EPA-600 /R- 9 3-0 64b
April 1993
o-EPA Research and
Development
PROCEEDINGS:
1991 S02 CONTROL SYMPOSIUM
Volume 2. Sessions 4 and 5 A
United States
Environmental Protection
Agency
Prepared for
Office of Air Quality Planning and Standards
Prepared by
Air and Energy Engineering Research
Laboratory
Research Triangle Park NC 27711
-------
TECHNICAL REPORT DATA
(Please read J/izin/eriotis on the reverse before coniplef'
1 REPORT NO. 7
EPA-600/R-93-064b
3.
4. TITLE ANOSUGTITLE
Proceedings: 1991 SC>2 Control Symposium, Volume 2.
Sessions 4 and 5A
6 REPORT DATE
April 1993
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Miscellaneous
8. PERFORMING ORGANIZATION REPORT NO.
TR-101054 (1)
9. PERFORMING OROANIZATION NAME AND ADDRESS
See Block 12
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
NA (Inhouse)
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Air and Energy Engineering Research Laboratory
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Proceedings; 1991
14. SPONSORING AGENCY CODE
EPA/600/13
15.supplementary notes AEERL project officer is Brian K. Gullett, Mail Drop 4, 919/541-
1534. Cosponsored by EPRI and DOE. Vol. 1 is Opening Session and Sessions 1-3,
Vol. 3 is Sessions 5B and 6, Vol. 4 is Session 7, and Vol. 5 is Session 8.
16. abstract p-poceedingg document the 1991 S02 Control Symposium, held December
3-6, 1991, in Washington, DC, and jointly sponsored by the Electric Power Research
Institute (EPRI), the U.S. Environmental Protection Agency (EPA), and the U.S. De-
partment of Energy (DOE). The symposium focused attention on recent improve-
ments in conventional S02 control technologies, emerging processes, and strategies
for complying with the Clean Air Act Amendments (CAAA) of 1990. It provided an in-
ternational forum for the exchange of technical and regulatory information on S02
control technology. More than 800 representatives of 20 countries from government,
academia, flue gas desulfurization (FGD) process suppliers, equipment manufac-
turers, engineering firms, and utilities attended. In all, 50 U. S. utilities and 10
utilities in other countries were represented. In 11 technical sessions, speakers
presented 111 technical papers on development, operation, and commercialization of
wet and dry FGD, clean coal technologies, and combined sulfur oxide/nitrogen oxide
(SOx/NOx) processes.
17. KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Pollution
Sulfur Dioxide
Nitrogen Oxides
Flue Gases
De sulf ur iz ation
Coal
Pollution Control
Stationary Sources
13 B
07B
21B
07A, 07D
21D
18. DISTRIBUTION STATEMENT
Release to Public
19. SECURITY CLASS (This Report)'
Unclassified
21. NO. OF PAGES
548
20. SECURITY CLASS (Thispage)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73) 5A~ 155
-------
EPA-600/R-93-064b
April 1993
Proceedings: 1991S02 Control Symposium
Volume 2. Sessions 4 and 5A
Electric Power Research Institute
B. Toole O'Neil
3412 Hillview Avenue
Palo Alto, CA 94304
For Sponsors:
U.S. Department of Energy
Charles J. Drummond
Pittsburgh Energy
Technology Center
P.O. Box 10940
Pittsburgh, PA 15236
U.S. Environmental Protection Agency
Brian K. Guile tt
Air and Energy Engineering
Research Laboratory
Research Triangle Park, NC 27711
-------
ABSTRACT
These are the Proceedings of the 1991 SO2 Control Symposium held December 3-6,
1991, in Washington, D.C. The symposium, jointly sponsored by the Electric Power
Research Institute (EPRI), the U.S. Environmental Protection Agency (EPA), and the
U.S. Department of Energy (DOE), focused attention on recent improvements in
conventional sulfur dioxide (SO2) control technologies, emerging processes, and
strategies for complying with the Clean Air Act Amendments of 1990. This is the
first SO2 Control Symposium co-sponsored by EPRI, EPA and DOE. Its purpose was
to provide a forum for the exchange of technical and regulatory information on SO2
control technology.
Over 850 representatives of 20 countries from government, academia, flue gas
desulfurization (FGD) process suppliers, equipment manufacturers, engineering
firms, and utilities attended. In all, 50 U.S. utilities and 10 utilities in other
countries were represented. A diverse group of speakers presented 112 technical
papers on development, operation, and commercialization of wet and dry FGD,
Clean Coal Technologies, and combined sulfur dioxide/nitrogen oxides (SO2/NOX)
processes. Since the 1990 SO2 Control Symposium, the Clean Air Act Amendments
have been passed. Clean Air Act Compliance issues were discussed in a panel
discussion on emission allowance trading and a session on compliance strategies for
coal-fired boilers.
ii
-------
CONTENTS
PREFACE xi
AGENDA xii
VOLUME 1
Opening Session
EPRI Perspective OS-1
EPA Perspective OS-5
DOE Perspective OS-9
Guest Speakers
Shelley Fidler - Assistant, Policy Subcommittee on
Energy and Power, U.S. Congress OS-11
Jack S. Siegel - Deputy Assistant Secretary, Office of Coal
Technology, U.S. Department of Energy OS-19
Michael Shapiro - Deputy Assistant Administrator, Office
of Air and Radiation, U.S. Environmental Protection Agency OS-29
Session 1 - Clean Air Act Compliance Issues/Panel 1-1
Session 2 - Clean Air Act Compliance Strategies
Scrubbers: A Popular Phase 1 Compliance Strategy 2-1
Scrub Vs. Trade: Enemies or Allies? 2-21
Evaluating Compliance Options 2-39
Clean Air Technology (CAT) Workstation 2-49
Economic Evaluations of 28 FGD Processes 2-73
Strategies for Meeting Sulfur Abatement Targets in the
UK Electricity Supply Industry 2-93
iii
-------
Compliance Strategy for Future Capacity Additions: The Role of
Organic Acid Additives
A Briefing Paper for the Status of the Flue Gas Desulfurization
System at Indianapolis Power & Light Company
Petersburg Station Units 1 and 2
Evaluation of SO2 Control Compliance Strategies at Virginia Power
Session 3A - Wet FGD Process Improvements
Overview on the Use of Additives in Wet FGD Systems
Results of High SO2 Removal Efficiency Tests at EPRI's High
Sulfur Test Center
Results of Formate Ion Additive Tests at EPRI's High Sulfur
Test Center
FGDPRISM, EPRI's FGD Process Model-Recent Applications
Additive-Enhanced Desulfurization for FGD Scrubbers
Techniques for Evaluating Alternative Reagent Supplies
Factors Involved in the Selection of Limestone Reagents for Use in
Wet FGD Systems
Magnesium-Enhanced Lime FGD Reaction Tank Design Tests
at EPRI's HSTC
Session 3B - Furnace Sorbent Injection
Computer Simulations of Reacting Particle-Laden Jet Mixing
Applied to SO2 Control by Dry Sorbent Injection
Studies of the Initial Stage of the High Temperature
Ca0-S02 Reaction
Status of the Tangentially Fired LIMB Demonstration Program
at Yorktown Unit No. 2: An Update
Results from LIMB Extension Testing at the Ohio Edison
Edgewater Station
iv
-------
VOLUME 2
Session 4A - Wet FGD Design Improvements
Reliability Considerations in the Design of Gypsum Producing
Flue Gas Desulfurisation Plants in the UK
Sparing Analysis for FGD Systems
Increasing Draft Capability for Retrofit Flue Gas Desulfurization
Systems
Development of Advanced Retrofit FGD Designs
Acid Rain FGD System Retrofits
Guidelines for FGD Materials Selection and Corrosion Protection
Economic Comparison of Materials of Construction of Wet FGD
Absorbers and Internals
The Intelligence & Economics of FRP in F.G.D. Systems
Session 4B - Dry FGD Technologies
LIFAC Demonstration at Poplar River
1.7 MW Pilot Results for the Duct Injection FGD Process Using
Hydrated Lime Upstream of an ESP
Scaleup Tests and Supporting Research for the Development
of Duct Injection Technology
A Pilot Demonstration of the Moving Bed Limestone Emission
Control (LEC) Process
Pilot Plant Support for ADVACATE/MDI Commercialization
Suitability of Available Fly Ashes in AD VAC ATE Sorbents
Mechanistic Study of Desulfurization by Absorbent Prepared
from Coal Fly Ash
Results of Spray Dryer/Pulse-Jet Fabric Filter Pilot Unit Tests
at EPRI High Sulfur Test Center
-------
Results of Medium- and High-Sulfur Coal Tests on the TVA
10-MW SD/ESP Pilot Plant
Evolution of the B&W Durajet™ Atomizer
Characterization of the Linear VGA Nozzle for Flue
Gas Humidification
High SO2 Removal Dry FGD Systems
Session 5A - Wet Full Scale FGD Operations
FGD System Retrofit for Dalhousie Station Units 1 & 2
Zimmer FGD System: Design, Construction, Start-Up
and Operation
Results of an Investigation to Improve the Performance and
Reliability of HL&P's Limestone Electric Generating Station
FGD System
Full-Scale Demonstration of EDTA and Sulfur Addition to
Control Sulfite Oxidation
Optimizing the Operations in the Flue Gas Desulfurization Plants
of the Lignite Power Plant Neurath, Unit D and E and Improved
Control Concepts for Third Generation Advanced FGD Design
Organic Acid Buffer Testing at Michigan South Central Power
Agency's Endicott Station
Stack Gas Cleaning Optimization Via German Retrofit Wet
FGD Operating Experience
Operation of a Compact FGD Plant Using CT-121 Process
VOLUME 3
Session 5B - Combined SOx/NOx Technologies
Simultaneous SOx/NOx Removal Employing Absorbent Prepared
from Fly Ash
Furnace Slurry Injection for Simultaneous SO2/NOX Removal
Combined SO2/NOX Abatement by Sodium Bicarbonate
Dry Injection
vi
-------
SC>2 and NOx Control by Combined Dry Injection of Hydrated
Lime and Sodium Bicarbonate 5B-67
Engineering Evaluation of Combined NOx/SC>2 Controls for
Utility Application 5B-79
Advanced Flue Gas Treatment Using Activated Char Process
Combined with FBC 5B-101
Combined SO2/NOX Control using Ferrous*EDTA and a
Secondary Additive in a Lime-Based Aqueous Scrubber System 5B-125
Recent Developments in the Parsons FGC Process for Simultaneous
Removal of SOx and NOx 5B-141
Session 6A - Wet FGD Operating Issues
Pilot-Scale Evaluation of Sorbent Injection to Remove SO3 and HC1 6A-1
Control of Acid Mist Emissions from FGD Systems 6A-27
Managing Air Toxics: Status of EPRI PISCES Project 6A-47
Results of Mist Eliminator System Testing in an Air-Water
Pilot Facility 6A-73
CEMS Vendor and Utility Survey Databases 6A-95
Determination of Continuous Emissions Monitoring
Requirements at Electric Energy, Inc. 6A-117
Improving Performance of Flushless Mechanical Seals in Wet FGD
Plants through Field and Laboratory Testing 6A-139
Sulcis FGD Demonstration Plant Limestone-Gypsum Process:
Performance, Materials, Waste Water Treatment 6A-163
Session 6B - dean Coal Demonstrations
Recovery Scrubber - Cement Application Operating Results 6B-1
The NOXSO Clean Coal Technology Demonstration Project 6B-17
vii
-------
Economic Comparison of Coolside Sorbent Injection and Wet
Limestone FGD Processes 6B-33
Ohio Edison Clean Coal Projects Circa: 1991 6B-55
Sanitech's 2.5-MWe Magnesia Dry-Scrubbing Demonstration
Project 6B-79
Application of DOW Chemical's Regenerable Rue Gas
Desulfurization Technology to Coal-Fired Power Plants 6B-93
Pilot Testing of the Cansolv® System FGD Process 6B-105
Dry Desulphurization Technologies Involving Humidification
for Enhanced SO2 Removal 6B-119
VOLUME 4
Session 7 - Poster Papers
Summary of Guidelines for the Use of FRP in Utility FGD
Systems 7-1
Development and Evaluation of High-Surface-Area Hydrated
Lime for SO2 Control 7-13
Effect of Spray Nozzle Design and Measurement Techniques on
Reported Drop Size Data 7-29
High SO2 Removals with a New Duct Injection Process 7-51
Combined SOx/NOx Control Via Soxal™, A Regenerative Sodium
Based Scrubbing System 7-61
The Healy Clean Coal Project Air Quality Control System 7-77
Lime/Lime Stone Scrubbing Producing Usable By-Products 7-93
Modeling of Furnace Sorbent Injection Processes 7-105
Dry FGD Process Using Calcium Sorbents 7-127
Clean Coal Technology Optimization Model 7-145
SNRB Catalytic Baghouse Process Development and Demonstration 7-157
Reaction of Moist Calcium Silicate Reagents with Sulfur Dioxide
in Humidified Flue Gas 7-181
viii
-------
Commercial Application of Dry FGD using High Surface Area
Hydrated Lime
Initial Operating Experience of the SNOX Process
Progress Report of the NIPSCO - Pure Air - DOE Clean Coal II
Project
Development of a Post Combustion Dry SO2 Control Reactor
for Small Scale Combustion Systems
Scrubber Reagent Additives for Oxidation Inhibited Scrubbing
Recovery of Sulfur from Calcium Sulfite and Sulfate
Scrubber Sludges
Magnesite and Dolomite FGD Technologies
SO2 and Particulate Emissions Reduction in a Pulverized Coal
Utility Boiler through Natural Gas Cofiring
Design, Installation, and Operation of the First Wet FGD for a
Lignite-Fired Boiler in Europe at 330 MW P/S Voitsberg 3 in Austria
VOLUME 5
Session 8A - Commercial FGD Designs
Mitsui-BF Dry Desulfurization and Denitrification Process
Using Activated Coke
High Efficiency, Dry Flue Gas SOx, and Combined SOx/NOx
Removal Experience with Lurgi Circulating Fluid Bed
Dry Scrubber - A New, Economical Retrofit Option for U.S.
Utilities for Acid Rain Remediation
Incorporating Full-Scale Experience into Advanced Limestone
Wet FGD Designs
Design and Operation of Single Train Spray Tower FGD Systems
Selecting the FGD Process and Six Years of Operating Experience
in Unit 5 of the Altbach-Deizisau Neckarwerke Power Station
Development and Operating Experience of FGD-Technique at the
Voelklingen Power Station
Advantages of the CT-121 Process as a Throwaway FGD System
ix
-------
Session 8B - By-Product Utilization
German Experience of FGD By-Product Disposal and Utilization 8B-1
The Elimination of Pollutants from FGD Wastewaters 8B-25
The Influence of FGD Variables.on FGD Performance and
By-Product Gypsum Properties 8B-47
Quality of FGD Gypsum 8B-69
Chemical Analysis and Flowability of ByProduct Gypsums 8B-91
Evaluation of Disposal Methods for Oxidized FGD Sludge 8B-113
Commercial Aggregate Production from FGD Waste 8B-127
x
-------
PREFACE
The 1991 SO2 Control Symposium was held December 3-6, 1991, in Washington,
D.C. The symposium, jointly sponsored by the Electric Power Research Institute
(EPRI), the U.S. Environmental Protection Agency (EPA), and the U.S. Department
of Energy (DOE), focused attention on recent improvements in conventional sulfur
dioxide (SO2) control technologies, emerging processes, and strategies for complying
with the Clean Air Act Amendments of 1990.
The proceedings from this Symposium have been compiled in five volumes,
containing 111 presented papers covering 14 technical sessions:
Session
Subject Area
1
Opening Remarks by EPRI,EPA and DOE Guest Speakers
1
Emission Allowance Panel Discussion
2
Clean Air Act Compliance Strategies
3A
Wet FGD Process Improvements
3B
Furnace Sorbent Injection
4A
Wet FGD Design Improvements
4B
Dry FGD Technologies
5A
Wet FGD Full Scale Operations
5B
Combined SOx/NOx Technologies
6A
Wet FGD Operating Issues
6B
Clean Coal Demonstratioins/Emerging Technologies
7
Poster Session - papers on all aspects of SO2 control
8A
Commercial FGD Designs
8B
FGD By-Product Utilization
These proceedings also contain opening remarks by the co-sponsors and comments
by the three guest speakers. The guest speakers were Shelley Fidler - Assistant,
Policy subcommittee on Energy and Power, U. S. Congress,
Jack . . S. Siegel - Deputy Assistant Secretary, Office of Coal Technology, U.S.
Department of Energy, and Michael Shapiro - Deputy Assistant Adminstrator,
Office of Air and Radiation, U. S. Environmental Protection Agency.
The assistance of Steve Hoffman, independent, in preparing the
manuscript is gratefully acknowledged.
The following persons organized this symposium:
• Barbara Toole O'Neil - Co-Chair, Electric Power Research Institute
• Charles Drummond - Co-Chair, U.S. Department of Energy
• Brian K. Gullett - Co-Chair, U.S. Environmental Protection Agency
• Pam Turner and Ellen Lanum - Symposium Coordinators, Electric Power
Research Institute
xi
-------
AGENDA
1991SO2 CONTROL SYMPOSIUM
Opening Session
Session Chain M. Maxwell - EPA
1-1
1-2
EPRI Perspective - S.M. Dalton
EPA Perspective - M. Maxwell
DOE Perspective - P. Bailey (no written manuscript)
Guest Speakers
Shelley Fidler - Assistant, Policy subcommittee on energy and
Power, U. S. Congress
Jack S. Siegel - Deputy Assistant Secretary, Office of Coal
Technology, U.S. Department of Energy
Michael Shapiro - Deputy Assistant Adminstrator, Office of Air
and Radiation, U. S. Environmental Protection Agency
Comments by:
Alice LeBlanc - Environmental Defense Fund
Karl Moor, Esq., Balch & Bingham
John Palmisano AER*X
Craig A. Glazer - Chair, Ohio Public Utilities Commission
Session 1 - Clean Air Act Compliance Issues/Panel
Session Moderator: S. Jenkins, Tampa Electric Co.
xii
-------
Session 2 -Clean Air Act Compliance Strategies
Session Chain Paul T. Radcliffe - EPRI
2-1 Scrubbers: A Popular Phase 1 Compliance Strategy, P.E. Bissell,
Consolidation Coal Co.
2-2 Scrub Vs. Trade: Enemies or Allies? J. Piatt, EPRI
2-3 Evaluating Compliance Options, J.H. Wile, National Economic
Research Association, Inc.
2-4 Clean Air Technology Workstation, D. Sopocy, Sargent & Lundy
2-5 Economic Evaluations of 27 FGD Processes, R.J. Keeth, United
Engineers & Constructors
2-6 Strategies for Meeting Sulfur Abatement Targets in the UK Electricity
Supply Industry, W.S. Kyte, PowerGen
2-7 Compliance Strategies for Future Capacity Additions: The Role of
Organic Acid Additives, C.V. Weilert, Burns & McDonnell Engineerir
Co.
2-8 IPL Petersburg 1 & 2 CAAA Retrofit FGDs, C.P. Wedig, Stone &
Webster Engineering Corp.
2-9 Evaluation of SO2 Control Compliance Strategies at Virginia Power,
J.V. Presley, Virginia Power
Session 3A Wet FGD Process Improvements
Session Chain David R. Owens - EPRI
3A-1 Overview on the Use of Additives in Wet FGD Systems, R.E. Moser,
EPRI
3A-2 Results of High SO2 Removal Efficiency Tests at EPRI's HSTC, G.
Stevens, Radian
3A-3 Results of Formate Additive Tests at EPRI's HSTC, M. Stohs, Radian
Corp.
3A-4 FGDPRISM, EPRI'S FGD Process Model-Recent Applications, J.G.
Noblett, Radian Corp.
3A-5 Additive Enhanced Desulfurization for FGD Scrubbers, G. Juip,
Northern States Power
3A-6 Techniques for Evaluating Alternative Reagent Supplies, C.V. Weilert
Burns & McDonnell Engineering Co.
3A-7 Factors Involved in the Selection of Limestones for Use in Wet FGD
Systems, J.B. Jarvis, Radian Corp.
3A-8 Magnesium-Enhanced Lime Reaction Tank Design Tests at EPRI's
HSTC, J. Wilhelm, Codan Associates
xiii
-------
Session 3B - Furnace Sorbent Injection
Session Chain Brian Gullett - EPA
3B-1 Computer Simulation of Reacting Particle-Laden Jet Mixing Applied to
SO2 Control by Dry Sorbent Injection, P.J. Smith, The University of
Utah
3B-2 Studies of the Initial Stage of the High Temperature Ca0-S02 Reaction,
I. Bjerle, University of Lund
3B-3 Status of the Tangentially Fired LIMB Demonstration .Program at
Yorktown Unit No. 2: An Update, J.P. Clark, ABB Combustion
Engineering Systems
3B-4 Results from LIMB Extension Testing at the Ohio Edison Edgewater
Station, T. Goots, Babcock & Wilcox
Session 4A - Wet FGD Design Improvements
Session Chain Richard E. Tischer - DOE
4A-1 Reliability Considerations in the Design of Gypsum Producing Rue Gas
Desulfurization Plants in UK, I. Gower, John Brown Engineers &
Constructors Ltd.
4A-2 Sparing Analysis for FGD Systems, M. A. Twombly, ARINC Research
Corp.
4A-3 Increasing Draft Capability for Retrofit Flue Gas Desulfurization
Systems, R.D. Petersen, Burns & McDonnell Engineering Co.
4A-4 Development of Advanced Retrofit FGD Designs, C.E. Dene, EPRI
4A-5 Acid Rain FGD Systems Retrofits, A.J. doVale, Wheelabrator Air
Pollution Control
4A-6 Guidelines for FGD Materials Selection and Corrosion Protection, H.S.
Rosenberg, Batelle
4A-7 Economic Comparison of Materials of Construction of Wet FGD
Absorbers & Internals, W. Nischt, Babcock & Wilcox
4A-8 The Intelligence & Economics of F.R.P. in F.G.D. Systems, E.J. Boucher,
RPS/ABCO
xiv
-------
Session 4B - Dry FGD Technologies
Session Chain Michael Maxwell /Brian Gullett/Norman Kaplan - EPA
4B-1 Poplar River LIFAC Demonstration,T. Enwald, Tampella Power Ltd.
4B-2 1.7 MW Pilot Results for Duct Injection FGD Process Using Hydrated
Lime Upstream of an ESP, M. Maibodi, Radian Corp.
4B-3 Scaleup Tests and Supporting Research for the Development of Duct
Injection Technology, M.G. Klett, Gilbert/Commonwealth Inc.
4B-4 A Pilot Demonstration of the Moving Bed Limestone Emission
Control Process (LEC), M.E. Prudich, Ohio University
4B-5 Pilot Plant Support for MDI/ADVACATE Commercialization, C.
Sedman, U.S. EPA
4B-6 Suitability of Available Fly Ashes in ADVACATE Sorbents, C. Singer,
U.S. EPA
4B-7 Mechanistic Study of Desulfurization by Absorbent Prepared from Coal
Fly Ash, H. Hattori, Hokkaido University
4B-8 Results of Spray Dryer/Pulse-Jet Fabric Filter Pilot Unit Tests at EPRI
HSTC, G. Blythe, Radian Corp.
4B-9 Results of Medium & High-Sulfur Coal Tests on the TVA 10-MW
Spray Dryer/ESP Pilot, T. Burnett, TVA
4B-10 Evolution of the B&W Durajet™ Atomizer, S. Feeney, Babcock &
Wilcox
4B-11 Characterization of the Linear VGA Nozzle for Flue Gas
Humidification, J.R. Butz, ADA Technologies, Inc.
4B-12 High SO2 Removal Dry FGD Systems, B. Brown, Joy Technologies, Inc.
Session 5A - Wet Full Scale FGD Operations
Session Chain Robert L. Glover - EPRI
5A-1 FGD System Retrofit for Dalhousie Station Units 1 & 2, F.W. Campbell,
Burns & McDonnell Engineering Co.
5A-2 Zimmer FGD System, W. Brockman, Cincinnati Gas & Electric
5A-3 Results of on Investigation to Improve the Performance and Reliabiity
of HL&P's Limestone Electric Generating Station FGD System, M.
Bailey, Houston Lighting & Power
5A-4 Full-Scale Demonstration of EDTA and Sulfur Addition to Control
Sulfite Oxidation, G. Blythe, Radian
xv
-------
5A-5 Optimizing the Operations in the Flue Gas Desulfurization Plants of
the Lignite Power Plant Neurath Unit D and E and Improved Control
Concepts for Third Generation Advanced FGD Design, H. Scherer,
Noell, Inc.
5A-6 Organic Acid BufferTesting at Michigan South Central Power Agency's
Endicott Station, B. J. Jankura, Babcock & Wilcox
5A-7 Stack Gas Cleaning Optimization Via German Retrofit Wet FGD
Operating Experience, H. Weiler, Ellison Consultants,,
5A-8 Operation of a Compact FGD Plant Using CT-121 Process, Y. Ogawa,
Chiyoda Corp.
Session 5B - Combined SOx/NOx Technologies
Session Chain Mildred E. Perry - DOE
5B-1 Simultaneous SOx/NOx Removal Employing Absorbent Prepared
from Fly Ash, H. Tsuchiai, The Hokkaido Electric Power Co.
5B-2 Furnace Slurry Injection for Simultaneous SO2/NOX Removal, B.K.
Gullett, U.S. EPA
5B-3 Combined SO2/NOX Abatement by Sodium Bicarbonate Dry Injection,
J. Verlaeten, Solvay Technologies, Inc. (124)
5B-4 SO2 and NOx Control by Combined Dry Injection of Hydrated Lime
and Sodium Bicarbonate, D. Helfritch, R-C Environmental Services &
Technologies
5B-5 Engineering Evaluation of Combined N0x/S02 Controls for Utility
Application, J.E. Cichanowicz, EPRI
5B-6 Advanced Flue Gas Treatment Using Activated Char Process
Combined with FBC, H. Murayama, Electric Power Development Co.
5B-7 SO2/NOX Control using Ferrous EDTA and a Secondary Additive in a
Combined Lime-Based Aqueous Scrubber System, M.H. Mendelsohn,
Argonne National Laboratory
5B-8 Parsons FGC Process Simultaneous Removal of SOx and NOx, K.V.
Kwong, The Ralph M. Parsons Co.
xvi
-------
Session 6A - Wet FGD Operating Issues
Session Chain Gary M. Andes - EPRI
6A-1 Pilot-Scale Evaluation of Sorbent Injection to Remove SO3 and HC1, J.
Peterson, Radian Corp.
6A-2 Control of Acid Mist Emissions from FGD Systems, R.S. Dahlin,
Southern Research Institute
6A-3 Managing Air Toxics: Status of EPRI PISCES Project, W. Chow, EPRI
6A-4 Results of Mist Elimination System Testing in an Air-Water Pilot
Facility, A.F. Jones, Radian Corp.
6A-5 CEM Vendor and Utility Survey Databases, J.L. Shoemaker,
Engineering Science, Inc.
6A-6 Determination of Continuous Emissions Monitoring Requirements at
Electric Energy Inc., V. V. Bland, Stone & Webster Engineering Corp.
6A-7 Improving Performance of Flushless Mechanical Seals in Wet FGD
Plants through Field and Laboratory Testing, F.E. Manning, BW/IP
International Inc.
6A-8 Sulcis FGD Demonstration Plant Limestone-Gypsum Process:
Performance, Materials, Waste Water Treatment, E. Marchesi, Enel
Construction Department
Session 6B - Clean Coal Demonstrations
Session Chain Joseph P. Strakey - DOE
6B-1 Recovery Scrubber Cement Application Operating Results, G.L.
Morrison, Passamaquoddy Technology
6B-2 The NOXSO Clean Coal Technology Demonstration Project, L.G. Neal,
NOXSO Corp.
6B-3 Economic Comparison of Coolside Sorbent Injection and Wet
Limestone FGD Processes, D.C. McCoy, Consolidation Coal Co.
6B-4 Ohio Edison's Clean Coal Projects: Circa 1991, R. Bolli, Ohio Edison
Emerging Technologies
6B-5 A Status Report on Sanitech's 2-MWe Magnesia Dry Scrubbing
Demonstration, S.G. Nelson, Sanitech Inc.
6B-6 Application of DOW Chemical's Regenerable Flue Gas Desulfurization
Technology to Coal Fired Power Plants, L.H. Kirby, Dow Chemical
6B-7 Pilot Testing of the Cansolv System FGD Process, L.E. Hakka Union
Carbide Canada LTD.
6B-8 Dry Desulfurization Technology Involving Humidification for
Enhanced SO2 Removal, D.P. Singh, Procedair Industries Inc.
xvii
-------
Session 7 - Poster Papers
Session Chair Charles Sedman - EPA
7-1 Summary of Guidelines for the Use of FRP in Utility FGD Systems, W.
Renoud, Fiberglass Structural Engineering, Inc.
7-2 Development and Evaluation of High Surface Area Hydrated Lime for
SO2 Control, M. Rostam-Abadi, The Illinois State Geological Survey
7-3 Effect of Spray Nozzle Design and measurement Techniques on
Reported Drop Size Data, W. Bartell, Spraying Systems Co.
7-4 High SO2 Removals with a New Duct Injection Process, S.G. Nelson, Jr.
Sanitech, Inc.
7-5 Combined SOx/NOx Control Via Soxal™, A Regenerative Sodium
Based Scrubbing System , C.H. Byszewski, Aquatech Systems
7-6 The Healy Clean Coal Project Air Quality Control System, V.V. Bland,
Stone & Webster Engineering Corp.
7-7 Lime/Lime Stone Scrubbing Producing Useable By-Products, D. P.
Singh, Procedair Industries Inc.
7-8 Modeling of Furnace Sorbent Injection Processes, A.S. Damle, Research
Triangle Institute
7-9 Dry FGD Process Using Calcium Absorbents, N. Nosaka, Babcock-
Hitachi K.K.
7-10 Clean Coal Technology Optimization Model, B.A. Laseke, International
Technology Corp.
7-11 SNRB Catalytic Baghouse Process Development & Demonstration, K.E.
Redinger, Babcock & Wilcox
7-12 Reaction of Moist Calcium Silicate Reagents with Sulfur Dioxide in
Humidified Flue Gas, W. Jozewicz, Acurex
7-13 Commercial Application of Dry FGD using High Surface Area Hydrated
Lime, F. Schwarzkopf, Florian Schwarzkopf PE.
7-14 Initial Operatiing Experience of the SNOX Process, D.J. Collins, ABB
Environmental System
7-15 Progress Report of the NIPSCO - Pure Air - DOE Clean Coal II Project, S.
Satrom, Pure Air
7-16 Development of a Post Combustion Dry SO2 Control Reactor for Small
Scale Combustion Systems, J.C. Balsavich, Tecogen Inc.
xviii
A
A
-------
7-17 Scrubber Reagent Additives for Oxidation Inhibited Scrubbing, J.
Thompson, Process Calx, Inc.
7_lg Recovery of Sulfur from Calcium Sulfite and Sulfate Scrubber Sludges,
J. Thompson, Process Calx, Inc.
7-19 Magnesite & Dolomite FGD Technologies, D. Najmr, Ore Research
Institute
7-20 SOx and Particulate Emissions Reduction in a Pulverized Coal Utility
Boiler through natural Gas Cofiring, K.J. Clark Aptech Engineering
Services
7-21 Design, Installation, and Operation of the First Wet FGD for a lignite
Fired Boiler in Europe at 330 MW P/S Voitsberg 3 in Austria, H.
Kropfitsch, Voitsberg
Session 8A - Commercial FGD Designs
Session Chain Robert E. Moser - EPRI
8A-1 "* Mitsui-BF Dry Desulfurization and Utility Compliance Strategies, K.
Tsuji, Mitsui Mining Company Ltd.
8A-2 High Efficiency Dry Flue Gas SOx and Combined SOx/NOx Removal
Experience with Lurgi Circulating Fluid Bed Dry Scrubber; A New
Economical Retrofit Option for Utilities for Acid Rain Remediation, J.
G. Toher, Environmental Elements Corp.
8A-3 Incorporating Full-Scale Experience into Advanced Limestone Wet
FGD Designs, P.C. Rader, ABB Environmental Systems
8A-4 Design and Operation of Single Train Spray Tower FGD Systems, A.
Saleem, GE Environmental Systems
8A-5 Selecting the FGD Process and Six Years of Operating Experience in
Unit 5 FGD of the Altbach-Deizisau Neckawerke Power Station, R.
Maule, Noell Inc.
8A-6 Development and Operating Experience of FGD Technique at the
Volkingen Power Station, H. Petzel, SHU-Technik
8A-7 Advantages of the CT-121 Process as a Throwaway FGD System, M.J.
Krasnopoler, Bechtel Corp.
xix
-------
Session 8B - By-Product Utilization
Session Chair: Charles E. Schmidt - DOE
8B-1 German Experience of FGD By-Product Disposal and Utilization, J.
Demmich, Noell Inc.
8B-2 The Elimination of Pollutants from FGD Wastewaters, M.K.
Mierzejewski, Infilco Degremont Inc.
8B-3 The Influence of FGD Variables on FGD Performance and By-Product
Gypsum Properties,F. Theodore, Consolidation Coal Co.
8B-4 Quality of FGD Gypsum, F.W. van der Brugghen, N.V. Kema
8B-5 Chemical Analysis and Flowability of By-Product Gypsums, L.Kilpeck,
Centerior
8B-6 Evaluation of Disposal Methods Stabilized FGD & Oxidized FGD
Sludge & Fly Ash, W. Yu, Conversion Systems, Inc.
8B-7 Commercial Aggregate Production from FGD Waste, C.L. Smith,
Conversion Systems, Inc.
xx
-------
Session 4A
WET FGD DESIGN IMPROVEMENTS
RELIABILITY CONSIDERATIONS IN THE DESIGN OF GYPSUM
PRODUCING FLUE GAS DESULPHURISATION PLANTS IN THE UK
I. P. Gower
John Brown Engineers and Constructors Ltd
20 Eastbourne Terrace, London W26LE
ABSTRACT England
Her Majestys Inspectors of Pollution (HMIP) Have stipulated that Flue Gas
Desulphurisation (FGD) plants built in the UK must achieve a minimum availability of
97.5% per boiler stream. This has resulted in FGD contracting companies carrying out
extensive Reliability, Availability and Maintainability (RAM) studies to prove that the FGD
plant designs can achieve the required availability.
John Brown carried out a number of reliability studies to prove the FGD plant
configuration selected would have an availability of over 97.5%. The following major
techniques were used:
* Established, Novel Features and Preferred Parts Analysis
* Buffer Storage Analysis
* Failure Modes, Effects and Criticality Analysis
* Fault Tree Analysis
In addition a failure and routine maintenance model was developed that predicted total
maintenance hours for an FGD plant, split by craft disciplines.
The interdependency between each model was studied in detail culminating in a
predicted availability figure of 99 + % for the average operating case. Sensitivity
analysees were completed for different operating and fault scenarios.
John Brown are confident that the original reliability objectives have been met by the
approach taken to develop a lowest life cycle cost plant.
k.
4A-1
-------
INTRODUCTION
The driving force for retrofitting Flue Gas Desulphurisation (FGD) equipment to coal and
oil fired power stations in the United Kingdom (UK) is the "Large Combustion Plant (LCP)
Directive" which has been agreed between all member countries of the European
Economic Community (EEC). In the UK all emissions from industrial processes and power
stations are monitored by Her Majestys Inspectorate of Pollution (HMIP). There are
currently no regulations specifically concerning the design of retrofit FGD plant,
however, HMIP for the first two retrofit FGD plants have stipulated that the FGD plant
must have an availability per boiler stream of at least 97.5%, measured on the following
basis:
Availability (A) = a / (b + c)
where:
a = Time that the unit FGD plant operates at the. required performance
b = Time that the boiler is operated whilst input conditions are within
agreed ranges including the time when the boiler is operated on
bypass
c = Time that the boiler is not operated due to a unit FGD outage
This paper discusses some of the reliability, availability and maintainability models that
John Brown Engineers and Constructors Ltd have utilised in ensuring that the design of
gypsum producing FGD plants for the UK will meet an availability of at least 97.5%.
4A-2
A
A
-------
DETERMINATION OF PLANT CONFIGURATION
The most recent Invitations To Tender (ITT) for FGD plants in the UK have been based
on performance type contracts. This has allowed contracting companies to:
* Fully optimise the plant design
* Concentrate on the system rather than components design
* Fully evaluate all options for the plant design
* Offer a lowest life cycle cost plant design
* Offer a fit for purpose plant design
This in essence has required all contracting companies to determine the configuration of
plant equipment that they felt met all the above points.
In response to these requests John Brown embarked on an extensive series of studies to
optimise the plant design with due regard to the performance contract requirements
whilst also ensuring that the onerous plant guarantee conditions set by the UK utility
companies were not compromised.
The series of studies started by defining the major systems and subsystems which
would comprise the plant design, Figure 1. These were taken as the basic building
blocks. Each system was then reviewed to identify the possible arrangements of plant
which could sensibly make the basis of a plant design. Figure 2 shows an example of
the configurations studied for the oxidation air compressor system. Each arrangement
was then subjected to an analysis, assessed against the following criteria:
* Availability
* Operability
* Maintainability
* Installed cost
* Operating cost
* Proven track record for equipment duty
* Proven track record for system configuration
* Criticality of equipment / system failure
* Can required turndown be achieved
k.
4A-3
-------
Each criteria was assessed against a 1 to 10 ranking, 1 = undesirable ranking, 10 =
desirable ranking. The criteria scores were weighted heavily in favour of a high
availability for the system configurations considered. In some cases technical conditions
could not be satisfied eg the turndown required for a particular equipment item could not
be achieved, for these cases the particular system configuration was not studied further.
In completing the first pass availability calculations for each system configuration
considered and the first pass estimate of total maintenance manhours, John Browns
Maintenance Manning and Availability system model was utilised (see later). Initial data
input into this model was taken from John browns inhouse database.
Results from these initial studies demonstrated that the availability of critical building
block systems aggregated across an overall plant scheme was in excess of 99%. This
backed up claims made for General Electric Environmental Services Inc (GEESI) designed
systems which have consistently demonstrated availability approaching 100%.
The process design developed by John Brown and GEESI for the UK FGD plants had
characteristics specifically selected to maximise plant availability and reliability, these
included:
* The use of a constant bleed slurry flowrate independent of sulphur content or
boiler load to prevent variability in transporting the slurry between the
absorber and common plant processing areas
* Limiting the number of throttling valves in slurry service
* Operation of the common plant processing equipment eg limestone mills in a
batch mode. This ensures stable flow and corresponding consistent
performance
* Ensuring sufficient buffer storage capacity of upstream liquors / slurries exists
to counteract on line maintenance requirements for downstream equipment
(see later)
* Use of dedicated recycle pumps for individual spray banks
* Use of Insitu Forced Oxidation (IFO) reducing the need for extra plant
equipment
These screening methods and process design considerations allowed John Brown to
quickly establish the range of system and subsystem configurations which could be
considered as part of the overall FGD plant design. The following set of activities
considered the interdependency of each of the chosen plant configurations from a
4A-4
A
-------
reliability viewpoint, resulting in figures for the overall plant availability, installed cost,
operating cost and impact on plant guarantees.
RELIABILITY CONSIDERATIONS DURING DESIGN
To design, procure and build an FGD plant which has a minimum guaranteed availability
of 97.5% requires an approach that extends beyond simple Fault Tree modelling; the
approach typically adopted for process plants in the past. John Browns approach
considered first the number of factors from which the FGD plant was termed
unavailable; these were basically tied to the required performance and operating
envelope determined by the utility companies and HMIP. For every hour the FGD plant
operated outside the agreed envelope the plant would be labelled "unavailable", refer to
the availability definition equation in the introduction. Major factors which made the
plant unavailable were:
* Sulphur dioxide removal below 90% across the system
* Gypsum quality outside target concentrations
* Reheated flue gas below a minimum temperature
The emphasis the reliability analysees required was therefore based around meeting the
level of performance guarantees set for the plant. The following lists the RAM activities
completed after the initial screening of plant configuration options. However, the Failure
Modes, Effects and Criticality analysis and the Fault Tree Analysis were the major
analysees which determined whether the effect of plant failures were acceptable for the
chosen plant design:
* Gathering of Detailed Reliability and Maintainability Data
* Established, Novel Features and Preferred Parts Analysis
* Buffer Storage Analysis
* Logistic Repair Delays Analysis
* Failure Modes, Effects and Criticality Analysis (FMECA)
* Fault Tree Analysis (FTA)
* Maintenance Manning and Maintainability Analysis
* Consequential Failures Analysis
It is important to understand how to approach each of the above analysees to ensure
that the correct level of detail is addressed to minimise the risk of a plant design that has
4A-5
-------
an unacceptably low availability. It is equally important to understand how to treat the
results of each analysis on an independent and interdependency basis. The following
sections describe the major RAM activities. The sequence in which they were completed
is shown in Figure 3.
GATHERING OF DETAILED RELIABILITY AND MAINTAINABILITY DATA
The data used in all the analysees completed by John Brown was based on a number of
sources, these included:
* John Browns in house databank
* Offshore Reliability Data (OREDA)
* Discussions with client companies of John Brown
* Discussions with General Electric Environmental Systems Inc (GEESI)
* Discussions with other GEESI Licencees eg Hoogovens Technical Services
(ESTS)
* Equipment Suppliers
* Published Articles eg EPRI reports
* National Centre for Systems Reliability - Reliability Databank
* Data from Riedersbach II generating station
* Reliability Engineering and Risk Assessment - Henley E.J. and Kumamoto H.,
Prentice Hall
During initial tendering for equipment John Brown required all potential suppliers to fill in
comprehensive pro formas for reliability and maintenance information. The type of
information requested included:
* Mean Time To Repair (MTTR) and Mean Time Between Failures (MTBF) for all
critical components
* Maximum Time To Repair for critical components
* Failure modes and effect of failure of critical components
* Maintenance requirements including repair team sizes, special equipment, on
and off line maintenance frequencies
* Spare parts lists
Requesting this information from suppliers had a twofold objective. It ensured both that
the unavailability was realistic for the components considered, and that the suppliers had
4A-6
A
A
-------
realistic guarantee targets to aim for their equipment. This would add value to claims
they had to make about the reliability of their equipment.
ESTABLISHED, NOVEL FEATURES AND PREFERRED PARTS ANALYSIS
In order to gain maximum confidence that the FGD system design and selected
components would achieve the required 97.5% availability per boiler stream, a review
was carried out to ensure that a successful track record for all proposed design
configurations and types of components existed. These formed the basis of the preferred
parts / systems for the FGD plant. Novel features were kept to a minimum and only
formed part of the design if there was absolute confidence in the reliability, operability
and maintainability features of the novel design configuration / component. These parts
of the design were studied further in detail in the Buffer Storage Analysis, FMECA and
Fault Tree Analysis.
BUFFER STORAGE ANALYSIS
The buffer storage analysis showed the effect of the storage of process liquors on
potential plant outages due to equipment repair or planned maintenance. Outage was
defined as the time when the FGD plant was not available to scrub the flue gas.
Equipment highlighted in the buffer storage analysis were those whose breakdown had a
direct impact on the plant when there was either no buffer capacity available or no
standby equipment. For each item of equipment considered in the buffer storage analysis
the following information was detailed:
* Hours required for repair of each expected breakdown
* MTTR for the breakdown
* Hours required for planned maintenance (off line only)
* Planned maintenance period
* Logistic Repair Delays
Equipment outage times were assessed against the associated buffer storage of the
equipment (if any). If the maintenance / repair time was greater then the buffer storage
time available the equipment was assessed in more detail. Where standby equipment
k
4A-7
-------
was in existence the analysis was not completed in detail as maintenance / repair times
did not directly result in FGD plant outage time. On line planned maintenance was not
considered in this analysis as it also does not result in outages. It was however
considered in the Maintenance Model. Planned off line maintenance of frequency longer
than time between major boiler maintenance periods was also excluded as this can be
carried out during the planned boiler and fgd plant shutdowns. In general repair times
and mean time between failures were supplied by preferred vendors for each item of
equipment. Maximum time to repair of equipment was also considered to review the
sensitivity of critical items.
The hours required for breakdown repair were divided into three sections in the buffer
storage analysis:
* MTTR
* Set up time to prepare equipment for maintenance
* Preparation time for equipment to be running at design conditions after
maintenance
The breakdown repair times were sectionalised in this way in order to ensure the total
repair times were realistic. The last two breakdown repair hours are termed Logistic
Repair Delays. An example of the Buffer Storage Analysis can be found in Figure 4.
LOGISTIC REPAIR DELAYS
Various aspects of maintenance or repair of plant were considered to determine the
correct downtime for plant failures. A piece of equipment ready to be put back into
service may not actually be operating at design conditions until some hours later.
Logistic repair delays were therefore incorporated into the fault tree analysis models to
ensure that a realistic availability was derived. Similar conditions applied to the other
models. Logistic repair delays were considered in the following areas:
* Drain down / purge / cooling time for relevant equipment items
* Mobilisation of repair teams
* Provision and commissioning of cranes, power or other maintenance related
items
* Restoration of an item to steady state operating conditions once it is repaired
4A-8
A
A
-------
* Time to restoration of process conditions from commencement of operation of
the item
Some of the logistic repair delays were concurrent. When this occurred it was the
longest logistic repair delay which determined the true downtime of an item of
equipment. An example of a Logistic Repair Delay analysis can be found in Figure 5.
FAILURE MODES, EFFECTS AND CRITICALITY ANALYSIS / FAULT TREE ANALYSIS
A full Failure Modes, Effects and Criticaiity Analysis (FMECA) of a plant takes each of
the process systems and breaks it down into its components in order to analyse the
failure mode of each component and the effect this can have on system availability ie
how critical it is to system operation. For the UK FGD plants a FMECA was initially
performed at a subsystem level to determine which are immediately critical to plant
operation. Those subsystems found to be critical were further analysed to determine the
component failures which contributed most to the unavailability of the plant. The FMECA
also incorporated the results of the Buffer Storage Analysis and the Maintenance
Manning and Availability System model. It was found that a number of failure modes
were consistent across most subsystems, these were:
Transfer failure The system did not pass on flow to the
downstream systems
Loss of buffer capacity The hold up available in the system was lost due
to a fault
Out of specification The process flow was out of specification due
to a fault
Loss of one stream For parallel systems a duty train failed
Once the effect of failures had been determined they were classified according to the
severity of the failure in terms of FGD plant unavailability or outage. Fault trees were
derived for those cases where guarantee conditions were violated as revealed by the
FMECA, either at a component or subsystem level. These trees were then solved to find
the combinations of failures which produced an overall system failure. Figure 6
demonstrates how the FMECA was applied to part of a limestone milling system.
k.
4A-9
-------
MAINTENANCE MANNING AND AVAILABILITY SYSTEM
To help identify overall life cycle costs of FGD plants John Brown developed a
methodology to assess the quantity of routine and failure maintenance required for an
FGD plant configuration. From this the size of the maintenance crew and maintenance
schedule were established. Plant failure and repair data was used to get a first pass
estimate of the availability of each individual building block (subsystem) and the whole
FGD plant. For the initial set of studies completed by John Brown the methodology used
defined a range of building block systems representing conventional systems used on
FGD plants. The building blocks were then collated to generate the various FGD plant
configurations considered. Routine maintenance schedules and failure data for all system
components were incorporated into a series of linked computer spreadsheets.
Maintenance of mechanical, instrument and electrical equipment was also addressed.
Figure 7 demonstrates how the spreadsheets comprising the Maintenance Manning and
Availability System model interlink.
The maintenance model methodology was based on two levels:
* Level 1 - A comprehensive component database which collates failure data
and routine maintenance schedules for all mechanical, instrument and
electrical equipment.
* Level 2 - A series of building block systems assembled from the individual
components. Selection of the appropriate numbers of each system represented
the FGD plant configuration studied. Individually tailored building blocks were
added to suit particular needs.
Building block systems which were critical to maintaining FGD plant availability, that is if
their failure immediately caused loss of availability, were given the highest priority for
performing failure maintenance on the system. Conversely, systems which did not
directly support FGD plant availability were given a lower maintenance priority as the
consequences of failure were less. Once the appropriate numbers of each building block
system were selected priorities for performing maintenance were assigned both for
routine maintenance and failure maintenance, these took the form of a critical and non
critical classification. The criticalities of the maintenance tasks were aggregated to
provide a means of assessing how maintenance could be scheduled through a year.
Typically the following was adopted:
* Failure maintenance of critical equipment must be performed immediately
4A-10
A
-------
Failure maintenance of non critical systems should be performed within one
week of failure
Critical routine maintenance should be performed within +1-2 weeks of
scheduled date (except for components that require maintenance on a weekly
basis)
Non critical routine maintenance can be deferred by + /- 2months from
programmed dates
Failure maintenance manhours were assessed from failure data ie MTBF and MTTR for
each component. Annualised failure maintenance manhours were derived from the
product of MTBF, MTTR and repair team size.
Annualised routine maintenance manhours were developed by deriving routine schedules
for all components on a weekly, monthly, 3 monthly, 6 monthly, yearly and three yearly
basis. A predictive maintenance approach was adopted assuming the use of condition
monitoring techniques.
Individual spreadsheets were developed in the maintenance model for each technician
discipline and each type of maintenance. The approach adopted for instrumentation
identified all instrument components used in an FGD plant and the number of each
instrument in each building block system. Component failure data and estimates of
annual routine maintenance per component were then used to derive annual instrument
maintenance totals for each building block. Mechanical routine maintenance schedules
were developed for all mechanical equipment items and aggregated per building block.
Failure data was also applied at the building block level to aggregate failure maintenance
manhours. The approach adopted for electrical maintenance identified the numbers of all
motors, motorised valves and panels in each building block to which individual routine
maintenance schedules were applied. Failure data was also applied at the component
level. Figures 8 and 9 demonstrate how failure and routine maintenance manhours are
developed for different types of instrumentation and gas path equipment.
CONSEQUENTIAL FAILURES
Consequential failures were analysed by a Hazard and Operability analysis (HAZOP) and
Fault Tree Analysis. In the HAZOP analysis that was completed the effect of upstream
factors such as high particulate loadings were considered on downstream equipment.
4A-11
-------
Fault trees were reviewed to see if any effect beyond the repair of the component itself
arose from equipment failure.
CONCLUSION
John Brown developed a system design for gypsum producing FGD plants in the UK that
was of "lowest life cycle cost" and of sufficiently high availability to comply with current
HMIP requirements.
To determine the reliability aspects of the plant design John Brown applied a number of
RAM techniques that included Fault Tree Analysis and Failure Modes, Effects and
Criticality Analysis. A number of other RAM related models were developed that
integrated with the FTA and FMECA, the results of which gave John Brown confidence
that the plants will perform to all guarantee requirements.
ACKNOWLEDGMENTS
John Evans - Senior Safety and Reliability Engineer with John Brown E&C Ltd, London.
REFERENCES
1. I. Gower, J. Evans and S. Reynolds. "Development of a Streamline Flue Gas
Desulphurisation Plant for Retrofit to a 2000 MW UK Power Station." Pittsburgh
Coal Conference. October 1991.
4A-12
A
-------
SYSTEMS
SUBSYSTEMS
FLUE GAS PATH
DAMPERS
BOOSTER FANS
REHEATERS
ABSORBER
RECYCLE PUMPS
MIST ELIMINATORS
OXIDATION AIR
GYPSUM DEWATERING
PRIMARY SEPARATION
FINAL DEWATERING (CENTRIFUGES)
CENTRATE COLLECTION
GYPSUM STORAGE
CONVEYING
STACKING
RECLAIMING
LOADING
LIMESTONE STORAGE
UNLOADING
CONVEYING
STACKING
RECLAIMING
LIMESTONE MILLING
LIMESTONE FEEDING
LIMESTONE MILLING
LIMESTONE CLASSIFICATION
LIMESTONE SLURRY STORAGE
UTILITIES
PROCESS WATER
FLUSH WATER
INSTRUMENT AIR
PLANT AIR
WASTE WATER TREATMENT SECONDARY HYDROCYCLONES
WASTE WATER STORAGE
WASTE WATER TREATMENTS
WASTE WATER DISCHARGE
SLUDGE DISCHARGE
Figure 1. Breakdown of FGD Plant into Systems and Subsystems
k.
4 A-13
-------
100%
100%
Oxidation Air to
Four Absorbers
Oxidation Air to
Four Absorbers
Oxidation Air to
Four Absorbers
Oxidation Air to
Four Absorbers
Oxidation Air to
Two Absorbers
Oxidation Air to
Two Absorbers
Oxidation Air to Oxidation Air to Oxidation Air to Oxidation Air to
One Absorber One Absorber One Absorber One Absorber
Figure 2. Configurations Analysed for Oxidation Air Compressors
4A-14
-------
IDENTIFY RANGE OF SYSTEMS AND SUBSYSTEMS
REVIEW EACH SYSTEM AGAINST IDENTIFIED CRITEREA
SELECT MOST APPROPRIATE SYSTEMS FOR INCLUSION IN FGD PLANT
COMPLETE ESTABLISHED, NOVEL FEATURES AND PREFERRED PARTS ANALYSIS
ISSUE TENDERS TO EQUIPMENT SUPPLIERS
RECEIVE COMPLETED RAM SCHEDULES FROM SUPPLIERS
INPUT SPECIFIC DATA TO MAINTENANCE MODEL
COMPLETE BUFFER STORAGE ANALYSIS
COMPLETE FMECA FOR ALL SUBSYSTEMS
COMPLETE FTA FOR ALL FAILURE MODES THAT IMPACT GUARANTEES
COMPLETE HAZOP AND COMMON MODE FAILURE ANALYSIS
REVIEW RESULTS OF ALL MODELS
REVIEW PLANT CONFIGURATION / DESIGN AS NECESSARY
Figure 3. Sequence of RAM Activities
4A-15
-------
RATCL1FFE ARM ASSESSMENT
BUFFER STORAGE ANALYSIS FOR BREAKDOWN REPAIRS
INCORPORATING LOGISTIC REPAIR DELAYS.
EQUIP HTTR MAX MAIN'CE CRANE DRAIN COOL EQPT DET PRE COMM'G EQPT HAND TOTAL ASSOC'D BUFFER AVAILABLE
No. DESCRIPTION (HRS) (HRS) MOBIL'N TIME TIME TIME ISOL DIAG WORK TIME RUNUP BACK B/DOUN TIME BUFFER TIME TIME
COMMENTS
M2508 GYPSUM 4
A/B STORE
S/PILE
CONVEYOR
1
2.0 0.0 0.0 0.0 1.0 0.0 0.0 1.0 0.0 0.7 7.7
2.0 0.0 0.0 0.0 1.0 0.0 0.0 1.0 0.0 0.7 4.2
2 0 2.0 0.0 0.0 0.0 1.0 0.0 0.0 1.0 0.0 0.7 5.7
12 24 2.0 0.0 0.0 0.0 1.0 0.0 0.0 1.0 0.0 0.7 15.7
5 0 2.0 0.0 0.0 0.0 29.0 0.0 0.0 1.0 0.0 1.0 36.0
M2601 DRAG LINK 8 0 2.0 0.0 0.0 0.0 1.5 0.0 0.0 1.0 0.0 0.5 11.5
A/B FEEDER
6 0 2.0 0.0 0.0 0.0 1.5 0.0 0.0 1.0 0.0 0.5 9.5
M2602 WEIGH BELT 6 0 2.0 0.0 0.0 0.0 1.5 0.0 0.0 1.0 0.0 0.5 9.5
A/B FEEDER
8 0 2.0 0.0 0.0 0.0 1.5 0.0 0.0 1.0 0.0 0.5 11.5
P1001 GYPSUM 5 8 2.0 0.0 0.0 0.0 29.0 0.5 0.0 0.5 0.0 1.0 36.0
A/B SLURRY
TRANSFER
PUMP
T2101, 46.5 38.8
ABSORBE
R.
T1001A/
B
T2101, 46.5 42.3
ABSORBE
R.
T1001A/
B
T2101, 46.5 40.8
ABSORBE
R.
T1001A/
B
T2101, 46.5 30.8
ABSORBE
R,
T1001A/
B
T2101, 46.5 10.5
ABSORBE
R.
T1001A/
B
SEE 0.0 -11.5
COMMENT
S
SEE 0.0 -9.5
COMMENT
S
SEE 0.0 -9.5
COMMENT
S
SEE 0.0 -11.5
COMMENT
S
ABSORBE 19.5 -16.5
R,
T1001
A/B
PLANNED MAINTENANCE TIME IS BASED
ON MANUFACTURERS REQUIREMENTS.
BREAKDOWN FAILURE MODE IS THE
PULLEYS, MTTF = 26200 HRS
PLANNED MAINTENANCE TIME IS BASED
ON MANUFACTURERS REQUIREMENTS.
BREAKDOWN FAILURE MODE IS THE
IDLERS, MTTF = 43800 HRS
PLANNED MAINTENANCE TIME IS BASED
ON MANUFACTURERS REQUIREMENTS.
BREAKDOWN FAILURE MOOE IS THE
COUPLING PINS, MTTF = 26200 HRS
PLANNED MAINTENANCE TIME IS BASED
ON MANUFACTURERS REQUIREMENTS.
BREAKDOWN FAILURE MOOE IS THE
BELTS, MTTF = 43800 HRS
CAT A TYPE REPAIR. PLANNED
MAINTENANCE TIME IS BASED ON
MANUFACTURERS REQUIREMENTS.
BREAKDOWN FAILURE MODE IS THE
REDUCTION UNITS, MTTF = 87600 HRS
STANDBY TRAIN EXISTS. E TYPE
REPAIR AS LOU MTTR. FAULT IS
LININGS & WEAR PLATES. MTTF = 35040
STANDBY TRAIN EXISTS. E TYPE
REPAIR AS LOU MTTR. FAULT
CONDITION IS THE CHAIN AND
SPROCKET FAILURE. MTTF = 26200 HRS.
STANDBY TRAIN EXISTS. E TYPE
REPAIR AS LOU MTTR. FAULT IS
CHAINS AND SPROCKETS. MTTF = 26200
STANDBY TRAIN EXISTS. E TYPE
REPAIR AS LOU MTTR. FAULT IS
LININGS & WEAR PLATES. MTTF = 35040
CAT A FAILURE. NO DIRECT
AVAILABILITY EFFECT AS STANDBY
EXISTS. FAILURE MODE = MECH SEAL
OR IMPELLER, VOLUTE & FRAME PLATE
Figure 4. Buffer Storage Analysis (Example)
-------
LOGISTIC REPAIR DELAYS P2105 A/B
CENTRATE PUMP
PREMAINTENANCE TIME
MAIN'CE MOBILISATION
CRANAGE TIME
DRAIN TIME
COOLING TIME
EQPT ISOLATION
DETAILED DIAGNOSIS
MAIN'CE PREWORK
=> TOTAL PREMAINTENANCE TIME
POST MAINTENANCE TIMES
COMM'G REPL ITEM 1.0 COMMENTS ROTATION CHECKS
2.0 COMMENTS SPARES EX STORES
6.0 COMMENTS IRON FAIRY REQD
0.0 COMMENTS
0.0 COMMENTS
29.0 COMMENTS CAT A TYPE REPAIR
0.0 COMMENTS
COMMENTS
29.0
0.0
EQPT RUNUP TIME
EQPT HANDBACK
POST MAINTENANCE TIMES=
0.0 COMMENTS
1.0 COMMENTS DEISOLATION
2.0
MTTR BREAKDOWN 12.0
MAX TIME TO REPAIR BDOWN 24.0
TOTAL MEAN BDOWN TIME 4 3.0
TIME AVAILABLE- MEAN 3.5
BUFFER TIME (IF AVAIL'E) 4 6.5
COMMENTS
CAT A TYPE FAILURE. FAILURE MODE IS MECH SEALS. MTTF = 240,000. ALSO,
BEARINGS MTTR = 12, MAX = 24, MTTF = >500,000, WEAR RINGS, MTTR = 32, MAX =
64, MTTF = 480,000 HRS. BUFFER BREAKDOWN T2101 27, ABSORBER 17, T1001 2.5.
FRESH WATER MAKEUP FOR LIMESTONE REQUIRED.
Figure 5. Logistic Repair Delays (Example)
k
4A-17
-------
>
REF No.
SYSTEM
FAILURE MODE
CAUSE OF
FAII.UKE
FAILURE EFFECT
DETECTION
COMMENTS
CRITICALITY OF PRIMARY GUARANTEE
1 2 3 4 5
LSI
Limestone
Storage And
Transfer
Loss Of Buffer
Capacity
No Limestone
Supply
No Limestone Feed.
Availability Only
While System Hold
Up Processed.
Inspection
12 Hour Buffer Capacity
In Duty Silo (average).
24 Hour Buffer Capacity
In Standby Silo.
18 Hour Buffer Capacity
In Limestone Slurry
Tanks.
3 3 P
Transfer Failure
Failure Of
Conveyor Or
Stacker
Switch To Standby
Conveyor
Operation Of
Conveyor Is
Supervised
Buffer Storage As
Above
6
Limestone Out Of
Specification
Wrong Material
Delivered
Effect Only If
Limestone Allowed
To Be Processed
Visual And
Laboratory Test
Limestone Outside
Supply Guarantee
Quality.
Possible Loss Of FGD
Capability.
Gypsum Quality Could
Be Effected.
3 3 P
LS2
Limestone Mill
Slurry Out Of
Specification
Agitator /
Hydrocyclone /
Mill Failure
Loss Of One Mill
Stream.
Switch To Other Mill
Stream
Density Of Slurry
Changes
Reactivity Partly
Depends On Particle
Size.
Switching Partly
Depends On Detection.
Buffer Storage As
Above.
6 6 P
l-oss Of One
Stream
Critical
Component
Failure
Loss Of Sluny
Supply.
Switch To Other
Stream.
Low Flow, Low
Density, Low
Level, Motor
Stopped.
6
Slurry
Distribution
Failure
Common Mode
Pump Failure
Loss Of Feed To All
Absorbers
pH Falls.
SO2 Removal
Falls.
Pump Trip
Indication.
This Will Have An
Immediate Effect.
1
Leakage
Lining / Tank
Failure
Switch To Other Tank
Or Mill Stream
Low Level
Indication And
Trips
No Immediate Effect On
Process.
