steel and consist of
four disc and donut absorption stages. A concentrated solution of sodium
sulfite is pumped from the sump in the base of each absorber to the top of
the absorber at a liquid-to-gas ratio of approximately 10 gal/Macf. The
scrubbing solution cascades through the absorption stages while the gas
flows countercurrent through the stages, to the top of the absorber. Two
100% capacity alloy (317LM) pumps per absorber continuously recirculate the
solution and bleed off the spent solution to the regeneration area.
Installed in the upper level of each absorber is a mist eliminator
constructed of 904L stainless steel. From the mist eliminator, gas exits
each absorber and passes through an in-line steam reheater, constructed of
Inconel 625, which provides 30°F of reheat. From there the gas enters a
common outlet duct to the 500-foot-tall acid brick-lined chimney.
Proportionate to the amount of SC>2 collected, the absorber bleed
stream is pumped to one of the two lime mixing tanks located in the
regeneration building. There, slaked lime from the six slakers (2 spare)
is added to the continuously agitated tanks to regenerate the sodium
sulfite and form by-product solids. Each lime tank overflows by gravity to
a 90-foot-diameter thickener which concentrates the calcium sulfite
solids. Regenerated scrubbing solution overflows by gravity to a single
regeneration surge tank, from which it is pumped on demand to the
absorbers.
The thickener underflow is pumped by air-operated diaphragm pumps to
four 33% capacity rotary drum vacuum filters. The filter cake is washed on
the vacuum filters with heated water to recover additional sodium. Sodium
remaining in the filter cake is replaced by adding soda ash (Na2CC>3) in
the regeneration return line from the surge tank.
The filter cake is transported by conveyor to one of the four mixers.
The cake is mixed with flyash and pulverized lime to increase strength and
reduce permeability.
The FGD by-product is conveyed to one of two radial stackers, serving
both Units 17 and 18, and is discharged to a temporary disposal area. Then
the material is loaded onto 50-ton off-road trucks for hauling to the
on-site product disposal area located about one mile east of Unit 17. The
material is placed and compacted to a maximum height of 67 feet. The
approximately 200 acre disposal area will be covered with soil and seeded
as each.section is completed.
5-6
-------
NIPSCO START-UP
NIPSCO's Schahfer Unit 17 started up on April 6, 1983, and the FGD
system was first operated on April 10. The unit was declared commercial on
April 28. Unfortunately, in addition to the FGD area problems which will be
discussed below, there have been numerous outages initiated by events
upstream of the FGD system. These outages have prevented the collection of
much meaningful data on cost or reliability in the FGD area. However,
official plant records indicate that the FGD system has theoretically
constrained the boiler load only 5.3% of the time so far.
Aside from the usual list of start-up trials and tribulations, there*are
a few aspects of the early operation that are noteworthy or unique.
ABSORBER PERFORMANCE
The disc contactor scrubbers are guaranteed to achieve 90% sulfur
dioxide removal with three stages in operation at a design liquid-to-gas
ratio of 10 gal/Macf. However, early spot checks based on the S02 monitor
indicated that efficiency was frequently in the 80's. Closer examination
showed that the design efficiency was not being achieved under certain
combinations of flow conditions, apparently due to maldistribution of the
flue gas. This had the effect of short-circuiting the bottom stage of the
absorber. In June and July, three different inlet baffling arrangements
were tested. The best of these arrangements was implemented on all
absorbers in August and unofficial performance tests on the individual
absorbers showed that 90% efficiency was being achieved routinely with three
stages. These results are given in Figure 3, which shows collection
efficiency as a function of number of stages in operation and percent of
design flow rate. These results correlate excellently with an existing
design model which expresses number of transfer units as a function of
liquid-to-gas ratio, superficial gas velocity, and certain dimensions of the
absorber. However, the cause of the gas maldistribution problem is still
something of a mystery in that other existing absorbers of a very similar
design have not experienced this problem and there is no significant
maldistribution of the velocity profile entering the absorber.
VIBRATION
During start-up, severe vibration was noted in the outlet ductwork and
reheater casing of the FGD system. Initially, effects of the vibration were
limited to the loss of the outlet instrumentation (pressure and temperature)
due to the instrumentation and instrumentation taps being cracked and/or
broken. Next, a crack was noticed in the absorber shell. The crack had
propagated from the toe of an internal weld. Further investigation
identified cracks throughout the outlet ductwork from the absorber outlet
elbow to the guillotine damper. All cracks started in an area where there
was internal or stitch welding performed. A dye penetrant test was
conducted on the outlet ductwork to ensure all cracks were found and
repaired.
5-7
-------
100
90
o
o
C\J
o
oo
80
A
A - 3 Stages
- Design Model
50
60
70 80 90
FLOW RATE (% OF DESIGN)
100
110
Figure 3. UNOFFICIAL COLLECTION EFFICIENCY VS. GAS FLOW RATE.
Extensive testing was carried out in conjunction with Stone and Webster
and NIPSCO to determine the cause of the vibration. Testing results
centered around the reheater and the turning vanes in the outlet elbow.
Vibration was occurring at the two resonant frequencies of the ductwork.
Further testing indicated that the excitation was flow related. The larger
of the two turning vanes was removed as shown in Figure 4, and blanking
plates were installed at the top of the reheater housing. These changes
greatly reduced the vibration, but it reappeared with somewhat lower
intensity.
As in the first instance, the source of the vibration could not be
identified. It appeared to be amplified by the steam tube reheaters and
certain dimensions of the ductwork that approximated multiples of the
wavelengths corresponding to the natural frequency of the structure. On
this occasion it was found during in situ air tests that there was boundary
layer separation at the diverging sidewall section of the ductwork between
the outlet elbow and the reheater (see Figure 4). Four foot wide test
panels installed at each side of the inlet to the reheater perpendicular to
the gas stream eliminated the vibration completely but were deemed
5-8
-------
impractical because of their obvious adverse effects on the reheater
performance. An alternative solution found by trial and error turned out to
be the installation of a perforated plate along the bottom of the outlet of
the elbow between the smaller turning vane (see Figure 4) and the bottom of
the duct. However, there is still no satisfactory explanation for why this
modification worked.
OUTLET PLENUM
REHEATER
BOUNDARY LAYER
SEPARATION
UPPER TURNING
VANE REMOVED
GUILLOTINE
PITTING CORROSION
PERFORATED PLATE
INSTALLED
ABSORBER
Figure' 4. ABSORBER OUTLET DUCTWORK ISOMETRIC VIEW
5-9
-------
MATERIALS OF CONSTRUCTION
During the engineering of the FGD system, NIPSCO evaluated two materials
of construction alternatives in some detail. The first alternative was
flake-glass-filled polyester resin linings for the absorbers with full
bypass and quench protection for the lining. The second alternative was all
allpy construction with no bypass. For the relatively high pH of the double
alkali process (6 to 7) and the relatively low chloride concentrations
anticipated (less than 6,000 ppm), 317LM stainless steel was selected as the
principal alloy for the absorbers. Based on this alloy selection, the alloy
absorber option evaluated better than lining when the implications of the
quench system, bypass, and associated dampers and draft control requirements
were taken into account.
An inspection of the absorber and ductwork internals in early August
revealed two areas of concern. First, the cracks in the outlet duct
mentioned in the preceding section were examined more closely. Test
specimens were removed and analyzed, but corrosion of the fracture surface
made it impossible to determine for certain if the cracks were initiated by
stress corrosion or fatigue. It is clear that the cracks propagated in
fatigue and the circumstantial evidence of the vibration and the nature of
the cracks strongly suggest that they were fatigue initiated. All of these
cracks have been ground out and repaired and only time will tell if the
elimination of the vibration has also eliminated the cracking.
A second area of concern is the small section of ductwork between the
reheater and the outlet isolation damper for each absorber module (about 7%
of the total outlet duct area; see Figure 4). The converging sidewalls in
this relatively short segment of ductwork show varied degrees of pitting.
These converging walls, which experience impingement of the flue gas stream
and intermittent saturated and unsaturated conditions due to the
intermittent operation during start-up, are obviously high potential sites
for corrosion. Test coupons have been installed in one of these duct
segments and will be monitored closely to provide data on possible alternate
materials in this area.
The outlet plenum and the absorbers themselves are overall in excellent
condition with only one or two localized areas of pitting identified at this
time.
CIPS UPDATE
Central Illinois Public Service Company's, Newton Unit 1 FGD system
distinguished itself in 1982 with a 97.8% availability record, the best
record of any high sulfur coal FGD system reporting to EPA for the full
year. Recent developments at CIPS that are of general interest are
discussed below.
5-10
-------
PROCESS MODIFICATIONS
The Newton double alkali FGD system began operation in a double loop
mode with a unique precooler loop which removed chlorides from the flue gas
before scrubbing for S02« The objective was to reduce reagent consumption
and protect the system from corrosion. Extensive problems associated with
this loop greatly reduced overall system availability during the first eight
months following start-up. These problems were substantially eliminated by
CIPS through modifications which eliminated the chloride removal function of
the precooler loop.
CIPS subsequently applied for a patent to cover these system
modifications, which included:
1. Changes in system pH;
2. Reaction residence times; and
3. Alteration of agitation and tip speed in the reaction tanks.
In May of this year, CIPS received Official Notice of Allowance for the
patent application and is currently working through its patent attorneys to
finalize the paperwork on the patent, which is expected to issue October 25,
1983.
SYSTEM PERFORMANCE
The Newton system was designed to maintain compliance with Subpart D of
the New Source Performance Standards. As such, it is required to meet a 1.2
Ib/MMBtu S02 emission limit, but has no requirement for removal
efficiency. Accordingly, no inlet S02 monitors are used on the system,
and there have been no Method 6 inlet/outlet efficiency tests conducted.
•
The coal burned in Unit 1 has an average sulfur content of approximately
2.8% (or about 4.5 Ib/MMBtu S02).
In conducting tests associated with optimization of scrubber performance
over the last -several years, it has been demonstrated that the absorbers can
achieve outlet loadings of less than 0.6 Ib/MMBtu. In addition, between
September 1980 and June 1982, five audits conducted by a US EPA contractor
demonstrated that outlet levels in the 0.3-0.6 Ib/MMBtu range had been
achieved.
LIMESTONE CONVERSION
CIPS is presently studying the feasibility of converting the lime-based
double alkali system at Newton to limestone-based. Obviously, such a
conversion would have significant cost benefits by substantially reducing
the reagent cost component of the operation. Concerns include the following
impacts such a conversion might have on the process itself:
5-11
-------
1. Will efficiency and system control be similar to present operation?
2. How will a conversion affect filter cake solids concentrations and
quality?
3. Impact on soda ash consumption and regenerated liquor quality.
All testing to date indicates such a conversion would be quite cost
beneficial and technically sound.
SIGECO UPDATE
In spite of the excellent performance record in terms of compliance
operation, availability, and cost compiled at A. B. Brown Station in its
four years of operation, SIGECO has continued to strive to reduce its costs
and further improve its reliability. In the last few years, this effort has
centered on improvements in equipmment design for those components which
caused the greatest percentage of the forced outages. Additional efforts
have centered on improvements in filter cake quality as has been reported at
the last two symposia.* Solids quality affects the quantity of material,
placement cost in the landfill, and the maintenance and reliability in the
material handling segment of the system. Additionally it relates directly
to reagent consumption.
In 1982, SIGECO initiated a program to improve the regeneration section
of the system. This program centered around crystal growth studies
conducted by Codan Associates and a series of equipment design changes
engineered by FMC and implemented by SIGECO plant engineering.
EQUIPMENT MODIFICATIONS
e
In late 1982 the design changes planned for A. B. Brown were detailed by
FMC. Beginning in the second quarter of 1983 they were implemented by
SIGECO plant engineering. The status of the major changes is summarized
below:
1. The existing lime reactor was modified in April in accordance with
Codan's recommendations. The average moisture content of the cake
for the month prior to the shutdown was 48.1%. For the month
immediately after the system was restarted in May, the moisture
content averaged 41.7%. This 6.4% increase in solids content
resulted in relatively dramatic improvement in the cake consistency.
Durkin, T. H., Wilhelm, J. H., Boward, W. L., "Recent Operating Results
with the Double Alkali Process," EPA Symposium on Flue Gas
Desulfurization, Hollywood, FL, May 1982.
Van Meter, J. A., Durkin, T. H., Legatski, L. K., "Operating Experience
with the FMC Double Alkali Process," EPA Symposium on Flue Gas
Desulfurization, Houston, TX, October 1980.
5-12
-------
2. A horizontal bed filter was installed in summer 1983 and started up
in early September. The horizontal filter replaces one of the
three original rotary vacuum filters, though it has sufficient
capacity to carry the system at full load by itself. The other two
rotary filters are being retained as back-up. The filter was
started up early in September, and it further improved the solids
content of the cake while increasing filtration rates to 75 - 100
lb/hr-ft2.
3. The original thickener rake has been replaced with a power lift
rake, and other modifications have been made to allow the thickener
to function as a reactor as well as a thickener. Late this year
the existing lime reactor will be converted into a lime slurry
storage tank, and the reaction step will be moved to the
thickener. The availability of a lime slurry storage tank and the
power lift thickener should eliminate the sources of two of the
most important causes of forced outages in the past.
SIGECO is optimistic that these changes will meet its stated objectives*
of improving the availability to 99% from the recent availability of 91.1%
and further reducing current operating costs of 4 mills/kWh by 0.5 - 0.7
mills/kWh.
CRYSTAL GROWTH STUDIES
The characterization of waste solids from FGD systems, both direct lime
and limestone and double alkali, is a deceptively complex subject.
Relatively large differences in solids quality between different commercial
FGD systems have been noted but not explained in any systematic way. The
problem is complicated by the lack of clear-cut criteria for solids
quality. For example, percent solids is not by itself a meaningful
criterion. A forced oxidation system producing a calcium sulfate (gypsum)
product at 80% solids generates the same mass of material for disposal as an
unoxidized calcium sulfite producing system that generates 60% solids.
Moreover, a higher percent solids material may or may not be superior in
terms of handleability, bearing strength or permeability. These latter
characteristics depend primarily on the size and shape of the crystals
produced rather than just solids content. However, when comparing cake
quality in chemically identical situations, percent solids is usually a fair
measure of solids quality.*
It has been found that the crystal size and shape in double alkali FGD
are primarily influenced by the following variables:
1. Reactor residence time
2. Shear rate of the agitator(s)
3. pH
4. Concentration of solids
5. Magnesium concentration
6. Entry point of the reactants
* Durkin, et al., op. cit.
5-13
-------
7. Lime slaker design and operation
8. Chloride concentration
9. Sulfate concentration
10. Sulfite and bisulfite concentrations
11. The presence of other crystal modifiers
12. Chemical composition of the lime used
It is outside the scope of this paper to review the influence of each of
these variables. However, it is important to note that the current study by
Codan showed that through minor changes in agitator speeds, operating pH and
points of addition of reactants to the reactors, the crystal size and
dewatering characteristics could be improved substantially. For the A. B.
Brown system, as with CIPS1 Newton station which was modified a few years
ago, it has been shown that the filter cake solids content can be increased
from 10 to 20 weight percent even though the chemical conditions at the two
systems are different and the modifications required to achieve the
improvement in solids quality are not identical. This improvement in solids
content corresponds to reducing the weight of water per weight of dry solids
by as much as half, and it results in much improved cake washing efficien-
cies and handling characteristics of the dewatered cake. The dewatering
rates are also improved substantially, resulting in smaller equipment sizes
required for thickening and filtration.
Figure 5, which is based on data obtained by Codan for CIPS a few years
ago, is an example of the type of improvement that can be made by adjusting
agitator speed. In this particular instance, the higher agitation speeds
cause a deterioration in filter cake solids concentration. By lowering the
agitator speed (shear rate) at the time the solids are precipitated, the
final crystal size is improved, thus improving cake solids concentration.
Similar effects were obtained with the use of high calcium lime and
magnesian lime, but the shape of the curves and the required operating pH
may be different. Reactor residence time also changes the shape of the
curve shown in Figure 5.
It is interesting to note that some of these reaction variables apply to
lime and limestone scrubbing systems also. This has been shown to apply
particularly to agitation, magnesium concentration, residence time, pH and
chemical composition of the lime or limestone used. The presence of
co-precipitates such as sulfate and magnesium has a marked effect on the
influence of agitation and detention time.
OPERATING RESULTS
Ultimately, there are really only three important criteria for a flue
gas desulfurization system. These are:
* Can it meet the legislated emissions criteria?
* Is it reliable and available? That is, can it meet the emissions
criteria consistently with acceptable levels of maintenance and
operator attention?
5-14
-------
• Is it cost competitive?
Double alkali clearly meets all of these criteria based on the three
systems discussed here when considered individually or as a group.
70.
65 _
00
Q
O
00
60__
UJ
55 _
50
Test Conditions
20 Min. Detention Time
CIPS Double Alkali System
200
I
400
\
600
800
1000
AGITATOR SPEED - RPM
Figure 5. EFFECT OF AGITATOR SPEED ON FILTER CAKE SOLIDS CONCENTRATION
EMISSIONS PERFORMANCE
The sulfur dioxide removal performance of all three systems is
summarized in Table 2.
5-15
-------
TABLE 2. UTILITY DOUBLE ALKALI SULFUR DIOXIDE REMOVAL SUMMARY
Station Name
Unit Sizes
Coal Sulfur Content, "/,
Inlet Sulfur Dioxide
Concentration
(ppm dry)
(Ib/MMBtus)
Outlet Sulfur Dioxide
Concentration
(ppm dry)
(Ib/MMBtu)
Efficiency (%)
Regulation
SIGECO
Brown 1
265
3.4-4.5
3,100-4,000
CIPS
Newton 1
575
2.8
4.5
NIPSCO
Schahfer 17
393
3-3.7
1,700-2,700
240-300
85-93
1.2 Ib/MMBtu
0.3-1.1*
132-215
0.49
92
1.2 Ib/MMBtu 90% and 0.62
Ib/MMBtu
*Newton 1 has achieved outlets of 0.3 to 0.6 in individual tests, but
normally operates at 1.0 to 1.1.
Even though the absorber designs and process conditions are slightly
different in each case, all three systems clearly meet their regulatory
requirements with room to spare.
AVAILABILITY
Availability results as reported in the current PEDCO data base for 1982
for A. B. Brown 1 and Newton 1 are summarized in Figure 6. However, the
record should reflect that SIGECO subsequently recognized that they counted
the system as available during two major outages totaling 30 days during
which duct and absorber lining repairs were made. They have recently asked
PEDCO to correct the official record to reflect this, but this change has
not yet been made in the PEDCO data base. The only significant forced
outage at Brown during 1982 was attributable to freezing the plant water
system during January.
For all of 1982, SIGECO's records indicate that Brown 1 was in
compliance 89 - 91% of the time based on hourly monitoring and material
balance methods respectively. More recently, in anticipation of future
increased load demand for Unit 1, they have begun to keep track of their
"full load capability." Full load capability represents SIGECO's assessment
of its ability to stay in compliance under boiler design maximum operating
conditions. SIGECO has retroactively estimated their full load capability
5-16
-------
for 1982 as 80%, and they are committed to a goal of raising it to 85%.
Given that Brown has no spare absorber and very little spare equipment in
other areas, this is an impressive objective.
%
ITY
Cfl
<
_l
i—i
<
>
<
IL
CO
0
1
CIPS NEWTON #1
_______ SIGECO BROWN #1
1 1
1 1
JJ
MONTH
Figure 6. 1982 PEDCO DOUBLE ALKALI SYSTEM AVAILABILITIES*
*SIGECO data include 30 days of scheduled outages
counted as available time
Availability results for all high-sulfur-coal FGD systems reporting
availabilities and operating costs to EPA from mid-1981 to mid-1982 are
summarized in Figure 7. The fuel sulfur contents are the reported design
basis. However, CIPS and SIGECO's actual sulfur contents are 2.8 and 3.8%
respectively. Costs include capital and various adjustments made by PEDCO
to put systems on a comparable basis. These results reflect the superior
availability record of the double alkali process in high sulfur coal
applications. It is too early to tell whether the NIPSCO system will
continue that record. However, the system is guaranteed to be 97% available
without the use of the spare absorber during a 90-day test period scheduled
later this year, and there is presently no reason to believe that it will
fail to pass that test.
5-17
-------
1001—
95
90
85
80
I
M
00
1 75
ss
I
70
65
50
98.3
97.8
92.9
60 —
55
96.0
Avg.
Notes
1. Data taken from EPA
Utility FGD Survey
2. * - pond disposal
(otherwise landfill)
3. Cost figures are
in 1981 dollars
Lime
Limestone
Double Alkali
Figure 7. FGD SYSTEM AVAILABILITY JULY 1981 TO JUNE 1982 (>2%S)
-------
OPERATING COSTS
Operating costs are difficult to compare because they are not always
recorded on a consistent basis and they reflect significantly different
design conditions, unit costs, and capital charges.
Capital costs vary more with respect to site specific design
requirements than they do according to process type. For similar design
criteria, all wet FGD systems have similar costs for reagent preparation,
ductwork, dampers, reheaters, and absorbers, which as a group typically
comprise 70-80% of FGD system cost. For consistent design criteria, none of
these components of the cost is more expensive for double alkali than for
any other wet FGD system.
Non-capital operating costs for 1982 for Brown 1 and Newton 1 are
reflected in Table 3. These numbers include disposal.. The total costs for
the two systems are very close, even though the respective circumstances are
slightly different. SIGECO has higher reagent costs, reflecting its higher
sulfur fuel, but lower operating labor and maintenance costs, primarily
because it does not include a stabilization system and the general
arrangement allows them to cover the area with less people. Power costs for
SIGECO are slightly less because of the double loop scrubber design at
CIPS. In all cases, these costs are lower than the values for the
comparable line items in the most recent EPRI sponsored process evaluation
which overall showed lime regenerated double alkali to be very competitive
with conventional limestone scrubbing.
TABLE 3. 1982 DOUBLE ALKALI OPERATING COST SUMMARY
Cost Item (mills/net kWh)
Reagent
Operating Labor
Maintenance
Power
Total
Sulfur Dioxide (Ib/MMBtu)
Inlet
Outlet
Collected
Load Factor (%)
A. B. BROWN 1
2.40
0.36
0.81
0.39
3.96
NEWTON 1
1.79
0.57
1.09
0.62
4.07
4.5
1.1
49
3.4
45
SUMMARY
Modifications to operating lime double alkali systems have improved
their operation. Costs have gone down and dependability up, so that the
process is the most satisfactory one available for use on high sulfur coal.
Other changes which are now being demonstrated will make future costs even
lower.
5-19
-------
MODIFICATIONS TO OPERATING SYSTEMS
Since the reaction between lime and the absorbed S02 occurs in a
precipitation tank that is not directly connected with the scrubber, the
double alkali process offers the flexibility of design to modify the
reaction variables without influencing the scrubbing operation. This
flexibility has allowed significant improvements in crystal size and sludge
dewatering characteristics at both the Newton and A. B. Brown stations. At
start-up, both of these units experienced problems with a variable-quality
filter cake and corresponding problems with cake washing efficiencies and
cake handling characteristics. Consequently, studies were undertaken to
improve the crystal size of the precipitated calcium salts. Both full-scale
units now produce a dry, and usually non-thixotropic, filter cake resulting
from relatively minor changes to the reaction tank system.
CURRENT COST AND DEPENDABILITY
Double alkali is clearly cost competitive (4 mills/kWh operating cost),
more dependable (greater than 95% available), and more efficient (capable of
attaining 90% removal) on high sulfur coal applications than any other
commercially demonstrated technology. The only issues clouding the future
of the process have been perceptions that solids quality is suspect and
reagent costs are too high. The former issue has been resolved by the
success of the modifications made by CIPS and SIGECO. The latter issue has
been disproved by the low actual costs experienced by users of the process.
FUTURE IMPROVEMENTS
The cost of the double alkali process will become even lower with the
imminent commercialization of limestone for regeneration in place of lime.
Use of limestone is made fairly simple by the flexibility of the double
alkali reaction system. Reaction tank design and chemical process variables
can be altered to complete the reaction of the limestone while still
providing good dewatering characteristics and without altering the scrubbing
operation. Consequently, the good scrubber system availability and
efficiency history of double alkali scrubbing systems can be maintained
while using the lower cost limestone for regeneration of the scrubbing
liquor.
In 1984 EPRI will sponsor a 3 megawatt demonstration of limestone double
alkali. NIPSCO has agreed to serve as a cosponsor for the program and will
be furnishing a site at Schahfer Station.
5-20
-------
PILOT EVALUATION OF LIMESTONE REGENERATED DUAL
ALKALI PROCESS
J. C. S. Chang, N. Kaplan
-------
INTRODUCTION
The sodium based dual alkali (DA) process is one of the second generation
flue gas desulf urization (FGD) technologies currently utilized in the United
States.!1) The DA process has the advantages of clear solution scrubbing,
high S02 removal efficiency, low liquid to gas (L/G) ratio, and high
rel iability. (2) All existing commercial-scale DA systems use lime to react
with spent scrubber effluent and regenerate the sodium sulfite scrubbing
solution. Since lime is produced from limestone by an energy intensive
calcination process, the direct use of limestone is increasingly more
attractive with the rising cost of energy. In recent years, unit weight lime
cost has been as much as 5 to 10 times the cost of limestone. It is estimated
that a significant cost savings could be realized by converting a DA FGD
system from lime to limestone.(3)
In order to promote the more advanced and more economical limestone DA
air pollution control technology, EPA's Industrial Environmental Research
Laboratory at Research Triangle Park, North Carolina (IERL-RTP, NC) has
sponsored a series of research projects. Factors influencing limestone
reactivity and system chemistry were identified by early laboratory work.
Limestone regenerations of simulated scrubbing solutions were studied by
bench-scale tests with promising results(4>5). The technical feasibility of
the limestone DA process was evaluated by conducting prototype testing with
the 20 MW facility at Gulf Power Company's Scholz Steam Plant.(6) Excellent
S0£ removal efficiencies in excess of 95 percent were achieved; limestone
utilizations were also high, over 97 percent. However, the soda ash
consumption of 0.29 moles of Na2C03/mole of $03 removed far exceeded
conventional consumption of 0.05 in a lime DA system.(?) Solids content in
the filter cake ranged between 35 and 45 percent, which was below the
anticipated 50 to 55 percent. The continuous operation was also limited by
the intermittent production of very fine, needle-shaped crystalline solids
with poor settleability. These fine solids did not settle in the thickener
and caused interruption of system operation or system upset, when they were
carried over from the thickener overflow to the absorber.
In order to support future full-scale operation and to promote system
feasibility, a pilot-scale dual alkali system was established by IERL-RTP, and
testing started in 1979. The pilot plant successfully simulated the ,
.prototype-scale testing at Scholz steam plant and later demonstrated the long
term operability of limestone dual alkali processes.^) Additional pilot
plant tests were conducted to study factors affecting filter cake quality and
soda ash consumption rate. This paper reports on the results of testing
during the period February 1982 through March 1983. This recent pilot plant
testing showed that significant improvement in soda ash consumption and filter
cake quality could be achieved with proper system control. The causes of
system upset by non-settleable solids were also identified and demonstrated in
tests. The objectives of this paper are to present the highlights of recent
pilot plant testing and to discuss, generally, limestone dual alkali
processes.
5-21
-------
TEST FACILITY
The absorber is a 23 cm (9 inch) diameter tray tower designed for 8 m-V
min flue gas capacity (0.1 MW). The overall process flow diagram is shown in
Figure 1. Flue gas from an oil/gas-fired boiler is drawn through the
absorption tower by an induced draft fan located immediately upstream from the
stack. Gaseous S02 from cylinders is fed into the ducting before the absorber
to spike the flue gas to predetermined concentrations of SO?- The oxygen
content of the flue gas is controlled by adjusting the boiler excess air. The
absorber contains a spray section (quench) followed by three trays in series
for gas/liquid contacting. The trays are sieve-type with 4.8 mm (3/16-inch)
holes on a 9.5 mm (3/8-inch) triangular spacing. Design pressure drop across
the trays is 20 cm (8 inches) of water. Regenerated scrubbing liquor from the
thickener hold tank is fed to the top tray and overflows a 1.3 cm (1/2-inch)
weir into a collection box and then to successive trays via internal
downcomers and weirs countercurrent to the flue gas flow. To prevent
short-circuiting of gas around the trays, liquor from the bottom tray is
collected in a weir box and overflows to the hold tank through a standpipe.
Spent scrubbing solution from the hold tank is pumped to the reactor
system based on liquid level control. Slurry flows through the four reactors
in series by gravity overflow. Limestone is fed to the first reactor as 40%
slurry. The feedrate is manually set as required for either pH or reactant
stoichiometry control. Soda ash is added to the second or the third reactor
as a dry powder (for ease of operation in a small pilot unit).
Reactor effluent slurry flows by gravity to the thickener centerwell.
Clarified liquor overflows from the thickener to the forward feed hold tank
from which it is pumped to the absorber.
Two dewatering systems, a rotary drum filter and a belt filter, were used
to evaluate the filterability of the thickener underflow slurry. The drum
filter is equipped with a single stage cake .washing device and the belt filter
with a three-stage countercurrent washing system.
PROCESS CHEMISTRY
The overall main reaction in a limestone regenerated DA process may be
represented by:
CaC03(s) + S02(g) H20 > CaS03(s) + C02(g) (1)
One mole of limestone reacts with 1 mole of S02 to produce 1 mole of calcium
sulfite and 1 mole of carbon dioxide, according to this equation.
5-22
-------
TO STACK
ABSORBER
S02
FLUE GAS f ^
u
££....!£?
£...fD»
GL.,^
nt*.
ni n TAMIT
'5 X — 3
•e
LIMESTONE S(
n 1
mm K* ACTOR
ROTARY DRUM
FILTER
BELT FILTER
THICKENER
Figure 1. Flow diagram of IERL-RTP dual alkali pilot plant.
5-23
-------
Reactions in the Absorber
The major chemical reactions that occur in the absorber include:
Absorption
so2
(g)
Oo - >. 0?
(aq) (3)
Desorption H2C03 - > C02 . +H20,ix (4)
(aq) (aq) (1)
Neutralization
S03". . + S02 + H20 - > 2HS03-
(aq) (aq) (1) (aq) (5)
HC03- , + S02/ x + H20, x - > H2C03/ s + HS03- x (6)
6 (aq) d(aq) (1) ^ ^aq) d (aq)
Oxidation
HS03- + 1/2 02 - > S04-- + H+ (7)
(aq) (aq) (aq) (aq)
S03" + 1/2 02 - » S04" (8)
(aq) £(aq) H (aq)
Absorption of S02 and desorption of C02 are governed by gas-liquid mass
transfer and equilibria between the S02 and C02 in the flue gas and in the
recirculated scrubbing liquor. If the alkaline species in the scrubbing
liquor (e.g., $03--), were reduced by low pH due to insufficient limestone
stoichiometry, a drop of S02 absorption rate will result. On the other
hand, higher L/G ratio and liquid holdup in the absorber can increase the S02
removal efficiency.
Neutralization of the absorbed S02 by aqueous alkaline species is
dominated by the essentially instantaneous reaction between S03-- and hydrated
S02. Oxidation of the sulfite species to sulfate ion is accompanied by the
absorption of oxygen in the flue gas. The oxidation reaction occurs
predominantly in the absorber. However, any contact between oxygen (from air
or flue gas) and aqueous sulfite or bisulfite will result in oxidation.
5-24
-------
Reactions in the Reactors
The primary chemical reactions that occur in the reactors include:
Dissolution and Desorption
CaC03, x + HS03- N > Ca++, N + S03" x + HC03- x (9)
(s) J (aq) (aq) J (aq) J (aq)
HC03- + HS03- > H2C03 + S03"
. (aq) (aq) (10)
Precipitation
Ca++, x + (l-x)S03-- x + xS04--, x + yH20, %
(aq) * (aq) * (aq) * (1)
Ca(S03) (S04) • yH20
1 "" A X \ ^ /
Limestone is dissolved in the reactors to replenish the aqueous alkaline
species which are consumed by the neutralization reaction in the absorber.
Most of the carbon dioxide resulting from limestone dissolution evolves
to the atmosphere from the reactor surface. The system product is a solid
solution, Ca(S03)1_ (S04)x • yH20, where x is proportional to the
oxidation level and ranges up to about 0.2. Limestone utilization is defined
as:
Percent utilization = 100 x TS/TCa = 100/stoichiometric ratio (13)
where TS: total sulfur in the product solids, moles
TCa: total calcium in the product solids, moles
Precipitation of solids occurs in accordance with the solubility product,
KSp, which is taken here to be equal tow):
Ksp = aCa++ . a$0 __ = 4.5 x 10'7 (mole/liter)"2 at 50°C (14)
where ar ++ and a^n __ are calcium and sulfite' ion activities, respectively.
Since the sulfite ion concentration in the scrubbing liquor is usually
maintained in the range of 0.1 to 0.5 M, the calcium ion concentration is
extremely low; i.e., sodium is the dominant cation in the scrubbing liquor.
Due to the low solubility, most of the carbon dioxide produced by the dissolu-
tion of limestone evolves almost immediately into the atmosphere. The
concentration of carbonate species (<0.05M) in the absorbent liquor is usually
orders of magnitude lower than that of sulfite species.
5-25
-------
The process chemical conditions can be reflected by two important
concentration ratios, sulfite to bisulfite and sulfate to sulfite. The
sulfite to bisulfite molar ratio and the pH are interdependent and can be
adjusted by limestone stoichiometry. The sulfate to sulfite molar ratio is
determined by the balance of oxidation and coprecipitation rates. The
absolute concentration of all species is reflected by the total sodium
concentration, which is controlled by Na2C03 addition.
RESULTS AND DISCUSSION
Pilot plant tests have been performed over the range of operating
conditions indicated in Table 1. The results demonstrate that solids with
reasonably good dewatering properties can be produced over a range of
simulated utility boiler operating conditions. Soda ash consumption rate
substantially lower than that in previous test programs can be achieved by
improved filter cake washing. System upset simulation tests were conducted to
identify the causes of system upset. It was demonstrated that system upset
could be avoided by proper design and control of the processes. The pilot
plant has been operated successfully for more than 10 months over the wide
range of operating conditions shown in Table 1 without any system upset. The
various aspects of system performance are discussed below.
SO? Removal
Over 92% S02 removal efficiency was readily achieved at 3000. ppm flue gas
S02 concentration. The two S02 monitoring techniques, ultraviolet and wet gas
analysis, were in good agreement. The S02 removal efficiency was a function
of system pH, sulfite ion concentration, and absorber internals. Material
balance indicated that the absorber "make-per-pass" (moles of S02 absorbed per
liter of absorbent flowing in the absorber) was in the range of 0.1 to 0.13
mole of S02 per liter of scrubbing solution, which was an order of magnitude
higher than that of slurry scrubbing processes. The high make-per-pass of the
pilot limestone dual alkali absorber was reflected by the low L/G ratios
(0.8-1.2 l/m^ or 6-9 gal/1000 acf utilized), as shown in Table 2, compared with
the high L/G (5.3-9.3 1/m3 or 40-70 gal/1000 acf, usually) of slurry scrubbing
processes. (9)
Limestone Utilization
High calcium, Fredonia limestone was utilized to regenerate the scrubbing
liquor in all testing described herein. The chemical composition and particle
size distribution are shown in Tables 2 and 3. In previous testing,
simulating conditions at the Gulf Power Scholz plant, Sylacauga limestone was
also tested in certain runs and proved to be acceptable.
As shown by Equation 9, bisulfite ion is the predominant acidic species
which reacts with the limestone and dissolves it. Figure 2 illustrates the
effect of acidity concentration on limestone utilization obtained from the
pilot plant tests with the pH of regenerated liquor between 6.4 and 6.5.
5-26
-------
Table 1. Operating Ranges of Limestone Dual Alkali Pilot Plant Tests
L/G ratio, 1/m3:
Limestone stoichiometry:
Scrubber feed pH:
Scrubber effluent pH:
Active sodium concentration3, M:
TOS, M
Flue gas 02> %
Sulfate to sulfite molar ratio:
Oxidation, %:
0.80-1.2 (6-9 gal/1000 acf)
1.0-1.2
6.0-6.8
5.4-6.3
0.48-1.5
0.34-1.1
5-10
1.4-3.9
6-20
aActive sodium is defined as 2[S03--] + 2[C03"] + [HCOs"] +
expressed as Na+ concentration. Active sodium is actually a misnomer
since Na+ does not participate in any of the dual alkali reactions but
is used for convenience.
5-27
-------
I
KJ
00
Table 2. Chemical Composition of the Limestone Used for
Dual Alkali Pilot Plant Testing
Chemical Composition,
wt. percent
Limestone
Fredonia
Location of
Quarry
Fredonia,
Kentucky
Calcium
(Ca)
35.8
Magnesium
(Mg)
2.5
Carbonate
(C03)
60.5
Acid
Insolubles
2.2
Table 3. Results of Wet Sieve Analysis of Limestone Used for
Dual Alkali Pilot Plant Testing
Weight Percent
Limestone -325 mesh -230 mesh -200 mesh -170 mesh -100 mesh
Fredonia 79.6 89.1 91.8 98.4 99.2
-------
100
NUMBER OF REACTORS = 3
REACTOR RESIDENCE TIME = 60 min (TOTAL)
REGENERATED LIQUOR pH = 6.4 - 6.5
95
90
<
N
C/J or
LU 85
80
75
0.1
0.2
ACIDITY, M
0.3
0.4
Figure 2. Effect of (thymolphthalein) acidity on limestone utilization.
5-29
-------
(Acidity in a DA system is defined as the concentration of all acid species
determined by titrating a quantity of solution with standard base to the
thymolphthalein end point of pH 10.6.) A minimum concentration of acidity at
G.2 M was needed to attain 90% limestone utilization. When the acidity
concentration was above 0.4 M, 95-100% limestone utilization could be
achieved. Since the acidity concentration of a concentrated mode DA process
is usually higher than 0.2 M, it was concluded that high limestone utilization
(greater than 90%) could be obtained in the DA process under normal operating
conditions.
Oxidation and Sulfate Balance
Two effects of oxidation were observed during the pilot plant tests,
sulfate ion accumulation and solids quality deterioration. The latter will be
discussed in the next section. The level of sulfate ion accumulation is
determined by the oxidation rate and sulfate removal rate. Sulfate ion can
leave the system either in liquids, wetting the waste product as sodium
sulfate, or in solids as calcium sulfite/sulfate crystals. The latter is more
desirable because sodium is not lost, and the waste cake will contain less
soluble salts which can be leached into the ground water near a disposal site.
Calcium sulfate is coprecipitated with calcium sulfite in the reactors to form
a "solid solution" as indicated by Equation 12. (The liquid phase calcium and
sulfate concentrations never exceed the gypsum solubility product, and no pure
gypsum phase forms.) The relationship between sulfate/sulfite molar ratio in
the liquid and in the precipitated solids is shown in Figure 3. Overall, the
pilot plant sulfate balance showed that the DA system was capable of keeping
up with oxidation rates up to 20% of the S02 removed--oxidation rates higher
than those anticipated for most medium- and high-sulfur coal applications. (An
oxidation rate of 20% is equivalent to a S04--/S03"" molar ratio of 0.25 in
the solids.) In other words, sulfate balance can be accomplished by calcium
sulf ite/sulfate coprecipitation with closed-loop operation for the medium- to
high-sulfur coal applications of the limestone DA processes.
Oxidation is an important but undesirable side reaction in DA processes
since high oxidation could convert the useful total oxidizable sulfur (TOS,
moles of sulfite ion plus bisulfite ion per liter of solution) species to
inert sulfate ion and deteriorate solids qualities. Efforts have been made to
characterize the oxidation reaction and to minimize its rate.
Short term (8 to 12 hours each run) pilot plant tests were performed by
recirculating Na2S03/NaHS03 solution around the absorber which was isolated
from the rest of the system. Oxidation tests were conducted at 3.5, 5.5, and
8.5% flue gas oxygen concentrations. The initial TOS concentration was set at
0.3 to 0.4 M and the solution pH at about 6.7. The oxidation rate increased
from 2.35 to 3.08 grnole/hr as the oxygen concentration was raised from 3.5 to
5.5%. Oxidation rate at 3.77 gmole/hr was obtained with 8.5% flue gas oxygen.
It was concluded that the oxidation rate in the DA absorber is a function of
flue gas oxygen concentration.
5-30
-------
I I I J
LIMESTONE DUAL ALKALI TESTS
• THIS WORK •
• SCHOLZ (6) 0
A ADL (4) . A
0.20
EC
ec
0.15
20.10
_l
a
CO
o
CO
0.05
>**
I 1.0 2.0 ' 3.0 4.0 5.0
REGENERATED LIQUOR SULFATE TO SULFITE MOLAR RATIO
Figure 3. Ratio of sulfate to sulfite in solids as a function of sulfate to sulfite
ratio in regenerated liquor.
5-31
-------
The effectiveness of sodium thiosulfate as an oxidation inhibitor was
also evaluated by pilot plant testing. It was suggested that the thiosulfate
ion might act as a radical scavenger to deactivate the radicals that promoted
the oxidation of sulfite and/or bisulfite ions. Results shown in Figure 4
indicate that the concentration ratio of thiosulfate ion to TOS was an
important factor determining the effectiveness of sodium thiosulfate as an
oxidation inhibitor. Insignificant effect was observed when the thiosulfate
to TOS ratio was below 0.1. If 50% reduction of the oxidation rate is
desirable, the thiosulfate to TOS ratio must be maintained at 0.5 or above,
under conditions tested.
In general, the action of thiosulfate as an oxidation inhibitor may be dependent
upon the ratio of thiosulfate to TOS as shown in Figure 4. Since the TOS
concentrations in DA systems are orders of magnitude higher than in limestone
slurry scrubbing systems, a much higher concentration of thiosulfate may be
required in DA systems. In limestone slurry scrubbing only a few hundred ppm of
thiosulfate was reported to be effective for oxidation inhibition. However,
for a concentrated sodium based limestone DA process, the TOS concentration is
expected to be in the range of 0.7 to 1.0 M. Therefore, a significant amount of
sodium thiosulfate would be required to cut the oxidation rate by 50%. The
feasibility of using sodium thiosulfate as an oxidation inhibitor will be
determined by economic trade-offs and other site specific factors.
Filter Cake Quality
Probably the most significant impact of oxidation on system performance
observed during the pilot plant tests was the change in filter cake quality.
As shown in Figure 5, the % insoluble solids in the filter cake dropped with
the increase of oxidation rate reflected by the sulfate to sulfite molar ratio
in the regenerated liquor. When the sulfate to sulfite molar ratio was below
1.5, filter cake % insoluble solids reached 60%. The % insoluble solids in
the filter cake varied inversely with the sulfate to sulfite molar ratio,
reaching about 35% at a sulfate to sulfite molar ratio of 4. Operation under
these conditions was considered to be undesirable because of the poor physical
properties of the waste and the high losses of sodium compounds.
Coulter counter analyses were conducted to measure the particle size
distribution in the reactor effluent at various sulfate to sulfite molar ratios.
However, no significant changes in particle size distribution were obtained, and
the mean particle diameter remained at about 20 urn even though the filter cake %
insoluble solids content dropped from 60 to 35 percent. The product solids were
further examined under a 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 6 and 7, the two product solids were
physically different. The photomicrographs shown in Figure 6 were taken at a
magnification of lOOOx. The solid "particles" shown in Figure 6 were actually
clusters composed of agglomerates of crystals. The "particles" which gave 60%
solids filter cake were composed of thicker, larger, and better defined crystals
or platelets. However, the 35% filter cake "particles" were agglomerates of
thinner, smaller, more needle-like crystals. The differences in crystal
thickness were illustrated clearly by the 5000x magnification photomicrographs
5-32
-------
I.OQ;
\
V FLUE GAS OXYGEN =3-4%
,_ °\ INITIAL TOS = 0.3-0.4 mM
< \o INITIAL pH =6.7-6.8
\ INITIAL SULFATE =OM
0.81 X OXIDATION RATE WITHOUT THIOSULFATE =2.35 gmol/hr
\
\
\
< 0.6h- N
2 \
\
0.4
a
X
o
0.2
0.2 0.4 0.6 0.8 1.0
THIOSULFATE/INITIAL TOS
Figure 4. Effectiveness of sodium thiosulfate as an oxidation inhibitor.
5-33
-------
ss
50
u
=; 40
35
30
1.0
1.5 2.0 2.5 3.0 3.5
REGENERATED LIQUOR SULFATE TO SULFITE MOLAR RATIO
4.0
Figure 5. Percent filter cake solids as a function of sulfate to sulfite ratio in the regenerated
scrubbing liquor.
5-34
-------
60 PERCENT SOLIDS FILTER CAKE
35 PERCENT SOLIDS FILTER CAKE
Figure 6. Limestone dual alkali solid product (1000x).
5-35
-------
60 PERCENT SOLIDS FILTER CAKE
35 PERCENT SOLIDS FILTER CAKE
Figure 7. Limestone dual alkali solid product (5000x).
5-36
-------
in Figure 7. An interpretation of Figure 7 is that crystal fractures (which
might be caused by the collision of agitator blades) and thin platelets or
needles in reactors were observed with 35% filter cake "particles" at high
,sulfate to sulfite molar ratio; and that these fine particles were responsible
for the poor physical properties of the filter cake. It may also be postulated
that the coprecipitated sulfate ion acts as an impurity which inhibits crystal
growth and causes crystal defects.
High Active Sodium vs. Low Active Sodium
In order to evaluate the effects of operating conditions on system
performances such as oxidation rate and filter cake % insoluble solids, long
term demonstration tests were performed at high (1.2 M) and low (0.48 M)
active sodi'um concentrations. The operating conditions for each run are shown
in Figures 8 and 9, respectively. Results summarized in Table 4 indicate that
at high active sodium the sulfate to sulfite molar ratio was 1.9 which was
considerably lower than the 3.9 obtained from low active sodium test. Filter
cake analyses also indicate that the 8% oxidation level at high active sodium
was. considerably lower than the 18% observed at low active sodium. The lower
sulfate to sulfite molar ratio and oxidation rate at high active sodium are
reflected by the higher filter cake % insoluble solids as shown in Table 4.