Figure 6. Failure Modes, Effects and Criticality Analysis (Example)
-------
FGD PLANT MMAS MODEL
SUMMARY BLOCK DIAGRAM
ELECTRICAL
COMPONENT
FAILURE DATA
10.//B
ROUTINE
MAINTENANCE
SUMMARY BY 10.11
BY DISCIPLINE
MECHANICAL
COMPONENT
FAILURE DATA
105
INSTRUMENT
COMPONENT
FAILURE DATA
104
ANNUAL MAINTENANCE
SUMMARY BY
DISCIPLINE
10.12
COMPONENT
QUANTITIES PER
BUILDING BLOCK
10.6
ELECTRICAL
COMPONENT
MAINTENANCE DATA
10.7/B/10
BUILDING BLOCK
SYSTEM
UNAVAILABILITY
10 J
AVAILABILITY
BUILDING BLOCK
DEFINITION
MAINTENANCE
BUILDING BLOCK
DEFINmON
MECHANICAL
COMPONENT
MAINTENANCE DATA
105
COMPONENT
QUANTITIES PER
BUILDING BLOCK
10.6
BUILDING BLOCK
CRmCALTTY
CRITERIA
10.4
SUMMARY OF ANNUAL
MAINTENANCE
MANHOURS
10.14
SUMMARY OF
PRODUCTION
AVAILABILITY
10.2
INSTRUMENT
COMPONENT
MAINTENANCE DATA
104
ANNUAL MAINTENANCE
MANHOUR BY ROUTINE
PRIORITY &
CRmCALTTY 10.13
Figure 7. Maintenance Manning and Availability System
-------
instrument Component Maintenance Data per Building Block
PRESSURE INSTRUMENTS |j FLOU INSTRUMENTS
T rans-
cont rol I er
Dp trans-
indt
or if ice
ventur i
insertion
turbine
ultrasonic
guages
mitters
switches
pilots
-------
Mechanical Routine Maintenance Schedules
1.10 ALLOT LIKED DUCTWORK DUCTWORK
Total
2.01 BOOSTER fANS & REGEN' BOOSTER FAN
HEAT EXCHANGER RECYCLE FAN
GAVO
EXTRA
Total
2.02 BOOSTER FAN & REGEN LUBE OIL COOL.FAN
HEXCK CONT'D
2.03 PURGE FAN
2.04 PURGE PUMP
2.05 QUENCH TANK
2.06 ABSORBER RECYCLE
SEAL FAN
LUBE OIL PUMP
HYDRAUL1C OIL PUHP
FAN (D)
EXTRA
PUMP
EXTRA
TANK
EXTRA
ABSORBER
PUHP
NOZ2LE ( 770 )
AGITATOR
NO. OF
COMPONENTS
ROUTINE MAINTENANCE Of MAJOR MECHANICAL COMPONENTS (EXCLUDING VALVES)
MANHOURS PER COMPONENT I TOTAL IMHRS. PER COMPONENT | AVERAGE
MONTHLY 3 MONTHLY 6 MONTHLY
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
1.0
1.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0 0
0.0
0.0
3.5
0.0
7.0
0.0
0.0
0.0
0.0
1.5
0.0
0.0
0.0
0.0
2.0
0.0
4.0
0.0
6.0
2.0
2.0
2.0
2.0
16.0
4.0
0.0
8.0
0.0
0.0
0.0
1.0
0.0
1.0
0.0
0.0
2.0
2.0
SUB-YEARLY
0.0
0.0
3.7
0.0
7.4
11.2
22.4
7.4
7.4
7.4
7.4
29.0
14.9
0.0
14.9
26.1
0.0
26.1
1.9
0.0
1.9
0.0
28.0
3.7
22.3
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
32.0
32.0
128.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
50.0
100.0
50.0
0.0
100.0
941.0
1091.0
50.0
50.0
0.0
0.0
200.0
20.0
0.0
40.0
0.0
0.0
0.0
2.0
0.0
2.0
200.0
0.0
50.0
20.0
31.0
31.0
15.5
0.0
31.0
291.7
338.2
31.0
31.0
59.5
59.5
181.0
12.4
0.0
12.4
0.0
0.0
0.0
0.6
0.0
0.6
62.0
0.0
15.5
37.2
Figure 9. Gas Path Equipment Maintenance Manhours (Example)
k.
4A-21
-------
Intentionally Blank Page
4-A-22
-------
SPARING ANALYSIS FOR FGD SYSTEMS
C. E. Dene
J. Weiss
Electric Power Research Institute
3412 Hillview Avenue
Palo Alto, California 94303
M. A. Twombly
J. Witt
ARINC Research Corporation
2551 Riva Road
Annapolis, Maryland 21401
Preceding page blank
4A-23
-------
Intentionally Blank Page
""--•v.
4A-24
-------
ABSTRACT
With the passage of federal clean air legislation, utilities will be evaluating the
capability of various flue gas desulfurization (FGD) system design configurations
and operating scenarios to meet sulfur dioxide (SO2) removal goals. The primary
goal in reviewing these alternatives will be to optimize SO2 removal capability in
relation to power production costs. The Electric Power Research Institute (EPRI) and
its contractor, ARINC Research Corporation, have developed an automated FGD
Analysis System that can evaluate competing FGD design alternatives in terms of
their SO2 removal capability and operating costs.
The FGD Analysis System can be used to evaluate different design configurations for
new systems or to calculate the effect of changes in component reliability for existing
FGD systems. The system is based on the EPRI UNIRAM methodology and
evaluates the impact of alternative FGD component configurations on the expected
unit emission rates. The user interactively enters FGD design data, unit SO2
generation-level data, and FGD chemical additive-level data for the design
configuration to be evaluated. The system then calculates expected SO2 removal
capability and operating cost data for operation of the design configuration over a
user specified time period. This paper provides a brief description of the FGD
Analysis System and presents sample results for three typical design configurations
with different redundancy levels.
Preceding page blank 4A-25
k.
-------
INTRODUCTION
With the passage of federal clean air legislation, utilities will be evaluating the
capability of various flue gas desulfurization (FGD) system design configurations
and operating scenarios to meet sulfur dioxide (SO2) removal goals. In order to
provide its member utilities with an automated capability for evaluating competing
FGD design alternatives, EPRI and its contractor, ARINC Research Corporation,
have developed the FGD Analysis System. The FGD Analysis System is a user-
friendly, personal computer (PC) software package which uses FGD design data, unit
SO2 generation-level data, and FGD chemical additive-level data to estimate
expected SO2 removal capability and operating cost data. This paper provides a brief
overview of the FGD Analysis System and presents sample results for three typical
design configurations with different redundancy levels.
OVERVIEW
The FGD Analysis System is designed for use on a 286 microprocessor-based PC with
at least 640 kilobytes of memory, a hard disk, enhanced graphics adapter (EGA), and
an EGA monitor. Users interact with the system through a graphical user interface
(GUI). The GUI provides easy-to-use menus and data forms to enter the required
data and perform evaluations. The GUI also provides utilities for viewing and
printing output reports and input files. Figure 1 illustrates the user data interaction
with the system for a typical design evaluation. The user enters FGD design data,
SO2 generation-level data, and FGD chemical additive-level data. The FGD analysis
system evaluates the FGD design and generates a report detailing the expected SO2
removal capability and operating cost data.
FGD Design Data
The FGD Analysis System provides a series of data forms for the user to enter the
FGD design data. The data are stored in an FGD design data file. The FGD Analysis
System evaluates FGD systems with an Availability Block Diagram (ABD) of the
form shown in Figure 2. (An ABD is one type of graphical representation of the
4A-26
A
A
-------
component configuration.) The user is able to vary the quantity and reliability data
for each individual component type. By varying component quantity data the user
can easily represent many different redundancy and sparing alternatives including:
hot-standby redundancy, cold- standby redundancy, and warehoused spares. The
user can also vary the reliability and maintainability data for a component to
determine whether a more reliable or maintainable component would provide a
better return on investment than adding component redundancy.
An example FGD design data file is shown in Figure 3. The FGD design data file
contains an ABD of the FGD system, the planned operating hours of the FGD
system, and the component quantity and reliability data for each component type.
The ABD contained in the file is a condensed version of the generic ABD shown in
Figure 2. Instead of drawing parallel components, the condensed ABD in Figure 3
represents the number of components in parallel with total and cold-standby
quantity figures placed outside the upper right corner of the rectangle representing
the component.
Unit SO? Generation-Level Data
The FGD Analysis System provides a series of data forms for the user to enter the
unit SO2 generation-level data. The data are stored in a SO2 generation-level data
file. A sample SO2 generation-level data file is shown in Figure 4. The user must
enter data characterizing the gas flow and SO2 flow in the gas emission states
produced by the power-generation unit serviced by the FGD. The number of gas
emission states (maximum of 40) may, or may not, correspond to the number of
power-generation states for the power-generation unit. Because the gas emission
state data will vary based on the type of coal being used, the user can vary the data to
investigate the effects of using different types of coal on expected emission rates and
removal efficiency.
FGD Chemical Additive-Level Data
The FGD Analysis System provides a series of data forms for the user to enter the
FGD chemical additive-level data. The data are stored in a chemical additive-level
data file. A sample chemical additive-level data file is shown in Figure 5. The
chemical additive-level data consists of the operating cost and the SO2 removal
efficiency curves associated with using each additive level. The user must enter
data for at least one additive level (Level 0)which corresponds to operation without
chemical additives. The user is allowed to enter additional data for two more
additive levels which correspond to operation with different chemical additives.
4A-27
-------
The user specifies when to use the chemical additives by setting additive thresholds.
An additive threshold is the operating absorber pump deficit (number of failed
absorber pumps) that will cause the use of chemical additives to become necessary.
By varying the additive thresholds, the user can investigate possible additive
strategies and their effect on expected emission rates and removal efficiency.
Evaluation Process
The FGD Analysis System uses the UNIRAM modeling methodology1 to calculate
the expected SO2 removal capability data. The UNIRAM modeling methodology is
an availability assessment methodology that allows the user to relate failures at the
component level to system- level capability losses. The FGD Analysis System uses
the analysis routines found in UNIRAM and FGD specific cold-standby combination
routines to calculate the expected liquid flow states of the FGD system at the absorber
level. These liquid flow states are then combined with the gas emission state data
and the chemical additive- level efficiency curve data to calculate the effective SO2
removal states for the FGD design. During the combination process, results are
calculated for the various absorber pump operating levels.
A weighted combination of the results is then calculated for each absorber. This
weighting process results in the overall SO2 removal figures for each effective state
and the state's associated additive- level probabilities. The additive-level
probabilities are then used to calculate the expected additive usage times and cost
figures. Once calculated, the SO2 removal data and operating cost data are used to
create an FGD design alternative evaluation report (as shown in Figure 6).
SAMPLE RESULTS
As an example of the application of the FGD Analysis System, consider the
evaluation of three typical FGD designs to determine which one provides the
greatest expected total SO2 removal efficiency. The three design configurations are:
• A single 100% capacity absorber with a maximum liquid flow rating of
176,000 GPM. The absorber will use eight active absorber pumps each rated
at 22,000 GPM (12.5% capacity). (ONEIOOA2)
* User's Guide for the UNIRAM Availability Assessment Methodology: Version 3.0 RP-3199-5.
Palo Alto, California: Electric Power Research Institute, October 1990
2 File Designator in output file.
4A-28
A
A
-------
• Two 75% capacity absorbers each with a maximum liquid flow rating of
132,000 GPM. The absorbers will each use four active absorber pumps
rated at 33,000 GPM (18.75% capacity). (TW075A22)
• Three 50% capacity absorbers each with a maximum liquid flow rating of
88,000 GPM. Two absorbers will be active and one will be maintained in a
cold-standby configuration. The absorbers will each use four active
absorber pumps rated at 22,000 GPM (12.5% capacity). (THREE50A22)
Each design configuration will also contain two 75% capacity reagent prep
components. The component reliability data shown in Figure 3 will be used for
each design configuration. The unit SO2 generation- level data and FGD chemical
additive-level data shown in Figures 4 and 5 will also be used for each design
configuration. The FGD Analysis System also provides a tool that will extract key
information from the design alternative reports and generate an FGD comparison
table. This allows the user to quickly compare competing alternatives based on the
key analysis results. The above mentioned design configurations were evaluated
and the results used to create a comparison table (as shown in Figure 7). The results
contained in Figure 7 show that using two 75% absorbers yielded the greatest
expected total removal efficiency. Its value of 0.9923 was significantly higher than
using either one 100% absorber or three 50% absorbers with one in a cold-standby
configuration.
Another possible design alternative would be to add redundancy at the absorber
pump level. Subsequent evaluations of the three example design configurations
were performed after adding one cold-standby absorber pump to each absorber
(alternative B) and after adding one warehoused spare absorber pump per active
absorber (alternative C). The FGD comparison table containing all nine design
alternatives is shown in Figure 8. The following observations can be made about
the results of the analysis shown in Figure 8:
• When absorber pump redundancy was considered, the design
configuration using two 75% absorbers still yields the greatest expected
total removal efficiency.
• The effect of a cold-standby absorber pump (alternative B) was essentially
the same as a warehoused spare absorber pump (alternative C).
The first observation is expected, since this provides the greatest installed
redundancy of equipment. The second observation is due largely to the relative
short meantime to restore (MTTR) used for absorber pumps in this set of data.
4A-29
-------
Although a MTTR of two hours has been reported both in the United States and
European FGD installations, longer MTTR times would result in lower overall
removal when using warehouse spares compared to cold stand-by pumps.
CONCLUSIONS
EPRI and its contractor, ARINC Research Corporation, have developed an
automated FGD Analysis System that can evaluate FGD design configurations to
determine expected SO2 removal capability and operating cost data. The system
provides easy to use menus and data forms to enter data and perform evaluations
quickly. The system allows the user to analyze the effects of sparing and
redundancy, different types of coal, and different additive-level strategies on FGD
SO2 removal efficiency and emission rates quickly and inexpensively.
The advantage of this system is first to provide an objective evaluation of how
different sparing and redundancy strategies effect the expected emissions for a given
time period. The second advantage is to predict emissions for operating FGD
systems. By calibrating the program to actual FGD performance at each operating
stage; i.e. lost pump, absorber. The model can be used to predict the SO2 emissions
with changing equipment performance characteristics. Thus, the operating strategy
(additive level, L/G, etc.) can be changed to compensate for the increase outage rate
of pumps and absorbers.
4A-30
A
A
-------
FGD Design Data
SO2 Generation-Level Data
FGD Chemical Additive-Level Data
S02 Removal Capability and
Operating Cost Data
FGD
Analysis
System
Figure 1. User Interaction With The FGD Analysis System
Absorber
Absorber
Reagent Preparation
Absorber Pump
Reagent Preparation
Absorber Pump
Absorber Pump
Absorber Pump
Absorber Pump
Absorber Pump
Figure 2. Generic FGD Availability Block Diagram
4A-31
-------
FGD Analysis System
FGD Design Data File
File N^me : SAMPLE.FDD
2/0
1
Reagent Prep
Absorber Pump
Planned Operating Hours
Component Name
Throughput Capacity (X)
Total Quantity
Cold-Standby Quantity
Mean Time Between Failures (h)
Mean Downtime (h)
Surge Time (h)
Delay Time (h)
Component Name
Throughput Capacity (X)
Total Quantity
Cold-Standby Quantity
Mean Time Between Failures (h)
Mean Downtime (h)
Surge Time (h)
Delay Time (h)
Minimum Pumps for Operation
Component Name
Throughput Capacity (X)
Total Quantity/Absorber
Cold-Standby Qty/Absorber
Mean Time Between Failures (h)
Mean Downtime (h)
Surge Time (h)
Delay Time (h)
MTTR (h)
Mean Logistics Time (h)
No. of Warehoused Spares
Active Serviced Components
Liquid Flow Rating (GPM)
: 8400.00
: REAGENT PREP
: 75.00
: 2
: 0
: 60000.00
: 200.00
: 0.00
: 0.00
: ABSORBER
: 100.00
: 1
: 0
: 10000.00
: 100.00
: 0.00
: 1.00
: 1
ABSORBER PUMP
12.50
8
0
5000.00
50.00
0.00
0.25
2.00
48.00
-1
0
22000.00
Figure 3. Example FGD Design Data File
4A-32
A
-------
FGD Analysis System
SO2 Generation-Level Data File
File Name : SAMPLE.SGL
Number of Gas Emission States : 3
Constant SO2 Concentration (PPM) : 0
FGD Inlet Temperature (deg F) 300.00
FGD Inlet Pressure (PSF) : 2116.20
State
1
2
3
Capacity
100.00
50.00
0.00
Probability
0.600000
0.300000
0.100000
Gas Flow
(CFM)
1465000.00
879000.00
0.00
SO2 Flow
(lb/h)
23275.00
13965.00
0.00
Figure 4. Example SO2 Generation-Level Data File
4A-33
-------
FGD Analysis System
Chemical Additive-Level Data File
File Name : SAMPLE.ALD
Number of Additive Levels
Number of Efficiency Curve Points
Additive Thresholds
Operating Costs ($/h)
3
6
1-1 2-2
1- 25.00 2- 50.00
Removal Efficiency
Point
L/G (gal/100 ft3)
Level 0
Level 1
Level 2
1
400.00
0.9960
0.9960
0.9960
2
150.00
0.9950
0.9950
0.9950
3
120.00
0.9600
0.9750
0.9750
4
90.00
0.8350
0.8950
0.9100
5
60.00
0.6400
0.7450
0.7800
6
0.00
0.0000
0.0000
0.0000
Figure 5. Example Chemical Additive-Level Data File
4A-34
A
A
-------
FGD Analysis System
FGD Design Alternative Evaluation Report
Filename : SAMPLE.DAE
Planned Operating Hours
Total SO2 Generated (lb)
Total SO2 Removed (lb)
Total SO2 Emitted (lb)
Total Removal Efficiency
Total Chemical Additive Cost ($)
8400.00
152497824.00
145871744.00
6626080.00
0.9565
14811.99
FGD Effective SO2 Removal States
Additives Probabilities
Emission Operating
State
Removal
Rate
Rem.
Time
No.
Prob.
Rate (lb/h)
(lb/h)
Eff.
(h)
ABS
Level 0
Level 1
Level 2
0.590118
22301.87
973.13
0.9582
4956.99
1
0.923484
0.073879
0.002638
0.003934
19551.99
3723.01
0.8400
33.05
1
0.923484
0.073879
0.002638
0.005948
0.00
23275.00
0.0000
49.96
0
0.000000
0.000000
0.000000
0.295059
13897.87
67.13
0.9952
2478.50
1
0.923484
0.073879
0.002638
0.001967
13895.19
69.81
0.9950
16.52
1
0.923484
0.073879
0.002638
0.002974
0.00
13965.00
0.0000
24.98
0
0.000000
0.000000
0.000000
0.100000
0.00
0.00
0.0000
840.00
0
0.000000
0.000000
0.000000
Additive Level Usage
Cost Rate Usage Time Usage Cost
Additive Level ($/h) (h) ($)
0 0.00 6912.33 0.00
1 25.00 552.99 13824.66
2 50.00 19.75 987.32
Figure 6. Example FGD Design Alternative Evaluation Report
4A-35
-------
FGD Analysis System
FGD Comparison Table
File Name : SAMPLE.FCT
Planned
Total SO2
Total SO2
Total
Additive -level
Operating
Removed
Emitted
Rem
Usage Time (h)
Design Alt.
Hours
(lb)
(lb)
Eff
Level 1 Level 2
ONEIOOA
8400.00
145871744.00
6626080.00
0.9565
552.99 19.75
TW075A
8400.00
151329904.00
1167904.00
0.9923
575.44 8.69
THREE50A
8400.00
147237520.00
5260288.00
0.9655
581.08 8.77
Figure 7. Comparison Table For Absorber Design Configurations
4A-36
A
A
-------
FGD Analysis System
FGD Comparison Table
File Name : SAMPLEl.FCT
Design Alt.
Planned
Operating
Hours
Total SO2
Removed
(lb)
Total SO2
Emitted
(lb)
Total
Rem
Eff
Additive
Usage
Level 1
-level
Time (h)
Level 2
ONEIOOA
8400.00
145871744.00
6626080.00
0.9565
552.99
19.75
ONEIOOB
8400.00
146085984.00
6411824.00
0.9580
25.07
0.53
ONEIOOC
8400.00
146082608.00
6415216.00
0.9579
34.93
0.07
TW075A
8400.00
151329904.00
1167904.00
0.9923
575.44
8.69
TW075B
8400.00
151385056.00
1112736.00
0.9927
14.48
0.12
TW075C
8400.00
151384160.00
1113648.00
0.9927
26.78
0.02
THREE50A
8400.00
147237520.00
5260288.00
0.9655
581.08
8.77
THREE50B
8400.00
147532800.00
4965024.00
0.9674
14.62
0.12
THREE50C
8400.00
147526784.00
4971024.00
0.9674
27.04
0.02
Figure 8. Comparison Table For Absorber Pump Redundancy Alternatives
4A-37
-------
Intentionally Blank Page
\
N,
4A-38
-------
Increasing Draft Capability for Retrofit Flue Gas
Desulfurization Systems
4A-39
Preceding page blank
-------
Intentionally Blank Page
4 A-40
-------
Robert D. Petersen,
Brian E. Basel,
Raymond J. Mosier
Burns & McDonnell Engineering Co.
4800 East 63rd Street
Kansas City, Missouri 64130
ABSTRACT
The retrofit installation of flue gas desulfurization (FGD) systems results in
significantly higher draft losses for existing generating stations.
Consequently, the means for increasing draft capability must be included in many
FGD retrofit projects. Consideration is given to several alternatives for
increasing draft capability. Alternatives are developed for new induced draft
(ID) fans to replace the existing ID fans and for new booster fans to supplement
the existing ID fans. Both centrifugal and axial fans are evaluated, as are
different means of fan volume control. Each alternative is evaluated on the
basis of technical merit and economics. Presented are the development of fan
alternatives and results of the technical and economic evaluations.
Preceding page blank
4A-41
-------
INTRODUCTION
The addition of flue gas desulfurization (FGD) systems to existing coal-fired
generating stations results in substantially higher draft requirements, because
of pressure losses through the FGD system and additional ductwork needed to
interface with existing facilities. Several alternatives should be developed
and evaluated for meeting the increased draft requirements for each station.
These include rebuilding or modifying the existing induced draft fans, replacing
the existing induced draft fans with new induced draft fans, and supplementing
the existing induced draft fans with new booster fans. In addition, an
evaluation should be made of different fan volume control methods. These
include variable frequency drives, hydraulic coupling drives, variable pitch
blades, two speed motors, and variable inlet vanes. The installation of new
larger fans will require that the boiler and existing draft components be
evaluated for their capability to withstand greater negative pressures than
produced by the existing fans. The evaluation of different fan alternatives
should include technical considerations, capital costs, energy costs, and
capacity charges.
Two case studies were developed and evaluated for retrofit FGD projects to
determine the best means of increasing draft system capability. The first case
is a 250-megawatt unit for which six alternatives are evaluated involving new
booster fans and ID fans. In the second case, six alternatives for adding fan
capability are evaluated for a 650-megawatt unit.
4A-42
A
A
-------
TECHNICAL APPROACH
Though each retrofit project is unique, the technical approach to determine the
means for increasing draft capability will be similar in most cases. The
approach will generally include the following tasks:
• Establish the existing draft fan operating parameters prior to
installation of the FGD system.
• Establish the capability and operating characteristics of the existing
induced draft fans.
• Estimate induced draft fan operating parameters with the FGD system in
service.
• Develop several alternatives for meeting the increased draft requirements
imposed by the FGD system.
• Investigate the various methods available for fan volume control.
• Evaluate the boiler implosion potential of each draft fan alternative.
• Evaluate the capital and operating costs for each draft fan alternative.
Fan Operating Parameters
As part of the effort to estimate draft system requirements once the new FGD
systems are in service, the existing fan design capacities and system
requirements must be investigated. Fan manufacturer's data sheets and
performance curves are the primary sources for this information. However, it
may be prudent to perform field testing of the existing draft systems to
determine actual gas flows, system resistance, and fan capability.
In addition to establishing gas flow requirements, the pressure needed to pass
flue gas through the FGD system absorbers and ductwork must be evaluated for
sizing of new fans or evaluation of existing fan capability. The addition of an
FGD system may increase the pressure losses through the draft system by 15 to 20
inches W.C. at unit maximum continuous rating (MCR). This pressure range is
conservative and should account for fouling and pluggage of absorber components
that could occur during normal operations.
The design of the absorber module has a major influence on the pressure drop
across the FGD system. Open spray tower designs can be expected to have lower
module pressure drop than a tower with internals such as trays or packing. The
design pressure should also include an allowance for pluggage and fouling of gas
stream components. A problem which can occur is pluggage in the packing or the
k.
4A-43
-------
mist eliminator section of the absorber. The pressure drop across this part of
the module is commonly monitored to gauge the amount of scale building up.
Remedial action should be taken to prevent an excessive amount of buildup;
however, the fans should be sized to develop the pressure needed to pass the gas
through a partially fouled absorber or mist eliminator section. The arrangement
and length of ductwork from the ID fans to the absorbers and from the absorbers
to the stack will affect the fan pressure requirements but the impact will be
minor compared to the absorber modules and mist eliminators.
To ensure that draft fans do not limit a boiler's performance, it is usual
practice to apply safety margins to the MCR fan requirements to determine the
fan test block specifications. Safety margins are intended to account for
unusual or unexpected operating conditions. For fans on new units, it is not
unusual to apply safety margins of 20 percent to MCR flow, 40 percent to MCR
pressure, and 25° F to MCR temperature. However, for retrofitting new fans to
existing units, actual operating conditions can often be assessed and lower
safety margins can be used to avoid oversizing the new fans. Oversized fans not
only result in increased equipment cost and increased potential for boiler
implosion, but also often result in higher operating costs due to decreased
efficiencies at reduced fan loads.
Fan Volume Control
Because flue gas volume varies with unit load, draft fans must be provided with
some form of volume control. Several methods are available including variable
inlet vanes, variable speed hydraulic couplings, variable pitch blades (axial
fans only), two speed motors used in conjunction with variable inlet vanes or
variable pitch blades, and variable frequency drives. Generally, methods which
vary fan speed to control volume will provide the lowest power consumption and
energy costs. However, increased efficiency is usually accompanied by increased
equipment capital costs. On fans sized with large flow margins between maximum
continuous rating and test block, two speed motors have been used to reduce fan
power consumption and wear by allowing the fan to operate at low speed for flows
up to maximum continuous rating. If lower margins are used with fans on
retrofit FGD projects, the benefits of two speed motors are somewhat diminished.
Variable Inlet Vanes. Variable inlet vanes are a proven means of volume control
for centrifugal fans. By varying the position of the inlet vanes, resistance to
flow can be increased or decreased as needed to modulate flow. By imparting a
4A-44
-------
swirl to the air stream in the direction of rotation of the fan, inlet vanes are
able to reduce power consumption while reducing flow.
The major advantage of variable inlet vanes is low capital cost. However,
depending on fan loading, the variable inlet vanes low capital cost may be
offset by high energy costs as the result of poor efficiency at low load
operation. Assuming a 10 percent margin on flow and utilizing two speed motors,
the combined fan/variable inlet vanes/motor efficiency for a typical fan is
approximately as follows:
CENTRIFUGAL FAN
VARIABLE INLET VANES
Flow Efficiency
% MCR %
100 79
90 76
80 67
70 56
60 47
50 37
40 28
30 20
20 13
Variable Frequency Drives. Another volume control option for centrifugal fans
is variable frequency drives. Two classes of variable frequency drives (VFDs)
are available - induction motor drives and synchronous motor drives. Both use
similar electronics to receive AC power at 60 Hz and convert this power to
variable speed motor output power. The major components of the VFD are an input
convertor, DC link reactor, and output convertor. The synchronous motor drives
offer higher efficiency and greater flexibility of application than do the
induction motor drives.
Benefits of using VFDs include the following:
1. Reduced energy costs because of the ability of the VFD to
maintain high efficiency over a wide range of loads.
2. Reduced maintenance outages because of the modular
electronics and detailed diagnostics features which are
available.
3. Increased equipment life due to equipment operating at
reduced speeds.
4. Improved process control and flexibility.
k.
k.
4A-45
-------
5. Soft equipment starts attained by allowing the VFD to
control motor acceleration rate, thus limiting inrush
current to the motor and potential voltage drop problems.
6. Reduced fault contribution because the VFD can limit the
amount of fault current contributed to the power grid by the
motor it controls.
Dual channel VFDs, with each channel having a 6-pulse input and 6-pulse output,
minimize the total harmonics imposed on the auxiliary power system and provide
the highest reliability. Each channel is connected to one of two separate sets
of windings in a two-winding, six-phase motor. By sizing each related channel
and motor winding for the fan maximum continuous rating, the fan can be run at
maximum continuous rating even with one of the channels out of service.
However, due to increased harmonics and fan torsional vibrations during single
channel operation, it may not be possible to operate in this mode for extended
periods.
VFDs provide the highest efficiency over the widest range of loads of any fan
volume control method utilized on centrifugal fans. Fan test block usually is
set at the VFD's full speed rating. Assuming a 10 percent design margin on fan
flow, approximate values for variable frequency drive efficiency, including
motor losses, are as follows:
VARIABLE FREQUENCY DRIVES
Fan & Drive Flow
Speed. %
% MCR
Efficiency. %
100
110
94.5
80
88
94.5
70
77
93.5
60
66
91.5
50
55
87.5
40
44
80.0
30
33
60.0
Variable Speed Hydraulic Couplings. Variable speed hydraulic couplings have
been widely used in utility applications and are a proven means of achieving
increased efficiencies at reduced loads. Hydraulic couplings consist of two
major components--an impeller connected to the driving (motor end) shaft and a
runner connected to the driven (fan end) shaft. Hydraulic fluid between these
4A-46
-------
two components transmits power from the motor shaft to the fan shaft. There is
no mechanical connection between the impeller and the runner. By adjusting the
amount of hydraulic fluid in the coupling, the driven shaft speed can be varied
infinitely over the design range of the coupling.
Benefits of using variable speed hydraulic couplings include:
1. Reduced energy costs at reduced fan loads.
2. Low capital cost as compared to variable frequency drives.
3. Unloaded motor starts, thus limiting the inrush current to
the motor and potential voltage drop problems.
4. Reduced wear of fan rotating components at lower operating
speeds.
5. Improved process control and flexibility.
The efficiency of a hydraulic coupling is a function of the amount of slip
(velocity differential) between the impeller and the runner. As is the case
with variable frequency drives, fan test block is established at the hydraulic
coupling's full speed rating. Considering a 10 percent design margin on flow
and the motor losses incurred, the combined hydraulic coupling/motor efficiency
is approximately as follows:
HYDRAULIC COUPLINGS
Fan & Drive Flow
Speed. % % MCR Efficiency. %
100 110 92
90 99 83
80 88 73
70 77 63
60 66 54
50 55 44
40 44 33
30 33 24
Variable Pitch Blades. Variable pitch blades are an efficient means of varying
gas flow with axial fans. By varying blade pitch, the fan volume can be varied
to accommodate changes in boiler load or upsets in boiler operation. Benefits
of using variable pitch blades include the following:
k.
4A-47
-------
1. Reduced energy costs because of the ability to maintain high
efficiency over a wide range of loads.
2. Rapid response to changes in boiler load or to upset conditions.
3. Improved process control and flexibility.
The efficiency of an axial fan with variable pitch blades will vary widely
within its operating range. Fan efficiencies for a typical axial fan is
approximately as follows:
AXIAL FAN WITH VARIABLE PITCH BLADES
Flow Efficiency. %
100 88
90 89
80 87
70 80
60 70
50 55
Boiler Implosion
New ID fans, sized to overcome existing draft losses plus the pressure drop of
the new FGD system, may have much greater head capability than the existing ID
fans. Therefore, the installation of new ID fans may increase the potential for
developing negative pressure excursions, in excess of original design pressures,
within the furnace, air supply system, air heater, particulate removal
equipment, and ductwork. The National Fire Protection Association (NFPA) 85C
Standard for the Prevention of Furnace Explosion/Implosions in Multiple Burner
Boiler Furnaces establishes minimum standards for the design, installation, and
operation of boilers and related equipment to prevent furnace implosion. NFPA
85C applies to new installations and major alterations, or extensions of
existing installations. It is likely that a FGD retrofit project will be
considered a major alteration of the draft system and will subject the furnace
to new draft conditions.
4A-48
-------
NFPA 85C offers the following two methods for minimizing the risk of negative
furnace draft in excess of furnace structural capability:
• Design the furnace so that the maximum head capability of
the induced draft fan system with ambient air does not
exceed the furnace design pressure.
• Provide a furnace pressure control system in accordance with
the standard and design the furnace for minimum design
pressure of -35 inches W.C.
Additional studies should be performed during the FGD project preliminary design
phase to determine the impact of new ID fans on the draft system. The pressure
capability of the existing boiler, air heaters, particulate removal equipment,
and ductwork must be evaluated and a determination made of the modifications
needed to meet the increased draft pressures imposed by the new ID fans. The
existing draft controls should also be reviewed to determine any modifications
which might be needed.
ALTERNATIVES FOR INCREASING DRAFT CAPABILITY
Case Study A
Case Study A considers the draft modifications required by the retrofit
installation of an FGD system on an existing 250-megawatt unit. The FGD system
is assumed to be an open spray tower design. For sizing of new fans, it was
assumed that the draft losses will be increased by 15 inches W.C. at maximum
continuous rating. The unit is assumed to have two existing centrifugal ID fans
equipped with two speed motors and variable inlet vanes. In sizing new fans,
design margins of 10 percent and 21 percent were applied to MCR flow and
pressure respectively to determine fan test block requirements. In all
alternatives, the fans are placed upstream of the FGD system. Fan operating
parameters are indicated in Table 1. Six alternatives for increasing fan
capability were evaluated.
Alternative No. 1. Replace the existing centrifugal ID fans with two new
centrifugal fans with variable frequency drives. Each fan would have a test
block rating of 502,000 acfm and 42.4 inches W.C. static pressure rise, and a
maximum continuous rating of 456,000 acfm and 35 inches W.C. static pressure
rise.
4A-49
-------
Alternative No. 2. Replace the existing ID fans with two new centrifugal fans
with hydraulic couplings and constant speed motors. Fan flow and pressure
ratings would be the same as for Alternative No. 1.
Alternative No. 3. Replace the existing ID fans with two new centrifugal fans
with variable inlet vanes and two speed motors. Fan flow and pressure ratings
would be the same as for Alternatives No. 1 and 2.
Alternative No. 4. Install two new centrifugal booster fans with variable
frequency drives and in series with the existing ID fans. Each new booster, fan
would have a test block rating of 502,000 acfm and 18.2 inches W.C. static
pressure rise, and a maximum continuous rating of 456,000 acfm and 15 inches
W.C. static pressure rise.
Alternative No. 5. Install two new centrifugal booster fans with hydraulic
couplings and constant speed motors, and in series with the existing ID fans.
Fan flow and pressure ratings would be the same as Alternative No. 4.
Alternative No. 6. Install two new centrifugal booster fans with variable inlet
vanes and two speed motors. Fan flow and pressure ratings would be the same as
for Alternatives No. 4 and 5.
Case Study B
Case Study B evaluates six alternatives for increasing the draft capability of a
650-megawatt unit to be retrofitted with an FGD system. Like Case Study A, this
case assumes an open spray tower absorber with a design pressure drop of
15 inches W.C. Design margins of 10 percent on MCR flow and 21 percent on MCR
pressure were used to size new fans. The placement of all fans is upstream of
the FGD system. Fan operating parameters are indicated in Table 2.
The unit is assumed to have two existing axial ID fans with variable pitch blade
control. Six alternatives were considered for increasing fan capability.
Alternative No. 1. Install two new axial ID fans with constant speed motors to
replace the existing ID fans. The new fans would be equipped with variable
pitch blades for flow control. Each fan would have a test block rating of
1,365,000 acfm and 42.4 inches W.C. static pressure rise, and maximum continuous
rating of 1,241,000 acfm and 35 inches W.C. static pressure rise.
4A-50
A
A
-------
Alternative No. 2. Install two new centrifugal fans with variable frequency
drives to replace the existing ID fans. Fan flow and pressure ratings would be
the same as Alternative No. 1.
Alternative No. 3. Install two new centrifugal fans with hydraulic couplings
and constant speed motors to replace the existing ID fans. Fan flow and
pressure ratings would be the same as Alternatives No. 1 and 2.
Alternative No. 4. Install two new centrifugal booster fans with variable
frequency drives and in series with the existing ID fans. Each fan would have a
test block rating of 1,365,000 acfm and 18.2 inches W.C. static pressure rise,
and maximum continuous rating of 1,241,000 acfm and 15 inches W.C. static
pressure rise.
Alternative No. 5. Install two new centrifugal booster fans with hydraulic
couplings and constant speed motors and in series with the existing ID fans.
Fan flow and pressure ratings would be the same as for Alternative No. 4.
Alternative No. 6. Install two new axial booster fans with constant speed
motors and in series with the existing ID fans. Variable pitch blades would be
provided on the new fans for flow control. Fan flow and pressure ratings would
be the same as Alternatives Nos. 4 and 5.
EVALUATION OF DRAFT FAN ALTERNATIVES
Capital costs, energy costs, and capacity charges were estimated to calculate
the present value revenue requirements for each of the alternatives considered
for Case Studies A and B. It was assumed that planned unit outages for the FGD
projects are sufficiently long to perform the tie-ins needed for each of the
alternatives, therefore outage related replacement power costs were not
considered. The evaluated costs are indicated in Tables 3 and 4.
Capital Cost
The total installed cost in 1991 dollars was estimated for each alternative.
Installed costs include the fans, fan volume controls, motors, switchgear,
wiring, foundations, and sound attenuation. For Case Study A, the lowest
capital cost of $1,640,000 is associated with Alternative No. 6 - New
Centrifugal Booster Fans with Variable Inlet Vanes. Of the Case Study B
alternatives, Alternative No. 6 - New Axial Booster Fans with Variable Pitch
Blade Control has the lowest capital cost of $3,915,000.
k.
4A-51
-------
Energy Costs
Energy costs were determined by calculating fan power consumption at 10 percent
increments of unit load. Fan, variable speed drive, and motor efficiencies were
evaluated at each load increment. Although the fans were sized for 15 inches
W.C. increase in draft losses at MCR with the FGD system in operation, energy
costs were calculated assuming a normal increase of 10 inches W.C. at MCR.
Total kilowatt hours per year for each alternative were calculated. Energy
costs were determined using a 1991 rate of 25 mils per kilowatt hour.
For Case Study A, Alternative No. 1 - New Centrifugal ID Fans with Variable
Frequency Drives - has the lowest annual energy cost of $442,014. For Case
Study B, Alternative No. 2 - New Centrifugal ID Fans with Variable Frequency
Drives - has the lowest annual energy cost of $1,236,387.
Capacity Charges
Capacity charges were calculated for the alternatives at a 1991 rate of $1600
per kilowatt of total fan power consumption at fan test block rating. For Case
Study A, Alternative No. 1 - New Centrifugal ID Fans with Variable Frequency
Drives - has the lowest capacity charge of $9,585,600. For Case Study B,
Alternative No. 1 - New Axial ID Fans with Variable Pitch Blade Control - has
the lowest capacity charge of $25,544,000.
Present Value Revenue Requirements
Present value revenue requirements (PVRR) were determined for each alternative
as follows:
1. An annual escalation rate of 5 percent was applied to all
equipment and construction costs.
2. Economic analysis was based on an annual discount rate of 11
percent.
3. Capital costs, annual energy costs, and capacity charges
assume an FGD system in-service date of January 1, 1995.
4. Capital costs, annual energy costs, and capacity charges in
1995 dollars are converted to levelized annual costs
assuming a unit retirement date of December 31, 2020.
5. The present value in 1995 of the levelized annual costs was
calculated to arrive at each alternative's PVRR.
4A-52
-------
The PVRR are shown on Tables 3 and 4. Case Study A alternatives range from a
lowest PVRR of $26,036,403 for Alternative No. 1 - New Centrifugal ID Fans with
Variable Frequency Drives - to a highest PVRR of $27,269,117 for Alternative
No, 2 - New Centrifugal ID Fans with Hydraulic Couplings. Of the Case Study B
alternatives, No. 6 - New Axial Booster Fans with Variable Pitch Blade Control -
has the lowest PVRR of $65,373,166. Case Study B, Alternative No. 3 - New
Centrifugal ID Fans with Hydraulic Couplings - has the highest PVRR of
$70,750,402.
CONCLUSION
Various alternatives are available for increasing capacity of the draft system
to meet the added pressure drop of a retrofit FGD system. Each application must
be analyzed carefully and specific alternatives developed and evaluated. The
best alternative for one unit's retrofit project may not be the best alternative
for another unit. Unit size, condition and capability of existing draft
equipment, FGD absorber design, unit loading, and unit life expectancy are among
the various factors influencing a utility's decision on how to best increase fan
capability to meet the requirements of the FGD system.
4A-53
-------
Table 1
Unit A - 250 MW
Induced Draft Fan Operating Parameters
Avg.
Unit
Load
MW
Avg.
Gas Flow
per Fan
ACFM
Draft
System
Resistance
in. W.C.
Alt. No. 1
Total
Efficiency
%
Alt. No. 2
Total
Efficiency
%
Alt. No. 3
Total
Efficiency
%
Alt. No. 4
Total
Efficiency
%
Alt. No. 5
Total
Efficiency
%
Alt. No. 6
Total
Efficiency
%
Exist. Fans
Total
Efficiency
%
Operating
Hours per
Year
250
456,000
30.0
83.2
73.2
79.4
83.2
73.2
79.4
77.8
1200
225
423,400
25.9
83.2
67.7
76.3
83.2
67.7
76.3
71.1
1200
200
391,000
22.1
83.0
62.5
66.6
83.0
62.5
66.6
60.0
1600
175
358,600
18.5
82.5
57.1
56.1
82.5
57.1
56.1
50.6
1200
150
326,300
15.3
81.6
51.5
46.5
81.6
51.5
46.5
40.8
1200
125
294,000
12.5
80.3
46.1
37.0
80.3
46.1
37.0
31.6
1000
100
261,500
9.8
77.9
40.3
28.4
77.9
40.3
28.4
23.8
400
75
229,100
7.5
74.8
34.8
19.8
74.8
34.8
19.8
17.1
100
50
196,700
5.6
69.5
29.1
13.2
69.5
29.1
13.2
11.6
30
Notes:
(1) Alternatives are as follows:
No. 1 - New centrifugal I.D. fans with variable frequency drives.
No. 2 - New centrifugal I.D. fans with hydraulic couplings.
No. 3 - New centrifugal I.D. fans with variable inlet vanes.
No. 4 - New centrifugal booster fans with variable frequency drives in series with existing I.D. fans.
No. 5 - New centrifugal booster fans with hydraulic couplings in series with existing I.D. fans.
No. 6 - New centrifugal booster fans with variable inlet vanes in series with existing I.D. fans.
(2) Existing fans are centrifugal I.D. fans with variable inlet vanes.
(3)Total Efficiency=Fan Efficiency x Drive Efficiency x Motor Efficiency.
-------
r
Table 2
Unit B - 650 MW
Induced Draft Fan Operating Parameters
Avg.
Unit
Load
MW
Avg.
Gas Flow
per Fan
ACFM
Draft
System
Resistance
in. W.C.
Alt. No. 1
Total
Efficiency
%
Alt. No. 2
Total
Efficiency
%
Alt. No. 3
Total
Efficiency
%
Alt. No. 4
Total
Efficiency
%
Alt. No. 5
Total
Efficiency
%
Alt. No. 6
Total
Efficiency
%
Exist. Fans
Total
Efficiency
%
Operating
Hours pei
Year
650
1,241,000
30.0
80.4
83.2
73.2
83.2
73.2
80.4
80.4
1700
585
1,149,000
25.7
77.3
83.2
67.3
83.2
67.3
77.3
77.3
1600
520
1,058,000
21.8
71.3
82.7
62.1
82.7
62.1
71.3
71.3
1200
455
966,000
18.2
65.4
82.3
56.4
82.3
56.4
65.4
65.4
1000
390
874,000
14.9
56.6
81.4
50.6
81.4
50.6
56.6
56.6
1000
325
783,000
12.0
47.8
79.6
44.9
79.6
44.9
47.8
47.8
500
260
691,000
9.3
37.2
77.0
39.0
77.0
39.0
37.2
37.2
100
195
599,000
7.0
31.4
73.5
33.2
73.5
33.2
31.4
31.4
50
130
508,000
5.1
22.0
66.9
27.4
66.9
27.4
22.0
22.0
30
Notes:
(1) Alternatives are as follows:
No. 1 - New axial I.D. fans with variable pitch blade control.
No. 2 - New centrifugal I.D. fans with variable frequency drives.
No. 3 - New centrifugal I.D. fans with hydraulic couplings.
No. 4 - New centrifugal booster fans with variable frequency drives in series with existing I.D. fans.
No. 5 - New centrifugal booster fans with hydraulic couplings in series with existing I.D. fans.
No. 6 - New axial booster fans with variable pitch blade control in series with existing I.D. fans.
(2) Existing fans are axial ID fans with variable pitch blade controls.
(3) Total Efficiency=Fan Efficiency x Drive Efficiency x Motor Efficiency.
-------
Table 3
Unit A - 250 MW
Induced Draft Fan Alternatives
Present Value Revenue Requirements
>
i
U1
ON
Alt.
No. 1
Alt.
No. 2
AK.
No. 3
AK.
No. 4
Ah.
No. 5
Alt.
No. 6
New Centrifugal I.O.
Fans with Variable
Frequency Drives
New Centrifugal I.O.
Fans with Hydraulic
Couplings
New Centrifugal I.D.
Fans with Variable
Inlet Vanes
New Centrifugal Booster
Fans with Variable
Frequency Drives
Capital Costs, 1991$s
New Centrifugal Booster
Fans with Hydraulic
Couplings
New Centrifugal Booster
Fans with Variable
Inlet Vanes
Fans
700,000
700,000
780,000
500,000
500,000
570,000
Variable Speed Drives
2,000,000
700,000
NA
1,300,000
460,000
NA
Motors
Incl. w/ Drives
400,000
600,000
Incl. w/ Drives
232,000
350,000
Equipment Foundations
160,000
160,000
150,000
140,000
130,000
120,000
Equipment Installation
200,000
250,000
180,000
150,000
175,000
140,000
Power Supplies
330,000
110,000
200,000
250,000
110,000
200,000
Controls & Instruments
25,000
25,000
40,000
35,000
35,000
40,000
Sound Attenuation
230,000
230,000
230,000
220,000
220,000
220,000
TOTAL
3,645,000
2,575,000
2,180,000
2,595,000
1,862,000
1,640,000
Annual Energy Cost, 1991 $s
442,014
590,315
588,592
578,278
627,857
627,347
Capacity Charge, 1991 $s
9,585,600
9,828,800
9,963,200
9,795,200
9,899,200
9,964,669
Present Value Revenue
Requirements, 1995 $s
Capital Cost
5,193,432
3,668,886
3,106,086
3,697,382
2,652,996
2,336,688
Energy Cost
7,185,312
9,596,059
9,568,060
9,400,389
10,206,350
10,198,052
Capacity Charge
13,657,659
14,004,173
14,195,667
13,956,299
14,104,479
14,197,760
TOTAL
26,036,403
27,269,117
26,869,813
27,054,070
26,963,825
26,732,500
Rank (lowest cost = 1)
[1]
[6]
[3]
[5]
[4]
[2]
Notes:
(1) Annual energy cost is based on 25 mils per kWhr.
(2) Capacity charge is based on $1600 per kW.
(3) Based on 1995 in-service date, 2020 unit retirement date, 5 percent annual escalation rate, and 11 percent annual discount rate.
(4) Alternatives No. 4, 5, and 6 energy costs and capacity charges include existing induced draft fans and new booster fans.
k.
-------
Table 4
Unit B-650 MW
Induced Draft Fan Alternatives
Present Value Revenue Requirements
Alt.
No. 1
Alt.
No. 2
Alt.
No. 3
Alt.
No. 4
Alt.
No. 5
Alt
No. 6
New Axial I.D.
Fans with Variable
Pitch Blade Control
New Centrifugal I.D.
Fans with Variable
Frequency Drives
New Centrifugal I.D.
Fans with Hydraulic
Couplings
New Centrifugal Booster
Fans with Variable
Frequency Drives
New Centrifugal Booster
Fans with Hydraulic
Couplings
New Axial Booster
Fans with Variable
Pitch Blade Control
Capital Costs, 1991 $s
Fans
3,100,000
1,250,000
1,250,000
1,100,000
1,100,000
2,000,000
Variable Speed Drives
N/A
4,100,000
1,150,000
2,250,000
1,010,000
N/A
Motors
1,620,000
Incl. w/Drives
1,110,000
Incl. w/ Drives
890,000
840,000
Equipment Foundations
410,000
450,000
450,000
420,000
420,000
350,000
Equipment Installation
350,000
300,000
350,000
250,000
275,000
300,000
Power Supplies
60,000
450,000
130,000
340,000
90,000
90,000
Controls & Instruments
25,000
25,000
25,000
35,000
35,000
35,000
>
i
Sound Attenuation
310,000
300,000
300,000
280,000
280,000
300,000
Ul
TOTAL
5,875,000
6,875,000
4,765,000
4,675,000
4,100,000
3,915,000
Annual Energy Cost, 1991 $s
1,413,660
1,236,387
1,590,150
1,354,348
1,472,694
1,413,660
Capacity Charge, 1991 $s
25,544,000
26,084,800
26,748,800
25,862,400
26,147,200
25,838,400
Present Value Revenue
Requirements 1995 $s
Capital Cost
8,370,759
9,795,569
6,789,220
6,660,987
5,841,721
5,578,131
Energy Cost
22,980,224
20,098,511
25,849,225
22,016,060
23,939,872
22,980,224
Capacity Charge
36,395,347
37,165,884
38,111,958
36,849,006
37,254,792
36,814,811
TOTAL
67,746,330
67,059,963
70,750,402
65,526,053
67,036,385
65,373,166
Rank (lowest cost = 1)
[5]
[4]
[6]
[2]
[3]
[1]
Notes: (1) Annual energy cost is based on 25 mils per kWhr.
(2) Capacity charge is based on $1600 per kW.
(3) Based on 1995 in-service date, 2020 unit retirement date, 5 percent annual escalation rate, and 11 percent annual discount rate.
(4) Alternatives No. 4, 5 and 6 energy costs and capacity charge include existing induced draft fans and new booster fans.
-------
Intentionally Blank Page
4A~ 58
-------
DEVELOPMENT OF ADVANCED RETROFIT FGD DESIGNS
C. E. Dene
Electric Power Research Institute
3412 Hillview Avenue
Palo Alto, California 94303
W. L. Boward
Sargent & Lundy
55 East Monroe St.
Chicago, 111 60603
J. G. Noblett
Radian Corporation
8501 Mo-Pac Blvd
Austin, TX 78759
R. J. Keeth
United Engineers & Constructors
5555 Greenwood Plaza Blvd.
EnglewoodCO 80111
k.
Preceding page blank
4A-59
-------
Intentionally Blank Page
4A-60
A
A
-------
ABSTRACT
The 1990 Clean Air Act Amendment is a dramatic departure from previous
legislation in that it affords the electric utility industry the flexibility to
achieve their portion of the sulfur dioxide reduction in a myriad of ways.
Each utility must look at its system overall. One strategy which may prove
beneficial is to remove as much SO2 as possible at facilities where there is an
existing flue gas desulfurization (FGD) system or where one is planned. In
response to this need EPRI is developing a family of advanced retrofit FGD
designs that incorporate recent advances in FGD technology. A range of
design options are being investigated to determine both the SO2 collection
capability and the relative cost impacts of each option.
Some of the design options considered include the use of trays, packing,
additional liquid flow rate, and additives to boost the removal efficiency.
These options are being investigated for limestone, and magnesium-
enhanced lime systems. The sensitivity of these designs to changes in coal
sulfur content, chloride content, unit size, gas velocity, and other factors are
being investigated to determine how the performance of a designs is changed
and the ability to meet compliance. This paper illustrates the type of analysis
used to develop the advanced designs and presents the sensitivity of a
Countercurrent spray tower design using limestone and forced oxidation to
changes in specific design input parameters such as boiler load, tower height,
and gas velocity.
Preceding page blank
4A-61
-------
INTRODUCTION
The 1990 Clean Air Act Amendment (CAAA) represents a dramatic departure
from previous legislation in that it affords the electric utility industry the
flexibility to achieve their portion of the sulfur dioxide reduction in a myriad
of ways. Each utility must look at its overall system. One strategy which may
prove beneficial is to remove as much SO2 as possible at facilities where there
is an existing flue gas desulfurization (FGD) system or where one is planned.
The purpose of this study is to investigate the cost of achieving various levels
of SO2 reduction, in $/ton of SO2 removed, for various wet FGD processes.
Collection efficiencies in excess of New Source Performance Standards (NSPS)
are being investigated to determine if there is a minimum of $/ton of SO2
removed.
The project has developed a methodology that utilizes two existing EPRI tools
to optimize SO2 Removal with respect to cost. FGD Process Integration and
Simulation Model (FGDPRISM) is used to determine the process design
parameters, which are then used to develop cost information with FGDCOST.
Through an iterative process the limits of a selected design can be
determined.
The overall project will evaluate the following process configurations:
1- Countercurrent, Limestone, Spray Tower, Forced Oxidation
2- Countercurrent, Limestone, Spray Tower, Inhibited Oxidation
3- Countercurrent, Limestone, Tray Tower, Forced Oxidation
4- Countercurrent, Limestone, Tray Tower, Inhibited Oxidation
5- Countercurrent, Mg-Lime, Spray Tower, Inhibited Oxidation
6- Countercurrent, Mg-Lime, Tray Tower, Inhibited Oxidation
7- Co-current, Limestone, Spray Tower, Forced Oxidation
8- Co-current, Limestone, Packed Tower, Forced Oxidation
4A-62
A
A
-------
9- Co-current, Mg-Lime, Spray Tower, Inhibited Oxidation
10- Co-current, Mg-Lime, Packed Tower, Inhibited Oxidation
However, rather than proceed with ten processes in parallel, the methodology
is first being tested on a single configuration. This paper will cover the
preliminary results from the configuration of a countercurrent spray tower
using limestone with forced oxidation.
DESIGN BASIS
Some aspects of the design basis are particular to each configuration (such as
% solids in the waste, or oxygen to SO2 ratio, etc.), but most aspects are
common to all configurations (e.g., coal composition, combustion conditions,
limestone type and grind, etc.). Table 1 shows the design basis that was
developed for the first configuration to be investigated, countercurrent spray
tower using limestone with forced oxidation.
The plant size chosen was a 600 MW (net) system using a single absorber
module. The absorber was assumed to be of circular cross section, constructed
of carbon steel with a rubber liner. The absorber diameter was calculated to be
58.8 ft. based on a gas velocity of 10 ft/sec. The coal combustion portion of the
FGDPRISM model was used to calculate the flue gas flow rate resulting from
using EPRI's standard high sulfur coal. A coal analysis is shown in Table 2.
For the detailed mechanical design of the absorber, the distance from the top
of the gas inlet to the first spray header was picked to be 10 ft, and the distance
between headers was set at 6 ft. The nozzles used were assumed to be 90
degree, hollow cone nozzles operating at 10 psi pressure drop producing a
2,000 micron Sauter mean droplet size. The number of nozzles per header
will vary as liquid-to-gas ratio is varied, to keep the flow per nozzle at about
300-350 gpm. Realizing that specific designs may vary with respect to these
parameters, they were chosen as sensitivity variables which will be
investigated to determine the impact on system performance and cost.
The total system oxidation level was set at 99%, using an oxygen to SO2 ratio
of 3.0, and a minimum sparger depth of 20 feet in the reaction tank. This
high oxidation level will allow concentration of the waste solids using a
vacuum filter to 85% or higher. Initial material balances using the
k
k.
4A-63
-------
FGDPRISM computer model showed that chloride levels for this
configuration (at 85% solids in the waste) would exceed 50,000 mg/1 in the
absorber liquid and would be even higher for higher concentrations of solids
in the waste. In order to simplify calculations for the base case a liquid
blowdown was specified to keep the circulating liquid chloride concentration
at 30,000 mg/1. This blowdown rate was changed as necessary for changes in
operating variables (such as additive concentration, L/G, coal composition,
etc.) to keep the circulating liquid chloride concentration at 30,000 mg/1. The
effect of circulating liquid chloride concentration on system performance was
also chosen as a sensitivity variable to be investigated.
METHODOLOGY AND EXAMPLE RESULTS
The overall approach to the project is to first run a series of FGDPRISM
simulations for each process configuration. Then, the results of the process
simulations are used as input to the FGDCOST model to calculate capital and
operating costs. Those results are then used to evaluate the tradeoffs in
system design considerations with costs. For three of the process system
configurations, a sensitivity analysis is planned, using both FGDPRISM and
FGDCOST. Both process design variables and cost parameters are being
investigated.
Since various L/G's can be used in conjunction with different number of
headers (and therefore different tower heights) to achieve a desired SO2
removal, a range of L/G's per header will be investigated. Figure 1 shows a
range L/G from 60 to 190 gal/Macf (satd), this corresponds to an L/G per
header of from 20 to 36, although typically most existing limestone spray
towers are limited to approximately 30 L/G per header.
FGDPRISM is used to prepare a series of curves that plot SO2 removal versus
Liquid to Gas Ratio (L/G) for various levels of organic acid additives. Figure 2
shows the effect of two levels of organic acid additive for a four header
system. The effect of additives are most pronounced at the lower L/G levels
but still provide increased removal at 120 L/G.
These results from the simulations are input to the FGDCOST model to
calculate capital and operating costs for each scenario. Figure 3 shows the
4A-64
A
A
-------
resulting costs compared to SO2 removal for a 4 header system with 0, 500,
and 1000 ppm DBA. The break in the curve represents the point at which an
additional pump is required on each header to achieve the higher L/G. Using
additives lowers the cost per ton of SO2 removed.
SENSITIVITY
Table 1 shows the variables chosen for investigating the sensitivity. As might
be expected the parameters that impact the sizing of equipment or structures
have the most effect on the costs. Therefore designs that minimize tower
height and diameter or the size of the reaction tank will result in reduced cost
at the expense of SO2 removal capability.
Typical designs for limestone spray towers use a gas velocity of 10 ft/sec
(saturated gas basis). This velocity has been chosen in the past as a means to
avoid mist eliminator scaling and plugging problems. However, with a
forced oxidation system, scaling is rarely a problem, and with improved mist
eliminator technology, operation at higher gas velocity may be feasible. The
impact of gas velocity on system performance is illustrated in Figure 4.
Header spacing is another important variable that will affect the tower size as
well as removal efficiency. The base design is 10 ft from the gas inlet to the
first header, and 6 ft between headers. The options investigated include 6 and
8 ft from the gas inlet to the first header, and 5 ft between four headers. Also,
several cases were run using the 3 header design, but increasing the distance
between the gas inlet and the first header to 16 ft, so that the overall tower
height remained constant (just the header arrangement within the two was
changed). Figure 5 shows the impact on cost of increasing the distance to the
first header for a design with 6 ft between headers; also shown is the affect of
using additives in the same system. Figure 6 illustrates the same relationship
for a system with only 5 ft between headers. It is interesting to note that the
effect of additives is much greater for the shorter tower. The use of additives
in the shorter tower increases the SO2 removal capability by 7-10%. The tower
with compact spray headers (5 feet between 4 headers) with 16 ft to the first
header (31 feet total spray zone) will achieve 97.5% SO2 removal with 1000
k.
4A-65
-------
ppm DBA at $366 per ton SO2 removed compared with a system with 6 ft
between headers and only 10 feet to the first header (28 foot spray zone) that
gets 94.6% SO2 removal at $373 per ton SO2 removed.
It is also important to consider the flexibility of the design to some of the
plant operating characteristics or requirements. For example, selecting a
design based on the lowest dollar per ton of SO2 removed may not be capable
of achieving the desired removal if it must be operated at a different design
point or it may not be the lowest cost system at those conditions. Therefore
the sensitivity of a design to such parameters as coal sulfur, boiler load, and
limestone quality need to be considered.
Figure 7 illustrates the impact on cost of reduced load operation. It is
interesting to note that the costs increase rapidly for low L/G system with no
additives. Therefore, the expected unit load capacity should influence the
selection of the best design point.
CONCLUSIONS
This methodology utilizing existing EPRI tools can identify lower cost wet
FGD system designs. The methodology can further identify the impact of
operating characteristics on overall cost. Limestone designs that feature closer
header spacing with extended inlet zones that use organic additives should be
capable of high SO2 removal at lower cost.
This is a methodology, and the results obtained by applying this methodology
will vary based on the underlying design criteria. The results reported here
represent cost trends utilizing large scale pumps and absorber where the
economy of scale will greatly influence the actual $/ton of SO2 removed. The
results obtained by using FGDPRISM is intended to indicate trends and
should not be expected to predict removal efficiency without calibration to a
specific design.
4A-66
A
-------
Table 1
Base Case Design Parameters and Sensitivity Ranges
Gas Velocity
10 ft/sec
8-12 ft/sec
Header Spacing
(dist to bottom header)
6 ft between
(10 ft)
5 ft
(6-16 ft)
Coal Sulfur
2.6%
3.0-3.5%
Ash Loading
0.02 gr/scf
0.05-0.10 gr/scf
Reagent Ratio
1.07
1.05-1.12
Limestone Grind
95% < 325 mesh
67-90% < 325 mesh
Limestone Reactivity
3
1-10
Recycle Slurry Solids
12%
10-15%
Solids Residence Time
15 hr.
10-20 hr.
Additive Type
DBA
Sodium Formate &
Formic Acid
Dissolved Chloride
30,000 ppm
10,000-50,000 ppm
Available Mg in Limestone
0.5% MgCC>3
0.2-1.0% MgC03
k.