The decrease of oxidation rate with the increase of active sodium concentra-
tion is probably due to the ionic strength effect. As shown in Table 4, the
high active sodium mode was operated with 4.5 M ionic strength which is 100%
higher than that of low active sodium. Data in the literature indicate that
both the solubility and diffusivity of oxygen in aqueous solutions can be
reduced by higher ionic strength through the change of liquid viscosity. As
a result, the mass transfer rate of oxygen from flue gas to scrubbing solution
is reduced, and the liquid-phase oxidation rate is also decreased.
Soda Ash Consumption
With a properly designed and operated system, the majority of the sodium
loss should occur with the liquor entrained in the filter cake. Sodium ion
concentration in the scrubbing liquor, filter cake % insoluble solids, and
filter cake wash efficiency are the three major factors determining the rate
of sodium consumption. The-typical sodium ion concentration of a limestone DA
process is expected to be in the range of 2 to 3 M. As shown in Figure 10,
the soda ash consumption will be significant if there is no filter cake
washing to recover the sodium species, especially when the filter cake %
insoluble solids is below 50%. However, the soda ash consumption can be
reduced substantially by filter cake washing with a reasonable wash
efficiency. The filter cake wash efficiency was defined as:
. . ,:,:. . /-, sodium content after washing \ v 1nfw /1(-\
washing efficiency = (l - sodium content before washing) x 100% (15)
A rotary drum filter with single stage washing was first used to evaluate
the washability of sodium compounds from the filter cake. Only 10 to 20% wash
efficiency was obtained. The poor wash efficiency was attributed to filter
cloth blinding or poor filter design. Most of the wash water did not
penetrate the cake and just flowed down off the filter cake surface.
5-37
-------
TO STACK
94XS02
REMOVAL
SCRUBBER FEED LIQUOR
TOTAL SULFUR 1.65 M
TOTAL OXIDIZABLE SULFUR 0.70 M
TOTAL ALKALINITY 0.50 M
TOTAL SODIUM 3.00 M
pH 6.60
Figure 8. Operating conditions of high active sodium (1.2 M) test.
5-38
-------
TO STACK
SCRUBBER FEED LIQUOR
TOTAL SULFUR
TOTAL OXIDIZABLE SULFUR
TOTAL ALKALINITY
TOTAL SODIUM
pH
LIMESTONE
SLURRY*
4.4 kg/hr
FILTER CAKE,
35%INSOLUBLE SOLIDS
0.89 M
0.34 M
0.14 M
1.SOM
6.40
Figure 9. Operating conditions of low active sodium (0.48 M) test.
5-39
-------
In an effort to reduce sodium losses by improving wash efficiency, a belt
filter with three stages of countercurrent washing was installed and then
evaluated. Three sets of filter cake washing tests were conducted. The
results, compared with wash efficiencies predicted by a wash model, are
shown in Figure 11. The wash efficiency, assuming countercurrent washing with
complete mixing of wash water and occluded liquor at each stage, can be
calculated by this model as follows:
wash efficiency = (l - J^" l ) x 100% where R>1 (16)
where R is the number of filter cake water displacements, and n is the number
of washing stages.
Table 4. The Effects of Active Sodium and Ionic Strength on
Oxidation and Resulting Filter Cake Quality at the
IERL-RTP Dual Alkali Pilot Plant
Active sodium, M
Total sodium, M
Ionic strength, M
Sulfate/sulfite molar ratio(a)
Oxidation rate(b), %
Filter cake insoluble
High Active
Sodium Mode
1.2
3.0
4.5
1.9
8
52
Low Active
.Sodium Mode
0.48
1.5
2.25
3.9
18
35
solids, %
Flue Gas 02, % 7.1 7.0
(a)in regenerated liquor
(b)based on filter cake
analyses
5-40
-------
0.35
0.30
0.2S
CM
o
CO
i
o
o
'0.20
3 0.15
V)
Q
O
CO
0.10
0.05
35
40 45 50 55
FILTER CAKE INSOLUBLE SOLIDS, percent
60
Figure 10. Soda ash loss rate through unwashed filter cake as a function of filter cake insoluble
solids and scrubbing liquor sodium concentrations.
5-41
-------
100
WASH WATER RATE
O • DISPLACEMENT x 2
O DISPLACEMENT x 3
90
80
1
70
60
50
a*
THEORETICAL WASHING
(COMPLETE MIXING EACH STAGE)
i
1 2
NUMBER OF WASHING STAGES
Figure 11. Performance of belt filter with countercurrent washing.
5-42
-------
Figure 11 shows that wash efficiencies lower than those predicted by the
model were obtained from the pilot plant. With two filter cake water
displacements and three washing stages, approximately 75% wash efficiency was
observed, which is 18.3 percentage points lower than the predicted 93.3% wash
efficiency. There are two reasons for the difference. First, the wash model
assumes that all sodium species are washable. However, data indicated that
about 10 to 15% of the sodium species are not washable. Coprecipitation and
crystal occlusion might explain why some of the sodium species contained in
the filter cake cannot be washed out. Secondly, the model assumes an even
distribution and complete mixing of wash water with the entrained cake liquor.
However, cake cracking causes some bypassing or channeling of wash water and
results in less than ideal mixing conditions.
The model indicates that the filter cake washing efficiency could be
maximized by further increasing the quantity of wash water and/or the number of
washing stages. Two displacements of wash water were used in these tests since
that was the volume of wash water permitted by material balance when limestone
was added to the DA reactors as 40% slurry (limestone in fresh water).
Material balance indicates that if limestone is fed as a dry powder, the wash
rate can be increased to three displacements, corresponding to an ideal wash
efficiency of 97.5%.
Pilot plant data indicated that the average filter cake insoluble solids
was about 52% at 3M total sodium concentration (Table 4). Figure 10 indicated
that the soda ash requirement to replenish the sodium loss with unwashed
filter cake should be about 0.2 mole N32C03 per mole of S0£ absorbed. If 75%
filter cake wash efficiency is achieved, the soda ash consumption rate is
predicted to be as low as 0.05 mole ^2603 per mole of SO? absorbed.
Simulation of System Upsets
System upsets are characterized by non-settling finely divided solids
suspended throughout the liquid in the system. The thickener cannot produce
clear regenerated liquor overflow to feed the absorber, and the system ceases
to function as a dual alkali process.
Pilot plant testing indicated that system upsets may be caused either
mechanically or chemically. Mechanical attrition of crystals produced poor
quality, non-settleable solids.(3»10) System upset was also observed to be
purely a function of chemistry; specifically, it would occur with sulfate to
sulfite molar ratios above about 4.
A mechanically caused upset was demonstrated by using the reactor design
shown in Figure 12a, which included long baffles and the agitators pumping
downward.* This design produced a quiescent zone between the baffles and the
5-43
-------
It has been reported by FMC Corporation that they experienced, if anything,
superior performance with down draft agitation. Due to the difference in
results cited by FMC Corporation and those of our pilot plant tests, and personal
correspondence with Dr. R. R. Lunt, an investigator of this technology, it is
clear that reactor design plays a major role in the operation of the system.
It is also clear that this effect is not completely understood. We have
reported the results of operation in this 0.1 MW pilot unit, and have merely
tried to explain the observed results. The design of our current reactor
system, however, is probably not optimal.
• ABSORBER BLEED
LIMESTONE SLURRY
J
OVERFLOW
• ABSORBER BLEED
-LIMESTONE SLURRY
OVERFLOW
Figure 12. Reactor flow patterns utilized for system upset simulation tests.
5-44
-------
wal 1 of the tank. Large crystals tended to settle in the quiescent zone and
accumulate in the reactors. As crystals were accumulating, the reactor
solids concentration increased and the filter cake quality deteriorated. It
was suspected that the increase of solids concentration favored crystal
attrition, and the attrition produced extremely fine particles. Eventually,
the overload of fine solids overflowed to the thickener and produced a system
upset. Solids quality was improved by using the continuous stirred tank
reactors with agitators pumping upward as shown in Figure 12b. When the
reactor solids concentration was 3% or below, the solids settling rate reached
1.5 to 2 cm/min vs. the 0.1 cm/min during upset conditions.
A chemical upset was produced by decreasing the soda ash make-up rate to
zero to let the sulfate to sulfite molar ratio increase. The solids
qualities, including solids settling rate and filter cake % insoluble solids,
deteriorated with the increase of sulfate to sulfite molar ratio. System
upset occurred when sulfate to sulfite molar ratio reached 4.3. Heavy doses
of soda ash were then added to the reactors in an attempt to reduce the
sulfate to sulfite molar ratio and to reverse the system upset. The sulfate
to sulfite molar ratio decreased gradually with associated improvement in
solids qualities. When the sulfate to sulfite molar ratio decreased below
1.5, the solids settling rate reached 2.5 cm/min, and filter cake insoluble
solids content reached 60%. This indicated that the chemical upset was
reversible by adjusting the system chemistry.
CONCLUSIONS
The following conclusions were drawn from the pilot plant studies:
1. The results confirmed that limestone dual alkali processes can
provide the advantages of clear solution scrubbing with high S02
removal efficiency at low L/G'and high limestone utilization.
2. The level of oxidation can be reduced by increasing the ionic
strength in scrubbing solutions. The removal of the oxidation
product, sulfate ion, can be accomplished by calcium sulfite/sulfate
coprecipitation as solid solution with closed-loop operation and
filter cake washing for medium- and high-sulfur coal applications.
3. The filter cake quality was affected by levels of oxidation or liquid
phase sulfate to sulfite molar ratios. Pilot plant tests
demonstrated that filter cake % insoluble solids can be improved from
35% to 52% by increasing the active sodium concentration from 0.48 M
to 1.2 M to reduce the oxidation level and sulfate to sulfite molar
ratio.
4. The soda ash consumption rate was influenced by scrubbing solution
sodium concentration, filter cake quality, and cake wash efficiency.
Pilot results achieved cake wash efficiency of 75% using three stages
of countercurrent washing. Material balance projected that a soda ash
consumption of 0.05 mole Na2C03 per mole S0£ removed should be
attainable with closed-loop operation and 75% cake wash efficiency.»
5-45
-------
5. System upsets assumed to be caused by mechanical, crystal attrition
may be avoided by better design and operation of reactors, agitators
and pumps. Chemically induced system upset caused by high sulfate to
sulfite molar ratio is reversible and can be avoided by controlling
the system chemistry in medium- and high- sulfur coal applications
with normal levels of oxygen in the flue gas.
Future EPA Test Program
The EPA/IERL-RTP test program with limestone dual alkali FGD systems is
continuing. The following activities are either proceeding or planned for
1984.
• Tests evaluating the magnesium ion effects on system performance
• Tests evaluating the chloride ion effects on system performance
• Limestone type and grind tests
• Optimization .of reactor design and operation
• Factorial tests to develop the interrelationship of system
performance and operating parameters and to optimize the process
design.
5-46
-------
REFERENCES
1. Smith, M.P., M.T. Melia, B.A. Laseke, Jr., and N. Kaplan, "Recent Trends in
Utility Flue Gas Desulfurization," Proceedings: Symposium on Flue Gas Desulfur-
ization -- Houston, October 1980, Volume 1, EPA-600/9-81-019a (NTIS PB 81-
24315677 April 1981.
2. Kaplan, N., "Summary of Utility Dual Alkali Systems," Proceedings: Symposium on
Flue Gas Desulfurization -- Las Vegas, Nevada, March 1979, Volume II,
EPA-600/7-79-167b (NTIS PB 80-133176), July 1979.
3. Chang, J.C.S., J.H. Dempsey, and N. Kaplan, "Pilot Testing of Limestone
Regeneration in Dual Alkali Processes," presented at the 7th EPA/EPRI
Flue Gas Desulfurization Symposium, Hollywood, FL, May 1982.
4. LaMantia, C.R., R.R. Lunt, J.E. Oberholtzer, E.L. Field, and J.R.
Valentine, "Final Report: Dual Alkali Test and Evaluation Program,
Volume II," EPA-600-7-77-050b (NTIS PB 272770), May 1977.
5. Oberholtzer, J.E., L.N. Davidson, R.R. Lunt, and S.P. Spellenberg,
"Laboratory Study of Limestone Regeneration in Dual Alkali Systems,"
EPA-600/7-77-074 (NTIS PB 272111), July 1977.
6. Valencia, J.A., J.F. Peirson, Jr., and G.J. Ramans, "Evaluation of the
Limestone Dual Alkali Prototype System at Plant Scholz - Final Report,"
EPA-600/7-81-141b, August 1981.
7. Durkin, T.H., J.A. Van Meter, and L.K. Legatski, "Operating Experience with
the FMC Double Alkali Process" in Proceedings: Symposium on Flue Gas
Desulfurization --Houston, October 1980, Volume 1, EPA-600/9-81-019a (NTIS
PB 81-243156), April 1981.
8. Lowell, P.S., D.M. Ottrner, K. Schwitzgebel, T.I. Strange, and D.W. DeBerry,
"A Theoretical Description of the Limestone Injection - Wet Scrubbing
Process, Volume I," U.S. Environmental Protection Agency, APTD 1287 (NTIS PB
193029), June 1970.
9. Henzel, D.S., B.A. Laseke, E.O. Smith, and D.O. Swenson, "Limestone FGD
Scrubbers: User's Handbook," EPA-600/8-81-017 (NTIS PB 82-106212), August
1981.
10. Boward, W.L. Jr., R.O. Petkus, and K.H. Wang, "FMC Limestone Double Alkali
FGD Process," Presented at the Meeting of AIChE in Cleveland, OH, August
1982.
5-47
-------
SESSION 6: FLUE GAS TREATMENT (COMBINED SO /NO )
A A
Chairman: J. David Mobley
Industrial Environmental Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC
-------
STATUS OF THE DOE FLUE GAS CLEANUP PROGRAM
J. E. Williams
-------
STATUS OF THE DOE FLUE GAS CLEANUP PROGRAM
by: John E. Williams
Project Manager, Flue Gas Cleanup
Pittsburgh Energy Technology Center
U. S. Department of Energy
Pittsburgh, PA 15236
ABSTRACT
The U. S. Department of Energy directs a substantial R&D effort to
develop advanced environmental control technology for coal-fired stationary
sources. The Flue Gas Cleanup Program is aimed primarily at post-combustion
cleanup of S02 and NO ; a small part of the program is directed at charac-
terization and improvea control of respirable particulates. The programmatic
goal established at Fossil Energy Headquarters is the development of technol-
ogy options for 90% removal of both S02 and NO while controlling particulates
to the New Source Performance Standards (NSPS). Emphasis is placed on devel-
oping process concepts that offer potential cost reductions of 20% to 25%,
compared to commercially available technology (Selective Catalytic Reduction
of NO followed by a wet limestone scrubber and the necessary baghouse or
electrostatic precipitator (ESP)). The commercialization or application
period is expected to be in the late 1980's or early 1990's.
The Flue Gas Cleanup Program is implemented by the Pittsburgh Energy
Technology Center (PETC). PETC is also responsible for implementation of
other related research programs, including coal preparation and direct coal
combustion, and has been involved in flue gas cleanup since the late 1950's.
The current DOE program is being conducted by a cross-section of indus-
trial organizations, not-for-profit research laboratories of universities,
national laboratories, and in-house research. Many of the projects are new,
and the research is just beginning. A chronology and overview of the program,
together with brief descriptions of the status of individual projects, are
given.
DISCLAIMER
Reference in this report to any specific commercial product, process, or
service is to facilitate understanding and does not necessarily imply its
endorsement or favoring by the United States Department of Energy.
6-1
-------
BACKGROUND
PROGRAM OVERVIEW
The Flue Gas Cleanup Program was initiated in DOE during 1979 as part of
the Advanced Environmental Control Technology (AECT) program, and originally
included the lime/limestone scrubber program, the advanced flue gas desulfur-
ization (FGD) program, and the NO /particulate program.
X
The program supports the development of post-combustion flue gas cleanup
technology. The goal of this program is to help foster the more widespread
use of coal by providing the technology necessary for its utilization in an
environmentally and economically acceptable manner.
The DOE Flue Gas Cleanup Program, like many other Federal research and
development (R&D) programs, was reoriented in the 1981-1982 period to empha-
size longer-range objectives. Accordingly, those FGD projects aimed at
developing improved near-term FGD technologies were curtailed, and attention
was shifted toward the needs of the future. Programmatic changes included
the following:
• Phaseout of the lime/limestone scrubber projects.
• Phaseout of the advanced FGD projects (dry scrubbing, regenerable
processes, etc.).
• Cancellation of the cooperative effort with EPA on the LIMB program.
• Redirection of the NO /particulate flue gas projects to longer-range
combined NO /SO /particulate projects.
X X
PROGRAM RATIONALE
In order to develop the new program rationale and to develop new long-
range program objectives, discussions were held within DOE and with EPA,
EPRI, TVA, and others.
Coal usage was projected to about double by the year 2000 (Table 1).*
With increased coal usage, emissions of S02, NO , and particulates could be
expected to increase correspondingly unless controlled. Emissions under four
possible control scenarios are listed in Table 2. From Table 2, it can be
seen that under current NSPS, while S02 emissions increase only moderately by
the year 2000, NO emissions about double in the same period. Because of
these analyses, tne increased international concern with acid precipitation,
and the historical pattern followed in Japan, it was concluded that NO
control will probably receive increased attention during this time period.
Interim R&D targets selected were the following:
• NO - 70% to 90% control level.
• SOl - NSPS (90% for high sulfur coal).
• Particulates - NSPS plus further restrictions on respirable
particles (<2 microns).
"DOE's Electron-Beam Irradiation Developmental Program", by Edward C.
Trexler, paper presented at the Seventh Symposium on Flue Gas Desulfuri-
zation (May 17-20, 1982).
5-2
-------
TABLE 1. FLUE GAS CLEANUP-PROJECTED- COAL CONSUMPTION
(QUADS/YR)
•
•
0
6
•
Sector
Utility
Industrial
Residential/Commercial
Transportation
Synthetic Fuels
Total
1980
11
3
0
15
.4
.9
.3
-
.6
1990
17
3
0
21
.3
.7
.2
-
.2
2000
23
4
0
0
28
.0
.6
.2
-
.4
.2
National Energy Policy Plan, 1983
TABLE 2. FLUE GAS CLEANUP POLLUTANT EMISSIONS VS.
CONTROL LEVEL (COAL-FIRED UTILITY)
Generation/Emission
r, -,-, o ^ T T T (1°6 tons Per year)
Pollutant Control Level * ? . '
1980 1990 2000
NO 0
x 1
2
3
S02 0
1
2
3
TSP 0
1
2
3
(uncontrolled)
(NSPS)
(NSPS/ low NO )*
(retrofit)* X
4
4
23
16
46
1
.9
.6
-
-
.1
.0
-
-
.0
.0
-
™
7
6
4
0
35
17
17
5
70
1
1
0.
.5
.4
.9
.9
.2
.8
.8
.2
.2
.1
.1
26
9
8
5
1
46
19
19
6
93
1
1
0.
.9
.1
.2
.2
.8
.5
.5
.9
.2
.2
.2
35
Control Levels:
Level 2 - NSPS with NO lowered to 0.1 ///MMBtu
Level 3 - Retrofit allxplants to Level 2 by 1990
6-3
-------
To further help establish the DOE Flue Gas Cleanup Program, several
studies were made to assess the current status of combined NO /SO /particu-
late technology. These studies identified technological approaches (sum-
marized in Figure 1); identified specific processes and their developmental
status (gas-phase processes are summarized in Figure 2, while liquid-phase
processes are summarized in Figure 3); and assessed potential problems as-
sociated with each type of process along with possible breakthrough keys
(summarized in Table 3). Key technical barriers with possible solutions
based on this study are summarized in Table 4.
Based on these studies, it was concluded that combined NO /SO /particulate
control technology was not a well developed area; existing processes appeared
to be very complicated and likely to be quite expensive. It was therefore
decided that the DOE Flue Gas Cleanup Program objectives should be focused on
development of simple and economical systems to meet emission limits expected
for the late 1980's through the 1990's. The approach was to emphasize combined
NO and S02 (and preferably also particulates) in a single reactor. Simple,
economical processes that minimize the use of premium fuels or their deriva-
tives (i.e., NHs) and that produce either a useful or benign waste product
should be sought. Selective catalytic reduction coupled with limestone
scrubbing (SCR/FGD) was selected as a yardstick by which to assess the poten-
tial of processes selected for evaluation under the DOE Flue Gas Cleanup
Program. Processes selected should have a potential savings of at least 20%
to 25% compared to SCR/FGD.
CURRENT PROGRAM DESCRIPTION
Using the guidelines established, the DOE Flue Gas Cleanup Program was
developed. The major focus is the evaluation of combined NO /SO /particulate
processes showing potential for meeting the criteria established. For discus-
sion purposes, the projects have been broken down into the following categories:
Energy Technology Center (ETC) and National Laboratory projects, electron-beam
(E-Beam) projects, support studies, recently completed projects, and projects
initiated in late-1983. Iii the following section, specific projects under
each of these categories will be discussed in more detail, including their
current status.
PROJECT DESCRIPTIONS AND CURRENT STATUS
ENERGY TECHNOLOGY CENTER AND NATIONAL LABORATORY PROJECTS
Current Energy Technology Center projects under the Flue Gas Cleanup
Program include projects with Pittsburgh Energy Technology Center (PETC),
Grand Forks Project Office (GFPO, formerly Grand Forks Energy Technology
Center), and Morgantown Energy Technology Center (METC). A brief summary of
the ETC and National Laboratory activities is shown in Table 5.
Pittsburgh Energy Technology Center Projects
PETC has been involved in flue gas cleanup research for more than twenty
years. Recent (1981-1983) activity under the DOE Flue Gas Cleanup Program
6-4
-------
SCR/Sorption
I
Ui
Thermal
D-NOX
+ FGD
SCR + FGD
Gas Phase
NOX Removal
E-Beam
Irradiation
Absorption/
Reduction
Electrolytic
Oxidation/SCR
Liquid Phase
NOX Removal
Oxidation/
Absorption/
Oxidation
Oxidation/
Absorption/
Reduction
Figure 1. Flue Gas Cleanup NOX/SOX Removal -Technological Approaches-
-------
SCR/Sorption
CARBON CATALYST (TAKEDA,
SUMITOMO, UNITIKA, B-F,
HITACHI) (2-3)
METAL OXIDE (EXXON)(2)
CuO (SHELL-UOP)(1)
Thermal D-NOX
EXXON (2)
DEVELOPMENTAL STATUS:
COMMERCIAL —(1)
PILOT — (2)
PDU — (3)
BENCH — (4)
SCR + FGD
• HITACHI-ZOSEN (2)
E-Beam Irradiation
NH3 INJECTION (EBARA)(3)
LIME INJECTION (R-C)(4)
Figure 2. Flue Gas Cleanup NOX/SOX Removal -Gas Phase Technology-
-------
Absorption/ Reduction
FE-EDTA (CHISSO, ASAHI,
KUREHA)(3-4)
FE-CH ELATE (3-4)
FE-NTA (5)
FES (SULF-X) (3)
BUTYL-N-PHOS./STRIP (PETC) (5)
Oxidation/Absorption/Oxidation
• Mg(OH)2 (KAWASAKI) (2-3)
DEVELOPMENTAL STATUS:
COMMERCIAL —(1)
PILOT — (2)
PDU — (3)
BENCH — (4)
CONCEPT — (5)
Electrolytic Oxidation/SCR
• FUSED SULFATE(GE/IONICS)
ON L,AI02(5)
Oxidation/Absorption/ Reduction
GAS PHASE OX. (MITSUBISHI,
ISHIK, CHIYODA, SUMITOMO,
MORETANA) (2-3)
PHOSPHATE ROCK (PIRCON
PECK) (3)
Figure 3. Flue Gas Cleanup NOX/SOX Removal -Liquid Phase Technology-
-------
TABLE 3. FLUE GAS CLEANUP NO /SO REMOVAL
x' x
TECHNOLOGY ASSESSMENT
Technology
Drawbacks to Process
Breakthrough Key
SCR/Sorption
- Carbon Cat.
- Metal Oxide
- CuO
SCR + FGD
Thermal D-NO
E-Irradiation
Absorption/
Reduction
Electrolytic-
Oxidation/SCR
Oxidation/
Absorption/
Reduction
Oxidation/
Absorption/
Oxidation
NH3 Cost Sensitivity-
Use High Carbon and NH3
Need Large Reactor
Need H-C for Regeneration
NH3 Cost Sensitivity-
High NH3 and Catalyst
NH3 Breakthrough
Need High/Close Temp. Control
High Cost of E-Guns
N03 Disposition (Lime Slurry)
NH3 Cost Sensitivity (NH3 Inj.)-
Need High L/G and Large R
Limited to High Sulfur Coal
Unknown
High Cost of Ozone or C102
Creates Waste N03
Need Catalyst if Insoluble
Absorbent
May Create Waste
Creates Waste N03
Alternate Source
More Reactive Sorbents
or Catalysts
Coal Reduction
Alternate Source
Improved Catalyst
Improved Catalyst
None?
Lower E-Gun $
Low Cost N03 Disp./Util
Alternate Source
Higher Reactivity
(Catalyst Dev.)
More Data
Low Cost Ozone
Low Cost N03 Disp./Util,
Improved Catalyst
Fixation
Low Cost N03 Disp./Util.
* Operating cost of processes that rely on NH3 to reduce NO will increase
significantly if NH3 cost increases by factor of 3.9 by 1*90, as predicted
for its natural gas feedstock.
6-8
-------
TABLE 4. FLUE GAS CLEANUP NO /SO /PARTICULATE REMOVAL
A A
KEY TECHNICAL BARRIERS
• NO
x
"NO" is generally insoluble and unreactive.
Solution:
A) Find additive for reducing NO to N2-
B) Find additive to convert NO to a water-soluble species.
• S02
A) Slurry sorbents (Ca base) require high L/G; scale and create sludge.
B) Solution sorbents (Na base) produce a highly soluble Na2S04 waste
(arid regions only).
Solution:
A) Dual alkali: solution sorbents are regenerated with Ca.
B) Forced oxidation used to convert sludge to gypsum.
C) Spray dryer produces dry waste materials.
• Particulates
High resistivity ash (ESP)
Solution:
Use precharger or add conditioners.
High AP (baghouse)
Solution:
Use electrostatic bag.
6-9
-------
TABLE 5. ETC AND NATIONAL LABORATORY ACTIVITIES
Organization Activity
PETC Experimental studies with fluidized-bed CuO absorber
and regenerator.
GFPO Calcium compound injection, followed by selective
catalytic reduction with high-temperature baghouse.
Evaluation of particulate control technology.
METC Measurement of ultrafine particulate matter and
mechanisms of aerosol formation.
ANL Enhancement of NO removal by aqueous scrubber
chemistries. Low-temperature, gas-phase reduction
of NO .
x
LBL Mechanistic studies: of coordination of NO to metal
chelates, reaction between nitrosyl metal chelates
and absorbed S02, and NO oxidation by 02 activated
on metal chelates.
6-10
-------
included testing of a lime spray dryer on high sulfur Eastern coal in a
500-lb coal/hr test facility; dry injection using nahcolite, trona, and
sodium bicarbonate; equilibrium studies of adsorption/steam stripping; and
bench-scale tests to evaluate the fluidized-bed copper oxide combined NO /SO
process in a six-inch-diameter fluidized-bed test unit.
The PETC fluidized-bed copper oxide process is chemically identical to
the Shell/UOP process, which was tested earlier under a joint DOE/EPA program
at Tampa Electric Company's Big Bend Station in North Ruskin, Fla. The
processes differ primarily in the contactor. The PETC concept uses a lower
cost fluidized bed, while the Shell/UOP process uses a parallel passage
fixed-bed reactor.
The PETC tests on the six-inch-diameter fluidized-bed system appeared so
promising that the process was scaled up and tested on the 500-lb coal/hr
test facility during 1983. Potential advantages and disadvantages of the
fluidized-bed design are shown in Figure 4. Testing has been completed and
major preliminary test results are summarized in Table 6.
The major problem encountered during the recent tests was sorbent attri-
tion. The sorbent support material is alumina spheres which are impregnated
with copper. Attrition losses of about 0.5% were observed during the recent
tests, which is over ten times greater than in similar tests in the six-inch
bench-scale unit. Such losses are a key variable in determining makeup
requirements, which, in turn, strongly impact the economic feasibility of the
process. The attrition losses observed are considered excessive. A test
program has been developed for FY84 to identify and correct the cause(s) of
the attrition problem.
A preliminary cost estimate for the fluidized-bed copper oxide process
to produce sulfuric acid was made by Science Management Corporation, using
the EPRI-recommended cost guidelines for a 500-MW utility boiler application
burning a 3.5% sulfur coal. The results indicate capital investment of
$93/kW to $98/kW, and levelized revenue requirements of 12 mills/kW-hr (with'-
out credit). Although the cost estimates have been carefully reviewed by
DOE, we plan te submit them for independent analysis by TVA or EPRI or both.
Grand Forks Project Office Projects
The GFPO manages the University of N-orth Dakota Energy Research Center
(UNDERC) in research primarily related to low-rank coals, including North
Dakota lignites. Prior activity under the DOE Flue Gas Cleanup Program
includes lime spray dryer studies and dry scrubbing in a baghouse, using both
trona and nahcolite applied to low-rank-coal-fired boilers. Limestone boiler
injection studies and characterization of low-rank-coal ashes have also been
included.
The GFPO is currently making an assessment of boiler injection of calcium
compounds followed by selective catalytic reduction using a high-temperature
baghouse. Small-scale tests have been conducted using an alternate material
for boiler injection to replace the limestone normally used. The material is
pressure-hydrated lime and shows promise for substantially improved S02
6-11
-------
Advantages
Absorber and regenerator can operate at different
temperatures.
No valve problems.
Constant absorber and regenerator offgases.
No fly ash buildup.
Disadvantages
Possible higher pressure drop.
Attrition potential.
Figure 4. Fluidized-Bed Design (CuO)
TABLE 6. RESULTS OF COPPER OXIDE TESTING IN THE
500-LB/HR COMBUSTION TEST FACILITY
S02 Removal, %
NO Removal, %
X
Inlet S02 (dry) , ppm
Inlet NO (dry), ppm
1
90
93
2354
520
Test Number
2
91
94
2680
500
3
91
93
3100
510
Bed Temperature, °F/°C 750/400 750/400 760/404
Mole S02 Removed per
Mole Available Cu 0.57 0.506 0.532
6-12
-------
removal, at lower stoichiometries, compared to limestone. For 1984, they are
planning to coordinate larger-scale field tests at the Hoot Lake Station,
where limestone boiler injection tests were conducted earlier.
Also in progress at UNDERC are studies to characterize low-rank coal
(LRC) particulates. This effort includes measurement of trace metal species
found in LRC fly ash.
Morgantown Energy Technology Center Projects
The METC project consists of a basic research study involving the mea-
surement of ultrafine particulate matter and mechanisms of aerosol formation.
The two-year study project will be initiated during 1984.
Argonne National Laboratory (ANL) Projects
ANL has initiated a research effort in 1983 for the enhancement of NO
removal by aqueous scrubber chemistries and by low-temperature, gas-phase
reduction of NO . Most of this year has been devoted to literature searches
and bench-scale tests of a variety of additives for wet scrubbers to enhance
NO removal. Work in FY84 will emphasize novel approaches to low-temperature
NO reduction.
x
Lawrence Berkeley Laboratory (LBL) Projects
For the past three years, the LBL effort under the Flue Gas Cleanup
Program has been directed toward basic aqueous chemistry studies pertaining
to reactions involved in combined NO /SO processes. The work has primarily
centered on the chemistry of aqueous metal chelate systems, similar to those
used in several Japanese wet S02/NO cleanup processes.
A
Workers at LBL have measured rate and activity coefficients, have devel-
oped metal chelates with at least an order of magnitude improved reaction
rates compared to NTA and EDTA, and have developed methods to avoid the large
catalyst losses and high water usage encountered earlier with these types of
processes in Japan.
ELECTRON-BEAM PROJECTS
The E-Beam projects were among the first initiated under the DOE Flue
Gas Cleanup Program. A paper entitled "DOE's Electron-Beam Irradiation
Developmental Program" was presented by Edward C. Trexler at the last FGD
Symposium in Hollywood, Fla. The specific projects were described in detail
at that time, and only a brief description and update of their current status
will be given in the following section.
There are two major projects: one with Research-Cottrell to evaluate
the E-Beam/lime spray dryer process, and another with Ebara International
Corporation to evaluate the E-Beam/ammonia injection process. Other E-Beam
projects include a kinetic study by the AVCO Everett Research Laboratory, and
a study by Florida State University of an E-Beam precharger to improve fine
particulate capture. The various projects are summarized in Table 7.
6-13
-------
TABLE 7. DOE SUPPORTED E-BEAM PROCESSES AND SUPPORT STUDIES
Program Elements
E-Beara/ Ammonia
Process
10-20,000 SCFM
PDU Slipstream
E-Beam/Lime
Process
10,000 ACFM
PDU Slipstream
E-Beam
Kinetics
Bench-Scale Tests
and Modeling
E-Beam
Overview
Independent
Studies
Eastern High
Sulfur Coal
Optimization Test
Reliability Tests
Fertilizer Studies
Ebara
Cost Share
2 Years
Eastern High
Sulfur Coal
Optimization Test
Nitrate Fixation
Tests
E-Gun Cost Reduc-
tion Studies
Research-Cottrell
Cost Share
2 Years
18 Months
6-14
-------
E-Beam chemistry is briefly summarized in Figure 5. A more detailed
discussion of the E-Beam chemistry and the E-Beam program can be found in the
paper by Edward C. Trexler mentioned above. Potential benefits are shown in
Figure 6, and comparative costs of E-Beam versus other possible options are
given in Table 8. However, caution is advised at this time, since the E-Beam
cost data are very preliminary. Costs are presently used as a guide to
assess the future potential of the technology -- not as a basis for judgment
concerning commercial application. The E-Beam process costs are highly
sensitive to the cost of the E-Beam guns, the relative cost of ammonia versus
the sale price of fertilizer by-product (for the ammonia injection process),
and the cost of by-product disposal (for lime spray dryer process). All of
these factors will be studied in detail during the future course of the
project effort.
Spray Dryer/E-Beam Project
This project is being conducted at TVA's Shawnee Steam Plant near
Paducah, Ky. Actually, two separate test programs are planned at Shawnee by
Research-Cottrell, one for DOE and another for TVA. The overall effort is
cooperatively funded by DOE, TVA, and Research-Cottrell. An overall sche-
matic of the process is shown in Figure 7. A dry landfill waste material is
produced from this process.
Due primarily to contractual problems, the project is approximately nine
months behind schedule. Installation of equipment at Shawnee also could not
be started until the contractual arrangements had been resolved.
Construction at Shawnee should be completed in October 1983, after which
shakedown tests are scheduled. The DOE test program will be conducted over a
six-month period.
Ammonia Injection/E-Beam Project
This project will be conducted by Ebara International Corporation at
Indianapolis Power and Light's E. W. Stout plant. (Originally, AVCO had the
lead role and designed the test facility under a previous DOE contract, with
Ebara in a support role.) The project will be cost-shared between DOE and
the private sector. Private sector funding has now been secured and includes
Ebara, the State of Indiana, several Indiana-based utilities, a coal company,
and a fertilizer company.
Contractual arrangements were completed in late September 1983, and
construction of the test facility has now been initiated. An overall
schematic of the process is shown in Figure 8. The test program will be
conducted over a twelve-month period. A dry fertilizer by-product is pro-
duced from this process.
E-Beam Kinetic Study
A study of E-Beam kinetics to develop a model of the process was initi-
ated in 1982. This effort is scheduled for completion in December 1983. A
final report will be issued and the model developed will be made available
for use.
6-15
-------
1. Flue gas is cooled and moisture added.
2. Electron beam bombards flue gas components to
produce neutral dissociated atoms.
3. Dissociated atoms react with NO and SO ; reactions
x x'
produce NO and SO acids.
r xx
4. NO and SO acids react with NHg (in Ebara process)
X X
and CaO (in Research-Cottrell process) to form
solid ammonium or calcium salts, respectively.
Figure 5. E-Beam Chemistry
6-16
-------
• Preliminary estimates show potential savings of
20% to 25% compared to lime/limestone scrubber
coupled with selective catalytic reduction (SCR).
• Dry system, which produces a dry product.
• Fertilizer product may be salable (for Ebara process).
• Dry product may be disposed of easily (for Research-
Cottrell process).
Figure 6. E-Beam Potential Benefits
TABLE 8. COMPARATIVE COSTS: E-BEAM VS. OTHER OPTIONS
SO and NO
X X
Control Approach
Lime/Limestone Scrubber
Plus Separate NO Control
A
Lime Spray Dryer Plus
Separate NO Control
A
Mature E-Beam Combined
NO /SO System
A A
Capital
Cost ($/kW)
255
220
205
Operating Cost
(mills/kW-hr)
22
19
15
Reference Footnote on page 2
6-17
-------
I
M
00
INLET FLUE
LIME
WATER
SPRAY
REACTOR
E-BEAM
MACHINE
M
REACTION
CHAMBER
FABRIC FILTER
COMPARTMENTS
SOLIDS RECYCLE
I.D.FAN STACK
TO SOLIDS DISPOSAL
REAGENT
PREPARATION
Figure 7. The Spray Dryer, Electron-Beam System
-------
I
M
v£>
BOILER
MECHANICAL
COLLECTOR
COAL AIR
FLY ASH
NH,
1
SPRAY
COOLER
WATER
E-BEAM
REACTOR
STACK
E-BEAM
GUN
ID FAN
PARTICLE
COLLECTOR
FERTILIZER
Figure 8. Ebara E-Beam Process Schematic Diagram
-------
E-Beam Precharging Study
This study is being conducted at Florida State University (FSU) under a
DOE contract initiated in late 1980. The precharger shows promise for substan-
tial improvement in fine particulate capture. Numerous mechanical problems
have been encountered in setting up the test apparatus, which is shown sche-
matically in Figure 9. Testing has been initiated and will continue during
the next two years.
SUPPORT STUDIES
Acoustic Agglomeration Study
This project was initiated with the Pennsylvania State University in
1980 to evaluate acoustic agglomeration as a means to improve capture of fine
particulates. The original work using various aerosol materials has been
completed and a final report is being published.
The results appear promising, but additional work was recommended to
improve performance and to run tests on an actual flue gas from an operating
coal-fired boiler. These additional data are needed to verify and improve
the model. A follow-on contract has been negotiated and further work is now
in progress.
Chemical Thermodynamic Studies
This project was initiated with the National Bureau of Standards (NBS)
in 1980 under the lime/limestone scrubber program. Under an Interagency
Agreement, NBS was to study the basic chemical thermodynamics and kinetics of
S02 reactions involved in FGD processes for the purpose of establishing a
data base for these reactions. As part of the redirection of the Flue Gas
Cleanup Program to combined NO /SO removal, the effort was redirected to
emphasize the chemical thermodynamics of reactions involved in combined
NO /SO processes, and to establish a data base for these reactions.
A A
Systems Study of Integrated Environmental Control Technologies
This project is being conducted by Carnegie-Mellon University, Pittsburgh,
Pa. The effort will primarily involve the expansion of existing computer-based
models capable of analyzing power plant control options for conventional air
pollution control technologies from pre-combustion through post-combustion.
Previous studies in this area have been sponsored by both EPA and DOE. The
current effort will be to integrate results from the advanced combined NO /
SO /particulate projects, as they become available, into the overall system
capability. This should provide even greater flexibility in assessing overall
control options to optimize multimedia environmental emissions, energy effi-
ciency, resource requirements, and economic costs. The model may also be
useful in helping to guide future research efforts. The project is just now
getting underway.
6-20
-------
Window
1
111
Window
III
Fan
) 1 y v \\
Particle Injection Section
Seam Tube
Ductwork Junctions
Collector
Plates Wires
Outlet Damper
PLAN VIEW
-1'-
-2'-
Figure 9. FSU Electron Beam Precipitator Test System
-------
PROJECTS NEARING COMPLETION
Spray Dryer/Limestone Injection Pilot Evaluation
This project was conducted for DOE by Babcock and Wilcox (B&W) at their
Alliance, Ohio, Research Center using high sulfur Eastern coal. The project
has been completed and the final report is being prepared. A separate paper
discussing the results will be presented by John Doyle, of B&W, at this
Symposium.
NOXSO Process Pilot Evaluation
Briefly, the NOXSO process utilizes a dry sodium/alumina sorbent oper-
ating in a fluidized-bed adsorber at about 120°C to obtain 90% removal of
both S02 and NO . The catalyst is regenerated by reaction with H2S at about
550°C, to produce elemental sulfur. The NO is converted to N£ during regen-
eration. A simplified process schematic is shown in Figure 10.
The present NOXSO catalyst is, in some ways, similar to alkalized alu-
mina, which was tested extensively by the U. S. Bureau of Mines at Pittsburgh
(predecessor to PETC). The primary problem with the original alkalized
alumina process was excessive sorbent attrition loss. The NOXSO sorbent is
prepared by impregnation of sodium on a very attrition-resistant support
supplied by the Davison Corp., while the original alkalized alumina was
prepared by co-precipitation. Additionally, the NOXSO absorption step oper-
ates at about 120°C, while the alkalized alumina absorption was done at about
325°C. Attrition was found to be a strong function of temperature as well as
other variables.
During the current bench-scale test program, which extends through
December 1983, a number of sorbents were tested for 5-20 cycles to identify
the impact of process conditions and sorbent composition on attrition, S02
and NO removal, and capacity. It appears from these tests that attrition
can be kept at acceptably low levels. S02 and NO removals are over 90%.
A
PROJECTS INITIATED IN LATE-1983
To solicit ideas for the newly reoriented program, a Program Research
and Development Announcement (PRDA) for Advanced Combined NO /SO /Particulate
Control Processes was issued in 1982. The solicitation resulted in fourteen
responses, which were evaluated and rated according to predetermined evalua-
tion criteria specified in the PRDA. Five of the most promising proposals
were selected early in 1983 for negotiation for contract award. For the most
part, these contracts have only recently been signed, and work is just begin-
ning. The five projects are described below.
Glow Discharge Irradiation Process Evaluation
The project is being conducted by Westinghouse R&D Center, Pittsburgh,
Pa. The process is based on the concept of glow discharge irradiation of the
flue gas stream to remove high levels of both the NO and S02 without ammonia
addition. In concept, the process is similar to the E-Beam irradiation
6-22
-------
I
NJ
PARTICULATE
COLLECTION
BOILER
FEED WATER
FUEL
TAIL GAS
INCINERATOR
.x^
- ]
x^-
\///^7/
t
t
—
ELEMENTAL
SULFUR
—
*V
k
•^
H,S
GENERATION
REDUCTANT
GAS
J
/,
Figure 10. NOXSO Process Flow Diagram
-------
processes. Potential advantages include the following: (1) less power is
required than for the E-Beam; (2) no additives, such as ammonia, were pro-
posed; however, one of the key questions is whether the by-product material
will be salable as a fertilizer without additives; and (3) the process is
potentially a simple, low-cost system that should require no radiation pro-
tection and that could even be installed in the existing ductwork.
A separate downstream particulate collection device would be required.
Small bench-scale tests are first planned to study the concept further. If
these are promising, a small pilot plant will be constructed and operated on
an actual coal-fired boiler flue gas for evaluation. The project is just now
getting underway, and if the process works as proposed, it could be a very
attractive, simple, low-cost system applicable to either new or retrofit
situations.
Moving-Bed Copper Oxide Process Evaluation
The project is being conducted by Rockwell International, Canoga Park,
Calif. The process chemistry is similar to those of the Shell/UOP fixed-bed
and the PETC fluidized-bed copper oxide processes. Since the proposed proc-
ess involves continuous sorbent circulation through a moving-bed filter, it
should retain all of the potential advantages of PETC's fluidized-bed approach,
with the added advantage of obtaining NO , 862, and particulate control in a
single reactor vessel.
Copper oxide processes have been tested adequately to demonstrate that
they work well for up to 90% control of S02 and NO . Therefore, the thrust
of this project will be to evaluate the capability of the proposed process to
also control particulate simultaneously with S02 and NO in a cost-effective
manner. This implies meeting the targeted control levels for S02, NO , and
particulates at a reasonable pressure drop and without excessive catalyst
attrition losses. The approach will be to modify and test an existing rela-
tively small-scale contactor/filter on a simulated coal-fired flue gas for
further evaluation. If these tests appear promising, the process may be
scaled up to a larger pilot plant size for testing on actual coal-fired flue
gas. The project is just now getting underway and is the only process being
tested with the potential to control all three -- S02, NO , and particulates --
in a single reactor vessel. x
Sulf-X (FeS-Based System) Process Evaluation
The project is being conducted by NUS Corporation, Gaithersburg, Md.
The process concept is not new. A test facility has been installed and
operated at the Western Center Hospital in Canonsburg, Pa., for the past two
years. The process is based on aqueous absorption of S02 and NO , using a
slurry of FeS that can be prepared from waste products. The spent reagent is
regenerated with coal or coke to produce sulfur. Results to date indicate
that up to 90% S02 control and over 40% NO removal can be achieved with the
present unit. Modifications of the absorber were proposed to increase NO
absorption to 90%. x
6-24
-------
Zinc Oxide Spray Dryer Process Bench-Scale Tests
The project is being conducted by Battelle Columbus Laboratory. The
process concept is based on the use of ZnO in a spray dryer to remove up to
90% of both S02 and NO . Only limited fixed-bed data were available, espe-
cially for NO control. Small bench-scale tests are planned at this time to
further evaluate whether or not the proposed concept shows potential for
further testing. The project is just now getting underway.
Zeolite Catalyst (Using Fe or Cu) Process Bench-Scale Tests
The project, is being conducted by Science Applications, Inc., San Diego,
Calif. The process concept is based on the use of a synthetic zeolite catalyst
impregnated with Fe or Cu, ammonia addition for NO control, and catalyst
regeneration to produce either sulfuric acid or elemental sulfur. Only
limited work has been done to identify possible candidates for an improved
catalyst material. Small bench-scale tests are planned at this time to
evaluate the proposed catalysts and determine if they show potential for
further testing. The project is just now getting underway.
SUMMARY
CURRENT PROGRAM
The current program supports the development of cleanup technology to
promote the use of abundant coal resources in an economical and environmen-
tally acceptable manner. Simple, economical processes capable of 90% control
of NO and S02, or showing improved capture of fine particulate, preferably
in a single reactor vessel, are emphasized. Projects include both DOE and
National Laboratory research and contracted efforts with industry, not-for-
profit research organizations, and universities. An effort is being made to
keep a balanced program between basic research studies, longer-range concepts
and ideas, and processes that are at the proof-of-concept stage. Much of the
current effort is really just now getting started.