4A-67
-------
Table 2
Base Coal analysis
Carbon
Hydrogen
Nitrogen
Sulfur
Oxygen
Chloride
Water
Ash
HHV
72.21%
4.79%
1.40%
2.60%
4.80%
0.12%
5.99%
9.09%
13,084 Btu/lb
4A-68
A
A
-------
a
>
o
£
a>
f
o
M
100
95
90
85
80
75
70
65
60
40
Figure 1.
SOAR Base Design
~i 1 1 1 1 1 1 r
i i—i 1 i i I | r
J I 1 i i I I i i I i
100 120 140
L/G, gal/1000 acf
•—
- 3 Headers
:
¦—
- 4 Headers
-
*—
- 5 Headers
-
I I I I I I I I I L
160 180 200
-------
£
a
>
o
E
a>
04
t
O
N
4
Figure 2.
SOAR Base Design
4 Headers
t 1 1 1 1 r
-i 1 r
n—r
-•— No Additive
¦— 500 ppm DBA
-a— 1000 ppm DBA
J I I I I I l_
I I ' I L_
J I I I I I
100 120 140
L/G, gal/1000 acf
160
180
200
-------
V
V
Figure 3.
LEVELIZED CONTROL COSTS vs. S02 REMOVAL
AND DBA CONCENTRATIONS
LEVELIZED CURRENT $/ton
500
480
>
460
440
420
75%
80%
85% 90%
S02 REMOVAL
95%
100%
0 PPM DBA
500 PPM DBA
-%r 1000 PPM DBA
UWTtD mIMiW ft COMiniCTQM
-------
Figure 4.
SO Removal vs. L/G,
2 '
Varying Gas Velocity
100
85
•— Base Case, 10 ft/sec (58.8 ft diameter)
-b— 8 ft/sec (65.7 ft diameter)
A— 12 ft/sec (53.7 ft diameter)
70 80 90 100 110 120 130 140 150
L/G, gal/1000 acf
-------
Figure 5.
LEVELIZED CONTROL COSTS vs. S02 REMOVAL
AND 1000 PPM DBA CONCENTRATION
LEVELIZED CURRENT */ton
6 FEET BETWEEN HEADERS
440
420
>
&
400
380
360
I
I
75%
80%
85% 90%
S02 REMOVAL
95%
100%
1000 PPM DBA
~ 8 FOOT SPACING
¥ 6 FOOT SPACING
X 10 FOOT SPACING
UMTED UWHiMI Ik CQNtlMUCTOM
-------
Figure 6.
LEVELIZED CONTROL COSTS vs. S02 REMOVAL
AND 1000 PPM DBA CONCENTRATION
LEVELIZED CURRENT */ton
5 FEET BETWEEN HEADERS
420
410
400
390
380
370
360
X
I
75%
80%
85% 90%
S02REMOVAL
95%
100%
1000 PPM DBA
X 10 FOOT SPACING
* 6 FOOT SPACING
0 16 FOOT SPACING
~ 8 FOOT SPACING
UMTCP EMIKEM * COMTBUCTTM*
-------
Figure 7.
LEVELIZED CONTROL COSTS vs. S02 REMOVAL
AND VARYING BOILER LOADS
LEVELIZED CURRENT $/ton
530 -
480 -
430 -
380
70%
75%
80% 85%
S02REMOVAL
90%
95%
100%
4 HEADERS/ALL WORK
~ 100% LOAD
mmm 4 HEADERS/HIGH OFF
X 80% LOAD
4 HEADERS/LOW OFF
0 60% LOAD
UWTED OMUPCEM I CONSTRUCTORS
-------
\
\
4A-76
-------
Acid Rain FGD System Retrofits
4A-77
Preceding page blank
-------
4^-78
-------
Legislation Allows Creative Economic Optimization
By Antonio J. DoVale, Jr., George D. Krause and James L. Murphy
WHEELABRATOR AIR POLLUTION CONTROL
INTRODUCTION
This paper provides an overview of economic optimizations to be
applied to comply with the Clean Air Act Amendments of 1990. It
addresses the options open to utilities based on the annual tonnage
emission concept, as opposed to the previous legislation for new units
requiring sulfur dioxide emissions on a Lb/MMBTU basis.
LEGISLATION SERVES AS THE BASIS FOR COST SAVING
On November 15, 1990, President Bush formally signed the Clean Air
Act Amendments of 1990. Title IV of these amendments addresses
Acid Rain Formations. The purpose "is to reduce the adverse effects of
acid deposition through reductions in annual emissions of sulfur
dioxide of ten million tons from 1980 emission levels", and of nitrogen
oxides emissions of two million tons relative to their 1980 levels.
Starting January 1, 2000, there will be an 8.9 Million Ton
S02 emission limit which will create an absolute cap for all
electric utility generating stations in the U.S.. Any new electric
generating capacity added to the national grid in the future must offset
its S02 emissions with equal reductions in S02 from existing
installations or unit retirements after the year 2000. Regulation is
centered around an emission allowance system for "Affected Units", as
opposed to the previous emission rates. "Affected Units" SO2 emission
allowances are tied to their individual unit 1985 - 1987 average coal
Preceding page blank
4A-79
k.
-------
firing rate, "Baseline". Increases in the coal firing rate above the
"Baseline", will require the unit to achieve more stringent removal
rates to meet a fixed S02 emission allowance tonnage. All Electric
Generating Utilities, Independent Power Producers and Cogenerators
must have emission allowances to operate.
The above new Clean Air Act serves as the basis for cost savings concepts
in the design of Flue Gas Desulfurization (FGD) systems for retrofit
applications. This legislation will affect as many as 58 electric utilities
across the United States, primarily in the Midwest and East (see Figure I).
Half of the total reduction must be achieved by 1997, if scrubbers are
utilized. As these utilities are gearing up for compliance, there are a
number of considerations which can significantly reduce the capital costs
for FGD retrofits in order to reach emission reduction requirements.
"Affected
Unit"
Locations
21 "Affected" States
58 "Affected" Utilities
261 "Affected Units"
FIGURE I - AFFECTED SOURCES PER LEGISLATION
While, at a glance, the new legislation appears to be severe, there is a
greater degree of emissions flexibility not evident in the Clean Air
Regulations of 1979. In fact, the Acid Rain Bill opens up new concepts in
dealing with the abatement which will make compliance less costly than
originally anticipated.
4A-80
A
A
-------
With respect to the Retrofit market, it is important to recognize that this
legislation requires a fixed S02 emission tonnage reduction per year,
versus the historic straight emission rate requirement (0.60 - 1.20 Lb
S02/MMBTU) to be met on an hourly, daily or weekly operation basis.
Therefore, given the requirements of the legislation, a utility's FGD system
has a greater degree of flexibility in achieving the outlet tonnage and does
not require the high levels of system availability as under the 1979
legislation. As a result, many of the wet scrubbing system's redundancies
can be eliminated, without jeopardizing the achievement of the legislation's
goals.
The above, combined with the fact that today there is a more thorough
understanding of the chemical, mechanical and instrumentation
technologies of wet scrubbing systems, provides various opportunities for
utilities to reduce initial capital costs, as well as operating costs for
compliance. Various options for capital reduction presented herein are
supported by previous studies conducted by the Electric Power Research
Institute (EPRI).
Indeed, individual utilities have to examine the various design options
depending on their own individual situations. For example, if a utility
designs to overscrub, how much redundancy needs to be built into the
system? Or if a utility plans to overscrub, but eliminates redundancy and
faces possible downtime, the utility may forfeit credits for overscrubbing
but could still meet the annual tonnage requirements.
ELIMINATING REDUNDANCY FOR COST SAVINGS
If we examine the opportunities for eliminating redundancies, it becomes
apparent that a utility can economically benefit while still meeting the
requirements of the legislation. Three specific areas of cost reduction for
the FGD system involve the SO2 removal (absorber) area, flue gas handling
and reagent preparation areas.
The SO2 removal area offers the greatest opportunity for reducing cost by
eliminating redundancy. Because the SO2 removal area involves many
components, including the absorber modules, reaction tanks, recycle
pumps, agitators, isolation dampers, and associated piping/equipment, it is
the largest capital cost of a new FGD (flue gas desulfurization) system.
k.
4A-81
-------
Consequently, the overall retrofit costs and site impacts will be reduced if
the capital cost and plot plan of the absorption process equipment is
reduced.
Since the largest single cost item in the removal area is the absorber
modules, the single most significant means of reducing costs is to reduce
the number of absorber modules. This can be done by eliminating the use
of a spare absorber, which then enhances the benefit of using larger
capacity absorber modules. Depending on the combination of fewer and
larger absorbers, this could mean a capital cost savings ranging from 12%
to 20%. An additional savings could be realized by the elimination of the
spare slurry recirculation pump with related piping, instrumentation and
electrical per absorber, which could amount to an additional 4% cost
reduction.
Table 1 presents a number of current Phase I FGD Retrofits in the planning
or design stage required by the Clean Air Act Amendment. It can be seen
TABLE 1
Preliminary Phase 1 Clean Air Act Retrofit Absorber Design
Utility
Units
MW
Absorbers
Capacity
Spare Stage
AEP
Gavin 1 & 2
1300 Ea.
6
20%
Yes
APS
Harrison 1-3
680 Ea
1 Ea.
100%
Yes
PSI
Gibson 4
650
2
67%
Yes
Pennelec
Connemaugh 1 & 2
900 Ea
2Ea.
50%
Yes
Niagara Mohawk
Huntley 1 & 2
218 Ea
1
100%
Yes
Kentucky Utilities
Ghent 1
550
3
50%
No
Virginia Power
Mt. Storm 1 - 3
550 Ea
2
50%
No
Owensboro Muni.
Smith 1 & 2
150/260
2
67%
No
Indianapolis P&L
Petersburg 1 & 2
250/450
1/2
100%/50%
Yes
Illinois Power
Baldwin 1 & 2
600 Ea
1 or 3
100%/33%
Yes/No
Centerior Energy
Eastlake 4 & 5
200/680
1/2
100%/50%
Yes
SIGECO
Culley 2 & 3
100/260
1
100%
Yes
Georgia Power
Wansley 1 & 2
900 Ea
2/1
50%/100%
N/A
4A-82
A
-------
that very few spare absorbers are being planned, but that oversizing
absorbers to provide a spare at higher reduced loads is more common for
the Retrofit market. In some cases the addition of the spare absorber is
partially offset by the elimination of the spare slurry recirculation stage.
Absorber module sizes as large as 680 MW have been utilized this year
domestically in lime service, but absorber modules in the 375 to 450 MW
range are more typical for limestone service in the domestic Retrofit
market. As domestic utility experience with larger absorber modules
increases, the use of larger limestone chemistry absorbers is more likely.
The majority of savings directly results from eliminating the spare
absorber, thereby requiring a smaller plot plan. In addition, having fewer
absorber modules reduces the branching and amount of inlet and outlet
ductwork which can improve flow distribution to and from the absorber
modules. Historically, the outlet ductwork, isolation slide gates and
dampers have been a source of problems with respect to materials of
construction selection and design. Reducing the ductwork area and quantity
of isolation devices minimizes the size of the potential problems and allows
for reducing the cost impact of increasing the quality of material selections
in this and other areas.
Early scrubber systems were generally constructed of carbon steel and
relied upon a variety of non-metallic linings throughout the system for
corrosion protection. While most were chemically suitable for the
corrosive service and performed well under laboratory and test
conditions, it soon became apparent that many had long term
operational limitations, sometimes requiring forced outages or
extended downtime for repair or replacement. In addition, non-
metallic lining installation often required a high profile sand blasted
substrate, controlled environmental conditions (Temperature and
Humidity) for application and special curing requirements.
Overall service life was uncertain, but it was typically expected that
replacement would be necessary two or three times over the life of the
plant with annual maintenance touch-up. Some linings exhibited poor
abrasion resistance or were subject to mechanical damage during
routine maintenance. Others were sensitive to temperature excursions
and few, if any, proved completely successful in areas where hot gas
mixed with treated gas, such as at the absorber inlet and the mixing
zone for bypass gas in the outlet ductwork. Few non-metallic options
k.
4A-83
-------
were available for a "solution" in these more aggressive corrosion areas
and stainless steel or, more frequently, nickel based alloys emerged as
the preferred material.
This alloy experience has been carried directly into the Retrofit market
and in many cases extended to include all parts of the system handling
saturated flue gas. The current reduced cost of alloy materials has
made their use much more attractive, especially when considering the
increased reliability they provide and the reduced temperature
excursion concerns, against the first cost as well as the cost of
maintaining and replacing the previous non-metallic lining materials
over the life of the plant.
A single alloy selection does not necessarily have to be used
throughout the FGD system. Consideration must be given to pH and
chloride exposure, along with the type of construction - wallpaper or
clad versus solid alloy.
In general, the higher quality nickel based alloys, C22/C276, are
currently preferred for the areas having the greatest corrosion
potential, the absorber inlet transition section and the outlet ductwork.
Either wallpaper or clad carbon steel is used for the outlet ductwork,
while solid alloy is preferred for the absorber inlet section. Outlet
isolation dampers typically are of the same alloy as the ductwork.
Unless chlorides are unusually high, alloy selection for the absorber
vessel and recycle tank is generally several steps down the alloy
ladder. These areas are constantly flushed, have minimal build-up and
moderate pH, permitting the use of higher grades of stainless steel
such as 317LMN or duplex stainless steel such as alloy 255. In some
cases more than one alloy might be used in a single scrubber, matching
corrosion resistance with conditions in different areas or zones (See
Figure II). Extremely high chlorides generally demand the use of very
high grade alloys, usually C22/C276. Because of cost, however,
wallpapering rather than solid alloy construction can be used. If there
are concerns about the relatively thin (1/16 inch) wallpaper in areas
subject to abrasion from slurry sprays or at the entrance to pump
suction nozzles, an increased thickness can be used in these areas,
allowing a wear surface while still maintaining the structural integrity
of the carbon steel vessel shell.
4A-84
A
A
-------
To Stack
Materials
of
Construction
¦
Severe
n
Severe/Moderate
M
Moderate
Hi
Moderat«/Mlld
Flu*
Gu Inlet
FIGURE E - ABSORBER CORROSION ZONES
Where solid alloy construction is used for absorbers, economic
considerations require that the cost be compared against wallpapered
designs. Stainless steels, either austenitic or duplex, are generally less
costly solid than wallpaper. If the lower or intermediate grades of
nickel based alloys are considered for solid construction, wallpapering
with a higher grade such as C22/C276 may prove to be a better choice.
Welding of stainless steel and high grade alloy materials requires
particular attention and is often a concern, especially for wallpaper
installations. Well written specifications, fully qualified welders and a
strong, enforced inspection and testing program can make this less of a
quality control problem than the surface preparation and installation
of non-metallic lining materials in use for many years.
The S02 removal equipment is one area of the FGD System where
regulatory flexibility could save millions of dollars while still accomplishing
the SO2 emissions reductions desired. It is industry practice to remain in
compliance with the 1979 legislation for new facilities, which require a
utility to install a spare absorber in order to have a system bypass. This
requirement is necessary so that the FGD system could remain in emission
L.
4A-85
-------
rate compliance with one absorber module off-line to maintain short
duration emission averages.
This above should no longer be necessary for two major reasons. One, new
Acid Rain Legislation can allow for catch-up capability. The regulation can
permit an absorber or entire FGD system to be bypassed for short periods,
as long as the tonnage emission requirements are met on an annual basis.
Therefore, a utility could overscrub S02 during normal FGD operations with
a no spare system, in order to offset periods when an absorber is out of
service for routine repair or maintenance. Two, FGD technology has
progressed such that a utility can have high availability without a spare
module. Indeed, there have been systems operating successfully for more
than a decade without spare absorbers for their FGD systems, while
maintaining availabilities in excess of 95%.
Reducing the number of absorber modules requires maintaining high levels
of absorber reliability and availability. This is achievable through
improved process chemistry knowledge and control. For example,
limestone chemical additives such as alkalinity enhancers and scale
inhibitors can be utilized to help on-line performance (increase S02
removal) and improve the maintainability of the absorbers by reducing
scale formation. Another chemistry option is the use of in-situ forced
oxidation to produce Disposal or Commercial Grade Gypsum, which also will
help minimize scale formation by operation at a beneficial chemistry
regime.
It is estimated that integrating organic acid additives into the design can
reduce the capital cost of the S02 removal equipment by 5%, due to
reduction in L/G (smaller recycle pumps) and the resulting smaller recycle
tank volumes. While there is an increase in operating cost for the organic
acids (Adipic and DBA), the overall operating costs can be reduced due to
lower power consumption for the reduced L/G. The total benefits of these
alkalinity enhancers can only be achieved if included during the initial
design of the FGD system. Historically, they have been added after
commercial operation to offset a design problem or fuel change. The
Kentucky Utilities Ghent Station retrofit includes the concept of DBA
addition to achieve 95% SO2 removal, while the base design L/G is for a
90% removal forced oxidized limestone design. This will assist Kentucky
Utilities in achieving overscrub capacity, if required, at a later date without
adding to the capital and power consumption at this time. Illinois Power is
4A-86
A
A
-------
considering a similar concept in their specification for the Baldwin Power
Station.
In addition, chemical additives have been developed, which can be utilized
in lime and limestone FGD systems that reduce the scaling characteristics of
the process. Scale inhibitors, such as emulsified sulfur (A second
generation to Sodium Thiosulfate addition) for in-situ thiosulfate formation
does reduce the tendency for scaling, but does not boost (increase
alkalinity) the SO2 removal performance. This additive can provide the
increased confidence needed to allow a utility to specify a no spare
absorber FGD system, by assuring a minimizing of offline absorber scale
cleanup. This concept has been utilized by Public Service of Indiana for the
Gibson Station, Unit 4 retrofit which is being supplied with two 67%
capacity (430 MW) absorber modules. Again, experience gained from
previous industry problems has been utilized to optimize the economics of
the Retrofit market.
The use of Flue Gas Reheat Systems (Indirect Air, Bypass, Inline or
Regenerative) has been shown to be of minimum benefit with respect to
elimination of moisture plumes and reduction of ground level SO2
concentrations. These Reheat systems can also have high operating costs,
with the exception of Regenerative and Bypass Reheat. A design that does
not utilize Reheat can save up to 2% of the FGD system capital cost and
open up additional materials of construction options for the outlet
ductwork and chimney liner, assuming saturation level temperature will be
present in the outlet ductwork for the normal operation mode. This can
have beneficial economic benefits considering a number of the Retrofits
will maintain their existing chimneys for bypass and build new chimneys
for the saturated gas operation. Although, to date retrofit FGD systems in
general have not taken advantage of this concept, which could allow for
FRP chimney liners, as well as outlet ducts.
In conjunction with the elimination of the spare absorber, the concept of
maximizing scrubber removal efficiency must be considered. The
advantage will be to develop catch-up capability in the FGD system. If the
annual average S02 removal requirement for the FGD system is, for
example, 87% to meet the S02 tonnage emission, a 95% absorber removal
design would allow for a net system availability of 91.6%. This would be
quite comfortable a margin for a no spare absorber FGD system based on
historical data. Availabilities at the 95% level, which is highly probable,
k.
4A-87
-------
would yield credits for the utility to apply to other units in their system or
for future years operation of this and other units.
In order to assure 95% S02 removal with limestone chemistry and high
inlet S02 concentrations, an efficient mass transfer contacting absorber
such as a Dual Flow Tray absorber should be utilized. The tray absorber
provides a high mass transfer froth level above the Dual Flow Tray
(Essentially a perforated plate designed for slurry and flue gas mass
transfer), which can not be achieved in a conventional Open Spray Chamber
(see Figures III and IV). An open spray chamber absorber achieves
removal by contacting the flue gas with slurry droplets sprayed from a
series of spaced spray header stages. Since a Dual Flow Tray absorber does
not require the number of spray levels utilized in an open spray chamber
absorber, its overall height can be lower and its L/G requirements will
be less. In addition, larger capacity slurry recirculation pumps can be
utilized, minimizing the total number of installed pumps. This can result in
FGD system capital cost savings of 1% to 4%. An additional benefit of the
tray absorber is the ability to significantly increase the SO2 removal
Dual
Flow
Two Stag* .
Mist Eliminator'
InM Wall
Slurry
Spray
Haadara
Tray
Tower
Presaturator
Haadar "
Agitators
FIGURE m - DUAL FLOW TRAY ABSORBER
4A-88
A
A
-------
performance of the absorber when using chemical additives (Such as adipic
acid or DBA). This will be of advantage during periods of unexpected high
inlet S02 concentration (due to fuel variations or changes) to achieve extra
catch-up capability for S02 tonnage emission requirements. An option for
possible consideration is a hybrid absorber design which can be utilized as
an open spray chamber or quickly modified (One day) to a Dual Flow Tray
Open
Spray
Tower
Two Stag*
Mist Eliminator"
Inlet Wall
Wash H«ad«rN
Presaturator
Header
V
>1
. Mist Eliminator
Wash Headers
Slurry
Spray
Headers
8-0*—
Agitators
FIGURE IV - OPEN SPRAY CHAMBER ABSORBER
absorber. This will allow a utility more flexibility in coal selection, without
having the burden of higher capital for a high sulfur open spray chamber
design.
4A-89
-------
MINIMIZING THE REAGENT PREPARATION SYSTEM COST
The flexibility of the proposed Acid Rain Legislation also impacts the need
for the traditional 100% spare reagent process capacity preparation area.
Studies indicate that the overall cost of a retrofit FGD system can be
reduced by decreasing the level of redundancy of the reagent preparation
system. The most significant method for reducing the capital cost is to
eliminate the spare ball mill system, which is typically the most expensive
item in a wet lime/limestone reagent preparation system. A secondary
cost reduction is the reduction in limestone slurry storage tank capacity
from 24 hours storage at maximum S02 removal to 16 hours capacity.
These options will also decreases a utility's plot requirements for this area.
The best scenario for this system is the use of two 60% or 75% ball mill
systems based on the maximum design sulfur condition at maximum load.
In the event one ball mill system is down for maintenance, the remaining
ball mill system would be able to grind the majority of the limestone
required for normal operation.
In addition to the plot plan savings, a utility can expect a 1% reduction in
the overall FGD system capital cost. The annual S02 tonnage emission
concept allows the option for flue gas bypass and reduced SO2 removal
performance, so that a power plant could continue to operate with reduced
reagent preparation system capacity even during periods of maximum inle
S02 concentrations without jeopardizing the annual tonnage requirement.
SAMPLE CASE: TECHNOLOGY MEANS TO REDUCE CAPITAL COSTS
Consider the following sample case, if a combination of the aforementioned
construction and chemical options are incorporated into a retrofit FGD
system. The sample FGD system is a standard 600 MW unit utilizing
limestone chemistry, initially costing $85 million, based on the 1979 Clean
Air Regulations' compliance with spare absorber,reheat and bypass. As can
be seen from Table 2, a total capital reduction of $ 18,700,000 resulted
from the utilization of some of the techniques discussed earlier in this
paper. The cost optimized FGD system will still achieve availability levels of
95% and comply with proposed Acid Rain legislation.
4A-90
A
A
-------
TABLE 2
SAMPLE COST REDUCTION SUMMARY
600 MW -- 3.5% Sulfur Coal
BASE COST (2+1 Absorbers)
REDUCTIONS
Absorbers w/o spare
(2+0 Absorbers)
Smaller Reagent Prep Capacity
Eliminate Reheat System
Tray Tower vs Open Tower
TOTAL REDUCTIONS
ADJUSTED FGD SYSTEM COST
PERCENTAGE OF BASE COST
$ 85,000,000
($ 15,000,000)
($ 700,000)
($ 1,250,000)
($ 1,750,000)
($ 18,700,000)
$ 66,300,000
78.0%
4A-91
-------
ADDITIONAL OPTIONS AND CONSIDERATIONS
There are other Acid Rain Retrofit options for utilities to consider in order
to meet compliance requirements. Since their inception, FGD designs can be
much more specialized to specific applications due to a better
understanding of the overall construction and chemical technologies. Units
today can now be designed to better meet specific needs than in the 1970s
and early 1980s.
Some larger utilities with both affected and unaffected units may opt to
meet annual emission requirements by eliminating many unnecessary
redundancies in retrofitted FGD systems and use the excess unaffected
generating capacity in the system to supply power during FGD system
down-time, if catch-up S02 margin is not available. Other "Affected
Utilities" may opt to spend more capital up front for a compliance strategy
to overscrub, maintaining fewer, but larger, absorbers and FGD systems.
These systems benefit from the 100% bypass capability, allowing an
absorber to be shut down for maintenance without paralyzing the whole
utility system.
For a utility that can not afford the capital investment to purchase the
required FGD systems, own and operate may be an economic measure to
consider. While the supplier retains the ownership and operation of the
system, this option permits a utility to possibly build a more reliable and
available FGD system without the up front capital outlay.
Modularization is also another consideration for utilities, particularly if the
plant has direct water access. This includes approximately 25-50% of the
"Affected Utilities". In this situation, the absorbers and ductwork can be
built in large sections off-site in a controlled facility environment and
shipped to the site to be erected as large modules. This scenario allows
reduction in field erection costs and schedule, provided water access is
convenient and the walk in distance is not to long.
If a utility does have direct river access, it may benefit in addition to
modularization, due to the possibilities in determining which reagent to
use, since waterway transport can significantly affect reagent economics.
While lime and thiosorbic lime are expensive from a chemistry standpoint,
these particular reagents allow for the least expensive capital outlay of the
4A-92
A
A
-------
FGD system. Utilities can save on the initial capital costs because of lime
and thiosorbic lime smaller absorber requirements via reduced L/G and
smaller recycle tanks. In addition, money can be saved on operating power
costs. The key parameter is the ability to get low long term lime costs.
The ability to operate more economically is enhanced by innovations in
computerized control systems and related instrumentation (i.e. PH probes,
density meters). Also, advancements in laboratory analytical methods
have allowed for optimization in operation and improved on-line time. In
addition, Distributive Control Systems (DCS) are less expensive than the
traditional analog systems used in the 1970s and early 1980s. These
systems can control and monitor the scrubbing operation more closely,
which helps to control operating costs and reagent usage.
Finally, there has been much discussion on the issue of waste and its
disposal. While in Germany wastes must be made into usable product,
particularly Commercial Grade Gypsum, this requires additional monies in
order to manufacture the finished product. On the other hand, the United
States, save for Florida, has made no official statement concerning the
disposable materials.
Disposal
Grade
Gypsum
J Rotary
¦Vacuum
Flltar
To Trucks
FIGURE V - DISPOSAL GRADE GYPSUM PROCESS
4A-93
-------
Due to the lack of legislation, perhaps the most economical approach is the
manufacture of Disposal Grade Gypsum (Figure V), because this approach
does not require any additional equipment or chemical additives and
produces a product that can be used as a landfill. Ultimately, this approach
offers the least expensive first cost for back-end disposal in comparison to
the historical stabilized sludge process. The upgrade to Commercial Grade
Gypsum (Figure VI) is easily accomplished by the addition of higher
oxidation air stoichiometry and changes to the secondary dewatering
system. A chloride/impurities purge may be required depending on coal
composition, which can create additional environmental permitting
requirements.
Commercial
Grade
Gypsum
Bait
Vacuum.
Flltar
Wash
Watar
Stack Out
Convayor
To Truck*
FIGURE VI - COMMERCIAL GRADE GYPSUM PROCESS
The increased complexity of the stabilized sludge process over the Disposal
Grade Gypsum process can be seen in Figures V and VII. In addition, the
stabilized sludge process requires the adding of lime to force the fixation
reaction, which increases the operating cost significantly of the limestone
FGD system waste processing.
4A-94
A
A
-------
Ou!
Sludge (2?
Stabilization
System
To Trucks
FIGURE vn - SLUDGE STABILIZATION PROCESS
Indeed, today's Retrofit wet scrubbing systems are more flexible, can be
tailored to meet unique operating conditions, and are more economical to
build and operate. Since the past legislation of 1979, there is a greater
degree of operating experience, improvements in chemical, construction
and instrumentation technologies. This solid data base, coupled with the
flexibility of the new legislation, offers utilities a variety of options to
reduce initial capital and operating costs in order to build a Retrofit FGD
system designed for optimum S02 removal performance.
4A-95
-------
Intentionally Blank Page
4 A- 96
A
-------
Guidelines for FGD Materials Selection and Corrosion
Protection
Preceding page blank
-------
Intentionally Blank Page
4A-9 8
-------
Harvey S. Rosenberg
Gerald 0. Davis
Barry Hind in
Battelle
505 King Avenue
Columbus, Ohio 43201
Paul T. Radcliffe
Barry C. Syrett
Electric Power Research Institute
3412 Hi 11 view Avenue
Palo Alto, California 94304
ABSTRACT
Under funding from the Electric Power Research Institute (EPRI), Battelle is pre-
paring a manual for the selection, specification, installation, and maintenance of
materials resistant to corrosion and erosion in flue gas desulfurization (FGD)
systems. The manual presents a decision logic for selecting appropriate materials
for the various FGD system components. The decision logic is based upon exposure
environment, previous field performance, and costs of construction materials for
these components. Information on items pertaining to the decision logic is
included in the manual. Techniques for avoiding materials problems are also pre-
sented. Retrofit considerations and new and innovative materials applications are
addressed. Also included with the manual is a diskette that contains a life cycle
cost analysis model, FGD materials data bases, and a bibliographic data base with
over 600 citations. The manual provides instructions for using the model and data
bases on a personal computer.
Preceding page blank
4A-99
-------
INTRODUCTION
EPRI is currently funding a research project at Battelle to develop guidelines for
the selection, specification, installation, and maintenance of materials resistant
to corrosion and erosion in flue gas desulfurization (FGD) systems. These guide-
lines are to be applicable to new FGD systems installed on new units or retrofitted
on existing units, and to the upgrading of existing FGD systems. The product of
this project will include a manual and personal computer (PC) software that can be
used by utility personnel in screening alternative materials of construction for
FGD system components and selecting the most cost effective methods for protecting
these materials from corrosion and erosion. The manual and software assess the
technical feasibility and potential costs of alternative materials, and present
those practices that utilities have found successful in avoiding or minimizing cor-
rosion and erosion. Current U.S. utility design, operating, and maintenance prac-
tices are being incorporated, as well as new and innovative applications of materi-
als advanced by European and Japanese utilities.
COLLECTION OF INFORMATION ON FGD MATERIALS
The following sources of information are being used to prepare the FGD materials
guidelines:
• Published literature
• EPRI research reports (both draft and published)
• FGD system suppliers
• Architect and engineering (A/E) firms
• Materials suppliers
• Component suppliers
• European and Japanese companies with experience in FGD materials.
4A-100
-------
Pertinent literature and EPRI reports were compiled into a computerized bibliog-
raphy. Details on the bibliography are provided later in this paper. U.S. utili-
ties with operational FGD systems were surveyed to obtain information on materials
performance, maintenance practices, and installed costs of FGD materials. This
information is being combined with information that Battelle obtained from previous
surveys for EPRI (1,2), as well as background information from Battelle's data base
on FGD systems, to produce a computerized data base on materials experience.
Details on the computerized data base are also provided later in this paper.
FGD system suppliers, A/E firms, materials suppliers and component suppliers were
also surveyed to obtain additional information. System suppliers and A/E firms
were requested to provide recommendations on materials selection and maintenance
practices for the various components, as well as estimates of the installed costs.
Materials suppliers were requested to provide manufacturing and installation proce-
dures, installation times, and installed costs. Component suppliers were asked
about materials of construction and cost as a function of size. Both materials and
component suppliers were queried with regard to maintenance requirements, expected
service life, and environmental limitations of their products. European and
Japanese companies were requested to provide the same type of information as their
American counterparts.
Letters and questionnaires were sent to a total of 162 companies. These included
82 utilities, 18 FGD system suppliers, 14 A/E firms, 52 materials suppliers, 72
component suppliers, and 24 European and Japanese companies. A summary of the
number and percent of responses by category is shown in Table 1. The overall
response was 38 percent, which is excellent for a survey of this type. The
response ranged from 48 percent for utilities to 14 percent for A/E firms. It
should be pointed out that the quality of the responses in each category ranged
from excellent (most of the requested information was provided) to poor (a very
limited number of questions were answered).
TYPES OF MATERIALS
A variety of construction materials have been used in FGD systems on utility boil-
ers. However, they can all be classified into only three major categories:
• Metals
• Organics
• Inorganics.
4A-101
-------
Metals ranging from carbon steel to nickel-base and titanium alloys have been used
in most components of FGD systems. Unlined carbon steel is used where the flue gas
is above the acid dew point and/or where alkaline conditions are maintained, such
as most of the components of spray dryer FGD systems, inlet ducts of all FGD sys-
tems, lime/limestone storage silos, and fresh lime/limestone storage tanks and
piping. Under acidic conditions, the carbon steel needs to be protected by a
lining or by other corrosion control techniques. Alternatively, corrosion-
resistant alloys are used under these conditions. Stainless steels (e.g., Type
316L, Type 317L, and Alloy 904L) have been used in FGD system components such as
prescrubbers, absorbers, spray nozzles, reheaters, and dampers. In critical loca-
tions with more aggressive environments, such as reheaters, outlet dampers, outlet
ducts, and gas mixing zones, nickel-base alloys (625, G-3, C-276, and C-22) and
titanium (grades 2 and 7) have been used. Where failure of an alloy occurs, it is
usually by pitting or crevice corrosion. General corrosion, erosion-corrosion, or
stress corrosion cracking can also occur.
Because of the relatively high cost of nickel-base and titanium alloys, a signifi-
cant cost savings can be realized by using thin claddings of these alloys over
carbon steel plates rather than solid alloy plates for ductwork and dampers. Clad
plate can be manufactured by hot rolling, explosive bonding, or welding the clad-
ding and base plate together. However, these methods are generally applicable only
for new FGD systems because existing systems already have the substrate (e.g.,
carbon steel) in place. A recently developed method that is applicable to both new
and retrofit construction is called "wallpapering". This method involves plug
welding thin sheets, usually 1/16 in. thick, of an alloy such as C-276 or C-22 to
the carbon steel substrate. The thin sheets are overlapped and seal welded toge-
ther. A resistance brazing process can be used for thin sheets of titanium alloys
because these alloys cannot be successfully plug welded to a carbon steel sub-
strate. Another option for titanium alloys, which is more expensive than resis-
tance brazing, is to attach the cladding to the substrate with titanium nuts and
bolts, and to seal weld around each bolt hole. Alloy claddings, of course, are
susceptible to the same corrosion problems as solid alloys. In addition, if inade-
quate welding causes flaws and pinholes in the seal welds between the overlapped
alloy sheets, then liquid can get behind the sheets and cause corrosion of the sub-
strate. Also, during plug welding of the alloy sheets to the substrate, care must
be taken to avoid diluting the alloy with the substrate material (e.g., iron),
thereby, reducing the corrosion resistance of the cladding. One approach is to
cover the plug weld with a seal welded patch of the alloy cladding.
4A-102
A
A
-------
Organic linings have been used extensively in various FGD components including pre-
scrubbers, absorbers, tanks, outlet ducts, and stacks. These linings include epox-
ies, vinyl esters, polyesters, fluoroelastomers, and rubber. The use of organic
linings is attractive because it can provide the lowest initial cost material for
many of these components. The linings are generally applied in liquid form by
brushing, rolling, or spraying. To provide an effective barrier, the organic
linings used in FGD systems are often applied in several layers. This decreases
the probability of through-thickness pinholes and achieves a greater permeation
thickness than typical coatings. Moreover, the linings often contain a flake-type
filler that further decreases the permeability and provides reinforcement. In
components such as venturi prescrubbers where high abrasion resistance is required,
organic linings reinforced with glass cloth or inert matting have been used. These
linings may also contain special abrasion-resistant fillers such as alumina to
improve wear resistance. Failure of organic linings can occur by blistering,
debonding, and wear from abrasive slurries. Excursions to hot flue gas tempera-
tures can damage or destroy organic linings. Fluoroelastomer linings can withstand
excursions up to 500°F and are resistant to chemical attack and abrasion. However,
they are relatively permeable to moisture and are generally more expensive than
other types of organic linings. Although rubber is an organic lining, it is often
placed in a separate category because it is applied in sheet rather than in liquid
form. Rubber linings offer excellent resistance to abrasion, but close attention
to product specification and application are required to assure a quality lining
job.
Fluoroelastomer, fluoropolymer, and fluorinated elastoplastic materials, as well as
rubber, are used in expansion joints, typically with fabric reinforcement. The
cost of the fluorine-containing materials may be an order of magnitude higher than
that of rubber materials, but life expectancy, temperature tolerance, and operating
performance can easily offset this factor. In most expansion joints, it is usually
the flexible gas-sealing material that fails. The causes of failure include
thermal degradation induced by temperature excursions or thermal cycling, acid
attack, and mechanical damage.
Fiberglass-reinforced plastic (FRP) is used to fabricate mist eliminators, piping
and spray headers, and stack liners. Various plastic resins are used, but the most
common one is polyester. Failure of FRP can be caused by temperature excursions,
abrasion, and, depending upon the resin, chemical attack, e.g., acid hydrolysis of
polyester.
4A-103
-------
Nonmetallic inorganic materials are used in prescrubbers, spray nozzles, slurry
pumps, outlet ducts, and stacks. The materials used for these components include
prefired bricks and shapes, ceramic tiles, hydraulically bonded concretes and
mortars, and chemically bonded concretes and mortars, all of which are used as
lining materials where temperature resistance, chemical resistance, or abrasion
resistance are required. Acid-resistant bricks are commonly used as construction
materials for stack liners. Alumina bricks are more abrasion resistant than acid-
resistant bricks and also will withstand hot sulfuric acid conditions; therefore,
they have been used to line venturi throats. Prefired silicon carbide shapes have
high abrasion and chemical resistance and are used for spray nozzles, atomizer
wheel inserts in spray dryers, and pump components, as well as for lining venturi
throats. Design aspects that could cause failure include insufficient thermal
expansion allowances, vibration, and thermal or mechanical shock. A potential
failure cause for acid-resistant brick linings is mechanical stress from shrinkage
during service. Also, several free-standing brick stack liners have been found to
lean due to differential moisture expansion of the bricks (3).
Hydraulic setting, cement-bonded concretes (hydraulic concretes) are used as lin-
ings for prescrubbers, outlet ducts, and stacks. These concretes contain aluminate
to withstand temperatures of 500°F without strength degradation. However, in
sulfuric acid solutions (pH < 4), the cement is attacked due to dissolution of the
calcium aluminate. This dissolution of the bond results in loosening of the aggre-
gate and erosive wear of the lining.
Chemically-bonded mortars are commonly used to bond acid-resistant brick stack
liners. Chemically-bonded concrete mixes can also be used as linings. They gener-
ally contain siliceous aggregates, a sodium silicate, potassium silicate, or col-
loidal silica bond phase, and a silicofluoride, phosphate, or organic bond gelling
agent. Since chemically-bonded concretes can withstand hot, acidic environments,
they offer an abrasion-resistant alternative to organic linings. However, they
have a higher permeability than most organic linings, so that any condensed acid
will reach the substrates to which they are applied. In many cases, an organic
membrane is used under a chemically-bonded concrete lining to provide additional
protection to the carbon steel substrate. While this measure increases the time
for acidic condensate to reach the substrate, it can complicate repair procedures.
When the membrane fails due to permeation by acidic species, the concrete must be
removed to repair or replace the membrane.
4A-104
-------
Borosilicate glass blocks are used in outlet ducts and stacks. The blocks are usu-
ally bonded with urethane asphalt. The combination of these two materials results
in a lining that is resistant to permeation by aggressive condensate. Moreover,
the thermal insulating properties of the glass block and the flexibility of the
membrane material contribute to the good performance of the lining. However, the
borosilicate glass can easily be mechanically damaged, and failures can occur due
to abrasion, puncturing, and cracking. Also, scale that forms on the lining cannot
be easily removed. These problems can be minimized by applying a topcoat of chem-
ically bonded mortar.
The large accumulation of operating experience with FGD systems has shown that
degradation of materials of construction is a major problem. FGD system components
of particular concern are absorbers, outlet ducts, and stack liners. The perfor-
mance of materials is critically dependent upon the environment to which they are
exposed. Therefore, factors that affect the environment, such as operating and
maintenance practices, and coal and makeup water composition, also influence the
performance of materials.
EFFECT OF OPERATING AND MAINTENANCE
PRACTICES ON MATERIALS PERFORMANCE
FGD operating practices that have a significant effect on materials performance are
those that have the greatest effect on the environment. The aggressiveness of the
environment is dependent upon temperature, chemistry, and abrasiveness. As shown
in Table 2, the ASTM has identified three levels of severity, from 1 (mild) to 3
(severe), for classifying each environmental factor (4). For example, the mildest
condition encountered can be coded 1-1-1 and represents a temperature below 140°F,
a pH of 3 to 8 with low concentrations of chloride and fluoride, and low velocity
fluid flow with no direct impingement of particulates (e.g., a duct wall). By
contrast, the most severe condition can be coded 3-3-3 and represents a temperature
above 200°F, an acid concentration greater than 15 percent with high concentrations
of chloride and fluoride, and very high velocity fluid flow with impingement of
entrained particulates (e.g., a venturi throat). A code of zero for a given envi-
ronmental factor denotes that the factor does not exist or is not applicable for a
specific operating condition.
Service conditions can vary in different regions of a given component. The inlet
of a venturi prescrubber can be exposed to the most severe environmental condition
(3-3-3), while the sump is exposed to a mild temperature and an intermediate acid-
ity and abrasiveness (1-2-2). Except for the wet/dry zone (if there is no
4 A-105
-------
prescrubber) and areas of direct spray impingement, the service conditions in an
absorber are usually mild. However, in FGD systems where the water loop is
completely closed, the chloride concentration in the scrubbing liquor can build up
to severe levels.
Information from the materials data base was used to prepare a summary of absorber
materials found in both closed and open loop lime/limestone FGD systems. These
data, presented in Table 3, show that the prevalence of absorbers constructed from
alloys (mainly Types 316L, 317L, and 317LM stainless steels) is about the same for
both closed and open loop systems. Closed loop systems would be expected to have a
lower percentage of alloy absorbers because of the higher chloride concentration
than in open loop systems. Possible explanations for this apparent discrepancy are
that other variables are skewing the results, the corrosion mechanisms are not
fully understood, or the optimum absorber material is not always selected.
Outlet ducts and stack liners (flues) are subjected to a variety of service con-
ditions depending upon the configurations of the ductwork. The various configur-
ations and service conditions are outlined in Table 4. The mildest operating
conditions (1-1-1) are encountered when there is no reheat and no bypass. More
severe conditions (2-2-2) are encountered under partial bypass operation. The
mixing zone for wet scrubbed gas and hot bypassed gas is exposed to an aggressive
environment due to condensation of acidic species (S03, HC1, and HF) from the
bypassed gas and to temperature variations within a very small area. Full bypass
operating conditions (3-0-2) can exceed the temperature limits of most organic
linings and, thus, cause them to fail.
The materials data base was also used to prepare summaries of outlet duct and stack
liner materials for wet lime/limestone FGD systems categorized into the five types
of ductwork configurations shown in Table 4. The results for outlet ducts and
stack liners are presented in Tables 5 and 6, respectively. As one would expect,
Table 5 indicates that for mild operating conditions (ductwork configuration
Types I, III, and V) a glass flake/polyester lining is the most prevalent material,
while for more severe operating conditions (Type II configuration), borosilicate
glass block and high alloy linings predominate. Because of the various combina-
tions of reheat and bypass that are possible with a Type IV configuration, the
materials in use run the gamut from carbon steel to high alloy linings.
Table 6 shows that acid-resistant brick and mortar is the predominant stack liner
material for all ductwork configurations except Type V. The reason Type V is an
4A-106
-------
exception is not immediately apparent. It may be related to the relatively small
number of FGD systems in this category, or to the possibility that existing bypass
ducts are not used in these systems. It is interesting to note that six of the
eight acid-resistant brick and mortar flues in the Type III category lean toward
the outer concrete shell (3). In the other two cases, the bypass ducts are never
used. Mixing scrubbed gas and bypass gas in the flue results in differential
moisture expansion of the bricks on opposite sides, which causes a lean.
If necessary, the chloride and fluoride concentrations in the absorber scrubbing
liquor can be minimized by the use of separate quench loop, thus reducing the need
for highly alloyed materials downstream from the quench section. The abrasive
effects of fly ash on FGD components can be minimized by using a high efficiency
particulate collection device upstream from a wet FGD system. On the other hand,
the aluminum in fly ash can form complex ions with fluorine and, thus, mitigate the
corrosive effect of acid fluorides on certain alloys. Therefore, some entrainment
of fly ash can be beneficial in these cases. However, one must be careful not to
create an FGD chemistry problem while solving an FGD materials problem. Aluminum-
fluorine complexes can cause loss of reactivity of the scrubbing agent in wet
limestone FGD systems.
Maintenance practices can also have a significant effect on FGD materials perfor-
mance. For example, frequent cleaning to remove scale or deposits in the system
can help minimize crevice corrosion. Maintenance practices for absorbers, outlet
ducts, and stack liners were documented using information obtained from the utility
survey. Table 7 is a summary of the maintenance information supplied for absorb-
ers. In general, absorber maintenance consists of periodic inspection and minor
repairs. The data are insufficient to determine the effect of maintenance prac-
tices on materials performance.
Surprisingly, from the reported data, organic absorber linings have the lowest
annual maintenance costs and alloys have some of the highest. Also, there is a
large discrepancy in the annual maintenance costs for the two FGD systems (Plant F
and Plant H) that have Type 317LM stainless steel absorbers. Although the mainte-
nance practices are very similar, evidently one system requires much more repair
than the other. The difference could be due to operating practices. Information
from Battelle's data base on FGD systems was used to compare operating practices
that may have an effect on the performance of absorber materials. Although Plant F
has a wet limestone FGD system and Plant H has a double alkali system, the pH and
chloride concentrations of the scrubbing liquors are about the same. The
4A-107
-------
difference in the performance of the absorbers may be attributable to the effects
of an erosive slurry spray at a high liquid-to-gas (L/G) ratio at Plant F as
opposed to a scrubbing solution at a low L/G ratio at Plant H. However, other
possible explanations include the effects of operating practices that may have been
overlooked, or the lack of a complete understanding of the corrosion mechanisms.
In most cases, maintenance practices for outlet ducts and stack liners are similar
to that for absorbers, i.e., periodic inspection and minor repairs. For outlet
ducts, the period between inspections at specific sites ranges from 3 months for an
organic resin lining to 2 years for a high alloy lining. Contrary to maintenance
for absorber vessels, annual maintenance costs for outlet ducts at these sites
range from nil for a high alloy lining to $40,000 for a vinyl ester lining. In the
latter case, the lining requires extensive repairs annually. With regard to stack
liners, annual inspections are typical. Annual maintenance costs range from nil
for an acid-resistant brick and mortar flue to $100,000 for major repairs to an
epoxy lining. However, it must be recognized that these costs are site specific
and depend upon the exposure environment as well as the type of material.
EFFECT OF COAL AND MAKEUP WATER
COMPOSITIONS ON MATERIALS PERFORMANCE
Several species in solution have been found to affect the corrosion of alloys in
FGD systems. However, most of these species are considered trace elements in
scrubbing liquors. Notable exceptions are chlorine, fluorine, and aluminum, which
are all present in coal. Depending upon the source, the makeup water can contrib-
ute chlorides and fluorides to the FGD system. When the FGD water balance is
closed loop, dissolved ionic species, such as chlorides, become highly concen-
trated. Chloride ions cause both general and localized corrosion of alloys. The
corrosion rate for a given alloy is a function of both chloride concentration and
pH.
Fluorides tend to form in the solid phase because of the relative insolubility of
calcium fluoride. However, calcium fluoride precipitates as very small particles
that are difficult to separate from the liquid phase and, thus, they accumulate in
the recycle slurry. The solubility of calcium fluoride is enhanced by low pH and
the presence of other ions in solution, especially chlorides. Fluoride ions alone
cause general corrosion, but not localized corrosion. However, fluoride ions can
aggravate chloride-induced general and localized corrosion at low pH. Aluminum,
dissolved from fly ash, can react with fluorides to form the complex ions A1F2* and
4A-108
A
A
-------
A1F**. If there is sufficient aluminum, then free fluoride will not be available
to contribute to corrosion.
If cooling tower blowdown is used as makeup water for the FGD system, then any
additives used to control scaling in the cooling tower will enter the system.
Various additives are used, but phosphorous-containing compounds are predominant.
The effect of these additives on materials performance as well as process chemistry
have not been thoroughly investigated.
MATERIALS GUIDELINES MANUAL
The information collected on FGD materials is being used to prepare a guidelines
manual, as previously mentioned. The outline for the manual is shown in Table 8.
The key section of the manual is Chapter 2, a presentation of the decision logic
for selecting appropriate materials for the various components of an FGD system.
The decision path leads the user through a step-by-step process for selecting
suitable materials for a given application. The decision logic is illustrated in
Figure 1.
The steps in the decision path, as shown in Figure 1, are listed below along with
the appropriate section(s) of the manual for obtaining information related to the
decision:
Chapter 6
Appendix A
Chapter 4
Appendix A
Chapter 4, Appendix F, Appendix G
Step
1.
Step
2.
Step
3.
Step
4.
Step
5.
Step
6.
Step
7.
Step
8.
Step
9.
Step 10.
Appendix C, Appendix D
Appendix D
A key part of the manual is Appendix D, an economic model for computing the life-
cycle costs of alternative materials for FGD components. This computerized model
is generally based on the methods of economic analysis presented in the current
4A-109
-------
version of EPRI's Technical Assessment Guide (TAG) (5), adapted to the special
considerations for materials selection decisions.
The general approach was to use the time value of money concept to determine the
net present value of the various costs associated with each alternative material
over the life cycle of an FGD system. Typically, these costs occur at different
times over the life cycle of the system. The methodology permits the user to
calculate the net total present value of the costs for each prospective material
and compare it to similarly calculated costs of alternative materials. In addi-
tion, the program allows calculation of the levelized annual cost for use of each
material. Such comparisons are valid because all costs are adjusted to the same
point in time, i.e., the present or "time zero", in each separate analysis. The
effects of inflation and extended plant shutdowns due to maintenance or replacement
of a given material are included in the procedure. State and federal taxes and the
discount rate (used to adjust costs that occur in the future to their present
values) are also included. Default values of these financial components are
included using values from the TAG, but they are variables that can be changed by
the user. The economic analysis model is a template to be used with Lotus 1-2-3®
spreadsheet software for IBM-compatible personal computers. The template and
instructions for the user are included on a diskette that will be supplied along
with the guidelines manual.
Appendix F contains a data base on FGD materials experience that was developed in
conjunction with the preparation of the manual. The data base is divided into
three sub-data bases, namely background information on each specific utility FGD
installation, materials performance data for each component in the FGD system, and
available materials maintenance and cost data for each component. The data base
and instructions for the user are also included on the diskette that will be sup-
plied with the manual. The data base can be used to generate tables summarizing
materials experience for each component in an FGD system. More importantly, the
data base can be used to uncover various correlations or relationships between
materials performance and FGD system design, operation, or maintenance.
Appendix G contains a bibliographic data base on FGD materials and includes over
600 citations. Each citation has seven fields that allow computerized sorting by
title, author(s), document source, publication year, language, abstract, and key
word(s). About 460 citations are in English, but 9 other languages are repre-
sented, of which German is the most prevalent with about 90 citations. About 80
citations do not have abstracts available and almost 200 did not provide separate
4A-110
-------
key-word lists. However, these citations can be accessed through the words in the
titles. The bibliographic data base and instructions for the user are included on
the diskette that will be supplied with the manual.
ACKNOWLEDGMENTS
The authors want to acknowledge the numerous people who supplied information for
the materials guidelines manual.
REFERENCES
1. H. S. Rosenberg et al. Construction Materials for Wet Scrubbers. CS-1736.
Palo Alto, California: Electric Power Research Institute, March 1981.
2. H. S. Rosenberg et al. Construction Materials for Wet Scrubbers: Update.
CS-3350. Palo Alto, California: Electric Power Research Institute, July
1984.
3. H. S. Rosenberg et al. Leaning Brick Stack Liners. GS-6520. Palo Alto,
California: Electric Power Research Institute, September 1989.
4. Manual of Protective Coatings for Flue Gas Desulfurization Systems. STP 837.
Philadelphia, Pennsylvania: American Society for Testing and Materials, March
1984, p. 7.
5. TAG Technical Assessment Guide. P-6587-L. Palo Alto, California: Electric
Power Research Institute, September 1989.
k
k
4A-111
-------
New
Wet
Component A
\
Retrofit
Component B
Wet/Dry
Type of FGD
Construction
Type of
Environment
Type of
FGD System
TVpe of
Component
7
/
Existing
Component C
Type
Temperature
Data
Available
Type of
Material
r Type of >
Information
on Materials
Performance
Type
Type
Type
Type of
Environment
Chemistry
Screening Tests
Necessary or
Desirable
Abrasive ness
4 iype n-
>
N>
Inorganics
Type of
Material
fK Features
No
Undesirable
Materials
Features
Undesirable
Materials
Features
Type of
Costs
Unacceptable
Life Cycle
Costs
Acceptable
Life Cycle
Costs
Inorganic A
Inorganic B
Type of
Acceptable
Materials
Organic A
Organic B
ARoy A
ARoy B
Step 1. Specify type of FGD construction
Step 2. Specify type of FGD system
Step 3. Select component
Step 4. Define environment
Step 5. Determine If data on materials
performance are available
Step 6. Perform screening tests If necessary or desirable
Step 7. List suitable materials
Step 8. Examine materials features
Step 9. Compute life cycle costs for desirable materials
Step 10. Select specific material
• Decision point
Figure 1. Decision Logic for Selecting Appropriate FGD Materials
k.
-------
Table 1
STATISTICAL SUMMARY OF SURVEY ON FGD MATERIALS
Type of Company
Utility
EPRI Member
EPRI Nonmember
Subtotal
FGD System Suppliers
Wet systems
Spray dryer systems
Both types of systems
Subtotal
A/E Firms
Materials Suppliers
Alloys
Organic linings
Inorganic linings
Subtotal
Component Suppliers
Ball mills
Centrifuges
Dampers
Expansion joints
Mist eliminators
Nozzles
Piping
Pumps
Slakers
Vacuum filters
Valves
Subtotal
European and Japanese Companies
Grand Total
Number of
Companies
Surveyed
45
37
82
10
4
_4
18
14
21
24
_7
52
3
2
6
10
2
4
5
18
3
2
17
72
24
262
Number of
Companies
Responding
22
17
39
3
1
_1
5
11
10
_1
22
0
0
3
5
1
1
2
2
3
1
_6
24
7
99
Percent of
Companies
Responding
49
46
48
30
25
25
28
14
52
42
14
42
0
0
50
50
50
25
40
11
100
50
35
33
29
38
k.
4A-113
-------
Table 2
ASTM CLASSIFICATION SYSTEM FOR ENVIRONMENTAL FACTORS
Level of
Severity
1
Environmental Factor
Temperature
Ambient to
140°F
104° to
200°F
Above 200°F
Chemistry
pH 3 to 8 with low
concentrations of
fluoride and
chloride
pH 0.1 to 3 with up
to 1,000 ppm F and
10,000 ppm CI
Greater than 15%
acid with high con-
centrations of
fluoride and
chloride
Abrasiveness
Low velocity fluid
flow with no direct
impingement of partic-
ulates (e.g., duct
wall)
High velocity fluid
flow, spray impinge-
ment, or strong agita-
tion (e.g., absorber
spray zone)
Very high velocity
fluid flow with
impingement of
entrained particulates
(e.g., venturi throat)
Table 3
SUMMARY OF ABSORBER MATERIALS IN WET LIME/LIMESTONE FGD SYSTEMS
Type of
Material on
Interior
FGD Systems in Each Category
Closed Water Loop
Open Water Loop
Surface
Number
Percent
Number
Percent
A1 loy
27
39.1
13
36.1
Rubber
21
30.5
5
13.9
Organic resin
17
24.6
18
50.0
Gunite
4
5.8
0
0.0
Total
69
100.0
36
100.0
4A-114
-------
Table 4
SERVICE CONDITIONS IN OUTLET DUCTS AND FLUES
Service Conditions*
Ductwork Configuration
Type I - No reheat and
no bypass
Outlet Duct
Absorbers
Stack
Type II - Bypass to
outlet duct for reheat
and/or malfunctions
~ Absorbers
Type III - Bypass to
stack for reheat and/or
malfunctions
Bypass
Type IV - Operational
reheaters; bypass to
outlet duct for
malfunctions
Stack
Absorbers
Reheaters
Bypass
Type V - Operational
reheaters; bypass to
stack for malfunctions
Stack
Absorbers
Reheefers
Bypass
Stack
1-1-1
1-1-1
1-1-1+ and 2-2-2*
0-0-0+ and 3-0-2*
1-1-1
1-1-1
0-0-0
2-1-1
2-1-1+ and 2-2-2*
0-0-0+ and 3-0-2*
2-1-1
2-1-1
0-0-0
Flue
1-1-1
1-1-1
2-2-2
3-0-2
1-1-1
2-2-2
3-0-2
2-1-1
2-2-2
3-0-2
2-1-1
2-2-2
3-0-2
Comments
Exposure only
to scrubbed gas
No bypass
Partial bypass
Full bypass
No bypass
Partial bypass
Full bypass
No bypass
Partial bypass
Full bypass
No bypass
Partial bypass
Full bypass
*Three-digit codes denote severity of temperature, chemistry, and abrasiveness,
respectively. Refer to Table 2 for a description of the code. A zero denotes
that a condition does not exist or is not applicable.
+Upstream from bypass junction.
^Downstream from bypass junction.
4A-115
-------
Table 5
SUMMARY OF OUTLET DUCT MATERIALS IN WET LIME/LIMESTONE FGD SYSTEMS
Material on Interior Surface
Carbon Steel
Type 316 and 317 stainless steels*
Alloy 904L
Alloy 6
Alloy C-276 or C-22
Titanium
Glass flake/polyester
Vinyl ester
Gunite
Ceramic tile
Borosilicate glass blocks
Total
Number of FGD Systems in Each Category
Ductwork Configuration*
10
II
3
11
1
2
5
7
1
8
41
III
2
1
1
IV
10
1
1
30
*Refer to Table 4.
+Includes Types 316L, 316LM, 317L, 317LM, and 317LN stainless steel.
7
3
13
4A-116
A
A
-------
Table 6
SUMMARY OF STACK LINER MATERIALS IN WET LIME/LIMESTONE FGD SYSTEMS
Material on Interior Surface
Carbon Steel
Alloy 625
Alloy C-276
Glass flake/polyester
Vinyl ester
Epoxy
Fluoropolymer
FRP
Gunite
Acid-resistant brick and mortar
Total
*Refer to Table 4.
Number of FGD Systems in Each Category
Ductwork Configuration*
10
II
3
1
31
42
III
8
IV
3
4
16
29
4
1
2
2
3
14
k.
4A-117
-------
Table 7
MAINTENANCE INFORMATION
Plant
Designation
A
B
C
D
E
F
G
H
I
J
K
Current Materials
Carbon steel lined with natural rubber in
the spray zone and glass flake/polyester
above and below this zone
Carbon steel lined with glass flake/
polyester
Carbon steel lined with hydraulic-bonded
concrete
Type 316L stainless steel
Alloy 904L
Type 317LM stainless steel with unspecified
areas replaced with Alloy C-22
Carbon steel lined with glass flake/
polyester and inert flake/vinyl ester
Type 317LM stainless steel
Carbon steel lined with epoxy
Carbon steel lined with Ferralium® 255
Carbon steel lined with chlorobutyl rubber
FOR ABSORBERS
Maintenance Practices
Blisters and some eroded rubber are
repaired at each outage; minor
repairs to glass flake/polyester at
each outage
Overspray existing lining every 18
months
Some repair is necessary every
6 months; about 1,000 man-hours per
year are required
Inspection/cleaning
Clean, inspect, and repair every 6
months
Check every 3 months and patch as
necessary
Inspect annually and make weld
repairs as necessary
Spot repair during outages
Clean during outages
None
Annual
Maintenance
Cost, $
1,000
35,000
High, but
covered by
extended
warranty
218,000 for
3 modules
Less than
10,000
0
-------
r
r
Table 7 Continued
MAINTENANCE INFORMATION FOR ABSORBERS
4*.
>
VO
Plant
Designation
M
N
0
P
Current Materials
Carbon steel lined with chlorobutyl rubber
Alloy G
Carbon steel lined with glass flake/
polyester
Carbon steel lined with neoprene
Carbon steel lined with mat-reinforced
polyester in high abrasion areas and glass
flake/polyester elsewhere
Type 316L stainless steel
Maintenance Practices
Some patching of rubber around
recycle pump suction
Annual inspection
During annual outages, patch 1 to 2
sq ft areas where the lining has
failed
Patch holes and cover as necessary
using special rubber
Inspect for abrasions annually and
repair as needed; spark test after 4
to 5 years of service life
Inspect for abrasion and repair as
needed, annually or during power
outages
Annual
Maintenance
Cost, $
20,000
1,500
5,000
-------
Table 8
OUTLINE FOR MANUAL ON GUIDELINES FOR SELECTION AND MAINTENANCE OF MATERIALS
FOR FGD SYSTEMS
Chapter Number and Title
1. Introduction
2. Decision Logic for Selecting Appropriate Materials
3. Overview of FGD Materials
Material A*
Manufacturers and suppliers
Trade names
Manufacturing procedures
Installation procedures
Mechanical integrity and/or resistance to corrosion and erosion
Operation and maintenance requirements (general)
Expected outage time
Expected service life
4. Materials Performance
Component A+
Description of component
Description of environment
Materials alternatives
Operation and maintenance requirements (specific to component)
Required outage time
Summary of previous experience
Successful applications
Problems and solutions
Relative cost/impact of problems on plant operations
Failure causes
5. Techniques for Avoiding Materials Problems
Successful maintenance practices
Corrosion protection techniques (other than coatings)
Cathodic protection
Inhibitors additives
Corrosion monitoring
Modification of environment
Temperature change (e.g., reheat, insulation, water spray)
pH buffering (organic acid additives)
On-line cleaning/washing
Additives for scale control
6. Retrofit Considerations
7. New and Innovative Applications
Appendices
A. Environmental Factors in Materials Selection
Types of FGD systems
Wet
Additives (e.g., pH buffering, oxidation inhibition)
Forced oxidation
Wet/dry
Dry
Environmental considerations
Temperature
Phases
4A-120
A
A
-------
Table 8 Continued
OUTLINE FOR MANUAL ON GUIDELINES FOR SELECTION AND MAINTENANCE OF MATERIALS
FOR FGD SYSTEMS
Chemistry
Fluid dynamics
Ranking system for severity of environment
B. Materials Failure Mechanisms (general discussion)
C. Comparative Cost Information
A1loys
Organics
Inorganics
D. Example of Life Cycle Cost Analysis
E. Materials Screening Tests
Laboratory tests
Field tests
F. Computerized Data Base on Materials Experience
G. Computerized Bibliography on FGD Materials
*The following types of materials will be included:
Alloys - solid construction, claddings (including wallpapering)
Organics - shapes, linings (resin and elastomer)
Inorganics - shapes, monolithic linings.
+The following types of components will be included:
Gas circuit - prescrubbers, absorbers, mist eliminators, reheaters, fans,
ductwork, expansion joints, dampers, stack liners
Slurry circuit - pumps, piping, valves, spray nozzles, tanks, agitators,
thickeners.
k.
4A-121
-------
Intentionally Blank Page
4 A-122
-------
ECONOMIC COMPARISON OF MATERIALS OF CONSTRUCTION
OF WET FGD ABSORBERS AND INTERNALS
W. Nischt
D. W. Johnson
M. G. Milobowski
Fossil Power Division
Babcock & Wilcox
Barberton, Ohio
4A-123
Preceding page blank
-------
Intentionally Blank Page
4A-124
-------
ABSTRACT
The Clean Air Act Amendments of 1990 have renewed interest in wet flue gas
desulfurization (FGD) in the U.S. This paper presents an economic comparison of the
common materials of construction for wet FGD spray towers, and a review of experience
with alloy, carbon steel elastomer lined, and carbon steel flakeglass lined
absorbers. An economic analysis of the common materials of construction for absorber
internals is also made. Experience with materials of construction for absorber
inlets, absorber trays, spray headers, mist eliminators, and absorber outlets is
included, as are recommendations for the most cost-effective materials of
construction.
INTRODUCTION
Materials of construction are a major design consideration for new FGD systems. Much
has been learned about materials for absorbers since the 1970s and early '80s, when
the majority of wet FGD systems were placed into operation in the U.S. Some of the
construction materials have stood the test of time, while others failed. This paper
reviews some of the options for construction materials for wet FGD absorbers and
their internals. Economic cost comparisons, generated from current pricing
information, are also presented.
The paper also describes Babcock & Wilcox's (B&W) experience with FGD system
materials of construction. Specifically, the following areas are addressed:
1. Inlet fluework and dampers
2. Absorber inlet wet/dry interface
3. Absorber spray wetted zone
4. Absorber recirculation tanks
5. Absorber moisture separator and outlet zone
6. Fluework downstream of bypass
7. Absorber internal supports
8. Absorber spray headers
9. Hist eliminator spray headers
10. Absorber mist eliminators
11. Absorber tray/gas distribution devices
A typical countercurrent spray tower is shown in Figure 1. The countercurrent spray
tower is the most commonly used design in the wet FGD industry. The tower can be
Preceding page blank
4A-125
-------
divided into zones depending on local corrosion and erosion characteristics. The
major zones to be discussed in this paper are also highlighted. Included as part
of the spray tower is B&W's patented absorber tray which, coupled with a centered
outlet cone, provides even gas flow distribution across the absorber.
BACKGROUND
Materials for early generation FGD systems were either 300 series stainless steels
(316 or 317) or carbon steel lined with elastomers, flakeglass, and/or refractory
materials. B&W has a range of experience with the various materials in FGD service
through 20 operating units. This experience includes 23 absorber towers constructed
of 316L stainless steel, 16 towers of 317L stainless steel, 22 towers of carbon steel
lined with elastomer, and two towers of carbon steel lined with polyester flakeglass.
These and other materials have been used in ancillary sections of the FGD system such
as fluework, inlet sections, internal components, and bypass sections.
In 1990, a formal status survey was conducted on all B&W-designed FGD systems in
operation. One goal of this survey was to update operating experience with various
materials in FGD service. The systems encompassed a diversity of FGD designs and
operating practices, including the use of limestone, lime, magnesium promoted lime,
and sodium based reagents. Also, the record of material applications for FGD
components in dealing with corrosion and erosion as a result of acid dewpoint,
chemical attack, and/or abrasion was reviewed. This data is presented in the
following sections in relation to current material selection criteria.
INLET FLUEWORK AND DAMPERS
The inlet fluework and dampers of the wet FGD system are located directly upstream
of the absorber tower, where they are exposed to dry, hot (300F+) flue gas. The
300F+ temperatures are well above the acid dewpoint of the gas, so A-36 carbon steel
is acceptable and the most economical material choice for the fluework and the damper
structural elements. Guillotine damper blades, which routinely remain raised and
exposed to nothing more corrosive than the atmosphere, can be fabricated from ASTM
242 steel with seal strips made from 300 series stainless steel. While more exotic
materials can be specified for this service, their use is extravagant unless tower
"spray-back" is a concern, or ambient conditions warrant their use.
The possibility of gas recirculation carrying spray-back into the inlet flue (Figure
2) depends primarily upon inlet configuration. The absorber tower design
incorporates an awning over the inlet and side curtains to minimize spray-back. To
protect against corrosion resulting from any tower spray that does penetrate the
inlet, B&W's practice is to extend the materials used for the wet/dry interface from
the tower five feet back into the inlet flue.
ABSORBER INLET WET/DRY INTERFACE
The tower inlet is classified as a wet/dry zone because it is exposed to both the
incoming dry, hot flue gas and the tower's recirculating slurry sprays. This zone
probably experiences the most severely corrosive conditions in the absorber system;
not only is it very acidic due to the presence of chloride and/or fluoride, but this
wet/dry interface fosters scale formation and the possibility of localized,
concentrated, under-deposit corrosion.
A variety of materials have been used in the wet/dry zone. The earliest FGD systems
used a polyester flakeglass corrosion barrier covered with three inches of refractory
for thermal protection. The polyester has held up quite well, but the refractory
4A-126
-------
often requires maintenance due to softening and the formation of cracks. After eight
years of operation one plant changed to a alloy C22/C276 wallpaper system from the
refractory over polyester flakeglass system. The remaining installations have an
average life of nearly 11 years and still use the original system, though repairs
are often made with materials other than the original.
One unique FGD system incorporates a pre-scrubber to remove HC1 and dust before the
absorber. This wet/dry zone is lined with borosilicate block covered with an
abrasion resistant tile in the spray zone. This unit has been commercially operating
for five years with only minor repairs required to replace dislodged tiles.
For most newer FGD systems, alloy metals and linings, rather than non-metallic
linings, are the materials of choice at the wet/dry zone. Primarily, this choice
is made due to high temperatures and to reduce maintenance requirements. Austenitic
stainless steels are suitable for applications with chloride concentrations below
20,000 ppm. Duplex stainless steels can be utilized in chloride concentrations to
45,000 ppm, and high nickel alloys such as alloy C-276/C-22 can be used for chloride
concentrations greater than 45,000 ppm. Due to evaporation of the slurry at the
wet/dry zone, concentrations of chloride ion in excess of 100,000 ppm are possible.
Many utilities are now specifying closed loop water systems in which chloride
concentration can reach or surpass 50,000 ppm in the recirculated slurry. Alloy
C-276/C-22 wallpaper (typically 1/16 inch thickness) over carbon steel is recommended
for the wet/dry interface. The alloy wallpaper approach is an effective means of
corrosion control at a reasonable cost. Table 1 illustrates relative material costs
for wet FGD absorbers. It shows the erected cost of C-276/C-22 wallpapered carbon
steel is comparable to non-metallic lined carbon steel or the higher alloyed
austenitic stainless steels, but is only half the cost of C-276/C-22 solid plate.
As with any lining system, proper application is key to the system's integrity.
Metallic wallpapering systems are no different than non-metallic lining systems in
that proper applicator training, supervision, and QA inspection/testing are
essential. Metallic wallpapering also provides greater longevity than a non-metallic
lining system, but it does have the drawback that any application flaws not detected
by inspection will, more often than not, go unnoticed until substrate material damage
is discovered. Application flaws in non-metallic linings are more readily apparent
prior to severe substrate damage, via lining blistering, cracking, or delamination.
ABSORBER SPRAY WETTED ZONE
The absorber spray wetted zone is unique in that it is subject to both corrosive and
abrasive conditions. Slurry spray nozzles must be positioned within the tower to
obtain complete cross-sectional area coverage, eliminating any possibility of
untreated flue gas channeling through the sprays. Spray impingement against the
tower shell does occur, and since the reagent slurry is typically 15% solids, shell
abrasion can result.
Experience in the absorber spray zone has been varied. Some early applications of
natural rubber did not last as long as expected because of improper compounding, poor
application, and deterioration of the lining due to aging in the scrubber
environment. Natural rubber has lasted from a low of less than four years (due to
application problems) to units still in service after more than 12 years. Generally,
repairs are made with chlorobutyl rubber or synthetic rubber to areas that have
failed after long-term exposure. The failures are primarily in the form of
bl istering.
It should be noted that non-metallic lining formulations have improved dramatically
over the past twenty years. Elastomer linings are a case in point. First generation
4A-127
-------
wet scrubbers used soft natural rubber linings that, while superior in abrasion
resistance, offered poor permeability resistance. Today, three-ply natural rubber
(utilizing a hard rubber layer sandwiched between two soft rubber layers), neoprene
rubber, and chlorobutyl rubber are widely accepted for FGD service. While neoprene
rubber has an advantage in that it is non-flammable, it is more costly and harder
to apply than other elastomers. Chlorobutyl rubber, with its 220F maximum service
temperature, excellent abrasion and permeability resistance, and competitive price
is B&W's recommended elastomer lining choice.
One FGD system that uses a three-ply rubber has been in service for over eight years.
The three-ply rubber requires about 10 to 20 patches every six months. B&W has also
used chlorobutyl rubber. After six years of service, the only rubber lining failures
in this system have been due to spray impingement on supports, and these areas are
now protected with 409/410 stainless steel wear plates. This could also be
accomplished with a double layer of rubber. Absorber modules lined with polyester
(flakeglass) have been in service for up to 13 years. These are inspected twice a
year and minor repairs are made as required.
For utility customers wary of non-metallic linings, but unable to economically
justify a solid C-276/C-.22 plate tower, the wallpapering or cladding concept is
gaining acceptance as an alternative material choice. In direct spray impingement
zones, thicker wallpaper or cladding is recommended.
During the survey, the only problem noted with solid stainless steel absorbers is
erosion at the point of the strongest impact from the spray nozzles. Units with
stainless construction have experienced no significant failures. Operators of units
that have been in service from seven to 18 years are only now considering
installation of wear plates or linings at spray impingement areas.
While a number of variables can dictate final material selection, cost (both major
capital and total life cycle, including replacement and maintenance costs) is by far
the predominant factor. Three hundred series stainless steels have seen considerable
use in absorber tower fabrication. But, with the growing interest in minimal
discharge, closed loop systems, scrubber liquor chloride levels are exceeding the
limits of austenitic stainless steels. As a result, non-metallic linings, duplex
stainless steels such as alloy 255 or high nickel alloys such as C-276/C-22 are the
only viable material options.
The most economical material of construction that can be supplied is elastomer lined
or glass flake-filled plastic lined carbon steel (flakeglass can be polyester or
vinylester based). In spray zone impingement areas, abrasion resistant formulations
or double lining thicknesses is recommended. The major shortcoming of non-metallic
lining systems is that their expected life is only one-third to one-half the life
of the FGD system, necessitating at least one complete re-lining operation at
considerable downtime and expense.
ABSORBER RECIRCULATION TANK
In the B&W absorber tower design, the recirculation tank is integral to the tower
structure. Therefore, material options for the recirculation tank are similar to
those for the tower's spray wetted zone. There are, however, a number of differences
that should be noted.
The absorber recirculation tank is an agitated vessel in which side-entry agitators
are used to keep the slurry solids in suspension. The tank shell abrasion is
considerably less severe than in the tower's spray wetted zone since the "cutting"
action of spray impingement is not present. But, the floor surface lining must be
able to withstand possible mechanical damage caused by annual outage maintenance and
inspections.
4A-128
-------
Acid resistant brick, block, or tile can be used to line a carbon steel recirculation
tank, but their costs are slightly higher than an elastomer or plastic lined vessel.
These ceramics are highly resistant to corrosion and erosion; however, the mortar
can become porous after long term immersion, necessitating an elastomeric or plastic
lining/membrane on the tower shell surface for corrosion protection. Also, vibration
and thermal cycling can cause the bricks or tile to come loose from the lined
vertical surfaces.
While stainless or alloy steels could provide excellent service in this application
depending on system chemistry, the lowest initial capital cost results from using
a glass flake-filled plastic lining over carbon steel. Use of elastomer lining is
the next best cost alternative. The principle drawback of such linings is their
limited expected life, generally 10-15 years. Both lining systems are also
susceptible to mechanical damage, so the installation of an acid brick overlay on
floor surfaces is suggested.
ABSORBER MOISTURE SEPARATOR AND OUTLET ZONE
The absorber moisture separator and outlet zone is exposed to a different environment
than those previously discussed. The abrasive action of the recycle slurry no longer
predominates. However, water for moisture separator washing combines with flue gas
residual sulfur dioxide to form sulfurous acid resulting in a highly corrosive
environment.
Because abrasion resistance is no longer a material selection criteria, the use of
an elastomer lining is unnecessary. A glass flake filled plastic lining over carbon
steel is typically recommended. If the absorber spray wetted zone is lined with
chlorobutyl rubber, a lining material interface will exist near the moisture
separators. By conditioning the plastic lining surface and overlapping the
chlorobutyl rubber onto it, a sound interface lining bond is achieved and system
corrosion resistance integrity maintained.
The outlet flue was a major maintenance area for early FGD systems. Along with
damper maintenance and cleaning necessitated by scale formation, problems in outlet
flues have resulted in a large portion of the outage time affecting system
reliability of existing scrubbers.
B&W's oldest operating FGD system, placed into service in 1973, has a 316L stainless
steel outlet flue with a refractory liner. This system has experienced minimal
problems. The refractory covered flakeglass lined outlet flue on another unit with
14 years of service was recently replaced with solid alloy C-276. A sister unit at
the same station, in operation for 11 years, still has the original
refractory/flakeglass lining.
Two units without bypass flues downstream of the scrubber have been in operation for
10 and five years, respectively, with the original polyester flakeglass lining over
carbon steel. The linings in these units are still in good condition requiring only
minimal routine maintenance.
Some units have required replacement of the original lining and/or the entire outlet
flue. This choice was made because of the comprehensive maintenance that becomes
necessary as the linings age. The deterioration of the lining in the outlet has been
most pronounced in areas where bypass gas mixes with the saturated gas leaving the
absorber tower. The materials of choice for outlet flue repairs include alloy C-
276/C-22 (solid, wallpaper, or clad); refractory covered polyester flakeglass;
fiberglass reinforced plastic (FRP) though used on a more limited basis, and
borosllicate glass block with a furan based mortar. When borosilicate glass block
is used, floors should be lined with other materials (such as alloy C-276/C-22)
because the glass block has very little abrasion resistance and is not suitable for
foot traffic.
4A-129
-------
Many existing units now have more than one type of lining in the outlet flue. This
1s because the extent of lining failure is dependent upon such things as: materials,
operating conditions, maintenance practices, and length of time in service. Lining
failures generally occur in isolated areas rather than as wholesale failures.
Therefore, repairs are made to match the new lining to the conditions at the various
locations between the outlet and the stack. One of the worst locations, flue work
downstream of the bypass, is discussed in the next section.
FLUEWQRK DOWNSTREAM OF BYPASS
On some FGD units, a flue gas bypass around the absorber tower is provided. This
bypass typically ties back into the tower's outlet fluework upstream of the stack.
The outlet fluework at this tie-in is exposed to the highly corrosive environment
of a mixture of scrubbed gas and hot bypassed gas.
Due to the high (300F) temperatures at this section of the flue, there are only a
very limited number of plastic linings suitable for this service. Better success
1n this application has been achieved using alloy C-276/C-22 wallpaper lining or
borosilicate block. Borosilicate block is an extremely light, closed cell,
inorganic, foamed, glass block that is virtually impermeable to acidic liquids and
gases at temperatures to 950F. One shortcoming, discussed earlier, to using
borosilicate block is that it is susceptible to mechanical damage - floor surfaces
have to be protected during maintenance inspections/operations.
ABSORBER INTERNAL SUPPORTS
Failure of absorber internal support linings is most often due to spray impingement.
These areas are repaired and protected by doubling the liner thickness or using a
replaceable shield material. Wholesale failure of absorber internal supports has
not been experienced in B&W systems.
Recently, there has been some interest in self-supporting headers. However, with
the large absorber towers which are being proposed today, this can be impractical.
Supports constructed of 300 series stainless steel, duplex stainless steels, high
nickel alloys, elastomer lined, or plastic lined carbon steel have been used,
depending on the chloride level of the recirculating slurry. Typically, the supports
provided are constructed of the identical material to the absorber shell.
ABSORBER SPRAY HEADERS
Absorber spray headers are exposed to the recirculating slurry on both the inside
and outside surfaces. Therefore, they must be made of abrasion and corrosion
resistant materials. B&W has utilized headers constructed of abrasion resistant
FRP and austenitic stainless steels. Headers constructed of high nickel alloys, and
rubber lined rubber covered carbon steel material have also been proposed.
Abrasion/corrosion resistant FRP is the most common material chosen for spray
headers. It offers excellent resistance to chemical attack from chlorides, however,
it is susceptible to failure due to erosion. By managing the grind of the limestone
and, therefore, the abrasiveness of the recirculating slurry (B&W recommends that
the grind be maintained at 95% passing 325 mesh), problems with abrasion can be
controlled. It is also important that the slurry spray nozzles be properly connected
to the spray headers. A small flange leak can grow into a fan spray causing severe
erosion in the affected area. The spacing between the nozzles and the headers is
critical.
4A-130
A
A
-------
Abrasion resistant FRP headers may require maintenance during their useful life;
fortunately, typical maintenance can be performed with the headers in place and is
usually on a small, isolated section of the header. If a section must be replaced,
it can be removed and a new section installed with a field joint.
Absorber spray headers constructed of austenitic stainless steels are suitable for
service with chloride concentration below 20,000 ppm. The header material can be
supplied in alloys up to 317LH material, however, the flanges are typically only
available up to 317L material. Spray headers constructed of austenitic stainless
steels are nearly maintenance free. In addition, less support structure is required
with metallic spray headers.
Absorber spray headers constructed of high nickel alloys can be utilized to chloride
concentrations greater than 100,000 ppm. Headers constructed of high nickel alloys
are similar to the austenitic stainless steels in that they are nearly maintenance
free.
Absorber spray headers constructed of rubber lined rubber covered carbon steel are
also available. These spray headers provide excellent corrosion protection similar
to abrasion resistant FRP and the high nickel alloys. The headers also provide
excellent erosion protection. However, rubber lining of both the inside and the
outside surfaces of the carbon steel pipe is intricate work, especially if the
header/branch geometry is complicated. Maintenance on rubber lined rubber covered
carbon steel headers will also be required and is more complicated than with FRP
headers. If an area of rubber lining fails, the carbon steel may quickly corrode,
and the failure is difficult to detect.
Table 2 shows an economic comparison of the spray header materials discussed above.
The abrasion resistant FRP is capable of surviving high chloride levels and offers
the lowest relative cost of all the materials.
HIST ELIMINATOR SPRAY HEADERS
Mist eliminator spray headers are not subject to abrasion, however, the pH on the
outside surface can be quite low. The low pH coupled with high chloride levels make
the environment in this area aggressive. The quality of the wash water will have
an impact on the pH and chloride level. If recycle water is blended with make-up
water to wash the mist eliminators, some level of chlorides will be present in the
wash water and, therefore, in contact with the mist eliminator spray headers.
Mist eliminator spray headers can be constructed of FRP, austenitic stainless steels,
or high nickel alloys. Rubber lined rubber covered carbon steel mist eliminator
headers are typically not available in the small size range of the branch headers.
One drawback to the FRP headers is that branch headers can be broken off from the
main header if they are stressed, usually during maintenance. Metallic headers are
not as susceptible to breakage.
The allowable chloride levels for austenitic mist eliminator spray headers are lower
than those recommended for the absorber spray headers. pH levels of 2.0 are not
uncommon in this area. At a pH of 2.0, 317LM pipe is suitable for a chloride
concentration of up to approximately 1000 ppm, while C-276/C-22 and FRP are suitable
to greater than 100,000 ppm chloride.
The relative costs for the mist eliminator spray header materials are shown in Table
3. Based on FRP's superior chemical resistance and lower cost, it is the material
of choice for this application. B&W has successfully utilized FRP spray headers on
nearly all of its installations. Austenitic stainless headers have also been used
with equal success, but their initial installed cost was higher than FRP.
4A-131
-------
ABSORBER MIST ELIMINATORS
Mist eliminators can be constructed of FRP, glass coupled polypropelene, austinetic
stainless steels, duplex stainless steels, or high nickel alloys. The most common
material for mist eliminators used in the U.S. is FRP. As with FRP or plastic
materials used elsewhere in the FGD system, the formulation must be carefully
selected to withstand the process environment. FRP mist eliminators have an expected
life of four to 10 years. Polypropelene mist eliminators have gained acceptance in
Japan and Europe in the original equipment market, but are only now gaining
acceptance in the U.S. Glass coupled polypropelene has been utilized in the
replacement market in the U.S. Polypropelene mist eliminators have an expected life
of four to five years.
FRP mist eliminators have an upper temperature limit of approximately 340F while
polypropelene's upper temperature limit is approximately 210F. Since the normal
operating temperature in an FGD absorber is typically below 130F, either mist
eliminator material is acceptable. When utilizing either polypropelene or FRP mist
eliminator material, care must be taken to design for upset conditions.
Use of metallic mist eliminators has not been as common as FRP. Low pH coupled with
the chloride levels found in the mist eliminator zone of an absorber have made
austenitic stainless steels such as 317LM inappropriate. The relative corrosion rate
of 300 series stainless steels in this environment is high. Mist eliminators
constructed of thin gauge stainless steels will not withstand the corrosive
environment. Further, stress and crevice corrosion would be accelerated in areas
where the austenitic stainless steels are mechanically formed into chevron shapes.
Duplex stainless steels and high nickel alloys can survive in the environment,
however, they are expensive.
Table 4 shows the relative cost of the mist eliminator material. Based on the
expected life and cost, B&W recommends the use of glass coupled polypropelene.
A major source of damage to mist eliminators occurs during maintenance outages. The
damage occurs either from foot traffic or high pressure washing. For all of the mist
eliminator materials discussed above, direct foot traffic is not recommended;
planking should be utilized to distribute the load. High pressure washing should
be carried out following the mist eliminator manufacturer's recommendations to avoid
damage.
ABSORBER TRAY/GAS DISTRIBUTION DEVICE
All of B&W's wet FGD absorbers utilize a patented absorber tray for flue gas
distribution across the absorber. A portion of the compartmentalized tray is shown
in Figure 3. The absorber tray is typically located between the first and second
spray level as shown in Figure 1. The absorber tray is subjected to saturated flue
gas and recirculating slurry. The recirculating slurry scours the tray, keeping it
clean, without significant erosion.
The absorber tray is normally constructed of austenitic stainless steels, duplex
stainless steels, or high nickel alloys depending on the chloride concentration and
pH of the recirculating slurry.
For chloride concentrations below 20,000 ppm, austenitic stainless steel is
acceptable. For chloride levels below 45,000 ppm, duplex stainless steels are
acceptable. High nickel alloys can be utilized for chloride concentrations greater
than 100,000 ppm.
An economic comparison of the absorber tray materials discussed above is shown in
Table 5. The relative pricing is essentially equivalent to the cost of the metallic
materials.
4A-132
-------
CONCLUSION
There are a large number of materials available for construction of FGD systems.
The selection of materials is dependent on a number of factors Including: operating
conditions (controlled or uncontrolled), corrosion, erosion, temperature
characteristics of the specific zones within the FGD system, initial cost, and
maintenance philosophies. Experience within the industry has eliminated many
materials due to problems associated with construction and maintenance. Preferred
materials include stainless steels, FRP, chlorbutyl or layered rubber, vinylester
and polyester flakeglass, tile and/or refractory materials, and borosilicate glass.
Cost information provided in this paper can be used as a guideline to help make
material decisions.
TABLE 1
FGD TOWER COSTS
The following tower plate (with stiffeners) cost comparisons are based on 0.25 inch
plate and knockdown construction. Costs are based on recent proposal quotes and
budget information received from B&W Purchasing and Construction. The least
expensive option is assigned a base cost 1.0. Costs of all higher priced options
are presented relative to the base cost of 1.0.
COST R
lATIO
MATERIAL
MATERIAL/
FABRICATION
ERECTION
TOTAL
CARBON STEEL
1.00
1.00
1.00
CS ELASTOMER LINED1
6.50
1.00
2.22
CS PLASTIC COATED1
5.50
1.00
2.00
316L STAINLESS
4.17
1.00
1.70
317L STAINLESS
5.00
1.00
1.89
317LM STAINLESS
5.50
1.00
2.00
317LMN STAINLESS
6.00
1.00
2.11
ALLOY C-276
16.50
1.19
4.59
ALLOY C—22
16.50
1.19
4.59
ALLOY 255
8.50
1.05
2.67
CS W/C-22 CLADDING
11.33
1.14
3.40
CS W/C-22 WALLPAPER
5.33
1.43
2.30
CS W/TILE
6.83
1.00
2.30
CONCRETE/BLOCK
11.83
INCLUDED
2.63
NOTES: (1) LINING/COATING COST VARIES BASED ON SURFACE AREA
AND SITE LOCATION. LINING MATERIAL COST INCLUDES
FIELD INSTALLATION.
4A-133
-------
TABLE 2
The least expensive option is assigned a base cost 1.0. Costs of all higher priced
options are presented relative to the base cost of 1.0.
COST COMPARISON OF ABSORBER
SPRAY HEADER MATERIALS
MATERIALS
COST
RATIO
CHLORIDE LIMIT
Abrasion Resistant FRP
1.0
> 100,000 ppm
Carbon Steel, Rubber Lined,
Rubber Covered
1.5
> 100,000 ppm
317LM Stainless Steel
2.5
< 20,000 ppm
Alloy C-276/C-22
6.8
> 100,000 ppm
TABLE 3
The least expensive option is assigned a base cost 1.0. Costs of all higher priced
options are presented relative to the base cost of 1.0.
RELATIVE COSTS OF HIST
ELIMINATOR SPRAY HEADERS
MATERIALS
COST RATIO
CHLORIDE LIMIT 0
pH OF 2.0
Abrasion Resistant FRP
1.0
> 100,000 ppm
317LM Stainless Steel
2.5
< 1,000 ppm
Alloy C-276/C-22
6.8
> 100,000 ppm
TABLE 4
The least expensive option is assigned a base cost 1.0. Costs of all higher priced
options are presented relative to the base cost of 1.0.
RELATIVE COST OF
MIST ELIMINATORS
MATERIAL
COST
RATIO
CHLORIDE LIMIT
TEMPERATURE
LIMIT
Glass Coupled
Polypropelene
1.0
> 100,000 ppm
210F
FRP
2.6
> 100,000 ppm
340F
317LM Stainless Steel
3.0
< 1,000 ppm
700F
Alloy C-276/C-22
7.6
> 100,000 ppm
700F
4A-134
A
A
-------
TABLE 5
The least expensive option is assigned a base cost 1.0. Costs of all higher priced
options are presented relative to the base cost of 1.0.
RELATIVE COST OF
ABSORBER TRAYS
MATERIAL
COST RATIO
CHLORIDE LIMIT
317 LMN Stainless Steel
1.0
< 20,000 ppm
Alloy 255
1.5
< 45,000 ppm
Alloy C-276/C-22
2.5
> 100,000 ppm
4A-135
-------
— Flue Downstream of Bypass
Moisture Separator
and Outlet Zone
Secondary
M.E.
Underspray
Mist Eliminator
Mist Eliminator
Bypass
Duct
Primary
M.E.
Overspray
Primary M.E. Underspray
^_^__Abaprb|r Spray Level ^
Absorber Spray Level
-frl-"—' ¦' ' "
Absorber Spray Level
Abosrber Tray
Spray
Wetted
Zone
Absorber Spray Level
Normal Liquid Level
Inlet Flue
and Damper
Inlet
Recirculation
Tank
Oxidation
Sparge Header
Figure 1
Absorber Spray Tower with Bypass
4A-136
-------
Moisture
Separator
Water
Wash
Nozzles
Patented Alloy
Perforated Tray
Figure 2 Absorber Cutaway View
Patented Alloy
Perforated Tray
Figure 3 Absorber Gas Distribution Device
4A-137
-------
Intentionally Blank Page
4A-138
-------
THE
INTELLIGENCE & ECONOMICS
OF
F.R.P.
(fiberglass reinforced plastics)
IN
F.G.D. SYSTEMS
E.J. BOUCHER
REINFORCED PLASTIC SYSTEMS
ABCO PLASTICS DIV.
136 Simsbury Road
AVON, CT. 06001
R.D. BRADY
COMPOSITE CONSTRUCTORS & ENGINEERS
365 118th AVE. S.E.
BELLEVUE, WA. 98005
Preceding page blank
4A-139
-------
Intentionally Blank Page
4A-140
-------
ABSTRACT
In a wet flue gas desulfurization system/ the architect/engineer/
system supplier, and utility (end-user) face a myriad of problems in
material selection. This paper sets forth some comparative costs and
intelligence factors which should enable the decision maker (materia
specifier) to consider the use of an abrasion/corrosion resistant
fiberglass reinforced plastic in piping/spray headers/ demist piping
outlet ducting/and absorber vessels.
INTELLIGENCE :
The advantages of a corrosion/abrasion resistant FRP piping system:
A. Lightweight /easy to handle as compared to metals.
Examples: wgt/lbs/ft FRP CSRL SS 3161
6" 3.0 19.0 18.9
12" 9.6 50.0 53.5
24" 23.7 95.0 171.1
FRP piping is 5 to 7 times lighter than corresponding metals.
B.Corrosion resistant : FRP is resistant to sulfur dioxide gases/
(300° f maximum exposure)/ Alkaline and chloride saturation. There
are no known "chemical attack"failures on FRP piping in wet FGD
systems.
C. Abrasion Resistance: Manufacturers offering abrasion resistant
piping systems have developed and tested a specialized liner to
handle the limestone slurry. (I.D. lined for recycle applications
and other applications outside the absorber; I.D & O.D. lined for
Preceding page blank
4A-141
-------
spray headers and other applications inside the absorber.) The
limiting factors involved in making a selection of abrasion resist-
ant FRP piping are:
• particle size of the slurry should not exceed 100 mesh (150
microns).
• velocity of the slurry should not exceed 12 ft/sec.
• % solids concentration should be in the 20/25% by weight range.
Most FGD system suppliers target their slurries to operate in the
range of 325 mesh (44 microns) and 5-8 ft/sec. - both well within
the operation range of an FGD designed FRP abrasion resistant liner.
CAVEAT : generic chemical grades of FRP do not have this type of
abrasion resistant liner and may not be suitable for your
slurry application. When in doubt/- ask the manufacturer
for test data and/or case histories.
D. Low installed cost : see the economic section of this paper.
E Additions/ deletions/ or repairs easily accomplished : no open
flame or complex equipment is needed to make an ajustment - repair
and maintenance kits are available from the manufacturer and many
alterations can be made inside the scrubber during a shutdown. Also/
some manufacturers have training programs for plant personnel to do
on-site fiberglass work.
F. Successful case histories; Abrasion resistant FRP is in operation
at more than 75 plant sites at this time.
G. Larger flow capacity: FRP piping will have a larger i.d. than CSRL
and/as a result/ pumps will work better and reduced horsepower may be
4A-142
A
-------
hidden advantage. There are case histories where pump abd/or line
size has been reduced .
H. TEMPERATURE : FRP designed for FGD applications can operate at
300° f in gas service and at 200° f for liquid service. Temperature
excursions of 375 -400° f have been tolerated by products of at least
one manufacturer. This property gives the system supplier/architect
engineer a larger safety factor than available with CSRL.
ECONOMICS:
In order that the writer provide a fair and impartial economic
evaluation/ quotations were solicited for both labor and materials
from several sources - see Acknowledgements 1,2 & 3/ and refs 1 -7.
Assumptions were as follows:
• Abrasion/corrosion resistant FRP
• Carbon steel rubber lined sch 40, (24" is sch 20) 1/4" nat rub.
• Stainless is 3161 sch 40
• local deliveries - at least 500 ft of pipe involved
• A simple and a complex piping configuration is used
In the simple configuration prices were obtained for sizes 6",12"/ &
24". (prices were checked against the mechanical estimating guides
listed in ihe refs. and in all cases the contractors prices were
lower and used here.)
k.
4A-143
-------
0 SIMPLE CONFIGURATION "fig.a"
lO'O"
FRP
10'-0"
©
STEEL
10'-0"
Abbreviations; M = material ; L = labor; T = total cost
CS = carbon steel RL = rubber lined FRP - fiber glass
reinforced pipe w/ abrasion resistant liner
( ) Brackets under the FRp cost indicate the material
$reduction achieved by removing the ^ flanges
shown in the steel configuration.
The "intelligent net" is the $ figure that has been achieved
by removing 4 flanges from the FRP assembly. The light weight#
easy handling aspect of FRP allows this feature not readily
available in metallic systems.
4A-144
A
A
-------
SIMPLE CONFIGURATION "fig.a"
6" size
FRP
M 1473.00
L
T
600.00
2073.00
Remove 4 figs (412.00)
Intelligent net 1661.00
CSRL
367.86
587.00 rl
2007.00
2962.86
(42%)
(78%)
SS 316 sch40
1923.90
3416.00
5339.40
(257%)
(321%)
% in brackets is the premium paid for this choice of material over
FRP.
12" size
FRP
M 5299.00
L
T
1080.00
*
6379.00
Remove 4 Figs. (1612.00)
Intelligent net 4767.00
CSRL
1298.00
2191.00 rl
3875.00
7364.00
(15%)
(54%)
SS 316 sch40
7758.00
6530.00
14288.00
(223%)
(299%)
24" size
M
L
T
FRP
17540.00
3600.00
21140.00
(32%)
Remove 4 figs. (5400.00)
Intelligent net 15740.00
CSRL
5214.00
2370.00 rl
8428.00
16012.00
(-32%)
( 1.7%)
SS 316 sch 40*
29225.00
13579.00
42840.00
(223%)
(272%)
* = sch 10 SS may be 30/40
% lower in this size
4A-145
-------
COMPLEX CONFIGURATION "FIG.B"
This configuration represents one (1) level of a six (6) level spray
tower and the associated recycle piping. The sketch is typical and
the additional sketches and pipe layouts were not included for space
saving reasons. The quotations received all included the six sketchs
and associated piping.
MATERIAL FRP CSRL SS316 sch -40
Total fab cost 1/486,331 1/ 817/ 796 3,486,980
22% premium 235% premium
OBSERVATIONS :
• FRP piping is a viable and cost saving alternate to rubber
lined pipe* stainless steel,and other alloys.
• Life expectancy of 10 years or more can be realized if the
slurry is kept at the manufacturer's recommended sizes and
velocities.
• Maintenance can be accomplished at plant sites with modest
training.
• Additions of nozzles, alterations,and other items can be
worked on the pipe inside the scrubber during shutdowns.
• Even without maintenance - headers that have been damaged can
be returned to the factory for reconditioning at a fraction
of the cost for new replacements.
4A-146
-------
l.D. VESSEL
TVPICAL.
SK-A-IOJ -THRU
SK-A-I08
COMPLEX CONFIGURATION "FIG.B
-------
OUTLET DUCTS/ VESSELS/AND STACK LINERS :
During the late 60's & 70'S/FRP became the material of construction
for many power plant stack liners and duct installations. The cool
wet/acid laden gases proved to be too severe for the conventional
brick and coated steel systems. FRP structures were designed and
manufacturing equipment was developed to produce large structures at
competetive costs.(sixteen foot to ninety foot diameter/ field wound
tanks/ ducts/ and liners are feasable). The ease of handling and
erection provides a shorter construction schedule/ on-site quality
control/and immediate vendor contact and communication. Many install-
ations have shown that installation times may be cut by 6 months or
more.
The FRP winding machinery is set up at the job-site and the modules
(tanks/ducts , and/or liners) are fabricated on site for installation
in-place. This unique /self-contained/ manufacturing capability
allows the FRP to be installed without interference with or from
other construction activities. Improvements in resin systems/engin-
eering/ field control/and history allow a product to be a viable
alternate to the mass and high costs of alloys/ coatings/ and other
labor intensive materials.
CONSIDERATIONS :
Vinylester resins contain terminal cross-linking over the entire
length of hte molecular chain - allowing the material to elongate
under stress and/ thus/ absorb mechanical and thermal shocks. The*... j
4A-148
-------
end result is a tough laminate/corrosion resistant/and capable of
long life at temperatures of 300° F. Quench systems could be used
if higher temperatures are anticipated.
EXPANSION : FRP generally expands 1 1/2 to 3 times the rate of steel
(actually about the same as copper or aluminum) depending upon the
size# resin system# glass loading, and other design considerations.
Since the material has a low modulus , the expansion can be handled
with expansion joints,anchors/ or incorporating the expansion problem
into the design.
THERMAL
RETENTION -
FLEXURAL STRENGTH,
PSI
Temp
•
ASTM 3299
conv. resin
Dera 470
510c
510n
Room
19000
29600
24000
23800
25000
150°
F
28500
23800
24000
200
F
27400
24500
24000
25600
225
F
14700
21000
24400
250
F
5000
24100
12000
18400
300
F
3200
21000
325
F
12000
350
F
8000
COMPARATIVE CYCLIC LOADING RESULTS
CYCLES TO FAILURE
Materials 30% ultimate stress 10% ultimate stress
904L 200 4000
C276 200 5000
Vinylester Composite 5000 100000
k.
k
4A-149
-------
FIRST COST COMPARISONS : Multipliers
MATERIAL STACK LINERS DUCT ABSORBER VESSELS
CSRL NA NA 1.15
FRP 1.0 1.0 1.0
316L 1.18 1.18 1.18
317LM 1.25 1.25 1.35
.C276 clad 1.35 1.35 1.75
C276 1.80 1.80 2.00
Brick* .83 NA NA
Basis : 300' tall stack liner/ zone 0 or 1, ( * = NA in zone 2B or
higher.
MAINTENANCE :
FRP does not rust/corrode, spall/ scale/ or need extra protection
from chemical attack. FRP is corrosion resistant inside and outside/
and requires no painting, cathodic protection, or other treatments.
If minor repairs have to be made, they, usually can be made on-site
with the manufacturer's repair kit. Overall, FRP has lower costs,
i.e. initial cost, installation cost, scheduling savings,and life-
cycle cost. (Cost comparisons should be made on an individual basis
since service conditions differ at each plant)
4A-150
-------
PRP USED IN PIPING,STACK LINERS,VESSELS/ AND OUTLET DUCTS
CAN OFFER THE UTILITY INDUSTRY (especially wet F.G.D systems)
substantial advantages in lieu of lined metals and exotic
alloys.
Frp installed cost data shows premiums of 15% to over 250%
are being paid by AE's, system suppliers,and utilities to
use lined products or alloys.
System suppliers and material engineers need to be aware of
the cost/performance advantages of FRP materials.
Actual case histories of 10 years or more are in existance
giving credence to the life cycle cost advantages of FRP.
FRP offers the plant maintenance personnel an "easy" mater-
ial to work, perform alterations, or repairs.
Significant additional savings can be experienced with
prefabrication, on-site fabrication, and in the case of
chimney liners - manufacturing inside the existing chimney.
4A-151
-------
ACKNOWLEDGEMENTS
1. James Brunelle/ Harry Grodsky & Co./ Mechanical Contractors
Springfield, Mass.
2. Lalit Mehra/ Ardco Companies/ Denver/ Colorado
3. D. Bell/ P. Bruce/ K. Eisner/ S.Tanner; Reinforced Plastic /
Systems/ ABCO Plastic Div./ Mahone Bay/ Nova Scotia/ Canada
REFERENCES
1. National Mechanical Estimator/ Ottaviano Technical Services
Melville,N.Y. 1985, pgs 4-13,4-14,4-16/4-17,
2- Means Plumbing & Piping Cost Data, R.S. Means Co. Inc., Kingston,
Mass. 1989 12th Edition/pgs 69/77/ 89-119.
3. Dow Chemical Installed Cost of Corrosion Resistant Piping Systems
Dow Chemical Co. Midland/ Mi.
4. Mechanical Contractors of America Estimating Guide/ 1979 rev 2.
5. Personal Communication/All Stainless Pipe & Fittings/ Cambridge,
Mass. 9/5/91
6. Personal Communication/ Tube Sales, Agawam/ Mass.;9/6/91
7. Quotation and personal communication/ Harry Grodsky & Co./
Springfield, Mass ., 9/5;9/20;9/23; 9/27; 9/30 -1991
4A-152
A
A
-------
Session 4B
DRY FGD TECHNOLOGIES
LIFAC DEMONSTRATION AT POPLAR RIVER
T. Enwald
Tampella Power Canada Inc.
2022 Cornwall Street
Regina, Saskatchewan
Canada S4P Osl
M.E. Ball
Saskatchewan Power Corporation
2025 Victoria Avenue
Regina, Saskatchewan
Canada s4P Osl
ABSTRACT
A full scale demonstration unit of Tampella*s LIFAC desulfurization process was
tested at saskPower's poplar River Power station unit #1. Parametric testing was
conducted to evaluate the S02 removal performance of the process in October to
December 1990. A 1500 hour operability test was performed in December 1990 to
July 1991 to assess impacts on the power plant's operation.
Capital and operating costs of the LIFAC process were evaluated based on the
demonstration unit•s costs.
Based on the performance of LIFAC at Poplar River, saskPower has committed to
installation of LIFAC on its shand Power station which is now under construction.
The startup of Shand Power Station and the LIFAC system are scheduled for the
summer of 1992.
4B-1
-------
BACKGROUND
In the early 1980*s Saskatchewan Power Corporation (SaskPower) recognized
that S02 control would be required for coal fired units in the late 1980's
and beyond. This requirement was anticipated even though the corporation's
facilities are fired on lignite with a relatively low sulfur content
(typically 0.4 to 0.8% S). In response to this need, SaskPower became active
in development of furnace sorbent injection as a control technology. By
1988, SaskPower had committed to the use of sorbent injection for SO control
at the shand Power Station which is scheduled for start-up in mid 1992.
Also in the early 1980's, Tampella Power Inc. of Finland began development
of a humidification process which would enhance the effectiveness of furnace
sorbent injection. The Tampella process is called LIFAC. LIFAC is an
acronym for Limestone Injection into the Furnace and reactivation of Calcium.
In mid 1989, SaskPower and Tampella Power came to an agreement to conduct a
demonstration project for LIFAC at SaskPower's 300 MWe Poplar River Unit #1.
The purpose of the demonstration unit was to prove LIFAC's suitability for
Canadian lignite and to prove its low investment costs.
The LIFAC process has also been installed at two power stations in Finland
and one in the Soviet Union. As well, a LIFAC reactor is being installed at
an iron ore pellet factory in the Soviet Union to clean approximately 290,000
scfm of flue gas. The LIFAC process at Richmond Power & Light's Whitewater
Valley Unit #1 in the USA is under construction as part of the Clean Coal III
program to demonstrate the process with high sulfur coal.
PROCESS DESCRIPTION
LIFAC is a combination of limestone injection into the furnace and
post-furnace humidification of the flue gas. when a sorbent such as
limestone is injected into a furnace, some of it combines with flue gas so2,
capturing it in a solid which can be collected in the precipitator. Sorbent
which does not react in the furnace can be reactivated by humidification of
the flue gas as it leaves the air heaters, and further S02 capture occurs.
For convenience, each of the two stages of so, control in this process are
discussed separately. A photograph in Figure 1 shows the arrangement of the
humidification reactor and sorbent silo external to the boiler house.
Furnace sorbent Injection
so, removal in the furnace itself is highly sensitive to thermal conditions
and residence time at the point of injection. In the case of the Poplar
River boiler, the injection conditions were known to be relatively
unfavourable. Gas temperature in the available injection zone is high and
residence times in the sulfation temperature window are short. Additionally,
an earlier test program had found very limited capture of so2 with limestone
in this furnace.
Considering the difficult furnace conditions, extensive modelling was
conducted to determine optimal injection arrangement and to establish
injection equipment parameters for this project. The resultant design
included six furnace injection nozzles and a booster air system to improve
mixing in the furnace. with a Ca/s ratio of 2:1, the model predicted
approximately 24% S02 capture using booster air; 16% capture without booster
air.
Figure 2 shows an outline of the PRPS furnace. The sorbent handling system
is shown in figure 3.
4B-2
-------
Humidi f ic at ion
SOz removal in the humidification section depends upon water droplets
contacting the unused sorbent particles in the ash. The process is therefore
sensitive to the amount of water injected, the life duration of the droplets
and the amount of sorbent available for reactivation. The amount of water
injected and the life time of the droplets are indicated by gas temperature,
and in this paper we will use "approach to saturation temperature" as our
indicator.
To maximize the effectiveness of the LIFAC system, it is operated at the
lowest practical approach to saturation temperature. To avoid potential
downstream corrosion problems, a small portion of hot furnace gas bypasses
the reactor to act as reheat. Another step in maximizing effectiveness is
the recirculation of ash from the precipitator back through the reactor,
thereby giving the sorbent additional opportunities for reaction.
Up to 20% of total boiler ash load can be collected in the reactor and
removed from the bottom hoppers. The actual amount depends upon reactor
outlet temperature.
A system of vibrators prevents accumulation of ash on the reactor walls.
Figure 4 shows the gas side flow diagram for the PRPS LIFAC installation and
figure 5 shows the humidification system with ash handling components.
TEST PROGRAM
The project schedule is summarized below:
July 1989
Start engineering
February 1990
Award first contracts
April 1990
Start on site
May 1990
LIFAC tied in during
Scheduled 3 week overhaul
September 1990
Start commissioning
October to December 1990
Phase I testing
December 1990 to July 1991
Phase II testing
Phase I - Parametric Testing
Immediately after commissioning, a series of testing was undertaken to
determine so2 capture capabilities of the system. This "phase I" testing
investigated the relationships between S02 removal and;
Ca/S ratio
Reactor approach temperature
Boiler Load
Furnace Excess Air
Booster Air
Limestone injection angle
Ash recycle
Furnace temperature
4B-3
-------
The most important parameters overall were found to be Ca/S ratio, reactor
approach temperature and ash recycle. Figure 6 shows the total S02 removal
as well as the furnace and reactor contributions to it as a function of Ca/S
ratio.
As predicted by the furnace modelling, so, capture in the furnace was fairly
limited. The model accurately predicted 16% removal in the furnace at Ca/S
ratio of 2:1 without booster air. The model predicted that removal would
improve to 24% with booster air, that improvement was not realized. No
inherent capture could be detected. Ca/S ratio was the primary variable in
furnace S02 removal, with furnace excess air showing a strong secondary
influence. The furnace S02 removal rate was considerably less than what has
been achieved in other Lifac installations of similar boiler size (Ekman, I.
et al., 1990).
The humidification section contributed by far the largest portion of the S02
capture, with its performance improving as reactor outlet temperature
dropped. The reactor was successfully operated at approach temperatures down
to 4°C. At this operating point, with an injected Ca/S ratio of 2:1, S02
removal across the reactor was 57%. Ash recirculation accounted for
approximately one quarter of that removal.
The resulting total SO removal (furnace & humidification) rate with Ca/S
ratio 2:1 was 63% at 300 MW load. By reducing load to 200 MW, overall
removal at Ca/S ratio of 2 improved to 68%.
Changing furnace parameters other than Ca/S ratio had no measurable effect
on S02 removal in the reactor.
Measurement of S02 throughout the testing was made by a dilution extraction
system which included a heated in-duct probe. Accuracy of these measurements
was confirmed by SaskPower's Technical Services and Research group who
operated an independent extraction type S02 monitor at each measurement
location for approximately 3 weeks.
A process trend of a typical start up during Phase I testing is shown in
figure 7. The effect of starting limestone and humidification can be seen
as quick drops in S02 levels.
Phase II - Operation
During the period from December 1990 to July 1991 the LIFAC system was in
service for over 1500 hours. The operation was not continuous, but depended
upon plant operating conditions, conditions of the LIFAC process equipment,
and time required to organize various tests. The primary objective of this
"Phase 2" testing was to determine impact of the system on the plant
operation, however the effectiveness of different sorbents was also evaluated
during this testing phase.
Impact on plant operations was assessed in terms of LIFAC system forced
outages, manpower requirements and effects on related systems.
Lifac System Forced Outages.
Table I lists forced outages which were experienced by the LIFAC system
during the 1500 hour operational test period. Although a calculated forced
outage rate based upon these values would appear high, analysis of the causes
shows that a mature plant, designed for the high ash coal used by SaskPower,
would be a very reliable system.
The majority of the forced outages (20 out of 27) are associated with the ash
handling system and are a result of design inexperience with high ash coal.
The experience gained with this plant has permitted designers of the system
for SaskPower's new Shand Power Station to select drives and equipment
capacities which more closely match actual conditions and which we are
confident will result in reliable operation at shand.
4B-4
-------
Two of the outages were associated with control fusing, a design error
corrected in the shand design.
The change from hoses to pipe in the injection system has eliminated outages
due to limestone leakage; the remaining outages reflect the need to negotiate
suitable safeguards into the limestone supply contract.
Manpower Requirements.
Maintenance. All maintenance performed during the 1500 hour test period
was arranged through the normal plant maintenance system, and the work
orders were logged and reported accordingly. Analysis of the work
orders shows that the mechanical trades were most affected by the LIFAC
system, with an average of 0.4 manhours per LIFAC operating hour, when
corrected for one event which we feel is not repeatable, projected
mechanical trade loading becomes 0.15 manhours per operating hour.
The mechanical maintenance projection of 0.15 manhours per LIFAC
operating hour does not include replacement of wear components in the
ash system, which for PRPS with high ash loading and very abrasive ash,
is likely to be an annual contract maintenance activity. The projection
also does not reflect improvements which would be seen with upgrading
the interface with the plant ash system as suggested earlier in this
paper.
The other trades together used 0.06 manhours per operating hour, for a
total routine maintenance requirement of 0.21 manhours per operating
hour. At an 80% operating factor, this translates to 1500 manhours per
year.
operating. Records kept for the period 1991 March 12 to 1991 July 2,
show the LIFAC system operated 885 hours, and utilized 173 hours of
operator time. This translates to 1400 manhours per year.
Test supervisor. One supervisor was dedicated to LIFAC operation and
maintenance throughout the phase II testing. Although it is noted that
part of the supervisor's efforts were directed towards testing, data
collection and other non-routine operating activities; his contribution
to the operation is assessed here as equivalent to one full time
position (operator).
Comparison With Industry standard For S02 Removal. Based on the
foregoing, total projected manpower for operating the PRPS LIFAC system
is approximately 4600 manhours per year, or roughly 2.5 positions.
An EPRI report "Impact of FGD Systems" dated July 1991, indicates a
typical staffing level of 4.3 people per 100 MW of scrubber, or 6.5
people for the 150 MW equivalent to the LIFAC system. We feel that this
comparison is not entirely fair to LIFAC, because the labour
requirements will not be strictly proportional to unit size; however
it is clear that the LIFAC system has substantially lower labour
requirements than the industry average for S02 removal equipment.
k.
k.
4B-5
-------
Summary of Phase II Testing
Results and conclusions of Phase II testing are summarized as follows:
-No generating unit outages were caused by the LIFAC system.
- Operation of the LIFAC system usually required no additional operating
or maintenance staff> however there were occasions when call outs were
required.
- Overall, plant staffing compliment would be increased by
approximately 2.5 positions if LIFAC is put into normal service.
- interface with the plant ash system could be improved by use of a
better slurry pumping system. This improvement would significantly
reduce the operator time, maintenance time and forced outage rate.
- The PRPS LIFAC system can be very reliable if the ash system is
upgraded. :
- Variations in the grinding of limestone did not significantly affect
the process.
- Variations in the source of limestone did not significantly affect the
process.
- Calcium Hydroxide was more effective in the furnace, equally effective
in the reactor.
FURTHER WORK
Two outstanding issues regarding the PRPS test program are ash system scaling
and stack opacity.
Ash System Scaling
Poplar River Power station uses a closed loop hydraulic ash disposal system.
The system normally operates using scaling inhibitors and scaling has been
limited to Ash water Recirculation pumps. The pumps typically require acid
cleaning once or twice per year.
scaling of Ash Water pumps escalated with the extensive addition of Calcium
to the ash by LIFAC. Acid cleaning at one point became a daily routine;
however when the ash lagoon reached critical conditions, a spontaneous cold
softening process occurred. The calcium precipitated in the lagoon, leaving
very "soft" water in the hydraulic system and scaling problems stopped. A
series of laboratory jar tests has been initiated to find what operating
conditions would maintain this cold softening mode. Scaling would not be
a consideration at shand or other stations with a dry ash disposal system.
Opacity
Stack gas opacity at Poplar River varies significantly throughout a given
day. The initiation of LIFAC operations coincided with some changes in coal
supply and the effect of LIFAC alone was not directly obvious. Since opacity
was measured only in the two unit common stack, additional opacity monitors
were installed after each of the two ID fans on the test boiler (one fan
draws humidified gas through "A" precipitator, the other draws non humidified
gas through "B" precipitator). These monitors came into operation near the
end of the test program, and a limited amount of dust loading measurement was
used to confirm their operation.
4B-6
-------
On e?ich start up of LIFAC, the meters have shown that putting LIFAC into
service increases opacity in the gas leaving the non humidified "B"
precipitator, while opacity of humidified gas leaving "A" precipitator is
dramatically decreased. When the LIFAC process is stopped, opacity on the
non humidified "B" side rapidly returns to previous values, while recovery
to previous values on the "A" side takes several days.
It is postulated that the sudden drop in ash resistivity due to
humidification resulted in precipitator plates shedding their ash layers, and
that it takes a day or two after stopping the process for the layer to
rebuild and again impair collection. Further investigation is planned,
however it is clear that LIFAC operation does result in a net opacity
improvement on the humidified gas side.
ECONOMICS (SHAND POWER STATION)
Based on the performance of LIFAC at PRPS, SaskPower has committed to
installation of LIFAC on its Shand Power Station which is now under
construction.
The Shand boiler was designed for furnace sorbent injection from the
beginning by the incorporation of a sorbent injection and mixing cavity.
This boiler cavity provides for gas/sorbent contact at optimal temperatures
and is expected to approximately double the furnace contribution to overall
SO, removal when compared to Poplar River. The reactor performance will be
affected by the cavity only in that a smaller quantity of unreacted sorbent
will be carried over to be reactivated.
The coal at Shand contains alkalies which absorb part of the sulfur during
combustion. Considering this inherent absorption along with the initially
low sulfur levels in the coal, removal of only 50% of the sulfur will allow
the plant emissions to meet the federal government guidelines (258 ng SO /J
or 0.6 lb/10 BTU).
Even considering the highly reactive furnace and the low overall removal
requirements, SaskPower's economic modelling showed that installation of one
LIFAC reactor (humidification of 50% of the gas stream) with overall capture
of 75% would provide a significant life cycle cost saving over straight
sorbent injection at shand. Addition of a second reactor did not add
sufficiently to the sorbent savings to offset increases in capital costs.
A summary of the cost comparison between furnace sorbent injection and LIFAC
options is presented in Table II. Approximate values for wet scrubbing are
also included for reference.
CONCLUSIONS
The LIFAC demonstration project at Poplar River Power Station showed that the
process can economically reduce S02 emissions from lignite fired boiler to
meet the Canadian federal government guideline of 258 ng/J.
The 1500 hour operability test showed that the process has a small manpower
requirement compared to an industry standard for FGD.
The installation and operation was carried out without affecting the
availability of the generating unit.
Based on the performance of LIFAC at PRPS, SaskPower has committed to
installation of LIFAC on its Shand Power Station which is now under
construction. The startup of Shand Power Station and the LIFAC system are
scheduled for the summer of 1992.
4B-7
-------
REFERENCES
1. Ekman, I., Asikainen, A., Lepikko, J., Desulphurization with Limestone
Injection Combined with Low-NQjt Combustion. Gen-Upgrade 90 Symposium,
March 6-9 1990, Washington, DC.
2. North American Electric Reliability Council, Impact of FGD Systems. July
1991.
4B-8
-------
Figure 1. The LIFAC reactor at Poplar River Power Station.
STEAM DRUM
58m
ECONOMIZER
EXTERNAL
economizer-
superheater
-------
BOOSTER
AIR FAN
a-
LIMESTONE INJECTION SYSTEM
POPLAR RIVER POWER STATION
BOILER I
LIMESTONE
LIMESTONE
WEIGH BELT
^VARIABLE SPEED
FEEDER
, TRUCK FILL
CONNECTION
BIN BOTTOM
BLOWER
TRANSPORT
1 AIR
Figure 3. Sorbent handling system.
FLUE GAS FLOW DIAGRAM
POPLAR RIVER POWER STATION
BOILER
REHEAT -
CONTROL
DAMPER
E3 INLET
DAMPER
BYPASS
DAMPER \
"I ^eactoTB
hp
STACK
-OUTLET
DAMPER
Figure 4. PRPR LIFAC gas flow diagram.
4B-10
-------
UFAC REACTOR AND AUXILIARIES
POPLAR RIVER POWER STATION
REACTOR
ATOMIZING
AIR
WATER
CT"
FLUE GAS —
FROM BOILER
ASH
CRUSHER
FLUE GAS TO
PRECIPITATOR
DRAG
CONVEYOR
SCREW CONVEYOR
SLURRY
WATER
MAIN
PLANT
ASH
SYSTEM
ASH
SLURRY
= TANK
ASH FROM ESP
Figure 5. Humidification System with ash handling components.
4B-11
-------
S02 Removal in the LIFAC Process
Poplar River Power Station
70-
TOTAL
eo-
*
* ""
s «-
cc
CM
O 30-
C0
FURNACE
20-
10-
0
Z6
3
as
4
as
1
1.6
2
Ca/S Ratio
Figure 6. S02 removal in the LIFAC process as a function of Ca/S ratio.
PRPS LIFAC
November 27,1991
2600"
160
2400-
-140
-120 £¦
S02 Leaving Furnace
2000-
3 1800-
s
cf 1600-
8
-100
-80
Reactor Temperature
-60
1400-
S02 Leaving Reactor
-40
1300-
-20
1000-
tcct
13:15
1336
1336
13:46
13:56
TIME
14:06
14:16
Figure 7. A typical start up during testing.
4B-12
-------
Table I
FORCED OUTAGES DURING 1500 HOUR OPERATION
Number of
Occurrences
Action Taken
Limestone Supply System
Limestone feeder plugged 1
Limestone hose failure 2
Foreign object in feeder 1
Limestone supply interrupted 1
None
Hoses changed to pipe
None
None
Ash Handling System
Conveyor overload - ash
amount exceeding design
Slurry tank plugged
Drop chute gate malfunction
Slurry tank crusher
13 Design values
increased for Shand
2 Crusher installed
3 Gate limit switches
type changed
2
Control System
fuse blown
Organize fuses in
different groups for
Shand
4B-13
-------
Table II
COST FOR CONTROLLING S02 @ SHAND
PRESENT WORTH OF LIFECYCLE COSTS
(MILOONS $CDN)
One LIFAC
Reactor
(CaC03)
Two LIFAC
Reactors
(CaC03)
Sorbent
Injection
Only
(Ca(OH)2)
Wet
Scrubber
Capital Costs
Reactor Cost
$10.0
$19.7
$0.0
$77
Sorbent Storage
0.3
0.3
0.6
incl
Sorbent Processing
0.0
0.0
3.3
incl
Air/Water Supply
0.2
0.4
0
incl
Electrical Supply
0.1
0.15
0.05
incl
Electrical Installation
0.7
1.2
0.2
incl
Foundations
0.1
0.15
0.05
incl
Controls
0.3
0.5
0.1
incl
Subtotal
11.7
22.4
4.3
77.0
Operating Costs
Sorbent Cost
$32.1
$20.9
$53.4
$7.3
Fan Energy
1.2
1.4
0
1.4
Reheat Energy
0
0.6
0
1.2
Labor & Maintenance
3.4
4.7
1.4
17.4
Ash Disposal
1.0
0.6
1.0
5.0
Subtotal
37.7
28.3
55.7
32.3
Total P.W. of Costs
$49.4
$50.7
$60.0
$109.3
Prepared by Saskatchewan Power Corporation
4B-14
-------
1.7 MW Pilot Results for the Duct Injection FGD
Process Using Hydrated Lime Upstream of an ESP
k.
k
4B-15
-------
Intentionally Blank Page
4B-16
-------
Charles A. Brown
Mehdi Maibodi
Lynn M. McGuire
Radian Corporation
2121 Fourth Avenue, #280
Seattle, WA 98121
ABSTRACT
As part of the U.S. Department of Energy's Flue Gas Cleanup program, 1.7 MW pilot
scale tests of the duct injection process for flue gas desulfurization were
conducted at Central Illinois Public Service Company's Meredosia Station. Testing
began in November 1989 and was completed in October 1990. The test results will
I
be used by another DOE contractor to produce a design manual for duct injection
technology.
Under base conditions of 2.0 reagent ratio, 30°F approach-to-adiabatic-saturation
temperature, no recycle, and lime injected upstream of humidification, an overall
system S02 removal of 40% was achieved. When lime was injected downstream of
humidification, overall system S02 removal decreased to 32 percent. The use of
recycle significantly increased S02 removal performance (to 54%) when used with
lime injection upstream of humidification, but had no effect when lime was
injected downstream of humidification. When calcium chloride (CaCl2) was added to
the humidification water and recycle was used, an overall system S02 removal of 72%
was achieved. However, the accumulation of damp deposits on the duct walls
increased with the use of chloride. Also, when CaCl2 was added without the use of
recycle, there was no significant increase in S02 removal performance. Other
process operating variables were studied, including approach-to-adiabatic-
saturation temperature, reagent ratio, recycle ratio, and inlet flue gas
conditions.