6-25
-------
STATUS OF S02 AND NOX REMOVAL IN JAPAN
J. Ando
-------
STATUS OF S00 AND NO REMOVAL IN JAPAN
2 x
Jumpei Ando
Faculty of Science and Engineering
Chuo University
Kasuga, Bunkyo-ku,
Tokyo 112, Japan
Flue gas desulfurizatian (EGD) capacity in Japan has increased rapidly
from 1970, reaching 10 Nm /h (33,000 MW equivalent) in 1976, and
increasing slowly since. For nitrogen oxide (NO ) removal, selective
catalytic reduction (SCR) of NO has been used in,addition to combustion
modification. The SCR capacityxincreased from 10 Nm /h in 1976 to
7 x 10 Nm /h in 1982. A combined flue gas cleaning system (including
SCR, electrostatic precipitator (ESP), and FGD) has been applied to
industrial boilers and furnaces since 1975 and to coal-fired boilers
since 1980 to remove 50-90% of NO and 90-95% of SO-. The combined
cleaning for coal costs 2-3 yen*/kWh including 7 years depreciation.
Many new coal-fired boilers with the combined cleaning system have been
planned because of their economic advantage over the use of low- or
high-sulfur oil with FGD, although the recent decrease in oil price has
reduced considerably the economic advantage. Efforts have been made to
improve the combined system mainly by (1) reducing FGD cost by simplifying
or modifying the process,.and by saving energy, (2) reducing particulate
emissions below 30 mg/Nm by ESP and wet FGD, and (3) improving gas
reheating after wet FGD.
Simultaneous S09 and NO removal processes were studied eagerly between
1973 and 1978 and applied to several small industrial boilers and furnaces
but have not been used at a large scale because of the problems involved.
1. General Aspects of FGD
Table 1 lists major constructors of FGD plants and numbers and capacities
of plants larger than 10,000 Nm /h that either are operational or will
be complete by the end of 1983.,. The plants total 470, and their combined
capacity reaches 111,000,000 Nm /h (equivalent to 36,000 MW). About 60%
of the capacity is accounted for by utility boilers, including 6,200 MW
for coal-fired boilers. About 50% of all the plants, in terms of capacity,
use the wet lime/limestone process to by-produce gypsum; 18% use the
indirect lime/limestone process (double alkali type) and modified lime/
limestone process; 10% use the regenerable process, by-producing sulfuric
acid, ammonium sulfate, and sulfur; and 22% use sodium scrubbing to by-
produce sodium sulfite or sulfate. About 80% of the sodium scrubbing
plants by-produce sodium sulfite for paper mills. In addition to the
262 sodium scrubbing plants listed in Table 1, there are nearly 900
smaller ones by-producing mainly waste sodium sulfite or sulfate liquor.
*August 5, 1983 exchange rate: 244 yen = $1
6-27
-------
TABLE 1. NUMBERS AND CAPACITIES (1,000 Nm /h) OF FGD PLANTS IN JAPAN BY MAJOR CONSTRUCTORS
(CAPACITY LARGER THAN 10,000 NmJ/h, OPERATIONAL OR TO BE COMPLETED BY END OF 1983)
I
to
00
By-product
Gypsum H0SO, ,
Plant constructor
Mitsubishi Heavy Industries (MHI)
Ishikawajima-Harima H.I. (IHI)
Hitachi Ltd. & Babcock-Hitachi (BHK)
Kawasaki Heavy Industries (KHI)
Mitsubishi Kakoki (MKK)
Oji Koei (Oji Engineering)
Fujikasui Engineering
Chiyoda Chem. Eng. Construction
Kurabo Engineering
Mitsui Miike Machinery Co. (MMMC)
Tsukishima Kikai Co. (TSK)
Ebara Corp. (Ebara Manufacturing)
Nippon Kokan (NKK)
Nippon Steel
Kobe Steel
Kureha Chemical Industries (KCI)
Showa Engineering & Showa Denko
Mitsui Metal Engineering
Dowa Koei (Dowa Engineering)
Sumitomo Chemical Engineering (SCEC)
Niigata Iron Works
J.G.C. Corp.
Total
Direct3
41
20
17
2
5
7
2
6
1
1
3
2
4
2
1
114
(25,019)
( 8,633)
(10,200)
( 174)
( 506)
-
( 4,144)
( 470)
,
( 3,653)°
( 80)
( 26)
( 197)
( 2.100)6
-
-
-
( 1,006)
-
-
( 265)
( 330)
(56,803)
S
I
Ja,S
o,
Indirect (NH4)_S04 Na_SO,
H ( 7
13 ( 4
5 (
4 (
11 ( 1
1 (
6 ( 1
1 (
3 ( 1
11 ( 1
66 (19
_
-
-
,966)
-
-
-
,375)
588)
-
430)
,520)
150)
-
,925)
40)
,002)
-
,113)
-
-
—
,109)
_
-
2 (
-
13 ( 6,
-
-
-
1 (
1 (
1 (
-
2 ( 1,
-
-
-
-
2 (
-
5 (
-
1 (
28 (10,
590)
606)
18)
520)
110)
990)
130)
748)
125)
837)
3
26
19
7
3
32
18
102
31
10
7
2
2
262
(
( 3
(
(
(
( 6
( 1
( 3
( 3
( 2
( 1
(
(
(24
292)
,240)
843)
256)
530)
,239)
,716)C
-
,912)C
-
,553)
,006)
-
-
-
,380)
370)
-
-
-
220)
-
,557)
Total
44 (
46 (
38 (
20 (
21 (
32 (
25 (
15 (
108 (
7 (
37 (
22 (
6 (
2 (
6 (
8 (
5 (
6 (
11 (
5 (
4 (
2 (
25,311)
11,873)
11,633)
8,396)
7,642)
6,239)
5,860)
4,845)
4,518)
4,173)
4,173)
3,552)
2,337)
2,100)
1,925)
1,420)
1,372)
1,136)
1,113)
748)
485)
455)
470 (111,306)
Wet lime/limestone process
Indirect and modified lime/limestone process
Including waste MgSO, liquor
j ^
Including one plant by-producing throwaway sludge
Using converter slag as the absorbent
-------
Most of the FGD plants were constructed between 1972 and 1977. The
growth of the FGD capacity has been slow since 1977 because the ambient
S09 concentrations have been lowered and not many new SO- sources have
been constructed due to the economic depression. The FGD plants
constructed since 1978 include 18 plants involving the wet lime/limestone
process with a total capacity of 14,600,000 Nm /h (4,800 MW, mainly
limestone process for coal-fired boilers) and 19 plants involving the
indirect and modified lime/limestone gypsum process with a total capacity
of 5,700,000 Nm /h (1,900 MW). Virtually no plants involving other
processes have been constructed since 1978.
Gypsum supply in Japan has been short since 1978 due to the increase in
demand and the decrease of by-product gypsum from wet-process phosphoric
acid production. Although gypsum by-producing by FGD costs 1.5-2 yen/kWh
(6-8 mills/kWh) including. 7 years depreciation, the process has been
more useful than other processes. FGD processes by-producing elemental
sulfur have been used at oil refineries by introducing the recovered
S0? into existing Glaus furnaces. A demonstration plant is under
construction at a coal-fired power station to by-produce elemental
sulfur by the activated carbon process.
2. FGD Plants for Coal-Fired Boilers
Table 2 lists coal-fired boilers and their FGD and SCR units. FGD has
been applied to nearly all coal-fired boilers except several small ones
in Hokkaido (Northern Island) not shown in the Table. All coal-fired ~
boilers have an ESP upstream of FGD to reduce particulates to 50-1,000 mg/Nm
(removal efficiency 95-99.8%). All FGD units for coal-fired boilers use
a wet lime/limestone process (there are six types, shown in Figure 1 and
Tables 2 and 3). Type I by-produces throwaway sludge; others by-produce
gypsum.
Most of the new FGD plants use Type IV which has a prescrubber with a
separate liquor loop. The prescrubber has the following functions:
(1) remove fluorine.in flue gas which tends to lower S0_ removal efficiency,
in combination with other impurities (Figure 2); (2) remove chlorine in
flue gas which tends to promote corrosion and lower the SCL removal
efficiency; (3) remove fly ash and increase the purity of By-product
gypsum; and (4) cool the gas to prevent scaling and to protect linings
in the scrubber. On the other hand, the prescrubber raises the FGD
plant cost by 3-7% and increases the wastewater; therefore, it may be
eliminated for coals with less impurities that depress the S07 removal
efficiency or when the adverse effect of the impurities is depressed by
certain additives.
Most of the new plants use a small amount of sulfur acid to reduce the
pH of the calcium sulfite slurry to 4.0-4.5 to attain complete oxidation
to by-produce good-quality salable gypsum; however, the units using the
Chiyoda jet bubbling reactor (JBR) process (Type V) and the Kawasaki
magnesium gypsum process (Type IV), can attain perfect oxidation without
sulfuric acid addition.
6-29
-------
TABLE 2. FGD AND SCR UNITS FOR COAL-FIRED BOILERS (LARGER THAN 75 MW)
I
LO
O
Boiler
Owner
Mitsui
Aluminum
Mitsui
Aluminum
EPDC
EPDC
EPDC
EPDC
EPDC
EPDC
EPDC
EPDC
Chugoku
Electric
Chugoku
Electric
Chugoku
Electric
Chugoku
Electric
Plant site
Orauta
Omuta
Takasago
Takasago
Isogo
I so go
Takehara
Takehara
Matsushima
Matsushima
Shimonoseki
Shin-Ube
Shin-Ube
Shin-Ube
No.
1
2
1
2
1
2
1
3
1
2
1
1
2
3
MW
156
175
250
250
265
265
250
700
5008
5008
175
75
75
156
N/Ra
R
N
R
R
R
R
R
N
N
N
R
R
R
R
Constructor
MMMC°
MMMC
MMMC
MMMC
IHId
IHI
BHK6
IHI
IHI, MMMC
BHK, MMMC
MHIh
MHI
MHI
MHI
FGD
Completion
1972
1975
1975
1976
1976
1976
1977
1983
1981
1981
1979
1982
1982
1982
Type"
I
II
II
II
III
III
III
III
IV
IV
IV
IV
IV
IV
SCR
Constructor Completion Type
BHK, KHIf 1980 L1
BHK 1983 L
MHI 1980 Hm
MHI 1982 H
MHI 1982 H
MHI 1982 H
-------
TABLE 2. FGD AND SCR UNITS FOR COAL-FIRED BOILERS (LARGER THAN 75 MW) (Continued)
Boiler
Owner
Chugoku
Electric
Chugoku
Electric
Hokkaido
Electric
Hokkaido
Electric
Joban
Joint
Joban
Joint
Kyushu
Electric
Kyushu
Electric
Sakata
Joint
Sakata
Joint
Shikoku
Electric
Shikoku
Electric
Plant site
Mizushima
Mizushima
Tomato-Atsuma
Tomato-Atsuma
Nakoso
Nakoso
Omura
Minato
Sakata
Sakata
Saijo
Saijo
No.
1
2
1
2
8
9
2
1
1
2
1
2
MW N/R3
125 R
156 R
3501 N
600 N
600J N
600J N
156 R
156 R
350 N
350 N
175 R
230 R
FGD
Constructor Completion
BHK 1984
BHK 1984
BHK 1980
MHI 1984
MHI 1983
MHI 1983
MHI 1982
MMMC 1983
MHI 1984
MHI 1985
KHI 1983
KHI 1984
Typeb
IV
IV
IV
IV
IV
IV
IV
III
IV
IV
VI
VI
Constructor
BHK
BHK
BHK
MHI
IHI
MHI
MHI
MHI
MHI
IHI
SCR
Completion
1984
1984
1980
1983
1983
1983
1984
1984
1983
1984
Type
H
H
L
H
H
H
H
H
H
H
-------
TABLE 2. FGD AND SCR UNITS FOR COAL-FIRED BOILERS (LARGER THAN 75 MW) (Continued)
Boiler
&>
1
u>
N3
Owner
Tohoku
Electric
Tohoku
Electric
Tokyo
Electric
Tokyo
Electric
Toyama
Joint
Toyama
Joint
Plant site
Sendai
Sendai
Yokosuka
Yokosuka
Toyama
Toyama
No.
2
3
1
2
1
2
MW
175
175
265k
265k
200
200
N/Ra
R
R
R
R
R
R
Constructor
BHK
BHK
MHI
BHK
Chiyoda
Chiyoda
FGD
Completion
1983
1983
1985
1985
1984
1984
Typeb
IV
IV
IV
IV
V
V
Constructor
BHK
BHK
MHI
BHK
BHK
BHK
SCR
Completion
1983
1983
1984
1985
1984
1984
Type
H
H
H
H
H
H
New or retrofit
See Figure 1 for type
Slitui Miike Machinery Company
Ishikawajima-Harima Heavy Industries
6Babcock-Hitachi K.K.
Kawasaki Heavy Industries
Subjecting 75% of the gas to FGD
Mitsubishi Heavy Industries
Subjecting 50% of the gas to FGD and 25% of the gas to SCR
Burns both coal and oil
k
Burns coal/oil mixture
Low-dust system
High-dust system
-------
(I)
T Ca(OH)2
*
H20
Slu2 removal efficiency of the
limestone-gypsum process (by Babcock-Hitachi).
6-33
-------
TABLE 3. CLASSIFICATION OF WET LIME/LIMESTONE PROCESSES
_ _
Type By-Product Absorbent Additive Prescrubber Sulfuric Acid
I
II
III
IV
V
VI
Sludge
Gypsum
Gypsum
Gypsum
Gypsum
Gypsum
Ca(OH)
CaCO
CaCO^
CaCO_
CaCO^
Ca(0fl)2
None
Catalyst
Occasional
None
None
MgO
None
None
None
Yes
Yes
Yes
None
None
Yes
Yes
None
None
With a separate liquor loop
To lower pH of calcium sulfite slurry to promote oxidation
Two large FGD plants for new coal-fired boilers have started operation
in 1983: one is the Takehara plant of EPDC with a 700 MW boiler using
a Type III scrubber system in one train; and the other is the Nakoso
plant of Joban Joint Power Co. with a 600 MW boiler using a Type IV
scrubber system in two trains. Both have been operated without trouble,
except a minor start-up problem with a fan at Nakosa, which has been
solved. Operating parameters of those plants are shown in Table 4 in
comparison with those of two new FGD plants for industrial oil-fired
boilers.
The plants in Table 4 are characterized by low power consumption—1.3-
2.1% as compared with 1.7-2.6% for conventional plants by-producing
gypsum. At the Nakoso and Sendai plants, power consumption decreases at
reduced boiler load, while at many conventional plants the power con-
sumption (%) tends to increase at reduced load. The Nakoso and Sendai
plants use improved fans that consume less energy at reduced load. Thus
the power consumption (%) can be reduced due to the decrease of the
pressure drop in addition to the decrease in the gas volume and the
slurry recirculation rate. The FGD cost for those plants may be about
1.5 yen/kWh including 7 years depreciation.
Efforts have been made for further improvement of FGD systems for coal.
Concerning gas reheating, the Ljungstrom gas-gas heater has been improved.
to give little gas leakage. Non-leaking gas reheaters have been tested. ' '
For particulate reduction, wet ESPs (downstream of wet FGD) have been
tested/ '
3. FGD Plants for Industrial Boilers Utilizing New Processes
The Yokkaichi plant of Mitsui Petrochemical is the first commercial
plant utilizing the Chiyoda 121 JBR process (Table 4). It has the
capacity to treat 250,000 Nm /h of flue gas from an oil-fired boiler and
6-34
-------
TABLE 4. OPERATING PARAMETERS OF NEW FGD PLANTS
(ABSORBENT: LIMESTONE. BY-PRODUCT: GYPSUM.
NUMBERS IN PARENTHESES ARE FOR REDUCED LOAD.)
Owner
Site, Boiler No.
Fuel
Boiler capacity, MW ,
Gas treated, 1000 Nm /h
FGD constructor
FGD start-up year
Inlet S02, ppm „
Inlet dust, mg/Nm
1st scrubber (prescrubber)
L/G, liter/Nm
pH
2nd scrubber (scrubber)
Absorbent pH
Additive
L/G, liter/Nm ,
Outlet S02, ppm ~d
Outlet dust, mg/Nm
S02 removal, % ,
Dust removal, %
CaC00 utilization, %8
J *
H?SO, consumption, liter/h
Wastewater, m /h
Gas reheating
Pressure drop, -mm H_0
Power consumption, %
Plant cost, yen/kW
EPDC
Takehara, 3
Coal
700
2200
IHI
1983
600-700
30-150
Spray
5
4.5-5
Spray
5.5-6
None
15
30-35
10-20
95
70-80
Over 99
10
25
GGHT
350,
2.1h
20,000
Joban
Joint
Nakoso, 8
Oil, Coal3
600
1980 (990)
MHI
1983
1000 (900)
150
Spray
2
1
Packed
5.5
None
20
40 (36)
45
96
70
Over 99
13
15
GGH & SGH?
300 (250)1
2.0 (1.7)1
21.0001
Mitsubishi
Petrochem.
Yokkaichi
Oil
85
250
Chiyoda
1982
1500
180
Venturi
1
1
JBR
3.5
None
35
35
98
78
Over 99
None
1.5
None
600
1.3
18,000
Toyo
Rubber
Sendai
Oil
14
40 (30)
Kureha
1982
1500
150
None
h
P.P.
5.5
S.S.°
2.7
5
75
99.7
50
Over 99
None
0
None
200 (170)
1.4 (1.3)
19,000
g
Oil - 90: Coal - 10, on calorie basis
Perforated plate
"Sulfosuccinate
At scrubber outlet
a
"By scrubbers
Gas-gas heater to 100C
Steam-gas heater to 120C
Including GGH
Including GGH and SGH
6-35
-------
was put into operation in May 1982. The process is characterized by the
use of a single reactor containing gypsum slurry at pH of nearly 4 into
which flue gas is injected through many nozzles, limestone powder is
added, and air is introduced for oxidation. The plant has no slurry
recirculation pumps, resulting in a low power consumption. The plant
cost is also low. A continuous centrifuge (super decanter) is useful
because gypsum crystals grow very well. The plant has been operated
without trouble. Although a small amount of scale forms on some of the
gas pipes, the scaling did not hinder its continuous operation for more
than a year. Chiyoda has been constructing a similar unit for Chita Oil
Co. and units for two 200 MW coal-fired boilers (Table 2).
Kureha Chemical modified the acetate lime gypsum process and has developed
the sodium sulfosuccinate limestone process, which has been applied to
the Sendai plant of Toyo Rubber Co. (Table 4). Actually anhydrous
maleic acid is added to the scrubber liquor, which readily reacts with
sodium bisulfite to form sodium sulfosuccinate. 'The role of the sulfo-
succinate may be similar to that of adipic acid used in the U.S.A.
Kureha claims that the decomposition rate of the sulfosuccinate is
slower than that of adipic acid, and that the decomposition products of
the sulfosuccinate give no odor while those from adipic acid may give
some odor. The Sendai plant has no prescrubber and gives no wastewater
at all. The plant has been operated with over 99% reliability. Power
consumption is low because of the small L/G ratio and low pressure drop.
4. NO Control Systems
Extensive combustion modifications have been applied in Japan to lower
NO from boilers to 160-400 ppm for coal, 60-130 ppm for oil, and 40-
80 ppm for gas to meet the emission standards by the Central Government.
In many regions, however, more stringent regulations by local governments
require further NO reduction. Many plants utilizing SCR of NO have
been constructed, as shown in Figures 3 and 4. These plants convert
NO to N by treating with NH at 300-400C in the presence of a catalyst:
X Z. j
4NO + 4NH- + 02 = 4N2 + 6H20
Nearly all of the SCR plants constructed before 1977 are for oil-fired
industrial boilers and furnaces, while nearly all of the plants constructed
since 1978 are for utility boilers. By the end of 1983, the total SCR
capacity will reach 80 x 10 Nm /h (27,000 MW equivalent), of which
about 80% is for utility boilers. By the end of 1984, the total SCR
capacity for utility boilers will reach 75 x 10 Nm /h (25,000 MW), of
which 60% is for oil, 21% for coal, and 19% for gas. The FGD capacity
for coal-fired boilers will increase fairly rapidly because most of the
new coal-fired utility boilers will need SCR (Figure 4).
6-36
-------
80
60
40
ApH 140 ^ Esp L-
,.« >
GGH
1 L ,nu
Low-dust System
jlOO
B
350-400 ., . 350-400 350-400 L_
1 "Ul, -v rpn x Apu .._>
) £sp > .CR ^ API! 14Q ^
GGH
c50
90 '
90
FGD
FGD
B: Boiler APH: Air preheater GGH; Gas-gas heater
Figure 5. Combination of SCR and FGD for coal-fired boilers
(figures show gas temperatures, °C).
6-37
-------
SCR units for coal-fired boilers are shown in Table 2. Most of the
units treat a flue gas from boiler economizer with full dust load at
350-400C (high-dust system, Figure 5) and remove 55-80%. of NO to meet
the regulations, using 0.56-0.83 mole NH to 1 mole NO. Plant operation
is easy and free of trouble. Three of the plants treat the flue gas
after particulate removal by a hot-side ESP (low-dust system, Figure 5).
The low-dust system has a slight tendency to deposit dust on the catalyst
and ammonium bisulfate in the air preheater, but the system may be
operated continuously for 1 year by applying soot blowing. Nearly all
of the SCR plants for coal-fired boilers have been constructed by Mitsubishi
Heavy Industries and Babcock-Hitachi using honeycomb catalyst and plate
catalyst, respectively. The catalyst life may be over two years for coal.
x
The cost of the SCR plant for utility boilers (yen/kW) for 80% NO
removal is 6000-8000 for coal, 4000-6000 for oil, and 2500 for gas. The
annualized operating cost (yen/kWh), including 7 yefurs depreciation, is
0.5-0.8 for coal, 0.3-0.4 for oil, and 0.2 for'gas.
5. Combined S09/N0 Removal (SCR/FGD Combination)
£. X
A combination of FGD and SCR has been applied to industrial boilers and
furnaces since 1975 (Table 5) and to coal-fired boilers since 1980
(Table 2). The types of combinations for industrial use are shown in
Figure 6.
TABLE 5. COMBINED FGD/SCR PLANTS FOR INDUSTRIAL USE
Owner
Shindaikyowa
Petrochem.
Kawasaki
Steel
Nippon
Yakin
Nippon
Kokan (NKK)
Tokyo
Metropolis
Gas
Site Source
Yokkaichi Oil-fired
boiler
Chiba Sintering
machine
Kawasaki Oil-fired
boiler
Keihin Sintering
machine
Sunamachi Incin-
erator
Capacity
Nm3/h
450,000
762,000
15,000
1,320,000
100,000
FGD
MKK-Wellman
V°4
MHI
Lime- gypsum
Sodium
scrubbing
NKK ammonia
scrubbing
Sodium
scrubbing
SCR
Hitachi
Zosen
Hitachi
Zosen
MKKb
NKK
Mitsui
Engineering
Year
Completed
1975
1976
1976
1979
1979
Throwaway liquor
Mitsubishi Kakoki Kaisha
6-38
-------
Shidaikyowa Petrochemical
400
Boiler
150
^
FGD
55
GGH
nri j
295 '—[ 400 ^
Nippon Yakin
Boiler
290
1
GGH
330
330
' H
* "L
370
370
Heat recovery
SCR
s
en
Iou
•*
Kawasaki Steel
Sintering
machine
IbO
•'
FSP
150'
j
FfiD
55
V
Wet
ESP
r
— >
« ,
40
GGH
•J
>
?9B
380
H
380 P
SCR
Nippon Kokan (NKK)
110
Sintering
machine
150
ESP
150
>
|
fTH
"
t
90
>
prn
rbu
— ^-
bO
bO
Wet
ESP
t
L
*?
|
rru
uun
T
> H 1 "> SCR
390 |
Tokyo "etropolis
Incine-
rator
150
>
FGD
55
V
x
Wet
ESP
120
>
cr»
GGH
280
350
350
[GGH~|: Gas-gas heater (F|: Heater
Wet electrostatic
precipitator
Figure 6. FGD/SCR combinations for industrial use
(figures show gas temperature, °C).
6-39
-------
The first plant for an oil-fired boiler was completed in 1975 for
Shindaikyowa Petrochemical at Yokkaichi. The plant treats the flue gas
first by wet FGD and then by SCR. This FGD/SCR combination is no longer
used for boilers because the system requires large amounts of fuel oil
for heating and still has an ammonium bisulfate problem with the gas-gas
heater.
Nippon Yakin's Kawasaki plant treats flue gas first by SCR and then by
FGD. This system is simpler and requires less energy, although a little
gas heating is needed because the gas temperature is 290C at the boiler
outlet. The SCR unit first used a ring catalyst in a random packing,
which caused dust plugging; this was changed to a parallel-flow catalyst
in 1978. The combined system has since been operated without trouble,
reducing NO from 200 to 35 ppm, SCL from 1000 to 50 ppm, and particulate
from 200 tox40 mg/Nm . The new catalyst has been in use for over 4 years.
The plant cost in 1976 was 63 million yen for FGD and 73 million yen for
SCR with the heaters. The operation cost, including 7 years depreciation,
is about 14,000 yen/kl of heavy oil. Since heavy oil and kerosene
currently cost 50,000 yen/kl and 70,000 yen/kl, respectively, giving a
20,000 yen/kl difference, the use of heavy oil with the combined cleaning
system is more advantageous than using kerosene without gas cleaning.
Sintering plants of Kawasaki Steel and Nippon Kokan, as well as an
incinerator of Tokyo Metropolis, have combined wet FGD, wet ESP,'gas
heater, and SCR (Table 5 and Figure 6). The temperature of the flue gas
from the sintering machines is about 150C. A low-temperature SCR
catalyst is not useful for gas because of the high S0_ concentration
which causes ammonium bisulfate to deposit on the catalyst. The flue
gas from the incinerator is also too dirty to be treated directly by
SCR. Therefore, the flue gases are first treated by FGD and wet ESP and
the cleaned gases are heated and subjected to SCR.
The Chiba plant of Kawasaki Steel, using lime-gypsum process FGD and a
pellet catalyst for SCR, reduces S02 from 300 to 10 ppm, NO from 250 to
20 ppm, and particulates from about 500 to below 10 mg/Nm .X The catalyst
was used for 4 years before it was replaced with a new catalyst which works
at a lower temperature for energy saving.
At Nippon Kokan's Keihin plant, S02 in the sintering machine flue gas is
combined with NH., in coke oven gas to by-produce ammonium sulfate. For
SCR, a special iron ore is used as the catalyst in a moving bed. By
the combined cleaning, SO- is reduced from 250 to 20 ppm, particulate is
reduced from 500 to 5 mg/Nm , and NO is reduced from 200 to 40 ppm.
The investment cost for the SCR plan? was 6.7 billion yen including the
gas-gas heater and a direct-fired heater.
The combined systems for the sintering machines have been operated
without trouble, although the operation cost may be high because a large
amount of energy is required for gas heating.
6-40
-------
At the Sunamachi plant of Tokyo Metropolis, flue gas from the city waste
incinerator is first subjected to sodium scrubbing and then to wet ESP,
gas heating, and SCR. A special honeycomb catalyst with a special
composition is used to remove odor together with NO . The plant has
been operated without trouble, removing over 90% of SCL and NO with
about 80% of the odor. The SCR unit cost 480 million yen, including the
gas-gas heater and a direct-fired heater. The SCR cost is about
0.45 yen/m of the gas including 10 years depreciation. About 50% of
the cost is for fuel for heating. Several similar plants are to be
constructed by Mitsui Engineering for incinerators.
For coal-fired boilers, economizer outlet gas at 350-400C is subjected
to SCR and then to air preheater and FGD (Figure 3). Possible problems
with the cleaning system are contamination of fly ash by ammonia for the
high-dust system, ammonium bisulfate deposit in the air preheater for
the low-dust system, and contamination of wastewater from FGD by ammonia
for both systems. Those problems have been solved by keeping ammonia at
the SCR reactor outlet below 5 ppm. Wastewater from FGD plants for
coal-fired boilers without SCR usually contains 100-600 mg/m of
nitrogen in the forms of nitrate and' others derived from NO in the flue
gas. With SCR, a small amount°of ammonia goes into wastewater but total
nitrogen content seems to be lowered, possibly because of the substantial
reduction of nitrogen compounds derived from NO in the flue gas.
X
6. Simultaneous SO,, and NO Removal
2 x
Various simultaneous SO™ and NO removal processes were tested between
1975 and 1979 when the combined SCR/FGD system was considered not feasible.
Several small commercial units were constructed using wet simultaneous
processes by-producing nitrates and sulfates. The simultaneous removal
processes, however, have difficulties (including economic disadvantages)
and have become less attractive as the combined SCR/FGD system has been
commercialized. The above wet processes produce large quantities of
waste by-products which would present major problems in Japan if applied
on a large scale. Wet processes by-producing elemental sulfur along
with ammonia or N« were tested, but seem too complex for commercial
application.
Some dry simultaneous removal processes were also given up because of
their high cost. However, two dry simultaneous removal processes,
activated carbon and electron beam, are to be further developed aiming
at commercial application. A demonstration plant of the activated
carbon process, with a capacity of treating 300,000 Nm /h of flue gas
from a coal-fired boiler, is under construction at Matsushima Station of
EPDC to remove over 90% of S02 with a 30-40% reduction of NO . The
plant will by-produce elemental sulfur by reducing the recovered
concentrated S02 in a reactor containing coal and Glaus furnaces. For
commercial use, very cheap carbon is needed because of the high carbon
consumption. The electron beam process, by-producing ammonium sulfate
and nitrate, may not be suitable in Japan where ammonium sulfate is in
oversupply and the nitrate has little use. The process may be useful in
a country which has a demand for the sulfate and nitrate for fertilizer,
if a large electron beam accelerator is commercialized for a reasonable
cost.
6-41
-------
7. Literature Cited
1. Tokyo Electric Power Co., NO Control Technology and Its
Application for Power Plants. US-Japan NO Information
Exchange Conference, Tokyo (May 1981).
2. Y. Nakabayashi, A. Tamura, T. Miyasaka, and T. Itoh, High
Performance Particulate Control of a Coal-Fired Thermal Power
Plant in Japan, Proceedings of the Fourth Coal Utilization
Technology Symposium. Coal Mining Research Center (August
1982) (in Japanese).
3. J. Ando, NO Abatement for Stationary Sources in Japan, EPA-
600/7-83-02? (NTIS PB 83-207639) (May 1983).
4. J. Ando, S0? Abatement for Coal-Fired Boilers in Japan, EPA-
600/7-83-028 (NTIS PB 83-225938) (May 1983).
6-42
-------
PANEL: THE ARCHITECT-ENGINEER - MIDDLEMAN
BETWEEN UTILITY AND FGD SUPPLIER
Chairman: A. V. Slack
President
SAS Corporation
Route 1, Box 69
Sheffield, AL 35660
-------
THE ARCHITECT-ENGINEER - MIDDLEMAN BETWEEN
UTILITY AND FGD SUPPLIER
E. W. Stenby, G. H. Dyer, P. R. Predick,
M. L. Meadows, D. B. Hammontree,
C. P. Wedig, R. Rao
-------
Panel Discussion. The A&E - Middleman
Between Utility and FGD Supplier
BY: E.W. Stenby
Stearns-Roger Engineering Corporation
Denver, Colorado
ABSTRACT
The A-E is hired to represent the owner through all phases of a scrubber
system selection and installation. In this capacity, the A-E is expected to
exercise good judgement in the areas of design and cost. The owner, depend-
ing on the size of his staff and his experience, may leave all or most deci-
sions to his A-E or he may participate a great deal. We feel it is very
important that the owner participate fully in all aspects of the scrubber
system design and installation, since he will be left to operate the system
after the vendor and A-E have gone home.
The A-E's objective is to specify and procure the most reliable scrubber
system at the low evaluated cost. The scrubber business, however, is highly
competitive and the vendor's objective is to win the award, knowing that low
cost is one of the primary selection parameters. The document used by all
parties to arrive at the appropriate selection is the specification. In
Steams-Roger's opinion, a good spec will detail not only basic design
criteria, but also minimum levels of quality in equipment components.
materials of construction and equipment sparing philosophy. A very important
area is performance guarantees. A conservative spec may tend to drive the
capital cost of the scrubber system upwards, but we consider it to be the
best means currently of achieving high operating reliability. However, a
very detailed spec can conflict with the owner's desire to obtain maximum
vendor exposure on equipment guarantees and performance guarantees. By dic-
tating too much of the system detail design, the A-E and owner take some of
the risk.
The current fixed price nature of the FGD business forces the vendor to
absorb the risk of cost over-runs. During the proposal stage the vendor
usually does minimal engineering to prepare his bid. Under these circum-
stances there will be a tendency for the successful vendor to protect his
profit by calling for an extra, everytime the A-E or owner makes a comment.
The vendor will try to minimize or prevent the owner and A-E from interfering
with the design and construction are in conformance with the spec and the
intent of the spec. Again, the spec becomes the all important document in
interpreting what is required and what is acceptable. It is absolutely
essential, therefore, that a good spec be developed.
In order to ensure that a good proposal be prepared, it is suggested that the
6-43
-------
owner consider paying for the proposal. In this manner, vendors will be
encouraged to do more engineering and provide better detail of his proposed
design. A limited bidder's list goes hand in hand with this approach.
One last thought... it must be remembered that there are many "gray" areas in
scrubber design and it is difficult, if not impossible, to get all the
details spelled out clearly during the bidding phase. A goal of the A-E is
to interpret and help coordinate the owner's requirements into positive
action by the vendor during the design and construction phase. The owner
also has a resposibility to know what he is buying and to apply standards of
operation and maintenance for the scrubber system that are consistent with
the rest of the plant.
The Role of Engineer-Constructors In Flue Gas Desulfurization
By G. H. Dyer
Bechtel Group, Inc.
San Francisco, CA
ABSTRACT
?irst, I would like to note that I have recast the title, since we
think of ourselves as being Engineer-Constructors rather than that of an
Architect-Engineers. Ve believe that to be a distinct difference in that it
provides an ability to provide a continuity between engineering and
construction activities that is vital to the timely and successful completion
of jobs involving complex systems.
Second, in the short time available, I would like to concentrate on the
Engineering end of this subject.
Lastly, as a representative of the Besearch and Engineering part of the
Bechtel Group of Companies, I would like to explain our views of developmental
aspects of Plus Gas Desulfurization as well as its production engineering
aspects.
To start off (SLIDE l) I would like to remind you of the very broad
definition of Engineering, as given in standard dictionaries.
Applying this definition to the subject of Flue Gas Desulfurization
(SLIDE 2) we can easily come up with a long list of more detailed functions,
all or most of which involve Engineering, in the professional sense. In fact,
we find people with these talents and experience throughout the various
portions of the industry, from the regulatory agencies to process development
activities, to equipment suppliers, systems engineers, constructors and plant
operators.
The issue I would like to address today, is where these functions
should best be performed. I hope to be able to show you that this question is
6-44,
-------
SLIDE 1
ENGINEERING!
THE APPLICATION OF SCIENCE AND MATHEMATICS EY WHICH THE PROPERTIES OF MATTER AND THE
SOURCES OF ENERGY IN NATURE ARE MADE USEFUL TO MAN IN STRUCTURES, MACHINES, PRODUCTS,
SYSTEMS, AND PROCESSES.
A. MERRIAM-WEESTER
REGULATORY
PROCESS DEVELOPMENT
SYSTEMS ENGINEERING
SLIDE 2
PROFESSIONAL ENGINEERING SERVICES REQUIRED
- ESTABLISH NEEDS, REGULATIONS, MONITOR RESULTS
- CONCEPTION, DEVELOPMENTS, SCALE-UP
- SYSTEMS CONCEPT (FGD VERSUS COAL SELECTION, CLEANING)
- FGD PROCESS FLOW SHEET
- EQUIPMENT SELECTION AND PROCUREMENT
- AUXILIARIES
- DETAILED DESIGN FOR CONSTRUCTION
EQUIPMENT MANUFACTURING
CONSTRUCTION
OPERATION
- DESIGN, FABRICATE, TEST
- METHODS
- CRAFT SUPERVISION
- INSPECTION, QUALITY CONTROL
- START-UP, SYSTEMS MANUALS, TRAINING
- TECHNICAL SUPERVISION
- REPAIR, MAINTENANCE
6-45
-------
answered differently for differing industries, differing kinds of technology,
and for different stages of technology development, and that the technical
nature of Plue Gas Desulfurization is sufficiently different from the rest of
the technology used by the electric utility industry so as to suggest valid
differences in how the related engineering should be conducted.
In presenting this thinkings I should also state that there is no
single aethod which is best under all conditions. Tfe do projects under many
different circumstances, and our project people are given a substantial degree
of freedom to structure their projects as they see fit for their specific
circumstances. If there is any hard Bechtel policy on the Batter, it is only
that this flexibility shall be aaintainedo
First, I would like to suggest soae principles (SLIDE 3) that are
generally true, regardless of the type of industry involvedo Fundamental to
these is th© aotioa that the ultimate plant otmer/operator must maintain
controlo To ©xercise this control he should staff with his own engineering
support capabilities to th© extent that he can do so without creating probleas
for hiaself, that is, either creating conflicts, or what I trill call
Inefficiencies..
A typical conflict is that the regulatory engineers aust not be on the
owner's staff, for there would be a clear potential for conflict of interest,,
However, the owner should have people available who can work closely with thea.
There are also soae conflicts with which the electric utility industry
is not generally familiar. These are due to the tightly held proprietary
positions that are maintained in soae highly competitive industries, such as
the pharmaceutical industry. As a result, almost all engineering is done
in-house and only non-proprietary related activities are conducted by others.
Beyond these conflicts, the main inefficiency seen by aost owners has
to do with the cyclic demands for engineering support - which are dependent
upon the nature of their capital plant additions. In the pharmaceutical
industry, aost expansions are small, or are aodifications to existing systems,
and are done by in-house engineers on a relatively steady state of employment
level. In the refinery and chemicals business there are similarities, and
strong owner engineering staffs, with outside assistance requested only for
their larger projects. In the electric power business, additions are usually
made in larger single steps, tending to present an increased cyclic staffing
level problem, and increased incentives to obtain outside support to avoid
major in-house staffing level cycles. I believe these principles are well
understood.
Another aajor factor in efficiently utilizing a particular engineering
talent is to consider the necessary inputs and feedback that an engineer needs
in order to do his job. Clearlyj, an engineer who is designing an iapeller for
a pump needs all kinds of interactions with laboratory tests on various
fluids, as well as with those who make castings and oachine aetal parts. It
would be inefficient to locate such an engineer at an electric utility site.
6-46
-------
Thus, in the evolution of our businesses ve have properly consigned this
general class of engineering activity to the equipment manufacturers site.
The same logic can be used for other classes of engineers, but vith
perhaps differing results. As a case in point, the systems engineer, the one
who puts several component parts into an operating process or system, should
be located at, and have his feedback derived from the site that has the
greatest involvement in the development and use of the system. For
pharmaceutical plants these engineers are clearly located in-house vithin the
operating company, due both to secrecy and efficiency aspects. In refinery
and chemical plants, there tends to be a mix, vith some proprietary systems
packages produced by organizations such as UOP, and vith other
state-of-the-art systems being developed by either in-house engineers - or by
outside engineering assistance if vorkload fluctuations are important. In
these cases, there are often many interties betveen the various systems
requiring that the plant ovners engineers design the overall system and then
procure separate chemical unit operation components of knovn performance value
and assemble them into a system using veil proven and previously demonstrated
engineering principles.
In electric utility plant design, the systems engineer of the past also
did the same thing. Eovever, the number of components vas smaller, and each
tended to be centrally manufactured elsewhere. Thus a system of major
equipment might have included separate components for coal pretreatment,
boilers, turbines, generators, and power transmission systems, and the ovners
systems engineer vas more involved in their selection and erection than he vas
in their internal technology development.
SLIDE 3
• OWNER CONTROL
• EXCEPTIONS
1. CONFLICTS - REGULATORY
- TIGHTLY HELD SECRETS
2. INEFFICIENCIES - CYCLIC STAFFING NEEDS
- EXPERIENCE FEEDBACK
- SYSTEMS INTERTIES
6-47
-------
Vhen a new technology is being first developed, it is even more
important that the engineer be located at the site where the development of
the technology takes place (SLIDE 4). In the case of Flue Gas Desulfurization
this development took place in two diverse ways. First, the EPA sponsored
development work at the Shawnee Test Facility. Secondly, numerous private
organisations saw an opportunity for new business and proceeded with their own
separate development programs. Since the EPA program was paid for by the
taxpayers, that technology has became open-art, while the privately developed
technology became proprietary to its developers. However, for lioe/limestone
processes, with some exceptions, Shawnee really showed the basic chemistry
while the other developments have been more oriented to the further
development of individual components of equipment to better handle a
particular unit operation, such as nozzle spray patterns and gas distribution,
vessel designs and mist elimination.
How that the technology of Flue Gas Desulfurisation has generally
matured, we believe that process, or chemistry related, guarantees are
becoming of lesser importance. This, of course, does not diminish the
importance of assuring the quality of workmanship and viability of individual
components which must continue in order to deliver the expected process
performance over the life of the plant. However, since Flue Gas
Desulfuri&ation is really a series of chemical unit operations, including
potential interties into other power plant sub-systems involving heat and
water utilization and waste solids rejection, we believe that the system tends
to start looking more like a refinery and that the owner's engineer - or his
contract engineer - should be in firm charge of designing the systems
involved. Inherently, as it has elsewhere, this will also tend to place less
emphasis on the use of externally supplied systems packages.
The developmental operation of Bechtel, which I represent, recognized
the probable evolution of these circumstances a number of years ago. As a
result, we sought out, and were pleased to have the opportunity, (SLIDE 5) to
play a key role in the basic development of lime/limestone Flue Gas
Desulfurization technology by designing, and then performing as the test
director for the EPA Shawnee Alkali Flue Gas Desulfurization Test Facility.
This gave us an unusual early insight into the basic chemistry as well as many
of the requirements for the equipment required to handle these chemical
engineering unit operations.
Ve were also fortunate to have the opportunity to work on Colstrip
Units 142, and were able to improve the system's concept by removing both
sulfur and particulate removal in a single set of venturi scrubbers, and in
doing so we were able to recover and utilize the alkalinity that inherently
existed in the fly ash - thus avoiding the use of extensive additional alkali
feed material.
¥e were also given another unusual opportunity on Colstrip 3 ft 4, which
needed much higher S(>2 removal than Units 1 ft 2, to develop a method for the
use of a more reactive alkali, Type S Lime. This systems approach, aided by
seal water pump packings that require much less water, have now allowed us to
6-48
-------
SLIDE i*
FLUE GAS DESULFUR1ZATION TECHNOLOGY DEVELOPMENT
• EPA
CHEMISTRY/PROCESS
• PRIVATE
PROPRIETARY COMPONENTS
• MATURE
» GUARANTEES
PROCESS
COMPONENTS
• CHEMICAL UNIT OPERATIONS
SYSTEMS INTERT1ES
• OWNER SYSTEMS ENGINEERING
SLIDE 5
BECHTEL ROLES IN FLUE GAS DESULFURIZATION DEVELOPMENT
• SHAWNEE
- DESIGN
- TEST DIRECTION
• COLSTRIP 1 & 2
- NO ESP
- ALKALINE ASH UTILIZATION
• COLSTRIP 3 & t
- REACTIVE TYPE S LIME
- WASTE WATER SINK
0 SPRAY DRY
- REACTIVE TYPE S LIME
6-49
-------
Bake the Flue Gas Desulfurisation system a waste water sink instead of a
source, vith a variety of other plant waste streams being evaporated in the
scrubber. Colstrip Units 3 A 4 will both employ these systems, with Colstrip
Unit 3 starting up this month.
Ve have also used these systeas insights as a basis to further develop,
and pilot test, technology to allow spray dry Flue Gas Desulfurization systems
to use such more reactive reagents, vith substantial reductions in the
resulting equipment size, elimination of recycle requirements, and use of
spray dry equipment on high sulfur coals as veil as on lov sulfur coals while
obtaining high SC>2 removal.
Lastly, ve are also fortunate to be substantially involved in nev
process developments for precombustion sulfur removal, including advanced coal
cleaning technology and the use of coal water mixtures.
It is equally important to note that, at least for our utility
applications, ve viev all of the above systems engineering as non-proprietary
open-art technology. ¥e do not charge license fees to our utility clients,
nor do ve limit our consideration of applicable process systems to that vhich
ve may have developed, or assisted in its development. Bather, ve believe it
is the engineer's job to continue to remain unfettered by commercial interests
and to be free to select, and design, the combination of processes that best
suits any particular owner for that particular owners specific site related
needs and opportunities.
In summary, (SLIDE 6) ve believe that the technology has nov reasonably
matured to a level where the utility needs to place greater emphasis upon
component quality rather than process performance guarantees, that much of the
technology is based upon open-art chemical engineering unit operations, and
that, because of the numerous interties that are necessary vith the rest of
the utilities system, the utility owners should start to place more emphasis
on control over the engineering - either through additions to their own staff
or by increased use of contract engineering services.
SLIDE 5
SUMMARY
TECHNOLOGY HAS MATURED
PERFORMANCE GUARANTEES
- COMPONENTS VERSUS PROCESS
CHEMICAL UNIT OPERATIONS
- OPEN-ART
- SYSTEMS INTERTIES
OWNER CONTROL OF SYSTEMS ENGINEERING
6-50
-------
THE ARCHITECT-ENGINEER — MIDDLEMAN BETWEEN
UTILITY AND FGD SUPPLIER
Opening remarks by: P. R. Predick
Acting Head
Mechanical Analytical Division
Sargent & Lundy
Chicago, Illinois 60603
ABSTRACT
Architect-Engineers (AEs) should serve as an extension of a utility's
engineering staff to procure an economical FGD system which provides
compliance with the applicable emission limits. To this end, AEs prepare
design criteria, write specifications, evaluate bids, administer contracts
and coordinate equipment interfaces. Sargent & Lundy's recommendations for
the procurement of a successful FGD system are summarized, and suggestions
are provided to illustrate these recommendations.