Preceding page blank
4B-17
-------
INTRODUCTION
The U.S. Department of Energy's Pittsburgh Energy Technology Center (DOE-PETC) is
sponsoring a Flue Gas Cleanup (FGC) program to promote the use of coal in an
environmentally and economically acceptable manner. One area of activity in the
FGC program is the development of low-cost S02 emissions control technologies that
can be installed on existing coal-fired power plants. A major focus of the FGC
program is on developing duct injection of calcium-based reagent into the flue gas
between the air heater and an existing electrostatic precipitator (ESP). The
goals for this technology are that it be suitable for retrofit to existing boilers
firing medium- to high-sulfur coal, and capable of a minimum of 50% S02 removal at
a cost of less than $500/ton of S02 removed.
Even though duct injection is an outwardly simple process, several key phenomena
are not well understood. Radian Corporation was contracted by DOE-PETC to
investigate the fundamentals of these phenomena in a 1.7 MW pilot plant. The
purpose of this investigation was to obtain a better understanding of the basic
physical and chemical phenomena that control (1) the desulfurization of flue gas
by calcium-based reagent, and (2) the coupling of an existing ESP particulate
collection device to a duct injection process.
Funding for the project was provided by DOE-PETC and the Illinois Department of
Energy and Natural Resources. The Electric Power Research Institute (EPRI) loaned
some equipment to the project, including the pilot ESP. Central Illinois Public
Service Company provided the host site at its Meredosia Station.
This paper includes a discussion of the final S02 removal results and a brief
discussion of the pilot plant operating experience. Final results cover the one
year test period from November 1989 to October 1990. An earlier paper (1)
discussed the background of pilot scale development for the duct injection
process, and described the 1.7 MW pilot plant at Meredosia in detail. ESP
4B-18
A
A
-------
performance results from the 1.7 MW pilot tests are presented in another paper
(2). These test results will be used by United Engineers & Constructors to
produce a design manual that will provide information necessary for technical and
economic evaluation and application of duct injection technology to pre-NSPS coal-
fired utility boilers (3).
PILOT PLANT DESCRIPTION
A simplified process flow diagram for the pilot plant is illustrated in Figure 1.
Flue gas for the pilot plant was obtained from a slipstream of 6300 actual cubic
feet per minute (acfm) withdrawn from the boiler #5 exhaust duct downstream of the
air heater. The boiler is a pulverized-coal, tangentially fired, 180 MW boiler
that fires medium-sulfur (3.2% S), low-chloride (<0.03% CI) coal. The flue gas S02
concentration typically was between 1500 and 2000 parts per million by volume
(ppmv). The flue gas temperature at the inlet of the pilot plant system was
controlled to 300°F for most tests.
Flue gas was treated in a horizontal test duct using dry, powdered, hydrated lime
[Ca(0H)2] injection and water spray from dual-fluid nozzles for humidification.
The test (iuct was 17.5 inches in diameter and provided approximately 1.5 seconds
residence time in the straight section before turning into the ESP. Particulate
matter was removed from treated flue gas in the pilot ESP.
RESULTS
SO, Removal Performance
Baseline Tests. The tests conducted at the Meredosia pilot plant were aimed at
determining how changes to various process parameters influence the ability of the
duct injection process to remove S02. Most tests were compared to a set of tests
at the following baseline conditions:
• Lime upstream of humidification;
• 300°F inlet gas temperature;
• 1800 ppmv inlet S02 concentration;
• 2.0 reagent ratio;
• 30°F approach temperature; and
• No recycle.
k.
4B-19
-------
Reagent ratio is defined as:
Reagent Ratio =
moles Ca(OH), in fresh reagent
moles S02 in flue gas
Percent S02 removal is defined as:
% S02 Removal
moles SO-, in - moles S02 out
moles S02 at the pilot plant inlet
For overall system removal and percent removal in the duct, "moles S02 in" equals
moles S02 at the pilot plant inlet. To calculate percent S02 removal in the ESP,
"moles S02 in" represents moles S02 at the inlet of the ESP. This ESP removal
contribution is calculated by dividing the amount of S02 removed in the ESP by the
system inlet S02 concentration, not by the ESP inlet concentration. Thus, the ESP
contribution can be added directly to the duct contribution to calculate overall
S02 removal.
Average baseline overall system S02 removal for four tests was 40%, with 27%
removal in the duct and 13% in the ESP. Baseline test results are plotted in
Figure 2. Repeatability of results for the baseline tests was very good; the 95%
confidence interval for overall S02 removal for an individual baseline test was ±3
percentage points. However, the goal of 50% S02 removal for this technology was
not achieved at baseline conditions.
System Configuration. The baseline configuration for the system was to inject
lime upstream of the humidification water sprays. The negative effect of
injecting lime downstream of humidification on S02 removal performance is
illustrated in Figure 3. Injecting lime 4 feet downstream of humidification
resulted in a significantly lower overall system S02 removal of 32%, with 24%
removal in the duct and a contribution of 8 percentage points from the ESP.
Injecting lime 20 to 24 feet downstream of humidification produced a similar
overall system S02 removal of 30%, although the split between the contributions
from the duct and the ESP were much different than in the 4-foot downstream case,
being 14% and 16%, respectively.
The lower overall system S02 removal results with the configuration of lime
injected downstream of humidification is attributed to decreased interception and
4B-20
A
A
-------
impaction, or "scavenging", of lime particles by water droplets. The high
velocities of water droplets exiting a two-fluid nozzle are calculated to decrease
rapidly within a few feet, decreasing the velocity differential between droplets
and lime particles. Moisture is required to promote the reaction between S02 and
lime [calcium hydroxide, Ca(0H)2], so that scavenging (interception and inertial
impaction) of lime by water droplets can achieve high S02 removal (4).
Reagent Ratio. As expected, the data plotted in Figure 4 show that increasing the
reagent ratio increases S02 removal performance. However, the goal of 50% overall
system S02 removal was not achieved by increasing only the reagent ratio from
baseline conditions to 2.9 moles calcium per mole S02. Only 44% overall system S02
removal was obtained. At a reagent ratio of 1.0 with other conditions at baseline
values, overall system S02 removal was 25 percent. Thus, incremental improvement
in S02 removal performance was much greater when increasing the reagent ratio from
1.0 to 2.0 than when increasing the reagent ratio from 2.0 to 2.9.
Reagent utilization is defined as:
Reaaent Utilization - mo1es Ca(0H)2 reacted with S02
Reagent Utilization - mo1e Ca(0H)2 in fresh lime
The cost of fresh Ca(0H)2 is one of the major operating costs for the duct
injection process. Therefore, it is desirable to maximize reagent utilization and
minimize the reagent ratio. A small incremental improvement in S02 removal
efficiency does not warrant a large increase in lime consumption. Lime
utilization decreased from 25% to 20% to 16% when increasing reagent ratio from
1.0 to 2.0 to 3.0, respectively.
Approach-to-Adiabatic-Saturation Temperature. The data plotted in Figure 5 show
that decreasing the approach-to-adiabatic-saturation temperature, or approach
temperature, results in improved S02 removal performance. The goal for overall
system S02 removal was achieved by reducing the approach temperature to 20"F while
holding other conditions at baseline levels, producing 52% S02 removal.
Lowering the approach temperature does not significantly affect operating costs.
However, using approach temperatures that are too low will result in significantly
increased operations problems from buildup of duct wall deposits.
4B-21
-------
Recycle. Because reagent utilization usually is low with the duct injection
process, the solids collected in the ESP contain a high fraction of unreacted
Ca(0H)2. A portion of these collected solids can be recycled back to the duct to
provide another opportunity for the Ca(0H)2 to react with S02. Recycling these
solids increases the total Ca(0H)2 content in the system without increasing the
addition rate of fresh lime. Therefore, any increase in S02 removal can be
achieved without increasing the cost of fresh reagent. However, an additional
solids handling system is required and the solids loading at the inlet of the ESP
is increased. Thus, the ESP particulate removal efficiency must be increased to
maintain the same particulate emission rate as before the retrofit of a duct
injection FGD system.
The use of recycle solids produced significantly improved S02 removal performance
when used in the configuration of lime and recycle solids injected upstream of
humidification. Using recycle solids at a ratio of 2.0 pounds recycle solids per
pound of fresh lime at baseline conditions resulted in 56% overall system S02
removal, as shown in Figure 6. Fifty-two percent overall system S02 removal was
achieved at a recycle ratio of 1.0 pounds recycle solids per pound of fresh lime.
I
The benefit of recycle was affected by the configuration of the system. As
illustrated in Figure 7, when solids were injected 20 to 24 feet downstream of
humidification, the use of recycle produced no improvement in S02 removal
performance. This observation may be explained by the hypothesis that
accumulation of calcium sulfite around a shrinking core of lime prevents further
reaction of the unused lime when the material is recycled. It appears that
wetting of recycle solids by scavenging with water droplets releases the available
lime in the recycle solids.
The data presented in Figure 8 illustrate that the incremental improvement in S02
removal performance obtained by adding a fixed amount of recycle solids was
greater at a higher reagent ratio of 2.1 (11 percentage points improvement) than
at a reagent ratio of 1.1 (5 percentage points improvement). At a lower reagent
ratio, more of the fresh lime is utilized in the first pass through the duct,
leaving less unreacted lime in the recycle material.
Also, as illustrated in Figure 9, there was less incremental improvement in S02
removal performance when recycle solids at a ratio of 2.0 pounds recycle per pound
of fresh lime were used at an approach temperature of 20°F (11 percentage points)
4B-22
A
A
-------
than at a an approach temperature of 30°F (15 percentage points). Again, the
explanation for this is that at lower approach temperatures, more fresh lime is
utilized in the first pass leaving less unreacted lime in the recycle material.
Chloride Addition. Calcium chloride (CaCl2) was added to the humidification water
during two tests to evaluate the effects of using a hygroscopic salt. A
hygroscopic salt causes the residual moisture content of the dried solids to be
higher, and the additional moisture allows the reaction with S02 to proceed.
The results from the two chloride addition tests are plotted in Figure 10. Only a
slight increase in S02 removal performance resulted from adding 0.9% CaCl2 to the
water. However, when recycle was used and 3.4% CaCl2 was added to the water,
overall system S02 removal increased dramatically to 72 percent. To increase
residual moisture content of the solids, and therefore the S02 removal efficiency,
the CaCl2 salt should be distributed throughout the solids in the duct. When no
recycle is used, CaCl2 in the humidification water droplets contacts only a
fraction of the lime solids through scavenging of lime particles by water
droplets. When recycle is used, all of the CaCl2 from the dried water droplets is
dispersed throughout the recycled solids, enabling most of the injected solids to
be moistened by the hygroscopic effect.
Unfortunately, adding CaCl2 increased the buildup of wall deposits and the duct
plugged repeatedly after only a very few hours of operation. Also, there were
operation problems from damp deposits on the ESP distributor plate, one ESP
penthouse, and the ESP hoppers. Both the improved S02 removal performance and the
increased operation problems with deposits are attributed to reduced droplet
evaporation rate and increased moisture content of the solids due to the
deliquescent nature of CaCl2. While CaCl2 could serve as a beneficial additive to
improve S02 removal in the duct injection process, more study is required to
determine a chloride addition rate that provides an improvement in S02 removal but
does not cause operation difficulties.
Inlet SO., Concentration. The literature contains conflicting information regarding
the effect of inlet S02 concentration on S02 removal performance in duct injection
systems using separate humidification and lime injection (5 and 6). A possible
explanation for this discrepancy is that the effect is related to the details of
k.
k
4B-23
-------
the process design and configuration, although data interpretation also may be
contributing to the observed difference.
The trend of decreasing S02 removal efficiency with increasing inlet S02
concentration from similar processes, such as spray drying and duct slurry
injection (7, 8, and 9), may not apply to dry injection because of the inherent
difference in the mechanism of contacting lime and water. With slurry injection,
all of the lime particles are contained in water droplets and are wetted. As the
inlet S02 concentration increases, more S02 must be transferred across nearly the
same droplet surface area. The droplet surface area does not change appreciably
because the amount of water required to cool the flue gas does not increase
significantly with S02 concentration. However, with the dry injection process,
lime particles must be scavenged by water droplets to become reactive. As the
inlet S02 concentration increases, more lime is injected to maintain a fixed
reagent ratio, and the increased lime particle concentration in the duct results
in more lime being scavenged by the water droplets.
In this study, two tests were run at markedly different inlet S02 levels. A high
concentration of 2990 ppm was produced by spiking with pure S02, and a low inlet
S02 concentration of 730 ppm was produced by diluting flue gas with ambient air
heated to 300°F. Since dilution with air also lowered the humidity and the
measured wet-bulb temperature of the flue gas during the low-inlet-S02 test, the
inlet flue gas temperature was dropped by 10°F so that the rate of humidification
water addition would be the same as for baseline tests. This eliminated water
addition rate as a variable that could influence S02 removal performance through
scavenging. The data plotted in Figure 11 show that little effect on S02 removal
performance was observed between inlet S02 concentrations of 730 to 2990 ppm.
Inlet Flue Gas Temperature. Operation of the duct injection process at different
inlet temperatures has a noticeable effect on S02 removal, as illustrated in Figure
12. This effect is attributable to the amount of humidification water required at
the different inlet temperatures. With other conditions at baseline levels,
overall system S02 removal increased from 33% to 43% as the inlet temperature was
raised from 260 to 340°F.
Residence Time. Two tests were made at baseline conditions, but the first two
fields of the ESP were turned off inadvertently during these tests. Thus, the
4B-24
A
A
-------
effect of increased residence time for gas/solid contact could be evaluated. The
data plotted in Figure 13 show that the increase in overall system S02 removal was
only 2 to 3 percentage points, which is within experimental error. Therefore,
increased residence time for gas/solid contact does not appear to have a major
effect on S02 removal performance. The more important effect resulting from
residence time in the duct may be more complete drying of wetted solids prior to
impinging on the duct walls at the first bend in the ductwork.
Duct Wall Deposits
The formation of duct wall deposits was a difficult problem to overcome during
most of the pilot plant operation. Several tests were ended prior to the desired
8 hours of data collection because the duct plugged with damp lime and fly ash
deposits on the walls.
Because the inside diameter of the duct was only 17.5 inches, it was easy to wet
the duct walls with the water spray. Nozzles were centered axially in the duct,
leaving a maximum of only 8.75 inches between the nozzle tip and the duct wall.
There wa$ no flexibility at Meredosia for pointing a nozzle away from the wall or
for increasing the spacing between nozzles and the wall. Larger facilities are
likely to utilize a manifold of several nozzles in parallel for humidification,
allowing the outside set of nozzles to be canted inward or to be placed well away
from the wall. It is unlikely that the severe difficulties with wall deposits
encountered at the Meredosia pilot plant would exist at larger facilities, but it
is not known if the difficulties can be avoided altogether.
Wall wetting was detectable by monitoring skin thermocouples installed around the
outside wall of the duct. Using wall temperatures, nozzles could be aligned
accurately to produce uniform high temperatures around the duct wall. Also, by
monitoring wall temperatures, it was determined that wall wetting did not cause a
gradual accumulation of solids deposits over the course of a test. Instead, skin
temperatures often would drop quickly after several hours of testing, indicating
that the deposits formed as a result of an incident or change that had occurred.
An typical example of a sudden drop in duct skin temperature is illustrated in
Figure 14. Sources of sudden changes that could produce wall wetting might
include:
k.
4B-25
-------
• Growth of deposits on the tip of the nozzle, which eventually affected
the water spray pattern;
• Accumulation of large agglomerates of deposits on the floor of the
duct caused by large air-foil shaped buildup that formed and grew on
the back of nozzle and its air and water tubing until it fell to the
floor of the duct;
• Changes in humidification water flow rate when steam sootblowing
started or stopped;
• Fluctuations in humidification water flow rate upon system startup or
when the water filter plugged; and
• Opening the port at the inlet of the test section to take wet-bulb
temperature measurements.
Initially, tests were performed using one nozzle positioned in the center of the
duct, with lime injected either upstream or downstream of the single nozzle. When
lime was injected upstream of the single nozzle, operation was possible for only 1
to 6 hours. Lime injection was moved 4 feet downstream of the humidification
nozzle, but wall deposits plugged the duct quickly in those tests also. When lime
was injected 20 to 24 feet downstream of humidification, operation was extended to
22 hours., Although extended operation could be achieved with lime injected well
downstream of humidification, S02 removal efficiency in this configuration was low.
Humidification with two nozzles staged in series was implemented after it became
clear that wall wetting could not be avoided when using a single nozzle. One
option for utilizing multiple small nozzles would be to arrange them in a parallel
configuration. However, hot-flow physical modeling and computational fluid
dynamics modeling indicated that two nozzles staged in series would produce less
wall wetting than two nozzles in a parallel configuration in the round duct. Use
of the two-stage series configuration was fairly successful and allowed testing
with lime injected upstream of humidification to continue, although problems with
wall wetting and deposits buildup were not eliminated completely.
Several different models of two-fluid nozzles were tried at Meredosia. Prior to
use in the pilot plant, nozzles were screened by measuring droplet size and
velocities on a spray test stand at the University of California at Irvine, and by
measuring unevaporated water flows in a hot-flow physical glass model. Nozzles
that produced the smallest droplet size distributions and a higher degree of
evaporation in the glass model were chosen for use in the pilot plant. In
general, the nozzles that consumed more air produced the smaller droplets.
4B-26
A
A
-------
Although air consumption rates were very high and may be uneconomical in full-
scale facilities, the objective for this study was to humidify the flue gas
without wetting the walls of the smal1-diameter duct to produce test results for
S02 removal performance. Therefore, the high air consumption rates were accepted
for this pilot study. The effort required to solve and optimize humidification in
the small duct probably would not be useful for large-scale facilities.
Although there were some differences in spray patterns produced by the different
nozzles, none of the nozzles could prevent wall wetting and deposit accumulation
in the single-stage configuration, and no significant differences in performance
between Lechler and Delavan nozzles were observed in the two-stage configuration.
Given a wel1-designed, properly sized two-fluid nozzle, other factors leading to
wal1-wetting, such as those described above, appeared to be more important than
the make and model of the nozzle.
CONCLUSIONS
The following conclusions are drawn from the duct injection pilot testing at
Meredosia:
• 40% overall system S02 removal can be achieved at baseline operating
conditions of 30°F approach temperature, 2.0 reagent ratio, 300°F
inlet temperature, with lime injected upstream of humidification, and
without using recycle.
• The ESP provides a significant contribution to overall system S02
removal. The contribution is 13 percentage points at baseline
operating conditions.
• Overall system S02 removal performance is reduced to about 32% when
lime is injected downstream of humidification at baseline conditions.
• Increasing reagent ratio from 2.0 at baseline conditions to 2.9
produces a small increase in overall system S02 removal efficiency, to
44 percent. Therefore, increasing reagent ratio alone does not appear
to be a practical means to achieve the goal of 50% S02 removal.
Decreasing the reagent ratio to 1.0 produces a large reduction in
overall system S02 removal efficiency, to 25 percent.
• Decreasing the approach temperature from 30°F at baseline conditions
to 20°F produces a significant increase in overall system S02 removal,
to 52 percent. Although there is increased potential for buildup of
damp wall deposits when using low approach temperatures, no
significant increase in wall deposits was observed during the 20°F
approach temperature test at the Meredosia pilot plant.
4B-27
-------
• Recycle significantly enhances S02 removal when added upstream of
humidification. At baseline conditions with 2.0 pounds recycle solids
per pound fresh lime, 55% S02 removal was achieved. When recycle and
lime are added downstream of humidification, recycle does not
significantly enhance S02 removal performance.
• Enhancement of S02 removal performance with the addition of recycle is
reduced at low reagent ratios and low approach temperatures. Lime
utilization during the first pass is higher under these conditions,
leaving less unreacted lime available in recycle solids.
• Changing inlet S02 concentration (from 750 ppm to 3000 ppm) does not
significantly affect S02 removal performance.
• Increasing inlet flue gas temperature from 260°F to 340°F increased
overall system S02 removal from 33% to 43% with other conditions at
baseline levels. This is attributed to the increased humidification
water flow rate, which increases wetting of lime particles.
• Adding moderate amounts of CaCl2 to the humidification water (0.9%
CaCl2 in the water) does not significantly enhance S02 removal
efficiency when recycle is not used.
• Adding large amounts of CaCl2 to the humidification water (3.4% CaCl2
in the water) significantly increases overall system S02 removal to 72%
when recycle is used, with other conditions at baseline levels.
• Adding a large amount of gas/solid contact time by turning off inlet
fields of the ESP does not significantly affect S02 removal
performance.
REFERENCES
1. C. A. Brown and L. M. McGuire. "Fundamental Investigation of Duct
Injection/ESP Phenomena: 1.7 MW Pilot Plant." Presented at the 1990 S02
Control Symposium, EPRI/EPA, New Orleans, LA, May 8-11, 1990.
2. M. D. Durham, T. G. Ebner, D. B. Holstein, C. A. Brown, and L. M. McGuire.
"Pilot Plant Investigation of ESP Performance and Upgrade Strategies for In-
Duct Sorbent Injection." Presented at the 9th EPRI/EPA Symposium on the
Transfer and Utilization of Particulate Control Technology, Williamsburg,
VA, October 15-18, 1991.
3. P. A. Ireland and C. E. Martin. "Duct Injection Technology Engineering
Design and Scale-up Criteria." Proceedings of the Sixth Annual Coal
Preparation, Utilization, and Environmental Control Contractors Conference,
Pittsburgh, Pennsylvania, August 6-9, 1990, pp. 352-358.
4. J. C. Kramlich, D. K. Moyeda, G. H. Newton, and R. Payne. "Rate Controlling
processes in humidification for Duct S02 Capture." Proceedings of the First
Combined FGD and Dry SO, Control Symposium, St. Louis, M0, October 1988.
5. M. Babu, J. College, R. Forsythe, R. Herbert, D. Kanary, D. Kerivan, and K.
Lee. 5-MW Toronto HALT Pilot Plant Test Results. DOE/PC/81012-T1-PT.I-A,
NTIS 1415924, December 1988.
4B-28
A
A
-------
6. G. M. Blythe, R. A. Smith, L. J. Muzio, C. A. Martin, and V. V. Bland.
Calcium In.iection Upstream of an Electrostatic Precipitator and a Fabric
Filter for Simultaneous SO, and Particulate Removal: Pilot and Bench-Scale
Results-Draft Final Report. Prepared for Electric Power Research Institute,
Research Project 2784-1, June 1989.
7. Tennessee Valley Authority, Ontario Hydro, Electric Power Research
Institute, Kentucky Energy Cabinet Laboratory. 10-MW Spray Dryer/ESP Pilot
Plant Test Program High-Sulfur Coal Test Phase (Phase III) Final Report.
TVA/0P/ED&T-88/35, July 1988.
8. G. M. Blythe, J. M. Burke, and R. L. Glover. Evaluation of a 2.5-MW Spray
Dryer/Fabric Filter SO-, Removal System. Electric Power Research Institute,
Report CS-3953, Palo Alto, CA May 1985.
9. L. S. Hovis, R. E. Valentine, B. J. Jankura, P. Chu, and J. C. S. Chang.
"E-S0X Pilot Evaluation." Presented at the EPA/EPRI First Combined FGD and
Dry S02 Control Symposium, St. Louis, MO, October 1988.
Venturi
ID Fan
Diverter
Gate ,
K
Recycle
Silo
Lime
Silo
Waste
Silo
S02
leater
Air
Heat
ixchanger
Waste
Solids
Weigh Belt
Feeder
Weigh Belt
Feeder ,
Air
Air
Horizontal Duct
so,
Air
Air
Air
Propane
Flue Gas
Burner
Water
\ A /
ESP
ESP
Figure 1. Simplified Process Flow Diagram for the Meredosia Pilot Plant
4B-29
-------
100%
£»
00
1
w
o
90%
80%
70%
— 60%
CO
>
o
| 50%
O
to
1800 ppm SO 2
300"F Inlet
No Recycle
30 T Approach
# Overall System
¦ Duct Contribution
A ESP Contribution
40%
30%
20%
10%
o% r 111111111111111
95% Confidence Limits
For Individual Test
0.5 1 1.5 2 2.5 3 3.5
Reagent Ratio
Figure 2. Baseline Test S02 Removal
Performance.
100%
90%
80%
70%
_ 60%
CD
>
O
J 50%
O
to
40%
30%
20%
10%
0%
1500-1800 ppm SO2
300*F Inlet
No Recycle
2.0 Reagent Ratio
|| Overall System
Duct Contribution
I I ESP Contribution
20 T Approach
30 T Approach
Ume Lime Ume Ume Lime Lime
Upstream Downstrm Downstrm Upstream Downstrm Downstrm
4 feet 20-24 feet 4 feet 20-24 feet
Figure 3. Effect of System Configuration on S02
Removal Performance.
-------
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Reagent Ratio
Figure 4. Effect of Reagent Ratio on S02
Removal Performance.
1800 ppm SO 2
300"F Inlet
No Recycle
30°F Approach
• Overall System
¦ Duct Contribution
Jk. ESP Contribution
1500-1800 ppm SO j
300*F Inlet
No Recycle
2.0-2.2 Reagent Ratio
# Overall System
¦ Duct Contribution
A ESP Contribution
1 1 ¦ 1 1 .... 1 ¦
1 1 1 1 1 ¦
I .
10 20 30 40 50
Approach Temperature,"F
60
Figure 5. Effect of Approach Temperature on S02
Removal Performance.
-------
100%
1800 ppm SO 2
300'F Inlet
2.1-2.2 Reagent Ratio
30°F Approach
90%
80%
Overall System
Duct Contribution
ESP Contribution
70%
60%
50%
40%
30%
20%
10%
0%
0.0
0.5
1.0
1.5
2.0
2.5
Recycle Ratio
Figure 6. Effect of Recycle Ratio on S02
Removal Performance.
100%
1500-1800 ppm S02
300"F Inlet
30°F Approach
2.0 Reagent Ratio
Overall System
iM Duct Contribution
I I ESP Contribution
Solids
Upstream
0% —
Solids
Downstream
20-24 feet
No 1:1 No 2:1
Recycle Recycle Recycle Recycle
Figure 7. Effect of Humidification
Configuration on Effectiveness of Recycle to
Enhance S02 Removal Performance.
-------
100%
03
i
OJ
w
*
1800 ppm SO 2
90%
-
300"F Inlet
-
30°F Approach
80%
.
Recycle No Recycle
-
# Overall System O
-
1 Duct Contribution ~
70%
A ESP Contribution A
_ 60%
CO
>
o
J 50%
QC
O
U)
40%
30%
20%
10%
0%
1:1 Recycle Ratio
- •
11%
2:1 Recycle Ratio
5%
r;:
0.0
0.5
o'
~
A A
1.0 1.5
Reagent Ratio
~
~ ~
2.0
2.5
Figure 8. Effect of Reagent Ratio on
Effectiveness of Recycle to Enhance S02 Removal
Performance.
100%
90%
80%
70%
- 60%
CO
>
o
J 50%
40%
30%
20%
10%
0% '
11%
1800 ppm S02 2.0-2.2 Reagent Ratio
300°F Inlet 2:1 Recycle Ratio
Recycle No Recycle
# Overall System O
¦ Duct Contribution ~
A ESP Contribution A
o
~
m
10
20 30
Approach, 'F
15%
40
50
Figure 9. Effect of Approach Temperature on the
Ability of Recycle to Enhance S02 Removal
Performance.
-------
100%
90%
80%
70%
- 60%
CD
>
O
| 50%
O
CO
DO
i
OJ
40%
30%
20%
10%
0%
cr
2.0-2.2 Reagent Ratio
300°F Inlet
30°F Approach
1.0:1 Recycle No Recycle
9 Overall System
o
¦ Duct Contribution
~
A ESP Contribution
A
o
95% Confidence
Limits For Individual
Baseline Test
~
A
X
_L
0.0 0.5 1.0 1.5 2.0 2.5 3.0
% CaCI2 in Humidification Water
i i i i
3.5
Figure 10. Effect of Chloride Addition on SO
Removal Performance.
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
2.0-2.2 Reagent Ratio
300"F Inlet
No Recycle
30°F Approach
9 Overall System
¦ Duct Contribution
A ESP Contribution
"95% Confidence
Limits For
"Individual Test
' ' ' 1 ' 1 1 ' ' ' 1 i . , . . i , . . .
500 1,000 1,500 2,000 2,500 3,000 3,500
S02 Concentration, ppm
Figure 11. Effect of Inlet S02 Concentration on
S02 Removal Performance.
-------
ca
I
OJ
Ol
100%
90%
80%
70%
- 60%
(0
>
0
1 50%
O
cn
40%
30%
20%
10%
0%
2.0-2.2 Reagent Ratio
1800 ppm SO 2
No Recycle
30°F Approach
# Overall System
¦ Duct Contribution
A ESP Contribution
95% Confidence
Limits For
Individual Test
a7 •
200 220 240 260 280 300 320 340 360 380 400
Inlet Temperature, "F
Figure 12. Effect of Inlet Flue Gas Temperature
on S02 Removal Performance.
.
-
1800 ppm SO 2
-
300"F Inlet
-
No Recycle
-
30'F Approach
_
Fields 1 & 2
-
Baseline Down
.
# Overall System O
.
¦ Duct Contribution ~
-
A ESP Contribution A
-
95% Confidence Limit!
.
For Individual
-
Baseline Test
~
P
¦
.... I .... I .... I .... I .
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Reagent Ratio
Figure 13. Effect of Increased Residence Time
in ESP Due to Fields Being Out of Service.
-------
feet downstream of nozzle #1
m
SOUTH
NORTH
2 3
test DURATION (hours)
Figure 14. Detect
ion Of „.n Wetting Using SM« Temperature Theses.
4B-36
-------
SCALEUP TESTS AND SUPPORTING RESEARCH FOR THE
DEVELOPMENT OF DUCT INJECTION TECHNOLOGY
L. G. Felix
J. P. Gooch
R. L. Merritt
Southern Research Institute
2000 Ninth Avenue South
P. O. Box 55305
Birmingham, AL 35255-5305
M. G. Klett
J. E. Hunt
A. G. Demian
Gilbert/Commonwealth, Inc.
P. O. Box 1498
Reading, PA 19603
4B-37
k.
-------
Intentionally Blank Page
4B-38
-------
ABSTRACT
The Department of Energy's (DOE) Duct Injection Test Facility (DITF), located in Beverly, Ohio,
at the Muskingum River power plant of Ohio Power Company, has been converted for testing
alternative duct injection technologies. Gilbert/Commonwealth, Inc., and Southern Research
Institute are working on the project under contract from the U.S. DOE. Originally, this facility was
one of the three DOE proof of concept test facilities constructed to evaluate duct injection
technology and was dedicated to rotary atomizer testing. For this project, the capability of
testing dry sorbent injection and slurry injection was added. The technologies under evaluation
include slurry injection of hydrated lime using dual fluid nozzles and pneumatic injection of dry
hydrated lime with flue gas humidification before, at, or after sorbent injection.
The goal of the two year test program is to gain a thorough and detailed understanding of the
scaleup and design requirements, operational data, and limitations of duct injection technology.
The test program has been structured to include a wide range of test parameters, system
components, process configurations, and measurement techniques.
After an initial shakedown period, the test program started in April 1990 and is scheduled to
continue until February 1992. To date, S02 removal data and concurrent ESP performance data
have been obtained during the injection of both dry and slurry-based sorbents. Also, a portable
fabric filter has been used to obtain S02 removal data during testing with dry and slurry-based
sorbent injection.
Preceding page blank
4B-39
-------
BACKGROUND
The Department of Energy's (DOE) Duct Injection Test Facility (DITF), located in Beverly, Ohio,
at the Muskingum River power plant of Ohio Power Company, has been converted for testing
alternative duct injection technologies. Gilbert/Commonwealth, Inc., and Southern Research
Institute are working on the project under contract from the U.S. DOE. Originally, this facility was
one of the three DOE proof of concept test facilities constructed to evaluate duct injection
technology and was dedicated to rotary atomizer testing. For this project, the capability of
testing dry sorbent injection and slurry injection was added. The technologies under evaluation
include slurry injection of hydrated lime using dual fluid nozzles and pneumatic injection of dry
hydrated lime with simultaneous flue gas humidification.
After an initial shakedown period, the test program started in April 1990 and is scheduled to
continue until February 1992. To date, S02 removal data and concurrent ESP performance data
have been obtained during the injection of both dry and slurry-based sorbents. Also, a portable
fabric filter has been used to obtain S02 removal data during testing with dry and slurry-based
sorbent injection.
TEST FACILITY
The DITF operates as a 12 MWe, 50,000 ACFM "slipstream" system on Unit 5 of the Ohio Power
Company's Muskingum River station, in Beverly, Ohio. The slipstream is taken from the existing
air preheater discharge, through ductwork to the DITF pilot electrostatic precipitator (ESP), is
passed to an induced draft fan, and is discharged into the Unit 5 ESP inlet. Test stations are
located at the system inlet, the ESP inlet, and the system outlet. Provision was made to bypass
the pilot ESP and allow gas to flow through a high-efficiency cyclone. These features are shown
in Figure 1, a schematic drawing of the DITF. Considerable flexibility was designed into the
facility to provide the capability of testing over a wide range of process conditions. These
capabilities include the following:
• Inlet gas temperature 275-330 °F.
• Inlet gas S02 concentration of 1100-3200 ppmv.
• Flue gas flow of 16,700-50,000 acfm (gas velocity of 20-60 ft/sec).
• Sorbent addition at a Ca/S ratio of 1.0-2.5.
4B-40
A
A
-------
• Duct residence time of 0.5-3.0 sec.
• Approach to adiabatic saturation temperature as low as 20 0 F or less.
• Recycle of spent sorbent for either slurry or dry sorbent injection.
RESULTS
The results presented below cover test results obtained through August 1,1991. Since testing
began on April 25,1990, three major series of tests have been conducted. During the first test
series, preliminary nozzle testing and in-situ droplet size measurements were performed. The
second test series focused on S02 removal tests with dry sorbent injection at various locations
before, at, and after flue gas humidification. The third test series involves the injection of lime
slurry and lime slurry-recycle ash mixtures. Considerably more tests have been conducted with
slurry injection because much higher S02 removals are possible compared to dry sorbent
injection. S02 removal tests were also conducted with a small portable sidestream fabric filter
(SSFF) during periods of dry sorbent and slurry-based testing.
Nozzle Testing
Wall Wetting Tests. A series of nozzle tests were performed to characterize the behavior of
Lechler Supersonic (LS) and Parker-Hannifin (PH) nozzles for a variety of nozzle configurations
(four or six nozzles), air flow rates, and flue gas velocities. The nozzle tests were carried out
with spray injection in the horizontal test section of the DITF, always with plain water rather than
lime slurry. The primary purpose of these tests was to determine how operating conditions
affected wall wetting by the water spray. For the purposes of these tests, wall wetting was
considered to have occurred if flue gas at the duct wall dropped to 130-135 °F (approximately
5-100 F above adiabatic saturation). To monitor wall temperatures along the horizontal test duct,
thermocouple arrays were placed every 4 ft on the center of each duct wall (top, bottom, left,
and right). Each thermocouple extends approximately 1 in. into the duct so that a gas
temperature at the duct wall is measured rather than the temperature of the duct wall itself.
The nozzles are mounted on two specially designed horizontal lances that are constructed from
Hastelloy™ to minimize corrosion. Each lance can carry up to five nozzles (center with two left
and two right locations); with appropriate mounting adapters, each nozzle can be independently
pointed to the left or right. The nozzle lances can be rotated up or down as desired.
For both the LS and PH nozzles, no lance or nozzle orientation was found that eliminated wall
wetting at approach temperatures of 35 °F or less. Also, for either type of nozzle the four-nozzle
array did not perform well compared to the six-nozzle array. While the LS nozzles are supposed
to reach choked (supersonic) flow at an air pressure of 60 psig, pressures of 80 to 100 psig
were required to minimize wall wetting. The PH nozzles did not seem to be strongly affected
by air pressure. Based on these results either type of nozzle would be suitable for the task of
humidification. However, the PH nozzle is a mechanically complex design whereas the design
of the LS nozzle is quite simple with few parts. Based on the simplicity of the LS design, the
k.
L.
4B-41
-------
decision was made to use LS nozzles for the majority of testing. The LS nozzles are also
smaller than the PH nozzles.
Droplet Size Distribution Measurements. In-situ measurements of droplet size distributions of
water sprays were made at the conclusion of the wall wetting tests. All measurements were
made directly downstream of the nozzle lances at downstream distances corresponding to
evaporation times of 0.1, 0.2, and 0.5 s. All of these tests were run in undiluted flue gas with the
incoming flue gas at approximately 320 0 F. Both LS and PH nozzles were tested at a variety
of water flow rates and air pressures. Because the results of these measurements are detailed
elsewhere(l), only a brief summary of the measurement methodology and our results will be
presented in this paper.
Two droplet sizing techniques were employed. The first was a SRI-developed Video Droplet
Analyzer (VDA). The VDA is an imaging system that provides on-line droplet diameter
measurement based on real-time measurement of the height (diameter) of droplet images
obtained by a high-speed strobe illuminator/video camera combination. The VDA was able to
provide data over a size range of from 25 to 450 //m. The second instrument was an Insitec
PCSV-P (Particle Concentration, Size, and Velocity Probe) which uses single particle light
scattering to determine particle size over two ranges: 0.5 to 2.5 /im and 3 to 40 /im.
For the LS nozzles, Sauter Mean Diameter (SMD) increased with increasing water flow and
decreased with increasing air pressure. Directly downstream of the nozzles log-normal fits to
the mass size distributions showed mass median diameters (MMD's) ranging from 30 to 50 //m
with geometric standard deviations of about 1.8 to 2.0. Downstream of the nozzles our results
indicate that after 0.5 s, about one-fourth to one-third of the water spray was still liquid with the
smaller size droplets suffering the highest depletion. At higher water flow rates (10 gpm) SMD
tended to increase with increasing downstream distance but with lower water flow rates (6 gpm)
SMD first increased than decreased with increasing downstream distance.
For the PH nozzles, SMD also increased with increasing water flow and decreased with
increasing air pressure. Directly downstream of the nozzles log-normal fits to the mass size
distributions showed mass median diameters (MMD's) ranging from 25 to 50 //m with geometric
standard deviations of about 1.7 to 2.0. As was the case for the LS nozzles, the results indicate
that after 0.5 s about one-third of the water spray was still liquid. Regardless of spray water flow
rate, SMD tended to increase with increasing downstream distance.
Duct Injection with an ESP
Dry Sorbent Injection. Dry sorbent refers to the powdered hydrate of lime~Ca(OH)2. All dry
hydrate testing was performed with hydrate purchased from Mississippi Lime Company. Typical
for this material is an available lime content of 94-96%, a BET surface area of 21.9-23.1 m2/g and
an LOI of 24.5%.
4B-42
A
A
-------
Four types of tests were conducted. In the first type of test, dry hydrate was injected without
any simultaneous attempt to humidify the flue gas. In the second type of test, dry hydrate was
injected downstream of the water spray nozzles. This is the non-scavenging mode where water
droplets are allowed to evaporate before they reach the point where the dry sorbent is injected
so that no interception of sorbent particles by water droplets can occur. In the third type of test,
dry hydrate was injected approximately 0.5 s of residence time upstream of the water spray
nozzles. This is the scavenging mode where the sorbent can disperse and react with S03
present in the flue gas (to give removal of sulfuric acid vapor, which may be beneficial if ESP
operation is adversely affected by space charge effects) before humidification. After
humidification, the sorbent and water move together down the duct with collisions occurring until
the water has evaporated. In the fourth and final type of test, dry hydrate was injected midway
between the upper and lower nozzle lances so that as the stream of dry hydrate dispersed it
had to pass through the nozzle spray. This variation of the scavenging mode, called coplanar
injection, is believed to provide better contact of sorbent and water. Most dry sorbent testing
was conducted using coplanar sorbent injection.
Tests using the first three modes of sorbent injection were conducted during July and August
of 1990 and the results of these tests have been reported elsewhere(l). Table 1 presents a
summary of these tests and shows the results of tests during which the highest sorbent
utilizations were achieved for each mode of sorbent injection. Tests conducted with coplanar
injection were conducted during May, June, and July of this year. The results of those tests are
shown in Table 2. All of the results shown in these tables were obtained with LS nozzles used
for humidification. These data suggest that S02 removal is maximized with coplanar injection.
However, as will be shown below, much higher S02 removals are possible with slurry injection.
Slurry Sorbent Injection. Considerably more time was devoted to slurry-based tests than to dry
sorbent injection. This is because much higher S02 removals were obtained with slurry injection
compared to dry sorbent injection and because essentially no wall wetting was observed with
slurry injection even at low approach temperatures (25 °F). Table 3 summarizes the main S02
removal data obtained during this testing. For more detail the reader is referred to the second
Topical Report on this project(1).
The conditions for maximum S02 removal that are shown in this table occur at an approach
temperature of 20-30 °F and a Ca/S ratio of 2.5. These conditions indicate removal of 70% of
the S02 at the ESP inlet and 85% of the S02 at the ESP outlet. These S02 removals are based
solely on gas-phase removal data and represent the upper limits of results to be expected, for
the conditions listed.
During the studies with slurry injection, extensive work was performed to determine whether the
extent of S02 removal found routinely by direct measurement was reliable. The work consisted
of:
1. Performance of "spike tests" in which S02 was injected ahead of the heated filters
in the in-situ probes through which samples are withdrawn for the gas sampling
system, and
k.
4B-43
-------
2.
Single-point collection of solid sorbent samples in an extractive probe designed to
quench the S02-sorbent reaction, followed by determination of the utilization of
calcium in the collected samples.
The spike tests indicated satisfactory performance of the gas sampling system, but the quench
probes indicated consistently lower calcium utilizations at the ESP inlet than those calculated
from gas-phase results. Figure 2 shows S02 removal as a function of Ca/S ratio as an area,
with the upper boundary based on gas-phase data and the lower boundary of the included area
determined by the quench probe solids samples. Two possible reasons for the disagreement
between the gas-phase and solids results at the ESP inlet are: 1) wall effects and 2) under-
representation of reentrained agglomerates of more highly utilized sorbent in the single-point
solids samples. This ambiguity is expected to be resolved with long-term tests which are
planned for later in this project. For the present, however, a conservative approach is to
estimate S02 removal at the ESP inlet of a full-scale plant from the analysis of solids samples.
Also, the DITF has a higher surface-to-volume ratio than does a full-size plant, so wall effects
would be proportionally greater at the smaller scale.
With respect estimating S02 removal across a full-scale ESP, S02 removal levels measured
across the ESP at the DITF should be duplicated in a full-scale ESP of the same relative size (in
terms of specific collecting area). Thus, a conservative estimate of overall S02 removal in a
full-scale installation can be obtained by adding the S02 removal measured across the ESP
(based on gas-phase data) to S02 removal in the duct (calculated from solids analysis). For
approaches in the range of 20 to 30 °F this method of estimation yields the following results:
Ca/S
Ratio
Duct
1.0
35
2.0
53
2.5
60
S02 Removal, %
ESP Total
15 50
16 69
17 77
Slurry Sorbent Injection with Recycle Ash. The process of recycling rather than discarding used
sorbent/ash solids as a means of increasing calcium utilization is common practice at spray
dryer installations in the utility industry. Recycling spent solids is especially important in the
economics of duct injection because slurry droplet size and in-duct residence times tend to
lower calcium utilization when compared to conventional spray dryers. However, encouraging
results have been obtained from our initial experiments with recycle ash/slurry mixtures. If the
recycle ratio, R, is expressed as the ratio of the mass of recycle ash solids to the mass of
pebble lime feed, R values ranging from 1.0 to 5.5 were investigated in single-pass experiments
in which a slurry containing both fresh Ca(OH)2 and recycle solids was injected (the recycle
solids were collected as the waste material from an earlier experiment with the fresh lime only).
These are called "single-pass" experiments because the recycled solids are not collected and
reintroduced into the slurry as would be the case in a continuous operation. Table 4
summarizes the results of these recycle experiments.
4B-44
A
-------
Solids collected from the ESP inlet hopper should provide a sample that is representative of
most of the total collected solids. A comparison of the utilizations based on ESP outlet gas
compositions and solids samples collected from the ESP inlet hopper is provided by the last two
columns in Table 4. In view of the fact that these were relatively short-term tests, agreement
between the gas and solid phase data is reasonably good.
To compare sorbent utilization with and without the addition of recycle ash, it is necessary to
know how utilization of Ca(OH)2 in recycle ash compares to the utilization of Ca(OH)2 in fresh
lime slurry. The available data are shown in Figure 3. The utilizations plotted on the vertical axis
in this figure are based on ratios of S02 removal at the ESP outlet to the total Ca(OH)2 available
to react (the sum of Ca(OH)2 in fresh lime slurry and Ca(OH)2 in the recycle ash, if present).
The Ca/S ratios on the horizontal axis in the figure are similarly based on the rate of injection of
Ca(OH)2 from fresh lime slurry and the rate of addition of Ca(OH)2 from recycle ash, if available.
There is considerable scatter in these data from any simple algebraic relationship. In principle,
part of the scatter might be attributed to variations in the approach temperature (here the range
was 23 to 40 °F), but the points found most distant from the others were not attributable to
exceptional approach temperatures. Our tentative conclusion is that, at a given value of the
Ca/S ratio, sorbent utilization is essentially the same, with or without the addition of recycle ash.
An example comparing operation of the DITF with and without recycle is as follows:
Under conditions that typically prevail at the DITF, the injection of a slurry of fresh
Ca(OH)2 at a Ca/S ratio of 1.0, without recycle, will cause an S02 removal of 50%,
as shown in Figure 3. In this instance, the waste sorbent will obviously have a
utilization of 50%.
Under the same conditions, except for a recycle ratio of 2.5, the injection of a
combination of fresh Ca(OH)2 and recycled Ca(OH)2 will produce an overall Ca/S
ratio of 1.20, assuming that the utilization of sorbent in each form remains at 50%.
However, the removal of S02 will increase from 50.0% to 59.9%, and the utilization
of sorbent in the discharged waste will also increase to 59.9%. Thus, for a given
expenditure of lime, recycle would increase the tonnage of S02 removed by the
factor of 0.599/0.500, or about 1.2.
If a fabric filter is used as the particulate control device, subsequent data will show that
significantly higher calcium utilizations are achieved with the fabric filter at the same approach
temperature compared with those measured with an ESP. In addition, higher recycle ratios
could be used without encountering the particulate emission limitations associated with small
ESP installations. An experimental effort is planned near the end of the current project which
includes the use of: 1) recycle on a continuous basis and 2) a sidestream device which
simulates pulse-jet fabric filter operation. This process configuration is expected to optimize
calcium utilization.
ESP Performance. A reliable evaluation of ESP performance with sorbent injection requires
several days of relatively stable process conditions. This time is required to allow an equilibrium
dust layer to accumulate that is representative of the process condition under study. To date
k.
k.
4B-45
-------
emphasis has been placed on parametric testing with relatively short test periods. Although
these conditions are not ideal for ESP performance testing, some useful information has been
obtained. Representative operating data for periods of both slurry and dry sorbent injection are
shown in Tables 5 through 7.
As Tables 5 and 6 show, with slurry injection, mass collection efficiencies usually averaged
averaged 99.9% or better and voltage-current relationships were usually normal, as would be
expected with a low-resistivity dust. As Table 6 shows, operating points near the operating
maximum for each TR set were routinely achieved. Occasional problems were experienced with
shorting due to buildups of low-resistivity dust which caused sparkover at low voltages.
Laboratory measurement of the resistivity of a sample of slurry ash suggests that for approach
temperatures of 500 F or less the resistivity of the ash/sorbent mixture is less than 109 ohm-cm.
As Tables 5 and 7 show, at approach temperatures of greater than 35 °F, dry sorbent injection
led to a degradation of the ESP electrical operating conditions. At a 45 °F approach, severe
corona current suppression was observed in the inlet field, to the point where corona onset and
sparkover voltages were separated by only 2 kV. Outlet emissions were also increased by a
factor of six. Laboratory resistivity measurements are planned for ash/sorbent samples from
these dry sorbent injection tests.
Duct Injection with a Fabric Filter
S02 removal tests were performed with a portable sidestream fabric filter (SSFF) that was
designed and fabricated by SRI for the U.S. DOE(2). This device is an updated and redesigned
version of a portable SSFF that SRI designed and built for the Electric Power Research Institute
to simulate the operation of a reverse-gas cleaned baghouse(3). Figure 4 shows a schematic
drawing of the SSFF. This device exposes a sample of filter fabric (0.67 ft2) to hot flue gas that
has been conveyed to the SSFF from a nearby process stream. The fabric is located in a
heated enclosure maintained at flue conditions. The SSFF has been used in the past to evaluate
flue gas conditioning and to screen potential filtration fabrics at full-scale utility baghouses(2,3,4).
Operation of the SSFF is controlled by a dedicated interactive data acquisition system (DAS).
For operation at the DITF the SSFF was modified to simulate pulse-jet cleaning. This
modification consisted of an external reservoir that was pressurized with compressed air and
exhausted through a quick-acting solenoid valve. A pipe directed the cleaning pulse so that it
was tangential to the clean side of the fabric. In order to preserve the dynamics of pulse-jet
cleaning a special insert was fabricated to fill most of the open area behind the fabric. For these
tests, a needle-felted Ryton(TM) fabric was used, typical of the fabrics used in full-scale pulse-jet
cleaned baghouses.
The SSFF was set up at the inlet to the ESP, approximately 1.5 s of residence time downstream
of the point of slurry injection. A sampling probe was located in the center of the duct and fitted
with a nozzle sized to convey an isokinetic sample to the SSFF when it was operated at an
air-to-cloth ratio of 4.0 ft/min. The sampling probe was heat-traced and in-situ thermocouples
were located at the inlet of the SSFF and near the surface of the fabric. When the SSFF was
4B-46
A
A
-------
operated at temperatures lower than the duct, the sample probe was set to operate at the lower
temperature, and the SSFF sample enclosure was maintained at the same temperature. For
these tests the SSFF was operated at temperatures from 0 to 45 °F above the adiabatic
saturation temperature.
Because the SSFF uses a single swatch of fabric, the fabric must be cleaned in a way that is
not representative of a full-scale baghouse where only a small portion of the bags (fabric area)
are pulsed during any one period of cleaning. Also, in a full-scale pulse-jet-cleaned baghouse,
cleaning is usually pressure-drop initiated (for example, at 5.0 in. H20), and cleaning pulses
continue until the pressure drop across the baghouse reaches some predetermined level (for
example, 4.0 in. H20). To obtain an average pressure drop across the SSFF between 4.0 and
5.0 in. H20, cleaning was initiated when the pressure drop across the fabric reached 9.0 in. H20.
Because the pressure drop across the SSFF fabric usually averaged 1.0 in. H20 after cleaning,
this yielded a time average-pressure drop of about 5 in. H20.
The SSFF was cleaned with two pulses (of 50 ms duration) 10 s apart. In laboratory
measurements, similar pulses have been measured to impart fabric accelerations of about 400 g,
typical of full-scale pulse-jet-cleaned baghouses. Pulse timing and duration were controlled by
the dedicated DAS.
During initial testing it was found that the dust cake accumulated during a filtering cycle was
always easily removed during cleaning. Indeed, a stoppage of flow through the fabric was
generally sufficient to reduce the pressure drop across the fabric to 1 in. H20 or less.
Throughout the entire test series, the pulse cleaning method described above maintained an
after cleaning pressure drop of approximately 1.0 in. H20.
Two major test series were conducted, corresponding to tests with slurry injection and tests with
dry sorbent injection. These tests were performed for a variety of Ca/S ratios (1.1 to 2.3) and
for approach to saturation temperatures (in the horizontal duct, at the ESP inlet) of 25, 35, and
45 °F. As indicated above, during many of these tests the SSFF was operated at an approach
temperature lower than that of the duct.
Whenever possible, ash samples were removed from the SSFF hopper for subsequent chemical
analysis to determine sorbent utilization for comparison with that obtained from gas-phase S02
removal data. As Figure 5 shows, for both dry injection and for slurry injection, there was usually
good agreement between the two methods of determining sorbent utilization.
Figure 6 shows data from the SSFF during tests with slurry injection at a Ca/S of 2.0 and
approach temperatures (in the duct) of 45 and 250 F. For the tests shown in Figure 6, the SSFF
was first operated for a period at duct conditions and then was cooled, for successive filtering
cycles, down to a 10 °F approach. This figure documents the pressure drop across the SSFF
fabric, the fraction of S02 removed across the SSFF, and the fraction of S02 removal due to the
duct and the SSFF, all at duct approach temperatures of 45 and 25 °F and SSFF approach
temperatures of 25, 15, and 10 °F. This figure shows that when the approach temperature
4B-47
-------
within the duct is held constant and the approach temperature within the SSFF is lowered, S02
removal (and sorbent utilization) is significantly increased. The constancy of the slope of the
pressure drop versus time curve (the same as drag versus weight because air-to-cloth ratio is
constant) in Figure 6 suggests that the filtration properties of the sorbent/ash mixture may be
independent of the approach temperature. Furthermore, because the slope of the A P versus
time curve appears to be constant throughout a filtering period, even at the high mass loadings
that are typical of duct injection, the dust cake formed in this process appears to be very
porous. Indeed, during this testing the temperature within the SSFF was reduced to near the
local adiabatic saturation temperature (during periods of both slurry and dry injection) without
causing an adverse effect on the filtering properties of the dust cake deposited on the fabric.
No condensation of water was observed. This behavior coincides with anecdotal reports of dry
FGD systems with fabric filters that have suffered excursions below the water dew point without
blinding or otherwise irreversibly damaging the filter bags.
Figure 7 summarizes the results of tests where the duct was held at a constant 250 F approach
and total S02 removal across the SSFF (averaged over the filtration period) was measured for
approach temperatures (within the SSFF) of 25 and 10 °F for a range of Ca/S ratios from
approximately 1.15 to 2.2. Also shown in this figure are S02 removals measured at the ESP inlet
and at the ESP exit. This figure and Figure 6 suggest two general observations: First, a fabric
filter can remove substantially more S02 than can be removed across an ESP under the same
conditions. Second, when a fabric filter is cooled (without additional humidification) below duct
conditions, much more S02 is removed than is possible with a fabric filter that is maintained at
duct conditions. These data indicate that a secondary cooling step, without the addition of
water, can reactivate nearly dry sorbent and significantly increase sorbent utilization. Unlike a
conventional spray dryer, long residence times are not required before the sorbent/ash aerosol
enters the fabric filter.
Figures 8 and 9 show total S02 removal across the SSFF (averaged over the filtration period)
as a function of the approach temperature within the SSFF for various Ca/S ratios. Figure 8
shows average S02 removal data for a 250 F approach in the duct for three Ca/S ratios. Figure
9 shows average S02 removal data for 25 and 45 °F approaches in the duct at a Ca/S ratio of
1.75 to 1.8. Figure 8 also shows total S02 removal measured across the ESP during these
tests. These data indicate that for Ca/S ratios of 1.75 or greater and a 25 0 F approach in the
duct, removal of essentially all of the S02 in the flue gas is possible if the SSFF is operated
within 10 0 F of the adiabatic temperature. Removal of almost 90% of the S02 in the SSFF was
achieved at a 45 °F duct approach and with a 10 °F approach in the SSFF. Additional
experiments with a larger-scale fabric filter and heat exchanger are needed to determine whether
these results can be achieved with a full-scale baghouse.
Finally, Figure 10 shows the results of similar tests conducted during dry sorbent injection at a
Ca/S ratio of 2.0 and at approach temperatures of 35 and 45 °F in the duct. As might be
expected, these data show that S02 removal levels are much lower with dry sorbent injection.
Inspection of Figure 10 shows that with the duct at a 45 °F approach, S02 removal in the duct
with slurry injection is equivalent to dry injection with the SSFF operated at a 10 °F approach.
4B-48
A
A
-------
The process of adding a temperature-controlled fabric filter particulate collection device to
enhance the efficiency of dry scrubbing systems for the control of sulfur oxides is a new
development. Therefore, we have acted to secure patent protection for this process under the
name of Dry Sorbent Reactivation.
MAJOR OPERATION AND MAINTENANCE EXPERIENCE
From startup of the DITF on April 25,1990, through July 31,1991, the DITF has completed 72
hours of nozzle testing, 471 hours of testing with dry hydrated lime injection, and 984 hours of
testing with slurry injection. This includes a three month outage by the Ohio Power Company
in the fall of 1990 to rebuild the Unit 5 boiler. During this time the DITF has experienced a
number of small problems typical of pilot plant operation. However, some other operation and
maintenance issues have emerged that may be more typical of full-scale operation. These
issues are discussed below.
Drv Injection System
Initially, dry sorbent was injected through a single 4-inch beveled pipe into the center of the
duct, approximately 2 ft upstream of the spray humidification nozzles. For recent testing, the
injection nozzle was extended so that the injection point is now in the same plane as that of the
humidification nozzles. This change has produced higher S02 removals (perhaps because of
better water-sorbent contact) and made it possible to operate for indefinite periods at an
approach temperature of 35 0 F at the ESP inlet. With the previous arrangement duct deposits
were a persistent problem at a 35 °F approach.
Slurry Injection System
Lechler Supersonic nozzles have been used for slurry injection since slurry testing began in
August 1990. The nozzles have required minimum maintenance as they are designed with
erosion-resistant silicon carbide inserts. There has been no detectable wear to date on these
inserts. The key to maintaining consistent nozzle performance has been adequate flushing when
shutting down plus periodic inspection and acid cleaning.
A major problem with slurry injection at the DITF occurs when dilute slurry solutions are used
(i.e., less than 20% solids). With such dilute slurries significant duct wall deposits can occur
even at moderate approach temperatures. Duct deposits are minimized only when using
concentrated slurries (20% solids or more). Extended tests of up to 120 hours and approaches
as close as 25 °F have been run with concentrated slurries without generating significant
deposits. In general, for similar test conditions, duct deposits appeared to be less significant
with slurry injection than with dry injection.
Duct Cleaning System
The duct cleaning system consists of a sonic horn, three soot blowers and duct hoppers 10 to
18 ft downstream of the slurry injection points, upstream of the first duct hopper. Since dry
k.
4B-49
-------
deposits are seldom found on the roof or sides of the duct, performance of the sonic horn is
difficult to evaluate. The soot blowers appear to function efficiently only when the duct deposits
are dry. Because duct deposits were found to start on the far end lip of the first duct hopper,
one of the soot blowers was relocated downstream of the duct hopper configured to blow
against the gas flow. This appeared to inhibit the formation of solids deposits on the hopper
lip. Additional soot bbwers were installed between the first and second duct hoppers to
minimize the formation of solids deposits in this area. Deposits can be controlled by these
systems at conditions that minimize deposition. However, at conditions which promote
deposition (such as dilute slurry injection at low approach conditions), these systems do not
control solids buildup in the duct. On these occasions, the DITF eventually must be shut down
and the horizontal test duct cleaned out before testing can be resumed.
ACKNOWLEDGEMENTS
This work was performed under U.S. DOE Contract No. DE-AC22-88PC88851. The authors
would like to thank the Project Officers, Dr. Soung Kim and Dr. Richard Tischer of the DOE
Pittsburgh Energy Technology Center, for their aid and support in carrying out this research.
The authors would also like to acknowledge the support personnel at the DITF and personnel
at SRI who helped direct and plan this work, particularly Dr. Edward Dismukes and Mr. Charles
Lindsey. The assistance of the SRI on-site staff, Mr. Terry Hammond and Mr. Randy Hinton, is
acknowledged with appreciation.
REFERENCES
1. L. G. Felix, E. B. Dismukes, J. P. Gooch, M. G. Klett, and A. G. Demian. "Scaleup Tests
and Supporting Research for the Development of Duct Injection Technology". Topical
Report No. 2, Task 3.1: Evaluation of System Performance. DOE Contract
DE-AC22-88PC88851, in review.
2. T. R. Snyder, P. V. Bush, and M. S. Robinson. Characterization and Modification of
Particulate Properties to Enhance Filtration Performance - Final Report. U.S. Department
of Energy Report No. DOE/PC/88868-T5; SRI-ENV-90-577-6666 (NTIS Accession No.
DE91004164/XAB), Southern Research Institute, Birmingham, AL, June 1990, 155p.
3. L. G. Felix and R. L Merritt. "Fabric Screening Studies for Utility Baghouse Applications."
In Proceedings: The Fifth Symposium on the Transfer and Utilization of Control Device
Technology, Volume 3. U.S. EPA/EPRI Report CS-1835-6, Kansas City, MO, August 1984.
4. L. G. Felix, R. L. Merritt, and K. S. Duncan. "Improving Baghouse Operation at the
Monticello Station." JAPCA, Vol. 36, No. 9, September 1986, pp.1075-1085.
4B-50
A
A
-------
Table 1
Summary of Most Favorable SO2 Removal Results
Obtained During Dry Hydrate Testing with Lechler Nozzles
Mode of Sorbent
Injection
No Humidification
Non-Scavenging
Scavenging
Ca/S
Approach
Inlet SO2
SO2 Removal, %
Ratio
r°n
(Dpnrt
ESP Inlet ESP Outlet
2.32
158
1350
12 17
3.15
26
1950
27 37
2.50
30
1900
42 53
Table 2
Summary of SO2 Removal Results Obtained
with Coplanar Injection (Lechler Nozzles).*
Approach
Ca/S
SO2 Removal, %
m
Ratio
ESP Inlet
ESP Outlet
25
0.95
24
30
1.44
33
36
1.97
41
47
45
1.00
13
18
1.53
19
24
2.03
23
32
* 2000 ppm of SO2 at the inlet.
Table 3
Summary of Data on SO2 Removal with Slurry*
Approach Ca/S SO2 Removal, %
(°F> Ratio ESP Inlet ESP Outlet
20-30 1.0 45 50
2.0 60 75
2.5 70 85
50 - 55 1.0 30 40
2.0 50 60
2.5 60 70
* Inlet SO2 levels from 1200 to 2800 ppm. SO2 removal
appeared to be unaffected by inlet SO2 concentration.