OPERATING EXPERIENCE FEEDBACK
by: Michael L. Meadows, P.E.
Black & Veatch Engineers-Architects
P. 0. Box 8405
Kansas City, Missouri 64114
ABSTRACT
One of the important activities in which the architect/engineer (A/E)
serves as an intermediary between the utility and the flue gas desulfur-
ization (FGD) system supplier is preparation of equipment purchase specifi-
cations. One of the A/E's many goals in preparing this procurement document
is to ensure that the bid specification requirements reflect current knowledge
gained from operation of existing FGD systems. Accurate and detailed inform-
ation about operating experiences at utility FGD systems is difficult and time
consuming to collect, but it is one of the best methods for improving FGD
system reliability and maintainability. As part of an effort to improve the
availability of operating FGD system experience information, Black & Veatch
has participated in several recent projects sponsored by the Electric Power
Research Institute to assist the utility industry. These projects have
included:
• Lime FGD Systems Data Book (Second Edition, CS-2781).
• Limestone FGD Systems Data Book (CS-2949).
• Operation and Design of Dampers in FGD Systems (RP-2250-1).
• FGD Systems Spare Parts. Program Practices (RP-2248-4).
6-51
-------
In addition, Black & Veatch is participating with Los Angeles Department
of Water and Power and the FGD equipment supplier in an Availability Improve-
ment Program for the new 1,500 MW Intermountain Power Project. This task
force has made recommendations on many facets of the FGD system now under
design which are intended to improve the reliability of the system. A
significant portion of this effort has included visits to several operating
FGD systems to evaluate and discuss the strengths and weaknesses of their
systems.
THE ARCHITECT/ENGINEER - MIDDLEMEN BETWEEN UTILITY AND
FGD SUPPLIER
Douglas B. Hammontree. P.E.
Burns & McDonnell
ABSTRACT
In administering the FGD system contract, as with any contract throughout
the plant, th? A/E must maintain a professional relationship with the vendor,
remaining impartial and committed to their interests. On the other hand,
keep in mind that we are hired and paid by the utility. At Burns & McDonnell
we act as an extension of the Owner's staff and we are totally dedicated to
the Owner's interests and needs.
I mention this to clarify our role as middleman and to shed some light on
the discussion of the problems encountered in this role. Hopefully, with a
good specification, the intermediary role is minimized. No problems should
develop unless deviations to the specifications occur.
But, of course, problems do occur. Nozzles plug, valves don't seal,
materials corrode, and mist eliminators do not. While these are problems
that must be faced, the biggest problem as intermediary is to integrate the
FGD system with the power plant itself.
With continuing developments in the FGD industry, current designs have
reached the stage where the process is adequate to meet removal efficiencies.
Equipment and instrumentation is, for the most part, workable. Problems
with materials construction are slowly being resolved. In general, reli-
ability problems of current FGD systems can be traced to lack of operator
understanding.
It is this situation in which the A/E as middleman, must provide guidance,
both to the utility and the vendor. Typically the utility h,as extensive
experience with the boiler/turbine operation. The FGD system presents them
with a chemical process plant "located out back by the stack." The A/E must
ensure that the utility is prepared to operate this system. We must let
them know what to expect. Maintenance and operation requirements are more
extensive than in the plant, and the utility must be made aware so that
appropriate staffing needs are met.
6-52
-------
Operators must be trained in the chemical nature of the system. They quickly
find that mechanical solutions are not always adequate for the problems that
arise. As with the boiler, where proper control id required to prevent
slagging and poor heat rates, there are certain chemical parameters in the
FGD system operation that must be monitored to keep the system operating
economically and efficiently. Appropriate training programs are vital to
meet this need.
With regard to the FGD system supplier, the A/E's goal as middleman is to
ensure that the equipment supplied is compatible with the rest of the power-
house. Mainteanance and operation personnel will be burdened enough without
having to learn about all new equipment can and should be by the same manu-
facturer throughout. Valve operators, solenoids, limit switches, etc., can
be the same. Rather than strict demarktion between powerhouse and FGD
system, computer control system should be identical. Logic diagram, flow
diagrams and process instrument diagram should be consistent in meaning, with
common tag and connection numbering.
By integrating the FGD system in this manner, training and operator under-
standing will be one step ahead. Process problems will more easily under-
stood and resolved.
The FGD industry is young with slightly more than 10 years' experience in
broad application. There will continue to be problems at any intallation.
It is the A/E's duty to ensure that the utility and the FGD supplies are
working together to provide for the proper training ind integration required
for an efficient and reliable system.
THE ROLE OF THE ENGINEER/CONSTRUCTOR IN FGD
by: Christopher P. Wedig
Stone & Webster Engineering Corp.
Boston, MA 02107
ABSTRACT
The role of the Engineer/Constructor (E/C) in FGD depends on the
contract between the utility and E/C. Usually, the E/C will provide to the
utility licensing service, studies, specifications, comparison of bids,
purchasing services, review and supervision of FGD supplier eneineerine anH
construction. In addition, the E/C provides design and engineering of the
electrical, controls, and structural areas, construction, start-up, and
maintenance training.
The following represents some of the many important concepts in perform-
ing the E/C work for the utility and being the intermediary between the
utility and FGD system supplier.
6-53
-------
1. Licensing of the coal fired power plant site should be based on
proven FGD technologies.
2. FGD and FGD waste studies should be performed in an unbiased manner
and based on E/C and utility "inhouse" information plus written
correspondence and meetings with FGD and FGD waste system suppliers.
3. The FGD and FGD waste specifications should be a combination
performance and design specification. The specification should
be a "tool" with which the utility and E/C purchase a highly
reliable system. Design parameters should be specified for
those areas in which problems have occured in operating systems.
The E/C should make frequent visits to operating FGD systems
and to manufacturers of components to ensure that he has know-
ledge of problem areas. Performance acceptance tests, guarantees,
and warranties should be emphasized. The specification should be
reviewed by the utility engineering, construction, and power
production departments.
4. The FGD supplier bidders list should be short and contain only
suppliers of proven technology and demonstrated highly reliable
operating systems. All FGD suppliers should be treated in an
unbiased manner. The bidders should bid to the base specification
and offer exceptions or options only when it will improve or not
effect system reliability. The award of contract to the FGD or
FGD waste system supplier should be based on a present worth
total evaluated cost basis which includes all costs of the system
supplier, utility, and E/C.
5. The E/C should carry out the utility's directions, while at the
same time providing reliable advice. The E/C should ensure that
the FGD system equipment is designed similar to the balance of the
power plant so that maintenance procedures are compatible and spare
parts are minimized. The E/C and utility should review and monitor
the engineering and construction work of the FGD supplier. The E/C
and FGD supplier should work as a team to ensure that interface
problems are minimized. The FGD supplier should engineer and con-
struct in accordance with the system purchase specification and
should inform his subsuppliers and subcontractors of the require-
ments of the purchase specification.
3. The utility, E/C, and FGD & FGD waste system suppliers should work
as a team in the start-up and maintenance training of the systems.
The utility should train the O&M personnel before the systems
become operational and participate in the start-up of the systems.
The E/C and FGD supplier should provide start-up services and
numerous training classes for the utility O&M personnel. The E/C
should ensure that the performance acceptance tests and supplier
guarantee/warranty requirements are met in accordance with the
6-54
-------
purchase specification. The FGD supplier should not be permitted
to leave the site until he has met all of the requirements of the
purchase specification. The E/C and FGD & FGD waste system
suppliers should be permitted to work with the utility during the
entire life of the systems to provide, on an as needed basis,
engineering and training services to ensure a highly reliable FGD
and FGD waste system.
"THE ARCHITECT/ENGINEER - MIDDLEMAN BETWEEN UTILITY AND FGD SUPPLIER"
Richard Rao
EBASCO Services, Inc.
Lyndhurst, New Jersey
ABSTRACT
As the "middleman," the architect-engineer has a responsibility, a role
and a salient position.
The responsibility of the "A/E" with respect to flue gas desulfurization
(FGD) is to define the term "good engineering practices" and apply its
principles throughout a project to obtain two fundamental objectives -
operational reliability and cost-effectiveness. The responsibility of the
A/E with respect to FGD is similar to the A/E's responsibility for other
technologies that comprise a steam-electric generating plant. The task of
this responsibility with respect to flue gas desulfurization, in comparison
to other technologies, is more difficult because of the relatively develop-
mental status of FGD technology and the fact that the utility, architect-
engineer and the FGD supplier can not unilaterally design this part of the
plant. Regulatory agencies and the general public scrutinize and approve
the adequacy of designs.
The role "of the A/E is to direct the process by which: (1) a philosophy
of design is developed by the utility, (2) the technical issues are identified
complete with the proper economic assessment, (3) necessary decisions are made
by the utility on a timely basis to meet project objectives, (4) competitive
bids are secured that address the technical issues and allow the inclusion
of the bidders expertise and ingenuity, (5) evaluations' are conducted in
in accord with the aforementioned responsibility, (6) operational results
are evaluated to measure achievement.
The position of the A/E enables him to input objective and credible
perspectives into the project. The A/E is not subject to the competitive
pressures of FGD suppliers. The position of the A/E enables him to perform
"power plant engineering" allowing the FGD supplier to concentrate upon the
core of the FGD system (the essentials to achieve performance). The A/E's
position is a perch from which credible perspectives of the impact of FGD
upon the overall plant design, cost and operation can be provided to
regulatory agencies and the general public.
6-55
-------
Concerning responsibility, the definition of "good engineering practices"
enconpasses identifying commercially available technologies commensurate
with the planned operation of the particular steam-electric generating unit,
establishing the proper degree of redundancy, space requirements and
enclosures to allow performance of necessary maintenance, and integrating
the FGD operation into the overall power plant operation.
Concerning the role of the A/E, as the director of the engineering
process he is the "chief designer." With respect to technical issues, the
primary issues are "dry " versus "wet" scrubbing, selection of suitable
materials of construction and determining the proper degree of redundancy.
From the developed design philosophy which begins with applying the economic
factors to design concepts, technical issues related to the degree of design
conservatism and cost can be settled. To the extent that such issues are
settled prior to the specification, the burden of bidding by the FGD
suppliers is reduced.
Corresponding to the increasing maturity of flue gas desulfurization
technology, the position of the A/E has become more significant. Specifica-
tions have become more detailed as the impact upon the overall plant
operations became more perceptible. The position of the A/E can increase
further in stature, if he assume*more responsibility for the "balance-of-
plant" aspects of flue gas desulfurization."
6-56
-------
SESSION 7: FGD CHEMISTRY
Chairwoman: Dorothy A. Stewart
Electric Power Research Institute
Palo Alto, CA
-------
INFLUENCE OF CHLORIDES ON THE PERFORMANCE
OF FLUE GAS DESULFURIZATION
W. Downs, D. W. Johnson,
R. W. Aldred, L. V. Tonty,
R. F. Robards, R. A. Runyan
-------
INFLUENCE OF CHLORIDES ON THE PERFORMANCE
OF FLUE GAS DESULFURIZATION
By:
William Downs
Babcock & Wilcox
Research and Development Division
Alliance Research Center
Alliance, Ohio 44601
Dennis W. Johnson
Babcock & Wilcox
Advanced Energy and
Environmental Systems Division
Barberton, Ohio 44203
Robert W. Aldred, L. Victoria Tonty,
Russell F. Robards, Richard A. Runyan
Tennessee Valley Authority
Energy Demonstrations and Technology
Chattanooga, Tennessee 37401
ABSTRACT
A pilot plant test program was performed to determine what effect high
concentrations of chloride might have on the performance of limestone-based
flue gas desulfurization (FGD) processes. This test program took place during
January and February 1983 at the Tennessee Valley Authority (TVA) Shawnee
Steam Plant. The specter of high chloride concentration arises from a trend
toward closed-loop operation. The influence of chlorides on S0_ absorption
was examined; 161 limestone tests were completed. The principal conclusion
that can be drawn from this test program is that chlorides generally inhibit
SO absorption. The severity of this effect can vary widely, however, depend-
ing upon the design and method of operation.
INTRODUCTION
One of the most serious concerns of current and future users of lime and
limestone-based (L/L) FGD systems involves the influence with which chlorides
(and to a lesse'r extent, fluorides) will affect the performance of these pro-
cesses. In response to these concerns, Babcock & Wilcox (B&W) approached the
Tennessee Valley Authority (TVA) in May 1982 to solicit interest in a joint
pilot plant project at TVA's Shawnee pilot plant facilities. TVA's interest
concerned limestone flue gas desulfurization (FGD) systems, whose closed-loop
water operations could be subjected to high chloride concentrations.
Prepared for presentation at Environmental Protection Agency/Electric Power
Research Institute Symposium on Flue Gas Desulfurization, New Orleans, La.,
November 1-4, 1983.
7-1
-------
Subsequently, the Electric Power Research Institute (EPRI) joined the project
and participated in developing a pilot plant test program to study the
influence of high chloride concentrations on SO absorption. Two objectives
of this test program included:
• To determine to what extent chloride would influence SO absorp-
tion if chlorides were permitted to build up to concentrations as
high as 60,000 ppm in the recirculation loop.
• To investigate whether certain system design parameters were
capable of reducing harmful effects of chloride.
An implicit objective of this test program was to evaluate to what extent
heterogeneous (liquid/solid) reactions take place within the absorber and to
what extent they might alleviate reductions in SO absorption that could
result when chlorides are added to calcium-based slurries.
PILOT PLANT FACILITIES
TVA's pilot plant facilities at the Shawnee Steam Plant include most of
the elements common to full-scale, limestone, FGD systems. The main facil-
ities utilized for this test program included a 30,000-ACFM pilot plant,
designated by TVA as Train 200, and a dry limestone preparation facility for
grinding limestone to desired fineness.
Some modifications to Train 200 were made for this test program. These
included installing a venturi scrubber placed ahead of the pilot absorber, as
well as modifications to the absorber itself. The venturi scrubber was in-
stalled to remove fly ash from the flue gas via a water spray. The absorber
is a square tower with a flow area of about 31 ft that was originally de-
signed to simulate a Turbulent Contact Absorber. This absorber was modified
for these tests to simulate B&W's L/L absorber with B&W sieve trays.
In addition to these changes, an SO vaporizer/injection system was
installed to permit variability in S0_ concentration in the flue gas.
BACKGROUND
Currently, most operating L/L FGD systems experience chloride levels in
the range of 2000 to 8000 ppm. In this range, few process-related problems
have been experienced. However, some architect/engineers anticipate future
FGD systems to far exceed these levels (30,000 to 120,000 ppm, for example).
How this could develop is best shown by considering a simplified power plant
flowsheet (Figure 1). Chlorides enter these systems witli coal and fresh water
makeup. Chlorides have the property of being easily and completely dissolved
in water. The only escape for chlorides from these water cycles depicted in
Figure 1 is via cooling tower blowdown and with the free moisture that leaves
with the scrubber sludge. Typically, for every 1 pound of coal that is burned
in the boiler, about 4 pounds of water are evaporated in the cooling tower,
and 1/2 pound of water is evaporated in the scrubber. Eastern bituminous
coals average about 800 ppm chlorides. Therefore, if the water contains about
100 ppm chlorides in this generalized example, then each pound of coal will
result in the addition of about 1.25 x 10 pound of chlorides.
7-2
-------
WATER CYCLES
1 IB
FRESH
WATER
Cl = 100 PPM
0.6 LB
FILTER
SLUDGE
FRESH
4 LB WATER
Cl = 100 PPM COOLING TOWER
SLOWDOWN
Figure 1. Simplified power plant back-end flowsheet.
Since blowdown from the cooling tower can be ecologically undesirable,
and since every pound of blowdown water requires an additional pound of
precious fresh water makeup, it is in the utilitys' interest to minimize that
blowdown. This fact provides motivation to use blowdown water as the princi-
pal source of water in the FGD system. This reduces the demand of the FGD
system for fresh water and reduces the amount of cooling tower blowdown.
Take this-example to its extreme. If all cooling tower blowdown is
directed to the FGD system, then the steady-state chloride concentration can
be shown [1] to depend only on the degree of sludge dewatering (Figure 2).
The fact that industry trends point toward both water conservation and
improved sludge dewatering brings into foeus the concern over chlorides.
A major concern regarding chlorides in the electric utility industry is
the implication that catastrophic corrosion could occur even with the use of
expensive alloys. As important as this concern is, however, the mission of
this project was not to explore the implications of chloride buildup upon
materials of construction. The literature is crowded with corrosion studies
as applied to FGD processes and is not repeated here.
Although the concern for chloride buildup is relatively recent as it
applies to FGD processes in the electric utility industry, chloride buildup in
various pulp and paper processes has been a problem of concern for many years.
Two methods used to remove chlorides in the pulp and paper industry are
discussed by Karjalainen, et al. [2] and Reeve, et al. [3].
7-3
-------
70.000
60,000
50.000
a.
a.
U
O
ui 40.000
a
E
3
O
til
« 30,000
20,000
10.000
ASSUMPTIONS
Cl IN COAL = 800 PPM
Cl IN WATER = 100 PPM
SO2 EFF = NSPS
1% SULFUR
IN COAL
FILTRATION
THICKENING
ONLY
20
30
40
50 60
PERCENT SOLIDS IN SLUDGE
70
80
90
Figure 2. Chloride levels achievable in closed-cycle operation.
Studies dealing with the implication of chloride buildup in FGD systems
are relatively recent. Much of the current interest in chlorides arises from
the fact that results reported in the past few years (1971 - 1981) have been
varied, inconsistent, and confusing. Bechtel [4] and Rader [1] reported that
SCL absorption improved slightly with increasing chloride concentrations up to
about 20,000 ppm. Others, notably Chang and Laslo [5], reported a very sub-
stantial decrease in SO absorption at chloride concentration ranges up to
180,000 ppm. To complete the trilogy, Rader [6] reported on an expanded pilot
plant test program at higher chloride concentrations (up to 80,000 ppm) and
showed that chlorides had no significant effect on SO absorption.
The fact that all these observations were made at various times on dif-
ferent pilot plants is testimony to the complex interaction and trade-offs
that contribute to the performance of lime/limestone scrubbers.
7-4
-------
The most obvious influence of chlorides in the FGD system is to decrease
the solubility of calcium salts because of the "common ion effect." This
situation results when HC1 produced in the furnace from the chloride in the
coal is collected in the absorber by, for example, the following simplified
reaction:
2 HCl(g) + CaC03(s) - - CaCl2(aq) + V + C02(g) (Rl)
Thus, dissolved calcium ion (Ca ), builds up in proportion to the chloride
concentration that, in turn, limits the solubility of basic species via:
KS l
(CaH
K
sp
(CaH
K
h+)
2
">
3
The result of this common ion effect is to reduce the dissolved alkalinity of
the scrubbing slurry.
Not only does the equilibrium solubility of limestone (and lime) decrease
with increasing chloride concentration, but the rate of limestone dissolution
decreases as well. Borgwardt [7] and Rochelle [8] have both shown that the
dissolution rate of limestone decreases sharply with increasing chloride con-
centration when titrated with hydrochloric acid. Borgwardt performed his
experiment at a pH of 6.0 and at chloride concentrations up to 10,000 ppm;
Rochelle did his at pH 5.0 up to 70,000 ppm. Both experimenters performed
tests with NaCl and MgCl (as well as CaCl ) and observed that these salts
were nearly as effective as CaCl in reducing the dissolution rate. Since
sodium and magnesium chlorides do not reduce the solubility of limestone by
the common ion effect, their influence on the dissolution rate must result
from some other effect. Rochelle has modeled the dissolution process and
concludes that the suppressed dissolution rate is the result of reduced
hydrogen ion diffusivity to the reaction surface of the limestone. Thus,
increasing chloride concentration should also reduce SO- absorption through
this negative limestone dissolution effect.
What influences exist to counter these harmful consequences of chlorides?
First, two instances where improvement was observed were actually artifacts of
the experiment. In the case of both Bechtel [3] and Rader [1], their pilot
plant experiments were performed at constant pH. As chloride concentrations
increased, the equilibrium pH decreased, resulting in the need to increase the
limestone feed rate to maintain the pH set point. Therefore, the increase in
SO- absorption was the result of increasing stoichiometry — not increasing
chlorides. Rader [6] reported that fact after his expanded test program was
run at constant stoichiometry.
7-5
-------
In a conventional L/L system, the steady-state pH of the recirculated
slurry operates at a set point somewhat above the chemical equilibrium pH of
the slurry. As a general rule, the difference between these two pH values
will decrease as the limestone stoichiometry increases. When chlorides are
added to the slurry, the equilibrium pH drops. Therefore, if the stoichiom-
etry is to be maintained constant as chlorides increase, then the set-point pH
must also be decreased. Rader [6] has hypothesized that the limestone disso-
lution rate may increase in this more acid environment and, in so doing, help
to moderate these negative influences of chlorides. The extent to which this
actually exists has not been experimentally examined in any controlled exper-
iments but is implied to be a factor in Rader's pilot plant results.
Jarvis [9] has presented perhaps the most compelling explanation to
indicate why experimenters have at times seen slight improvement or at least
little degradation in SCL absorption at elevated chloride levels. Specific-
ally, he has found that the total sulfite solubility actually increases with
increasing calcium chloride concentration up to a point and reaches a maximum
around 30,000 ppm Cl . This means that SO absorption can increase with
increasing chlorides even if the dissolved alkalinity in the absorber reduces
slightly. In effect, the make-per-pass capacity of the slurry increases.
Computer models of this complex chemical equilibrium among SO , ., CO , .,
CaCl . ., and HO,,,, can be shown to support this experimentai^obserVafion.
A B&W-modified form of the Radian chemical equilibrium model suggests that the
enhanced sulfite solubility is principally the result of ion pairing between
Ca and HSO- as CaHSO., .
THEORY
SO absorption into the absorber slurry is enhanced by chemical reaction
with dissolved species. The dissolved species that are thought to participate
include:
S°2(g) == S02(aq) <»«
S00/ . + CaSO.0 + HO . ' HSO ~ + HSO~ + Ca** (R3)
2(aq; 4 2 3 4
S°2(aq) + S04 + H2° :=rHSOo + Hso/. (R4>
SO . + CaSO ° + H00 ^==:2HSO,~ + Ca"^ (R5)
2^aq^ J
S00/ . + CaHCO/1" . ' HSO + Ca + C00 (R6)
2(aq) j
SO + HCO ~ :==:HSO ~ + C00 (R7)
2(.aq; j
SO., , + SO * + HO 3=-:2HSO ~ (R8)
2(.aq; J 2 J
Slurry entering the absorber at the spray nozzles will tend to approach
equilibrium among the following species: C0?, .., CaCO-, ,., CaSO--1/2 H-O, -.,
CaSO -2H20, ., and H 0/pH = set point. The equilibrium8composition can bl;
computed for example, by B&W's modification of the Radian's chemical equilib-
rium model. However, one uncertainty is what_yalue to assign to the C0~
partial pressure. Values ranging from 3 x 10 atm to 1.0 atm are possible.
7-6
-------
Increasing CO partial pressure increases the bicarbonate concentration. This
benefits SO absorption via Reaction 7. Most investigators assume a value of
0.1 to 1.0 atm. The sum of the molar concentration of these reacting species
is often called the "dissolved alkalinity."
A question of fundamental importance to the design of limestone absorp-
tion systems is whether within the S09 absorber the reaction system is limited
to homogeneous liquid-phase reactions, such as Reactions 3 through 8. The
question reduces to whether sufficient time exists for calcium carbonate
and/or calcium sulfite to dissolve within the absorber via:
CaC03(s) , z • Ca + C03 (R9)
HO
CaS03-l/2 H20(g) . 2 ' Ca4"1" + S03= + 1/2 H20 (RIO)
TVA [10] has reasoned that Reaction 9 can be a significant contributor to the
alkalinity of a limestone system but that Reaction 10 is probably unimportant,
except in lime FGD systems.
Alternatively, direct reaction between H and limestone has been
postulated at the limestone surface to proceed via:
H+ + CaC00, .. ' Ca"1^ + HCO ~ (Rll)
3(s) 3
While the homogeneous reactions can be assumed to be instantaneous, the
heterogeneous reactions (R9 through Rll) are finite. Further, the rates of
these reactions will depend on the surface area of the solid species. Thus,
it follows that participation by Reaction 9, for example, will be benefitted
by fine grinding of limestone and by maximizing residence time of slurry
within the absorber.
Therefore, one of the major questions addressed in this pilot plant test
program was to determine whether harmful effects of chlorides might be
alleviated by designing the FGD system to promote the heterogeneous reactions
(R9 through Rll) in the absorber.
Finally, we should consider the SO /limestone reaction system for forced
oxidation tests. For these tests where oxidation is nearly complete, Reac-
tions 5, 8, and 10 cannot participate. Further, the tendency to blind
limestone via the following mechanism is heightened since the system is
deficient of sulfite crystals to receive precipitating sulfite.
SO.., , + HO - -HSO ~ + H+ - -SO " + 2H+ (R12)
2(aq; 2 J J
HO
+ Ca'r+ (CaC00) - ( (CaC00 ) 1 (R13)
7-7
-------
Countering these effects is the fact that the equilibrium and interfacial
SO vapor pressure are substantially reduced under forced-oxidation operation.
This should enhance SO absorption. On balance, it seems reasonable to expect
that S0? performance under conditions of forced oxidation may be more sensi-
tive than the natural oxidation tests to the limestone grind size.
TEST PROGRAM PLAN
A factorialized test plan was devised for this program. The variables
selected and the levels to be tested were:
VARIABLES LEVELS
Mode of Oxidation Natural Forced
Limestone Grind: Fine 95% through 325 mesh 95% through 325 mesh
Coarse 80% through 200 mesh 80% through 200 mesh
Inlet SO Concentration 2000 ppm, 4000 ppm 2000 ppm, 4000 ppm
L/G 45, 90 gal/MCF 45, 90 gal/MCF •
Recirculation Tank Residence Time 5.6, 11.2 min 5.6, 11.2 min
Liquid Holdup 0, 3, and 6 inches 0
Chloride Concentration 0, 15,000, 30,000, 15,000, 30,000,
45,000, 60,000 45,000, 60,000
The specification for limestone grind was intended to evaluate the
extremes of normal commercial practice as were the specifications for L/G and
tank residence. The specification for SO concentration at the inlet was
intended to examine the usual range for eastern bituminous coals. Actually,
the specification of 2000 ppm was to be obtained by burning eastern Kentucky
coal in Unit 10 at the Shawnee Steam Plant. In fact, during the test program,
actual SO level varied from about 2200 to 2600 ppm because of variations in
factors like sulfur content of the coal, boiler load, excess air, etc.
The three levels of "liquid holdup" expressed as 0, 3, and 6 inches
represent the approximate hydraulic head in inches of water of retained slurry
in the absorber. Actually, the physical variable was the number of sieve
trays in the absorber. Zero holdup represents the spray tower configuration
with no tray. The 3-inch holdup represents the use of one sieve tray, and the
6-inch holdup represents the use of two sieve trays.
The five levels of chloride are spread uniformly over the full range that
we anticipate might be experienced even under extreme conditions. The zero
level really represents the open-loop concentration, which could amount to
several hundred ppm.
In addition to the variables selected for study, careful consideration
was given to the absorber parameter settings that were to be held invariant.
Tower velocity of 10 ft/sec and 15% suspended solids were selected as normal
absorber settings. Spray nozzle pressure of 4 psi is somewhat low by FGD
standards (10 - 15 psi is more usual). The nozzles used in this program were
selected mainly to reduce the wall effect within the absorber.
7-8
-------
The most difficult operating parameter to select was for limestone
stoichiometry*. A limestone stoichiometry set point of 1.10 (+0.01) was
established for most tests. There are distinct difficulties in attempting to
operate a pilot plant at such a low stoichiometry. These include:
• SO performance may be very sensitive to small variations in
stoichiometric ratio.
• Control of stoichiometry is more difficult as the stoichiometric
ratio set point is lowered.
Despite these legitimate concerns, our overriding objective was to model
current industrial practice in FGD. Most limestone-based FGD systems are
designed to operate at low stoichiometries (1.1 to 1.15). Most chloride/FGD
studies to date have been performed at relatively high stoichiomet.ry (greater
than 1.3). Considering these facts, we decided to specify the stoichiometric
ratio at 1.1, with the understanding that if operating the pilot plant became
too difficult, the set point would be raised.
TEST RESULTS
The influence of chlorides on SO absorption was analyzed in four parts:
• Natural oxidation tests using finely ground limestone
• Natural oxidation tests using coarsely ground limestone
• Forced oxidation tests using fine limestone
• Forced oxidation tests using coarse limestone
Statistical analysis of 161 tests produced the correlations in Table 1. The
number of transfer units (NT) as utilized in these correlations is related to
S02 absorption (fgo ) by:
-In (1 - fgo ) = NT (E6)
The influence of chlorides on SO absorption can be thought of as a neg-
ative enhancement in a manner similar to Dankwertz's enhancement factor.
Laslo [11] has adapted that procedure by defining a term that we will call the
"diminution factor" as follows:
N at Cl = Cl
° ' at Cl • 0 (E7)
*Limestone stoichiometry (or stoichiometric ratio) refers here to the molar
ratio of calcium carbonate added to the FGD system to the moles of sulfur
dioxide absorbed by the system. This term is sometimes referred to in the
FGD literature as reagent ratio.
7-9
-------
Table 1
S02 ABSORPTION CORRELATIONS
Natural Oxidation Plus Fine Grinding
1/3
rSR"1>0l f -4 2
NT = -^Q-J I I ~ 0.0762 + 0.001496 (Cl/1000) - 0.4559 x 10 (Cl/1000)
+ 0.5346 TH + 0.02509 L, - 0.8835 x 10~4 (T..)(Y0.. ) - 1.365 x 10~6(L,)(YC. )
N f N S02 f S02
- 6.628 x 10~5 (C1/1000)(L£) (E2)
R = 96.7% a = 0.09227
Natural Oxidation Plus Coarse Grinding
1/3
rs - i.oi r
NT = 0 ~ °*1352 ~ °'0°2088 (Cl/1000) - 0.7832 x 10 (Cl/1000)
2
+ 0.2350 T + 0.02938 L_ - 0.8456 x 10~5 (T)(Yon ) - 0.3669 x 10~5(L,)(Yon )
N f N SO- f SO
+ 1.200 x 10~5 (C1/1000XL ) (E3)
R = 94.1% a = 0.0132
Forced Oxidation Plus Fine Grinding
1/3
2
FSR ' 1'°1 f -4
NT = -^Q-J - I 0.1393 + 0.003258 (Cl/1000) - 0.5008 x 10 q (Cl/1000)
+ 0.02718 (L,) - 0.3834 x 10~A (Y ) - 0.3210 x lo"5 (L )(Y )
r S0_ r s u_
- 0.4856 x 10~A (Lf)(Cl/1000) (E4)
R = 98.6% a = 0.0444
Forced Oxidation Plus Coarse Grinding
1/3
2
rs - i.o"] r
NT - -^o~i I I ~ °-1850 + 0.004229 (Cl/1000) - 0.2217 x 10~4 (Cl/1000)
+ 0.03421 (L,) + 0.1434 x 10~4 (YCA ) - 0.5426 x 10"5 (L,)(Y... )
t so2 t so2
- 1.343 x 10~4 (Lf)(Cl/1000) . (E5)
R » 98.2% a = 0.059
7-10
-------
The influence of chlorides was analyzed in this manner because of the
linear relationship between the diminution factor and the mass transfer
coefficient [11]. To relate the overall system SO absorption efficiency as a
function of the diminution factor, we simply combine Equations 6 and 7 and
rearrange to get:
JS(X
- £
so.
(100)
(E8)
To illustrate the influence that chlorides may have on SO performance,
Figure 3 was prepared from Equation 8; two circumstances are illustrated. The
curves sloping downward from right to left illustrate the loss in efficiency
of scrubber systems designed for 90% and 70% SO efficiency, respectively.
Conversely, the other two curves illustrate to what efficiency new scrubbers
must be designed to perform at these efficiencies if a chloride buildup is
anticipated.
100
90
80
70
O
O
v>
50
40
30
^DESIGN EFF. TO ACHIEVE 90%
—
DESIGN EFFICIENCY TO ACHIEVE 70%
EFFICIENCY OBTAINED FOR 90% EFF. DESIGN
EFFICIENCY OBTAINED FOR 70% EFF. DESIGN
0.4
O.S
0.6 0.7 0.8
DIMINUTION FACTOR
0.9
1.0
Figure 3. Relationship between diminution factor and SO
efficiency for high- and low-sulfur coals.
7-11
-------
The diminution factors were evaluated from the relationships in Table 1
and are presented in Figure 4. Although the influence of chlorides is signi-
ficant, the impact was generally less severe than Laslo observed. The
observed influence of such variables on S09 concentration and spray flux*
support the general contention that chlorides will affect performance by
reducing dissolved alkalinity. Note, for example, that during the natural
oxidation tests (Figures 4a through 4f), the diminution factor had its lowest
value at the lowest spray flux and highest SO . At these conditions, the dis-
solved alkalinity is most rapidly depleted. These figures also reveal the
dramatic influence of holdup. The diminution factor is greatly suppressed by
including sieve trays, which increase the residence time of the recirculated
slurry and, in so doing, permit Reactions 9 through 11 to participate in the
absorption process.
The use of trays in the coarse-grind, natural oxidation tests is even
more effective. This observation probably results from the fact that with
coarsely ground limestone in the spray tower configuration (no trays), the
heterogeneous reactions (R9 through Rll) participate to a lesser extent than
with finely ground limestone. This results in an even faster depletion of
alkalinity in the absorber. Then, as holdup is provided by the trays, these
reactions became progressively more important.
The influence of chlorides on SO absorption during the forced oxidation
tests is somewhat more difficult to interpret. During the fine-grind test,
S0» removal enhancement was actually observed. This enhancement could be due
to magnesium and sodium impurities, which tend to dissolve some sulfates as
MgSO and Na SO, , thereby increasing the dissolved alkalinity. Since all
forced oxidation tests were performed without sieve trays, it is difficult to
ascertain to what degree limestone dissolution participates to moderate the
depletion of dissolved alkalinity as chlorides increased. However, the more
severe effect of chloride on S02 absorption when using coarsely ground lime-
stone may indicate that Reactions 9 and/or 11 may be significant contributors.
If suppression of the chloride diminution factor with increasing holdup
is the result of improved limestone dissolution (as opposed to improved
calcium sulfite dissolution), then fine grinding should also contribute
significantly. To examine this question, we next define a "limestone grind
enhancement factor" by:
N with fine grind
1 N with coarse grind
where all other independent variables are constant. The fine-grind enhance-
ment factor is therefore the quotient of Equation 2 divided by Equation 3 for
natural oxidation and Equation 4 divided by Equation 5 for forced oxidation.
The results of this analysis are shown in Figure 5 and reveal that the influ-
ence of fine grinding is most significant for the spray tower arrangement.
*Spray flux is the normalized slurry recirculation rate in the absorber
(gpm/ft of absorber flow area).
7-12
-------
\\\
V
I
I
20 40
Cl. PPM * 10'3
lal
\\
\
eo
TWO TRAYS
Cl. PPM » 10
Ib)
NATURAL OXIDATION AND FINE GRIND
20 40
Cl. PPM » 10'3
Ic)
V
V
1
1
20 40
Cl, PPM > 10 3
Id)
20 40
Cl. PPM « 10'3
20 40
Cl. PPM « 10-3
m
_J
60
NATURAL OXIDATION AND COARSE GRIND
5. 0.6
1
Igl
FORCED OXIDATION AND FINE GRIND
----- Yso - 2600 PPM
— —.— =4000 PPM
S 0.9
* 0.8
0
20 40
Q. PPM . 10-3
FORCED OXIDATION AND COARSE GRIND
Figure 4. Influences of chlorides on SO. absorption.
7-13
-------
1.B
1.6
14
z
Ul
O
z
5
c
O
NO TRAYS
1.0
20 40
ci. PPM «i
-------
However, the benefit of fine grinding reduces as holdup increases. Taken as a
whole, the benefits of fine grinding do not change significantly with chloride
concentration. Fine grinding seems to be significantly more beneficial when
SO concentrations are high (4000 ppm) and spray flux is low. This fact sup-
ports the contention that Reaction 9 and/or reaction 11 do participate in the
SO,, absorption process.
Next, we examine the influence of forced oxidation on system performance.
Recall that forced oxidation offers a trade-off; neither dissolved sulfite nor
solid sulfite is present to provide a source of alkalinity. On the other
hand, S00 vapor pressure at the liquid-gas interface is negligible.
To examine the influence of forced oxidation, we again define an enhance-
ment factor, specifically the "forced oxidation enhancement factor," by:
N , forced oxidation
N , natural oxidation
(E10)
The results of this analysis are depicted in Figure 6 for finely ground
limestone. We cautiously concluded that for conditions of severely depleted
dissolved alkalinity (high SO concentration and low-slurry spray rates), the
absence of calcium sulfite is significant and produces a reduced level of
performance. Conversely, for conditions where dissolved alkalinity is not
severely taxed, forced oxidation is beneficial to S07 removal. These tests
also seem to suggest that the benefits of scrubbing with an oxidized slurry
improve slightly with increasing chloride concentrations. This further sug-
gests that during natural oxidation tests, calcium sulfite dissolution could
be inhibited as the chloride concentration increases. Inhibiting calcium
sulfite dissolution by calcium chloride has been observed by Rochelle [12].
•&"
EC
0
t-
u
5 1.2
z '
Ul
s
111
u
z ,
< *
I 1.0
z
Ul
Z
o
Q
0 0.8
6
Ul
U
c
o
u.
A A
NO TRAYS
FINE GRIND ^
& *
^ A
A s*
^^
i" ^ °
^ A f^^^^^^^ Q
^ ^-" Q^" ^^^ '
i^^^^^^^ -~ ^^
n ^^^
*~ ~^^^
^^ ^^
^^."^ O
a**-**
^
-
A L, = 52, Yso - 2600 PPM
O L, = 26. Yso 2 = 2600 PPM
D Lf = 26, Yso = 4000 PPM
2
I I
0 20 40 6
Cl. PPM x 10'3
Figure 6. Forced-oxidation enhancement of SO. absorption
with fine limestone.
7-15
-------
SUMMARY AND CONCLUSIONS
/
The influence of chlorides on FGD performance in limestone processes
reduces the rate of SO' absorption. This diminution results from decreased
dissolved alkalinity and from reduced ion diffusivity. The negative impact of
chlorides can be suppressed by increasing the spray flux (or L/G), by increas-
ing holdup in the absorber to promote dissolution of calcium sulfite and
limestone, and by increasing the limestone fineness. Fine grinding of
limestone was a particularly effective way of suppressing the chloride
influence during forced oxidation tests.
For example, these analyses suggest that an FGD system (either natural or
forced oxidation) designed to achieve 90% SO removal in a spray tower employ-
ing finely ground limestone (95% passing 325 mesh) and an L/G of 90 gal/MCF
should experience an S09 efficiency loss of only, about 2% to 3% at a chloride
level of 30,000 ppm.
On commercial FGD absorbers, the influence of spray flux (or L/G) upon
the diminution factor may be greater than observed here. With the use of low-
pressure nozzles in this test program, the total SO mass transfer rate was
somewhat less than would be expected for commercial installations. For a
given liquid spray flux, therefore, the extent of alkalinity depletion in the
pilot program was less than would be expected in a commercial FGD system.
Consequently, the influence of chlorides was probably also reduced.
The influences of finely ground limestone on the diminution factor were
most noticeable in the spray tower mode (no trays) under conditions of severe
alkalinity depletion. The use of trays, high spray flux, and low SO concen-
trations tended to reduce the significance (enhancement) of fine grinding.
Commercial FGD processes may be even more responsive to fine grinding than the
Shawnee facility, because limestone grinding was done in a dry, closed-loop,
ball-mill system. This system appeared to produce a broader particle size
distribution than conventional, wet, closed-loop, ball-mill systems. The net
effect is that the steady-state limestone surface area of the residual lime-
stone in the recirculating slurry had a lower specific surface area than
expected for finely ground limestone (95% passing 325 mesh).
The influence of chlorides in an absorber utilizing a fully oxidized
slurry containing finely ground limestone was generally less severe than for
the comparable case with a natural oxidized (partially oxidized) slurry. The
opposite was true with coarsely ground limestone.
The principal conclusion that can be drawn from this test program is that
chlorides do generally inhibit SO absorption. However, the severity of this
effect can vary widely, depending upon the design and method of operation. The
influence of chlorides can be minimized by applying the factors noted here.
7-16
-------
NOTICE
This publication was prepared by Babcock & Wilcox (B&W) and the Tennessee
Valley Authority (TVA) as an account of work sponsored by B&W, TVA, and the
Electric Power Research Institute (EPRI). Neither B&W, TVA, EPRI, members of
EPRI, nor persons acting in behalf of B&W, TVA, or EPRI: (a) makes any war-
ranty or representation, express or implied, with respect to the accuracy,
completeness, or usefulness of the information contained in this report, or
that the use of any information, apparatus, method, or process disclosed in
this report may not infringe privately owned rights; or (b) assumes any lia-
bilities with respect to the use of, or for damages resulting from the use of,
any information, apparatus, method, or process disclosed in this publication.
This paper does not necessarily reflect the views and policies of the TVA.
NOMENCLATURE
Cl = Chloride ion concentration, ppm in liquid phase
Ego - SO absorber efficiency under the influence of chlorides, %
fg0 = SO- absorber efficiency in the absence of chlorides, fractional
f0. = SO. absorber efficiency, fractional
S02 2
K = Solubility product of CaCO
Kg » Solubility product of CaSO -1/2 HO
K = Solubility product of CaSO •2H 0
Lf - Slurry spray flux, gpm/ft
N = Number of transfer units
SR = Stoichiometric ratio (reagent ratio), moles CaCO- added/mole
SO. absorbed
T •=• Number of sieve trays
YSO = Inlet S09 concentration, ppm
( ) = Activity, molarity
a = Diminution factor
6, ** Fine-grind enhancement factor
-------
REFERENCES
1. P. C. Rader, R. W. Hanson, D. C. Borsare, "Design of Lime/Limestone Flue
Gas Desulfurization Systems for High Chlorides," presented at Coal
Technology '81, Houston, Texas, Nov. 1981.
2. P. 0. Karjalainen, J. E. Lofkrantz, and R. D. Christie, "Chloride Buildup
in Kraft Liquor Systems," P&P Mag, of Canada, 73, 12, 95, Dec. 1972.
3. D. W. Reeve, et al., "The Effluent Free Bleached Kraft Pulp Mill,"
Part IV of "The Salt Recovery Process," P&P Mag, of Canada, 75, 8, 67.
Aug. 1974.
4. Bechtel Corp. (H. N. Head), "EPA Alkali Scrubber Test Facility: Advanced
Program, Third Progress Report," Cont. No. 68-02-1814, Program Element
No. EHE624, EPA-600/7-77-105, Sept. 1977.
5. J. C. S. Chang and D. Laslo, "Chlordie Ion Effects on Limestone FGD Sys-
tem Performance," presented at the EPA/EPRI FGD Symposium, May 18, 1982.
6. P. C. Rader, D. C. Borsare, and D. Frabotta, "Process Design of
Lime/Limestone Systems for High Chlorides," presented at Coal Technology
'82, Houston, Texas, Nov. 1982, page 33.
7. R. H. Borgwardt, "Effect of Forced Oxidation on Limestone/SO Scrubber
Performance," EPA Symposium on Flue Gas Desulfurization, Hollywood,
Florida, EPA 600/7-78-058, 1978.
8. G. T. Rochelle, P. K. R. Chan, and A. T. Toprac, "Limestone Dissolution
in Stack Gas Desulfurization Processes," Draft Final Report to EPA, EPA
Cooperative Agreement R806251, April 1982, page 71.
9. J. B. Jarvis and T. W. Trofe, "Effect of High Dissolved Solids on Bench-
Scale FGD Performance," prepared for presentation at EPA/EPRI Symposium
on Flue Gas Desulfurization, New Orleans, Nov. 1983.
10. G. A. Hollinden, "Chemistry of Lime/Limestone Scubbing Liquor from Power
Plant Stack Gases," Presented at 35th Annual Meeting of the International
Water Conference, Pittsburgh, Pa., Oct. 30 - Nov. 1, 1974.
11. D. Laslo and E. Bakke, "The Effect of Dissolved Solids on Lime and Lime-
stone FGD Scrubber Chemistry," prepared for presentation at EPA/EPRI
Symposium on Flue Gas Desulfurization, New Orleans, ,Nov. 1983.
12. G. T. Rochelle and C. H. Tseng, "Dissolution and Crystallization of
CaSO -1/2 HO," Progress Report - FGD Research Group, Univ. of Texas,
JuneJ23, 1983.
7-18
-------
EFFECT OF LIMESTONE GRINDING CIRCUIT ON
FGD PERFORMANCE AND ECONOMICS
J. D. Colley, 0. W. Hargrove, Jr.,
D. A. Stewart
-------
EFFECT OF HIGH DISSOLVED SOLIDS ON BENCH-SCALE
FGD PERFORMANCE
by: James B. Jarvis and Timothy W. Trofe
Radian Corporation
Austin, Texas 78766
Dorothy A. Stewart
Electric Power Research Institute
Palo Alto, CA 94303
ABSTRACT
To minimize wastewater treatment costs, utilities are evaluating the
use of cooling tower blowdown as makeup water for wet limestone FGD systems.
If closed loop operation is employed, the dissolved species in the makeup
water can be concentrated to significant levels. Additional ions may enter
the scrubbing liquor through the absorption of chemical species in the flue
gas. An important example is the absorption of HC1 produced during the
combustion of high chloride coal.
The effect of various high total dissolved solids (TDS) solutions on
S02 removal and other system variables was evaluated in a bench-scale
limestone FGD system. Five salts including CaCl2, MgCl2, NaCl, MgS04,
and Na£S04 were evaluated under both natural and forced oxidation
conditions.
Additional laboratory testing was conducted to isolate specific effects
observed in the more complex bench-scale tests. These tests were designed to
illustrate the effect of high TDS solutions on chemical equilibria and
gas-liquid mass transfer in the absorber.