4B-51
-------
Table 4
Summary of SO2 Removal Results
Obtained During Recycle Ash/Slurry Testing
(Lechler Nozzles)
Ca/S Ratio
Fresh Slurry &
Slurrv Recvcle Ash
Recycle
Ratio
Approach
S02
Removal (%)
ESP Outlet
Utilization, % —
Gas Phase Solids,
ESP Outlet Inlet He
0.62
1.35
4.7
33
58
43
49
0.64
1.03
2.2
39
55
53
44
0.69
1.72
5.5
26
67
39
46
0.73
1.55
3.5
28
70
45
46
0.80
1.43
3.7
37
62
43
46
0.81
1.38
3.3
35
56
41
0.85
1.29
2.0
37
56
43
37
0.86
1.59
3.3
24
73
46
51
0.97
1.66
2.0
29
64
38
35
1.15
1.44
1.0
36
58
40
39
ESP
Table 5
ESP Operation During Slurry and
Coplanar Dry Sorbent Injection*
Sorbent Ca/S
Type Ratio
Approach
Temp.m
35
45
35
45
SCA
ffl2/kacfm)
Slurry 2
Dry 2
* 2000 ppm of SO2 at system inlet
402
417
336
345
Mass Eff.
Outlet
Inlet Flow
(%)
fqr/dscf)
fscfm}
99.96
0.0040
31660
99.95
0.0040
31257
99.94
0.0060
36750
99.61
0.0411
36150
Table 6
ESP Operating Points During
Slurry Injection
Approach
(°F)
35
45
Inlet Field
fk\rt (nfiJcmZ)
51.8
45.9
37
36
Second Field
(M) lnA/cm2)
42.6
42.6
60
60
Third Field
(k\ft (nA/cm2^
46.1
45.0
82
78
Outlet Field
(KV) fnA/cm2)
44.7 87
44.7 89
4B-52
A
A
-------
Table 7
ESP Operating Points During
Coplanar Dry Sorbent Injection
Approach
(°F)
35
45
Inlet Field
(k\A (nA/crr^
Second Field
(M) (nA/cm2)
71.3
60.0
6
<1
53.0
52.4
28
33
Third Field
(k\fl (nA/ctr^
52.3
45.1
51
39
Outlet Field
IkV) (nfiJcm2)
50.3 64
42.3 53
^0.
iO
VERTICAL
TEST
SECTION
INOUCED
DRAFT FAN
AIR
DILUTION SYSTEM
Figure 1. Schematic Drawing of the Duct Injection Test Facility.
4B-53
-------
100
90
80
70
5? 60
5
E 50
a>
0C
CM
S 40
30
20
10
20 - 30°F Approach
50 - 55°F Approach
Ca/S Ratio
Figure 2. Upper and Lower Limits of SO2 Removal at the ESP Inlet
Based on Gas-Phase and Solids Analyses.
«
N
60
55
50
25
.~
' O
= 45
c
ID
¦e 40
o
w
35
30
~ Recycle
o Fresh Slurry
~ ~ ~o
~ *
~ ~
o 0
—I I ¦ I ¦ 1 I 1 I I I 1 I I 1 1 1 1 L_
1.00 1.20 1.40 1.60 1.80 2.00
Ca/S Ratio
Figure 3. Comparison of Sorbent Utilization Based on Gas-Phase Data from Slurry
Injection Tests with and without the Addition of Recycle Ash.
4B-54
A
-------
VIEWPORTS
EXHAUST
OVEN
SAMPLE PROBE
VACUUM PUMP
POWER
REVERSE AIR
HEAT EXCHANGER-
FABRIC SAMPLE
HOLDER
CONDENSER/COOLER
THERMOCOUPLES
VIEWPORT
POWER LINE
o 0,
SAMPLE
-REMOTE ASH
HOPPER
REVERSE FLOW
FABRIC FILTER
SAMPLER CONTROL*
Figure 4. Schematic Diagram of the Sidestream Fabric Filter Device.
en
"en
CO
c
<
co
;g
o
OT
E
E
C
o
CO
N
70
60
50
40
30
20
10
~J—I—J—J—3—J J—J—1—|—1—I—1—1—J—1—I—I—I—|—1—I—1—I—1—I—1—1—I—1—I—I—r
¦ ~ i ¦ ' ' » ' « ' i i i i I i ~ i »
o Slurry
~ Dry
J i i i i I i i i i L
0 10 20 30 40 50 60 70
Utilization from Gas Phase Data, %
Figure 5. Comparison of Sorbent Utilizations Determined from Gas-Phase Data and
from Chemical Analyses of SSFF Hopper Samples. Data from Tests
Conducted with Dry Sorbent Injection and with Slurry Injection.
k.
4B-55
-------
SSFF TESTS - CA/S = 2.0
Inlet SO2 = 2000 ppm, Duct Approach = 45°F
Total
SSFI
AP
25°F Approach
15°F Approach
10°F Approach
¦—r~
18
10
1 1 1
11
I l~
12
-r~
13
i
14
1 I 1
15
-r-
16
Time of Day (March 12,1991)
-r~
17
19
SSFF TESTS - Ca/S = 2.0
Inlet SO2 =2000 ppm, Duct Approach = 25°F
25°F Approach
15 F Approach
10 F Approach
Time of Day (March 13,1991)
Figure 6. SSFF Data Obtained During Slurry Tests in which the Approach
Temperature within the SSFF was Lowered from 25 to 10°F in Two Steps.
Data Shown for Duct Approach Temperatures of 45 and 25°F.
4B-56
A
A
-------
40
30
20
~ SSFF, 10°F Approach
a SSFF, 25°F Approach
a ESP, 25° F Approach
o DUCT, 25°F Approach
' ¦ ' i i i
1.0
1.2
1.4
1.6 1.8
Ca/S Ratio
2.0
2.2
2.4
Figure 7. SO2 Removal as a Function of Ca/S Ratio for a Duct Approach
Temperature of 25°F and SSFF Approach Temperatures of 25 and 10°F.
Data Obtained During Slurry Injection Testing.
100
CO
B
E
o
CO
>
<
80
65
25°F Approach in Duct
a Ca/S = 2.10
• Ca/S = 1.75
¦ Ca/S = 1.42
5 10 15 20 25
SSFF Approach Temperature, °F
30
Figure 8. SO2 Removal Averaged over a Filtration Period in the SSFF as a Function
of the Approach Temperature within the SSFF for Three Ca/S Ratios at an
Approach Temperature of 25°F in the Duct. ESP SO2 Removal Data Also
Shown. Data Obtained During Slurry Injection Tests.
k.
k
4B-57
-------
100
95
13
g
E
®
cc
w
O
to
®
o>
2
®
90
85
80
70
65
75 — Ca/S = 1.75-1.80
a 45°F Approach in Duct
• 25°F Approach in Duct
10
15
20
25
30
SSFF Approach Temperature, °F
Figure 9.
SO2 Removal Averaged over a Filtration Period in the SSFF as a Function
of the Approach Temperature within the SSFF for Approach Temperatures
in the Duct of 25°F and 45 °F. Data Obtained During Slurry Injection
Testing.
TB
>
o
E
-------
A Pilot Demonstration of the Moving Bed Limestone
Emission Control (LEC) Process
4B-59
L
-------
Intentionally Blank Page
4B-60
-------
Michael E. Prudich
Shailendra N. Reddy*
Ohio University
Ohio Coal Research Center
Athens, OH 45701
Kenneth W. Appell
John D. McKenna
ETS, Inc.
1401 Municipal Road NW
Roanoke, VA 24012
ABSTRACT
The Limestone Emission Control (LEC) process is a unique system employing a moving bed
of standard quarry-sized limestone to remove S02 from coal-fired boiler flue gases.
The LEC system has the potential to enable boiler operators to burn high-sulfur coal
in both an economically and environmentally acceptable manner.
This paper describes an ongoing program funded by the Ohio Coal Development Office in
which a 2 MW LEC system has been installed and is currently being operated and tested
on the slip stream of the Ohio University coal-fired steam plant located in Athens,
Ohio.
The LEC process has demonstrated the capability of extremely high performance for SOz
removal (>90%) at a cost that is favorable when compared with wet scrubbing and spray
drying technologies. This paper presents a detailed description of the LEC process,
test results to date, and system economics.
k.
Mr. S.N. Reddy is currently with ICF, Inc.,9300 Lee Highway, Fairfax, VA, 22031.
Preceding page blank
4B-61
-------
INTRODUCTION
PROCESS DESCRIPTION
The Limestone Emission Control (LEC) process is a unique system employing standard
quarry-sized limestone to remove S02 from coal-fired boiler flue gases. In the LEC
process, hot flue gases (<350°F) are contacted with a bed of 1/32 to 1/4 inch limestone
granules covered with a thin film of water. Sulfur dioxide is absorbed from the flue
gas into the water film where it subsequently reacts with dissolved limestone. A
layer of reaction products, primarily calcium sulfate and calcium sulfite, forms on
the surface of the limestone granules as the reaction proceeds.
PROCESS DEVELOPMENT
In the early 1970s, the U.S. Department of Energy initiated bench-scale testing of dry
limestone for flue gas desulfurization. Shale and Stewart (1) reported that S02
removal efficiencies of greater than 90% for extended periods of time were possible,
and that the limestone granules were regenerated easily with mild agitation. This
early work resulted in U.S. Patent No. 3,976,747 entitled "Modified Dry Limestone
Process for Control of Sulfur Dioxide Emissions" (2).
ETS, Inc. began development work on the present LEC system in mid-1982. In June 1983,
ETS and the U.S. EPA agreed that the LEC was potentially a simple, low cost FGD
system. It was agreed to proceed along three paths: (1) ETS performed a series of
bench-scale slip stream tests to verify the basic S02 emission reduction. Results
showed greater than 90% S02 removal; (2) EPA/IERL requested that TVA do a preliminary
economic analysis of the LEC system. The resulting analysis was in agreement with
preliminary ETS economic studies showing that the LEC process offers significant
economic advantage over both spray drying and wet scrubbing systems; and (3) EPA/IERL
initiated an in-house laboratory bench-scale study of the LEC process aimed at
verification of the ETS slip stream tests.
Although Step 3, EPA/IERL's in-house work, was not formally published, ETS arrived at
two significant conclusions based on the EPA bench-scale testing program. First,
EPA's bench-scale tests using simulated flue gases adequately substantiated ETS' slip
stream test results and Shale's early work at DOE. And second, although regeneration
of the spent limestone was demonstrated, the regeneration was not accomplished as
easily as that reported by Shale and Stewart, i.e., more vigorous grinding was
required.
In the fall of 1986 the Ohio Coal Development Office awarded a research grant to Ohio
University to design, install and operate a small, semi-batch fixed-bed LEC pilot
plant (400 ACFM) on a slip stream of flue gas from the 70,000 lb/hr stoker boiler
providing steam to the Ohio University campus. Figure 1 shows a simplified flow
schematic of the fixed-bed LEC pilot unit. The flue gas first entered a spray chamber
where it was conditioned to the desired temperature and humidity. The conditioning
was achieved by adding water via an atomizing nozzle and/or by injecting live steam
into the chamber. The conditioned gas then entered the LEC reactor, passing downward
through a fixed-bed of limestone. The reactor contained a removable basket filled
with limestone. The basket was 22 inches square and permitted the limestone bed depth
4B-62
A
A
-------
(the bed dimension in the direction of flue gas flow) to be varied between 0 and 24
inches. The scrubbed flue gases were then drawn out from the bottom of the limestone
bed by an induced draft fan and returned to the duct leading to the stack.
Over 100 experimental runs were conducted on the LEC fixed-bed pilot unit between May
28, 1987 and December 18, 1987 (3,4,5). Three different Ohio limestones covering the
full range of Ohio limestone compositions were used in these tests. The program
studied the effects of varying limestone bed depth (from 6 to 18 inches), superficial
flue gas velocity (from 1.0 to 2.0 feet per second), flue gas humidity, flue gas S02
concentration, and flue gas temperature. All three of the limestones used in the
study were able to achieve greater than 90% S02 removal in the LEC process. Many tests
showed removals of up to 99%.
The fixed-bed LEC unit was shown to perform effectively for inlet S02 concentrations
in the flue gas ranging from 500 ppm to 3500 ppm. At an inlet S02 concentration of 700
ppm, a 6 inch deep LEC bed was able to sustain removals of over 90% for 16 hours and
removal of over 50% for greater than 24 hours (Figure 2). Even at 3500 ppm S02, the
6 inch LEC bed (the smallest bed tested) was able to sustain an S02 removal efficiency
of greater than 90% for about 1 hour and a removal efficiency greater than 50% for
more than 2 hours.
Two factors were identified which limit LEC operation. These were (1) drying of the
limestone surface and (2) build-up of the reaction product on the reactive limestone
surface. A dry limestone surface is effectively non-reactive under LEC processing
conditions. The build-up of reaction product on the limestone surface results in a
decrease in the reaction rate and eventually renders the limestone inactive. Two
operational modes were used to overcome deactivation by drying. Operating the process
with a saturated flue gas would halt evaporation and thereby prevent the limestone bed
from drying out. An overbed water spray could also be used to keep the bed wet.
Deactivation due to surface blinding cannot be avoided but it can be overcome through
the removal of the reaction products. In the small pilot plant studies the reaction
products were removed from the reacted limestone through off-line attrition in a
stirred ball mill. After attritting the stone it was returned to service. The
reactivated stone showed the same S02 removal activity as the virgin limestone.
This paper describes initial results for a larger (5000 ACFM), continuous moving-bed
LEC pilot plant that is currently being operated at Ohio University. A detailed
description of the moving-bed LEC facility is given below.
PROJECT OBJECTIVES
The primary goal of the current study is the demonstration of the techno/economic
capability of the LEC system as a post-combustion FGD process capable of use in both
existing and future coal-fired boiler facilities burning high-sulfur coal. The
specific objectives of the study include:
• Demonstrate the LEC capability of operating at high inlet S02
concentrations typical of those encountered in the burning of
a 3.5 wt% sulfur coal.
l.
k
4B-63
-------
• Generate long-term operational information that would provide
optimization parameters for a full-scale demonstration
project.
• Develop correlations for on-stream time and performance
changes for both S02 removal and particulate removal
efficiency.
• Demonstrate limestone regeneration techniques.
• Analyze both the waste material and its disposal/resale
options.
• Provide an economic analysis/comparison of the LEC process
with respect to current FGD systems.
SYSTEM DESCRIPTION
A continuous moving-bed LEC pilot plant sized to handle 5000 ACFM has been installed
on the slip stream of the 70,000 lb/hr stoker boiler providing steam to the Ohio
University campus. A simplified flow scheme of the moving-bed LEC process is shown
in Figure 3. A detailed schematic of the pilot plant is given in Figure 4. Limestone
is added to the bed via a 32 foot bucket elevator which is supplied with limestone
from the feed hopper, the make-up hopper, the vibrating screen, or the recycle chute
by means of a feed screw. The make-up hopper is used to compensate for limestone lost
due to reaction with S02 and it's subsequent removal as reaction product. The
limestone bed itself has dimensions of 14 inches X 36 inches X 128 inches (Figure 5).
The depth of the bed in the direction of the flue gas flow is 14 inches. The LEC
reactor internals consist of inlet and outlet louvers, overhead bed sprays and outlet
screening. The inlet and outlet louvers are four inches long and are inclined at a
75° angle with a 3/8 inch overlap between louvers. The flue gas inlet and outlet
plenums on the sides of the reactor are sized such that there is uniform distribution
of the gas across the whole bed. The flue gas is drawn from the duct leading to the
stack from the electrostatic precipitator. The superficial velocity of the flue gas
through the LEC reactor can be varied from 0.5 to 1.6 feet per second. The live
hopper at the top of the bed is provided so as to prevent the leakage of ambient air
into the outlet gas plenum. An induced draft fan is used to draw the processed flue
gas from the outlet plenum. The processed flue gas is directed to the stack.
Limestone moves vertically downward through the bed with a velocity sufficient to
prevent bridging of the limestone which may occur when the flue gas dries up the bed.
The limestone removed from the bottom of the bed is sent to the discharge screw via
a 19 foot bucket elevator. The discharge screw has three outlets, any one of which
may be opened. The first outlet directs the limestone to the recycle chute which
leads directly to the feed screw and back to the reactor. The second outlet leads to
the reactivation system. The build-up of reaction products on the surface of the
limestone tends to retard the rate of reaction per unit limestone surface area. To
prevent this surface blinding from rendering the limestone inactive, the limestone
granules are passed through a reactivation device that removes the reaction product
4B-64
4
A
-------
layer by either dry or wet abrasion, exposing a fresh layer of reactive limestone.
This mixture of reactivated limestone and abraded material is then passed through a
vibrating screen were the abraded material is removed as a waste. The reactivated
limestone is then directed to the feed screw and returned to the top of the LEC bed.
The third discharge screw outlet is directed to the waste hopper. Normal LEC
operation directs the limestone through the reactivation system.
The S02 concentration in the inlet flue gas can be varied from 500 to 3000 ppm. Early
experimental runs using this pilot plant have shown that the LEC process has a
capability of removing more than 90% of the S02 contained in the inlet flue gas. The
LEC process is considered to be potentially attractive for enabling coal-fired power
plants to meet stringent S02 emission standards.
EXPERIMENTAL RESULTS AND DISCUSSION
A total of 30 experimental runs were conducted between May 29, 1991 and July 25, 1991
using the LEC moving-bed pilot unit. The length of each experimental run ranged from
30 minutes to 290 minutes. The primary variables studied were:
• Superficial flue gas velocity - 0.5, 1.1, and 1.6 ft/s.
• Limestone bed velocity - 4.4 ft/hr, 13.7 ft/hr, and 21.6
ft/hr (these limestone bed velocities correspond to in-bed
limestone residence times of 2.4 hr, 0.8 hr, and 0.5 hr).
• Inlet S02 concentration - 500 to 1100 ppm.
• Water addition rate - 0 to 2.0 gallons/minute (includes both
overbed sprays and flue gas humidification water).
As mentioned earlier, the S02/CaC03 reaction is a liquid-phase reaction and therefore
a thin layer of water must be present on the limestone granules to allow significant
S02 removal to occur. The run philosophy to date has been to operate the LEC bed so
that the limestone exiting the bed is essentially dry. This is accomplished by
controlling the position of the drying front that exists in the limestone bed (Figure
6 (6)). As the limestone approaches the bottom of the bed, the drying front moves
closer to the outlet side of the reactor. This results in there being less available
wet limestone for S02 removal. Although this results in relatively poor S02 removal
values in the lower portion of the bed, the overall average bed removal remains high
because of the excellent performance of the top and middle portions of the bed.
The results obtained from the experimental runs performed to date have demonstrated
S02 removals in excess of 90% and, often, S02 removals in excess of 99%. Performance
plots for three experimental trials (910723A, 910724A, and 910725A; all with inlet S02
concentrations of about 1000 ppm) are shown in Figures 7, 8, and 9. These Figures
also show the influence of the mechanical difficulties that have prevented us from
achieving long-term operation. The outlet design of the LEC moving-bed pilot plant
relied only on the outlet louver to retain the limestone in the moving bed. As it has
turned out, the original outlet louver design did not accomplish it's purpose. During
k.
k
4B-65
-------
operations, limestone was drawn from the bed into the outlet plenum and subsequently
into the I.D. fan. In order to prevent this limestone loss from the bed and to
protect the I.D. fan, a screen was attached to the inside (limestone bed side) of the
outlet louvers. The screen now keeps the limestone in the bed. However, the presence
of the screen has introduced another operational problem. Damp fines now are
deposited on the screen, plugging the screen and significantly increasing the pressure
drop across the limestone bed. The blinding of the screen occurs most quickly at the
top of the bed (normally the most active part of the bed). Due to the screen blinding
at the top of the bed, the flue gas begins to flow preferentially through the bottom
portions of the bed (normally the least active part of the bed). This action results
in a simultaneous increase in pressure drop across the bed, decrease in S02 removal
performance, and decrease in flue gas flow rate. This behavior can be observed in
Figures 7, 8, and 9. The outlet louvers and outlet plenum have been redesigned.
Modifications for these systems will be made during an August/September shutdown.
It has been observed that significant limestone surface abrasion occurs within the LEC
bed itself. In traveling downward through the LEC bed, the limestone particles seem
to provide most of the granule on granule abrasion necessary to reactivate the
limestone. If this effect turns out to be significant, the external reactivating
attritor (equipment item #5 in Figure 4) may be eliminated from the process flow
sheet. The degree and effectiveness of this self-reactivation will be examined and
quantified during the next phase of LEC experimentation.
ECONOMICS
The LEC process and conventional limestone scrubbing have been compared on an
equatable basis using flue gas conditions that would be expected at the outlet of the
electrostatic precipitator (ESP) on a 500 MW coal-fired power plant. The equipment
list for the LEC case was obtained by conceptually scaling up the current LEC moving-
bed pilot plant at Ohio University. The equipment list for the wet limestone scrubber
case was obtained from an EPRI technical report (7). Equipment costs for both the LEC
and wet limestone scrubber were generally obtained using empirical estimating charts.
The LEC was found to have a definite economic advantage in both direct capital costs
and operating costs. The LEC equipment capital cost of $12,290,000 represented a 48%
savings when compared to the wet limestone scrubber process equipment cost of
$23,534,000. On an absorption tower basis, a capital cost advantage of 46% for the
LEC occurs due to its simpler mechanical design. Operating costs are also smaller for
the LEC process since a slurry is not required, resulting in a lower water
consumption. The solids material handling associated with the LEC requires less
electricity than that of the slurry handling system required by the wet limestone
scrubber. Table 1 lists the operating and equipment costs associated with each
process area for both FGD systems along with the total annualized costs. A similar
economic analysis has been performed by TVA (8).
4B-66
A
A
-------
CONCLUSIONS
The LEC moving-bed pilot plant project has demonstrated its capability as a high-
performance S02 removal system. S02 removals as high as 99.9% have been achieved for
periods of well over one hour at inlet flue gas flow rates of up to 3200 acfm. This
shows that the chemistry behind this road-grade limestone-based technology is valid.
Mechanical difficulties, i.e., limestone blow-over and exit screen plugging have
prevented long-term LEC operations. Outlet louver modifications made over the
August/September shutdown should eliminate these mechanical problems. Experimental
runs scheduled for after the shutdown include high S02 operations, reactivation
studies, and long-term operational studies.
The next step in the commercialization of the LEC process is the demonstration of the
technology on an electric utility boiler as either a stand-alone desulfurization
device or as an in-situ insert into the last field of an existing ESP. A conceptual
schematic of a generic compact moving-bed LEC is shown in Figure 10. A conceptual
schematic of a compact moving-bed LEC inserted into the space occupied by the last
field of an ESP is shown in Figure 11.
ACKNOWLEDGEMENTS
The authors would like to acknowledge the financial support of the Ohio Coal
Development Office under grants CDO/R-86-24 and CDO/D-88-49.
k
k
4B-67
-------
REFERENCES
1. C. C. Shale and G. W. Stewart. "A New Technique for Dry Removal of SOj, "
presented at the Second Symposium on the Transfer and Utilization of Particulate
Control Technology." Denver, CO. July 23-27, 1979.
2.- C. C. Shale, "Modified Dry Limestone Process for Control of Sulfur Dioxide
Emissions." U.S. Patent 3,976,747. 1976.
3. M. E. Prudich, K. W. Appell, M. J. Visneski, J. D. McKenna, D. A. Furlong, J.
C. Mycock, J. F. Szalay, and J. E. Wright. Small Pilot Plant Demonstration of
ETS' Limestone Emission Control System. Vols. 1 & 2. Final Report: OCDO Grant
No. CDO/R-86-24, May 1988.
4. K. W. Appell. A Mathematical Simulation of ETS' Limestone Emission Control
Process Using the Method of Characteristics; Fixed-Bed Configuration/Gas-Phase
Mass Transfer Control. M.S. Thesis. Ohio University, November 1989.
5. M. J. Visneski. Modeling of the Low Temperature Reaction of Sulfur Dioxide and
Limestone Using a Three Resistance Film Theory Instantaneous Reaction Model.
Ph.D. Dissertation. Ohio University, March 1991.
6. S. N. Reddy. A Mathematical Simulation of ETS' Limestone Emission Control (LEC)
Process Using a Moving Bed Configuration. M.S. Thesis. Ohio University, August
1991.
7. R. J. Keeth, M. J. Krajewski, P. A. Ireland. Economic Evaluation of FGD
Systems. Volume 5: The NOXSO and SOXAL Sodium-Based Processes and Four
Additional Calcium-Based Processes. CS-3342-V5. Palo Alto, California:
Electric Power Research Institute, October 1986.
8. An Economic Evaluation of the ETS Limestone Emission Control Process for Removal
of Sulfur Dioxide from Power Plant Flue Gas. Report prepared by the Tennessee
Valley Authority for the U.S. Environmental Protection Agency, 1983.
4B-68
-------
TO
STACK
ID FAR
<3
FLUE
GASES
OVERHEAD
SPRAYS
±±±
n
ATOMIZING
NOZZLE
STEAM
IHJECTOKS
.. LIMESTONE
BED
REACTOR
DRAIN
Figure 1. Simplified Schematic of the Fixed-Bed
LEC Process.
871216B
871214B
¦+->
c
-------
WET
LIMESTONE
BED
FLUE
GAS
1
DESULFURIZED
FLUE
GAS
I
DRY
LIMESTONE
BED
Figure 3. Simplified Schematic of the LEC Process.
4B-70
A
A
-------
r
r
03
I
LEGEND
1-Feed Hopper
2-Waste Hopper
3-Ha)ce-up Hopper
4-Screen
5-Fluid izer
6-Bucket Elevator
7-Bucket Elevator
8-Inlet Gas Plenum
9-0utlet Gas Plenum
10-Reactor
11-ID Fan
12-Bed Twin Screws
13-Live Hopper
14-Inlet Gas
15-0utlet Gas
16-Reactor Screw
17-Discharge Screw
18-Feed Screw
19-Vaste Screw
Figure 4 : Schematic of the Moving Bed LEC Pilot Plant.
-------
LIMESTONE
Figure 5. Bed Geometry of Moving-Bed LEC Pilot Plant.
4B-72
-------
1 DO -
CD
O
CD
Q.
90-
O
>
O
E
O
E
-------
OUTLET S02
CONCENTRATION
WET ZONE
DRY ZONE
0 7 14
BED DEPTH (inches)
Figure 6. Example drying front and outlet S02
concentration profiles.
100-
90-
CL
« 80^
"6
>
o
£ 70-1
0>
Cd
CN
O 60-
GO
50-
S02 Removal
Intet Go a Flow
Run No. 910723A
f Delta P
o
.o Csl
-oo
Q_
Q
X!
-cm
-------
mmm
Figure 10. Schematic of Conceptual Compact LEC Bed.
(a) Top view. (b) Cross-sectional face
view. Arrows indicate flue gas flows.
Shading indicates limestone bed. Flue
gas is distributed and collected through
louvered plenums.
Traveling
Tripper
ESP body
Flu©
Gas
Inlet
Flue
•GK3
Outlet
LiiMStona
Outlet
Conveyor
Figure 11. Conceptual Schematic of an LEC Occupying the
Space of the Last Field in a Electrostatic
Precipitator.
k.
4B-75
-------
TABLE 1
SUMMARY OF RESULTS OF A COMPARATIVE ECONOMIC ANALYSIS
OF AN LEC AND A WET LIMESTONE SCRUBBER
FOR A 500 MW FACILITY
RAW MATERIAL AND UTILITY CONSUMPTION LEC LSS
Limestone - tons/hr 32.7 33.5
Quicklime - tons/hr 0 2.5
Water - gpm 249 710
Steam - lb/hr 222,000 222,000
Operating Power - hp 8,769 13,058
Waste - tons/hr 86 130
EQUIPMENT COSTS - S1000 (19891
Area 1 - Material Handling 1,316 1,486
Area 2 - Feed Preparation 0 1,805
Area 3 - Gas Handling 3,897 3,326
Area 4 - S02 Absorption 3,716 11,690
Area 5 - Stack Gas Reheat 2,524 2,524
Area 6 - Waste Handling/Reactivation 837 2 .703
Total Equipment Costs 12,290 23,534
ANNUALIZED COSTS - S1000 (19891
Total Operating and Maintenance 24,445 34,349
Total Costs 27,872 40,601
4B-76
A
A
-------
Pilot Plant Support for ADVACATE/MDI
Commercialization
4B-77
k
-------
4B-78
-------
Charles B. Sedman
Michael A. Maxwell
U.S. Environmental Protection Agency
Air and Energy Engineering Research Laboratory
Research Triangle Park, NC 27711
Brent Hall
Acurex Corporation
P.O. Box 13109
Research Triangle Park, NC 27709
ABSTRACT
The sorbent chemistry and process development of the ADVAnced SiliCATE
(ADVACATE)/Moist Dust Injection (MDI) process are discussed for power plant
applications. Recycle and milling have proved essential for integrating
sorbent preparation into the overall process concept. Simultaneous flue ga
humidification and sulfur dioxide (SOo) absorption are achieved within 1
second residence time by integrating ADVACATE with the Flakt MDI concept.
Recent pilot plant efforts to optimize and integrate process unit operation
are outlined. Field evaluation at the 10 MWe scale by Tennessee Valley
Authority (TVA) and ABB-Flakt is scheduled for early 1992.
This paper has been reviewed in accordance with the U.S. Environmental
Protection Agency's peer and administrative review policies and approved fo
presentation. The contents of this article should not be construed to
represent Agency policy nor does mention of trade names or commercial
products constitute endorsement or recommendation for use.
Preceding page blank
-------
In response to the perceived need for a low-cost sulfur dioxide (SO2)
control process, easily retrofitted on existing U.S. utility boilers, the
U.S. EPA's Air and Energy Engineering Research Laboratory has fostered the
development of a lime-based, duct injection process called the ADVAnced
SiliCATE (ADVACATE) process. Through research agreements with the
University of Texas, contracts with Acurex Corporation, and a cooperative
research and development agreement with ABB-Flakt, the ADVACATE process has
evolved to a simple add-on technology with minimal disruption to the
existing facility.
The heart of the ADVACATE process is the on-line production of a calcium
silicate gel from lime and dust collector product which has high surface
area, thin layers of free lime [Ca(OH)2], and substantial moisture, allowing
for simultaneous in-duct absorption of acid gases and flue gas cooling. By
merging the ADVACATE sorbent with Flakt Moist Dust Injection (MDI)
technology, the pitfalls of duct deposits, low calcium utilization, and
maintenance of nozzles for humidification are avoided.
Within the next 10 months, ABB-Flakt and the Tennessee Valley Authority
(TVA) will conduct a 10 MW0 evaluation of the ADVACATE/MDI process; the
remainder of this paper focuses on the evolution of ADVACATE and on research
leading to, and in support of, the planned field evaluaton.
BACKGROUND
Much of the published work regarding lime/silica reactions has application
in the production of cement products and consists of relatively low
temperature (~25°C) studies. Earlier work by Iler [1] showed that the
solubility of silica in water:
(Si02)x + 2 H20 > (Si02)x_1 + Si (OH)4
was increased from about 200 ppmw at a pH of 7 and 22°C to nearly 1500 ppmw
at a pH of 10 and 200°C. Iler [2] also demonstrated that the presence of
alumina dramatically decreased the dissolution rate and solubility of
silica. Greenberg [3] found that the dissolution rate of silica is highly
dependent upon the surface area of the solid silica. Several subsequent
investigations [4, 5, 6] have identified different reaction products—
silicate hydrate gels at T <100°C and silicate hydrate crystals at T
>100°C—depending on calcium ion concentration and slurry temperature.
For work specifically using fly ash as the silica source, Jozewicz and
Rochelle [7] first reported the presence of high surface area materials in
solids precipitated out of lime/fly ash mixtures, and that these solids were
considerably more reactive than lime toward S02. Peterson [8] extended the
work of Jozewicz and Rochelle, concluding that:
• Mechanically attrited fly ash showed an increased
dissolution rate of silica, proportional to the fly
ash surface area; alumina dissolution rate was
unaffected.
• High calcium fly ash contains more soluble silica than
low and medium calcium content ashes.
• Calcium aluminates are unreactive toward S02.
• Calcium silicates are very reactive toward S02.
• The presence of sulfates (contained in recycled
solids) depresses the calcium aluminum reaction,
4B-80
-------
leading to an increased rate of calcium silicate
formation.
Jozewicz, et al. [9] conducted further efforts in increasing silica
dissolution (and sorbent reactivity) concluding that:
• Operation above 100°C via pressure hydration
dramatically decreased time for optimum SC>2
reactivity.
• Above 160°C the reactivity toward SC>2 decreased,
presumably due to the formation of crystalline calcium
silicate hydrates.
PROCESS DEVELOPMENT
Earlier work by Jozewicz, et al. [10] indicated that the most promising
process concept consisted of:
(1) preparing the sorbent as a slurry of water, fresh lime, and
recycle solids,
(2) filtering the sorbent into a moist cake, and
(3) mixing the cake (at 50-60% moisture) with about twice that amount
of dry recycle solids to produce a free-flowing sorbent of -20%
moisture.
Subsequently sorbent was produced using a pressure hydrator and cement mixer
for two field evaluations reported previously. In one the LIMB process at
Edgewater was improved on a 1.0 MWe slipstream from a base 65% SO2 removal
at calcium-to-sulfur-ratios (Ca/S) of 2.0 to 99% SO2 removal with silicate
enhancement at recycle ratios of 2.0 [11]. The other was performed on a 0.4
MWe slipstream using Flakt MDI technology and showed in-duct and baghouse
SO? removals of 80 and 95+%, respectively, at Ca/S ratios of slightly less
than 1.0 [12] .
With EPA working with ABB-Flakt, the process has evolved in the following
ways:
(1) The pressure hydrator has been supplanted by two slurry tanks in
series with a vertical mill, and
(2) The slurry is no longer filtered; rather, the slurry is mixed in
a paddle-type or pug mill device with direct discharge into
ductwork. The resulting process has been named ADVACATE/MDI and
is discussed further in the last section of this paper.
CURRENT RESEARCH
In support of anticipated commercial application, focus has been on improved
unit operations and integration of these operations into a continuous
process. The milling of fly ash has allowed production of higher surface
area sorbent, resulting in better reactivity toward SOo and the ability to
carry more water and flow freely. Figure 1 shows the bench-scale reactivity
improvement and Figure 2 the subsequent improvement in sorbent utilization
on an 85 Nm /min pilot facility at Research Triangle Park. Typical sorbent
production with milling and atmospheric (90°C) slurrying now requires 2-3
hours.
Mixing of slurry with recycle solids has become a focal point of research.
Recently a series of runs using a Sprout-Bauer mixer were performed which
4B-81
-------
showed acceptable operation at up to 44% moisture (Table 1). An AshTech
mixer is currently being evaluated under similar process conditions. Both
mixers are extensively used for wetting fly ash or mixing slurries and dry
solids for disposal.
As reported at the Ninth Particulate Control Symposium, pilot-scale data on
electrostatic precipitator (ESP) performance with high loadings of ADVACATE
showed surprisingly good performance [13]. Tables 2 and 3 show the impacts
of increased ESP loadings with fly ash and ADVACATE sorbent, respectively.
Figure 3 illustrates that ADVACATE does not increase submicron particle
loading with increased sorbent feed rate, while submicron particles increase
linearly with increased fly ash injection rate. It is hypothesized that
ADVACATE sorbent tends to agglomerate in-duct prior to entering the ESP, and
does not show the characteristic suppression of corona current of increased
fly ash loading. The results were modeled using ESPVI Version 4.0 for ESPs
of 40 to 80 m /rrr/s size (200 to 400 SCA in nonmetric units) . Figure 4
suggests that addition of ADVACATE sorbent will have no impact at the higher
SCA value, but may cause a modest increase in dust emissions at the lower
value.
COMMERCIAL DEVELOPMENT
On November 22, 1991, TVA and ABB-Flakt signed an agreement to cooperatively
evaluate the ADVACATE/MDI process on a 10 MWe prototype system under
construction at TVA's Shawnee Station in Paducah, KY. Previously in 1991,
ABB-Flakt was granted an exclusive, worldwide license to market the ADVACATE
technology using fly ash, and TVA received a government license for use of
ADVACATE within the TVA system.
The 10 MWe system at Shawnee, depicted in Figure 5, is the system envisioned
for commercial application. The duct will be fitted with a paddle mixer and
an existing ESP will undergo isolation of the first field hopper for
providing recycle solids. Approximately 25% of the recycle solids will be
mixed with fresh lime in a hold tank, pumped through a vertical mill, and
reacted for an additional 1.5-2 hours at 80 to 90°C. Slurry is then pumped
to the mixer mounted on the duct and distributed over the remaining 7 5% of
the dry solid recycle, resulting in a "damp" solid of 30 to 50% moisture.
The product solids drop into the duct, where the flue gas is cooled to
within 10-15°C of saturation within 0.5 second. [Previous pilot-scale
injection/humidification using the MDI concept has not formed duct
deposits.]
The system behaves similarly to spray drying technology in that:
(1) SO2 loading controls the fresh lime feed rate,
(2) flue gas evaporative capacity and desired ATsat control the rate
of water addition, and
(3) slurry density and desired product moisture control the relative
amounts of dry recycle solids fed to the mix tank and MDI mixer,
respectively.
The Shawnee evaluation is scheduled to begin the first quarter of 1992 using
flue gas from combustion of a 2.7% sulfur midwestern coal. Performance
goals include:
• >90% SO2 capture at Ca/S <.1.2.
• ESP emissions not to exceed 13 ng/J concurrent with
targeted SO2 removal.
• Sustained process operation at the above levels,
4B-82
A
A
-------
sufficient to offer commercial process guarantees.
Additionally, during the first test phase, ending October 1992, it is
planned to acquire:
• information on minimum ESP size to achieve target dust
emission levels.
• information on mill power consumption to achieve SC>2
and particulate goals simultaneously.
• data on nitrogen oxide (N0X) and hazardous pollutant
capture.
If successful, future work beginning in late 1993 will address alternative
sorbents such as fluidized bed combustion wastes and lime kiln dust, coals
with varying chlorine and sulfur contents, and operation with more advanced
sorbent preparation concepts.
PILOT SUPPORT FOR FIELD EVALUATION
EPA and Acurex Corporation are currently renovating an existing 0.7
MWg pilot plant and pilot ESP to provide smaller-scale simulation of field
pilot and commercial applications of ADVACATE/MDI technology. Bench-scale
efforts are addressing various fly ashes suitable for sorbent preparation.
Alternative silica sources, such as glass, for applications on non-fly ash
sources, and additives to enhance absorption of NOx and air toxic materials
such as volatile organic compounds (VOCs) and mercury will be addressed in
future research efforts.
REFERENCES
1. R.K. Iler. The Chemistry of Silica. John Wiley & Sons, New York, NY,
1979.
2. R.K. Iler. "Effect of Adsorbed Alumina on the Solubility of Amorphous
Silica in Water." J. of Colloid and Interface Science, 44, 1973,
399-408.
3. S.A. Greenberg. "Thermodynamic Functions for the Solution of Silica
in Water." J. Phys. Chem., 61, 1957, 196-197.
4. F.M. Lea. The Chemistry of Cement and Concrete. Chemical Publishing
Company, New York, NY, 1971.
5. H.F.W. Taylor. The Chemistry of Cements. Academic Press, New York,
NY, 1964.
6. S.A. Greenberg, T.N. Chang, and E. Anderson. "Investigation of
Colloidal Hydrated Calcium Silicates, I. Solubility Products, J. Phys.
Chem., 64, 1960, pp. 1151-1157.
7. W. Jozewicz and G.T. Rochelle. "Fly Ash Recycle in Dry Scrubbing."
J. Env. Prog., 1986, 218-223.
8. J.R. Peterson. "Hydrothermal Reaction of Lime with Fly Ash to Produce
Calcium Silicates for Dry Flue Gas Desulfurization." Doctoral Thesis,
University of Texas, Austin, TX, December 1990.
9. W. Jozewicz, et al. "Silica-Enhanced Sorbents for Dry Injection
Removal of S02 from Flue Gas." JAPCA, 38, 1988, 1027-1034.
k.
k
4B-83
-------
10. W. Jozewicz, et al. "Development and Pilot Plant Evaluation of
Silica-Enhanced Lime Sorbents for Dry Flue Gas Desulfurization."
JAPCA, 38, 1988, 796-805.
11. J.C.S. Chang, et al. "Reactivation of Edgewater LIMB Solids by the
ADVACATE Process for In-Duct SO? Removal." In: Proceedings: 1990
S02 Control Symposium, Vol. 2, EPA-600/9-91-015b (NTIS PB91-197228):
4A-43 through 57, May 1991.
12. C.B. Sedman, et al. "Commercial Development of the ADVACATE Process
for Flue Gas Desulfurization." Presented at the 25th IECE Conference,
Reno, NV, August 16, 1990.
13. C.B. Sedman, et al. "Evaluation of Pilot ESP Performance with
Elevated Loadings from Sorbent Injection Processes." Presented at the
Ninth Particulate Control Symposium, Williamsburg, VA, October 1991.
4B-84
A
A
-------
CM
I
o
ca
O
o
E
J
w
o
E
c
o
e
0)
>
c
o
0
jn
1
O
ca
O
0 1 2 3
Surface Area Ratio of Ground Ash to Unground Ash
Figure 1. Lime Conversion Vs. Increased Fly Ash Surface Area
L
4B-85
-------
=3
c\j
100
-*
80
o
®o
-------
CO
E
z
D)
"O
03
O
_l
-------
100
90
-- FLY ASH+ADVACATE LOW
80
FLY ASH+ADVACATE HI
70
60
50
40
30
20
10
0
20
80
100
Figure 4. Projected ESP Performance--ADVACATE Process Impact
-------
£>-
ca
i
00
vo
MIXER
BOILER
TO MIXER TO MIXER
MIX
TANK
lEACTOR
~ STACK
DISPOSAL
Figure 5. ADVACATE/MDI Process
-------
Table 1
SUMMARY
OF SPROUT-
¦BAUER MIXER
TESTS
Dry
ADVACATE
Mixer
Solids
Slurry
Speed
Reject
Residual
Test
A' K
B,
c,
Percentage,
Moisture,
Number
Variable3
lb/hr®
lb/hr
rpm
%
%
1
MS
511
522
60.0
10.40
41.5
2
MS
511
528
84 .6
14.38
41. 6
3
SFR
499
480
77 .3
14.34
42.0
4
SFR
499
660
77.3
28.62
49.0
5
DSFR
255
480
77.3
48.50
54 .0
6
DSFR
415
480
77 .3
10.16
44.0
a MS - Mixer Speed
SFR - Slurry Feed Rate
DSFR - Dry Solids Feed Rate
b 1 lb = 0.454 kg
Table 2
SUBMICRON PARTICLE BEHAVIOR VS. FLY ASH INJECTION RATE
Fly Ash Outlet Submicron
Injection Rate Particle Loading Particle Loading
g/min (q/Nnv' ESP Off, q/Nm ESP Off, q/Nm
75 (3.28) 1.607 0.0306
150 (6.57) 3.147 0.0881
300 (13.13) 6.883 0.2065
550 (24.07) 8.445 0.3378
Table 3
SUBMICRON PARTICLE LOADING WITH FRESH SORBENT
Outlet Submicron
ADVACATE Injection Particle Loading Particle Loading
Rate, q/min ESP Off, q/Nm ESP Off, q/Nm
150 0.878 0.014
300 1.060 0.013
750 0.534 0.021
1000 0.355 0.012
4B-90
A
A
-------
Suitability of Available Fly Ashes in AD VAC ATE
Sorbents
4B-91
-------
Intentionally Blank Page
4B-92
-------
Carl Singer and Wojciech Jozewicz
Acurex Corporation
Environmental Systems Division
P.O. Box 13109
Research Triangle Park, North Carolina 27709
Charles B. Sedman
Air and Energy Engineering Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
ABSTRACT
The suitability of various coal fly ashes to form ADVAnced siliCATE (ADVACATE) hydrated
calcium silicate sorbents for sulfur dioxide (S02) was investigated. Fly ashes derived from
major coal fields in the United States were obtained from commercial power plants. Both fly
ash and sorbent samples were extensively characterized. Fly ashes were hydrated with
calcium hydroxide [Ca(OH)2] at similar conditions, and the resulting dry sorbents were tested
for reactivity in a bench scale sand bed reactor. The sand bed reactor was operated at 64°C
(147°F), 60% relative humidity, and 1000 ppmv S02 in nitrogen. Sorbent reactivity varied
greatly depending on the fly ash source. Sorbents were also prepared from fly ashes ground
in a bench scale attritor prior to hydration with Ca(OH)2 and tested for reactivity. Sorbents
prepared from ground fly ash were not always more reactive than sorbents made from
unground fly ash at the same hydration conditions.
This paper has been reviewed in accordance with the U.S. Environmental
Protection Agency's peer and administrative review poiicies and approved for
presentation and publication.
Preceding page blank
4B-93
-------
INTRODUCTION
Following several years of process development under the sponsorship of the U.S.
Environmental Protection Agency (EPA), the ADVAnced siliCATE (ADVACATE) process is in
a commercialization stage. As various geographical locations are being considered for the
potential installation of ADVACATE, it becomes increasingly important to understand how
varying properties of fly ash can affect the reactivity of fly ash/calcium hydroxide [Ca(OH)2]
(ADVACATE) sorbent with sulfur dioxide (S02). The content and type of glass, mineralogy,
and chemical composition of fly ash may vary significantly depending on the coal source,
operation of the boilers, and collection of the fly ash. Additionally, the chemical composition
may vary from one particle to the other even within a bulk sample of a given fly ash [1],
Production of ADVACATE sorbent involves formation of high surface area, low crystallinity
calcium silicate gels by slurrying fly ash with Ca(OH)2. The rate of gel formation is limited by
the dissolution rate of silica from fly ash glass [2]. Dissolution is strongly increased by
increases in surface area of fly ash, temperature, and pH. In the presence of alumina, the
dissolution rate and solubility of amorphous silica decrease dramatically [3]. The presence of
gypsum increases calcium in solution by binding with alumina. More calcium in solution
results in increased calcium silicate formation [4]. Because of the above mentioned
variabilities in fly ash properties and in the number of ADVACATE sorbent production
parameters they affect, the source of fly ash can have significant effect upon ADVACATE
sorbent reactivity.
Previous fundamental work [5] on hydrothermal reaction of Ca(OH)2 with fly ash showed that
low-calcium fly ashes produced ADVACATE sorbents with much higher surface area than ones
produced with high-calcium fly ash. The lower surface area of sorbent produced with high-
calcium fly ash was explained by the high alumina content, which competes for calcium with
ADVACATE sorbent, and makes unreactive low surface area calcium aluminates. Results
indicated that sorbents made with low-calcium fly ashes were more reactive toward S02 than
sorbents made with high-calcium fly ashes based on total sorbent calcium. Previous studies,
however, indicated that high calcium fly ashes produced superior sorbents based on added
Ca(OH)2 [2,6]. In addition to having varying content of amorphous silica available for reaction
with calcium to form ADVACATE sorbent, fly ashes from the same class may have varying
amounts of sodium or other alkali metals (affecting pH during dissolution) or varying contents
of sulfate compounds such as gypsum (combining with alumina and preventing it from reacting
with calcium).
Fly ash grinding has been shown [7,8] to increase the surface area and reactivity with S02 of
ADVACATE sorbent produced at the same conditions of time/temperature of slurrying and fly-
ash-to-Ca(OH)2 weight ratio as sorbent made with an unground fly ash. However, no data
are available on the effect of ground fly ash type on the reactivity with S02 of ADVACATE
4B-94
A
A
-------
sorbent. Considering the dissolution rate of silica, fine ground particles having high surface
area would be preferable for ADVACATE sorbent formation, while coarse unground particles
with lower surface area would be less desirable. Chemical dissolution may, however, vary
with particle size and there may be a tradeoff between particle size and composition; silica,
alumina, et al.
This paper reports results of an investigation of the suitability of fly ashes derived from major
U.S. coals to form ADVACATE sorbents by measuring reactivity with S02 in a bench scale
reactor. As a result of this characterization, correlations will be derived between fly ash
properties and ADVACATE sorbent reactivity with S02. This paper also gives results of an
evaluation of the effect of fly ash grinding on the reactivity of ADVACATE sorbents produced
from various low-calcium eastern and high-calcium western fly ashes. By carefully comparing
the reactivity of ADVACATE sorbents produced with ground and unground fly ash of the same
origin, information is obtained on the enhancement of reactivity by grinding.
EXPERIMENTAL
Fly ashes were obtained from commercial coal fired electric utilities from across the U.S.
Plants were solicited based on geographic spread to obtain a representative cross section
from U.S. coals. For this evaluation, 31 fly ashes were received: 15 from coal basins in the
Eastern and Midwestern U.S. (primarily Appalachian and Illinois basins), 14 from coal basins
in the Western U.S. (primarily the Powder River basin), and 2 from combustion of a mix of
Eastern and Western basin coals. Plant-supplied information identified the fly ashes by basin
and state of coal origin (see Table 1). Extensive ADVACATE work has been done [8] with
Detroit Edison's Clinch River fly ash (West Virginia coal, 5.4% calcium ash) which will be used
for comparison.
The goal of this study was to determine the effect, if any, of fly ash source on reactivity of
ADVACATE sorbents, not the applicability or cost of installation to any specific utility. It was
therefore necessary to maintain constant sorbent preparation procedures and put aside the
variations in operating conditions implicit in using coals of various sulfur and ash contents.
Fly ash was slurried with reagent grade hydrated lime [Ca(OH)2] (Fisher C-97: 15 m2/g
surface area, 0.10 cm3/g pore volume) at a weight ratio of 3:1 for 3 hours at 90°C. Slurry
concentration was maintained at 1 part solids to 15 parts water. At the end of this hydration
period, the slurry was vacuum filtered through 0.7 |im paper and dried to halt hydration
reactions. Most sorbents were dried in a microwave oven although a few required drying in
a vacuum oven due to high iron content.
ADVACATE sorbents were also made using ground fly ash at the above mentioned hydration
conditions. Fly ash was ground in a Union Process Model 01 attritor, a vertically oriented ball
mill, at 35 wt % slurry for 10 minutes. Power supplied to the mill was kept constant, implying
equal amounts of work applied in each batch. The ground fly ash was immediately vacuum
filtered and dried in an oven prior to being stored for later use.
The sorbents were tested for reactivity toward S02 in a bench scale sand bed reactor (see
Figure 1) described in detail in previous papers [9,10]. Simulated flue gas is humidified and
passed through the reactor. The sorbent to be tested is dispersed in sand to prevent
channeling. The reactor is immersed in a hot water bath to maintain constant reactor
temperature. The flue gas exits the reactor and passes through a wet bulb chamber to
k.
4B-95
-------
measure humidity and through a condenser to remove water prior to sending the sample to
a ThermoElectron Model 40 S02 analyzer. The breakthrough curve is integrated to determine
S02 absorption. Sorbents were tested at the following conditions: reactor temperature 64 °C,
relative humidity 60%, and 1000 ppmv S02 in nitrogen. Reactivity is expressed as conversion
based on the added Ca(OH)2 fraction (external conversion) in the sorbent. Reagent grade
Ca(OH)2, used as a baseline sorbent, yielded 22% conversion.
Sorbents and fly ashes were characterized for surface area and porosity by the Brunauer,
Emmett, and Teller (BET) method of nitrogen adsorption/desorption. Selected sorbents and
fly ash were also characterized using a scanning electron microscope (SEM). Elemental
composition of the fly ashes was determined by X-ray fluorescence. Alkalinity of unground fly
ashes was investigated by monitoring the pH obtained by slurrying 1 g fly ash in 100 g water
at 90°C for 8 minutes. Loss on ignition (LOI) was measured by heating fly ash to 950 °C for
15 minutes and measuring percent weight loss.
RESULTS AND DISCUSSION
Fly ash is generally characterized by both chemical composition and strength development of
cement pastes for suitability as a pozzolan. Class C fly ash must contain at least 50 wt %
network formers (silica, aluminum, and iron expressed as oxides) while Class F fly ash must
contain 70 wt % network formers [11 ]. Class C fly ash is obtained from burning subbituminous
or lignitic coals, while Class F fly ash is obtained from burning bituminous or anthracite coals.
Class F fly ashes are generally low in calcium content, while Class C fly ashes are generally
high. Sulfur is not to exceed 5 wt % [as sulfur trioxide (S03)] in either classification.
Normalized composition from X-ray fluorescence was used to classify ash in this study (see
Table 2). Characterization of strength development was beyond the scope of current work on
ADVACATE sorbents. All the fly ashes from the Eastern U.S. were classified as Class F.
Class F fly ashes will be referred to as eastern type fly ashes and fly ashes lower in network
formers will be referred to as western type fly ashes in the remainder of this paper. The
majority of western type fly ashes were Class C with the notable exception of two Fort Union
basin fly ashes which were low in network formers and high in sulfur. One of the fly ashes
from the Midwestern U.S. basin was eastern type and the other was western type. Two fly
ashes (FAS-2 and FAS-17, identified as originating from the Powder River basin) appear to
be from a lignite coal due to the high levels of sodium in the fly ash. FAS-11, represented as
a TX lignite, was a Class F fly ash and was not particularly high in sodium or calcium content.
Although the coal basins were in the Western U.S., Unita and Centralia basins are bituminous
in nature producing eastern type fly ash. The two fly ashes from combustion of mixed eastern
and western coal were high in network formers indicating eastern type fly ash.
Sand bed reactivity of ADVACATE sorbent made from the unground eastern type fly ashes
is shown in Figure 2. Sorbents made from Appalachian basin fly ashes yielded an average
of 48% conversion. Sorbents made from Illinois basin fly ashes averaged a slightly lower
conversion, 42%. While there is large variability in the reactivity of sorbents prepared from
either basin's fly ash, there is 95% confidence that these basins produce ADVACATE sorbents
of different reactivity. Average reactivity for all sorbents made with eastern type unground fly
ash was 44% conversion.
Sand bed reactivity of ADVACATE sorbents made from unground western type fly ashes is
shown in Figure 3. Sorbents made from Powder River basin fly ash averaged 52%
4B-96
A
A
-------
conversion. FAS-2 and FAS-17 appear to be anomalous from this basin due to the
extraordinary levels of sodium present. Sorbents made with Fort Union fly ashes yielded an
average 62% conversion. Similar to the note above for eastern type fly ash, there is large
variability in reactivity for western type fly ash within the same basin. These western basins
cannot be distinguished as producing sorbents from different reactivity populations. Average
reactivity for all sorbents made with western type unground fly ash was 54% conversion. On
the average, western type fly ashes produced more reactive sorbents than eastern type fly
ashes [based on the added Ca(OH)2j.
SEM photographs indicate that fly ashes are predominantly smooth non-porous spheres. A
typical field is shown in Figure 4. Char present in the fly ash is expected to have high specific
surface area and pore volume [12,13]. BET pore volume correlates
well with LOI, assumed to be predominantly char, for the eastern type fly ashes in this study
(see Figure 5). LOI is low for the western fly ashes and probably indicates moisture or free
Ca(OH)2 as opposed to char. The presence of char in the fly ash, expected to be inert toward
ADVACATE formation, effectively reduces the fly-ash (siliceous material)-to-Ca(OH)2 ratio in
the ADVACATE sorbent. Corrections for this inert material can be made by subtracting LOI
from the fly ash when considering fly-ash-to-Ca(OH)2 ratio for FAS sorbents. Figure 6
indicates that variations in reactivity for unground eastern type fly ashes are much greater than
can be expected from changes in fly-ash-to-Ca(OH)2 ratio. A better explanation of the effect
of char on ADVACATE reactivity is that ADVACATE materials become trapped in the pores
of the char, limiting reaction with S02. A linear regression of sorbent reactivity with fly ash
pore volume and fly ash S03 for Appalachian and Illinois basin fly ashes indicates similar
dependence on char,-1800 and -1600%/(cc/g), respectively. This regression also indicates
a positive correlation of reactivity with S03 in the fly ash indicating that recycle may have a
positive effect on reactivity even with 100% utilization. The dependence on S03 appears to
be different in the two basins, indicating different solution chemistry. These results indicate
that fly ash composition influences ADVACATE sorbent reactivity.
The high alkalinity of the western type fly ashes likely influenced sorbent reactivity since the
rate of silica dissolution increases with higher solution pH [14]. Western fly ashes that
developed higher pH in slurry also developed higher sorbent reactivity when hydrated with
Ca(OH)2 (see Figure 7). This trend was not predominant with the eastern type fly ashes.
Several fly ashes were ground prior to hydration in attempts to increase the dissolution rate
of silica and thereby promote ADVACATE sorbent reactivity. Previous work with Clinch River
fly ash had indicated that increased fly ash grinding increased ADVACATE sorbent reactivity
at similar hydration conditions [7]. The average conversion using ground eastern type fly
ashes was 42% while the average conversion using ground western fly ashes was 55%.
Figure 8 illustrates that grinding fly ash prior to hydration with Ca(OH)2 did not always improve
reactivity. A marked improvement in reactivity was seen by grinding some fly ashes while a
decrease in reactivity was seen
with others. For eastern type fly ashes, changes in reactivity effected by grinding are inversely
related to the alkalinity of the unground fly ash as measured by pH development (see
Figure 9). A large fraction of soluble species, including alkalies, are probably washed from
the fly ash in the bench grinding process which filters the ground fly ash and uses only the
solids. This would have the effect of inhibiting pH development in the Ca^H^fly ash slurry
and thereby decrease the dissolution rate of silica. This effect would not be seen in pilot plant
or commercial operation because the slurry would progress from the mill to a hydration tank.
To test this hypothesis, FAS-26, a high alkalinity eastern type fly ash, was ground and the
4B-97
-------
filtrate was used in the subsequent hydration. This process was expected to exaggerate
improvement because all the filtrate was used while only 13% of the ground fly ash was used
in the hydration. This filtrate had a pH of 11.2 and started to precipitate on heating to 90°C.
Reactivity of the resulting sorbent was slightly higher, 40% conversion, compared to ground
FAS-26 fly ash sorbent made without this filtrate, 36% conversion. Reactivity of sorbent made
with unground FAS-26 was 51% conversion, substantially higher than either sorbent made with
ground FAS-26. This suggests that grinding may change the solution chemistry by exposing
inner surfaces, thereby making chemicals accessible for reaction other than those accessible
in unground fly ash. Commercial operation with recycle materials will additionally change
solution chemistry by supplying sulfate to tie up alumina compounds.
Previous work with Clinch River fly ash (Appalachian basin) based ADVACATE sorbent
indicated that reactivity increases with increasing sorbent surface area using ground and
unground fly ash at various hydration conditions [8]. Figure 10 illustrates that this correlation
cannot be used with confidence with all ADVACATE sorbents or even with other Appalachian
basin fly ash ADVACATE sorbents. The consistent increase of reactivity with increasing
sorbent surface area shown with Clinch River fly ash based ADVACATE sorbent was not
always seen with other fly ash sources, or even with the ADVACATE sorbent from the same
fly ash on an unground and ground basis (see Table 2; e.g., FAS-1, FAS-8, FAS-12). If
reactivity is controlled by the surface area or sites available for absorption, changing chemical
composition in the ADVACATE sorbent gel as a result of varying fly ash source may account
for this variability.
CONCLUSIONS
1. All fly ashes tested produced ADVACATE sorbents more reactive than
baseline Ca(OH)2.
2. A large variability in ADVACATE sorbent reactivity was seen using fly
ash from different sources. The average reactivity of ADVACATE
sorbents made with western type fly ashes was higher than the
average reactivity of sorbents made with eastern type fly ashes. The
reactivity of ADVACATE sorbents made from Appalachian basin fly
ashes was higher than ADVACATE sorbents made from Illinois basin
fly ashes.
3. A correlation was observed between sorbent reactivity, fly ash pore
volume, and fly ash S03 content for ADVACATE sorbents made from
eastern type fly ashes and Ca(OH)2. High fly ash pore volume, an
indication of char, was associated with decreased reactivity, while high
fly ash S03 was associated with increased reactivity.
4. Western type fly ashes produced sorbents which increased in reactivity
as the alkalinity of the fly ash increased.
5. Grinding fly ash prior to hydrating ADVACATE sorbents did not always
improve reactivity over ADVACATE sorbents made with the same
unground fly ash.
4B-98
A
A
-------
REFERENCES
1. Roy, D.M., K. Luke, and S. Diamond, "Characterization of Fly Ash and Its Reactions in
Concrete," Materials Research Society Symposium Proceedings. 43, 1985.
2. Jozewicz, W., and G.T. Rochelle, "Fly Ash Recycle in Dry Scrubbing," Env. Proa.. 5,
pp. 218-223, 1986.
3. Greenberg, S.A., "The Depolymerization of Silica in Sodium Hydroxide Solutions,"^
Phvs. Chem.. 61, pp. 960-965, 1957.
4. Wild, S., M. Hadi, and G.L. Ward, "Expansion Mechanisms and Cementation in Cured
PFA-Lime Composites. Part 1: Development of Mechanical Strength and Expansion,"
Adv. Ceram. Res.. 3, (10), pp. 55-62, 1990.
5. Peterson, J.R., "Hydrothermal Reaction of Lime with Fly Ash to Produce Calcium
Silicates for Dry Flue Gas Desulfurization," Ph.D. Dissertation, University of Texas at
Austin, 1990.
6. Peterson, J.R., and G.T. Rochelle, "Aqueous Reaction of Fly Ash and Ca(OH)2 to
Produce Calcium Silicate Absorbent for Flue Gas Desulfurization," J. of Envir. Sci. and
Tech., 22, 1988.
7. Petersen, T., "An Experimental Study of Fly Ash Utility in Flue Gas Desulfurization,"
Licentiate Thesis, Lund Institute of Technology, 1988.
8. Hall, B., C. Singer, W. Jozewicz, C.B. Sedman, and M.A. Maxwell, "Current Status of
ADVACATE Process for Flue Gas Desulfurization," Presented at A & WMA 84th Annual
Meeting, 1991.
9. C. Jorgensen, J.C.S. Chang, and T.G. Brna, "Evaluation of Sorbents and Additives for
Dry S02 Removal." Environ. Progress. 6:1, p. 26, 1987.