The results of this investigation show that the performance of the FGD
system is determined by the effect of the added salt on: 1) the concentra-
tion of dissolved ions in the scrubbing slurry, 2) the solubility of dis-
solved S02, 3) mass transfer enhancement through the formation of bisulfite
ion pairs with the added cation, 4) sulfate-bisulfate buffering in high
sulfate systems, and 5) the mass transfer rate of diffusing species through
the liquid film at the gas-liquid interface. The bench-scale test results
also compare favorably to the results from similar tests conducted at TVA's
Shawnee Test Facility in Paducah, KY. and EPA's pilot facility at Research
Triangle Park, NC.
Prepared for Presentation at EPA/EPRI Symposium on Flue Gas Desulfurization,
New Orleans, November 1-4, 1983.
7-19
-------
INTRODUCTION
This paper summarizes selected results from a laboratory study funded by
the Electric Power Research Institute to examine the effect of high total
dissolved solids (TDS) on S02 removal in wet limestone FGD systems.
Dissolved species such as sodium, magnesium, and chloride can enter the
scrubber liquor if process streams such as cooling tower blowdown are used as
makeup water to the FGD system. Alternatively, dissolved solids can enter
the scrubber liquor through the absorption of soluble species in the flue
gas. This situation occurs, for example, when HC1 produced during the
combustion of high chloride coal is absorbed in the scrubbing liquor along
with S02« HC1 reacts with limestone or lime in the scrubber slurry
resulting in an accumulation of CaCl2. If closed-loop operation is used,
the dissolved solids from either of the above sources can be concentrated to
levels which are considerably higher than those presently observed in most
systems.
The accumulation of dissolved solids alters scrubber chemistry and can
affect S02 removal, calcium sulfite and sulfate crystal morphology and
size, and other process operating characteristics. The remainder of this
paper centers on the effect of high TDS solutions on S02 removal. The data
presented here was obtained using Radian's bench-scale FGD system. Five
salts including CaCl2, MgCl2, NaCl, MgS04, and Na2S04 were
evaluated. Results are presented in three subject areas including:
1) Laboratory batch tests designed to illustrate the effect of high
TDS solutions on chemical equilibria and gas-liquid mass transfer;
2) S02 removal for natural and forced oxidation bench-scale FGD
system tests at constant limestone reagent ratio for each of the
five salts tested; and
3) Comparisons between the bench-scale FGD system results and the
results from similar tests conducted at TVA's Shawnee Test Facility
in Paducah, KY. and EPA's pilot facility at Research Triangle Park,
NC.
LABORATORY BATCH TEST RESULTS
A series of laboratory experiments were performed to investigate the
effect of each of the tested salts on chemical equilibria and gas-liquid mass
transfer:
• The liquid-side and gas-side mass transfer coefficients, kpa and
kga, were determined for the bench-scale bubbler contactor in
both a low TDS solution and a concentrated NaCl solution. The
purpose of these tests was to determine the effect of the high TDS
solution on the physical mass transfer rate of dissolved
7-20
-------
• Experiments were conducted to determine the effect of IDS composi-
tion and concentration on the solubilities of dissolved S02 and
total sulfite at constant S02 partial pressure.
The results of these laboratory tests illustrate several important
aspects of the effect of high IDS solutions on SC>2 removal. As a package,
these results are extremely useful in interpreting the results of the more
complex bench-scale FGD system tests presented later in this paper.
DETERMINATION OF THE BUBBLER SCRUBBER MASS TRANSFER COEFFICIENTS
The local gas-side and liquid-side mass transfer coefficients were
determined for the bubbler contactor used in the majority of the bench-scale
FGD system tests. The gas-side coefficient, kga, was measured by once-
through scrubbing with a 0.1 N NaOH solution. The liquid-side coefficient,
k^a, was measured by once—through scrubbing with a 0.5 N HC1 solution. In
each case, the mass transfer coefficients were measured in a low TDS solution
(pure acid or base) and in a concentrated NaCl solution (150,000 ppm Cl~ as
NaCl). The purpose of these tests was to determine if the higher density and
viscosity of the NaCl solution would result in a change in the scrubber mass
transfer characteristics.
With the techniques utilized in this experiment, a change in k»a
O
would represent a change in the interfacial area available for mass transfer.
The experimental results show that kga decreased by only about 5 percent
with the addition of the concentrated NaCl solution indicating that the
increased solution density and viscosity did not have a significant effect on
the gas-liquid contacting characteristics of the bubbler absorber. It should
be emphasized that this result is specific to the bubbler absorber and may
not hold true for all scrubber configurations. In a spray tower, for
example, high TDS solutions could affect nozzle performance and result in an
increase in droplet size. This could change kga significantly by reducing
the total surface area available for mass transfer.
In strong acid, the liquid-side mass transfer coefficient, kna, is a
function of the rate of surface renewal or turbulence in the liquid film and
the diffusion rate of dissolved S02« The experimental results show that
k^a decreased about 50 percent with the addition of the concentrated NaCl.
This result indicates that the increased density and viscosity of the NaCl
solution has a substantial effect on the turbulence within the liquid film
and/or the diffusion rate of dissolved S02« It is impossible to determine
from the experimental results which of these factors is most affected. In
any case, the decrease in k^a should result in a substantial decrease in
S02 removal, particularly in cases where liquid-phase alkalinity is low.
EFFECT OF HIGH TDS SOLUTIONS ON S02 SOLUBILITY
An additional factor affecting the physical mass transfer of dissolved
S02 (H2S03) is the ^803 concentration at the gas-liquid interface.
7-21
-------
Since the bulk liquid H2S03 concentration in typical FGD slurries is
generally very low, the liquid phase ^803 concentration at the gas-
liquid interface fixes the concentration driving force for H2S03
diffusion.
In conventional mass transfer theory, the gas-phase SC>2 concentration
and the liquid phase H2S03 concentration are in equilibrium at the gas-
liquid interface and can be related by an expression similar to Henry's Law.
The concentration of dissolved S02 (112803) was measured by experiment
for a variety of IDS solutions at constant 862 partial pressure. These
solutions consisted of acidified water and acidified solutions containing
concentrated NaCl, MgCl2, CaCl2, MgSO^ and Na2SC>4 (about 138,000
ppm Cl~ or S04=). The experimental results show that the equilibrium
dissolved 862 concentration for each of the concentrated salt solutions is
about 25 percent (on a weight basis) below that of the acidified water alone.
This "salting out" effect, together with the effect of increased viscosity on
the liquid-side mass transfer coefficient, suggests that the physical mass
transfer rate of 802 should decrease significantly with increasing ionic
strength.
EFFECT OF HIGH TDS SOLUTIONS ON TOTAL SULFITE SOLUBILITY
Thus far, the discussion of the effect of high TDS solutions has been
limited to mass transport by physical absorption and diffusion. The other
component of S02 mass transport in FGD systems is 802 mass transfer
enhancement by chemical reaction. Species which are alkaline with respect to
H2S03 react with H2S03 to form HS03~. This reduces the H2S03
concentration in the liquid film and increases the driving force for
H2S03 diffusion. Reactions which remove HS03~ from solution also
enhance S02 removal by allowing additional ^803 to dissociate via the
hydrolysis reaction:
S02 + H20 J H+ + HS03- (1)
In low TDS solutions, the important alkaline species include
and HC03~ which can be present initially in the liquid phase or result
from the dissolution of calcium sulfite and limestone. In high TDS solu-
tions, however, additional reactions can occur which enhance 802 removal.
The most important of these reactions include the formation of ion pairs
involving HS03~ an<* sulfate-bisulfate buffering in high sulfate solu-
tions. High ionic strength solutions also affect S02 removal enhancement
by changing the activity coefficients of other dissolved species.
The alkalinity provided by ion pairing and sulfate-bisulfate buffering
was measured for each of the five salts studied during this program by
equilibrating the pure salt solutions with an N2~S02 gas stream at fixed
802 partial pressure. The 802 absorbed into solution exists as
H2S03, HS03~, SQ3=, and ion pairs containing HS03~ and
S03=. The observed total sulfite concentration is a measure of the
ability of the salt solution to enhance 802 removal through the hydrolysis
7-22
-------
reaction, the formation of ion pairs, and sulfate-bisulfate buffering in high
sulfate solutions. In practice, the alkalinity of the scrubber slurry is
augmented by S03= and HC03~ initially present in solution and
dissolution of limestone and calcium sulfite.
The experimental results illustrated in Figure 1 show the total sulfite
concentration absorbed into the salt solution versus the chloride or sulfate
concentration of the salt tested. Note that the measured total sulfite
concentrations have been divided by the S02 partial pressure (about 1.38 mm
Hg ave.) for the purpose of normalizing for small variations in the gas phase
SC>2 concentration. The results show that the total dissolved sulfite
concentration is a strong function of the composition and concentration of
the salt tested. For the chlorides, an increase in total dissolved sulfite
is seen up to about 30,000 ppm Cl~. The amount of dissolved sulfite falls
off at higher chloride concentrations. The largest degree of enhancement at
the 30,000 ppm Cl~ level is observed with CaCl2. The least enhancement
occurs with NaCl. Total dissolved sulfite concentrations in the sulfate
solutions were considerably higher than those seen in the chloride solutions
due to sulfate-bisulfate buffering.
The effect of TDS composition and concentration on total sulfite solu-
bility plays a significant role in determining the S02 removal level in FGD
systems. These effects are quite apparent in the bench-scale FGD system
results which follow.
BENCH-SCALE FGD SYSTEM TEST RESULTS
WITH HIGH TDS SOLUTIONS
A total of 36 bench-scale tests were performed to evaluate the effect of
TDS composition and concentration on S02 removal. The core of this test
series consisted of 10 runs in which the effect of each of the five salts
tested was evaluated under both forced and natural oxidation conditions.
These runs were performed^t a constant limestone loading of about 10 g
CaC03/l which is approximately equivalent to a limestone reagent ratio of
1.10 at 15 wt. percent solids. The limestone used in these tests was high
calcite Edwards limestone (100% through 200 mesh, 95% through 325 mesh).
A schematic diagram of Radian's bench-scale FGD system is given in
Figure 2. Synthetic flue gas containing N2, 02, C02, and S02 was
saturated with H20 to about 40°C and fed to the absorber. For most tests,
the flue gas composition (dry basis) was 77% N2, 7% 02, 16% C02, and
2000 ppm S02.
The bench-scale scrubber process flow diagram is essentially the same as
that for a full-scale FGD system. Slurry from the hold tank is fed to the
absorber where it contacts the flue gas. The scrubber effluent is recycled
back to the hold tank. Closed-loop operation is achieved by sending a
portion of the hold tank slurry to an in-line filter. Clear filtrate is then
returned to the hold tank. The temperature of the hold tank was maintained
at 50°C by recycling part of the slurry through coils- in a heated water bath.
7-23
-------
1000-
900-
300
200 -
100
Dissolved S02 (H2S03) only
20K 40K 60K 80K 100K
Chloride or sulfate concentration, ppm
120K
140K
Figure 1. Comparison of total sulfite solubility at constant S02 partial
pressure for NaCl, MgCl2, CaCl2, Na2S04, and MgS04.
7-24
-------
DRY GAS
METER
D.I.
H3O
• VENT
MANIFOLD
V )PRESSURE
GAUGE
H2C
^—•v
TCV—
C/
(*-
p*
^SL
T
> SATURATOR
COLUMN ABSORBER
I ^ icarKCn Hen
•*N y
O-i:
* *
•?
9V AC
^
BUBBLER. OR
SPRAY TOWER)
x-*-v
—
\ .
V
/ \
^ i
THETA SENSOR
SO, ANALYZER RANGES
0-SOOppm
0-ISOOppm
0-SOOOppm
DOWNSTREAM IMPINGERS
• VENT
•*-tXl ?N,
*-CX] {0,.N,(AIR|
^-«l—ICQ, (TC
•*-CX] 1 S07. N,
ABSORBER RECYCLE
(SLURRY OR
CLEAR LIQUOR)
FILTRATE RECYCLE
Figure 2. Schematic diagram of the bench-scale FGD system.
7-25
-------
A bubbler contactor was used for most of the bench-scale tests
conducted in this program. The bubbler was designed such that L/G,
gas-liquid contact time, and the residence time of slurry within the
contactor could be varied independently. The flow characteristics within the
scrubber are best described as plug flow with respect to the gas phase and
back-mixed with respect to the liquid phase. In this configuration, the
bench-scale scrubber is intended to simulate the SC>2 removal
characteristics of a tray tower with a single tray.
The S(>2 removal results for the natural and forced oxidation runs are
shown in Figures 3 and 4, respectively. For the natural oxidation runs in
Figure 3, the initial and final total sulfite concentrations for each test
are also listed.
For NaCl, a slight enhancement in S02 removal is seen for both forced
and natural oxidation at chloride levels of about 15,000-20,000 ppm. S02
removal decreases substantially, however, at higher chloride levels. The
initial enhancement in S02 removal is the result of the increase in total
sulfite solubility as shown in Figure 1. The subsequent decrease in S02
removal is the result of at least three factors:
1) A decrease in total sulfite solubility above the 20,000 ppm Cl~
level (see Figure 1),
2) A decrease in the liquid-side mass transfer coefficient with
increasing chloride concentration, and
3) A decrease in dissolved S02 (H2S03) solubility with
increasing ionic strength.
Note that the sulfite concentration has little effect on the change in SC>2
removal with NaCl since the low and high chloride level sulfite concentra-
tions are nearly the same for both the natural and forced oxidation runs.
Enhancement in S02 removal is also observed for MgCl2 at around
25,000 ppm Cl~ for both the natural and forced oxidation runs. The level
of enhancement is somewhat higher than that seen for NaCl. This result is
expected based on the total sulfite solubility results in Figure 1. At
chloride levels above 25,000 ppm Cl~, S02 removal decreases as a result
of the same factors causing decreased 802 removal with NaCl. The decrease
in S02 removal for the natural oxidation run with MgCl2 was not as severe
as with NaCl, however, due to an increase in liquid phase sulfite at high
MgCl2 concentrations.
Based on the total sulfite solubility results in Figure 1, considerable
enhancement in 802 removal should be observed with CaCl2. In fact, the
level of enhancement should be the highest of the three chloride salts
7-26
-------
90
80
70
60
4, and MgSC>4 on S02
removal for natural oxidation at constant limestone reagent ratio.
7-27
-------
100
90
80
70
60
S 50
u
u
11
Ol.
40
30
20
10
MgCl2
L/G = 80 gal/103 SCF
Limestone Loading = 10-12 g CaC03/1
NG = 2.2
Sulfite concentrations are less than
35 ppm for all tests.
j L
30K 60K 90K
Chloride or Sulfate Concentration, pom
120K
150K
Figure 4. Comparison of the effect of NaCl2, CaCl2, NaSC^, and MgSO^ on S02
removal for forced oxidation at constant limestone reagent ratio.
7-28
-------
tested. However, the data in Figures 3 and 4 show that no enhancement was
observed with CaCl2 at chloride levels around 25,000 ppm. With CaCl2,
the enhancement expected from the increase in total sulfite solubility is
offset by a decrease in the feed liquor concentrations of total SC>3= and
C03=. As shown in Table 1. a decrease in total S03= and C03=
occurs with increasing Ca"1"* due to the common ion effect. The net result
of these two competing factors is little, if any, effect on SC>2 removal for
the bench-scale tests at chloride levels up to 20,000 ppm.
At high chloride concentrations, S(>2 removal for natural oxidation
with CaCl2 is considerably below that for either MgCl2 or NaCl. As
indicated in Figure 3, this is probably caused by the differences in the
S03= concentrations in solution. In forced oxidation, however, the
S03= concentration is low for all three chloride salts and the high
chloride S02 removals are much closer together.
One other consideration involving CaCl2 should be mentioned. In
operating FGD systems, high CaCl2 concentrations will most likely occur as
a result of the combustion of coal containing chlorine. The combustion
product, HC1, will be absorbed in the scrubber along wtih S02 and will
consume a portion of the alkalinity in the scrubbing liquor. The effect of
the absorption of HC1 on SC>2 removal was not considered in the bench-scale
study.
The natural oxidation results in Figure 3 for MgS04 and Na2S04
show that the SC>2 removals increased significantly from the baseline level
and approached the maximum possible S02 removal level for the contactor.
This result is not surprising in view of the high sulfite concentrations in
the scrubbing liquor. What is surprising, however, is that S02 removals in
the forced oxidation tests in Figure 4 were only slightly below the
corresponding natural oxidation tests even though the liquid phase sulfite
concentrations were very low. The high S02 removals for the forced
oxidation tests with S04= are the result of the large increase in total
sulfite solubility (Figure 1) caused by the combination of ion pair formation
and sulfate-bisulfate buffering.
COMPARISON OF BENCH-SCALE, SHAWNEE, AND RTF HIGH CHLORIDE TEST RESULTS
One of the benefits of bench-scale testing is that it allows the
evaluation of important process variables at costs which are well below those
of similar tests on larger units. In addition, the flexibility inherent to
bench-scale systems along with better control over process variables allows
tests to be conducted which are impossible at full-scale. To be of value,
however, the bench-scale tests must adequately predict the results seen with
larger units.
The results of three test programs in which the effect of CaCl2 on
S02 removal was evaluated are compared in the following paragraphs. These
programs include the bench-scale FGD system tests described in this paper, a
high CaCl2 test program conducted by Babcock and Wilcox at TVA's Shawnee
7-29
-------
TABLE 1. EFFECT OF CaCl2 ON THE CONCENTRATION OF DISSOLVED SOLIDS
Bench-scale
Run No.
TDS-13
Natural
Oxidation
TDS-14
Forced
Oxidation
CaCl2 Cone. ,
ppm Cl~
13
1,780
4,330
8,460
16,600
32,800
63,000
89,400
121,000
19
1,700
8,850
16,500
31,200
66,700
104,000
140,000
Total S03=,
ppm
167
142
101
86
80
62
46
40
25
14
6
—
4
—
—
—
3.5
ppm
1938
—
—
—
—
—
—
—
102
1923
—
—
—
—
—
—
63
Total 003=,
ppm
184
—
—
—
—
—
—
—
76
108
76
89
—
59
—
—
41
7-30
-------
prototype facility, and high IDS testing conducted by Peabody Process Systems
at EPA's pilot facility at Research Triangle Park (RTP). The effect of
CaCl2 on S(>2 removal is given in Table 2 for each of the three test
programs. Two types of comparisons are presented including:
1) The effect on SC>2 removal of an increase in CaCl2 from 0 to
60,000 ppm Cl~ for both forced and natural oxidation, and
2) The difference in SC>2 removal between forced and natural
oxidation at the 0 and 60,000 ppm Cl~ levels.
The most striking result shown in Table 2 is the dramatic decrease in
S02 removal in the RTP natural oxidation test. In this test, a decrease of
26 percent S02 removal was observed which is more than twice the decrease
seen in either the bench-scale or Shawnee test results. This result is also
evident in the comparison of the difference in SC>2 removal between forced
and natural oxidation. All three test programs show that S02 removal with
forced oxidation is above that for natural oxidation. However, the
difference of 18 percent S02 removal in the RTP tests is greater than four
times that seen in either the bench-scale or Shawnee tests. This occurred
despite the fact that the limestone reagent ratio in the natural oxidation
tests (1.35) was well above that used in the forced oxidation tests (1.20).
From the available data, it would appear that excessive limestone
blinding had occurred in the RTP tests. Several additional bench-scale tests
were performed to investigate the cause of the rapid decrease in natural
oxidation S02 removal at RTP. These tests included:
• Operation with limestone from the RTP facility,
• A determination of the effect of hold tank C02 partial pressure
(at 80,000 ppm Cl"),
• Simulation of potential limestone blinding in the scrubber effluent
line, and
• A determination of the effect of calcium sulfite dissolution in the
scrubber.
Although these tests increased our understanding of the effect of
CaCl2 on FGD system operation, none of the test results indentified the
cause of the drastic decrease in S02 removal in the natural oxidation RTP
tests. Calcium sulfite precipitation within the scrubber was observed at
higher chloride levels and limestone reagent ratios. However, this did not
result in limestone blinding.
The data in Table 2 show reasonable agreement between the bench-scale
and Shawnee results. For forced oxidation, the bench-scale and Shawnee
results show decreases of 7 and 8 percent S02 removal, respectively, at the
60,000 ppm Cl~ level. For natural oxidation, the Shawnee results show a
7-31
-------
TABLE 2. COMPARISON OF S02 REMOVAL RESULTS WITH CaCl2 FOR THE
BENCH-SCALE, SHAWNEE, AND RTF TEST PROGRAMS*
Decrease in % SO?
Removal for an Increase Difference between Forced
in CaCl2 from 0 to Oxidation and Natural
60,000 ppm Cl~ Oxidation % S02 Removal**
Natural Forced 60,000
Oxidation Oxidation 0 ppm Cl~ ppm Cl~
Radian Bench-scale 775 4
Tests***
Shawnee Tests**** 11 8 1 4
RTP Tests***** 26 9 2 18
* All results are for constant limestone reagent ratio tests.
** Forced oxidation S02 removals were higher than those for natural
oxidation in all cases.
*** Tests performed with a bubbler contactor at a limestone loading of 10 g
CaC03/l which is approximately equivalent to a limestone reagent
ratio of 1.10 at 15 wt.% solids.
**** Test results generated from Radian's regression model of the
experimental data. S02 removals estimated at the following run
conditions: inlet S02 = 2500 ppm, L/G = 80 gal/103 ACFM, one
gas-liquid contacting tray, fine grid limestone, and a limestone
loading of 10 g CaC03/l (limestone reagent ratio = 1.10).
***** Tests performed in a three-stage turbulent contact absorber. Limestone
reagent ratios were 1.35 for natural oxidation and 1.20 for forced
oxidation (1).
7-32
-------
slightly greater decrease; 11 percent versus 7 percent for the bench-scale
tests. The most likely reason for this difference can be seen by comparing
the liquid phase concentrations of Kg"*""1" and S03= typically seen at very
low chloride levels:
Bench-Scale Shawnee
Mg++, ppm . 50-60 200-250
803=, ppm 150-160 600-650
The above data show that the Shawnee scrubber liquor typically had
higher NgSOj concentrations than were used in the relatively "clean"
bench-scale scrubbing liquor. That is, a slight amount of magnesium
enhancement of S02 removal was present in the Shawnee tests. Therefore,
the initial (low chloride) liquid phase alkalinity in th'e Shawnee liquor was
somewhat higher than that of the bench-scale tests. With the addition of
CaCl2, however, MgS03 in solution is quickly converted to MgCl2 through
the following reaction:
MgS03(aq) + CaCl2(aq) t CaS03(s)4- + MgCl2(aq) (2)
At high CaCl2 levels, the liquid phase concentrations (and liquid
phase alkalinity) of the two scrubbing liquors are nearly identical. In both
cases, the S03= concentrations are very low. Therefore, over the range
of 0 to 60,000 ppm Cl~, the net decrease in liquid phase alkalinity is
somewhat greater for the Shawnee liquor than for the bench-scale liquor. The
result is a slightly greater decrease in S02 removal for the Shawnee
natural oxidation tests.
Data have been obtained at the RTF pilot facility on the effect of NaCl
and MgCl2 on S02 removal(2). These data are compared to the bench-scale
results for both forced and natural oxidation in Figures 5 through 8.
Operating conditions for these tests are summarized below.
Radian Tests RTF Tests
Inlet S02, ppm 2000 2500
L/G, gal/105 cfm 80 60
Contactor Bubbler (single tray) TCA
Limestone reagent ratio Fixed at lOg CaCC>3/l 1.1-1.2
The results in Figures 5 through 8 show reasonably good agreement
between the bench-scale and RTF data.
CONCLUSIONS
1) Chloride affects S02 removal in wet limestone FGD systems through at
least four mechanisms. First, the concentrations of important dissolved
7-33
-------
90 r-
o
0)
1-1
CM
O
CO
c
0)
o
80
70
60
50
Bench-scale data
.RTF data
40
I
30K
60K
9 OK
120K
150K
Figure 5.
Chloride concentration, ppm
Comparison of bench-scale and RTF data for the effect
of NaCl on natural oxidation SC-2 removal.
90
I
d>
o
co
c
0)
a
t-i
ai
PM
80
70
60
50
Bench-scale data
•RTF data
40
Figure 6.
30K
60K
90K
12 OK
150K
Chloride concentration, ppm
Comparison of bench-scale and RTF data for the effect
of NaCl on forced oxidation S02 removal.
7-34
-------
90 r
n)
i
O
CO
4J
C
0)
O
S-I
0)
80
70
60
50
Bench-scale data
RTF data
40
_L
30K
60K
90K
120K
150K
Chloride concentration, ppm
Figure 7.
Comparison of bench-scale and RTF data for the effect
of MgCl2 on natural oxidation SC>2 removal
90r-
30
70
2 removal.
7-35
-------
species can be affected, depending on which chloride salt is present. In the
case of CaCl2, the concentrations of S03=, S04=, and C03= are
reduced due to the common ion effect. Second, the physical solubility of
SC>2 decreases with increasing ionic strength. This reduces the dissolved
S02 concentration at the gas-liquid interface and decreases the driving
force for diffusion of dissolved 802 through the liquid film. Third, the
total sulflte solubility of the salt solution is increased through ion
pairing involving bisulfite and the salt cation. For the chlorides, a maxi-
mum in the total sulfite solubility is seen around the 30,000 ppm Cl~
level. Finally, mass transfer rates through the liquid film are decreased
due to a reduction in the diffusion rates of dissolved species and/or the
turbulence in the liquid film.
2) S02 removal is enhanced in both natural and forced oxidation systems
with the addition of sodium or magnesium sulfate salts. In the natural
oxidation case, improved S02 removal can be attributed to increased
alkalinity in the form of liquid phase sulfite. S02 removal under forced
oxidation conditions approaches that with natural oxidation. In forced
oxidation, the enhancement in S02 removal with increasing sulfate can be
attributed to a substantial increase in total sulfite solubility at the
gas-liquid interface through sulfate-bisulfate buffering and the formation of
bisulfite ion pairs.
3) Bench-scale tests can be used to predict the effects of high TDS solu-
tions on the operating characteristics of larger FGD units. Bench-scale test
results compare favorably to pilot-scale test results from Shawnee for the
effect of CaCl2 on S02 removal. The bench-scale results with CaCl2
also compare favorably with pilot-scale tests at RTF for forced oxidation.
However, the drastic decrease in S02 removal with CaCl2 for natural
oxidation at RTF cannot be explained on the basis of either the bench-scale
results or the Shawnee results. Good agreement between the bench-scale and
RTF results is also seen in natural and forced oxidation tests with MgCl2
and NaCl.
LITERATURE CITED
1. Chang, J.C.S., and D. Laslo, "Chloride Ion Effects on Limestone FGD
System Performance," in EPRI CS-2897, pp. 224-254, March 1983.
2. Laslo, D., J.C.S. Chang, and J. D. Mobley, "Pilot Plant Tests on the
Effects of Dissolved Salts on Lime/Limestone FGD Chemistry," presented
at the EPA/EPRI Symposium on Flue Gas Desulfurization, New Orleans,
November 1-4, 1983.
7-36
-------
PILOT PLANT TESTS ON THE EFFECTS OF DISSOLVED SALTS
ON LIME/LIMESTONE FGD CHEMISTRY
D. Laslo, J. C. S. Chang, J. D. Mobley
-------
Pilot Plant Tests on the Effects of Dissolved
Salts on Lime/Limestone FGD Chemistry
by: Dennis Laslo
Peabody Process Systems, Inc.
835 Hope Street
Stamford, CT 06907
John C. S. Chang
Acurex Corporation
P.O. Box 13109
Research Triangle Park, NC 27709
J. David Mobley
U.S. Environmental Protection Agency
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
ABSTRACT
This paper presents an overview of pilot plant testing from September
1981 through June 1983 of the effects of dissolved salts on a lime or
limestone flue gas desulfurization (PGD) system at the Environmental
Protection Agency's Industrial Environmental Research Laboratory, Research
Triangle Park, North Carolina. Tests were conducted using a three-stage
turbulent contact absorber (TCA) with a typical gas flow rate (G) of 465
m /hr (0.1 MW) and absorbing slurry chloride ion (Cl~) concentrations
ranging from 160 to 180,000 ppm.
The PGD processes investigated include conventional lime/limestone,
magnesia enhanced limestone, and _limestone with two-tank forced oxidation.
Data indicate that the effects of Cl on the performance of the absorber are
a function of the cations associated with Cl and scrubber operating
conditions. The accumulation of calcium chloride caused decreased system pH
and SIX removal efficiency, occasional decreases of slurry settling rate,
and increased gygsum scaling potential. When magnesium was the cation, the
increase of Cl concentration improved SCL removal efficiency at Cl~
concentrations below 40^000 ppm. No significant effects were observed using
sodium chloride at Cl concentrations less than 50,000 ppm. However, when
Cl concentrations were greater than 70,000 ppm, S02 removal efficiency
and system pH declined with the accumulation of either magnesium or sodium
chloride. Significant decreases in SCL removal efficiency were also
observed when lime was used in the natural oxidation mode with high inlet
SCL concentrations. Calcium chloride had minor effects on the performance
of a DBA enhanced limestone scrubber. Most gypsum specifications required for
wallboard manufacturing were met by washing cake using a pilot belt filter.
7-37
-------
ACKNOWLEDGEMENTS
The authors wish to express sincere gratitude to the following
individuals for their continued support and assistance throughout the test
program:
Michael A. Maxwell - Environmental Protection Agency
Gary Rochelle - University of Texas
Even Bakke - Peabody Process Systems, Inc.
Professor Rochelle was retained as a technical consultant for portions of
this project.
Coauthor Mobley was the EPA project officer.
INTRODUCTION
High chloride ion (Cl ) concentrations in the scrubbing liquor are
possible with tightly closed water loops in flue gas desulfurization (PGD)
systems. If closed-loop operation is assumed (i.e., the only water leaving
the PGD system is through evaporation and filter cake moisture), material
balance calculations, for high chloride coals, indicate total dissolved solids
(TDS, usually the combination of calcium, magnesium, and sodium salts) can
accumulate to levels exceeding 50,000 ppm* (1,2). Forced oxidation
intensifies the TDS accumulation through improved cake dewatering, and
dissolved solids can reach worst case levels in excess of 150,000 ppm (1,2).
The origin of the chlorides can be from the coal, as HC1 in the flue gas is
absorbed, or from chlorides in the makeup water. The latter is especially
important if cooling tower blowdown (CTB) is utilized for makeup water. In
order to evaluate the effects of Cl on PGD system performance, a test
program was developed to conduct pilot plant tests under simulated high Cl
conditions.
This paper presents an overview of a pilot plant test program under the
sponsorship of Peabody Process Systems, Inc., and the U.S. Environmental
Protection Agency's (EPA) _Industrial Environmental Research Laboratory
(IERL-RTP) to test the Cl effects on wet PGD system performance covering
the period from September 1981 through June 1983. For a thorough coverage of
the test work, theory, and results, refer to the final report (3). The test
One ppm is equivalent to one mg/1.
7-38
-------
facility is located at Research Triangle Park, North Carolina.
Corporation is the major contractor and facility operator.
Acurex
TEST FACILITIES
Two process configurations, natural and forced oxidation, shown
respectively in Figures 1 and 2, were employed for the tests. The scrubber,
located at EPA/IERL-RTP, is a three-stage turbulent contact absorber (TCA)
with 465 m /hr flue gas capacity (0.1 MW). No flyash was present in the
flue gas, which was drawn from a gas-fired boiler and injected with pure SCL
and HC1 as required. The oxidizer consisted of a 30 on diameter tower
containing slurry at a depth of 5.5 m and was sparged with air from the tower
bottom. A bleed stream of the slurry from the hold tank was directed to the
clarifier and then processed by a rotary drum vacuum filter to remove the
precipitated waste slurry. All filtrate was returned to the scrubber in order
to maintain closed-loop operation. The system was manually controlled in a
feed forward mode by maintaining a constant limestone stoichiometric ratio of
moles of reagent added per mole of SCL absorbed (when lime was used,
constant pH control was utilized). A small amount of excess reagent was also
added to neutralize absorbed HCl gas during some tests.
To Stack
Scrubber
Limestone
Flue Gas
Additives
Filter
Figure 1. Flow diagram for single-loop natural oxidation tests in
the 0.1 MW limestone pilot plant.
Granulated or powdered calcium chloride, magnesium chloride, or sodium
chloride was added to the liquid inventory in order to bring the system to the
desired Cl~ level. HC1 was (at specified times) injected into the inlet
7-39
-------
duct at predetermined levels in order to simulate flue gas generated by the
combustion of high chloride coal. The limestone feed rate was adjusted
manually by a trial and error method until steady-state stoichiometric ratios
were achieved. Normally, each run consisted of at least 25 hours of
continuous stable operation, after which scrubber performance and system
conditions at each Cl level were established by averaging operational data
logged hourly.
The gas phase SCL concentration was monitored continuously using a
DuPont 400 S02 analyzer and occasionally checked by wet gas titration.
Liquid and solid samples of important species such as calcium, sulfite,
sulfate, carbonate, magnesium, and sodium were analyzed.
To Stack
-1
__»
Lim
Additives
estone
Make-
up
Tank
1
r
Hold
Tank
Filter
Sump
Tank
Air
Oxidizer
Figure 2. Flow diagram for in-loop forced oxidation tests in
the 0.1 MW limestone pilot plant.
7-40
-------
METHOD OF APPROACH
The effect of Cl on the FGD system was studied in three ways:
o All system parameters were held constant while the concentration of Cl~
in the scrubbing liquor was increased.
• At a given Cl~ concentration, either the liquid-to-gas ratio (L/G) or
the limestone feed rate was increased until the base case S0_ removal
efficiency was obtained.
• The Cl concentration was held constant while system parameters were
varied individually.
For the third case, comparisons of the effects on S0~ removal are
difficult since the base case S02 removal changes significantly when L/G,
limestone stoichiometry, or the absorber packing height are changed. Laslo
and Bakke (4) outlined a method for comparing S02 removal efficiencies based
on the logarithmic ratio of the fractional S02 remaining in the flue gas at
high Cl ,_ f , to the fractional SO- remaining at the base case conditions
with no Cl ,
Mathematically, this ratio, R, can be expressed as a percentage:
_ ln( 1- f ) 0 (1)
R ~ x 100-°
For all other tests, evaluation of effects was performed by examining
system performance with and without Cl addition or by quantitatively
determining the L/G or excess limestone required to return to the base case.
Major areas of testing during the reporting period are summarized below:
« Cl~ effects on the limestone FGD system utilizing the following salts
alone or in combination:
- Calcium chloride
- Magnesium chloride
- Sodium chloride
7-41
-------
• Evaluation of methods to regain base case S02 removal efficiency at high
calcium chloride concentrations including:
- Increasing L/G
- Increasing limestone feed rate
- Adding organic acid
• Effects of calcium chloride on a lime based FGD system
« Effects of calcium chloride on a magnesia enhanced limestone PGD system
» Effects of calcium chloride on an organic acid enhanced limestone PSD
system
• Washability evaluation of a gypsum cake using a horizontal belt filter
Each of these will be covered separately in the following text. By far,
most testing was performed in a limestone based system using calcium chloride
and thus will be treated in greater detail.
RESULTS AND DISCUSSION
Cl~ EFFECTS ON LIMESTONE PGD SYSTEMS
Calcium Chloride
Recycle slurry pH decreased as the concentration of calcium chloride
increased. It has been shown that the change in pH closely follows the
calculated equilibrium H concentration in the natural oxidation mode of
operation (1). The drop in pH in the forced oxidation mode was not
significant until the Cl concentration exceeded 50,000 ppm. Calculating
the H concentration during forced oxidation was not possible due to the
stripping of C02 gas from the slurry. At the same Cl concentration, the
pH decrease in the natural oxidation mode was significantly more than
operation in the forced oxidation mode.
A decrease in S0» removal efficiency was observed with the increase of
calcium chloride concentration for both the natural and forced oxidation
modes. As shown in Figure 3, the decrease was not as severe for the forced
oxidation mode.
The natural oxidation data in Figure 3 was obtained using a high base
case pH of 6.0 and a high limestone stoichiometry of 1.3-1.4 moles of CaCO.,
7-42
-------
per mole of S02 absorbed. Additional tests were performed at a lower base
case pH of 5.4-5.6, summarized in Table 1, and a series of tests run at high
and low make-per-pass, summarized in Table 2. The make-per-pass, or the
amount of SCL gas absorbed per liter of slurry, was minimized by simply
removing the spheres from the TCA.
90
so
8?
5 70
UJ
60
SO
40
30
G: 465 mVhr
UG: 8.0 l/m1
Inlet SO,: 2500 ppm
Limestone Stoichiometric Ratio 1.2 (forced oxidation)
1.35 (natural oxidation)
Forced Oxidation
CaCI2
a Original Data
• Replicates
Natural Oxidation
30 60 90 120
CI~Concentration, ppmx 10 '
150
180
Figure 3. Comparison of natural and forced oxidation S02
removal efficiencies as a function of Cl .
Table 1. RESULTS OF LOW BASE CASE pH, NATURAL OXIDATION TESTS
Test I.D.
NB-13
N2-4
N5-10
NB-14
N2-HM1
N5-HM5
Inlet
SO.
(ppffl)
2500
2500
2500
700
700
700
Cl
Cone.
(ppm)
*
20,000
50,000
*
20,000
50,000
Limestone
PH
5.6
4.9
4.6
5.4
5.2
5.0
Stoich.
(molar)
1.16
1.13
1.16
1.11
1.07
1.13
L/G.
(1/nO
8.0
8.0
8.0
3.0
3.0
3.0
S00
Removal
(%)
68
65
60
77
75
68
R
from Eq.
(%)
100
92
80
100
94
77
(1)
* Base Case
7-43
-------
Table 2. RESULTS OF HIGH/LOW MAKE-PER-PASS NATURAL OXIDATION TESTS
Test I.D.
Inlet
!°2
, (ppfi)
Cl
Cone.
(ppm)
pH
Limestone
Stoich. L/G.,
(molar) (1/nT)
so2
Removal
(%)
R
from Eq. (1)
(%)
High Make-per-pass
NB-8
N5-HM1
NB-9
N5-HM3
2500
2500
500
500
*
50,000
*
50,000
5.8
5.0
5.9
5.1
1.2
1.2
1.1
1.2
11.2
11.2
7.2
7.2
89
81
92
84
100
75
100
73
Lew Make-per-pass
NB-10
N5-LM1
NB-11
N5-LM3
2500
2500
500
500
*
50,000
*
50,000
5.8
4.9
5.9
5.4
1.2
1.1
1.1
1.2
10.7
10.7
4.5
4.7
56
47
52
43
100
77
100
77
* Base Case
Since it is difficult to compare tests at different base case conditions,
R from Equation (1) was calculated for each test and is presented in both
Tables_l and 2. Figure 4 summarizes the values for R plotted as a function of
the Cl concentration for forced oxidation and the high base case pH natural
oxidation data from Figure 3, as well as the data from_Tables 1 and 2. It is
apparent that R has a linear correlation with the Cl concentration and is
independent of inlet SO concentration, make-per-pass, liquid holdup in the
absorber, and liquid-to-gas ratio. Figure 4 also indicates that the intial
base case pH is significant, since tests run with an initial pH of 6.0 (or a
limestone stoichiometry of 1.3-1.4) have a slope significantly steeper than
the tests run with an initial pH of 5.5-5.9 (or a limestone stoichiometry of
1.1-1.2).
In addition to decreases in SO- removal efficiency and pH, occasional
decreases in the solids settling rate were observed in the natural oxidation
mode at high Cl . Little effect of Cl on solids settling was observed
during the forced oxidation mode. Also, with the accumulation of calcium
chloride, the relative saturation of gypsum in the scrubber system increased
to levels conducive to scaling; however, scale was rarely observed in the
pilot plant.
7-44
-------
100
80-
60-
40-
20-
A ^ Forced Oxidation
I
Natural
Oxidation
30
—i 1 1—
60 90 120
CrConcentration, ppm x 10'1
Key to Test ID'S
O NB1 through N10-1
a N5-LM1
x N5-LM3
? N5-HM1
O N5-HM3
A FB3 through F18-1
• N2-4, N5-10
* N2-HM1, N5-HM5
Limestone
Stoichiometry
1.3-1.4
1.1
1.2
1.1
1.2
1.2
1.1-1.2
1.1
150
180
Figure 4. R, from Equation (1), vs. Cl~ concentration.
Magnesium Chloride
When using magnesium chloride as the source of Cl , a drop of system pH
was observed at a rate similar to that obtained during calcium chloride
testing, as the Cl concentration increased. Figure 5 shows pH data for all
tests as a logarithmic function of the sum of calcium and magnesium cations.
It is apparent that the magnesium ion and the calcium ion behave similarly in
affecting the equilibrium pH. The_accumulation of magnesium chloride improved
SCL removal efficiency when the Cl concentration was below 40,000 ppm, as
shown in Figure 6. However, a significant decrease in SCL removal
efficiency occurred at Cl concentrations greater than 50,000 ppm. Figure 6
also shows there was no effect of Cl when a 50% molar mixture of magnesium
chloride and calcium chloride was used — indicating that the positive effect
of magnesium was cancelled by the negative effect of calcium chloride.
As shown in Figure 7, during forced oxidation, no significant effects of
magnesium chloride on SCL removal efficiency were observed until the Cl
concentration exceeded 70,000 ppm, after which a slight drop in efficiency
occurred. Also shown in Figure 7 is a more severe effect of the mixed calcium
and magnesium salts due to the lack of magnesium enhancement during the forced
oxidation mode.
7-45
-------
0.
JC
•D
0
6.0
5.9
5.8
5.7
5.6
5.5
5.4
5.3
5.2
5.1
5.0
4.9
4.8
4.7
4.6
•
A A
.
O
A OD
.
Co
•
Natural Oxidation • A»
Limestone
ft
• CaCh •
' D MgCU °
• MgCh + CaCh
& NaCI
. A NaCI + CaCh .
o Magnesia Enhanced
• D
10
50 100 500
[Ca++] + [Mg++],mM
1000
Figure 5. Comparison of hold tank pH's as a function of
Mg and Ca cations.
Ui
I
0)
cc
90
80
o 70
60
50
40
30
Natural Oxidation
G: 465 ri^/hr
L/G:8.0 l/m3
Inlet SO:: 2500 ppm
MgCb
30
60 90 120
Cl~ Concentration, ppm x 10
150
180
Figure 6. Conparison of natural oxidation S02 removal
efficiencies as a function of Cl .
7-46
-------
90
iS 80
o
c
0)
o
0)
cc
70
60
50
40
30
CaCb
Forced Oxidation
G: 465 m]/hr
L/G: 8.0 l/m3
Inlet SCh: 2500 ppm
MgCb
MgCU
+ CaCh
30 60 90 120
Cl~ Concentration, ppm x 10"
150
180
Figure 7. Comparison of forced oxidation SO- removal
efficiencies as a function of Cl~.
Sodium Chloride
In the natural oxidation mode, no significant effects on SO,, removal or
pH were observed with the accumulation of sodium chloride until the Cl con-
centration reached 50,000 ppm. The pH data are also plotted in Figure 5 which
indicates that the pH is a function of the cations Mg and Ca only. As
shown in Figure 8, when the Cl concentration exceeded 50,000 ppm, drops in
S0_ removal were observed. When forced oxidation was employed, as shown in
Figure 9, no effects on SO,, removal were observed until the Cl concentra-
tion exceeded 70,000 ppm.
METHODS USED TO COUNTERACT THE EFFECT OF Cl ON S02 REMOVAL EFFICIENCY
Increasing the L/G Ratio
Test results show that at 40,000 ppm Cl and a limestone stoichiometry
of 1.3, the L/G ratio would have to be doubled in order to regain the base
case SO- removal efficiency. Although never tested, the amount of L/G
increase should be substantially lower at lower limestone stoichiometries due
to the aforementioned reduced effects on
stoichiometries are utilized.
SO,, removal when low limestone
7-47
-------
G: 465 m'/hr
L/G: 8.0 l/m1
Inlet SO:: 2500 ppm
Natural Oxidation
A
LJJ
To
I
0)
K
30
60 90 120
Cl~ Concentration, ppm x 10
Figure 8. Comparison of natural oxidation S0« removal
efficiencies as a function of Cl .
90
80
70
LJJ
5 60
-------
Increasing the Limestone Feed Rate
Results of testing are summarized in Table 3. As can be seen, there is a
significant difference in the amount of excess alkali required to restore base
case SO- efficiency between the high and low limestone base case stoichio-
metry runs, but the quantity required in either case is unacceptably high.
Table 3. EXCESS LIMESTONE REQUIRED TO RESTORE BASE CASE S02 REMOVAL
Limestone Stoich-
Base Case
% SO,
Removal
80
89
56
92
52
Inlet
?2
(ppffi)
2500
2500
2500
500
500
Limestone
Stoich.
(molar)
1.3-1.4
1.16
1.19
1.14
1.22
Cl
Cone.
(ppm)
40,000
50,000
50,000
50,000
50,000
Packing iometry Required
to
Height Regain Base Case
(on) % S02 Removal (molar)
53 3.0
65 1.6
0 1.9
65 2.3
0 2.1
Addition of Organic Acid
A mixture_of organic acids, DBA, was added to the absorbing slurry at
50,000 ppm Cl to determine the amount of acid required to restore the base
case S02 removal efficiency. As seen in Figure 10, the amount required is
only 9.2 meq/1. Further testing indicated that DBA is very effective in
counteracting the effects of Cl with forced oxidation,_requiring only 9.4
meq/1 to counteract the effect of Cl at 160,000 ppm Cl .
EFFECTS OF Cl ON A LIME BASED FGD SYSTEM
The results of testing with lime are interesting in that the depression
in SO2 removal was identical to that of the limestone tests using high
limestone stoichiometry, as shown in Figure 11. The lime tests were run at
constant pH and a stoichiometric ratio approaching 1.0. Since pH control was
used, only the absorber effluent pH declined when the Cl concentration
increased. When the inlet S0_2 concentration was lowered to 500 ppm, there
was no noticeable effect of Cl on the SO_ removal efficiency up to
100,000 ppm Cl~.
7-49
-------
90
85
S? 80
5*
o
'5 75
UJ
o
»
6
65
60
55
50
G: 465 m'/hr
Cl~: 50,000 ppm
UG: 8.0 Km3
Inlet S02: 2500 ppm
L/SS.R:1.4
Base Case SO, Removal Efficiency
Natural Oxidation
Limestone
CaCI2
6 8 10
Total Organic Acid, meq/l
12
100 200 300 400 500 600 700 800 900
Equivalent Adipic Acid Concentration, ppm
1000
Figure 10. Effect of organic acid concentration on S02
removal efficiency at 50,000 ppm Cl~ and
natural oxidation.
EFFECTS OF CALCIUM CHLORIDE ON A MAGNESIA ENHANCED FGD SYSTEM
Addition of magnesia to a conventional limestone FGD system increases the
concentrations of two dissolved sulfite species, SO ~ and MgSO^, which
react with the absorbed SO and enhance S02 removal: The accumulation of
calcium chloride_ affects the magnesia enhanced limestone FGD system by
reducing both SO- and MgS03 through chemical equilibrium effects.
Pilot plant tests were conducted to evaluate the effects of calcium chloride
on the performance of a magnesia enhanced limestone FGD system. Results
indicate that when calcium chloride was initially added to the system, a sharp
drop of SO^removal efficiency and a slight decrease in system pH _were
observed^ The sharp drop in S02 removal reflected the reduction of SO.,"
and MgSO_ in the scrubbing liquor. As shown in Figure 12, further
increases in the calcium chloride concentration resulted in less of an effect
on S02 removal since the dissolved sulfite species have been depleted.
7-50
-------
80
o
§ 70
LU
15
S
a>
cc
60
50
CaCI2
Natural Oxidation
Lime/Limestone
G: 465 m'/hr
UG:8.0 l/m3
Inlet SO:: 2500 ppm
x Limestone
• Lime
30
60 90 120
Cl~ Concentration, ppm x 10"
150
180
Figure 11. Effect of Cl~ on SCL removal
efficiency in a line scrubber.
100
90
I
0)
cc
70
60
CaCI2
Natural Oxidation
Magnesia Enhanced Limestone
G: 435 m'/hr
L/G: 10.0 l/m3
Inlet SO:: 2500 ppm
Mg+ + : 2000 ppm
10 20 30 40
CI~Concentration, ppm x 10 '
50
Figure 12. Effect of Cl on SCL removal efficiency
in a magnesia enhanced scrubber.
7-51
-------
EFFECTS OF CALCIUM CHLORIDE ON A DBA ENHANCED FGD SYSTEM
The operation of a DBA enhanced limestone FGD system has the advantage of
high S02 removal efficiency, and low limestone stoichiometry. In order to
simulate the DBA enhanced limestone scrubber, the base case pH was controlled
at 5.2, limestone stoichiometry was less than 1.1, and S02 removal
efficiency was 90% for both natural and forced oxidation tests. Short-term
tests were utilized to evaluate the effects of Cl on system performance.
Long-term tests were conducted to evaluate DBA consumption and decomposition
SO., removal efficiency and
and forced oxidation modes. In order to
the base case SO,, removal efficiency, DBA concentrations had to be
at high Cl concentrations.
decreased in both the natural
maintain
increased as the Cl concentration increased, as shown in Figure 13. In
addition, plugging and scaling problems were encountered during the natural
oxidation mode:_ hard scale formed on the scrubber bottom internals while
operating at Cl concentrations above 50,000 ppm and a phenomenon similar to
"limestone blinding" was encountered while operating at a Cl concentration
of 120,000 ppm. There was no noticeable effect of high Cl on the
degradation rate of DBA, but an increase in the DBA coprecipitation rate was
observed at high calcium chloride concentrations during natural oxidation
testing as shown in Table 4.
1100
1000
T3
'o
D.
03
Q.
a.
CO
O
900
800
700
600
500
G: 465 m3/hr
L/G: 8.0 l/m1
Inlet SO:: 2500 ppm
Forced Oxidation
DBA Enhanced
Limestone
20 40 60 80
CI~Concentration, ppm x 10~3
100
Figure 13. DBA concentration required to reach 90% S02
removal at high concentrations of Cl .
7-52
-------
Table 4. COMPARISON OF UNACXXUNTABLE ORGANIC ACID LOSSES
AT THE RTF LIMESTONE PGD PILOT PLANT
Natural Oxidation Forced Oxidation
(g/hr) (g/hr)
DBA losses at low Cl7.4-8.014.4
concentration
.(less than 8,000 ppm)
DBA losses at high Cl" 10.4 13.9
concentrations*
* Cl_ = 30,000 ppm for natural oxidation
Cl = 50,000 ppm for forced oxidation
WASHABILITY. EVALUATION OF GYPSUM CAKE USING A PILOT HORIZONTAL BELT FILTER
The marketability of gypsum produced from a forced oxidized PGD system is
affected by its purity, especially with regard to dissolved solids. Gypsum
generated from high Cl FGD systems must be washed thoroughly to meet the
purity specifications of wallboard manufacturers. A belt filter, with the
potential of three-stage countercurrent washing, was utilized to determine the
washability of gypsum cake produced during the DBA testwork described
previously.
Results of these tests are shown in Figure 14 for both TDS and Cl~.
Also shown is the calculated theoretical washing efficiency obtained with
ideal washing. These data indicate that the pilot belt filter approaches
ideal washing. Chemical analysis of the resulting washed cake was performed
at three different system Cl concentrations. Results, shown in Table 5,
indicate that commercial grade gypsum suitable for wallboard manufacturing
could be produced from slurry containing up to 80,000 ppm Cl . Free
moisture content was slightly high, but optimizing moisture was not in the
scope of work.
7-53
-------
100
g 90
'jo
UJ
O)
_c
I/I
I
CO
O
80
70
Predicted by ideal countercurrent
washing model
O Washing efficiency based on Cl
A Washing efficiency based on
water soluble salts
1 2
No. of Washing Stages
Figure 14. Performance of a belt filter with countercurrent
washing and five wash displacements
Table 5. COMPARISON OF GYPSUM QUALITIES AFTER BELT FILTER WASHING
Free Water, %
Combined
Water, %
CaSO '1/2 H-0,
CaCof, %
CaO, %
so3, %
Na, ppm
K, ppm
Mg, ppm
Cl, ppm
TDS, ppm
Required
Quality*
<10
>19.8
% <2
<2
-31-33
>44
<75
<75
<50
<120
<600
Analysis
50,000
14.4
21.4
1
1
32.6
46.1
5
5
8.0
90.6
136.3
of Sample from Slurry with
ppm** 80,000
15.4
19.6
1
1
32.6
45.9
5
5
6.0
98.6
147.5
Cl of
ppm** 120,000 ppm***
15.7
20.7
2.1
3.7
33.8
45.0
5
5
8.0
292.4
451.8
* Specified by U.S. Gypsum for wallboard quality gypsum
** Washed with 5 displacement water and 3-stage countercurrent washing
*** Washed with 5 displacement water and 2-stage countercurrent washing
7-54
-------
CONCLUSIONS
• The S02 removal efficiency declined as calcium chloride accumulated in
the limestone PGD system in both the natural and forced oxidation modes of
operation. The decline was independent of absorber parameters such as
liquid-to-gas ratio, packing height, and inlet S02 gas concentration
within the ranges tested. Tests with natural oxidation and a base case pH
of 6.0 had a greater drop in S0_ removal than tests at a lower base case
pH or tests run with forced oxidation. Significant decreases in S0_
removal were also observed when lime was substituted for limestone in the
natural oxidation mode.
• The effect of Cl is cation dependent. Evidence _from conventional
limestone testing indicates that below 50,000 ppm Cl~, S0« removal is
not adversely affected when Mg or Na is the cation.
The system pH decreased as the Cl~ concentration increased when Mg
or Ca was the cation in both the natural and forced oxidation modes.
In the natural oxidation mode, pH was found to follow equilibrium H
concentrations. No significant change in system pH was observed when
Na was the cation.
• The accumulation of calcium chloride depressed S02 removal efficiency
which could be restored by adding excess limestone, small amounts of the
organic acid DBA, or by increasing the liquid-to-gas ratio.
• The accumulation of calcium chloride in a magnesia enhanced limestone PGD
system significantly reduced S02 removal efficiencies due to equilibrium
effects on dissolved alkalinity.
• The concentration of organic acid in a DBA enhanced limestone P3D system
had to be increased as the calcium chloride concentration increased in
order to maintain a constant S02 removal efficiency. An. increase in the
DBA coprecipitation rate was measured at high calcium chloride
concentrations, but no increase in the DBA degradation rate was measured.
• Gypsum, washed with a three-stage countercurrent belt filter, met each
specification required for wallboard manufacturing (except free moisture)
with feed slurry Cl concentrations as high as 80,000 ppm.
7-55
-------
REFERENCES
1. Chang, J.C.S., and D. Laslo, "Chloride Ion Effects on Limestone FGD
System Performance," "Proceedings: Symposium on Flue Gas Desulfurization,
Volume 1," pp. 224-254, EPRI CS-2897, March 1983. May 18, 1982.
2. Hargrove, O.W., et al., "The Effects of Process Water Selection on
Lime-Limestone Flue Gas Desulfurization Chemistry," Electric Power
Research Institute Report CS-2451, July 1982.
3. Chang, J.C.S.,"Pilot Plant Tests of Chloride Ion Effects on Wet FGD System
Performance," Draft Final Report, EPA/Acurex Contract No. 68-02-3648, June
1983.
4. Laslo, D., and E. Bakke, "The Effect of Dissolved Solids on Limestone FGD
Scrubbing Chemistry," Presented at the ASME 1983 Joint Power Generation
Conference, September 27, 1983.
7-56
-------
MODELING OF S02 REMOVAL BY LIMESTONE SLURRY
SCRUBBING: EFFECTS OF CHLORIDES
P.K. Chan, G. T. Rochelle
-------
MODELING OF SO REMOVAL BY LIMESTONE SLURRY SCRUBBING;
EFFECTS OF CHLORIDES
by: Pui K. Chan and Gary T. Rochelle
Department of Chemical Engineering
The University of Texas at Austin
Austin, Texas 78712
ABSTRACT
A model of limestone slurry scrubbing with staged contacting has been
developed by integrating gas/liquid mass transfer of SO , CO , and 0 and
dissolution of limestone and calcium sulfite solids. The model was used to
predict SO,, removal as a function of NaCl and CaCl^ accumulation in the
solution. Experimental data from three different pilot plants were accur-
ately simulated. Chloride accumulation reduces S09 removal by its effect on
the S0_ hydrolysis equilibrium. Calcium accumulation reduces SO removal by
its effect on S0_ and HCO. in the scrubber solution. Sulfite oxidation in
the scrubber can obscure the effects of CaCl~ accumulation. The model
predicts the effect of dibasic acid (glutaric) on SO. removal with solutions
containing 0.7 M CaCl-.
INTRODUCTION
Limestone slurry scrubbing is the most widely used flue gas desulfuriza-
tion process. Typically it includes a countercurrent spray scrubber with a
large slurry recycle from a hold tank and a slurry bleed to solids dewatering
and waste disposal. The scrubber is usually designed to include gas/liquid
mass transfer of S09, CO , and 0~, limestone (CaCO ) dissolution, and calcium
sulfite (CaSO») dissolution. The hold tank is designed for dissolution of
the rest of the CaCO,. and crystallization of CaSO. and calcium sulfate
(CaSO,). In some systems, oxidation of sulfite to sulfate is forced by
adding air to the hold tank.
BACKGROUND FOR SCRUBBER MODELING
Early effort in simulating this process was limited to a model based on
gas/liquid/solid equilibria (Lowell et al., 1970). Rochelle and King (1977)
integrated the SO. mass transfer rate over the changing gas-phase concentra-
tion in a column, assuming constant dissolved alkalinity and no SO- back-
pressure. Performance data obtained in 10-MW scrubbers have been correlated
by McMichael et al. (1976) and Head (1977).
Prepared for presentation at EPA/EPRI Symposium on Flue Gas Desulfurization,
New Orleans, November 1-4, 1983.
7-57
-------
Mehta (Mehta and Rochelle, 1983; Mehta, 1982) integrated SO gas/liquid
mass transfer and CaCO, dissolution as a function of changing gas and
solution composition across the length of a column. His model for S0_ mass
transfer included both gas- and liquid-phase resistance and simulated T.iquid-
phase mass transfer by approximate surface renewal theory with equilibrium
reactions (Chang and Rochelle, 1982a,b; Weems, 1981). CaCCL dissolution was
modeled by mass transfer with equilibrium reactions using film theory (Chan
and Rochelle, 1982; Toprac and Rochelle, 1982; Rochelle et al., 1983).
Solution equilibria were calculated by the Bechtel-modified Radian
equilibrium program (Epstein, 1975).
This work extends the effort of Mehta by adding gas/liquid mass transfer
of CO and 09 and CaSO, dissolution by mass transfer with equilibrium reac-
tions (Tseng"and Rochelle, 1983)'. The equilibrium procedures used by Mehta
have been simplified, and the model has been extended to include countercur-
rent staged contacting. The model used in this work does not include
simulation of the hold tank, and calculates SO. removal from a specified
scrubber inlet solids and solution composition.
BACKGROUND FOR EFFECTS OF CHLORIDES
Extensive experimental work has been performed on the effects of
chlorides on S0? removal in limestone slurry scrubbing. HC1 from flue gas
and chloride salts from makeup water can accumulate as dissolved CaCl ,
NaCl, or MgCl^ in the scrubber system. Pilot plant work on these effects
has been performed at four facilities:
3
1) EPA-RTP - Research Triangle Park, North Carolina, 7.8 m /min TCA
scrubber, work performed by Acurex and sponsored by EPA and Peabody (Chang
and Laslo, 1983; Acurex, 1983; Laslo et al., 1983).
3
2) Radian - Austin, Texas, 0.02 m /min bubbler, work performed by
Radian Corporation and sponsored by the Electric Power Research Institute
(Jarvis et al.,1983 a,b).
3
3) CE - Windsor, Connecticut, 34 m /min, spray tower, work performed
and sponsored by Combustion Engineering (Rader et al., 1982).
3
4) Shawnee - Paducah, Kentucky, 780 m /min spray/tray scrubber, work
performed by TVA, Radian, and Babcock and Wilcox (B&W) and funded by TVA,
EPRI, and B&W (Downs and Robards, 1983).
The data from these four sources are not in complete agreement on the
effects of chloride. With forced oxidation, 0.7 M CaCl (50,000 ppm Cl~)
reduced SO removal by comparable amounts at EPA-RTP (80-85 to 73%), Radian
(80 to 75%J, and Shawnee(66 to 58%). However, with natural oxidation, 0.7 M
CaCl reduced SO removal more at EPA-RTP (80 to 60%) than at Radian (75 to
70%) or Shawnee t65 to 54%) and had no effect at CE (50 to 50%).
This paper is an attempt to resolve these and other apparent discrepan-
cies in the experimental data by simulation of SO removal with the inte-
grated scrubber model. Most of the simulation has concentrated on the
7-58
-------
EPA-RTP and Radian data with effects of CaCl and NaCl. The Shawnee data
were unavailable at the time of this analysis. Data on the effects of
MgCl were not simulated because of a fault in the equilibrium data base for
Mg activity coefficients.
MODEL DESCRIPTION
The performance of CaCO slurry scrubbing has been modeled by numerical
solution of differential and algebraic equations representing three
processes of mass transfer enhanced by equilibrium reactions:
1) SO., C0_, and 0- absorption/desorption
2) Limestone (CaCO«) dissolution
3) . Calcium sulf ite (CaSO ) dissolution
These three processes depend on and affect the solution concentrations of
total S09, SO , CO , and Ca. Scrubber performance has been determined by
solution of the rate and material balance equations to give gas and solution
composition at every point in a staged scrubber.
SOLUTION EQUILIBRIA AND DIFFUSIVITIES
Rigorous solution equilibria were calculated once for the scrubber
inlet solution by the Bechtel-modif ied Radian Equilibrium Program (BMREQ)
(Lowell et al. , 1970; Epstein, 1975). These equilibrium results were then
simplified to eliminate ion pairs and incorporate activity coefficients into
the equilibrium constants. For example, the simplified equilibrium constant
for the reaction,
HSO~ * H+ + SO^
K = [H+][S03]/[HSO~] (1)
is given using concentrations by BMREQ at the ionic environment of a given
simulation,
] + [CaSO!] + [MgSO°]...)
K = - - =2— - ^ - (2)
([HS03] + [CaHS03]...)
With the simplified equilibria, the species SO. includes the concentrations
of SO., CaSO°, and other sulf ite ion pairs in the rigorous equilibria. The
simplified equilibria included only 13 solution species: H , OH , SO- ,
_ — _ — i
HSO S0~ CO , HCO , C0~, HA, HA , A~, HSO,, and S0~ In addition, Ca ,
Na , and Cl were included for charge balance, but participated in no solu-
tion equilibria. Mass balance in this system is represented by the
7-59
-------
components SO (SO + HSO + SO ), CO- (CO + HCO~ + CO ), S0~ (S0~ + HSO~),
HA (A~, HA~, HA; e.g., adipic acid), Ca++, Na , and Cl~. The use of
simplified equilibria is a good approximation to the extent that ionic
strength and Ca concentration do not change significantly across the
scrubber.
BMREQ was modified to include the solution species CaHSO. and NaHSO°
with the rigorous equilibria:
- = 0.073 (3)
aCaHSol"
NaHSO°
2.53 (4)
With these equilibria, the program accurately predicts the solubility of S09
gas at 1.8 x 10~ atm measured by Radian (Jarvis et al., 1983b) in NaCl and
CaCl solutions with up to 4 M chloride.
Based on data by Tseng and Rochelle (1983), the CaSO .^H 0 solubility
product has been reduced by 75% to give:
K r cr> = a a = 2.637xlO~7 M2 (5)
Sp'CaS°3 S0; Ca++
Based on data by Chan and Rochelle (1982), the effective solubility of
limestone is limited by sulfite with the apparent equilibrium,
? IRA
K rorn r en = a 4.4. a - a - = 8.95x10 M (6)
sp,CaC03.CaS03 Ca++ SQ- CQ-
Oxygen gas solubility was given by the equilibrium (Int. Crit. Tables):
H = P /[O ] = 1058.7 atm/M at 50°C (7)
2 2
A set of estimated diffusion coefficients was taken from Mehta and
Rochelle (1983). Dif fusivities for use with the simplified equilibria were
calculated as weighted averages based on concentrations from rigorous
equilibria. For example,
D =(simplified) = <£D.[j])/Z[j] (8)
so3 J
where j = S0~, CaSO°, MgSO°. Dif fusivities for use in gas/liquid mass
transfer were calculated as weighted averages of their square roots:
7-60
-------
D0.5
GAS/LIQUID MASS TRANSFER OF SO , CO , AND 0_
SO and 0 absorption and CO desorption/absorption were modeled by
two-film theory with an approximation of surface renewal theory to estimate
enhancement of liquid-phase diffusion by equilibrium reactions (Chang and
Rochelle, 1982a,b; Mehta and Rochelle,=1983) . The total flux of a component
(such as [S02]T = [S02] + [HSO~] + [S03]) in the liquid film is given in
approximate surface renewal theory by:
ko 05
Flux = -^ SD ' A[j] (10)
D J
DS°2
where /\[j] is the concentration difference of a single species across the
film, tj]± - [j]b.
The flux of SO through the gas film is given by:
Fluxso2 • ypso2 - Pso2jl> (11>
This must be equal to the flux of S0? plus sulfate through the liquid film:
VPSO -PSO? .> = -O ^A^.] + ^AtSO^.]) (12)
2 2)1 Dso2
where SO . includes S0? , HSO~, and S0~, and S0~ . includes S0~ and HSO~.
The solution and gas concentrations of SO- at tne-' gas/liquid interface must
be in equilibrium:
[SO ] = H P (13)
2 i bU2J,
The flux of CO is obtained from Eq. 10 with j including CO , HCO~, and
C0_. Gas film resistance is neglected for CO giving:
[C02]±- HPC()2 (14)
The finite-rate reaction of CO , H_0, H , and HCO is assumed to be at
equilibrium in the liquid film.
Absorption of 0 is assumed to occur with instantaneous oxidation of
sulfite at the gas/liquid interface. The rate of absorption is assumed to
be a constant factor, E, times the rate equivalent to physical absorption of
7-61
-------
oxygen. Therefore both oxygen absorption and sulfite oxidation are given as
the flux of the component sulfate (SO, and HSO,):
k° k?
Flux ^=J(D0.5 A [so'] +D°-5_A[HSO;]) = 2 E-^ D°'5 H P
D^ 02 02 02
(15)
The flux of other non-volatile components and charge is taken to be_
zero. Therefore, Eq. 10 gives relationships for adipic acid (H^A, HA , A )
and for all charged species where [j] is equal to the product or charge and
molarity.
When combined with solution equilibria for acid/base reactions, Eqs. 10
to 15 can be solved numerically to obtain bulk and interface compositions
and fluxes of CO,,, SO., and SO, (07). These calculations require specifica-
tion of bulk solution component concentrations and the system parameters k?,
k , and E, or an equivalent set of information.
O
SOLID/LIQUID MASS TRANSFER
Limestone and CaSO_ dissolution are both modeled as steady-state mass
transfer between the solid surface and the bulk solution (Chan and Rochelle,
1982; Mehta and Rochelle, 1983; Tseng and Rochelle, 1983). In this case the
flux of a component is proportional to the first power of the diffusivity,
and is given in terms of the film thickness as:
Flux = |ZD A[j] (16)
For limestone^ the rate of dissolution is the flux of Ca or of carbon-
ate components (C0~, HCO~). The reaction of H with HCO is assumed to be
negligible in the ooundary layer, so C0« is not includea as a diffusing
species. By material balance, the flux of Ca must be equal to the carbon-
ate flux. The fluxes of charge and of the components sulfite, sulfate, and
adipic acid are zero. The solution at the limestone surface is assumed to
be saturated to calcite, giving:
(17)
In the presence of sulfite, the solution at the limestone surface has the
apparent composition given by Eq. 6. In both cases the K values are
adjusted to reflect the effect of activity coefficients and ion pairs.
For CaSO_ dissolution, the dissolution rate is the flux of Ca which
must be equal to the flux of sulfite (S0~, HSO~, SO ) plus the flux of
sulfate (S0~, HSO~). By stoichiometry, tne ratio of sulfate and sulfite
fluxes must be equal to ratio of CaSO, and CaSO. in the hemihydrate solid
solution (up to 20% CaSO,). The fluxes of carbonate (C0_, HCO~), adipic
acid (A , HA , H_A), and total charge must be zero. At tne CaSD. surface,
7-62
-------
the solution is saturated with CaSO- as given by Eq. 5.
If the bulk solution composition is given, flux for limestone or CaSCL
dissolution can be found by numerical solution of the set of equations
representing the fluxes, solid/liquid equilibria, and acid/base solution
equilibria. The parameters CaCCL reactivity and CaSO reactivity are
specified to give the effective combination of solid surface area and film
thickness in the scrubber. Quantification of these parameters will be
described later. Both of the solid dissolution models permit apparent
crystallization if the driving force is negative. However, these results
are not rigorous, and care should be taken when using calculated results
when CaSO crystallization occurs in the scrubber.
SCRUBBER INTEGRATION
The results calculated in this paper assume a countercurrent scrubber
with several stages of contacting. In each contacting stage, the solution
is well-mixed and the gas is in plugflow. In a single stage the liquid
composition is equal throughout to the outlet liquid composition; therefore,
the rates of CaSO and CaCO dissolution are constant. However, the gas
composition varies, and the rate of gas/liquid mass transfer must be numer-
ically integrated.
With a guessed outlet solution composition and known outlet gas compo-
sition for a single stage, the inlet gas composition is determined by
numerically integrating the three differential equations:
dP k° (ZD?'5A[j])
dir = JL~-±5 (18)
8 kg Dso2
where i = SO , CO., and 0 (SO,). N is the number of gas-phase mass
transfer units ana can be defined by.
k a ZP
Ng= -i^-I (19)
For each step in the integration the values of ED." Afj] are calculated by
solving numerically the algebraic equations for fluxes and equilibria in
gas/liquid mass transfer.
For each stage in the scrubber the following algebraic equations were
solved numerically to determine the outlet liquid composition from a given
inlet liquid composition and outlet gas composition:
T,in) ' 2 % . - P0, > (20)
2, in 2,out
7-63
-------
([CO ] -[CO ] ) = (P - P ) + K N
G 2 I, out 2 T,in CO- . CO^ CaCO_ go
(22)
(23)
At each stage an initial guess of outlet composition was generated. Then
the gas-phase integration was performed and a library routine (IMSL/ZSPOW)
was used to solve the nonlinear Eqs. 20 to 23.
Since outlet gas composition is not usually known with a multistage
contactor, an additional convergence loop is needed. With a guessed value
of outlet S09 gas composition, each stage of the scrubber is solved from top
to bottom. 'The scrubber inlet S00 gas composition is compared with its
specified value, and the guess of outlet gas composition is updated. This
procedure continues until satisfactory convergence is achieved.
SOLIDS REACTIVITY PARAMETERS
The CaSO and CaCO reactivity parameters used in Eqs. 20 to 23 are
defined such that:
Kr rn = *T <24>
CaCO 6
where A is the solids surface area per volume solution, 6 is the apparent
film thickness, equal to particle radius for particles less than 10 ym in
diameter, and T is the slurry residence time (sec) in the scrubber per
transfer unit. These reactivity parameters can be derived from particle
size distributions or from batch dissolution data. Limestone reactivity is
given by:
KCaCO = ( - ^ ^~) CsT (25)
CaC°3 tzD.AtCO. .] S
J J > J
where C is the total calcium solids concentration in the slurry (M) and the
relativl reactivity (U/tZD. A [CO .]) is calculated by the method of Toprac
and Rochelle (1982) from batch dissolution data. Figure 1 gives the relative
reactivities of the limestones used at EPA-RTP and Radian for most of the
high chloride studies. These values were calculated from values measured by
batch dissolution at pH 4.0 with CO,, sparging (Radian) or at pH 5 with N?
sparging (EPA-RTP).
7-64
-------
? 10
-3
o
Kl
- 10
,-4
O
-------
RESULTS
INPUT SPECIFICATIONS
The following input data and parameters are needed to model a specific
scrubbing operation:
1. Gas and solution inlet composition
2. Liquid to gas ratio, L/G
3. Number of contacting stages and gas-phase transfer units, N
O
4. Ratio of liquid and gas film mass transfer coefficients, k°/k
x/ g
5. CaCO- and CaSCL reactivity, K_ rr. and'Kr
33 CaCO. CaSCL
6. Oxidation enhancement factor, E
Table 1 outlines the values of parameters used to model the Radian,
EPA-RTP, and CE data. Wherever possible actual pilot plant data for inlet
gas and solution compositions and for L/G were input to the model. N and
kVk were adjusted as necessary to simulate SO removal for each set? of
liquid and gas flowrates and scrubber intervals. Where possible, values of
CaCO., reactivity were selected that would be near the values estimated from
measured relative reactivity, scrubber liquid holdup, and slurry concentra-
tion. CaSO reactivity was chosen to be zero with forced oxidation, and up
to 10 sec/cm with natural oxidation. The oxidation enhancement factor was
adjusted to simulate observed levels of natural oxidation and varied from 1
to 7, consistent with kinetics observed by Ulrich (1983).
RADIAN
Figure 2 gives calculated SO removal for Radian results in the one-
stage bubbler with NaCl and CaCl . N was set at 2.2 to simulate observed
SO removal by Na?CO scrubbing. The parameter k^/k was estimated to be
0.28 atm/M from observed SO removal by HC1 solution^ however, an adjusted
value of 0.9 atm/M was necessary to simulate correctly SO removal with
CaCO slurry and forced oxidation. The value of kJJ/k estimated from
absorption in HC1 may be unreliable because of an unknown level of oxidation,
assumed to be. zero. .With forced oxidation, CaCO- reactivity was estimated
to be 2.3x10 sec/cm assuming 30 seconds liquia residence time in the
bubbler. With natural oxidation, the CaCO reactivity was reduced to 10
sec/cm to simulate SO removal accurately. This adjustment may reflect a
moderate level of limestone blinding. CaSO reactivity was also taken to be
1(T sec/cm . J
The level of natural oxidation observed in the Radian tests was 40 to
60%. This relatively high level of oxidation may occur in the hold tank or
in the scrubber itself. With no hold tank oxidation, an oxidation
enhancement factor of 2 to 4 was required to simulate observed levels of
7-66
-------
TABLE 1. EXPERIMENTAL CONDITIONS AND MODEL PARAMETERS
Radian
EPA-RTP
CE
Scrubber
Specification
Gas Rate
(m /min)
Scrubber Cross
Section (cm )
Gas composition (atm)
S02xlO
°22
Temperature (°C)
Number of stages
N
g
to?/k (atm/M)
-x, g
L/G (M"1)
E
Forced Oxidation
Natural Oxidation
Forced Oxidation
Natural Oxidation
Limestone
Bubbler
5.28 cm
Liquid Depth
0.018
11.4
2.1
0.12
0.01-0.07
50
1
2.2
0.9
0.25
2
2.3xl05
10
°5
1.4
Turbulent Contactor
Absorber, three 18-cm
beds, 4-cm spheres
7.8
410.0
0.5; 2.5
0.06
0.04-0.08
60
3
6.9
0.2
0.23
2
6xl03*5xlp4
5x10 ,10
£
1.2-1.4
Spray Absorber
Single Nozzle Level
4 spray levels
i
34
3716.0
0.5; 1.5
0.06
0.05-0.09
60
3
1.8
0.4
0.25
1,7
0
0
1.48-1.87
Stoichiometry
7-67
-------
solid oxidation. This enhancement factor also simulated the observed levels
of S09 removal in most of the experiments with clear solution scrubbing.
The results in Figure 2 were calculated with an oxidation enhancement
factor of 2.
As shown in Figure 2, the S0? removal for natural oxidation with NaCl
or CaCl9 and forced oxidation with CaCl is simulated well by these condi-
tions. However, with NaCl and forced oxidation the model predicts S09
removal 2 to 10% higher than the experimental values.
EPA-RTP
Simulated results with the 3-stage TCA at EPA's Industrial Environmental
Research Laboratory in Research Triangle Park (RTF), N.C., are presented in
Figures 3 and 4. The number of gas-phase mass transfer units was estimated
to be 6.9 by absorption of SO in Na CO., solution. The paramater k°/k was
found to be 0.2 atm/M by absorption of CO into Na CO_/NaHCO solution? An
oxidation enhancement factor of 2 was necessary to simulate observed levels
of natural oxidation. Assuming 7 seconds liquid residence time in the
scrubber, the estimated limestone reactivity was 2x10 sec/cm at 1.2
stoichiometry and 4x10 sec/cm at 1.4 stoichiometry. An even value of
5x10 sec/cm was used to simulate the forced oxidation data at 1.2
stoichiometry.
S09 removal with forced oxidation and CaCl? or NaCl was simulated well
by this set of parameters with chloride concentration greater than 20,000
ppm (Figure 3). However at low chloride it was necessary to reduce limestone
reactivity to 6x10 sec/cm in order to simulate SO removal. This reduction
is consistent with the tendency to get "limestone blinding" in systems with
forced oxidation at higher pH.
S09 removal with natural oxidation was overpredicted with the estimated
value or limestone reactivity. With an ordjer-of-magnitude lower reactivity
(10 sec/cm at 1.4 stoichiometry and 5x10 sec/cm at 1.2 stoichiometry),
the SO removal is simulated reasonably well over the entire range of
chloriae concentration (Figure 3). The predicted S09 removal at low
stoichiometry was less than at high stoichiometry primarily because of
reduced bicarbonate concentration in the scrubber inlet solution. CaSO«
[t t j
reactivity had little effect on S09 removal up to 10 sec/cm , and that
value gave the best fit of S09 removal over the entire range of chloride.
Figure 4 shows how the model simulates the observed effects of buffer
additive at 50,000 ppm chloride and 1.4 limestone stoichiometry. The
byproduct dibasic acid (DBA) used in these tests was mostly glutaric acid,
but it was modeled by assigning to it the same properties as adipic acid.
Assuming 1 mole of DBA as analyzed by the silicic acid method is equivalent
to 1 mole adipic acid, Figure 4 shows that the model overpredicts the effect
of DBA. However, with 0.65 moles adipic acid/mole DBA, the model predicts
accurately the effect of DBA.
7-68
-------
100
90
„ 8QI
3*
~ 70
o
s
LJ
CC
CM
O
CO
60
50
40
30
20
= 6xlO;
Oxidation Stoichiometry
Forced 1.2
Natural 1.4
Natural 1.2
30
60 90 120 150 180
CHLORIDE (g/Z)
Figure 3. Simulation of EPA-RTP Data on the Effects of
CaCly. Curves Calculated Using Conditions in
Table 1.
100
95
90
_ 85
~80
| 75
UJ
cc
(M
O
CO
70
65
60
55
50
o
1 Mole DBA =
1 Mole Adipic
1 Mole DBA =
0.65 Mole Adipic
I
_L
2 4 6 8 10 12 14
DIBASIC ACID (meq/J)
Figure 4. Effect of Dibasic Acid on S02 Removal with 0.7 M CaCl2,
Natural Oxidation, and 1.4 Limestone Stoichiometry (EPA-
RTP data), Curve Calculated Using Conditions in Table 1
(K = K = 104 sec/cm2 ) .
CaCO,
CaSO,
7-69
-------
COMBUSTION ENGINEERING (CE)
The simulation of CE data is given in Figure 5. The CE data were
collected in a spray tower with very short liquid-phase residence time.
Therefore, the CaCO. and CaSO reactivities were taken to be zero.
The simulated SO removal was insensitive to N and kp/k , but varied
significantly with the oxidation enhancement factor, E. N and k^/k were
arbitrarily set at 1.8 and 0.4 atm/M, respectively. The dara were simulated
best by E equal to 7, giving a calculated oxidation level near 100%. The
observed oxidation level was 60 to 90%, but these were short experiments and
the solids oxidation may not have reached steady state. Figure 5 shows that
the model simulates S0_ removal at both 500 and 1500 ppm S09 in the inlet
gas.
DISCUSSION
There are several effects evident from these theoretical and experi-
mental results. When chloride accumulates in scrubber solution, it
increases the ionic strength and thereby changes activity coefficients for
most solution species. If it shows up with Ca , it can also affect
solid/liquid equilibria directly. These chloride effects can interact with
the effects of bicarbonate accumulation and sulfite oxidation. The
effectiveness of buffer additives is also reduced by the pH effects of CaCl
accumulation. Limestone blinding does not appear to be caused by chloride
accumulation, but does affect interpretation of results with natural
oxidation.
EFFECTS OF IONIC STRENGTH
Chloride salts with any cations affect solution equilibria by changing
the ionic strength and resulting activity coefficients. The hydrolysis of
SO is one of the important equilibrium reactions enhancing SO. absorption
(Chang and Rochelle, 1981):
so2 + HZO t H+ +'HSO~
[H+] [HSOJ
~ (26)
h
so2
The apparent equilibrium constant for this reaction varies substantially
with ionic strength as shown in Figure 6. These results include the effects
of ion pairs and show somewhat improved equilibrium in CaCl. and MgCl,,
solutions. Figure 2 shows that NaCl accumulation reduces S09 removal in a
manner similar to its impact on K, .
EFFECTS OF CA++ CONCENTRATION
If Na or Mg is available in lower equivalent concentration than Cl ,
7-70
-------
20
0 20 40 60 80 100
CHLORIDE (g/£)
Figure 5. Simulation of CE Data, Curves Calculated
Using Conditions in Table 1.
0
Figure 6.
20
40 60 80 100
CHLORIDE (g/J)
120
Apparent Equilibrium Constant for S02
Hydrolysis in CaCl2 and NaCl Solutions.
7-71
-------
the chloride will accumulate in solution as Cad-. Ca has specific effects
on the solid/liquid equilibria and rate processes that determine the pH and
concentration of dissolved alkalinity (HCO,. and SO ) in the scrubber inlet
solution and the scrubber itself. S0_ absorption is enhanced by dissolved
alkalinity, so Ca can affect SCL removal.
The concentrations of SO and HCO are controlled by an approach to
equilibrium with CaSO.:
CaS03(s)
Ca
++
S0
[SO.] = K
sp.CaSO,
(27)
and the equilibrium with calcite:
CaCO (s) + CO + HO t Ca + 2HCO
[HCO.]2 = K . _
3 calcite
'CO,
(28)
Therefore, both [HCO.] and [S0_] should decrease as [Ca ] increases.
Table 2 shows the calculated dissolved alkalinity in the scrubber inlet
solution. Typically there is less than 2 mM_ dissolved alkalinity in_
experiments using forced oxidation, because S0_ is oxidized and CO-/HCO is
stripped by the sparging air. With natural oxidation the typical T.ever of
dissolved alkalinity was 2 mM. The only experiments giving higher levels of
dissolved alkalinity were those at EPA-RTP with less than 25 mM Ca and
high limestone stoichiometry (NB-1, 5.2 mM;-NN-1, 6.1 mM). The comparable
experiments at Radian gave only 3 mM alkalinity, probably because CO easily
escaped from the small hold tank, giving lower HCO alkalinity and only 0.06
atm CO. vapor pressure, compared to 0.25 atm at EPA-RTP. Furthermore, the
EPA-RTP experiment at low limestone stoichiometry also gave only 3.7 mM
alkalinity with less than 25 mM Ca , because of the low hold tank pH.
Therefore only the EPA-RTP experiments with natural oxidation show the
strong effects of CaCl accumulation on SO removal, because of the combined
effects on the SO hydrolysis reaction and the dissolved alkalinity.
The effect of Ca on dissolved alkalinity in the scrubber is more
difficult to quantify. However, the experiments with forced oxidation show
that Nad accumulation affects SO removal less than CaCl accumulation,
even though there is no. significant dissolved alkalinity in the inlet solu-
tion in either case. These results are consistent with a specific effect of
Ca on the dissolved alkalinity in the scrubber.
In general high dissolved alkalinity in the absence of chloride is most
likely to occur with high pH (high limestone stoichiometry) under conditions
7-72
-------
TABLE 2. HOLD TANK EQUILIBRIA
Run
No.
EPA-RTP
NB-1
Nl-1
N2-1
N3-1
N4-1
N5-1
N5-9
N10-1
NN-1
NN-3
NN-2
Mg-6
Mg-7
Mg-8
(mM)
Na+
(mM)
p.H
- Natural Oxidation
10
140
270
390
550
670
670
1400
23
24
350
360
46
280
-
-
-
-
-
-
-
-
710
1420
72.$
Mg
290
580
290
6.0
5.5
5.3
5.2
5.1
4.9
5.1
4.6
5.9
5.9
5.2
5.4
5.1
5.1
*
X
(atm)
- High
0.28
0.18
0.20
0.22
0.24
0.19
0.16
0.24
0.28
0.18
0.20
0.25
0.30
0.22
0
Saturation
CaSO CaCO
(MxlO U)
Alkalinity
Total 3
Limestone Stoichiometry
0.9
1.0
0.8
0.9
0.9
1.7
1.1
0.5
1.3
1.3
1.7
1.3
1.6
0.7
0.6
0.6
0.5
0.5
0.5
0.2
0.5
0.4
0.8
0.6
0.4
1.3
1.0
0.6
18
8
10
10
12
24
10 •
45
8
7
12
5
11
11
5.2
2.2
2.0
2.0
2.2
2.0
1.9
1.5
6.1
4.4
2.4
6.3
8.6
1.9
3.8
0.9
1.3
1.4
1.5
0.9
1.2
1.3
4.4
2.6
1.2
1.7
1.5
1.2
EPA-RTP - Natural Oxidation - Low Limestone Stoichiometry
LS-B
LS-2
LS-5
24
280
680
5.6
4.9
4.6
0.18
0.18
0.17
2.1
1.0
0.8
0.4
0.07
0.05
45
59
93
3.7
1.3
0.9
1.1
0.5
0.4
Radian - Natural Oxidation - High Stoichiometry
TDS-13
TDS-19
TDS-18
TDS-13
TDS-19
TDS-18
20
20
20
1880
2160
42
4330
6.3
6.2
6.3
4.9
4.9
5.8
0.08
0.08
0.06
0.06
0.05
0.07
1.2
1.0
1.1
0.7
-
1.4
1.8
0.9
1.3
0.9
0.9
0.6
1.8
3.6
1.9
8
9
8
3.6
2.7
2.7
1.2
-
1.6
2.3
1.6
.1.6
0.9
0.7
0.8
EPA-RTP - Forced Oxidation
FB-1
FB-2
F5-1
F8-1
FN-5
FN-3
16
12
672
1130
29
28
-
-
-
-
3130
2540
6.6
7.1
5.7
5.5
5.6
5.8
0.04
0.01
0.03
0.05
0.02
0.01
0.01
0.06
0.06
0.2
0
0
2.
3.
0.
2.
0.
0.
0
4
1
8
04
05
0
0
0
0
23
10
.7
.1
.6
.9
2.3
1.9
0.1
1.7
0.2
0.3
2.3
1.8
0.08
1.6
0.2
0.3
Radian - Forced Oxidation
TDS-14
TDS-17
TDS-14
TDS-17
18
18
2370
27
6.7
6.7
5.4
4270 6.2
0.02
0.02
0.01
0.01
-
0.08
0
0.06
2.6
2.2
3.3
6.1
0.4
0.4
0.9
0.1
1.4
1.2
0.6
0.6
1.4
1.2
0.6
0.6
CaS00 Saturation=a_, ++
J Ca
=/2.6xlO
-7
Saturation=a., ++
Ca
i.6xlO
-10
7-73
-------
where CCL is not lost from the hold tank. High dissolved alkalinity can
also result from excess Na or Mg in the scrubber system. With either of
these conditions there can be a dramatic decrease in S0« removal if an
accumulation of chloride causes Ca concentration to increase from less
than 25 mM to more than 100 mM.
Table 2 gives calculated thermodynamic parameters for EPA-RTP and
Radian scrubber inlet solutions. With natural oxidation, the scrubber inlet
solution is essentially saturated to CaSCL, with calculated CaSO«
saturations varying from 0.7 to 1.7. With nigh limestone stoichiometry
(1.4) and natural oxidation, the CaCO,. saturation of the scrubber inlet
solution is also nearly constant. It varies from 0.4 to 0.8 based on
calcite solubility. The calculated vapor pressure of CO varies from 0.18 to
0.28 atm and is also approximately constant. To the extent that CaCO_
saturation and the CO^ vapor pressure are constant, pH is given
approximately by the equilibrium:
CaCO. + 2H+ £ Ca++ + CO. a2 = K P* /[Ca++] (29)
3 2 R+ C02
As shown in Table 2, the ratio, a +/[Ca ] varies from 5x10 to 2.6x10
in EPA-RTP experiments at high limestone stoichiometry with natural oxidation
and CaCl or NaCl.
Table 2 includes data calculated from three runs with MgCl at EPA-RTP.
It appears that Mg has the same effect as Ca on the driving force for
limestone dissolution. The values of CaCO- saturation and a^+/[Ca ] were
calculated by substituting [(a ++) + (a^ ++)] and [Ca ] + [Mg ] for a ++
and [Ca ] , respectively. Witn this acrjustment CaCO« saturation and
a +/[Ca ] are in the same range as runs with CaCl .
n 2.
BUFFER ADDITIVES
Additives such as adipic acid enhance SO. removal with or without high
chlorides by buffering between the inlet and outlet pH of the scrubber, or
between the bulk solution pH and the pH at the gas/liquid interface. Because
Ca depresses scrubber inlet pH and bulk solution pH, a larger concentra-
tion of adipic acid (or DBA) is required to provide equivalent buffer
capacity and S0_ removal with an accumulation of CaCl . However, as shown
in Figure 4, DBA is still effective in enhancing SO removal.
LIMESTONE REACTIVITY
Forced oxidation was observed to enhance SO removal by 5 to 15% in the
EPA-RTP data, 3 to 7% in the Radian data, and hardly at all in the Shawnee
data. These observations have no major impact on the interpretation of the
chloride effects, but pose some difficult questions about the understanding
of general scrubber performance. From the EPA-RTP data, it would appear
that limestone reactivity is substantially depressed with natural oxidation,
since the SO removal with natural oxidation ia best simulated by reducing
the limestone reactivity a factor of 10 to 5x10 . The Radian data show less
7-74
-------
of an effect, requiring a reduction of a factor of 2, but were collected in
a contactor with a liquid-phase residence time three times greater than in
the EPA-RTP TCA scrubber. The Shawnee data were collected at 1.1 limestone
stoichiometry, where limestone reactivity with natural oxidation is negli-
gible in any case, so forced oxidation would not be expected to enhance SO.
removal.
SCRUBBER OXIDATION
In all cases an oxidation enhancement factor of 2 to 7 was required to
simulate the observed levels of scrubber oxidation. This means that 0.
absorption in the scrubber must be enhanced by its fast reaction with
dissolved SO species in the liquid-phase boundary layer. This reaction is
probably catalyzed by Fe and/or Mn (Ulrich, 1983; Ulrich et al., 1983).
The calculated results for the CE experiments demonstrate that levels
of scrubber oxidation near 100% can substantially enhance S0_ removal and
may cancel out the effects of chloride accumulation. High scrubber oxidation
serves to reduce total dissolved sulfite in the scrubber outlet solution and
permit satisfactory SO removal at lower scrubber outlet pH.
CONCLUSIONS
1. The integrated scrubber model is able to predict relative effects of
NaCl, CaCl , and dibasic acid on SO removal.
2. Chloride accumulation reduces S00 removal by reducing the apparent
equilibrium constant for SO hydrolysis.
3. Ca accumulation reduces S0? removal by reducing the concentration of
HCO_ and S0_. The effect of CaCI^ accumulation on S0« removal is greatest
in systems that start with a high level of HCO_ and S0\. The HCO- concentra-
tion can depend on CO loss from the hold tank.
4. Scrubber oxidation approaching 100% can have a significant impact on SO
removal.
5. Oxygen absorption in typical scrubbers is 2 to 7 times faster than
simple physical absorption because of fast reaction with dissolved SO .
6. Limestone reactivity in the scrubber can be lower in systems with
natural oxidation than in systems with forced oxidation.
ACKNOWLEDGEMENTS
This paper was prepared with the financial support of EPA Purchase
Order No. 2D5065NASX.