10. W. Jozewicz, J.C.S. Chang, T.G. Brna, and C.B. Sedman, "Reactivation of Solids From
Furnace Injection of Limestone for S02 Control," Environ. Sci. Technol.. 21:7, p. 664,
1987.
11. ASTM, "Standard Specification for Fly Ash and Raw or Calcined Natural Pozzolan for
use as a Mineral Admixture in Portland Cement Concrete." 1989 Annual Book of ASTM
Standards, volume 04.02. pp 296-298, 1989.
12. Diamond, S., "Particle Morphologies in Fly Ash." Cement and Concrete Research. 16,
pp 569-579, 1986.
13. Schure, M.R., P.A. Soltys, D.F.S. Natusch, and T. Mauney, "Surface Area and Porosity
of Coal Fly Ash," Environ. Sci. Technol.. 19, pp 82-86, 1985.
14. Her, R.K., The Chemistry of Silica. John Wiley & Sons, New York, 1979.
4B-99
-------
-VAC
WATER
SYRINGE
PUMP
IevIporaton
AIR HEATER
ROTAMETER
A
scvn2
GAS
EXHAUST
so2analyzer
If
I
¦
\
CP
RECORDER
WATER
TRAP
(ICE COOLED)
WET BULB
THERMOMETER
SAND
BED
REACTOR
THERMOSTATED
WATER BATH
Figure 1. Sand bed reactor system
x
o
of
O
©
o
E
o~
CO
©
o
E
c
o
"c/>
k_
©
>
c
o
o
J**?*? JJ'J'J' cfjrj'j'jpj'jyj
P71 Unita
^ W. Interior
111 Centrailia
¦ Warrior
TX Lignite
Appalachian
11 Mix
H Illinois
Figure 2. Reactivity of ADVACATE sorbents prepared from unground eastern
type fly ash [fly-ash-to-CafOH^ ratio 3:1, slurried for 3 h at 90 °C]
4B-100
A
A
-------
CO 0.4
Powder River
Fort Union
Figure 3. Reactivity of ADVACATE sorbents prepared from unground western
type fly ash [fly-ash-to-Ca(OH) ratio 3:1, slurried for 3 h at 90 °C]
Figure 4. SEM photomicrograph of an eastern type fly ash (FAS-3)
as received x 1000
k.
4B-101
-------
0.014
i? 0.012
© 0.01
3
> 0.008
0.006
0.004
0.002
0.04 0.06
Loss on Ignition, g/g
Eastern FAS Western FAS
Figure 5. Influence of char on fly ash pore volume
x
o
la
O
©
o
E
o~
£0
o
E
c
o
"w
a>
>
c
o
O
Fly-ash-to-Ca(OH)2 ratio
Clinch River
• -B - ¦
Ground Clinch River
• • • A- ¦ •
Eastern FAS
O
Figure
6. Effect of altering fly-ash-to-Ca(OH)2 ratio by accounting for char in
unground eastern type fly ashes
4B-102
-------
0.75
X
o
"to
O
®
o
E
3"
CO
®
o
E
c
o
"(/>
Q)
>
C
o
o
9 9.2 9.4 9.6 9.8
Unground Fly Ash Alkalinity, pH
Powder River Fort Union
~ A
Figure 7. Effect of alkalinity on ADVACATE reactivity for western type fly ash
Improved Reactivity
9*:
Decreased Reactivity
0.4 0.5 0.6 0.7
Unground Sorbent Conversion, mole SCymole Ca(OH)2
Eastern Type Fly Ashes Western Type Fly Ashes
~ O
0.8
Figure 8. Comparison of reactivity of ADVACATE sorbents made with ground and
unground fly ashes [fly-ash-to Ca(OH)2 ratio 3:1, slurried for 3 h at 90 °C]
k.
4B-103
-------
6 8
Unground Fly Ash Alkalinity, pH
Figure 9. Effect of fly ash alkalinity on the change in ADVACATE sorbent
reactivity due to grinding fly ashes
0.8
eg
0.7
0.6
A
0.5
0.4
0.3
0.2
0
20
25
Sorbent Surface Area, m2/g
Clind^River Appalachian Other Ashes
Figure 10. Scatter of the sorbent-reactivity-to-sorbent-surface-area
relationship by using different unground fly ashes
4B-104
A
A
-------
TABLE 1. FLY ASH SOURCES
Designated
State
Basin
Type
Designated
State
Basin
Type
FAS-1
Mix
Powder River
/Appalachian
eastern
FAS-17
MT
Powder River
western
FAS-2
MT
Powder River
western
FAS-18
UT
Unita
eastern
FAS-3
KY
Appalachian
eastern
FAS-19
MO
Western
Interior
western
FAS-4
WY
Powder River
western
FAS-20
Appalachian
eastern
FAS-5
AL
Warrior
eastern
FAS-21
eastern
FAS-6
IL
Illinois
eastern
FAS-22
KY
Appalachian
eastern
FAS-7
KY
Illinois
eastern
FAS-23
Mix
Powder River
/Illinois
eastern
FAS-8
PA
Appalachian
eastern
FAS-24
WY
Powder River
western
FAS-9
KY
Illinois
eastern
FAS-25
Powder River
western
FAS-10
WY
Powder River
western
FAS-26
MO
Western
Interior
eastern
FAS-11
TX
TX Lignite
eastern
FAS-27
ND
Fort Union
western
FAS-12
IL
Illinois
eastern
FAS-28
ND
Fort Union
western
FAS-13
WY
Powder River
western
FAS-29
WA
Centralia
eastern
FAS-14
WY
Powder River
western
FAS-31
OH
Appalachian
eastern
FAS-15
WV
Appalachian
eastern
FAS-32
OH
Appalachian
eastern
FAS-16
ND
Fort Union
western
k
k
4B-105
-------
TABLE 2. FLY ASH COMPOSITION AND REACTIVITY
FAS-1
FAS-2
FAS-3
FAS-4
FAS-5
FAS-6
FAS-7
FAS-8
Conversion
0.35
0.50
0.48
0.46
0.36
0.49
0.36
0.49
FASAa
7.29
1.45
2.02
1.03
2.43
1.59
2.93
1.84
FAPVb
0.0091
0.0065
0.0033
0.0026
0.0038
0.0035
0.0078
0.0055
SSAC
10.66
25.02
13.73
14.39
9.90
7.51
5.95
5.88
SPV'1
0.0437
0.0963
0.0532
0.0612
0.0385
0.0397
0.0336
0.0336
Ground
Conversion
0.58
0.61
0.59
0.50
0.44
0.33
0.41
0.42
Ground FAS A
9.44
5.06
3.45
12.63
3.76
3.02
Ground FAPV
0.0159
0.0212
0.0099
0.0516
0.0095
0.0131
Ground SSA
6.73
8.85
17.08
14.90
14.95
10.30
11.97
Ground SPV
0.0349
0.0470
0.0769
0.0883
0.0569
0.0569
0.0584
LOI
7.73%
3.24%
2.74%
3.83%
5.04%
4.31%
PH
5.20
9.35
7.39
9.47
7.89
8.95
9.13
8.68
Si203
48.35
25.67
51.34
35.51
49.20
57.76
50.27
47.28
ai2o3
28.15
16.12
26.26
15.76
19.84
25.32
18.71
20.03
Ti02
1.44
1.10
1.31
1.26
1.02
1.16
1.04
0.96
Fe203
5.15
4.70
6.05
5.63
23.59
20.45
19.73
20.30
CaO
1.85
15.25
1.02
21.13
2.69
3.71
3.71
3.86
MgO
0.92
4.26
1.10
3.65
1.01
0.88
0.76
0.99
k2o
1.31
0.56
2.40
0.30
2.55
2.00
1.55
2.13
Na20
0.54
8.47
0.32
1.16
0.48
0.27
0.43
0.79
S03
0.85
3.02
0.25
1.09
0.99
1.06
0.82
1.80
Total
88.56
79.15
90.05
85.49
101.37
112.61
97.02
98.14
base/acid
0.13
0.78
0.14
0.61
0.43
0.32
0.37
0.41
SigOg/A^Og
1.72
1.59
1.95
2.25
2.48
2.28
2.69
2.36
dolomite
0.28
0.59
0.19
0.78
0.12
0.17
0.17
0.17
(continued)
4B-106
-------
TABLE 2. FLY ASH COMPOSITION AND REACTIVITY (CONT)
FAS-9
FAS-10
FAS-11
FAS-12
FAS-13
FAS-14
FAS-15
FAS-16
Conversion
0.42
0.48
0.49
0.46
0.46
0.56
0.41
0.72
FASAa
1.19
1.32
0.73
1.52
1.09
1.64
2.32
3.48
FAPV6
0.0029
0.0061
0.0011
0.0034
0.0030
0.0056
0.0083
0.0088
SSAC
7.12
10.29
12.27
6.24
10.06
12.11
7.30
21.54
SPV*1
0.0372
0.0594
0.0481
0.0344
0.0428
0.0629
0.0386
0.1170
Ground
Conversion
0.38
0.50
0.43
0.36
0.52
0.62
0.33
0.75
Ground FAS A
4.41
5.38
16.43
Ground FAPV
0.0133
0.0189
0.0613
Ground SSA
10.24
16.96
29.69
Ground SPV
0.0494
0.0915
0.1318
LOI
3.21%
0.95%
4.53%
0.73%
0.50%
4.89%
pH
8.28
9.41
8.49
9.10
9.45
9.57
6.55
9.79
Si203
59.68
28.67
51.34
51.34
27.81
31.66
47.92
7.08
ai2o3
19.46
17.19
28.15
22.86
14.17
17.80
25.51
4.08
Ti02
1.54
1.39
1.40
1.17
1.25
1.33
1.30
0.27
Fe203
3.42
4.80
13.61
14.00
5.75
4.17
6.59
3.92
CaO
7.67
27.42
1.55
3.02
28.12
24.49
1.05
10.51
MgO
1.96
4.31
0.83
0.86
6.55
5.77
0.96
2.47
k2o
0.86
0.19
2.40
2.59
0.24
0.38
2.87
0.48
Na20
0.51
0.73
0.17
0.74
2.14
2.52
0.23
28.17
S03
0.16
1.43
0.36
1.08
2.12
2.08
0.30
24.18
Total
95.26
86.13
99.81
97.66
88.15
90.20
86.73
81.16
base/acid
0.18
0.79
0.23
0.28
0.99
0.74
0.16
3.98
S^Og/A^Og
3.07
1.67
1.82
2.25
1.96
1.78
1.88
1.73
dolomite
0.67
0.85
0.13
0.18
0.81
0.81
0.17
0.28
(continued;
4B-107
-------
TABLE 2. FLY ASH COMPOSITION AND REACTIVITY (CONT)
FAS-17
FAS-18
FAS-19
FAS-20
FAS-21
FAS-22
FAS-23
FAS-24
Conversion
0.70
0.46
0.46
0.48
0.41
0.53
0.43
0.49
FASAa
3.59
2.37
1.85
1.74
6.71
1.84
1.34
9.53
FAPV*5
0.0044
0.0040
0.0045
0.0055
0.0104
0.0036
0.0034
0.0491
SSA°
22.44
7.24
13.49
11.31
7.67
6.94
8.44
11.42
SPV6
0.1160
0.0318
0.0603
0.0366
0.0362
0.0373
0.0467
0.0408
Ground
Conversion
0.53
0.39
0.45
Ground FASA
5.50
Ground FAPV
0.0279
Ground SSA
14.81
Ground SPV
0.0720
LOI
3.01%
5.25%
3.05%
pH
9.78
9.45
9.45
5.94
2.86
5.86
9.57
9.45
Si203
21.39
52.20
27.17
46.21
46.63
54.34
35.72
28.45
ai203
16.40
17.54
13.05
27.02
26.26
28.34
14.91
18.27
Ti02
1.18
0.95
1.13
1.36
1.24
1.39
1.01
1.52
Fe203
3.77
3.36
4.87
4.30
6.82
6.81
18.44
5.36
CaO
16.79
6.03
26.58
1.07
0.95
1.29
19.73
29.38
MgO
4.99
1.07
3.81
0.41
0.66
0.75
2.90
5.72
K20
0.46
0.94
0.22
2.12
1.40
1.67
0.60
0.13
Na20
9.88
1.04
0.79
0.08
0.31
0.18
1.08
1.74
CO
O
CO
4.46
0.45
1.59
0.38
0.51
0.39
1.99
3.22
Total
79.32
83.58
79.21
82.95
84.78
95.16
96.38
93.79
base/acid
0.92
0.18
0.88
0.11
0.14
0.13
0.83
0.88
Si203/AI203
1.30
2.98
2.08
1.71
1.78
1.92
2.40
1.56
dolomite
0.61
0.57
0.84
0.19
0.16
0.19
0.53
0.83
(continued;
4B-108
A
A
-------
TABLE 2. FLY ASH COMPOSITION AND REACTIVITY (CONT)
FAS-25
FAS-26
FAS-27
FAS-28
FAS-29
FAS-31
FAS-32
Conversion
0.45
0.51
0.59
0.54
0.52
0.41
0.38
FASAa
1.48
1.09
2.14
0.82
5.40
3.13
1.11
FAPVb
0.0043
0.0021
0.0049
0.0020
0.0316
0.0116
0.0039
SSAC
13.18
6.29
10.38
7.19
5.57
8.05
6.37
SPV*1
0.0558
0.0351
0.0552
0.0427
0.0335
0.0470
0.0377
Ground Conversion
0.36
0.41
0.43
Ground FASA
5.40
2.62
6.67
Ground FAPV
0.0235
0.0108
0.0172
Ground SSA
7.48
12.10
5.45
10.99
Ground SPV
0.0434
0.0540
0.0300
0.0500
LOI
0.62%
0.26%
5.76%
1.21%
PH
9.01
9.77
9.83
9.59
8.04
2.40
2.82
Si203
28.88
52.20
22.89
31.55
42.25
47.28
54.01
ai2o3
16.34
19.84
8.47
10.35
19.84
23.24
25.04
Ti02
1.14
0.96
0.43
0.49
2.92
1.30
1.22
Fe203
5.78
44.18
5.65
7.21
4.80
24.59
19.02
CaO
18.75
5.19
14.97
15.74
5.32
1.85
1.60
MgO
3.93
0.93
4.15
3.80
0.77
1.11
0.93
k2o
0.67
1.52
1.22
0.92
0.38
1.81
1.42
Na20
3.24
0.28
11.50
6.71
2.07
0.58
0.26
CO
o
in
1.48
1.60
8.43
3.10
0.20
1.60
0.68
Total
80.21
126.70
77.71
79.87
78.55
103.36
104.18
base/acid
0.70
0.71
1.18
0.81
0.21
0.42
0.29
SijOg/AljOg
1.77
2.63
2.70
3.05
2.13
2.03
2.16
dolomite
0.70
0.12
0.51
0.57
0.46
0.10
0.11
aFASA = Fly Ash Surface Area, m2/g CSSA = Sorbent Surface Area, m2/g
bFAPV = Fly Ash Pore Volume, cm3/g dSPV= Sorbent Pore Volume, cm3/g
k.
4B-109
-------
Intentionally Blank Page
/
4B-110
-------
MECHANISTIC STUDY OF DESULFURIZATION
BY ABSORBENT PREPARED FROM COAL FLY ASH
Hideshi Hattori
Hideaki Kumagai
Department of Chemistry
Faculty of Science
Hokkaido University
Sapporo 060, Japan
Tomohiro Ishizuka
Tsutomu Ueno
Department of Research and Development
The Hokkaido Electric Power Co.,Inc.,
461-6, Satozuka, Toyohira-ku, Sapporo 004, Japan
Preceding page blank
L.
4B-111
-------
Intentionally Blank Page
X
4B-112
-------
ABSTRACT
A model absorbent prepared from SiC^, AlgOg, Ca(OH)2 and CaSO^ exhibited a high
activity for desulfurization of a model flue gas containing NO, SC^j Og, C02» and
H2O. The rate of SO2 absorption is accelerated by the presence of 1^0 and N0X in a
1 ft 1 fi 1ft
flue gas. The tracer studies in which N 0 0 and O2 were used indicate that
SO2 is oxidized by NC>2 and not by C^. A simplified scheme for desulfurization is
proposed as follows.
NO + l/202
SO2 + N02
CaO + SO3
- no2
S03 + NO
CaS04
The state of N02 acting as an oxidizing reagent of S02 was suggested to be in the
- 2-
forms of N02 and monodentate nitrato complex(NOg ) by IR.
One of the roles role of A1 in the absorbent is suggested to construct Ettringite
structure during preparation process, which increases a dispersion of CaO on heat
treatment above 130°C as well as to form micro pores so that SO2 can easily
penetrate into the absorbent.
Preceding page blank
4B-113
-------
INTRODUCTION
The SOx absorbent prepared from a mixture containing coal ash, Ca(0H)2> and CaSO^
exhibits a high activity for flue gas desulfurization in coal burning power
station. The charateristics of the absorbent are ; a high utilization percentage
of Ca as in the form of CaSO^ is attained when NO and 1^0 are present, and
absorbent shows high activity when the absorbent is dried above 130°C before use.
As main components of coal ash are Si02 and A^Og, we have prepared model
absorbents from Si02, A^Og, Ca(0H)2 and CaSO^, and studied the mechanisms of the
desulfurization by isotopic tracer method and characterization of the absorbent.
EXPERIMENTAL
Preparation of absorbent
The absorbent(sample 1) was prepared from a mixture containing Al(0H)g 39.9wt %,
Si02*4.5H20 13.5 wt %,Ca(0H)2 27.7 wt %, and CaSO^^I^O by kneading with water
followed by aging under a steam at 100°C and drying at 200°C. In addition to this
absorbent used for most of the experiments, four absorbents (sample 2-4) were
prepared as summarized in Table 1.
Temperature Programmed Desorpton Study
The sample was outgassed at 130°C and exposed to 20 - 50 Torr of a mixture of SO2,
NO2 and ^^2, or SO2, N^O^O, and O2, or ^NO^. Then TPD was run at heating rate
of 10°C/min. The evolved gases were analyzed by mass spectrometory.
Activity Test
The activity of the absorbent was examined under the following conditions: gass
composition, SO2 450 ppm, NO 250 ppm, O2 6 %, CO2 13 %, 1^0 10 %, N2 balance;
reaction temperature 130°C, SV=6000 h
4B-114
A
A
-------
Table 1
Model absorbent
Sample No. Starting material Remarks
1 A1(0H)3, SiC>2> Ca(0H)2> CaSO^, Standard sample
2 Si02-Al203 668m2g_1, Ca(OH)2, CaS04, S.A = 42 m2g_1
3 Si02-Al203 66m2g-1, Ca(OH)2, CaS04 S.A = 96 m2g-1
4 SiOg-AlgOg gel,Ca(0H)2j CaSO^,
5 N^SiOg, A^SO^, Ca(0H)2> CaSO^ Ettringite structure
RESULTS AND DISCUSSION
The activity of the sample 1 was equivalent to the absorbent prepared from coal
ash. The SC>2 is fixed in the absorbent in the form of CaSO^, which was confirmed
by XRD. The Ca utilization percentages for the samples 2-5 are plotted against
time on stream in Fig. 1. For all the samples, the utilization percentage
increased with the reaction time, no saturation of SO2 inclusion being appreciated
in 5 h.
As SC>2 is fixed in the absorbent in the form of CaSO^, oxidation of SO2 should be
involved in the SO2 absorption process. To examine where the oxidizing regent
18
originate from, TPD study using 0 as a tracer was performed. If the oxygen atom
to oxidize SO2 originates from molecular oxygen, the SO2 desorbed from the sample
18 18
exposed to a mixture of SO2, N02, and C>2 would contain a large amount of 0. In
contrast, if the oxygen atom originates from N0X, the SO2 from the sample exposed
to a mixture of SO2, N^O^O and would contain a large amount of ^®0. The TPD
plots of the desorbed SO2 are shown in Figs. 2 and 3. The evolved SO2 appeared
above 900°C originate from CaSO^. For both samples, the amount of SO2 exceeded
18
that of SO 0, since the absorbent contains CaSO^ as one of the components.
k.
k
4B-115
-------
18
However, it is clear that SO 0 deorbed in the larger amount for the sample exposed
to N*®0*®0. It becomes more evident if the fraction of S0^®0 in the total S0£ is
1 o
calculated. The fractions of SO 0 were 0.048 and 0.003 for the sample exposed to
1 fi 1 ft 1 fi
a mixture of SO2, N 0 0 and O2 and for the sample exposed to a mixture of SO21
18 . 1 ft
N02> and O2, respectively. The fraction of SO 0 was higher for the former
sample by more than ten times. Therefore, it is concluded that SO2 is oxidized to
SOg by N0X and not by 0£. The reaction sequence can be expressed as follows.
CaO
^ CaSO
4
In this reaction sequence, NO and N0£ in the forms of adsorbed state act as
catalyst for SO2 oxidation. The catalysis occurs on the surface of the absorbent.
The states of NO and NO2 in the absorbent were examined by IR spectroscopy. Fig. 4
shows IR spectrum observed when NO2 was adsorbed on the sample 1 and its change
caused by introduction of SO2. The bands at 1342 and 1460 cm-^ were assigned to
monodentate nitrato complex. The band at 1342 cm ^ decreased on introduction of
S02. This indicates that monodentate nitrato cmplex is capable of oxidizing SO2.
When a mixture of NO and S02 was adsorbed on the samples, the IR bands appeared
1249 -1261 cm ^ for all samples. These bands are assigned to NO2 • When NO2 was
2-
formed, SO 4 was simaltaneously formed in the samples, which was confirmed by TPD.
It is suggested that NO2 species are also capable of oxidizing SO2.
The reaction scheme is postulated as shown below.
4B-116
A
-------
(CaO)
NO
>• NO (ad) -tk
~
CaO
*> SOg(ad)
* CaSO^
(CaO)
so2
*¦ S02(ad)-S
One of the characteristic features of the absorbent is high utilization of Ca in
the absorbent. To attain a high utilization percentage, existence of Si and Al in
the absorbent is required. It is known that Ettringite structure is formed from a
mixture containing A^fSO^)^, CaSO^, Ca(0H)2, and water glass. The prepared sample
5 has Ettringite structure when dried at 100°C, but loses the structure detected by
VRD on heating above 130°C. Ettringite structure has columns as shown in Fig. 5.
Below 100°C, the columns are filled with water molecules. The absorbent exhibits
high Ca utilization percentage after heat treatment above 130°C. Even Ettringite
structure disappeares in XRD pattern on heat treatment, the colunms structure is
considered to be retained. A high Ca utilization percentage is considered to be
due largely to the existence of the column structure, which enable SO2 to penetrate
into the absorbent. In addition, Ca is highly dispersed in the absorbent through
Ettringite structure formation. Therefore, the role of Al is suggested to form
Ettringite structure in the precursor of the absorbent. Though the highky
dispersed Ca in the forms of CaO and Ca(0H)2 should also show a high activity, the
formation of Ettringite structure in the preparative procedures is a key to prepare
an efficient absorbent.
k.
4B-117
-------
100
0)
w 80 ~
«!
0 1 2 3 4 5
Time on Stream / h
Figure 1 Calcium Utilization Percentage vs. Reaction Time
Samples 2(#), 3 (Q) , 4(A), 5(A), and ra0 ~
-------
Temperature / C
Figure 2 TPD Plot for SC2 <£) and
SO^O (Q) fro* Adsorption of
the Mixture ( SO2 + NO2 + ^02 )
700
800
900
1000
Temperature / C
Figure 3 TPD Plot for SOg
SO^O (O) fro* Adsorption
the Mixture ( SC>2 +
(0) and
of
+ o2)
-------
13,42
1655
1900
1700
1500
1300
Wave Number / cm ^
Figure 4 IR Spectra of (a) Adsorbed on the Saaple 1
and (b) SO2 admitted after (a)
-------
cd
ro
Al
Figure 5 Ettringite Structure
-------
Intentionally Blank Page
4B-122
-------
RESULTS OF SPRAY DRYER/PULSE-JET FABRIC FILTER PILOT UNIT TESTS
AT THE EPRI HIGH SULFUR TEST CENTER
Gary M. Blythe
Joseph R. Peterson
Andrew D. Burnette
Radian Corporation
8501 MoPac Boulevard
Austin, Texas 78759
Richard G. Rhudy
Electric Power Research Institute
3412 Hillview Avenue
Palo Alto, California 94303
Preceding page blank
L
4B-123
-------
Intentionally Blank Page
4B-124
A
-------
ABSTRACT
At the High Sulfur Test Center in western New York, EPRI, along with a number of
co-sponsors, is conducting research and development for flue gas desulfurization
(FGD) technologies in medium- and high-sulfur coal applications. This paper
describes the results of tests conducted in the latter half of 1990 and during 1991
on a spray dryer FGD pilot unit. In all of these tests, a pulse-jet fabric filter
was used as the downstream particulate control device.
Results presented show the effects of atomizer feed configuration, coal sulfur
level, coal chloride level (and/or calcium chloride spiking), and other performance
enhancements on system S02 removal and lime utilization performance. The results
show that, particularly for medium-sulfur coal applications, high levels of S02
removal (90% and greater) can be achieved at high lime utilization values.
Results are also presented demonstrating the successful operation of a high ratio
(3.5 acfm/ft2 air-to-cloth ratio) pulse-jet fabric filter downstream of the spray
dryer. Pressure drop and particulate control performance compare favorably with
that of a more conservatively sized, and hence more expensive, reverse-gas fabric
filter design. During the past year and a half, testing has been conducted with
relatively low-cost acrylic bags in the pulse-jet fabric filter. The pressure drop
and particulate control performance of these bags compare quite favorably with that
of other, more expensive fabrics. Also discussed are results from applying a vinyl
ester coating in the outlet plenum of the fabric filter. To date, the coating has
proven very effective as a means of controlling corrosion there.
Preceding page blank
4B-125
-------
INTRODUCTION
Since 1987, EPRI has operated the High Sulfur Test Center (HSTC), which is dedi-
cated to FGD research for medium- and high-sulfur-coal utility applications. The
facility is located at the New York State Electric and Gas Company's Kintigh Sta-
tion. Several organizations have cosponsored the facility during the time period
discussed in this paper. Those cosponsors include the Empire State Electric Energy
Research Corporation, New York State Energy Research and Development Authority,
New York State Electric and Gas Company, Consolidation Coal Company, and the U.S.
Department of Energy.
The facility includes a 4-MW wet scrubber pilot unit, a 0.4-MW wet scrubber mini-
pilot unit, and the subject of this paper, a 4-MW spray dryer FGD pilot unit. A
paper presented at the First Combined FGD and Dry S02 Control Symposium in St.
Louis in 1988 described many of the results from the spray dryer pilot unit, pri-
marily for operation in conjunction with a reverse-gas fabric filter particulate
collector (!). These results showed that the spray dryer/reverse-gas fabric filter
combination was capable of achieving 90% overall S02 removal or greater for spray
dryer inlet S02 levels as high as 2500 ppmv (equivalent to about 3.5% to 4% sulfur
coal). Also, the reverse-gas fabric filter was able to maintain particulate emis-
sions levels well below the 0.03 lb/106 Btu required by current New Source Perfor-
mance Standards (NSPS) while operating at acceptably low pressure drop values.
A primary objective of the HSTC is to research and develop measures to lower the
costs and improve the reliability of existing or emerging FGD technologies. The
cost of a particulate control device represents a substantial portion of the
capital investment required to install a spray dryer FGD system. Reverse-gas
fabric filters in utility applications—with or without spray dryer FGD systems-
tend to be conservatively sized with design air-to-cloth ratios in the range of 1.5
4B-126
A
A
-------
to 2.0 acfm/ft2. Pulse-jet fabric filters are generally sized for much higher
air-to-cloth ratios, in the range of 3.5 to 4.0 acfm/ft2, and hence, their capital
costs are as much as 30% lower than for a reverse-gas unit sized to treat the same
volume of gas (2).
In 1988, a pulse-jet fabric filter (PJFF) was installed at the HSTC. This unit
made it possible to directly compare pulse-jet fabric filter particulate control
technology to reverse-gas fabric filtration and electrostatic precipitation. For a
spray dryer FGD system, the primary issues regarding the use of a pulse-jet fabric
filter rather than a reverse-gas unit are: what are the impacts of the higher air-
to-cloth ratio and more frequent cleaning of the pulse-jet unit on S02 removal; and
what are the impacts of the spray dryer upstream on the pressure drop and corrosion
tendencies of the pulse-jet unit?
A paper presented at the 1990 S02 Control Symposium presented the results from
about six months of spray dryer/PJFF performance (3). Those results showed the
system S02 removal performance was equivalent with either the reverse-gas or the
pulse-jet fabric filter downstream of the spray dryer. The PJFF did not appear to
be adversely affected by the spray dryer upstream. Pressure drop and particulate
control performance remained acceptable. A few bag failures occurred, but these
appeared to be attributable to poor quality control during bag manufacture.
Since May 1990, the pilot unit has continued to operate in the spray dryer/PJFF
configuration. Spray dryer operation has focused on optimizing the process to
lower costs in medium- to high-sulfur-coal applications, and the PJFF is being
evaluated with low-cost acrylic bags installed. The following sections describe
results from the current operating period.
PILOT UNIT PROCESS DESCRIPTION
Figure 1 is a simplified process flow diagram for the spray dryer pilot unit. Flue
gas is isokinetically extracted from the outlet duct on the Kintigh Station boiler.
The normal sulfur content of the coal fired (2.8%) produces a flue gas S02 concen-
tration of about 1600 to 1800 ppmv. The S02 concentration of the spray dryer inlet
gas can be raised by spiking with a small amount of S02 to allow testing at pilot
unit inlet S02 levels of 2000 ppmv and greater, or the concentration can be lowered
to 1500 ppm by bleeding in a small amount of air. An electric heater is used to
L
4B-127
-------
control the inlet flue gas temperature. Flue gas enters the spray dryer, where it
contacts a finely atomized slurry of fresh lime and recycled solids. The water
content of the slurry evaporates, cooling the flue gas. Simultaneously, the lime
in the slurry reacts with flue gas S02. Flue gas leaves the spray dryer from the
side of the vessel through an outlet duct that is centered above the bottom of the
cone. The entrained solids, containing equilibrium moisture levels, continue to
react with flue gas S02. In fact, S02 removal in the particulate control device
contributes significantly to the system S02 removal. Gas leaving the particulate
control device passes through an induced draft fan, then returns to the Kintigh
Station duct.
As mentioned above, the feed to the atomizer is a mixture of lime slurry, recycle
solids, and makeup water. The rate at which this slurry is fed to the atomizer is
controlled to reduce the flue gas temperature at the spray dryer outlet to within a
desired approach to adiabatic saturation. The fresh lime portion of this feed
slurry is produced from pebble quicklime as a 24 to 26 wt.% solids content slurry
in a paste slaker. The fresh lime content of the atomizer feed slurry is con-
trolled to result in the desired reagent ratio. The reagent ratio is defined as
the moles of fresh lime in the atomizer feed slurry, neglecting any remaining lime
in the recycled solids, divided by the number of moles of S02 in the system inlet
flue gas. Makeup water is added to the feed slurry because, at a given reagent
ratio, the amount of water associated with the fresh lime slurry is generally not
sufficient to reduce the flue gas temperature to the desired level. Solids col-
lected in the particulate control device are recycled to raise the solids level in
the combined atomizer feed slurry. This recycle is implemented both to improve
overall lime utilization and to improve droplet drying characteristics. Recycle of
solids from the particulate control device is preferred because solids which drop
out in the spray dryer contain oversize chunks that tend to cause pluggage in the
atomizer feed system.
The makeup water and recycled solids can be combined with the fresh lime slurry to
produce the atomizer feed slurry in two ways: all three streams can be combined in
a single tank and fed directly to the atomizer from that tank, or the makeup water
and recycled solids can be combined in one tank, then mixed with a separate fresh
lime slurry stream in the feed line to the atomizer. Results from both modes of
operation are discussed in this paper.
4B-128
-------
PILOT UNIT EQUIPMENT DESCRIPTION
The pilot unit is sized to treat 12,800 acfm of flue gas at 300°F. The spray dryer
vessel has a 13-ft diameter and an 11-1/4-ft straight side with a 45* cone bottom.
The vessel volume of 2160 ft3 results in a flue gas residence time of about 12
seconds when cooling 12,800 acfm of gas to within a 20*F approach to adiabatic
saturation. A single rotary atomizer at the top of the vessel is equipped with a
250-mm (9.8-inch) diameter atomizer wheel with eight 6-mm (0.24-inch) diameter
nozzles, and is driven by a variable-speed 50-hp motor. At the normal operating
speed of 14,000 rpm, the tip speed at the outer edge of the wheel is about 600
ft/sec, which is within the range of commercial practice for utility spray dryer
FGD systems.
The pulse-jet fabric filter (PJFF) particulate control device consists of a single
compartment with 124 oval-shaped bags arranged in a circular array. Figure 2
illustrates the general arrangement of the PJFF. Cleaning is accomplished with
short bursts of relatively low pressure (about 10 psig) compressed air while the
PJFF is on line. The compressed air is distributed to the bags by a manifold which
rotates just above the circular array of bags. This design avoids the cost and
complexity of the piping, solenoid valves, and Venturis normally required to con-
trol and distribute cleaning air pulses in high-pressure cleaning applications.
The felted bags are 20 ft long with a circumference of approximately 15-1/2 inches.
Each bag has a filtering area of approximately 25 ft2. The bags are installed on
cages that have an oval cross-section, with the longer dimension oriented parallel
to the direction of rotation of the cleaning manifold. Three bag materials have
been tested to date, felted Daytex (Ryton batt with a Rastex scrim), felted Ryton/
Ryton (Ryton batt with a Ryton scrim), and felted acrylic (acrylic batt with an
acrylic scrim). The acrylic bags are temperature sensitive; the fabric is subject
to dimensional stability problems if operated continuously at temperatures above
approximately 280#F. However, the fabric is quite durable and relatively inex-
pensive. Spray dryer applications, with low fabric filter inlet flue gas temper-
atures, may represent an ideal case for the use of this fabric.
The cleaning program for the PJFF is quite flexible. Cleaning can be initiated by
time between cleans, or it can be triggered by a set-point pressure drop. The
duration of each cleaning can be controlled to last a set period of time (hence
setting the number of pulses per cleaning period) or to clean down to a set
4B-129
-------
pressure drop value. With all 124 bags installed and a flue gas flow rate which
results in the design 12-second residence time in the spray dryer, the air-to-cloth
ratio at the PJFF is about 3.5 acfm/ft2. This value is approximately double that
which is typical of utility practice for reverse-gas fabric filters.
The pilot unit is well instrumented with provisions to measure flue gas flow rate
and S02 and 02 concentrations at the spray dryer inlet, outlet, and particulate
control outlet; slurry flow rates; spray dryer and particulate control device pres-
sure drop; and flue gas temperatures throughout the system. Manual measurements
are made on an hourly basis to measure the flue gas wet bulb temperature. On a
less frequent basis, manual measurements are also made to measure flue gas particu-
late mass loadings and particle size distributions.
RESULTS
A number of results from approximately 16 months of pilot unit operation are pre-
sented, all in the spray dryer/PJFF configuration. First S02 removal, then par-
ticulate control performance results are presented. Also discussed is a parallel
program to study spray dryer waste solids properties. Several field cells filled
with medium- to high-sulfur spray dryer wastes are being evaluated.
S0: Removal
Effect of Reagent Ratio. Figure 3 summarizes typical S02 removal performance for
the spray dryer/PJFF system, with the data plotted as percent S02 removal versus
reagent ratio. These data are for Mississippi Lime reagent and reflect operation
of the spray dryer with an inlet S02 level of 1500 ppmv, an inlet temperature of
300'F, and a 20*F approach to adiabatic saturation at the outlet. This inlet S02
level corresponds to a coal sulfur level of about 2.5%. In all of these tests, the
atomizer feed slurry, consisting of a mixture of fresh lime slurry, additional
makeup water, and solids recycled from the pulse-jet fabric filter hopper, was
mixed and fed to the atomizer from a single tank.
Three curves are illustrated in Figure 3. The top curve represents overall system
S02 removal performance, and the curve just below represents the percent removal
across the spray dryer. The bottom curve represents the PJFF contribution to over-
all S02 removal. This value is calculated by dividing the decrease in flue gas S02
4B-130
-------
content across the PJFF by the spray dryer inlet S02 content. By defining the PJFF
contribution in this manner, the spray dryer removal and the PJFF contribution may
be summed to get the overall system S02 removal.
The results in Figure 3 illustrate S02 removal performance for reagent ratio values
from about 1.0 to 1.3. The system S02 removal varies from about 80% at a 1.0 rea-
gent ratio to about 95% at a 1.3 reagent ratio. The contribution of the PJFF to
overall removal appears to be important in achieving 90% or greater removal. The
S02 removal across the spray dryer is limited to 85% or less, while the PJFF con-
tribution amounts to about 10 to 12 percentage points.
Effect of Inlet S0: Concentration. Figure 4 shows the effect of inlet S02 concen-
tration on spray dryer/PJFF system S02 removal performance. For a 2500 ppm inlet
S02 level (corresponding to a 3.5% to 4% sulfur coal), 90% overall S02 removal is
achieved at these operating conditions at a reagent ratio just less than 1.6. As
the inlet S02 levels drops to 2000 ppm, corresponding to about a 3% sulfur coal,
90% overall removal is achieved at a reagent ratio just under 1.3. At 1500 ppm,
corresponding to about a 2.5% sulfur coal, this level is achieved at a reagent
ratio of less than 1.2.
The difference in lime requirements to achieve 90% S02 removal at 2000 ppm and 1500
ppm inlet S02 levels may be somewhat understated in Figure 4, though. As mentioned
earlier, to achieve inlet S02 levels of 1500 ppm, the normal flue gas from the
Kintigh Station must be diluted by 10% to 15% with air. This has the effect of
raising the flue gas oxygen content and lowering the carbon dioxide and moisture
contents. More importantly, though, it lowers the flue gas HC1 content. As will
be discussed below, the calcium chloride produced when HC1 is removed from the flue
gas is an important performance additive in the spray dryer FGD system. The data
points in Figures 3 and 4 for 1500 ppm inlet S02 level operation would likely have
shown higher S02 removals at each reagent ratio tested had the flue gas HC1 content
not been diluted by 10% to 15%.
Effect of Calcium Chloride Addition. The effects of elevated chloride levels on
spray dryer FGD system performance have been reported in numerous reports and tech-
nical papers (4, 5, 6, 7). The effects are reported to be due to the deliquescent
nature of calcium chloride, which means that the compound is hygroscopic and highly
4B-131
-------
water soluble. Calcium chloride concentrates in the droplets in the spray dryer,
slowing overall droplet evaporation times, thus enhancing S02 removal performance.
In the solids collected in the particulate control device, elevated calcium chlor-
ide levels lead to higher residual moisture levels, and thus, enhanced S02 removal
levels there.
Figure 5 illustrates the effects of higher chloride levels (expressed as weight
percent chloride in the spray-dried solids collected in the pulse-jet fabric fil-
ter) on system S02 removal performance at a 2500 ppm inlet S02 level. The coal
fired at the Kintigh Station averages about 0.09 wt.% chloride. For operation at a
2500 ppm inlet S02 level and a 1.6 reagent ratio, this produces a chloride level of
about 0.4 to 0.5 wt.% in the spray-dried solids. The data in Figure 5 show that,
if the chloride level is raised to about 0.75 wt.%, the reagent ratio required to
achieve 90% overall S02 removal can be reduced from just under 1.6 to about 1.3.
By raising the chloride level all the way to 3 wt.%, 90% overall S02 removal can be
achieved at a 1.0 reagent ratio.
Figure 6 illustrates the effects of recycle solids chloride level on system S02
removal performance at an inlet S02 level of 1500 ppm. At this inlet S02 level,
the effects of elevated chloride levels are less pronounced, most likely because
the lime reagent utilization is already relatively high at the baseline chloride
level. At the baseline chloride level of 0.4 to 0.5 wt.%, 90% overall S02 removal
is achieved at a reagent ratio of about 1.1. By increasing the recycle solids
chloride level to 1.5 wt.%, 90% overall S02 removal is achieved at a 1.0 reagent
ratio. A further increase in recycle solids chloride level to 3.0 wt.% produces no
measurable improvement in S02 removal performance.
Remember that the baseline recycle solids chloride level produced from the 0.09
wt.% coal chloride content was reduced by about 10% to 15% by the air dilution of
the flue gas required to lower the pilot unit inlet S02 concentration to 1500 ppm.
The data plots for the 1.0 and 1.1 reagent ratios suggest that the baseline S02
removal levels in Figures 3 and 4 would have been 2 to 3 percentage points higher
if this dilution had not occurred.
Elevated chloride levels can be achieved in three manners. Coals with a higher
chloride content can be fired, or waters with a high chloride content can be used
4B-132
A
A
-------
for makeup to the process (but probably not for lime slaking), and/or calcium
chloride can be spiked into the system. Of these techniques, calcium chloride
spiking will most likely be the most costly. Coals with a higher chloride content
can often be purchased at lower prices than low chloride coals, and high-chloride
content waters are generally less expensive to use than high-quality, low-chloride
content makeup waters. Calcium chloride, however, must be purchased as a bulk
commodity chemical.
We have used the S02 removal performance data in Figures 5 and 6 to determine the
cost effectiveness of adding calcium chloride as a means of reducing system lime
consumption. The results of these calculations are summarized in Table 1. These
calculations show that, notwithstanding the capital cost impacts of installing a
calcium chloride injection system, calcium chloride can be cost effective as a
means of reducing spray dryer FGD system lime consumption even at the lower 1500
ppm inlet S02 level.
For a 300-MW plant, a 2500 ppm spray dryer inlet S02 level (3.5% to 4% coal sul-
fur), a 0.09% coal chloride content, and a 92% overall S02 removal level, raising
the recycle solids chloride level from 0.45 to 1.5 wt.% is most cost effective and
produces a net reduction in reagent costs of about $1.1 million per year. This
represents almost a 25% reduction from the cost of lime reagent for the base case.
For a 1500 ppm inlet S02 level, an increase from 0.5 to 0.75 wt.% chloride in the
recycle solids is most cost effective. In this case, the savings are predicted to
be $100,000 per year, which is only about 5% of the base case lime reagent cost.
Again, as mentioned above, if a less expensive chloride source is available, it may
be cost effective to operate at higher chloride levels in the recycle solids and
even higher lime utilization values.
Effect of Atomizer Feed Configuration. Host of the spray dryer pilot testing at
the HSTC has been conducted with the atomizer feed slurry being mixed and fed from
one tank. That is, lime slurry, recycle solids, and additional makeup water are
mixed in this tank in carefully controlled proportions, such that when the combined
slurry is fed to the atomizer at a rate to produce the desired spray dryer outlet
temperature, the desired reagent ratio will simultaneously be achieved. Testing
had been conducted in this mode because it was assumed that S02 removal performance
would be maximized by this configuration, in that the highest possible recycle
4B-133
-------
ratio (lbs of recycle solids per lb of fresh lime fed) could be achieved. However,
this feed mode is limited by a slow response time to changes in spray dryer inlet
flue gas temperature and/or S02 concentration. Much of the slurry in the tank must
be used before the concentrations of fresh lime, recycle solids, and makeup water
can be changed to the appropriate values for the new flue gas conditions.
The alternative is to mix makeup water and recycle solids in one tank, and store
slaked lime slurry in another. The atomizer feed then consists of a mixture of
these two slurries. The lime slurry feed rate is controlled to achieve the desired
reagent ratio, while the recycle slurry feed rate is used to control the spray
dryer outlet temperature. These two slurries can be mixed in a small tank with a
very short residence time, then fed to the atomizer, or they can just be combined
in the atomizer feed line. This feed mode is advantageous in that the spray dryer
FGD system has the capability of responding immediately to changes in system inlet
flue gas temperature or S02 concentration. However, it was assumed that for high-
sulfur-coal conditions, S02 removal would be lower in this feed configuration than
in the single-tank configuration. At high-sulfur-coal conditions, most of the
water fed to the spray dryer is that in the fresh lime slurry feed stream, which is
limited by viscosity constraints to about 25 to 30 wt.% solids. The ability to
recycle solids becomes limited by the fact that little additional water, besides
that associated with the lime slurry, can be fed to the atomizer while maintaining
the desired approach to adiabatic saturation at the spray dryer outlet.
However, in spite of these recycle ratio limitations, it is apparent that utilities
prefer the ability of the FGD system to respond quickly to flue gas condition
changes. Most vendors of spray dryer FGD systems in the U.S. offer a feed configu-
ration that is similar to the "dual-tank" method described above. Consequently, it
was decided to conduct a series of tests to document the effects of atomizer feed
slurry configuration on system S02 removal performance. The results of these tests
are summarized in Figures 7 and 8 for operation at 1500 ppm and 2500 ppm inlet S02
levels, respectively.
The results plotted in Figure 7 show no measurable effect of atomizer feed config-
uration on S02 removal performance. This is not surprising as, at a 1500 ppm inlet
S02 level, the water associated with the lime slurry accounts for less than half of
the total water feed rate requirement. Thus, the ability to recycle solids and
4B-134
A
A
-------
improve overall lime utilization is not significantly compromised by operating in
the dual-tank configuration.
The results in Figure 8 show a significant effect of atomizer feed configuration on
S02 removal performance at a 2500 ppm inlet S02 level and reagent ratio values of
1.3 and 1.6. At both reagent ratios, a penalty of about 7 to 10 percentage points
is realized when operating in the dual-tank rather than the single-tank mode. This
is also not surprising because, as the inlet S02 level increases, the water asso-
ciated with the lime slurry represents a greater percentage of the water fed to the
spray dryer, and the ability to recycle becomes greatly constrained. For example,
when the single-mix-tank tests are conducted with a 40 wt.% solids level in the
atomizer feed slurry for 2500 ppm inlet S02 conditions and a 1.6 reagent ratio, a
recycle ratio of 1.5 lbs of recycle solids per lb of fresh lime solids is achieved.
In comparison, in the dual-tank configuration, when the lime slurry is fed to the
atomizer at 24 wt.% solids and the recycle solids at 40 wt.%, the recycle ratio at
these conditions is reduced to only 0.5.
Since the lower S02 removal in the dual-tank atomizer feed configuration appears to
be due to the limited recycle ratio, a test was conducted to determine whether feed
slurry conditions could be changed to allow higher recycle ratios in this feed con-
figuration. In this test, the lime slurry solids content was raised to 26 wt.%,
and the recycle slurry solids content was raised to 50 wt.%. The results of this
test are included in Figure 8. With these changes, the recycle ratio was increased
from 0.5 to 1.3, and the overall S02 removal at a 1.6 reagent ratio was improved
from about 85% up to 93%. These results indicate that the difference in S02 remo-
val performance noted between the single- and dual-tank atomizer feed modes at a
high inlet S02 concentration of 2500 ppm can be virtually eliminated by operating
with the lime and recycle slurries at the maximum wt.% solids levels achievable.
For higher inlet S02 levels and/or for lower spray dryer inlet flue gas tempera-
tures, where there will be lower recycle slurry flow rates to achieve a given spray
dryer outlet temperature, the difference between the two feed modes may persist,
though.
Other Effects on S0= Removal Performance. In past operation of the spray dryer FGD
pilot plant at the HSTC, it had been noted that on occasion the S02 removal per-
formance of the system would improve markedly with no apparent explanation. For
k.
4B-135
-------
example, the data in Figure 4 show that, for a 2500 ppm inlet S02 level and a 1.6
reagent ratio, the overall S02 removal performance across the pilot unit was gen-
erally about 92% to 93%. On occasion, though, overall removal levels of 98% to
99% were measured at these conditions. After thorough investigations, it became
apparent that these increases were not due to lime source effects or coal chloride
variations. Also, solid waste product analyses confirmed the increased lime util-
ization values indicated during these incidents, discounting the possibility of
measurement errors being the cause.
After conducting numerous pilot unit tests and laboratory analyses, we now appear
to be able to control this effect, although we do not yet fully understand it. We
have found that, at these baseline conditions for the pilot unit (2500 ppm inlet
S02, 300*F inlet temperature, 1.6 reagent ratio, and 20#F approach to adiabatic
saturation at the spray dryer outlet), the overall S02 removal performance can be
varied from nominally 92% to 98% by careful control of the atomizer feed slurry
preparation process. Further testing is currently under way to develop a better
understanding of this effect and to document the effect over a wider range of pilot
unit operating conditions. Because these tests are on-going and the concept may be
patentable, only limited information can be presented here. However, Figure 9 sum-
marizes recent pilot unit data which illustrate the ability to control and repeat
this effect. In the figure, condition "A" represents normal spray dryer pilot unit
operating conditions and S02 removal performance, while condition "B" represents
tests where the conditions are controlled to result in elevated S02 removal perfor-
mance. While there is some scatter in the S02 removal performance data for the two
sets of conditions, the ability to improve S02 removal efficiencies by changing to
the "B" conditions is consistent.
Although this investigation is still underway, these first results are very encour-
aging in their implications for the future use of spray dryer FGD technology in
medium- and high-sulfur-coal applications. For example, it may be possible to re-
duce the lime requirements to achieve 90% overall S02 removal for a 2500 ppm inlet
S02 level, from the value of 1.5 to 1.6 as indicated in Figure 3 to a value of 1.3
or less. This would represent a 15% to 20% reduction in lime requirement with
little or no additional capital and operating cost impacts.
4B-136
A
A
-------
PJFF Operation
Pressure Drop Performance. Three different bag fabrics have been used in the PJFF
during spray dryer FGD testing: Daytex, Ryton/Ryton, and acrylic. At the begin-
ning of the spray dryer/PJFF configuration testing, Daytex bags were in service in
the PJFF. The bags had operated about six months prior to the spray dryer/PJFF
tests under baseline (fly ash only) conditions and HYPAS conditions (in-duct flue
gas humidification, followed by dry hydrated lime injection upstream of the PJFF).
At the beginning of the spray dryer/PJFF test period, it was noted that the Daytex
fabric had shrunk considerably during the previous six months of operation, such
that the fabric was stretched tightly over the cages. This shrinkage was later
attributed to improper heat treatment of the fabric prior to sewing the bags.
However, the shrinkage had the effect of greatly reducing the effectiveness of the
pulse-jet cleaning, as the fabric was stretched so tightly that it could not flex
as the cleaning pulse moved down the length of the bags.
Because these bags were not being cleaned effectively, PJFF air-to-cloth ratios
were held to 3.0 to 3.5 acfm/ft2. After nearly three months of spray dryer/PJFF
operation with the Daytex bags, the tubesheet pressure drop averaged 7.0 to 7.6 in.
H20 at an air-to-cloth ratio of 3.0 acfm/ft2 in spite of continuous cleaning of the
bags.
The second set of bags tested were constructed of a 19 oz/yd2 Ryton/Ryton fabric
that was singed/glazed on the outside (collecting) surface to minimize the poten-
tial for formation of thick or nodular residual dustcakes. To further ensure that
these bags would clean effectively, it was verified that the fabric was properly
heat set before the bags were sewn, the bags were sized for a looser fit on the
cages than the original Daytex bags, and 8-wire rather than the original 14-wire
cages were used.
These bags cleaned very effectively. After nearly three months of spray dryer/PJFF
operation, the tubesheet pressure drop averaged 4.5 in. H20 at an air-to-cloth
ratio of 4.0 acfm/ft2. During this period, the PJFF cleaning program was set for
cleaning to initiate at a tubesheet pressure drop of 5.0 in. H20, and the PJFF
cleaned approximately every 12 minutes.
L
4B-137
-------
The Ryton/Ryton bags suffered a number of failures that appeared to be due to abra-
sion wear along the vertical support cage wires. The abrasion wear was determined
to be a result of the relatively loose fit of the bags to the cages and inadequate
support of the bag material by the 8-wire cages. Consequently, when the PJFF was
rebagged in October 1990, the new acrylic bags were installed on more supportive,
14-wire cages, as were the original Daytex bags, and the bags were sized for a
tighter fit to the cages. Also, a considerable amount of effort went into quality
control to ensure a consistent fit of the bags to the cages. The acrylic bags were
singed on the collecting surface as were the Ryton/Ryton bags.
After nearly 7000 hours of operation downstream of the spray dryer, and at an air-
to-cloth ratio of 3.5 acfm/ft2, the pressure drop across the acrylic bags averages
5.5 in H20. Neither the Daytex nor the Ryton/Ryton bags were operated for this
duration downstream of the spray dryer, so it is not possible to compare pressure
drop performance by fabric type. The best comparison is to previous data for the
reverse-gas fabric filter operating downstream of the spray dryer. The average
tubesheet pressure drop for the PJFF with the acrylic bags is about 1 in. H20
greater than that of the reverse-gas fabric filter after a similar period of opera-
tion with new bags and downstream of the spray dryer. The reverse-gas fabric fil-
ter was cleaned every 90 minutes, but operated at an air-to-cloth ratio of only 1.6
acfm/ft2. This value is less than half the air-to-cloth ratio of the PJFF. From
this perspective, the pressure drop performance of the PJFF compares very favorably
with that of the reverse-gas unit over this relatively short period of time.
Pilot testing during the remainder of 1991 and in 1992 will continue to quantify
PJFF pressure drop performance downstream of the spray dryer with the acrylic bags.
Particulate Control Performance. Particulate removal data have been collected with
all three bag material types tested in the PJFF to date: Daytex, Ryton/Ryton, and
acrylic. The Daytex bags were in operation a total of nine months, but only about
three of those months were for operation downstream of the spray dryer. Over that
three-month period, PJFF outlet emissions remained quite low with measured emission
rates ranging from 0.002 to 0.010 lb/106 Btu (0.0007 to 0.0052 gr/acf) and averag-
ing 0.003 lb/106 Btu (0.0018 gr/acf). This represents substantially lower emission
rates than required by the current NSPS, in spite of the fact that the solids added
in the spray dryer raised the PJFF inlet loading to approximately 7 to 9 gr/acf.
4B-138
A
A
-------
The Ryton/Ryton bags were used for about three months in spray dryer FGD service in
1989 before the spray dryer was shut down to allow testing of another dry FGD tech-
nology. The PJFF outlet emissions measured were quite variable over this period
due to several problems related to the fabrication and installation of these bags.
The spray dryer/PJFF was operated with these bags again in 1990 for a period of
four months. During this time period, several bag failures occurred as a result of
abrasion along the 8-wire support cages, as described previously. However, outlet
emissions during periods where there were no failed bags in the PJFF ranged from
0.003 to 0.005 lb/106 Btu.
The pulse-jet fabric filter was rebagged in October 1990, this time with relatively
low-cost acrylic bags. For comparison purposes, the Ryton/Ryton bags for this
pilot unit cost about $90 each, while the acrylic bags cost about $40 each. A con-
siderable amount of effort went into quality control in specifying and fabricating
these bags. This quality control has proven very worthwhile; in over a year's
operation, not one of these 124 bags has failed, and pulse-jet fabric filter outlet
emissions have remained at 0.002 lb/106 Btu or less.
These data indicate that a PJFF should have no difficulty in meeting NSPS particu-
late emissions rates downstream of a spray dryer in FGD service. They also illu-
strate that, similar to experience with reverse-gas units, it is important that
proper quality control be exercised in the fabrication, installation, and mainte-
nance of the PJFF bags.
PJFF Corrosion. This is another area where the PJFF compares quite favorably with
the reverse-gas unit. At the 1988 FGD symposium, it was noted that there were two
modes of corrosion in the reverse-gas fabric filter downstream of the spray dryer.
One was general corrosion of the compartment walls on the clean side of the bags,
and the other was the failure of stainless steel bag clamps.
The design of the PJFF favorably addresses both of these corrosion problems.
Because there are no stainless steel components in the PJFF, no chloride-related
corrosion problems have been noted to date in the operation of the PJFF downstream
of the spray dryer. The PJFF has experienced some general corrosion of the walls
near cool spots on the clean side of the bags. However, the design of a PJFF is
such that the majority of the wall surface area is on the dirty side of the bags,
4B-139
-------
where excess lime solids are available to further insulate the metal as well as to
neutralize any dilute acid that may form due to moisture condensation. Thus, cor-
rosion tendencies must be dealt with over a much smaller area than in a reverse-gas
unit (e.g., about 10% of the affected area in a comparably sized reverse-gas unit).
During 1989 and 1990, a number of protective coatings were evaluated as a means of
controlling corrosion in this area. A catalyzed vinyl ester coating from Glidden
was selected as the most cost effective of the coatings in this initial evaluation.
When the PJFF was rebagged in October 1990, the PJFF outlet plenum area was sand-
blasted, and the vinyl ester coating was applied. After about one year of opera-
tion, this coated area remains relatively corrosion free, and the coating appears
to be unaffected by the flue gas conditions downstream of the spray dryer.
WASTE DISPOSAL CHARACTERISTICS EVALUATION
One goal of this pilot-scale evaluation of spray dryer/PJFF technology for medium-
to high-sulfur-coal applications is to determine disposal properties for the solid
waste stream produced. Recently, with the assistance of NYSEG, three test cells
were filled with spray dryer FGD system solid wastes to begin a field evaluation of
waste properties. The evaluation will determine properties such as strength devel-
opment, water permeation rates, and runoff and leachate water quality under real-
world landfill conditions.
These three cells were filled with wastes from operation at a 2000 ppm S02 concen-
tration to the spray dryer. Two were at baseline chloride levels and a reagent
ratio to achieve about 90% overall S02 removal. The third was filled with wastes
from a test conducted at a 1.0 wt.% chloride level in the recycle solids, also at a
reagent ratio to achieve 90% overall S02 removal. The 1.0 wt.% chloride level was
determined to be the cost-effective optimum for calcium chloride spiking at this
inlet S02 level. Wastes from an elevated chloride level test were selected for
evaluation because previous lab data suggested that higher chloride levels might
adversely affect waste solids strength development, permeability, and leachate
quality (4).
4B-140
A
A
-------
FUTURE TESTING
Spray dryer/PJFF tests are scheduled to continue into 1993. Most of the testing
will consist of short-term parametric tests to evaluate the S02 removal capabili-
ties of the process for a variety of medium- and high-sulfur-coal operating condi-
tions. Performance additives and other methods of improving S02 removal and lime
utilization by the process will be evaluated in an attempt to further reduce the
costs of future applications of this technology for medium- and high-sulfur coals.
During these short-term tests, a number of measurement efforts are planned to
determine the removal of air toxics by the spray dryer/PJFF pilot unit.
The final portion of this test program will include one or more longer-term tests
(approximately three months each). This testing will provide the opportunity to
better observe the pressure drop performance of the PJFF at conditions more like
utility spray dryer FGD operation. This is an important consideration that will
affect the evaluated costs of a spray dryer FGD system employing a PJFF for par-
ticulate control.
SUMMARY
After the most recent 16 months of operation of the spray dryer/PJFF pilot unit in
medium- to high-sulfur-coal FGD service, the following has been observed:
• The overall S02 removal performance of the FGD system is quite
dependent on spray dryer inlet S02 concentration. At the operating
conditions for these tests, the reagent ratio required to achieve
90% overall S02 removal across the pilot unit dropped from about 1.5
for a 2500 ppm inlet S02 concentration to about 1.1 at a 1500 ppm
inlet concentration.
• Calcium chloride appears to be a cost-effective additive for lower-
ing system reagent costs. For a 92% overall S02 removal level at a
2500 ppm inlet S02 concentration, calcium chloride addition can
lower system reagent costs by nearly 25%. At a 1500 ppm inlet S02
level, the reduction is less significant at 5%. However, if other
sources of chloride into the system are available (e.g., higher coal
chloride content, high-chloride content makeup waters), larger
reductions in lime reagent use may be realized with little or no
costs associated with raising the recycle solids chloride level.
• At a 1500 ppm system inlet S02 level, system S02 removal performance
does not appear to be affected by the atomizer feed slurry prepara-
tion configuration. As the inlet S02 level increases, a penalty in
S02 removal performance may occur when preparing the lime slurry and
k.
k
4B-141
-------
recycle slurry feeds in separate tanks rather than in a single tank.
If the weight percent solids levels in the two slurries is raised to
maximum levels though, such a penalty may be avoided.
• Further optimization of the atomizer feed preparation scheme, which
is currently under investigation, may yield further reductions in
the amount of lime required to achieve 90% overall S02 removal.
• After about a year of operation with new acrylic bags, low particu-
late emission rates and acceptable pressure drop values have been
measured for operation at a 3.5 acfm/ft2 air-to-cloth ratio.
These observations indicate that spray dryer/fabric filter FGD systems can be used
to achieve S02 control efficiencies of 90% and greater. At medium-sulfur-coal
conditions (1500 ppm inlet S02 concentrations), 90% S02 removal can be achieved at
lime utilization values approaching 90%. These results also indicate that pulse-
jet fabric filters can be used as an alternative to conservatively sized reverse-
gas units in utility spray dryer FGD applications. Since pulse-jet units are
capable of operating at air-to-cloth ratios that are approximately twice those of
reverse-gas units, the capital costs of the particulate control device can be re-
duced by about 30%. Hence, capital costs for the entire FGD system can be signifi-
cantly reduced by the use of a pulse-jet fabric filter rather than a reverse-gas
unit. Furthermore, the smaller size of a pulse-jet unit significantly reduces the
land area requirements for the particulate control device.
ACKNOWLEDGMENTS
The work reported in this paper is the result of research carried out in part at
EPRI's High Sulfur Test Center (HSTC) located near Barker, New York. We wish to
acknowledge the support of the HSTC cosponsors over this time period: New York
State Electric and Gas, Empire State Electric Energy Research Corporation, New York
State Energy Research and Development Authority, Consolidation Coal Company, and
the U.S. Department of Energy. The cosponsors provide valuable technical review of
the work in progress as well as funding for test center operations.
REFERENCES
1. Blythe, G., L. Lepovitz, and R. Rhudy. "Results from EPRI HSTC High Sulfur
Spray Dryer Pilot Tests," presented at the EPA/EPRI First Combined FGD and Dry
S02 Control Symposium, St. Louis, Missouri, October 1988.
4B-142
A
-------
2. Carr, Robert C. "Pulse-jet fabric filters vie for utility service." Power.
December 1988.