The modelling work was financially supported by EPA and EPRI under the
direction of J. David Mobley and Dorothy Stewart, respectively. The simula-
tions would not have been possible without detailed data provided by James
Jarvis (Radian), John Chang (Acurex), and Phil Rader (Combustion Engineering).
7-75
-------
NOTATION
a. - activity of species j
23
a - interfacial area, cm /cm
2 3
A - solids surface area per volume solution, cm /cm
C - concentration of calcium solids in feed, M
S 2
D - diffusivity, cm /sec
E - oxidation enhancement factor
2
2
Flux - Flux, gmol/cm -sec
G - gas flow rate, gmol/sec-cm
H - Henry's Law constant, M/atm
I - ionic strength, M
j - general solution or solid species
K - equilibrium constant
2
Kr rn - limestone reactivity, sec/cm
n
2
K - calcium sulfite reactivity, sec/cm
-
2
K . - solubility product for solid j , M
SP»J 2
k - gas-phase mass transfer coefficient, gmol/sec-cm -atm
6 2
k° - liquid-phase mass transfer coefficient, liter/sec-cm'
2
L - liquid flow rate, liter/sec-cm
M - molarity, gmol/liter
N - number of gas phase transfer units - k a ZP /G
& O
P - partial pressure, atm
P - total pressure, atm
pH - negative logarithm of H activity
T - liquid holdup, sec/N
o
t - residence time of solid particle in scrubber, sec
U - utilization of limestone
Z - height of scrubber, cm
6 - film thickness, cm
A - difference between bulk and interface or bulk and surface
quantities
[ ] - concentration, M
7-76
-------
Superscripts:
0 - ion pair, degrees
Subscripts:
i - interface (gas/liquid)
b - bulk
s - surface of solid
in - inlet to scrubber
out - outlet of scrubber
T - total quantity of a component
LITERATURE CITED
Acurex Corporation, "Pilot Plant Tests of Chloride Ion Effects on Wet FGD
System Performance," Draft Final Report, EPA Contract No. 68-02-3648,
June 1983.
Chan, P. K., and G. T. Rochelle, "Limestone Dissolution: Effects of pH, CO
and Buffers Modeled by Mass Transfer," ACS Symp. Ser. , 188. 75 (1982).
Chang, C. S., and G. T. Rochelle, "SO Absorption into Aqueous Solutions,"
AIChE J., J27_, 292-98 (1981).
Chang, C. S., and G. T. Rochelle, "Effects of Organic Acid Additives on S0?
Absorption into CaO/CaCO Slurries," AIChE J., 28, 261 (1982a).
Chang, C. S., and G. T. Rochelle, "Mass Transfer Enhanced by Equilibrium
Reactions," Ind. Eng.,Chem. Fund., 2^, 379-385 (1982b).
Chang, J.C.S., and D. Laslo, "Chloride Ion Effects on Limestone FGD System
Performance," in EPRI CS-2897, pp. 224-254, March 1983.
Downs, W., and R. F. Robards, "B&W - TVA High Chloride Test Program,"
presented at the EPA/EPRI Symposium on Flue Gas Desulfurization, New
Orleans, November 1-4, 1983.
Epstein, M., "EPA Alkali Scrubbing Test Facility: Summary of Testing
Through October 1974," EPA-650/2-75-047 (PB 244-901) (1975).
Head, Harlan, "EPA Alkali Scrubbing Test Facility: Advanced Program, Third
Progress Report," EPA 600/7-77-105 (PB 274-544) (1977).
Jarvis, J., et al., Draft Final Report on High Chloride Work, EPRI 1983b.
Jarvis, J., D. A. Stewart, and T. Trofe, "Effects of High Dissolved Chloride
on Bench Scale FGD Performance," presented at the EPA/EPRI Symposium on
Flue Gas Desulfurization, New Orleans, November 1-4, 1983a.
Laslo, D., J.C.S. Chang, and J. D. Mobley, "Pilot Plant Tests on the Effects
of Dissolved Salts on Lime/Limestone FGD Chemistry," presented at the
EPA/EPRI Symposium on Flue Gas Desulfurization, New Orleans, November
1-4, 1983.
Lowell, P. S., D. M. Ottmers, K. Schwitzgebel, T. I. Strange, and D. W.
DeBerry, "A Theoretical Description of the Limestone Injection-Wet
Scrubbing Process," _!, U.S. Environmental Protection Agency, APTD 1287,
PB 193-029 (1970).
7-77
-------
McMichael, J. W., L. S. Fan, C. Y. Wen, "Analysis of Sulfur Dioxide Wet
Limestone Scrubbing Data from Pilot Plant Spray and TCA Scrubbers,"
Ind. Eng. Chem. Proc. Des. Dev., 15, 459, (1976).
Mehta, R. R., and G. T. Rochelle, "Modeling of SO- Removal and Limestone
Utilization in Slurry Scrubbing with Forced Oxidation," presented at
the AIChE National Meeting, Houston, March 27-31, 1983.
Mehta, R. R., "Modeling of SCL Removal and Limestone Utilization in Slurry
Scrubbing with Forced Oxidation," M.S. Thesis, University of Texas,
Austin, TX (1982).
Rader, P. C., D. C. Borsare, and D. Frabotta, "Process Design of
Lime/Limestone FGD Systems for High Chlorides," presented at Coal
Technology '82, Houston, December 7-9, 1982.
Rochelle, G. T., and C. J. King, "The Effect of Additives on Mass Transfer
in CaCO or CaO Slurry Scrubbing of SO from Waste Gases," Ind. Eng.
Chem. Fund., j^, 67 (1977).
Rochelle, G. T., P. K. Chan, and A. J. Toprac, "Limestone Dissolution in
Flue Gas Desulfurization Processes," EPA 600/7-83-043 (PB 83-252-833)
(1983).
Toprac, A. J., and G. T. Rochelle, "Limestone Dissolution in Stack Gas
Desulfurization Processes - Effect of Type and Grind," Env. Prog.,
J.(l), 52 (1982).
Tseng, P. and G. T. Rochelle, "Dissolution Rates of CaSO ," paper in
preparation, 1983.
Ulrich, R. K., "Sulfite Oxidation under Flue Gas Desulfurization Conditions:
Enhanced Oxygen Absorption Catalyzed by Transition Metals," Ph.D.
Dissertation, University of Texas at Austin, December 1983.
Ulrich, P. K., R. E. Prada, and G. T. Rochelle, "Oxygen Absorption into
Bisulfite Solutions Containing Single and Synergistic Metal Catalysts,"
presented at the AIChE National Meeting, Houston, March 27-31, 1983.
Weems, W. T., "Enhanced Absorption of Sulfur Dioxide by Sulfite and Other
Buffers," M.S. Thesis, University of Texas at Austin (1981).
7-78
-------
INFLUENCE OF HIGH DISSOLVED SOLIDS ON PRECIPITATION
KINETICS AND SOLID PARTICLE SIZE
F. B. Meserole, T. W. Trofe, D. A. Stewart
-------
INFLUENCE OF HIGH DISSOLVED SOLIDS ON PRECIPITATION KINETICS
AND SOLID PARTICLE SIZE
by: Frank B. Meserole and Timothy W. Trofe
Radian Corporation
Austin, TX 78766
Dorothy A. Stewart
Electric Power Research Institute
Palo Alto, CA 94303
ABSTRACT
This paper presents results of a study to screen the effects of high
concentrations of dissolved ions on the precipitation of calcium sulfate
dihydrate, gypsum, and the solid solution of sulfate with calcium sulfite
hemihydrate. A series of precipitation measurements were conducted in the
presence of combinations of magnesium, sodium, calcium, chloride, and sulfate
at total dissolved solids levels up to 240,000 mg/L.
Significant differences in the precipitation rates and habit and size of
the precipitated solids were observed for several of the test solutions as
compared to precipitation from dilute solutions. Gypsum precipitation rate
in high TDS solutions was accelerated in high TDS solutions, especially those
containing chloride ion. The calcium sulfite-sulfate hemihydrate solid
solution precipitation rate was faster in sulfate ion solutions.
These results suggest that the operation of FGD systems at high
dissolved solids concentrations can alter the precipitation kinetics.
Attempts to model these effects will require the incorporation of the
concentrations of specific ions in the kinetic relationships.
INTRODUCTION
Sulfur dioxide (S02) is removed from flue gases of coal fired power
plants by a variety of commercial processes. At present, the lime and
limestone flue gas desulfurization (FGD) processes are the most advanced of
the throw away FGD systems (I_»j2,_3)» These systems remove S02 from flue gas
and convert it to calcium sulfite and calcium sulfate sludge. A calcium
sulfite-calcium sulfate sludge is usually produced in natural oxidation
systems, while forced oxidation systems produce primarily calcium sulfate
sludges.
The chemical reaction kinetics which produce these solids and the
solids' respective physical characteristics are especially important in,
designing and operating FGD systems. The rate at which calcium sulfite
7-79
-------
and calcium sulfate precipitate is important in controlling scaling on
scrubber surfaces and in designing rection tanks (.4».5)« Also, precipitation
rates are important in determining the particle size distribution of the
respective sulfur sludge. Particle size distribution has been shown to be a
major factor affecting the settling and dewatering properties of calcium
sulfite and calcium sulfate sludges. The poor dewatering properties of
calcium sulfite and, to a lesser extent, calcium sulfate are well documented
and are responsible for most of the problems associated with sludge disposal
in FGD systems.
Recently, more emphasis has been placed on water conservation and
multiple water usage in power plant design and operation. As a result, it is
becoming more common for makeup water to the FGD system to come from such
sources as cooling tower blowdown, and other power plant wastewater streams.
This operating approach has increased the total- dissolved solids (TDS) in
some FGD liquors to levels greater than 100,000 mg/L. The increased
concentration of total dissolved solids in the FGD system has raised
technical questions concerning the application of design and operating
criteria for FGD systems gathered at low TDS levels to these new high TDS
operating systems. Especially, it has not been determined what effects TDS
has upon the precipitation kinetics of calcium sulfite and calcium sulfate.
A major difference in precipitation rate may have a damaging effect (causing
severe scaling) or it could be beneficial (smaller reaction tank design) to
the system design. In addition, there is a need to determine whether the
chemical composition associated with high TDS scrubber liquors influences the
size and shape of calcium sulfite and sulfate crystals precipitating from
these high TDS solutions.
This paper presents the results from a laboratory study funded by the
Electric Power Research Institute to screen a broad range of high TDS
solutions to determine the relative effect of each solution on the
crystallization of calcium sulfate and calcium sulfite. The laboratory study
focused on determining the effect of high TDS solutions on the precipitation
rate and the crystal size and habit of each calcium-sulfur salt. These
results are presented in the following text.
PRECIPITATION OF GYPSUM IN HIGH TDS SOLUTIONS
The experimental approach used to study the precipitation of gypsum in
high TDS solutions, followed by a discussion of the results of these studies
is presented in the following paragraphs.
EXPERIMENTAL APPROACH
A mixed suspension mixed product removal reactor (MSMPR) was used to
study the effect of high TDS solutions on gypsum precipitation kinetics and
crystal habit and size modification. Figure 1 shows a diagram of the 2-liter
MSMPR reactor. The reactor was constructed from Plexiglas and incorporates
7-80
-------
STIRRER
HIGH IDS LIQUOR FEED \ EFFLUENT SLURRY
PH ELECTRODE •
UCL2 FEED
IMMERSION HEATER
HIGH TDS
LIQUOR FEED
BAFFLE STRIP
NA2S04 FEED
SIDE VIEW
EFFLUENT SLURRY
TOP VIEW
Figure 1. Mixed Suspension-Mixed Product
Removal Reactor
7-81
-------
a central draft tube with three radial baffle strips. This design, coupled
with a single shaft propeller type stirrer mounted along the axis of the
draft tube insured a well mixed slurry. The reactor temperature was
maintained by an immersion heater controlled by using a thermocouple probe
inserted in the reactor. Figure 2 shows the schematic diagram of the
experimental setup used in the high TDS tests.
Reagent grade chemicals were used to prepare the feed stocks for all of
the crystallizer experiments. The chemical compositions of the high TDS
solutions used in the study are shown in Table 1.
The crystallizer slurry was sampled directly from the drawdown stream by
rapidly filtering a known volume through a .45* Millipore* membrane filter.
The filtrate was diluted with deionized water to prevent additional
precipitation and saved for chemical analysis. The solids collected on the
filter were rinsed, dryed, and weighed to determine the weight percent solids
of the slurry according to the following equation:
„., .. Weight Solid x 100 (1)
Solid = TT-T; a ,. n1
Volume of Slurry
TABLE 1. GYPSUM CRYSTALLIZATION—HIGH TDS SOLUTIONS TESTED
Chemical Analysis (mg/L)
Solution
Identification
240K Mix
120K Mix
30K Mix
20OK CaCl2
250K MgCl2
250K NaCl
250K
200K
Ca+
—
—
—
72,000
—
—
—
—
Mg*4" Na+
39,200 23,200
20,700 12,200
4500 2700
—
63,200
101,000
78,200
41,300
S04
104,000
49,700
13,300
—
—
—
165,000
165,000
ci-
68,800
37,700
7800
127,900
186,800
156,000
—
—
7-82
-------
oo
HIGH TDS
LIQUOR
STIRRER
"a.
88)
HE
c
c
EATER
H
£
PL
WATER BATH
MSMPR
REACTOR
PUMP
CaCL,
Na2SO4
TIMER
SAMPLE
Figure 2. Calcium Sulfate MSMPR Precipitation Rate Experimental Apparatus.
-------
The actual sampling was started after five slurry residence times to
monitor the approach to steady state. The crystallizer was operated and
samples taken for at least 8-10 residence times for each test. Each test was
continued until steady state operation was reached and the sampling was
completed. Photomicrographs of selected solid samples were taken for particle
size analysis. The characteristic dimension measured was the longest axis of
a particular crystal. A typical plot of particle number as a function of
size is shown in Figure 3. The linear growth rate of the crystals grown in
the MSMPR reactor was determined from the slope of a best-fit line. A
detailed description of this techinque of measuring crystal growth rates in
the MSMPR reactor is presented elsewhere by Randolph and Larson (6).
The precipitation rate, Pr, far each experiment was calculated by
Equation 2 which can be derived from a material balance expressions
Pr » (UiJ - [Xf])Q/Ms
where Pr = reaction rate of Ca or 804, mmole/g min L
[X-jJ = initial concentration of Ca or 864, mmole/L
[Xf] = final reactor effluent concentration of Ca or 804, mmole/L
Q = steady state reactor flow rate, L/min.
Ms = mass of solids in reactor, g
The gypsum precipitation, nucleation, and crystal growth rates are all
functions of the relative saturation of CaS04 2H20. The relative
saturation of gypsum is defined as:
aCaaSO a H 0
Relative Saturation (RS) = 4 2 (3)
K
sp
where aca = activity of calcium ion
aSOi. = activity of sulfate ion
ajj20 = activity of water
Ksp = solubility product of CaS04 2H20
The high TDS solutions used in this study required a more empirical approach
to measure gypsum relative saturations since ion activities could not be
calculated accurately. The gypsum relative saturation or supersaturation was
calculated as follows:
7-84
-------
1000
800
600
400
Q.
O
Q_
200
o
<£
c
_>^
'35
C
-------
„, [Ca ] [SO. ]
RbCaSO -2H 0 = —§S 4_ss
[Ca"^] [SO. ]
eq 4 eq
where [Ca ]ss = reactor steady state calcium concentration, mmole/1
[CA++]eq = calcium concentration in equilibrated slurry,
mmole/1
[S04=]ss = sulfate steady-state sulfate concentration,
mmole/1
[S04=]eq = sulfate concentration in equilibrated slurry,
mmole/1
When [Ca] is much less than [804] in the high IDS liquor, Equation (4) can
be reduced to:
[Ca ,
-2H20 = - 8S1 (5)
[Ca ,
eq]
Depending upon the solution conditions, Equations (4) and (5) were used to
calculate the gypsum relative saturation in each precipitation rate test.
The steady-state calcium and sulfate concentrations were determined for each
test. The respective equilibration concentrations were determined by
equilibrating slurry samples for a 2-3 week period at 50°C.
GYPSUM PRECIPITATION IN HIGH TDS SOLUTIONS STUDY RESULTS
The experimental results for the gypsum precipitation rate tests in high
TDS solutions are summarized in Table 2. A plot of precipitation rate versus
calcium sulfate supersaturation for all of the 240K tests is shown in Figure
4. Three distinct bands of data (I, II, III) are shown. Data band I
corresponds to the NaCl and MgCl2 solutions. Data band II represents the
Na2S04, MgS04, and mixed liquor compositions, while data band III shows
the CaCl2 data. The precipitation rate for a given relative saturation in
the NaCl, and MgCl2 test liquors was found to be between 2-3 times greater
than the rate measured in the MgS04, Na2S04, and mixed liquors. The
third data band (III) shown in Figure 4 represents the 240K CaCl2 solution
tests. Note the high supersaturation levels achieved during these runs.
However, the corresponding precipitation rates were much lower than would be
expected by extrapolating results of the other conditions to these
7-86
-------
TABLE 2. EXPERIMENTAL RESULTS—PRECIPITATION OF CaS04 2H2) IN HIGH TDS LIQUORS
00
—i
Calcium Material Balance
Run
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Scrubber Liquor (TDS)
Mixed 26 K
Mixed 28K
Mixed 28 K
Mixed 120K
Mixed 120K
Mixed 120K
Mixed 235K
Mixed 235K
Mixed 110K
Mixed 110K
Mixed 110K
Mixed 110K
Mixed 110K
Mixed 110K
Mixed 110K
Mixed 110K
Mixed 110K
Mixed 110K
Mixed 110K
CaCl2 192K
CaCl2 192K
CaCl2 192K
NaCl 257K
NaCl 257K
NaCl 257K
NaCl 257K
MgCl2 255K
HgCI2 255 K
MgCI2 255 K
Na2S04 243 K
Na2S04 243 K
Na2S04 24 3 K
Na2S04 243K
Na2S04 243K
MgS04 206K
MgS04 206 K
MgS04 206K
Dilute Solution
Feed
mmole.mln"^
2.19
1.88
1.93
3.39
3.78
3.12
2.05
2.62
3.28
• 3.00
1.74
5.61
1.36
1.37
0.91
1.41
2.05
2.62
3.28
1.19
1.63
1.64
7.98
7.35
8.10
3.75
911
8.31
9.83
3.83
3.39
4.27
3.52
3.23
3.27
2.73
2.25
5.78
Effluent
nmole.mln"!
1.65
1.54
1.58
2.28
2.21
2.34
1.49
1.42
1.41
1.24
1.27
1.33
0.63
0.78
0.88
0.78
1.49
1.42
1.41
0.74
0.72
0.74
5.33
5.77
5.65
2.52
7.61
6.55
6.26
2.54
2.57
2.49
0.97
2.36
1.80
1.84
1.80
3.66
Delta
nrnole.mln-'
0.54
0.34
0.35
1.11
1.57
0.78
0.56
1.20
1.87
1.76
0.47
4.28
0.74
0.59
0.09
0.38
0.56
1.20
1.87
0.45
0.91
0.90
2.65
1.58
2.45
1.23
1.49
1.76
3.57
1.29
0.82
1.78
2.56
0.87
1.47
0.89
0.45
2.12
Temp.
(°C)
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
PH
4.80
4.35
4.45
6.05
6.20
6.15
6.25
6.30
6.18
5.95
6.25
6.40
6.28
6.22
5.95
6.32
6.15
6.38
6.40
5.50
5.45
5.22
3.50
3.48
3.55
3.50
3.80
4.10
3.90
6.40
6.32
6.25
6.45
6.10
6.80
6.75
6.82
5.50
Stlrrer
Speed (rpm) **
900 1
900 1
900 1
900 1
900 1
900 1
900 1
900 1
900 1
900 1
900 1
900 1
900 1
900 1
900 1
900 1
900 1
900 1
900 1
900 2
900 2
900 2
900
900
900
900
900
900
900
900
900
900
900
900
900
900
900
900
psun US
.63
.54
.60
.41
.40
.46
.41
.35
.34
.36
.40
.35
.32
.20
.26
.22
.41
.35
.34
.00+
.20*
.36*
.19
.28
.24
.07
.35
.25
.20
.47
.46
.43
.20
.41
.38
.42
.36
.56
Precipitation Rate
(mmole.g-l .m1n-l)
0.74
0.70
0.61
0.61
0.72
1.19
0.85
0.79
0.79
0.63
0.72
0.63
0.38
0.46
0.25
0.38
0.85
0.79
0.79
1.74
.38
.96
.02
.50
.23
0.40
1.34
0.79
0.92
0.94
0.86
0.74
0.13
0.46
0.69
0.71
1.01
0187
-------
i
00
oo
2.0
c
"E
_
o
c
.o
CO
+-»
'o.
1.5
1.0
0.5
0.0
DMixed ONaCI AMgCI2 +MgSO4 0Na2SO4
1.2
IEHighSO4~
Dilute Solution
IHHighCa
+
+
I
2.2
2.4
1.4 1.6 1.8 2.0
Gypsum Supersaturation
Figure 4. Gypsum Precipitation Rate Versus Supersaturation for all 240K High IDS Tests.
-------
same high supersaturations. This may be explained by the fact that in the
presence of high calcium concentrations the sulfate concentrations are
significantly reduced and can thus become diffusion rate limited.
«
The level of total dissolved solids (TDS) in actual scrubber liquors is
expected to increase as FGD systems switch to makeup waters such as cooling
tower blowdown. From a design and operations standpoint it is especially
important to know what the effect of increasing the TDS in a scrubber liquor
will have on the calcium sulfate precipitation rate. The mixed solution
consisting of Na, Mg, Cl, and 804 was tested at 3 TDS values (30,000 mg/1,
120,000 mg/1, 240,000). The overall precipitation rate was found to increase
as the TDS level was raised from 30,000 mg/1 to 240,000 mg/1. Further data
analysis showed that the increase in precipitation rate was primarily a
result of increase in crystal growth as opposed to nucleation. The effect of
TDS level on the enhanced gypsum growth rate in the mixed liquor solution is
shown in Figure 5. The differences in relative saturation among the TDS
levels was accounted for qualitatively.
HIGH TDS EFFECTS ON GYPSUM SIZE AND HABIT
Changes in crystal habit and size are important in determining the
physical properties of crystal slurries. The settling and dewatering
characteristics of gypsum sludge produced in the flue gas desulfurization
process are of special int erest to the design and operating engineer. The
effect of high TDS levels on gypsum crystal habit and size was previously
unknown. The following paragraphs report the results obtained by this study
on the effect of liquor chemical composition on gypsum crystal size and
habit.
Table 3 lists the high TDS liquor solution compositions along with the
observed crystal habit and the gypsum relative saturation at which the
crystals were grown. A key showing the different shapes of the monoclinic
gypsum system appears at the bottom of the table. Scanning electron
photomicrographs were taken of each test solution solids and are shown in
Figures 6 and 7. Each photomicrograph was taken at about 200X magnification
and a micrometer scale is shown between the arrows of each photomicrograph
for reference.
Gypsum precipitation in dilute solution (Figures 6a and 7a) exhibited
agglomerate crystal structures reflective of the high relative saturation
growth conditions. MgCl2, MgS04, and the mixed high TDS liquors all
produced very needle-like (acicular) crystals (Figures 6a, 7c and 7d).
Alternatively, the NaCl and Na2S04 solutions produced gypsum crystals
more columnar in habit as shown in Figure 6b and 7b, respectively. The
CaCl2 liquor had the most dramatic effect on gypsum crystal habit
modification. Figure 6c shows the CaCl2 solutions gypsum crystals grown at
comparably higher relative saturations, RS=2.2. These gypsum crystals were
7-89
-------
8
E
a.
1 3
O
240K Mix
120K Mix
>30K Mix
5 10 15 20 25
Total Dissolved Solids (mg/l)x 1(H
Figure 5. Growth rate vs IDS level for mixed
liquor solutions.
7-90
-------
TABLE 3. GYPSUM CRYSTAL HABIT IN HIGH TDS LIQUOR SOLUTIONS
i
VD
Solution Identification
in mg/l
CaSO4. 2H2O Crystal Habit
Dilute Solution (Base Case)
250K NaCI
250K MgCI2
200K CaCI2
250K Na2SO4
200K MgSO4
240K Mix
Agglomerate
Agglomerate/Columnar
Acicular
Lamellar
Agglomerate/Columnar
Acicular
Acicular
CaSO4. 2H2O
Relative
Saturation
1.55
1.30
1.25
2.20
1.45
1.40
1.30
Lamellar
Tabular
Equant
Columnar
Acicular
-------
Figure 6. Scanning electron micrographs of gypsum solids
(a)
(b)
(c)
(d)
Acicular/agglomerate gypsum from dilute solution,
Columnar/agglomerate gypsum from Nad (240K)
Lamellar gypsum from CaCl2 (240K)
Acicular gypsum from MgCl2 (240k)
7-92
-------
pJretJ: s(c:;n1;9 Tr micr°9raphs °f wsum *°"<*
- J l£. (a) Acicular/agglomerate
Columnar/agglomerate
Acicular gypsum from
Acicular gypsum from mixed liquor (240K)
7-93
-------
much more lamellar in habit. Of all the inorganic solutions tested, Nad,
Na2S04, and CaCl2 had the most effect on crystal habit modification.
In both the NaCl and Na2SC>4 liquor solutions the gypsum crystals
precipitated were columnar and dendritic even when precipitated at low
relative saturations. The more columnar crystal habit is thought to impart
better settling and dewatering properties to the respective slurries.
Conversely, the gypsum crystals precipitated from the CaCl2 liquor were
lamellar in habit, which is not well suited for settling and dewatering.
PRECIPITATION OF CALCIUM SULFITE HEMIHYDRATE IN HIGH TDS SOLUTIONS
The following section reports the experimental approach and results of
the part of the screening study devoted to determining the effects of high
TDS soultions on calcium sulfite crystallization.
EXPERIMENTAL APPROCH
The prevention of localized precipitation is of primary importance in
designing and executing precipitation rate experiments. Generally, the mixed
suspension-mixed products removal reactor (MSMPR) has been used to measure
precipitation rates under steady-state conditions. Satisfactory MSMPR
reactor operation requires complete and rapid mixing of the reactant feeds to
prevent localized precipitation in the region where the reactant streams mix
with the bulk solution. This is especially critical when precipitating
slightly soluble salts such as calcium sulfite.
Some initial MSMPR calcium sulfite precipitation tests were conducted at
several reagent feed rates. The precipitation rates and the calcium sulfite
relative saturations were measured. The precipitated calcium sulfite solids
were examined by microscopy. The data from these initial MSMPR experiments
indicated that localized precipitation of calcium sulfite was occurring in
the MSMPR reactor. Therefore, an experimental approach more closely
resembling an actual FGD scrubber system was adopted for the calcium sulfite
precipitation rate tests.
Figure 8 shows a schematic of the bench-scale S02 scrubbing apparatus
used to conduct the calcium sulfite precipitation rate tests. The apparatus
consisted of a flue gas mixing manifold, gas-liquid contactor, reaction tank,
and a calcium carbonate slurry hold tank. Sulfur dioxide gas (S02) was
blended with nitrogen gas and contacted with a calcium carbonate slurry. The
CaC03 slurry was used as a source of alkalinity to neutralize the S02
absorbed from the gas mixture as shown in Equation 6.
CaC03(s) + 2S02 (aq.) + 2H20 ? Ca"*"* + 2HS03 + H2C03 (6)
The calcium from Equation (6) is then available to precipitate calcium
sulfite-calcium sulfate solid solution according to the following equation:
+ (l-x)S03 + (x)SO + 1/2H20 •*• Ca(S03).._ (SOi+J • 1/2H20 (7)
J.^X X
7-94
-------
H
0
DRY GAS
MEIER
(^
J, V
' MONITOR
.PRESSURE
' (^^ ,«"
1 Ld M'°
« i
* AB
s~~*\ COLUMN
(rcV— ^ o-iaovAC
^
OVHA SCIENCES
SO, ANALYZER RANGES
OlOOOppm -»-VENT
OJOOOppm
0 10.000 ppm
MAIIIf OlD
»)PHCSSUH£
GAUGE /-"v •
Limestone $— {— )
feed XX
HOIOMEIER "* ~/^>k
41611/mln "*"IX1 «N» >Jv
ROIOME1ER
.024 im/n.ln
-^(Xl ?0,.N,(AIH| ,
i. —U
ROIOMEIER
1 54831/niln
--00 — fco, 1 vW/ ,_,
WATER
BA1H
_ ROIOMEIEH
.100 4.12 Wmln
•*-O<3 — 1 SO,.NJ
^
vL/ i
SORBERIPACKEO
OORBUBBLER)
V'
/\
/ \
(_\
^
VATIIBIE |[
SPEED 1
MOTOR 1
n i:
1
DOWN'
L^
L.l
STREAM IMPINGERS
REACTOR
_ BLEED STREAM
0<}-fc- OR MAKE UP
WATER ADDITION ,-.
cJo
^
jc)
^
Figure 8. Schematic Diagram of the Bench-Scale S02 Gas Scrubbing Apparatus
Used for Calcium Sulfite Precipitation Studies.
7-95
-------
The calcium sulfite precipitation rate was determined directly from the
S02 pickup rate in the bubbler contactor. This assumption is correct only
if the reaction tank slurry circulating through the contactor is saturated
or supersaturated with respect to calcium sulfite. At the beginning of each
test it was first verified that saturation was achieved by periodically
analyzing the total sulfite-sulfur concentration until a constant value was
reached.
The total sulfite-sulfur and calcium concentrations measured at steady
state and those concentrations measured later from the equilibrium slurry
samples were used to calculate the approximate calcium sulfite relative
saturations during each test as shown in Equation (8).
[Ca
CaS03'l/2H20 R.S. «
CALCIUM SULFITE PRECIPITATION IN HIGH TDS SOLUTIONS STUDY RESULTS
Seven different high TDS solutions were tested on the bench-scale system
and are listed in Table 4. Also shown for each test liquor was the
steady-state concentrations of total sulfite and calcium, the calcium sulfite
precipitation rate, and the calcium sulfite relative saturation obtained for
each test. The dilute solution test was used as a base case for direct
comparison with the other high TDS solutions. Of the seven high TDS liquors
tested, three contained the common anion chloride (NaCL, MgCL2» CaCl2)«
The precipitation rate data obtained for each of these solutions is plotted
along with the dilute solution data in Figure 9. The high chloride
precipitation rate data points all fall around the dilute solution base case
indicating that high chloride concentrations had little effect on calcium
sulfite precipitation rate. However, a similar plot of the high sulfate test
soltuion data, Figure 10, shows some enhancement in the calcium sulfite
precipitation rate for a given reltive saturation when compared to the dilute
solution base case.
Since large amounts of calcium sulfite seed crystals were retained in
the crystallizer during each test, crystal growth and nucleation rates could
not be determined individually. Therefore, the effects of high TDS are
reported as changes in the overall precipitation rate only. However, it is
clear that the precipitation rate of calcium sulfite in high liquors is
approximately double the precipitation rate for a given relative saturation
when compared to the dilute solution base case and high chloride liquors.
The crystal habit of calcium sulfite precipitating from FGD systems can
vary from lamellar crystal shape to crystals resembling rosettes and
clusters. It has been noted that the lamellar shaped crystals precipitate
more frequently from limestone reagent FGD systems and the rosettes arise
primarily from lime-reagent FGD systems.
7-96
-------
TABLE 4. CALCIUM SULFITE HIGH IDS PRECIPITATION RATE RESULTS1
Steady-State
Reactor Concentration
Run No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Designation
Dilute Solution
Dilute Solution
Dilute Solution
Dilute Solution
Dilute Solution
240K Nad
240K NaCl
240K NaCl
240K NaCl
240K NaCl
240K NaCl
240K NaCl
240K NaCl
240K NaCl
240K NaCl
240K MgCl 2
240K MgCl 2
240K MgCl 2
240K MgCl 2
240K MgCl 2
50K CaCl 2
50K CaCl 2
50K CaCl 2
50K CaCl 2
24K Na2SO,,
24K Na2SO,
24K Na2S04
24K Na2SO,
24K Na2SO,
24K MgSO,.
24K MgSO..
24K MgSO,
24K MgSO..
24K Mix
24K Mix
24K Mix
Sulfite
mg/L
485
660
755
890
965
420
570
620
700
725
775
425
550
595
680
925
1,355
1,420
1,520
1,525
85
100
110
80
3,530
4,330
4,700
5,320
5,840
3,380
3,665
'4,240
4,810
1,325
1 ,740.
2,080
Calcium
mg/L
410
440
485
540
555
295
360
410
440
470
500
320
370
410
440
590
810
870
960
950
17,040
16,910
16,360
16,380
510
530
540
525
545
650
630
700
680
640
800
830
CaS03-l/2H20
Precipitation
(mmoles/g-min)
.008
.017
.022
.025
.030
.010
.013
.016
.019
.022
.025
.012
.017
.020
.023
.014
.021
.027
.026
.028
.016
.024
.028
.011
.018
.026
.034
.038
.042
.031
.038
.041
.048
.021
• .025
.029
CaS03-l/2H20
Relative
Saturation
3.05
4.50
5.75
7.40
8.28
1.89
3.20
4.00
4.75
5.25
5.95
2.08
3.14
3.77
4.62
2.58
5.21
5.88
6.92
6.88
6.11
6.97
7.60
4.18
2.79
3.54
3.92
4.31
4.92
3.89
4.08
5.25
5.76
1.50
2.45
3.00
]A11 tests were conducted at 48°C and pH 5.5
2K = 1,000 mq/1
7-97
-------
i
vo
00
0.05
•| 0.04
g.
0.03
•5*
"o
0)
"co
DC
I °-°2
CS
0.01
0
0
Dilute Solution
Figure 9.
89 10
CaSO3. 1/2 H2O Relative Saturation
240K NaCI • 240K MgCI2 A 50K CaCI2
Precipitation Results in High Chloride Test Solutions
-------
VO
VD
0.05
c
"E 0.04
6)
O
0.03
0)
OS
DC
o 0.02
'•§
O
0)
0.01
0
0
Dilute Solution
8
Relative Saturation
24K Na2SO4 A 24K MgSO4 • 24K MIX
10
Figure 10. CaS03'JsH20 Precipitation Results in High Sulfate Test Solutions
-------
Calcium sulfite is known for its poor settling and dewatering properties
causing solids disposal problems in some FGD systems. Settling and
dewatering characteristics are influenced by many physical factors including
particle size and crystal habit. Very little research has been done to
determine the effect of high IDS solutions on calcium sulfite crystal habit
modification. As part of this screening program, a test sequence was made to
screen and evaluate the habit modifying potential of high IDS liquors on
calcium sulfite. These tests were carried out by the slow addition of .5
molar solutions of CaCl2 and Na2SC>3 into a stirred, nitrogen purged,
20-liter carboy filled with the high IDS test solution. The reagent addition
rate was set at .3 to .4 millimeters/minute and continued for up to 30 hours.
The slow reagent addition rate allowed the calcium sulfite relative
saturation to increase slowly in the carboy until nucleation occurred. The
results of these tests are shown in the photomicrographs in Figures 11 and
12. The crystals precipitated from high chloride liquors are compared with
those precipitated from dilute solution in Figure 11. The chloride liquors
all produced lamellar type crystals. The NaCl liquor solution produced
crystals with a greater length to width ratio resembling crystals more
acicular in habit (Figure He). The calcium sulfite precipitated from the
high sulfate liquors shown in Figure 12 contained no lamellar crystals. The
calcium sulfite solids consisted of rod shaped and globular crystals with
some foliated growth patterns evident. These crystals appeared to have more
mass per unit area than the crystals precipitated from the dilute solution
and high chloride liquor solutions.
SUMMARY AND CONCLUSIONS
CALCIUM SULFATE CRYSTALLIZATION IN HIGH TDS SOLUTIONS
The results of this screening study showed that the precipitation rate
of calcium sulfate was affected by high TDS solution composition. These
findings are summarized below.
1. Gypsum precipitation rates in 240,000 mg/1 NaCl and MgCl2
solutions were five times greater than in dilute solution for a
given relative saturation.
2. Gypsum precipitation rates measured in 240,000 mg/1 Na2SO^,
MgS04, and mixed liquor solutions were 2-3 times greater than in
dilute solution for a given relative saturation.
3. The CaCl2 high TDS solution (240,000 mg/1) suppressed the gypsum
precipitation rate compared to dilute solution.
4. Calcium sulfate dihydrate was found to be the only hydrate stable
in the solutions tested.
5. The crystal shape was found to be affected differently between the
high TDS solutions tested.
7-100
-------
(c)
(d)
Figure 11.
Photomicrographs of Calcium Sulfite Precipitated from HTDS Solutions
(a) Dilute Solution (b) 50K CaCl2
(c) 240K NaCl (d) 240K MgCl2
7-101
-------
—
(c)
(d)
Figure 12.
Photomicrographs of Calcium Sulfite Precipitated from HTDS Solutions,
(a) Dilute Solution (b) 24K NazSO,,
(c) 24K MgSO,, (d) 24K Mix
7-102
-------
The increase in precipitation rate was found to be dependent upon IDS level
in tests using a mixture of Na, Mg, Cl, and 804. Gypsum crystal growth
rate measured in this mixed solution composition showed a linear increase in
growth rate as a function of IDS level. At a IDS level of 200,000 mg/1 the
crystal growth rate was enhanced 400% compared to the dilute solution base
case.
These results suggest that FGD systems currently operating in the
10,000-25,000 mg/1 TDS range are not yet at a TDS level where the calcium
sulfate precipitation and growth rates are significantly affected by solution
TDS levels. However, these preliminary data indicate that an increase in the
TDS level of scrubber liquors containing Na, Mg, Cl, and 864 may be
accompanied by an increase in the gypsum growth rates. An increase in the
growth rates of gypsum (for a given relative saturation) in the FGD system
can have two potential benefits for FGD operations. First, a larger particle
size distribution of scrubber solids should result for a given operating
relative saturation. A larger particle size distribution could improve the
solids settling and dewatering characteristics of the FGD scrubber slurry.
Second, an increase in crystal growth should regulate the gypsum relative
saturation in the scrubber at a lower level for a given sulfur dioxide
removal. A lower gypsum relative saturation in the reaction tank could
decrease the gypsum scaling potential in the absorber.
CALCIUM SULFITE CRYSTALLIZATION IN HIGH TDS SOLUTIONS
The precipitation rate of calcium sulfite in high TDS solutions was
found to depend on solution composition only for those solutions containing
sulfate ion. In high TDS solutions of Na2S04, MgS04, and a mixture of
Mg, Na, Cl, and 804, the calcium sulfate precipitation rate was
approximately two times the rate measured in chloride based high TDS
solutions and in dilute solution. The calcium sulfite precipitation rates
were measured over a wide range of relative saturations. Typically, the
relative saturation varied from 2 to 8 for each test solution. These
relative saturations are typical of those encountered in many FGD systems.
These preliminary results indicate that the precipitation rate of calcium
sulfite in FGD system will be less affected by increases in the TDS level in
the recirculating slurry than that of gypsum.
Solutions containing sulfate were found to influence the calcium sulfite
crystal size and habit. Calcium sulfite precipitated in the presence of high
concentrations of sulfate were rod shaped or globular with some foliated
growth patterns. Whereas, calcium sulfite precipitated in high TDS chloride
solutions or in dilute solution were lamellar or acicular in habit.
Interpretation of these data are difficult, since these batch screening tests
were carried out without knowing the precise relative saturation of calcium
sulfite for each test solution. Additional experimentation under
steady-state precipitation conditions will be required to accurately evaluate
the effect of high TDS solutions on calcium sulfite crystal size and habit.
7-103
-------
LITERATURE CITED
1. Slack, A. V., Scrubber Survey: A Lime/Limestone Trend, Electrical World,
Vol. 182 (1974).
2. Rosenberg, H. S., How Good is Flue Gas Desulfurization?, Hydrocarbon
Processing, Vol. 57 (1978).
3. Elliott, T. C., S02 Removal from Stack Gases, Power, Vol. 118 (1974).
4. Smith, B. R., F. Sweet, The Crystallization of Calcium Sulfate, Journal
of Colloid and Interface Science, Vol. 37 (1971).
5. Liu, S. T. , G. H. Nancollas, The Kinetics.of Crystal Growth of Calcium
Sulfate Dihydrate, Journal of Crystal Growth, Vol. 6 (1970).
6. Randolph, A. D., M. A. Larson, Journal of the Institute of Chemical
Engineering, Vol. 8 (1962).
7-104
-------
EFFECT OF HIGH DISSOLVED SOLIDS ON BENCH-SCALE
FGD PERFORMANCE
J. B. Jarvis, T. W. Trofe, D. A. Stewart
-------
EFFECT OF LIMESTONE GRINDING CIRCUIT ON FGD PERFORMANCE AND ECONOMICS
by: J. D. Colley and 0. W. Hargrove
Radian Corporation
Austin, Texas
D. A. Stewart
Electric Power Research Institute
Palo Alto, California
ABSTRACT
Results of several recent EPRI sponsored programs investigating technical
and economic issues of limestone preparation for SOz scrubbing are presented
in this paper. Variables important in the selection of a limestone for FGD
application are discussed. The most common method used by vendors for sizing
limestone ball mills is identified. Correlation of this method with the
chemical, physical, and petrographic characteristics of a number of different
limestones was investigated.
Pilot-scale testing with a wet ball mill, air-swept ball mill, tower
mill, ring roller mill, and hammer mill was conducted with three of these
limestones. Selection of the stones was based^ on the strength of the mate-
rial. A hard, medium and soft stone were used so that the effect of this
variable on each machine's performance could be measured. Data were col-
lected on the performance of these machines in grinding these limestones to
various particle size distributions. Product samples collected from each
were tested in laboratory equipment to measure the effect of grinding method
on limestone dissolution rate.
Full-scale testing at Central Illinois Light Company's Duck Creek Unit
No. 1 measured the effect of the limestone mill circuit operation on the
product particle size distribution. Testing also quantified the effect of
the limestone size distribution on its utilization in the scrubbers. The
economic tradeoffs of producing the finer size product were estimated based
on data collected during the on-site testing.
INTRODUCTION
As of March 1983, there were 106 operational utility FGD systems in the
United States. An additional 105 systems were under construction or in the
planning stage at that time for a total FGD controlled generating capacity of
103,219 MW. Approximately one-half of the existing FGD controlled utility
plant capacity use limestone as the alkaline reagent. This fraction is not
projected to change by much through the remainder of this century. Based on
7-105
-------
current FGD industry surveys, the anticipated demand for limestone FGD sys-
tems will add an equivalent of six new 500 MW installations per year through
1991.
' Using average industry figures, the estimated annual consumption of
limestone by utility FGD systems will be almost 40 million tons in 1991. De-
pending upon the inflation rate for limestone and the average delivered cost,
from $700 million to $900 million will be spent annually by the utility in-
dustry to purchase this quantity. Factors which affect the cost efficient
utilization of this reagent are therefore of significant interest to the
utility user. To provide more information in this area, the Electric Power
Research Institute (EPRI) has recently sponsored two programs, directed by
Dorothy Stewart, which investigate efficient use of limestone in flue gas
desulfurization:
• Flue Gas Desulfurization Studies: Limestone Grindability
• Flue Gas Desulfurization Reagent Preparation Study
The first study had two major objectives: 1) determine the range in
grindability (ease of grinding) and investigate correlations between lime-
stone grindability and its physical, chemical, and petrographic character-
istics; and 2) determine types of tests and data required to size and design
ball mill grinding equipment for limestone FGD applications.
The objective of the second study was to develop design and economic
information concerning the preparation of lime and limestone for FGD applica-
tion. Only information on the limestone phase of this program is presented
in this paper. Results of pilot-scale grinding equipment testing, lab and
bench-scale work, and full-scale utility testing are presented. The paper is
organized into two major sections, the first dealing with the grindability
program and the second discussing the results of the limestone preparation
program.
LIMESTONE GRINDABILITY
The grindability of a reagent limestone is a characteristic being given
ever increasing attention in the optimization of controllable variables in
limestone FGD systems. As will be shown in the full-scale testing section
of this paper, S02 removal efficiency and limestone utilization can be in-
creased as the average particle size in the feed is reduced. However, there
is an exponential increase in the costs associated with production of smaller
particle sizes. And as will be seen in this section, natural variations in
stone strength have a major impact on the cost of grinding. This section pre-
sents the results of the grindability study, including the selection criteria
used to choose the 30 limestones in the study, geologic descriptions for
selected limestones, results of the grindability testing, and results of the
ball mill sizing survey.
7-106
-------
LIMESTONE SELECTION CRITERIA
The samples selected for this study were chosen primarily to assess the
grindability characteristics of a cross section of the major carbonate de-
posits in the U.S. with commercial potential. No minimum CaC03 content was
specified. In fact, some magnesium limestones and dolomites were purposely
included with the high-calcium limestones in order to determine any possible
effect of magnesium-carbonate phases on grindability behavior. Of secondary
consideration was acquiring stones with a wide variety of physical character-
istics, i.e., variable grain size and types and proportions of transported
and precipitated constituents.
Because of the nature of carbonate deposits, there are significant
lateral and vertical variations in chemical and physical properties over
relatively small distances. Therefore, a further selection consideration was
the consistency between chemical and physical trends noted within the
selected formations and the trends observed at the quarry level. This
comparison was made visually, by direct comparison of grab samples of the
crushed stone, by conversations with quarry operators and/or other knowledge-
able individuals, and through review of available chemical analyses.
Limestone samples were obtained in three ways. A number of the samples
selected were on hand in Radian's limestone library in sufficient quantities
to conduct all proposed analyses. These samples were screened for consis-
tency as described above. Other samples were newly obtained from quarries
identified by R. C. Freas of Limestone Products Corporation or geological
survey personnel of appropriate states. In these cases, quarry operators
were contacted and asked to supply a minimum 100 pound sample of the forma-
tion or unit of interest, along with any pertinent information on the local
geologic setting and quarrying operations. The level of detail of this
latter information is highly variable, depending primarily on the individual
background of each quarry contact. For the remaining samples, available in-
formation suggested that the target formations could be too compositionally
variable to be sampled by quarry personnel. In these cases, a geologist
travelled to the quarry to obtain the material with the quarry personnel's
assistance. The thirty limestones selected for study are listed in Table 1.
The geologic formation, quarry location, and Radian limestone library code
are given for each.
LIMESTONE SAMPLE DESCRIPTIONS
All of the samples subjected to grindability testing were described with
respect to their geologic history and in-hand physical appearance in the pub-
lished report on this program (1). Descriptions of only three of the lime-
stones are given below. These stones were the ones chosen for study in the
Reagent Preparation Program. When possible, details of quarry-specific
geology are included with the more general description of each sampled unit's
regional characteristics. Any uncertainties regarding sample representative-
ness are also discussed.
7-107
-------
TABLE 1. SELECTED LIMESTONES
Geologic Formation
Quarry Location
Sample Code No.
Aragonite
Manlius
Monteagle
Manlius
Vanport
Racine Dolomite
Beekmantown
Newala
Fredonia Mem., St. Gen.
Fredonia Mem., St. Gen
Redwall
Ocala
Newala
High Bridge Group
Crystal Pass Mem., Sultan
Crystal Pass Mem., Sultan
Salem
Franklin Marble
Bethany Falls, Ls., Swope
Dundee
Murphy
Edwards
Furnace Creek
Fredonia Mem., St. Gen.
Detroit River
Brassfield
Iceland Spar
Kimmswick
Dundee
Undiff. Pennsylvania
Ft. Pierce, FL
Jamesville, NY
Tiftonia, TN
Jamesville, NY
Butler, PA
Chicago, IL
Blue Ball, PA
Birmingham, AL
Cave-In-Rock, IL
Cave-In-Rock, IL
Peach Springs, AZ
Sumterville, FL
Birmingham, AL
Maysville, KY
Henderson, NV
Henderson, NV
Stinesville, IN
Sparta, NJ
Liberty, MO
Rogers City, MI
Tate, GA
Georgetown, TX
Lucerne Valley, CA
Fredonia, KY
Trenton, MI
Piqua, OH
Chihuahua, Mexico
Illmo, MD
Rogers City, MI
Wedone, UT.
023-B
047
023-A
067
051
007
043-A
069-0
069-B
070
074
043-B
034
019-A
019-B
066
075
072
055-A
073
048
039
022
076
010
071
055-B
068
Monteagle Formation
The Mississippian Monteagle Formation occurs throughout a large area of
south-central Tennessee. It is time-correlative with the Ste. Genevieve
Formation and the underlying St. Louis Limestone. The Monteagle Formation is
extremely variable in its lithologic characteristics.
The Monteagle sample used for this study was collected at the Stone Man
Quarry, Tiftonia, TN. The exposed stratigraphic section comprises, from top
to bottom, approximately 2 feet of overburden; 18 feet of Bangor Limestone,
48 feet of interbedded Harselle shale and limestones; 180 feet of Monteagle
Formation; and 40 feet of St. Louis Formation.
The sample collected from Stone Man was taken from stockpiled material
consisting predominantly of the lower 100 feet of Monteagle Formation (high-
calcium limestone with minor interbedded shale and chert nodules). A typical
7-108
-------
sample consists of dark gray, massive, fossiliferous, medium crystalline,
medium calcarenite.
Salem Formation
The Salem Formation is a Mississippian limestone that contains the thick-
est units of high calcium stone exposed in Indiana. The formation crops out
from Montgomery County to the Ohio River and into Kentucky. It is overlain by
the St. Louis Limestone and underlain by the Harrodsburg Limestone.
The Salem Formation was deposited in a shallow marine environment char-
acterized by frequent fluctuations in shoreline. This depositional setting
resulted in a geometrically complex network of lenticular bodies of fossil-
rich material within an overall wedge-shaped formation. Total-formational
thickness may, in places, exceed 80 feet.
The Salem sample used in this study was obtained from Indiana Cal-Pro,
which produces high-calcium limestone from an underground mine located about
0.5 miles southwest of Stinesville, IN. Here, a 91 foot thick section of
Salem Limestone is overlain by approximately 16 feet of St. Louis Limestone
and underlain by 35 feet of Harrodsburg Limestone. The mined stone is ex-
tremely consistent in both physical and chemical characteristics. A typical
hand-specimen is described as a light gray, massive, fossiliferous, medium
calcarenite.
Kimmswick Formation
The Ordovician Kimmswick Formation (subgroup in Illinois) crops out
mainly along the Mississippi River from northwestern Illinois to southwestern
Missouri and is also exposed along the White River in Arkansas. Formational
thickness decreases to the south. In Illinois, thicknesses vary from about
150 to 200 feet, but in Missouri the range is 25 to 120 feet.
The Kimmswick Subgroup in Illinois includes the Wise Lake and Dunleith
Limestone units, of which the former is the more widely quarried. Composi-
tionally, the Kimmswick varies from fine-grained high-calcium limestone to
cherty and argillaceous limestones and dolomites.
In Missouri and Arkansas, the Kimmswick is a major source of high-calcium
stone. It typically contains only minor impurities (principally chert) and
exhibits only local dolomitization.
The Kimmswick sample used in this study was obtained from West Lake
Quarry and Materials Company in Illmo, Missouri. Quarry-specific details were
not obtained; however, the sample supplied is high-calcium limestone composed
of approximately 96 percent CaC03. A typical sample is a light gray to tan,
massive, fossiliferous, medium calcarenite.
GRINDABILITY TEST RESULTS
The grindability of the limestone samples was determined using a labora-
tory ball mill procedure developed at Radian. The procedure was patterned
7-109
-------
after the Hardgrove method for measuring coal grindability and was intended to
produce relative grindability data for the rapid and inexpensive analysis of a
large number of different limestone samples. Briefly, the test method uses a
specific size jar mill, specific size and amount of ceramic grinding media,
and specific size distribution and amount of limestone feed. The limestone is
dry ground at a preset mill rpm for three timed periods such that the total
number of mill revolutions per test can be calculated. The ground limestone
product from each timed grinding is screened to determine the weight percent
passing through a USA Standard No. 200 mesh sieve. A plot is then constructed
of mill revolutions versus percent material passing through 200 mesh. The
number of revolutions required to grind 80 percent of the sample to -200 mesh
is determined from the plot. The grindability index for each sample tested is
then defined as the quotient of 50,000 divided by the number of mill revolu-
tions required to grind the sample to 80 percent passing through a 200 mesh
sieve.
Figure 1 presents the results of the grindability tests made on each of
the 30 limestones in the study. The grindability index is inversely propor-
tional to the ease of grinding with the hardest or highest strength stones
having the smallest index and the lower strength stones having higher numbers.
As the figure shows, the range of indices measured for the limestones tested
was from 2.1 (hardest stone) to 15.5 (softest stone). These results indicate
that there is a significant variation in limestone strength in the U. S.
population of stones. Stone strength or grindability is a major factor af-
fecting the size of ball mills. This variable is therefore an important
consideration in designing and operating limestone preparation equipment.
To assess correlations between grindability and other physical, chemical,
and petrographic characteristics, each limestone was analyzed for: CaC03,
MgCC>3, and acid insoluble material; crystalline phases; dissolution rate;
grain size; specific gravity; and thin section petrography. Six samples were
also tested using the Bond Work Index method. (The Bond Work Index is the
most widely used measurement for sizing ball mills, as will be discussed
later.) The results from the tests using the grindability index and the Bond
Work Index procedures were compared to assess the sensitivity of the two
methods to differences in grindability and to determine any relationship
between the two methods. Figure 2 presents a graph of the observed relation-
ship between Bond Work Index and grindability index.
Correlations between grindability and the other variables were evaluated
by measuring the strength of correlation coefficients for linear relation-
ships and the probability of those relationships being statistically signifi-
cant. The only significant correlations observed were between grindability
index and: 1) Bond Work Index, 2) percent CaC03, 3) percent acid insolubles
(inerts), and ,4) percent quartz. The correlation between grindability index
and Bond Work Index was very strong and very significant, which validates re-
sults from the laboratory procedure. The statistical significance of the
correlation does not indicate a cause-and-effect relationship, however, since
both methods are direct measures of grindability. Although the correlations
between grindability index and percent CaC03, inerts, and quartz are statis-
tically significant, the correlations are not very strong. And, percent
7-110
-------
15 10 5
Grindability Index
2.1
4.4
5.3
5.6
6.7
6.8
7.0
7.5
8.0
8.1
8.1
8.1
8.2
8.3
8.6
8.9
9.7
9.8
10.8
11.3
11.5
12.0
12.5
12.8
12.9
13.0
13.9
14.5
15.5
15.5
Aragonite (FL)
Manllus Fm. (NY)
Monteagle Fm. (TN)
Manlius Fm. (NY)
Vanport mem./Clarion Fm. (PA)
Racine Dol&mite (IL)
Baekmantown Grp. (PA)
Newala Fm. (AL)
Fredonia mem./Sla. Ganevieve Fm.(IL)
Fredonia mem./Sta. Ganevieve Fm. (IL)
Redwall Fm. (AZ)
Ocala Grp. (FL)
Nawala Fm. (AL)
High Bridge Grp. (KY)
Crystal Pass merrUSultan Fm. (NV)
Crystal Pass mem./Sultan Fm. (NV)
Salem Fm. (IN)
Franklin Fm. (NJ)
Bethany Fall* mem JSwope Fm. (MO)
Dundee Fm. (Ml)
Murphy Fm. (GA)
Edwards Fm. (TX)
Furnace Fm. (CA)
Fredonia mem./Sla. Genevleve Fm. (KY)
Detroit River Grp. (Ml)
Brassfleld Fm. (OH)
Iceland Spar (Mexico)
Klmmswick Fm. (MO)
Dundee Fm. (Ml)
Undlfferentiated Pennsylvanian (UT)
Geologic Unit (Location)
14-
12-
10-
a-
4-
Klmmswlck
O
Higher Grindability Index Value*
and Lower Bond Work Index Values
Indicate 'easier to grind* stone.
Edwards O
Salem O
Redwall. Ocala O
10
12
14
Bond Work Index
Figure 1. Range of Grindability Indices Observed
Figure 2. Relationship of Grindability Index
With Bond Work Index
-------
CaC03, inerts, and quartz are themselves strongly correlated; i.e., they are
interrelated and are not independently varying. In this study, no physical,
chemical, or petrographic characteristics of limestones were correlated
strongly enough with grindability to be a predictor of grindability.
Survey of Ball Mill Sizing Techniques
Since its development, the ball mill has found use in a wide variety of
grinding applications. Operational characteristics of this grinding machine
have made it by far the most popular selection for limestone FGD. Despite
the fact that ball mills have been in use for over 100 years, techniques for
designing and sizing them have had little basis in theory. As a result,
grinding and particle size reduction has developed more as an art than a
science for not only the ball mill, but for most other size reduction equip-
ment. The sizing of ball mills has been largely a question of applying em-
pirical equations or factors based on accumulated experience (2) .
Results of the equipment manufacture and literature survey indicate that
the sizing-by-energy method, developed principally by Fred C. Bond, is cur-
rently the most popular approach used for designing limestone ball mills for
FGD applications. This technique assumes that the energy per ton to break a
given feed size to a desired produce size is constant for a selected lime-
stone. Then, the energy per ton and a scale-up relationship involving mill
diameter are used to size the mill. The energy input to the test mill is
measured from power and grinding time, scaled to the production mill diam-
eter, and the production mill length estimated from an equation relating
power to mill length and diameter.
These relationships form the basis for sizing mills using the Bond Work
Index approach. With this method, a standard laboratory test procedure is
used to gather information to calculate the Bond Work Index by an empirical
formula. This Work Index is then used to estimate the required mill power
per ton of product by the following equation:
/ 1 \°'5
10 WI ,(-^-) - -=M (1)
where: E = kilowatt-hours per ton,
WI = Bond Work Index,
P = micron size that 80% of the product passes, and
oU
F0_. = micron size that 80% of the feed passes.
oU
This required specific energy (E) is then used to size the mill based on a
power-to-size relationship derived from mechanical analysis of a tumbling
mill (3) :
W = KpLD2'5 (2)
7-112
-------
where: W = power to rotate a rumbling mill,
K = a constant for a given set of mill conditions,
p = ball density,
L = mill length, and
D = mill diameter.
And since,
E = W/Q (3)
where: Q = desired production rate,
substituting fives:
Q = KpLD2>5/E (4)
('
This relationship is used to choose the mill length and diameter to give the
desired product rate.
Therefore, the required mill power per ton of product (specific grinding
energy) and production rate determine the size of the mill. Since the spe- t
cific grinding energy is directly proportional to Bond Work Index, the effect
•of limestone grindability on specific energy can be estimated. Using Figures
1 and 2, the range of Bond Work Indices for the 30 test stones is approxi-
mately 5 to 14. Substituting these into Equation 1 and using a PQQ of 36
microns (equivalent to a product with 90% passing 325 mesh) and an FQQ of
8,500 microns (^1/2 inch top size raw limestone), the specific energy re-
quired to grind stone with a Work Index of 5 is 8 kW-hrs/ton and for a stone
with a Work Index of 14 is 22 kW-hrs/ton. Limestone strength can therefore
have a significant effect on the amount of energy required to grind the mate-
rial to a size distribution typical of that used in a wet limestone scrubber.
As an example of the effect that stone strength has on the size of the
ball mill and motor, two mill systems were designed for a conceptual 500 MW
plant firing a high sulfur (3.5% by wt) coal. The mills were sized to pre-
pare 24 hours worth of limestone for the scrubbers with the unit at maximum
boiler load in one 8-hour shift. The mill system was capable, therefore, of
producing 90 tons per hour of limestone, of which 90 percent would pass a 325
mesh screen. For the soft stone with a Bond Work Index of 5, the mill re-
quires 940 hp. The mill grinding the stone with a BWI of 14 requires ap-
proximately 2600 hp. Actual motor horsepowers will be higher due to motor
efficiency and gear drive losses for both cases.
Mill length and diameter are chosen based on the previously described
relationship between these dimensions and the required mill horsepower
(Eq. 4). Each mill manufacturer has his own application philosophy which
dictates the length to diameter (L/D) ratio for a given set of conditions.
7-113
-------
However, it is known that as the diameter of the mill is increased for a
given application, the specific energy to grind the material to the desired
product size decreases. There is a limitation to this effect, though. For
limestone preparation in FGD applications, the mill length is usually 15 to
30 percent larger than mill diameter.
A representative size for a mill which requires 940 hp for grinding is
14 feet in length and 11 feet-6 inches in diameter. And a mill requiring
2600 hp for grinding is 18 feet in length and 15 feet-6 inches in diameter.
In terms of annual power costs, the mill grinding the softer stone will con-
sume 2.0 x 106 kW hrs. of electricity (65% plant capacity factor) costing
$90,000 annually, and the mill grinding the harder stone will consume 5.6 x
106 kW hrs. of electricity costing $250,000 annually. Plant power costs are
assumed to be 45 mils/kW-hr.
Based on these examples, limestone grindability has a significant effect
on mill circuit economics. More detailed cost information is being generated
at the time of this paper as part of the FGD Reagent Preparation program.
The scope of the cost study covers not only the effect of stone grindability
but also the effect of product size distribution and type of grinding machine.
The results of the cost study will be available in the final report issued on
the Reagent Preparation Study.
FGD REAGENT PREPARATION STUDY
The preparation of lime and limestone reagents for conventional wet
scrubbing FGD systems has typically not been a unit process for which there
is readily available design inforamtion. To obtain information that utili-
ties could use in planning or designing new FGD systems or in optimizing
existing grinding equipment, EPRI initiated this study. This paper discusses
the results available to date on the phase of the program concerning the
preparation of limestone for wet FGD applications.
PROGRAM OBJECTIVES
The objective of this program was to obtain technical and economic data
useful for the selection and design of limestone grinding systems. The type
of equipment studied included the major designs available for commercially
grinding limestone. The size of the equipment tested was chosen so that
scale-up to commercial size was possible.
Both wet and dry media grinding were studied. The four basic grinding
machines examined were: 1) ball mill, 2) ring roller mill, 3) hammer mill,
and 4) tower mill.
RESULTS OF PILOT SCALE GRINDING TESTS
To gather design information to allow scale-up to commercial size, test-
ing was conducted at equipment manufacturers' test facilities using pilot-
scale equipment. The ball mill and ring roller mill tests were performed at
Kennedy Van Saun Corporation's Danville, .Pennsylvania test plant. The hammer
7-114
-------
mill testing took place at Mikropul Corporation's Summit, New Jersey test
plant. And the tower mill testing was conducted at the Koppers Company test
facility in York, Pennsylvania.
Three different limestones were tested to investigate the effect of lime-
stone strength or grindability on the performance and costs of the mills.
The limestones selected for this study were also investigated as part of the
Limestone Grindability program discussed earlier. They were the Kimmswick,
Salem, and Monteagle stones. The following sections discuss the results of
the testing in order of mill type.
Ball Mill Testing
The ball mill study included pilot-scale testing of:
• an open circuit wet ball mill,
• a wet ball mill in closed circuit with a hydrocyclone,
• a wet ball mill in closed circuit with a high frequency
vibrating screen, and
• an air-swept ball mill in closed circuit with a centrifugal
classifier.
The ball mill used for open and closed circuit wet ball milling tests was
an overflow discharge type mill having a shell inner diameter of 3 feet and
an effective length of 5 feet. During testing, the mill was fitted with cor-
rugated liners, giving an effective mill diameter of 2.69 feet. The mill was
driven by a 15 hp hydraulic motor. The grinding charge consisted of almost
3000 pounds of 2 inch, 1-1/2 inch, and 1 inch diameter steel balls. The mill
was operated at 34 rpm and a strain gauge was used to measure mill pinion
shaft torque which was used to compute mill pinion shaft power. The hydro-
cyclone used for wet, closed circuit ball mill testing had an inner cylinder
diameter of 3 inches, an apex diameter of 5/8 inch, and a vortex finder of
3/4 inch diameter. Two screen cloths were used with the high frequency vi-
brating screen; one with a screen opening of 111 microns and the other with a
screen opening of 156 microns. The air-swept ball mill had an inner shell
diameter of 3-1/2 feet and a length of 5 feet. This mill was fitted with
corrugated liners also, giving it an effective inner diameter of 3 feet. The
mill was driven by a 25 hp motor. The grinding media consisted of about 4300
pounds of 1-1/2 inch, 1-1/4 inch, 1 inch, and 1/2 inch diameter steel balls.
The mill was operated at 34 rpm and mill power was determined from a watt
meter connected to the mill motor and corrected for drive train losses. The
classifier used to close circuit the mill was a centrifugal "H" classifier.
A fan was used to generate the air flow necessary to convey the ground stone
from the mill to classification.
The test variables for each test circuit were:
7-115
-------
Open Circuit Wet Ball Mill
• three limestone types
• three mill slurry percent solids
• three mill feed rates
,• two mill feed size distributions
Closed Circuit Wet Ball Mill with the Hydrocyclone and Vibrating
Screen
• three limestone types
Air-Swept Ball Mill
• one limestone type
• three air flow rates
In addition to gathering actual operating information on the performance
of the pilot wet ball mill operating in closed circuit with a hydrocyclone or
screen, this circuit was modelled using the fundamental mathematical rela-
tionships which have been developed describing grinding in ball mills. This
approach was taken for budget and time reasons since a comprehensive test
program would be very expens-ive, even at the pilot-scale. It was important
to verify the validity of the model and to do this, some testing was geared
specifically toward validation. Laboratory testing was also included in the
program to determine the breakage kinetics parameters for each limestone.
The data analysis and assessment methodology consisted of comparing data
generated from the actual tests performed on the wet ball mill with model-
simulated data to demonstrate the validity of the model. Then, the model was
used to generate further data over a broader range of operating conditions.
These results were then analyzed to determine the important relationships
used to evaluate ball mill performance which are those between specific
grinding energy, throughput, and product size.
The model was validated for each of the three limestones at various
throughputs and at two levels of feed size for the Salem and Monteagle
stones. Figure 3 presents one comparison between actual test results and
model predicted values for the Kimmswick stone. This level of agreement was
also seen for the other two stones which satisfactorily validated the ability
of the model to predict the performance of the 3 ft. x 5 ft. wet ball mill
over a wide range of operating conditions.
Figures 4, 5 and 6 present the results of the open and closed circuit
ball mill simulations for the Kimmswick, Salem, and Monteagle stones, re-
spectively. The results display the relationship between specific energy and
product particle size passing 200 mesh. Specific energy is defined as mill
7-116
-------
o
o
CM
i
H
Ul
H
u
a
(d
Oi
EH
1
O
at
a.
100
90 .
80
70
60
50
40
O EXPERIMENTAL
PREDICTED
0.2
0.4
0.6
0.8 1.0 1.2
PRODUCTION RATE (STON/HR)
Figure 3. Production Rate vs. Product Fineness for Kimmswick Limestone
Grinding in the KVS 3 ft. x 5 ft. Wet, Open Circuit Ball Mill
5
100
1 90
80
« 70
en
60
50
40
30
0 10
1. Open Circuit
2. Closed Circuit with
Cyclone
3. Closed Circuit with
Screen
20
30 40
50
60 70
80
90 100
SPECIFIC ENERGY, E (hp-hr/ston)
Figure 4. Specific Energy Requirements for Size Reducing 3/8" x 0 Kimmswick
Limestone in the Wet Ball Mill
7-117
-------
o
z
<
CO
Crt
3
M
§
OS
o.
100
90
80
70
60
50
40
30
1. Open Circuit
2. Closed Circuit With
Cyclone
3. Closed Circuit With
Screen
10 20 30 40 50 60 70 80
SPECIFIC ENERGY, E (hp-hr/ston)
90
100
Figure 5. Specific Energy Requirements for Size Reducing 3/8" x 0 Salem
Limestone in the Wet Ball Mill
3
Id
g
U
a.
S
a.
100
90
80
70
60
50
40
30
1. Open Circuit
2. Closed Circuit With
Cyclone
3. Closed Circuit With
Screen
10 20 30 40 50 60 70 80
SPECIFIC ENERGY, E (hp-hr/ston)
90
100
Figure 6. Specific Energy Requirements for Size Reducing 3/8" x 0
Monteagle Limestone in the Wet Ball Mill
7-118
-------
drive shaft power divided by circuit throughput. The results show the marked
advantage of closed circuit grinding compared to open circuit grinding for all
three stones. Recall that a reduction in specific energy not only decreases
the mill power consumption but also the size of the mill to produce equivalent
rates.
About 70 hp-hr/ton are required to grind the Kimmswick stone from a 3/8"
x 0 feed size to a product size distribution of which 90 percent would pass
200 mesh with the mill operating in an open circuit mode. With the harder
Monteagle stone, the open circuit mill would require about 100 hp-hr/ton to do
the same reduction. Closing the circuit with either hydrocyclones or screens
reduced the specific energy for the Kimmswick stone to 10 to 12 hp-hr/ton,
depending on the type of classifier. A similar reduction was seen for the
Monteagle. This improvement is due to the classifier:
1) removing coarse particles from the mill product and returning
them for further grinding, and
2) removing particles from the circuit that are already fine
enough, thereby preventing bvergrinding.
The screen classifier gave slightly better performance than the hydrocyclone
based on limited classifier testing performed at the KVS test facility. The
screens were able to give a sharper classification than the hydrocyclone.
The result is that slightly lower (by about 10 percent) specific energy re-
quirements were seen for equivalent product sizes.
The results also show the difference between the required energy to pro-
duce a given particle size due to stone grindability. Comparing the Kimms-
wick and Monteagle stones, the model predicts about 12 hp-hr/ton are required
for the softer Kimmswick and 16 hp-hr/ton for grinding the harder Monteagle
stone to a 90 percent passing 200 mesh size. Using the Bond Work Index
method, the energy predicted to grind the Kimmswick from a similar feed size
to product size is 14 hp-hr/ton and for the Monteagle it is 25 hp-hr/ton.
The lack of agreement between the model and Bond methods for the Monteagle
stone is probably due to the inability of the Bond method to account for the
greater effect of feed size on mill productivity for a hard stone versus a
soft stone.
Table 2 compares the results of the model simulating grinding two differ-
ent feed sizes of the Monteagle and Kimmswick to a product of which 90 percent
passes 200 mesh using a hydrocyclone. It is seen from this table that de-
creasing the feed size for a "hard" limestone such as the Monteagle causes a
significant decrease in mill power relative to the same change for a "soft"
limestone such as the Kimmswick.
Air-Swept Ball Mill Test Results
The test work performed on the pilot air-swept ball mill was much more
limited than the testing with the wet ball mill. The reason for this is that
the application of this type of grinding device for preparing limestone for
FGD use is almost nonexistent. When compared to wet ball milling in
7-119
-------
TABLE 2. EFFECT OF LIMESTONE FEED SIZE ON MILL SPECIFIC ENERGY TO
PRODUCT 90 PERCENT PASSING 200 MESH
Limestone
Monteagle
Kimmswick
95 Percent Passing, In.
3/8
1/4
3/8
1/4
Specific Energy,
16.1
13.7
12.7
12.2
hp-hr/ton
identical situations, the dry mill requires approximately 30 percent greater
energy input according to past experience. The reason for this is that ground
material coats the balls inside the mill and cushions the ball-to-ball impact,
reducing grinding efficiency. Also, there is no advantage to producing the
limestone in a dry form, since it must be eventually added to an aqueous
slurry in the scrubber reaction tank. In addition, the feed material to the
mill must be dried completely to prevent caking and packing of the limestone
inside the mill. Finally, grinding limestone to very fine particle sizes
typical of that required for limestone FGD systems even further decreases the
grinding efficiency of the dry ball mill compared to a wet mill. This is
caused by agglomeration of the already finely ground material, preventing
ball-to-ball nipping of individual particles which is the primary size reduc-
tion mechanism.
Table 3 presents the results of the dry ball mill tests which were made
using the Monteagle limestone. The Monteagle stone was chosen since the
harder limestone would present less of a problem for the mill in terms of the
powder cushioning effect.
TABLE 3. AIR-SWEPT BALL MILL TEST RESULTS
Mill Throughput,
Ibs/hr
Air Flow, acfm % Passing 200 Mesh
Specific Energy,
hp-hr/ton
300
580
1030
1000
1500
1900
90
78,
71
188
. 101
61
The results of the tests indicate that the air-swept ball mill was capable of
producing a suitable FGD' product size distribution at very high specific
energies. The reason for high energy requirements could be due to too low of
an air flow to either effectively convey the limestone out,of the mill or to
promote efficient classification in the classifiers. However, the fan power
represented at least 25 percent of the total power requirements. Therefore,
while increasing fan power may increase classifier and overall mill circuit
efficiency, the specific energy required by an air-swept ball mill is expected
to be substantially higher than the closed circuit wet ball mill.
Ring Roller Mill Test Results
The grinding media for the ring roller mill was 5 inch steel balls.
The mill has an adjustable internal classifier, three adjustable load springs
7-120
-------
and is driven by a fifteen hp motor through a V-belt drive. Mill power was
determined from a watt meter connected to the mill motor. The ring roller
mill grinds under dry conditions. Air is used to sweep fines from the mill.
Testing with the ring roller mill was limited for the same reasons as
was the dry ball mill testing. Table 4 presents results of the ring roller
mill tests with the Kimmswick and Monteagle stones.
TABLE 4. RING ROLLER MILL TEST RESULTS
Limestone
Monteagle
"
it
Kimmswick
11
ti
Mr Flow
acfm
840
1140
1640
820
1100
1600
Throughput
Ibs/hr
220
470
680
260
440
1060
% Passing
200 mesh
57
51
37
61
55
49
Specific Energy
hp-hr/ton
105
54
45
87
55
30
The results indicate that the ring roller mill was not capable of producing
an FGD quality grind, even at very high specific energies. As with the air-
swept ball mill, this may be due to insufficient air flow to convey the fines
from the mill and/or poor classifier performance.
Hammer Mill Testing
A fine grinding hammer mill was tested at Mikropul Corporation's Summit,
New Jersey test facility to investigate its ability to produce a limestone
size suitable for FGD use. The limestone is ground by contacting fresh feed
with a high speed rotor to which are attached fixed "T" head hammers. Rotor
speed was varied over a range of from 5000 to 7000 rpm, giving extremely high
peripheral speeds. Air sweeps the fine material from the grinding chamber
upward to a centrifugal type mechanical classifier which may be operated at
various speeds.. In addition, the length of the classifier blades can be
changed to investigate the effect of blade length on system performance.
Rotor amps, classifier amps and fan power were monitored to calculate total
system horsepower. Fan power was approximately one-half of total system
power.
All three test stones (the Monteagle, Salem, and Kimmswick) were studied
in the hammer mill. Mill operating conditions which were varied included
rotor speed, classifier speed, and classifier blade length. Table 5 presents
selected results from the testing with Monteagle and Kimmswick limestone.
The results show that the hammer mill was capable of producing a very
fine grind (100 percent passing 200 mesh) for both the Monteagle and Kimms-
wick stones. Reducing rotor speed, classifier speed, and classifier blade
length resulted in a coarser product, but the resulting increase in through-
put lowered the specific energy. There was a noticeable effect on specific
energy due to stone strength. Specific energy was anywhere from 20 to 40
7-121
-------
TABLE 5. HAMMER MILL TEST RESULTS
Rotor Speed
Limestone rpm
Monteagle
it
ti
ii
ti
it
Kimmswick
ti
ii
it
M
it
6000
6000
5000
5000
6000
5000
6000
6000
5000
5000
6000
5000
Classifier Speed Blade % Passing Specific Energy
rpm Length 200 Mesh hp-hr/ton
2000
1000
2000
1000
1000
2000
2000
1000
2000
1000
1000
2000
Long
Long
Long
Long
Short
Short
Long
Long
Long
Long
Short
Short
100
93
99
88
78
98
100
98
99
93
81
99
68
48
69
44
43
47
56
33
58
30
31
27
percent lower when grinding the Kimmswick stone compared to the Monteagle.
The hammer abrasion index was measured for the Monteagle and Kimmswick stones
also. The Monteagle index was 175 microns/5 pounds of material ground and the
Kimmswick index was 105 microns/5 pounds. These numbers are useful in pre-
dicting the wear rate and therefore the life of the hammers. In general, the
specific energies which were measured for each test were much higher than
those for equivalent size products produced in closed circuit wet ball mill-
ing.
The test with the Kimmswick stone at a rotor rpm of 5000, classifier rpm
2000, and with short classifier blades that produced a grind at a specific
energy that approaches to within a factor of 2 the performance of the closed
circuit wet ball mill. This suggests that with further tuning, the hammer
mill may be able to produce an equivalent particle size with specific ener-
gies approaching those required in the wet ball mill. It should be noted
that this was seen only for the softest stone, though.
Tower Mill Testing
A pilot-scale tower mill was tested at Koppers Company, York, Pennsyl-
vania facility. The tower mill is a wet grinding device that relies on a
ball charge to accomplish the necessary breakage and size reduction of a
particular material. However, unlike the ball mill which tumbles the ball
charge, the tower mill consists of a vertical cylindrical chamber which con-
tains the ball charge. Fresh feed and classifier oversize material is fed to
the top of the chamber. A rotating screw flight agitates the ball charge,
which grinds the limestone in this case by a combination of ball-to-ball,
ball-to-screw, and pebble-to-pebble attrition. The finely ground particles
that are generated are carried upward through the chamber and overflow into
the coarse classifier. Overflow from this classifier is fed to a spiral
classifier. The product is taken from the spiral classifier, and oversize
material is recycled back to the grinding chamber. Mill throughput and
classifier operation were varied to produce different product size
7-122
-------
distributions for the Monteagle, Salem, and Kimmswick stones tested. The test
approach taken with the tower mill was somewhat different than with the other
mills. Instead of running a comprehensive evaluation of mill operating condi-
tions, the circuit was controlled to produce specific size distributions for
direct comparison to the closed circuit wet ball mill. Table 6 presents the
results of the tests.
TABLE 6. TOWER MILL TEST RESULTS
Product Size
Limestone
% Passing 200 Mesh
% Passing 325 Mesh
Specific Energy
hp-hr/ton
Kimmswick
Kimmswick
Salem
Salem
Monteagle
Monteagle
93
100
93
100
97
99
65
88
68
88
70
85
6
11
6
12
10
14
Examining the results in the table indicate that the tower mill is capa-
ble of producing a very fine particle size (99+ percent passing 200 mesh) at
what are lower specific energies than the closed circuit wet ball mill. As
with the other mills, stone strength also affects the specific energy required
to achieve a given product size for the tower mill. Also, significantly
greater energy is required to produce a finer particle size, as was expected.
LABORATORY TESTING
The work at CILCo's Duck Creek Station (discussed in the next section)
demonstrated the beneficial effect that a fine particle size distribution has
on full-scale FGD performance. The major objective of this laboratory work
was to document the effect that different grinding devices can have on FGD
system performance exclusive of particle size effects. The hypothesis tested
was that different mills would produce products with different fissure char-
acteristics due to different breakage mechanisms. Different fissures might
result in different limestone dissolution rates, even for samples with iden-
tical particle size. If this is true, then a mill which creates a faster
dissolving product would not have to be designed to obtain the same size
product as a mill which creates a slower dissolving product.
Approach
The approach taken was to use laboratory-scale equipment to determine
the effect different grinding mills would have on FGD performance. Two dif-
ferent bench-scale systems were employed: 1) a limestone reactivity setup
developed previously for EPRI (Figure 7), and 2) a bench-scrubber arrangement
which scrubs a simulated flue gas (Figure 8).
The nucleus of the reactivity apparatus is a stirred tank reacter with
pH and temperature control. For each test, a known amount of limestone was
added and allowed to dissolve while maintaining reactor solution pH at the
7-123
-------
Craduaced
Feed Tank
Variable
Speed
Scirrer
Mecering
Pump
r
0
ccor
~xxv^
1 —
Thermos cat
Ouclec
Sample
Concainer
Figure 7. Limestone Reactivity Apparatus
Figure 8. Schematic Diagram of the Bench-Scale FGD System
desired setpoint by controlled addition of 0.025N HC1 stock solution. Liquid
is withdrawn through a filter at the same rate the acid solution is added,
thus creating a liquid flow-through, batch-solids reactor. By monitoring the
outlet flow rate and calcium concentration, the cumulative calcium dissolved
and the dissolution rate can be computed as a function of time. Dissolution
rate reported at 50 percent of the limestone dissolved has been a useful
method of comparing different limestone types and particle size distributions.
7-124
-------
The bench-scale wet scrubber equipment consisted of a one-inch diameter
countercurrent gas-liquid contactor and a two-liter reaction tank for lime-
stone dissolution and sulfur salt precipitation. A simulated flue gas con-
taining S02, 62, N2, and C02 was fed to the bottom of the contactor at a
controlled rate of 0.5 scfm. Slurry from the reaction tank was pumped to the
contactor at a rate equivalent to an L/G of approximately 50 gal/1000 acf.
Temperature of the reaction tank slurry was maintained at 50°C (122°F) by an
external heater. Limestone was fed to the reaction tank as a 20 percent
slurry to maintain a pH of 5.8. The duration of each run was at least six
hours, with some runs being extended to three days. Limestone utilization
was the primary measurement made to document the effect of particle size on
FGD system performance since the S02 removal was roughly equivalent for all
5.8 pH runs.
The first series of experiments were designed to show that the lime-
stone reactivity and bench scrubber test results are consistent with
trends measured at the full-scale facility. Once this was demonstrated, the
remainder of the experimental work was conducted on the bench reactivity ap-
paratus.
Results
A comparison of bench-scale and full-scale test results are reported in
Table 7. The trends measured in the lab are consistent with those measured
at CILCo, although the bench scrubber does not exactly duplicate the CILCo
limestone utilization. The range in utilization was 58 to 93 percent (coarse
to fine) at CILCo and 67 to 80 in the bench scrubber. At CILCo, the medium
and fine grinds achieved very similar results, while the bench scrubber re-
sults showed the medium grind to be more of a mid-point in utilization. The
bench reactivity tests showed the dissolution rate (measured at 50 percent
limestone dissolved) of the fine and medium grinds to be very close and sub-
stantially greater than that for the coarse grind. This trend is similar to
the full-scale results. >
TABLE 7. COMPARISON OF BENCH-SCALE AND FULL-SCALE
LIMESTONE* UTILIZATION RESULTS
Limestone Size
(% passing 325 mesh)
90 (fine)
80 (medium)
70 (coarse)
Bench Reactivity
Limestone
Dissolution Rate
at 50% Dissolved
(mg/min)
19.2
18.8
11.7
Bench Scrubber
Limestone
Utilization
(%)
80
75
67
CILCo
Limestone
Utilization
(%)
93
89
58
*A11 tests conducted with Kimmswick limestone and 5.8 pH setpoint.
Once the bench reactivity test had been demonstrated to reflect trends
observed in full-scale systems, a test matrix was set up to measure differ-
ences in limestone ground in different mills. The following variables were
investigated:
7-125
-------
Limestone Source - Monteagle (hard)
Kimmswick (soft)
Size Fraction - +325, -200 mesh (^56 y mass mean diameter)
+400, -325 mesh (^44 p mass mean diameter)
Grinding Mill Type - wet ball mill
tower mill
air-swept ball mill
ring roller mill
hammer mill
The results of the tests are summarized in Table 8. The only variable
found to have a statistically significant effect on dissolution rate was the
particle size. Neither the type of mill nor the type of limestone was found
to have a major effect on the dissolution rate. Consider, for instance, the
Kimmswick 325-400 mesh runs. The range of dissolution rates for three differ-
ent mills was 5.2 to 5.5 mg/min with an 0.2 difference found in a duplicate
set of runs. The same size fraction with the Monteagle limestone yielded a
5.0 to 5.4 mg/min range in dissolution rate. On the other hand, the 200-325
mesh samples gave 3.3 to 3.8 mg/min dissolution rates for both limestones and
four different mill types.
TABLE 8. RESULTS OF LIMESTONE REACTIVITY TESTS
Dissolution Rate
Mesh Size Limestone Source Mill Type mg/min at 50% dissolved
325-400
325-400
325-400
325-400
325-400
325-400
325-400
325-400
200-325
200-325
200-325
200-325
200-325
200-325
Monteagle
Monteagle
Monteagle
Monteagle
Kimmswick
Kimmswick
Kimmswick
Kimmswick
Monteagle
Monteagle
Monteagle
Monteagle
Kimmswick
Kimmswick
ASBM
Hammer
WBM
Tower
WBM
Tower
Tower
RRM
Hammer
WBM
WBM
Tower
Hammer
Tower
5.4
5.0
5.2
5.3
5.3
5.4
5.2
5.5
3.4
3.6
3.6
3.8
3.3
3.7
Based on these results, it appears that a limestone grinding mill should
be selected based on capital and operating costs required to produce a reason-
ably fine limestone particle size distribution (80 to 90 percent passing 325
mesh). It does not appear that differences in breakage mechanisms of the
mills considered in this study had a measurable effect on FGD system perfor-
mance when equivalent limestone size fractions were tested.
7-126
-------
CILCO DUCK CREEK FULL SCALE TESTING
In 1982, the Central Illinois Light Company (CILCo) initiated a program
to improve performance of the FGD system at its Duck Creek Station. As part
of the overall effort, Radian, EPRI and CILCo decided that improving limestone
grindability circuit performance should be examined as a means of improving
FGD system performance. Three aspects of the FGD system were used as guide-
lines to measure the performance: SC>2 removal, limestone consumption, and
mist eliminator scaling.
System Description
The Duck Creek Station generates about 400 MW of electrical power while
at full load. The four rod-deck scrubber modules treat about 1.4 million acfm
(130°F, saturated) of flue gas which contains about 2500 ppm S02 prior to
entering the scrubbers. Reagent preparation and waste disposal are common to
all modules. Figure 9 shows a simplified flow schematic.
Mist Eliminator
Wash Water
Reclaimed
Water
To
Absorbers
And Ash Sluice
Figure 9. Duck Creek FGD System
A 10 ft. x 18 ft. wet ball mill (rated capacity of 40 tons per hour)
grinds the limestone, and four 10-inch diameter hydrocyclones were initially
used to control product size distribution. Product slurry is stored in a
125,000 gallon tank prior to being fed to the scrubber reaction tanks.
Columbia Quarry supplies the limestone, which is mined from the Kimmswick
formation.
7-127
-------
A pH feedback loop controls the limestone feed rate to the reaction
tanks. The pH setpoint in each tank can be maintained within +0.05 units
with the on-off controller. The four 150,000 gallon reaction tanks provide
necessary time for dissolution of the limestone and precipitation of calcium-
sulfur salts. The slurry solids level is maintained at about 12 percent by
bleeding slurry to the disposal pond and returning clarified liquor from the
pond.
Reaction tank slurry is pumped to the top of the scrubbers at a rate of
about 15,000 gpm. The slurry contacts the flue gas countercurrently with the
seven rod-decks providing the turbulence required to achieve S02 removal. At
full-load conditions, the liquid-to-gas ratio is about 40 gal/1000 acf, and
the gas-side pressure drop is about 12 inches w.g.
A two-stage horizontal mist eliminator is located downstream of the last
rod deck in each absorber to remove entrained slurry. Periodic washing re-
moves collected solids. Flue gas exits the system without reheat through a
lined wet stack.
Approach
The approach taken in this phase of the program was to modify grinding
circuit operating variables, monitor the effect on limestone product particle
size distribution, and then measure the effect the change in limestone product
had on FGD system performance. Limestone mill/classifier variables which were
examined include:
mill throughput
mill slurry solids
level of ball charge in mill
top-size of balls in mill
raw limestone feed size
diameter of classifier barrel
classifier pressure drop (or throughput)
• solids content in classifier feed
• diameter of vortex finder and apex
The only equipment modifications were the changes in the vortex finder,
apex, and classifier barrel. Total capital expenditures were only about
$7,000.
Results
Time did not permit extensive parametric testing, so it is not possible
to isolate and measure the effect of each variable on mill/classifier system
performance. However, the changes which had the greatest effect on limestone
product particle size at Duck Creek were all related to the classifier:
reduction in classifier barrel diameter from 10 inches
to 6 inches
reduction in the slurry solids in the classifier feed from
^50 to ^45 wt%.
7-128
-------
increase in classifier pressure drop from 15 to 30 psi (in-
crease in angular velocity within hydrocyclone)
The modifications in mill/classifier operation resulted in a finer lime-
stone product particle size distribution. Prior to any changes, 86 percent of
the product passed a 200 mesh screen and 73 percent passed 325 mesh screen.
After the system had been optimized, 99 percent of the product passed 200 mesh
and 90 percent passed 325 mesh.
The effect the finer limestone had on FGD system performance is summari-
zed in Table 9. At a pH setpoint of 5.8, the limestone utilization increased
from below 60 percent to above 90 percent as the product size decreased. An
alternate method of summarizing the results shows that S02 removal (measured
at 85 percent limestone utilization) increased from about 72 percent to about
84 percent as the grind improved.
TABLE 9. RESULTS OF LIMESTONE GRINDING SYSTEM OPTIMIZATION
Limestone S02 Removal at
Particle Size Utilization at 85% Utilization
Grind
Coarse
Medium
Fine
% <200 mesh
86
94
99
% <325 mesh
70
81
90
5.8 pH (%)
58
89
92
(%)
72
82
84
The improved mill/classifier performance allows the Duck FGD System to
operate at a higher pH while still maintaining better limestone utilization
than previous operation. The finer limestone particle size distribution re-
sulted in: 1) increased S0£ removal efficiency, 2) decreased limestone con-
sumption for equivalent time of operation, and 3) decreased scaling on the
mist eliminator.
Increased S02 Removal—With the finer limestone grind, the FGD system
could maintain a higher pH setpoint than originally designed. Prior to the
optimization, the design pH setpoint had been 5.6. The S02 removal at full
load was about 74-77 percent (82^55 percent required by regulation). The fine
grind allows the system to operate at 5.8 pH and achieve 82-84 percent S02 re-
moval at full load.
Decreased Limestone Consumption—Prior to the mill/classifier optimi-
zation, the 5.6 pH setpoint resulted in about 77 percent limestone utiliza-
tion. This will result in limestone consumption of 160,000 tons per year if
the plant operates at the CILCo goal of 75 percent capacity factor. (Note
also that the S02 removal is below that required by regulation under these
conditions). After optimization, the limestone utilization was 93 percent at
5.8 pH, which translates to 144,000 tons per year of limestone consumed. The
savings of 16,000 tons per year of limestone is worth about $220,000 per year.
Decreased Mist Eliminator Scaling—Prior to the mill/classifier optimiza-
tion program, the mist eliminators had to be manually cleaned every four days
to prevent excessive solids buildup. This cleaning required about 400 main-
tenance hours per week, and part of the flue gas had to bypass the FGD system
7-129
-------
because of leaking isolation dampers. The improved limestone utilization
which resulted from the finer limestone, combined with improved wash water
quality, resulted in greatly decreased mist eliminator fouling. Both good
limestone utilization (>85%) and wash water subsaturated with respect to gyp-
sum were required. Neither can be used solely to improve mist eliminator
fouling. These improvements virtually eliminated the need for manual clean-
ing of the mist eliminators.
CONCLUSIONS
Conclusions drawn from the limestone grindability and limestone reagent
preparation programs can be summarized as follows:
• Grindability can vary widely from one limestone formation to
the next.
The limestone grindability lab method correlates well with
the Bond Work Index as a measure of limestone strength.
Other physical properties did not correlate strongly with
BWI.
Grindability (BWI) is widely used to size ball mills for
FGD application. The extremes of stone strength found in
the grindability study result in a factor of 3 difference
in the operating costs (specific energy) and a significant
difference in size of mill required.
BWI does not effectively account for the effect changes
in feed size of harder limestones has on specific energy
requirements.
The tower mill and wet ball mill operating in a closed
circuit offer the least energy intensive options for grind-
ing limestone. Pilot investigations showed the tower mill
to have an advantage in specific energy requirements.
If designed to produce equivalent PSD's, the type of mill
does not affect limestone dissolution rate.
PSD is an extremely important design and operating variable
in limestone FGD systems. Scrubber S02 removal, limestone
consumption, and reliability are all affected.
REFERENCES
1. Flue Gas Desulfurization Chemistry Studies: Limestone Grindability,
Radian Corporation, Oct. 30, 1982.
2. Stern, Arthur L. , A Guide to Crushing and Grinding Practice, Chem.
Engineering, 69 (25), 129-146 (1962).
3. Austin, L. G., Understanding Ball Mill Sizing, Ind. Eng. Chem. Process
Des. Develop., Vol. 12, No. 2, (1973), 121-129.
7-130
------- |