3. Blythe, G., L. Lepovitz, R. Rhudy, and R. Carr. "Results of EPRI's High Sul-
fur Test Center Spray Dryer/Pulse-Jet Fabric Filter Pilot Tests," presented at
the EPRI/EPA 1990 S02 Control Symposium, New Orleans, Louisiana, May 1990.
4. Blythe, G.M., J.M. Burke, D.L. Lewis, and C.M. Thompson. Field Evaluation of
a Utility Spray Drver System. EPRI CS-3954, Final Report, Electric Power
Research Institute, Palo Alto, CA, May 1985.
5. Karlsson, H.T., et al. "Activated Wet-Dry Scrubbing of S02." Journal of the
Air Pollution Control Association. Vol. 33, No. 1, January 1983, pp. 23-28.
6. Felsvang, K., H. Spannbauer, and P. Gedbjerg. "Scrubbing of Medium to High
Sulfur Coal - Industrial Operation Experience with Spray Dryer Absorbers."
1990 SO, Control Symposium. Vol. 2. Sponsored by U.S. EPA/EPRI, New Orleans,
LA, May 8-11, 1990.
7. Barton, R.A., et al. "S02 Removal Performance Improvements by Chloride Addi-
tion at the TVA 10-MW Spray Dryer/ESP Pilot Plant." 1990 SO.. Control Sympo-
sium. Vol. 2. Sponsored by U.S. EPA/EPRI, New Orleans, LA, May 8-11, 1990.
Table 1
COST EFFECTIVENESS OF CALCIUM CHLORIDE ADDITION
Plant Size
: 300 MW
Coal Heat
Content:
12,500 Btu/lb
Lime Price:
$50/ton
Coal Ash:
8%
Hours of Operation:
7,000
CaCl2 Price:
$214/ton
Chloride
S02
Lime
CaCl2
Total
Net
Reagent
Level
Removal
Cost
Cost
Cost
Savings
Ratio
(wt.%)
(%)
($MM/vr)
($MM/vr)
(SMM/vr)
($MM/vr)
3.8% S. 0.
09% CI. 2500
ppm Inlet
SO,
1.6
0.45
92
4.7
4.7
1.35
0.75
92
4.0
0.1
4.1
0.6
1.18
1
92
3.5
0.2
3.7
1.0
1.1
1.5
92
3.2
0.4
3.6
1.1
1.05
3
92
3.1
0.9
4.0
0.8
2.3% S. 0.
07% CI. 1500
ppm Inlet
SO,
1.17
0.5
92
2.1
--
2.1
1.1
0.75
92
1.9
0.1
2.0
0.1
1.07
1
92
1.9
0.1
2.0
0.1
1
1.5
92
1.8
0.3
2.0
0.0
1
3
92
1.8
0.6
2.4
-0.3
4B-143
-------
Temperature
Control
Flue Gas
From Boiler1
Clean
Flue Gas
Fabric
Filter
Or
ESP
Spray
Dryer
S02 t-
Injection
Lime
Silo
To Ash
Disposal
| Recycle
Silo
Lime
Slurry
Slurry
Feed
Pump
Paste Slaker
Make-up
Water
Slurry
Mix Tank
Lime
Storage
Tank
Figure 1. Simplified Process Flow Diagram for the HSTC
Pilot Spray Dryer FGD System (Single-Tank Atomizer Feed
Preparation Mode Illustrated)
Figure 2. Illustration of the Bag and
Cleaning Manifold Arrangement in the Low-
Pressure, High-Volume Cleaning Pulse-Jet
Fabric Filter at the HSTC
-------
4*
03
i
i-»
4*
CJl
100
90
80
70
? 60
_i
<
O 50
CM
o
40
30
20
10
/
LEGEND
1500 pfxn InM SOz
~ Overall Removal
I SDA Removal
A PJFF Removal
300 *F, 20 *F Approach to Saturation
3.6 Air-lo-Cloth Ratio, Miaeiaalppi Ume
a08 - 0.10 art % Coal Chloride
(0.3 to as wt.% a In Recycle Solids
1 Lines Repraeent Regreaalon FH of Data
I »
I I I
¦ till
0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9 2.1
REAGENT RATIO
(mols CafOHJj/mol S02 In)
100
90
80
70
£ 60
_i
<
>
O 50
UJ
oc
oN 40
(/)
30
20
10
A'
LEGEND
2500 2000 1500
ppm ppm ppm
~ ~ 0 Overall Removal
I D Q SDA Removal
AAA PJFF Removal
300 *F, 20 *F Approach to Saturation
3.5 AIMo-Cloth Ratio, Mississippi Lima
Lines Represent Rsgresslon Fit ol Data
¦ ¦ ¦ '
A H A A
0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9 2.1
REAGENT RATIO
(mols CafOH^mol S02 In)
100
90
80
£
J 70
<
>
O
2 60
UJ
oc
CM
O 50
w
< 40
oc
UJ
>
o 30
20
10
LEGEND
2500 ppm Inlet S02
4 1.6 Reagent Ratio
| 1.3 Reagent Ratio
A 1.0 Reagent Ratio
300 *F, 20 *F Approach to Saturation
3.5 AiMoCioth Ratio, Msslssippi Ume
Dual-Tank Configuration
Unas Represent Visual Fit to Data
_L
_U
_I_
0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
CI CONTENT OF RECYCLE SOLIDS (WT. %)
Figure 3. S02 Removal Results for
the HSTC Spray Dryer/PJFF Pilot
Unit for a 1500 ppm Inlet S02 Level
Figure 4. S02 Removal Results for
the HSTC Spray Dryer/PJFF Pilot
Unit for 1500, 2000, and 2500 ppm
Inlet S02 Levels
Figure 5. Effect of Recycle Solids
Chloride Content on S02 Removal
Across the HSTC Spray Dryer/PJFF
Pilot Unit for a 2500 ppm Inlet S02
Level
-------
£>-
Cd
h-»
ON
100
90
80
? 70
-j
<
> 60
O
2
HI
a 50
CM
O
0)
_j 40
<
0E
HI
>
O
30
20
10
LEGEND
1500 ppm InM S02
I 1.3 Raaganl Ratio
• 1.1 Raaganl Ratio
A 1.0 Raaganl Ratio
300 *F, 20 'F Approach to Saturation
3.5 Alr-to-CkXh Rat to, Miaalaalppl Lima
Dual-Tank Configuration
Una* Rapraoant Viaual FH ol Data
_L
_L
_L
0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
CI CONTENT OF RECYCLE SOLIDS (WT. %)
Figure 6. Effect of Recycle Solids
Chloride Content on S02 Removal Across
the HSTC Spray Dryer/PJFF Pilot Unit
for a 1500 ppm Inlet S02 Level
100
JL.
LEGEND
1500 ppm Inlat SOz
~ Ovarall Ramoval
I SDA Ramoval
A PJFF Ramoval
' 300 *F, 20 *F Approach to Saturation
> 3.5 Alr-to-CMh Ratio, Maalaaippl Lima
Point* ara for Dual-Tank
Unas Rapnamt Ragrawlon FH
of Ba—Una Slngla-Tank Data
_l L
_1_
JL.
I I I
_1_
0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9 2.1
REAGENT RATIO
(mols Ca(OH)2/mol S02 In)
Figure 7. Comparison of SO, Removal
Results for the HSTC Spray Dryer/PJFF
Pilot Unit for Single- versus Dual-
Tank Atomizer Feed Configurations at
a 1500 ppm Inlet S02 Level
-------
100
£>-
OJ
i
I-*
£>-
90
80
70
d 60
_j
<
>
O 50
UJ
oc
W 40
8
30
20
10
Dual-Tank
High Solids
Content
JL
LEGEND
2500 ppm Inlet S02
~ Overall Removal
I SDA Removal
A PJFF Removal
300 "F, 20 "F Approach to Saturation
15 AIMo-Clolh Ratio, Misalssippl Urn
Points am lor Dual-Tank
Unn Rapraaant Ragrasaion FH
ol Baaallna Singla-Tank Ma
I I
¦ it'll
X
I I
0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9 2.1
REAGENT RATIO
(mols CafOH^mol S02 In)
Figure 8. Comparison of SO, Removal
Results for the HSTC Spray Dryer/PJFF
Pilot Unit for Single- versus Dual-
Tank Atomizer Feed Configurations at
a 2500 ppm Inlet S02 Level
100
90
80
70
60
50
40
30
20
10
LEGEND
Condition Condition
"A" "B"
0 ~ Overall Removal
~ I SDA Removal
A A PJFF Removal
• 300 *F, 20 °F Approach to Saturation
• 3.5 Air-to-Cloth Ratio, Miaalaslppi Lima
¦ 2500 ppm Inlet SOi Laval
• Unas Rapraaant Ragraaalon Fit of
Baaallna Singla-Tank Data
IT
« ' tii 'iiiiiii¦i'i
0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9 2.1
REAGENT RATIO
(mols CafOH^mol S02 In)
Figure 9. Effect of Atomizer Feed
Slurry Preparation Conditions on S02
Removal Across the HSTC Spray Dryer/
PJFF Pilot Unit for a 2500 ppm Inlet
S02 Level
-------
Intentionally Blank Page
4B-I48
-------
RESULTS OF MEDIUM- AND HIGH-SULFUR COAL TESTS
ON THE TVA 10-MW SD/ESP PILOT PLANT
Thomas A. Burnett and Edward J. Puschaver
Tennessee Valley Authority
Robert A. Barton and Charles W. Dawson
Ontario Hydro
Richard G. Rhudy
Electric Power Research Institute
Gary Blythe and Katherine L. Heineken
Radian
Michael D. Durham
ADA Technologies, Inc.
Preceding page blank
4B-149
-------
Intentionally Blank Page
4B-150
-------
ABSTRACT
Since 1987, TVA has conducted a pilot-scale evaluation of
spray dryer (SD) flue gas desulfurization (FGD) technology,
for potential retrofit with an existing electrostatic
precipitator (ESP) for a medium- to high-sulfur coal
application. The Electric Power Research Institute (EPRI)
and Ontario Hydro (OH) are co-funders of this research. Some
results from this program have been presented at two previous
FGD symposia. This paper presents the results of testing
conducted in the latter half of 1990 and during 1991.
Previous results showed that the technology was capable of
achieving 95% S02 removal levels for a medium-sulfur (2.7%)
coal with a high chloride content (0.25 wt%) or with calcium
chloride spiking to equivalent levels. Recent results
presented in this paper indicate that calcium chloride
spiking can be employed to achieve 95% SO2 removal levels
with a high-sulfur (4.0%) coal as well.
In several previous papers, we have presented evidence that
ESP's downstream of SD FGD systems can be limited in
performance by a low resistivity reentrainment problem.
Recent results presented in this paper provide additional
confirmation of this effect, but also provide strong evidence
that high chloride levels in the spray dried solids (2.0
percent CI or greater) can virtually eliminate this problem.
Results of tests with other ESP performance additives are
also presented.
Preceding page blank
4B-151
-------
INTRODUCTION
Since 1987, TVA, along with co-sponsors EPRI and OH, has been
evaluating the SD/ESP FGD technology in a 10-MW pilot plant
located at the Shawnee Test Facility (STF) near Paducah,
Kentucky. A flow diagram for the 10-MW SD/ESP pilot plant at
the STF is shown in Figure 1.
The recent medium-sulfur coal phase of the SD/ESP test
program was designed to evaluate the SO2 and particulate
removal performance of the system at various chloride
levels. Previous work at the STF and elsewhere had
demonstrated that the chloride level in the SD/ESP system was
an important determinant of the SO2 removal efficiency,
perhaps more important than some other major process design
variables (e.g., inlet flue gas temperature, recycle rate,
etc.).
The major objectives of this test phase were to determine the
effect of the chloride level on the SO2 and particulate
removal efficiencies in the SD/ESP system over the ranges of
interest for the other variables and at differing coal sulfur
levels. The chloride level was changed by spiking the
recycle slurry with calcium chloride and/or by switching
coals.
Although the SO2 removal efficiency data were the highest
priority, the particulate removal efficiency and ESP outlet
4B-152
-------
grain loading data were also becoming a much more important
factor in this test program. Previous work at the STF had
indicated that high SO2 removal efficiencies (>90 percent)
were feasible at relatively low lime stoichiometrics
[1.20 moles Ca(OH)2/mole inlet SO2], but that the ESP
outlet particulate emission levels could be a problem.
These objectives were initially addressed through a
statistically-designed test program where chloride level and
the major process design variables (inlet flue gas
temperature, approach-to-saturation temperature, and lime
stoichiometry) were varied over the primary range of
interest. Chloride levels were varied in these tests by
spiking the SD feed slurry with a 35 weight percent calcium
chloride solution. The coal being burned contained
approximately 0.1 percent chloride.
In later test plans with other medium-sulfur coals and
high-sulfur coals, the effect of chloride on both SO2 and
particulate removal efficiency was explored further, also
using the calcium chloride spiking system.
S02 REMOVAL RESULTS
Medium-Sulfur Coal
The major finding of the medium-sulfur (2.7 percent) coal
work was that the chloride level in the system does have a
significant effect on the SD/ESP SO2 removal performance.
The magnitude of this effect was found to be best described
in terms of the chloride level in the recycle solids rather
than the equivalent coal chloride level. Although the
equivalent coal chloride level is much easier to understand
and use, the actual chloride level in the recycle solids
provides a more precise description of the chloride effect.
For example, for a given coal chloride level, changes in the
4B-153
-------
coal sulfur or ash level or changes in the SD operating
conditions can affect the chloride level in the recycle
solids and thereby affect the SO2 removal efficiency in the
SD/ESP system. These changes might not be adequately
described if the equivalent coal chloride variable was used.
At the baseline SD/ESP conditions of 320°F inlet flue gas
temperature, 18°F approach-to-saturation temperature, and a
1.3 mole Ca(OH)2/mole inlet SO2 lime stoichiometry, and a
medium-sulfur coal at the lower end of the sulfur range (2.2
percent S or 4.0 lbs S02/MBtu), the overall system SO2
removal efficiency was found to increase from about
80 percent at an equivalent coal chloride level of 0.02
percent to about 89 percent at a coal chloride level of 0.10
percent, to a maximum of about 95 percent at a coal chloride
level of 0.20 percent. These equivalent coal chloride levels
correspond approximately to chloride levels in the recycle
solids of 0.1, 0.7, and 1.4 percent, respectively. Further
increases above the 0.20 percent equivalent coal chloride
level seemed to result in only marginal increases in the
overall system SO2 removal efficiency.
For the medium-sulfur coal at the upper end of the range (2.7
percent S or 5.0 lb S02/MBtu) at the same baseline
condition (320°F, 18°F, 1.30 moles Ca(OH)2/mole inlet
SO2), the overall system SO2 removal efficiency was found
to increase from about 89 percent at a coal chloride level of
0.10 percent to about 96 percent at a coal chloride level of
0.20 percent, to a maximum of nearly 98 percent at a coal
chloride level of 0.30 percent. The corresponding chloride
levels in the recycle solids for these equivalent coal
chloride levels were approximately: 0.6, 1.1, and 1.7
percent, respectively.
Initially, the data were modeled as a function of the coal
chloride level (or the equivalent coal chloride level for
calcium chloride spiking tests) since this was a chloride
4B-154
-------
value that was easy to measure. Given the hypothesized
explanation for the chloride effect on the SD/ESP
performance, however, it was realized that the equivalent
coal chloride level was not the best variable to describe the
chloride effect. The best variable for predicting the SO2
removal efficiency in the system was found to be the chloride
level in the recycle solids since this reflects the actual
chloride level in the system.
This actual chloride level in the SD/ESP system is a function
of the lime stoichiometry and the coal sulfur, chlorine, and
ash levels. The coal chlorine level is important for obvious
reasons and will not be discussed further. The importance of
the coal sulfur level is due to the inherent characteristics
of the SD system. The amount of water that can be evaporated
in the SD is a function of the "spray down" temperature,
which is essentially independent of the coal sulfur level.
Higher coal sulfur levels require a higher lime slurry flow
rate to remain the same lime stoichiometry and the higher
lime slurry flow rate results in a corresponding reduction in
the recycle slurry flow rate. With less internal recycle,
the chloride level in the solids decreases even though the
equivalent coal chloride level may be the same. Thus, the
equivalent coal chloride level could be used as a major
variable to predict SO2 removal efficiency, if the coal
sulfur level remained the same. However, even with the same
coal, the chloride level in the recycle solids provides a
better fit of the data. The ash level in the coal also
affects the chloride level by increasing the portion of inert
material collected in ESP, there by "diluting" the chloride
level and crowding out some of the chloride-containing solids
that would otherwise be recycled.
Figure 2 is a plot of the overall system SO2 removal
efficiencies actually achieved during these chloride
evaluation tests and the projections made by an empirical
model developed from the test data. In this figure, the
4B-155
-------
overall system SO2 removal efficiency is shown as a
function of the chloride level in the recycle solids at
various lime stochometries for an 18°F approach-to-saturation
temperature, a 320°F inlet flue gas temperature, and
1,700 ppm inlet SO2 level. Since only a one-half factorial
test plan (with a full set of replicate tests) was run, some
conditions were not tested and thus, data were not available
for all of the conditions.
In this empirical model developed from the test data, the
overall system SO2 removal efficiency is a strong function
of: lime stoichiometry, chloride level in the recycle
solids, and approach-to-saturation temperature. It is also a
function of the inlet flue gas temperature and SO2
concentration, although to a lesser extent. The approach-to-
saturation temperature and chloride level variables are
included both separately and also as part of separate
two-factor interaction terms. The coefficient of
determination (R2) for the model developed from the data
and plotted in Figure 2 is 0.94, indicating that the model is
very good at explaining the variability in the data. The
mean square error (MSE) for this model is 2.3 percentage
points, indicating the model projections at the 95 percent
confidence level have an error band of +4.6 percentage points.
High Sulfur Coal
A series of tests were run with two separate high-sulfur
(8.0 lb S02/MBtu), very low chloride (<0.04 percent)
coals. With no calcium chloride spiking, the overall system
SO2 removal efficiencies were relatively low, i.e., only
about 75-80 percent at an 18°F approach-to-saturation
temperature and a 1.30 lime stoichiometry. However, spiking
with calcium chloride dramatically improved the SO2 removal
performance of the SD/ESP system. With spiking to the
equivalent of 0.7 percent chloride in the high-sulfur coal
(2.5 percent CI in the recycle solids), the overall system
4B-156
-------
SC>2 removal efficiency was nearly 95 percent at a 320°F
inlet flue gas temperature, an 18°F approach-to-saturation
temperature, and a 1.30 lime stoichiometry. Increasing the
lime stoichiometry to 1.60 moles Ca(OH)2/mole inlet SO2
with all other variables at the same level only increased the
overall system SO2 removal efficiency to 98 percent.
PARTICULATE REMOVAL RESULTS
The particulates and the flue gas characteristics produced by
the SD should provide for excellent ESP operating conditions.
The flue gas temperature decreases due to the SD operation,
which results in a significant increase in the ESP electric
field strength in those fields that are not already current-
limited. It also results in a 15 percent increase in the
effective specific collection area (SCA) in the ESP because
of the reduction in the flue gas temperature and volume
compared with fly-ash-only operation. The flue gas
temperature decrease also results in a decrease in both the
flue gas velocity and viscosity in the ESP. The combination
of this flue gas cooling and the increased moisture content
also lowers the resistivity of the particulate matter
entrained in the flue gas entering the ESP. Finally, the
size of the particulate matter entering the ESP also
increases from about 5 micron MMD mass mean diameter (MMD) at
baseline conditions to approximately 15 micron MMD with the
SD in operation. All of these factors should lead to
enhanced ESP performance.
The operation of the SD does, however, significantly increase
the grain loading in the flue gas entering the ESP. From a
typical value of 0.5-1.0 grains/acf for fly-ash-only
operation at the STF (this fly ash loading is low due to the
upstream mechanical collectors), the inlet grain loading
increases to 5.4-9.7 grains/acf, i.e., by a factor of nearly
10, with the SD operation. At other sites with a higher
baseline ESP inlet grain loading, the percentage increase in
4B-157
-------
grain loading with SD operation would be somewhat smaller,
but the outlet grain loading would be nearly the same.
Medium-Sulfur Coal
In spite of the improvement in most ESP operating parameters
and the increase in the amount of what is apparently easily
collected material, the particulate removal efficiency in the
ESP did not improve once the SD began operating. The tenfold
increase in the inlet grain loading to the ESP typically
resulted in a similar tenfold increase in the outlet grain
loading. Clearly, unless the ESP performance could be
improved, this increase in the particulate emissions could
prevent the widespread application of this technology in
retrofit situations.
The attempts to improve the ESP performance followed two
simultaneous tracks. First, the mechanical and operating
characteristics of the ESP were studied in detail. The
primary purpose of this inspection was to check the
mechanical condition of the ESP to reduce the sneakage and
reentrainment losses. Several minor problems (e.g., missing
hopper baffles, too much clearance between baffles and the
walls and floor, etc.) were corrected.
Second, it was noted in the data that there was an apparent
correlation between the emission rate (in lbs/MBtu) and the
chloride level in the system at the lower
approach-to-saturation temperature (18°F). No similar trend
was noted in the data taken at the 28°F
approach-to-saturation temperature.
This apparent chloride effect on the ESP performance was
hypothesized to result from the hygroscopic nature of calcium
chloride. As the concentration of chloride in the
particulate matter was increased, the amount of moisture in
the particulate matter increased. This increased moisture
4B-158
-------
level made the particulates more cohesive (i.e., "sticky"),
such that the particulates were better held, both to
themselves and to the plates and thus, were less likely to
become reentrained. This apparent chloride effect on ESP
performance was further evaluated in additional testing.
The baseline particulate removal efficiency with a properly
designed and operated ESP, i.e., after the ESP modifications
were made, was measured at 99.73 percent in a normal
fly-ash-only operating mode. With the SD operating at low
recycle solids chloride levels (0.2 percent or less), the ESP
particulate removal efficiency was found to range from 99.80
to 99.85 percent, depending on the test conditions. However,
this ESP performance was still not sufficient to avoid
prevention of significant deterioration (PSD) problems due to
the increase in particulate emissions after the SD is
retrofit. As shown in Table 1, even though the ESP achieved
a slightly higher particulate removal efficiency with the SD
in operation, the large increase in the inlet grain loading
led to an increase in the outlet grain loading and the
particulate emissions from the system.
To maintain the same level of particulate emissions after a
SD system was retrofit on a utility boiler, the particulate
removal efficiency in this case had to be increased to about
99.98 percent. Since the SD/ESP had demonstrated that it
could achieve SO2 removal efficiencies of 90-95 percent at
lime stoichiometries of 1.2 to 1.3, the focus of the test
plan shifted to improving the performance of the ESP.
A series of tests were conducted to determine the extent to
which chloride spiking could be used to reduce the
particulate emissions from the ESP even further. At a
chloride level in the recycle solids of 3.5 percent (0.5
percent on an equivalent coal chloride basis), the
particulate removal efficiency was boosted to the target
level of 99.98 percent and the particulate emission rate was
4B-159
-------
reduced to 0.005 lb/MBtu. Thus, addition of calcium chloride
was found to enhance the ESP performance, such that the
outlet grain loading from the SD/ESP system is comparable to
that achieved by the ESP at fly-ash-only conditions.
Compared to the removal levels for the low-chloride tests
(99.80-99.85 percent), it is clear that the increase in the
chloride level resulted in marked improvement in ESP
performance. However, this affect was only demonstrated at a
relatively close approach-to-saturation temperature (18°F).
To further understand this theory for the chloride effect on
ESP performance, a short series of tests was run at various
chloride levels in the recycle solids and other SD
conditions. The resulting data is shown in Figure 3 where
the penetration (100-percent removal efficiency) data from
this short series of tests are plotted as a function of the
chloride level in the recycle solids for a medium-sulfur (2.2
percent) coal. Although there is some scatter in this
limited amount of data, perhaps due to the difficulty in
accurately analyzing for the chloride level in the recycle
solids, the data appears to show a consistent improvement in
performance as the chloride level increases. Note that since
the penetration is plotted on a logarithmic scale, the
increase in particulate removal efficiency is much more
significant than a cursory review of this figure might
indicate.
The data in this figure also seems to indicate that there is
a threshold level below which, slight increases in the
chloride level do not seem to affect the ESP performance.
Once the chloride level in the recycle solids increases above
this threshold value, which is somewhere between 0.6 and 1.7
percent, the chloride effect on ESP performance is readily
apparent. These two chloride levels are comparable to
equivalent coal chloride levels of 0.1 and 0.2 percent for
this 2.2 percent sulfur coal as shown in Figure 3.
4B-160
-------
The similar data from some of the earlier chloride spiking
tests, which were run before the ESP modifications were made,
are also shown in this figure, as the 2.7 percent sulfur
coal. Although both sets of data show that increasing the
chloride level in the recycle solids improves the ESP
performance, these tests at the higher sulfur level show
higher particulate emissions for each chloride level. The
difference in the ESP performance between the two series of
tests is being attributed to the ESP modifications that were
made before the lower sulfur coal tests were run.
One other series of medium-sulfur coal tests was conducted in
an attempt to improve the ESP performance. Another additive,
which had previously been successfully tested in a
laboratory-scale SD system, was evaluated at the STF. The
additive was also found to enhance the ESP performance as
shown in Table 1. With this additive, the particulate
removal efficiency increases from the baseline SD value of
99.81 percent to 99.93 percent. Since this additive is
thought ' to enhance the ESP performance by a different
mechanism than chloride, future test work combining calcium
chloride and this additive is planned.
High-Sulfur Coal
A series of high-sulfur coal tests were run to determine if
this chloride effect on the ESP performance continued at
higher sulfur levels. The results of these tests are shown
in Table 2. At the fly-ash-only condition (i.e., with the SD
out of service), the particulate removal efficiency in the
ESP was 99.87 and the outlet grain loading was 0.002 gr/acf.
This corresponds to an emission rate of 0.008 lb/MBtu.
In the second test in the series with the SD operating, but
no calcium chloride addition, the ESP particulate removal
efficiency remained essentially constant at 99.86 percent.
However, because of the higher inlet grain loading with the
4B-161
-------
SD operating (11.3 versus 1.9 gr/acf), the outlet grain
loading increased to 0.014 gr/acf, a seven-fold increase.
The corresponding emission rate was 0.045 lb/MBtu.
However, by spiking the SD/ESP system with sufficient calcium
chloride to be comparable to a coal containing 0.7 percent
chloride (3.5 percent chloride in the recycle solids), the
particulate removal efficiency increased to 99.96 percent and
the outlet grain loading declined to 0.003 gr/acf. The
corresponding emission rate was 0.009 lb/MBtu, very near the
goal of maintaining emissions at the fly-ash-only level.
DISCUSSION OF RESULTS
The SO2 and particulate removal performance of the SD/ESP
system can be enhanced by adding chloride to the system.
This chloride can come from the coal, calcium chloride
addition, or by using high-chloride content makeup water.
Potential sources of high-chloride water could include
brackish or seawater or high-chloride content blowdown from
other operations. The purge stream from a prescrubber in a
wet limestone scrubbing system might be an appropriate
source, if available, especially since the alternative
methods for treating this prescrubber blowdown stream are
expensive.
The key variable for tracking the effects of increased
chloride levels on SO2 and particulate removal appears to
be the chloride level in the recycle solids. The SD and
overall system SO2 removal efficiencies in the SD/ESP
system improve as the chloride level in the system
increases. This improvement is particularly apparent with
the first incremental increases from a low baseline chloride
level. The effectiveness of chloride in enhancing the SO2
removal efficiency in the SD/ESP system is thought to be due
primarily to the ability of calcium chloride to slow the
4B-162
-------
evaporation of water from the slurry droplets and producing
higher residual moisture levels in the material collected in
the ESP.
However, once the chloride level in the recycle solids
exceeded about 2.0 percent, there was no further, significant
increase in the overall system SO2 removal efficiency.
This threshold level in the recycle solids was achieved when
the equivalent coal chloride level for a medium-sulfur coal
(4.0-5.0 lbs S02/MBtu) exceeded 0,20-0.25 percent.
However, the equivalent coal chloride level required to
produce the 2.0 percent chloride level in the recycle solids
is dependent on the actual sulfur level in the coal and the
SD operating conditions, primarily lime stoichiometry. For a
high-sulfur coal (8.0 lb S02/MBtu), this threshold chloride
level in the recycle solids was achieved when the equivalent
coal chloride level exceeded about 0.4-0.5 percent.
Similarly, the ESP performance increases with increasing
chloride level in the SD/ESP system. In contrast to the
chloride effect on SO2 removal efficiency, the ESP
performance continued to increase with increasing chloride
level up to the highest chloride levels tested, although at
that level the particulate removal efficiency was 99.99
percent and further increases are unlikely. The
effectiveness of chloride in enhancing the ESP performance is
thought to be due to the hygroscopic nature of calcium
chloride. The resulting particulates tend to be "sticky" and
thereby reduce reentrainment problems.
Based on the data from the STF, this ESP enhancement begins
when the chloride level in recycle solids exceeds about
0.6-1.7 percent, which is equivalent to a coal chloride level
of 0.1-0.2 percent for a medium-sulfur coal. This ESP
performance continues to increase as the chloride level in
recycle solids increases up to the maximum 3.5 percent level
4B-163
-------
tested. In one test when the chloride level in the recycle
solids inadvertently exceeded 4 percent, materials handling
problems (a plugged screw conveyor) prevented completion of
the test.
The chloride effect on both SO2 and particulate removal
efficiency seems to require a close approach-to-saturation
temperature (18-23°F) to maximize its effectiveness. This
close approach will require more careful monitoring of the SD
operation to prevent upset conditions from causing operating
problems.
The use of calcium chloride to enhance the performance of the
SD/ESP does, however, result in another adverse side effect
beside this increases potential for plugging. The atomizer
wheel must be fabricated from a material that is resistant to
chloride attack. The STF has used a titanium wheel for more
than 15 months at extremely high chloride levels with no
apparent damage to the wheel.
4B-164
-------
ADDITIONAL READING
1. T. A. Burnett, et al., 10-MW Sprav Drver/ESP PilotPlant
Test Program. Medium-Sulfur Coal Test Phase (Phase I\H .
Tennessee Valley Authority, Chattanooga, Tennessee, 1990.
2. M. D. Durham, et al., "Identification of Low-Resistivity
Reentrainment in ESP's Operating in Dry Scrubbing
Applications," Eighth Symposium on Transfer and
Utilization of Particulate Control Technology, San
Diego, California, 1990.
3. R. A. Barton, et al., "SO2 Removal Performance
Improvement by Chloride Addition at the TVA 10-MW Spray
Dryer/ESP Pilot Plant," SO2 Control Symposium, New
Orleans, Louisiana, 1990.
4. M. D. Durham, et al., "Improved ESP Performance in Spray
Dryer Applications Through the Use of Additives," Annual
Meeting of the Air Waste Management Association,
Vancouver, Canada, 1991.
5. T. A. Burnett, et al., 10-MW Sprav Drver/ESP Pilot Plant
Test Program. Medium-Sulfur Coal Chloride Evaluation
(Phase V). Tennessee Valley Authority, Chattanooga,
Tennessee, 1991.
4B-165
-------
ESP
Boiler
Spray
Dryer
Coal
.D. Fan
Stack
Air
H2O
To Disposal
Tank
Lime
Tank
Bin
Slaker
Tank
Tank
Figure 1. Spray Dryer/ESP Process with Chloride Addition
-------
120
110
100
90
80
70
60
50
40
30
20
10
~ 1.00 SR
A 1.30 SR
O 1.60 SR
Other Conditions
320°F Inlet Temperature
18°F Approach Temperature
1,700 ppm SO2
Maximum Recycle
0.2 0.7 1.2 1.7 2.2 2.7
Chloride Level in Recycle Solids, %
3.2
Figure 2. Overall System SO2 Removal Efficiency as a Function of
Chloride Level and Lime Stoichiometry
4B-167
-------
~£>
03
i
t—i
ON
oo
8*
C
O
0
C
0
Q.
0.1
0.10% CI
Sulfur
0.5% CI
0.01
T
12 3 4
Chloride Level in Recycle Solids, %
Figure 3. Effect of Chloride on ESP Performance
-------
Table 1. Chloride Effect on ESP Removal: Medium-sulfur Coal
ESP Grain
Particulate
Emission
Loading, gr/ac/
Removal
Rate,
Inlet
Outlet
Efficiency, %
Ib/MBtu
Fly Ash Only
0.56
0.002
99.73
0.005
Baseline SD
6.93
0.013
99.81
0.043
(No Chloride)
Chloride
6.81
0.002
99.98
0.005
Enhanced SD
Other Additive
10.31
0.006
99.93
0.019
-------
Table 2. Chloride Effect on ESP Removal: High-sulfur Coal
ESP Grain
Particulate
Emission
Loading, gr/ac/
Removal
Rate,
Inlet
Outlet
Efficiency, %
Ib/MBtu
Fly Ash Only
1.92
0.002
99.87
0.008
Baseline SD
11.28
0.014
99.86
0.045
(No Chloride)
Chloride
9.96
0.003
99.96
0.009
Enhanced SD
-------
EVOLUTION OF THE B&W
DURAJET™ ATOMIZER
S. Feeney,
R. Martinelli,
R. Myers
Environmental Equipment Division
The Babcock & Wilcox Company
20 S. Van Buren Avenue
Barberton, Ohio 44203
R. Bailey
Alliance Research Center
The Babcock & Wilcox Company
1562 Beeson Street
Alliance, Ohio 44601
B-171
k.
-------
Intentionally Blank Page
4B-172
-------
ABSTRACT
Dry flue gas desulfurization FGD was pioneered in the late 1970's.
From early years, when y-jet atomizers were standard, through the larger ranges of
DuraJet™ atomizer sizes (up to 100 MW) available today, the dual-fluid atomizer has
undergone a metamorphosis.
This paper reviews the history of dual-fluid atomizer technology. Also presented is
the progression from oil-fired y-jet to front-mix DuraJet™ atomizer technology for
humidification, dry FGD, dry acid gas removal and oil firing applications.
HISTORY
In the late 1970s, dry scrubbing processes for use in utility flue gas desulfurization
systems began to gain increasing acceptance.
Babcock & Wilcox's (B&W's) initial involvement with dry scrubbing began in 1977(1).
At that time, Basin Electric Power Cooperative was in the process of soliciting bids
for a dry scrubbing system at Antelope Valley Unit 1, a 440 MW unit. In response to
this development, an 8000 acfm pilot was designed and placed into operation at Basin's
W.J. Neal Station located in Velva, North Dakota.
In addition to considering such variables as low pressure drop, simplicity of
operation/maintenance, and flexibility in equipment when designing a dry scrubbing
system, major emphasis had to be placed on atomization quality(z). The y-jet dual-
fluid atomizer was believed to be well suited for this dry scrubbing application.
Based on the promising results from the pilot tests, in the fall of 1978, Pacific
Power and Light Company (known now as Pacificorp Electric Operations) agreed to make
a site available for the construction of a 20 MW demonstration plant at the Jim
Bridger Station located in Point of Rocks, Wyoming. The demonstration unit, sized
for 120,000 acfm of flue gas, included a S0Z absorber module, electrostatic
precipitator and baghouse. Though originally designed for six y-jet atomizers,
ultimately only one atomizer, centrally located in the front wall of the S0Z absorber,
formed the basis of the larger capacity y-jet.
Two commercial units were sold about this time. Basin Electric's Laramie River Unit
3, a 570 MW dry scrubber was scheduled for initial operation in the fall of 1982.
Colorado-Ute Electric Association purchased a 450 MW dry scrubber system, with initial
operation slated for the fall of 1983. Both commercial units utilized multiple y-jet
atomizers in a co-current, horizontal flow absorber. These y-jets were similar to
those tested at the Bridger demonstration unit.
While the y-jet dual fluid atomizer proved to be efficient and easy to maintain, the
small orifice holes tended to become plugged. In an attempt to alleviate this
condition, in 1985, a new type of I-jet atomizer was designed and tested, unlike the
Preceding page blank
k.
4B-173
-------
rear-mix I-jet atomizer being sold for fuel-oil atomization. This was quickly adopted
as the atomizer of choice due to its larger exit orifice holes, and lower pressure
operation.
In 1986, as part of the Clean Coal Technology LIMB (jjmstone Injection Multistage
fiurner) demonstration, a humidification system was designed. This used multiple
"mini" DuraJet™ atomizers housed within an airfoil lance assembly. For
humidification applications, the "mini" DuraJet™ atomizer must produce much finer
droplets to facilitate evaporation, since much less residence time is provided between
the plane of atomization and precipitator inlet. The mini-DuraJet™ atomizer end cap
is roughly one inch in diameter and about 1% inches long. In 1987, multiple "mini"
DuraJet™ atomizers were incorporated into the design of the 5 MW E-S0X pilot at Ohio
Edison's Burger Station in Dilles Bottom, Ohio, another Clean Coal Project.
In 1989, at the urging of Basin Electric Power Cooperative, a large DuraJet™ atomizer
(nominal 100 MW size) was built, installed and operated in a downflow, dry scrubber
module at the Antelope Valley Station. With 54 exit orifices, and an end cap
approximately 18 inches in diameter and 5 inches in height, this proved the viability
of operating a large dual-fluid atomizer in a commercial application while maintaining
S02 performance.
Also in 1989, operations began on the Limestone Injection Dry Scrubber (LIDS) pilot,
which utilized a single low pressure drop "mini" DuraJet™ atomizer. Pressure drop
was reduced by altering the air and exit orifices. The final report on this completed
pilot project has been issued.
Currently, the Department of Energy's (DOE) Duct Injection Prototype development work
continues, with concentration on computer modelling of Duct Injection phenomenon, and
atomizer testing at the Alliance Research Center in Alliance, Ohio.
CONSTRUCTION
The y-jet atomizer used for early pilot work at the W.J. Neal Station is shown in
Figure 1.
The inner barrel of this early dry scrubber y-jet, which carried the slurry, was about
Vi inch in diameter. The annular air chamber, surrounding the inner slurry barrel,
was approximately 1% inches in diameter. The end cap was drilled with six (6)
entrance orifices each for air and slurry, in very close proximity to one another.
The major benefit to this design was its simple construction. There were a total of
six exit orifices, each roughly % inches in diameter.
While adequate for the initial pilot tests, the demonstration work at Bridger
necessitated the use of a larger capacity y-jet atomizer. Initial testing utilized
a single-tiered y-jet atomizer which had 16 exit orifices. To avoid the tendency for
the sprayed jet to collapse, vent air was drawn through the center of the atomizer.
While the jet did not collapse in on itself, rapid deposition of solids on the end
cap initiated the search for an improved y-jet atomizer.
The single-tiered y-jet atomizer installed at Laramie River was replaced soon after
start-up with the two-tiered y-jet atomizer design (Figure 2). A total of 12 orifices
for both air and slurry were required, with two distinct planes for atomized slurry
to exit. The benefit afforded by this design was the capability of varying the
included angles of spray between the two planes to optimize the mixing between the
flue gas flowing through the turbo-diffuser and atomized spray from the y-jet.
Numerous drilling patterns were tested through the years, with the goal of maximizing
atomization quality and spray pattern, while reducing slurry deposition.
4B-174
A
A
-------
Eventually a drilling pattern was adopted for use on full scale units. With an eye
toward eliminating plugging, the first commercial scale front-mix I-jet, later to
become known as the DuraJet™ atomizer, was tested at Colorado-Ute Electric
Association's Craig Station, Unit 3 (Figure 3). The DuraJet™ atomizer designed for
this application contains only six larger exit orifices. The mixing of atomizing
medium and slurry initially takes place immediately upstream of the end cap in the
mix chamber.
With its ease of fabrication, excellent wear resistance, and simple maintainability,
the commercial DuraJet™ atomizer, capable of approximately 10 MW of dry scrubbing
capacity per atomizer, has been operational in two commercial units for over six
years.
Start-up of the LIMB project at Ohio Edison's Edgewater Station was accompanied by
degraded precipitator performance. A humidification system was designed to lower ash
resistivity. Much finer atomization was required than that of the commercial scale
DuraJet™ atomizer. The patented "mini" DuraJet™ atomizer was developed and installed
as part of this Clean Coal Project. Figure 4 shows multiple "mini" DuraJet™
atomizers housed within the low gas side pressure drop airfoil lance assembly.
The large capacity 100 MW DuraJet™ atomizer which was given an operability test in
one module at Basin Electric's Antelope Valley Station is shown in Figure 5.
WEAR RESISTANT MATERIALS
The y-jet atomizer initially tested at the W.J. Neal Station's pilot dry scrubber
incorporated wear-resistant tungsten carbide exit orifice inserts. While exhibiting
acceptable wear characteristics when atomizing lime slurry alone, wear rates increased
when atomizing recycled solids. Once testing began at the Jim Bridger 20 MW
demonstration dry scrubber, the abrasive/erosive nature of ash recycle material became
obvious. A program was initiated to test various materials under actual operating
conditions, attempting to find the most suitable wear-resistant material for this
application. Among the materials tested were different grades of aluminum oxides,
silicon carbides, tungsten carbides and titanium diboride.
After months of testing, sintered silicon carbide provided the longest wear-life:
approximately 21 days, however this material was deemed unacceptable from a commercial
operations standpoint.
The next ceramic material investigated in the y-jet atomizer showed negligible wear
after approximately 31/a months of operation. These Black Ceramic™ inserts were
installed in the y-jet atomizers at Colorado-Ute's Craig Unit 3, and are the current
wear-resistant materials used in the DuraJet™ atomizers installed today. The inserts
are 5/16" thick, while the wear pad is 3/8" thick.
In the future, investigation of advanced wear-resistant materials will continue, which
may offer even greater longevity for the DuraJet™ atomizer.
TESTING
During the development of the DuraJet™ atomizer, performance tests were conducted at
the Alliance Research Center Atomization Test Facility. This large scale facility
was constructed in 1982 for the purpose of determining the droplet size distribution
of atomized sprays from both mechanical and dual-fluid atomizers.
4B-175
-------
The Atomization Test Facility consists of a spray chamber along with an assortment
of auxiliary equipment and instrumentation. The spray chamber has a cross-section
of 8 X 8 feet and is 10 feet long. Two 4X8 foot windows mounted in opposing walls
provide visual access to the atomized spray for laser diagnostic measurements and
visual observation. Air supplied by a forced draft fan enters the spray chamber
through the windbox and carries the atomized droplets to a series of demisters. The
droplets collected by the demisters are collected in a hopper for recycling or
disposal. The clean air is then vented to the atmosphere.
The atomization facility is capable of testing full scale mechanical and dual-fluid
atomizers with liquid capacities in excess of 20 gpm. The size distribution of the
atomized droplets are measured with a Malvern Droplet and Particle Size Analyzer.
This system spans a nominal size range of 11 to 1100 microns, which can be adjusted
depending on the specific application.
The original DuraJet™ atomizer evolved from design criteria that reflected operating
requirements and experiences at Laramie River and Craig. These criteria included the
design pressures and flow rates, average droplet size and spray distribution,
resistance to wear and plugging and non-catastrophic failure due to a plugged air hole
or discharge hole. The design criteria for the Craig atomizers, summarized in Table
1, were the most restrictive due to their low operating pressures.
Figure 6 provides a comparison of the droplet size performance of the DuraJet™
atomizer and the y-jet atomizer that it replaced. This figure shows the spray average
Sauter Mean Diameter as a function of the atomizer air-to-water mass flow ratio. As
indicated, the DuraJet™ atomizer provided a smaller average droplet size at any given
air-to-water ratio compared to the y-jets that were previously used commercially.
Figure 7 provides a comparison of the liquid pressure characteristics of the two
atomizers over a range of liquid flow rates. With the y-jet atomizer, liquid flow
was limited to about 772 gpm. The DuraJet™ atomizer provided a full range of flows,
up to 17 gpm, at pressures lower than the maximum available. The end result of the
development effort, the DuraJet™ atomizer, operates at pressures 10 to 30 psig lower
than the y-jet atomizer, with equal or better drop size distribution at the same or
lower A/W ratios. This was accomplished while increasing internal and discharge
orifice diameters by factors of three or more.
During the development of the DuraJet™ atomizer, the operating characteristics were
analyzed to determine a general technique for sizing the atomizer for different
applications. These techniques were used to design the "mini" Durajet™ atomizer for
the LIMB Humidifier installed at Ohio Edison's Edgewater Power Plant located in
Lorain, Ohio.
This "mini" DuraJet™ atomizer design was significantly different in terms of capacity
as well as required droplet size. Liquid capacity was reduced from 16.8 gpm to 0.8
gpm, the air-to-water ratio was increased from 0.11 to a design of 0.45, and the
required Sauter Mean Diameter was decreased from 55 microns to a target 25-35 microns.
The droplet size performance for the "mini" DuraJet™ atomizer used for LIMB
Humidification, along with the DuraJet atomizer developed for commercial dry scrubbers
are given in Figure 8. Comparison of the data for the two atomizers shows that the
Sauter Mean Diameter of the larger atomizer is greater, but only by about 15 microns
at the same A/W ratio. About half of this variance in droplet size can be attributed
to the difference in operating pressures of the atomizer.
The data in Figure 8 show the DuraJet™ atomizer can be scaled in capacity and droplet
size to accommodate a wide range of applications.
4B-176
-------
ROTARY VS. DUAL-FLUID ATOMIZATION
The use of rotary atomization for dry scrubber applications has been the predominant
choice in utility applications*3*. When the majority of utility dry scrubbers were
installed in the U.S., the choice of rotary atomization was due to the large base of
experience in spray drying processes.
In rotary atomization, the feed slurry is fed by gravity into the cavity of a high
speed (8,000-10,000 rpm in utility FGD systems) rotating disc. The centrifugal energy
created by the spinning disc creates droplets through shear forces at the wheel
perimeter. The wheel diameter, rotational speed and number of inserts are based on
the slurry flow rate and droplet size required.
A typical atomizer assembly consists of a 3600 rpm motor coupled to a speed-increasing
gearbox, providing the higher rotational speeds of the atomizer wheel required in dry
scrubbing applications. The atomizer wheel can be directly driven off the gearbox,
or an atomizer spindle can be coupled to the gearbox to isolate the gearbox from hot
flue gases entering the reaction vessel(4). Both of these designs find commercial
application. As rotational speed increases, droplet size decreases.
In dual-fluid atomization, compressed air at 80-120 psig provides the energy necessary
for fine droplet production. As the compressed air exits the nozzle orifice, the
rapid increase in velocity creates shear forces that break the fluid into fine
droplets. In general, as air pressure increases, droplet size decreases.
Early experiences with rotary atomizers in dry scrubbing applications indicated that
the design criteria used would need to be improved and refined to meet the needs of
the air pollution control industry. In order to minimize the use of fresh lime, it
is desirable for utility applications to recycle collected solids back through the
absorber vessel. In addition, the nearer the dry scrubber can be operated to the
saturation temperature of the flue gas, the greater the improvement in S02 removal
efficiency and lime utilization. These requirements placed heavy demands on rotary
atomizer installations.
Some industry analysts(5) felt there were three fundamental conditions essential for
maintaining rotary atomizer longevity:
1. Minimization of vibration and load imbalance.
2. Atomizer wheel design to maximize life and reliability.
3. Hatching droplet size to the design of the spray dryer.
The following discussion centers on Item 1. In dry scrubbing applications, rotary
atomizers have been known to experience gearbox failure due to high temperature,
excessive starts/stops, inadequate lubrication and excessive vibration. While
coupling misalignment and shaft imbalance may cause vibration, more likely causes are
believed to be slurry buildup within the atomizer wheel, and the erosive nature of
recycled solids slurry causing uneven wear on exit orifice surfaces. While the wheel
is spinning at high speed, the inserts become worn preferentially on one side.
One solution offered by some manufacturers is the periodic 90° rotation of the
inserts, yet this may prove difficult in some designs due to the inherent immobility
of an insert once it is installed. Even if the rotary atomizer has design provisions
for minimizing wheel imbalance, the potential for drive-train damage can not be
completely alleviated, since slurry buildup and erosion are inevitable.
k.
4B-177
-------
In a dual-fluid atomizer, the rotating machinery responsible for providing the energy
necessary for atomization is an old power plant standby: centrifugal compressor.
Compressors have been around as long as power plants themselves, and maintenance
people are familiar with them, since compressed air is used for general services,
instrumentation, sootblowing and other duties. The dual-fluid atomizer itself has
no moving parts. Where the typical rotary atomizer can be composed of over one
hundred different parts (not including the drive motor), requiring many special tools
for assembly and maintenance, the DuraJet™ atomizer used in commercial dry scrubbers
is composed of a total of 24 pieces. See Table 2.
Considering that the end cap assembly is delivered as one separate piece, where the
inserts, wear pad and retainer rings are pre-assembled, the simplicity of design and
maintenance (a single 15 gpm atomizer weighs less than sixty pounds and can be handled
by one operator without hoist provisions) needs to be viewed favorably in contrast
to the maintenance intensive rotary atomizer.
In 1989, at the request of Basin Electric, a dual-fluid atomizer was developed as a
possible replacement for one rotary atomizer at the Antelope Valley Station AVS. As
stated previously, the AVS has two 440 MW units, each unit with five scrubbing
vessels, one of which is a spare. The driving force behind this request was the high
cost of maintenance at AVS, versus the much lower maintenance cost with the dry
scrubber at Basin's Laramie River Station Unit 3, a 570 MW system.
During March and April of 1989, operation of a large (100 MW) dual-fluid atomizer was
carried out at AVS.
Several separate tests to determine S02 removal were run under a variety of
conditions: low air-to-slurry ratio, low approach-to-saturation-temperature, low and
high Ioad.
While the results of this testing demonstrated the viability of dual-fluid atomizers
as a substitute for rotary discs in a large downflow dry scrubber module, they also
showed a need for more development work aimed at lowering the air consumption and
pressure, further reducing operating costs.
TECHNOLOGY APPLICATION
The DuraJet™ atomizer currently in commercial use at both Basin Electric's Laramie
River Station Unit 3, and Colorado-Ute's Craig Station Unit 3 have been shown
previously. At these stations, multiple atomizers are inserted into a spray chamber
where atomization is co-current to gas flow. These atomizers operate with slurry
pressures in the range of 40-60 psig, and air pressures in the range of 60-80 psig.
The air/slurry ratio is typically in the range of 0.1-0.14. Typical droplet size from
this commercial atomizer at these air/slurry ratios and pressures is 55 microns SMD.
Based on an EPRI report(6), at 100,000 pounds per hour slurry flow, the power
consumption for a Niro Atomizer is approximately 500 KW. Similarly, at 0.12 pounds
of air per pound of slurry, and 125 psig of air compressor discharge pressure, the
DuraJet™ atomizer would require an equivalent electrical power input for atomization.
The LIMB Humidifier uses a "mini" DuraJet™ atomizer, for extremely fine atomization,
which requires more power per mass of fluid atomized when compared to the commercial
DuraJet™ atomizer. The atomizer array is shown in Figure 9.
4B-178
-------
Since humidification requires rapid drying (1-2 sec.), significantly smaller droplets
are required than that for dry scrubbing. Pressures at the atomizer are typically
in the range of 100-120 psig for air, and 80-105 psig for water.
The "mini" DuraJet™ atomizer was also installed and operated at the 5 MW E-S0X pilot
constructed at Ohio Edison's R.E. Burger Station(7). Similar in many ways to the LIMB
humidifier with its multiple "mini" DuraJet™ atomizers housed within air foil lance
assemblies that incorporate the use of nacelles to direct shield air, it differs in
its orientation, and the fact that slurry is being atomized. This concept of a
DuraJet™ atomizer housed within a nacelle where vent air is carried, centered within
a low pressure drop air foil lance assembly, has been used again in the LIDS pilot.
This atomizer is fitted into a downflow dry scrubber followed by a fabric filter, and
has shown tremendous promise as a retrofit technology for high sulfur coal
applications.
Finally, it is interesting to note that the DuraJet™ atomizer, which evolved from
oil-firing y-jet atomizer technology, has been modified and installed in oil-firing
applications. Termed the LPA lighter, these modified slurry DuraJet™ atomizers are
operated at Kansas City Power & Lights LaCygne Station, and Otter Tail Power in
Minnesota. Lower opacity levels have resulted from use of the LPA oil lighter. The
air/oil ratio is roughly 0.25 to 0.30.
For main oil firing, installations at Sermide in Italy (18 burners) and Baltimore Gas
& Electric (8 burners) have proven the viability of the modified DuraJet™ atomizer
in main oil applications up to 180 million Btu/hr. The modified DuraJet™ atomizer
is also being used as a start-up oil gun at VEPCO's Bremo Bluff Station. Steam is
acceptable as the atomizing medium when fuel oil is the fluid atomized. For main oil
applications, standard operation maintains a 20 psi differential between steam and
oil over the load range, with full load oil pressure typically 120 psig, and steam
at 140 psig. Lower pressures are possible for special applications. A 0.1 ratio of
steam-to-oil is used.
FUTURE APPLICATIONS
Despite the broad range of applications and sizes of DuraJet™ atomizers, along with
the significantly lower power consumption in relation to other dual-fluid atomizers,
continual improvements in atomization quality are our goal.
Improved atomization for environmental processes represents increased surface area
for reaction and evaporation. In the combustion of fuel oil, it represents improved
combustion efficiency and lower emissions.
B&W has investigated new low pressure drop DuraJet™ atomizer designs. The objective
of this research is to alter the atomizer exit orifice and air orifice shapes to allow
more efficient use of the atomizing energy. More efficient use of atomizing energy
may be realized in terms of improved atomization quality without changing operating
pressures/atomizing fluid flow rates, or maintaining atomization quality with
reductions in operating pressure/atomizing fluid flow rates. Air pressure reductions
of up to 11% have been achieved while the air-to-water ratio and droplet size remained
constant.
The low pressure drop DuraJet™ atomizer will be incorporated into the humidifier
design as part of the Clean Coal III integrated Dry S02/N0x emissions control system
to be installed at the Public Service Company of Colorado's Arapaho Station Unit 4
located in Denver, Colorado.
4B-179
-------
SUMMARY
From oil-fired y-jet to front-mix DuraJet™ atomizers, the evolution of dual-fluid
atomization technology continues. As a member of the DOE Duct Injection Technology
Development Program, we remain actively involved in atomizer development, atomizer
array testing, along with further refinement to the second generation duct injection
computer model.
REFERENCES
1. W. Downs, W.J. Sanders, and C.E. Miller. "Control of S02 Emissions by Dry
Scrubbing". Proceedings of American Power Conference. April 1980.
2. T.B. Hurst. "Dry Scrubbing Eliminates Wet Sludge". Presented to Joint Power
Conference, 1979.
3. G.F. Burnett and B.E. Basel. "The Status of Dry Scrubbing in the United
States". Presented to Air Pollution Control Association, June 1985.
4. P.G. Maurin, H.J. Peters and V.J. Petti. "A Comparison of Two-Fluid Nozzle
Rotary Atomization for Industrial Dry Scrubbing Systems - Design, Operation &
Economics". Presented to American Institute of Chemical Engineers, 1982.
5. M.A. Remillard and C.J. Stempeck. "Evolution of Rotary Atomization Design for
Spray Dry Scrubbing". Proceedings of the American Power Conference. 1989.
6. EPRI Report CS-3954, May 1985.
7. K.E. Redinger, L.S. Hovis, F.C. Owens II, J.C.S. Chang and J.M. Wilkinson.
"Results from the 5 MWe E-S0X Pilot Demonstration". Proceedings of 1990 SO,
Control Symposium. May 1990.
4B-180
A
J
-------
Figure 1. Y-jet atomizer for dry scrubbing pilot.
' ; : ¦:;::TnT^]THTf "
'41 ri5l P{ ,6rJ'rl7l'' ' " 81 ' ' '9l'
rtt'STCOTT *. *Uit#
Figure 2. Two-tiered y-jet.
4B-181
-------
Black Ceramic
Mix Chamber Insert
(Optional)
Mix Chamber
Black Ceramic
Exit Onf/ce Insert
(Optional) Retaining Nut
Atomizing Medium
(Air or Steam/
Oil/Slurry
Black Ceramic
End Cap Insert
(Optional)
Atomizing Medium
(Air or Steam)
Large Diameter Ports
Multi-Spray
Nozzle End Cap
Figure 3. Commercial-scale DuraJet™ (10-17 gpm).
INCHES
Atomizing medium (air)
Mini I-Jet III™ atomizer
Naceile
Leading edge
Water (fluid)
Shield air
Figure 4. Mini-DuraJet™ for humidification.
4B-182
-------
18" Diameter
9' Length
Figure 5. 100 gpm DuraJet™ atomizer.
MK20 Y-Jet
DuraJet
Air-to-Water Flow Rate Ratio (lb/lb)
Figure 6. DuraJet™ drop size performance compared to
the y-jet atomizer it replaced.
4B-183
-------
MK20 Y-Jet at 120
psig Air Pressure
DuraJetat80
psig Air Pressure
0
10
Liquid Flow Rate (GPM)
15
20
Figure 7. DuraJet™ flow characteristics compared to
the y-jet atomizer it replaced.
0.4-1.0 GPM
at 100-120 psig
Air-to-Water Flow Rate Ratio (lb/lb)
Figure 8. Performance of two DuraJet™ atomizers of
different capacities.
4B-184
A
A
-------
Figure 9. Humidifier airfoil array.
4B-185
-------
Design Limits
Observed Performance 1
Max. Air Pressure
80 psig
80 psig
Max. Liquid Pressure
80 psig
70 psig
Max. Liquid Flow
16.8 gpm
16.8 gpm
Full Load Air-to-Slurry Ratio
0.125
0.107
Droplet Size (SMD)
60/ym
55/ym |
Table 1. Atomizer Design Criteria for the Craig Station Compared to DuraJet™
Atomizer Performance.
Atomizer Body
Packing Gland Nut
Following Ring
Inner Barrel
Outer Barrel
Adapter Coupling
Adapter Coupling Lock Nut
Mix Chamber Assembly (2 pieces)
Sprayer Header
End Cap Assembly (14 pieces)
Table 2. Commercial DuraJet™ Atomizer Part List
4B-186
A
-------
CHARACTERIZATION OF THE LINEAR VGA NOZZLE FOR FLUE GAS
HUMIDIFICATION
James R. Butz
James A. Armstrong
Timothy G. Ebner
ADA Technologies, Inc.
304 Inverness Way South, Suite 110
Englewood, CO 80112
William A. Walsh, Jr.
250 North Bay Street
Manchester, NH 03104
4B-187
-------
Intentionally Blank Page
4B-188
-------
ABSTRACT
A novel spray nozzle for the humidification of flue gas in ducts is under development by ADA
Technologies, Inc. for the US Department of Energy's Pittsburgh Energy Technology Center.
The Variable Gas Atomization (VGA) nozzle has a linear geometry where the water droplets
are discharged along a line, rather than from a point source as in current commercial
nozzles. In the patented design the thickness of the liquid sheet is controlled by the pressure
difference between the liquid and air supplied to the nozzle. This liquid sheet thickness is
key to the size of droplets generated in the nozzle. After the liquid sheet is formed, it is
immediately sheared on both sides by high velocity air streams to generate small droplets.
Subscale nozzles have produced sprays with a Sauter Mean Diameter (SMD) in the range of
30 to 40 /inn, at an air-to-liquid mass flow ratio (ALR) of 0.4 lb/lb. The advantage of the linear
VGA nozzle lies in typical operating air pressures that are less than 40 psig, compared to 70
to 150 psig for existing humidification nozzles. This reduced air pressure can result in
significant operational cost savings.
The development project has consisted of three major tasks: optimization of subscale
models of the nozzle; the design, fabrication and test of a prototype nozzle module; and the
preliminary design of a commercial scale humidification system that incorporates the
prototype linear VGA nozzle. The optimization addressed both the physical design of the
nozzle as well as the flowrates of air and water. The prototype nozzle has been designed
and fabricated and is now in test in a lab fixture as well as in an actual flue gas duct. The
commercial design has been started, and is awaiting performance data from the prototype
testing to move forward.
In the remaining six months of the development project, the testing of the prototype will be
completed, and the preliminary commercial installation will be designed. A final report will be
prepared to document the activities of the program. The linear VGA nozzle technology will
then be moved into the commercial marketplace.
Preceding page blank
4B-189
-------
INTRODUCTION
In a US Department of Energy (DOE) funded Phase I Small Business Innovative Research
grant, ADA Technologies, Inc. (ADA) demonstrated the feasibility of the linear Variable Gas
Atomization (VGA) nozzle technology for application to humidification of flue gas streams.
The linear VGA nozzle is a dual-fluid unit that operates at air pressures less than 40 psig and
air-to-liquid mass flow ratios (ALRs) below 0.5. The patented design uses the difference in
pressure between the water and air components to control the thickness of the linear slit
from which the liquid sheet is discharged. This sheet thickness is important because it
directly impacts the size of droplets produced in the nozzle.
There are three major objectives for the current development effort: to optimize the
performance of the linear VGA nozzle via small changes in the design; to design and test a
prototype linear VGA nozzle; and to develop a commercial design for an in-duct
humidification system that utilizes the performance features of the linear VGA nozzle. Some
of the activities to be carried out in fulfillment of each objective run concurrently, while
completion of others requires information and results from previously completed tasks. The
final result of this project will be a prototype system design that is ready to move into the
commercial arena for use in the humidification of flue gas.
Optimization of the current linear VGA nozzle in the first task assures that the design carried
forward to the prototype development is the most efficient and effective in the production of
droplets for humidification. The prototype design addresses the technical issues for a full
scale installation which were irrelevant in the earlier work. These include the optimum scale-
up size for a single nozzle section, installation structural support, flow controls for air and
water supplied to the nozzle, characterization of the prototype droplet size distribution, and
operational experience in an actual power plant duct. The commercial system design will
take into account the results of the prototype development, and will tackle additional aspects
of materials and cost reduction for mass production of a linear VGA nozzle design, flow
control for a duct humidification system, and plans for the identification of and agreement
with a commercial development partner.
Subscale nozzle configurations in both stainless steel and aluminum were evaluated by
measurement of the spray droplet size and velocity distribution. The Phase Doppler Particle
Analyzer (PDPA) employed a laser technique to determine the velocity and diameter of
individual spray droplets as they travelled through a sample volume. The instrument was
computer controlled and could handle up to 50,000 droplets per second. The sample
volume was quite small, so that in order to obtain data representative of the entire spray,
measurements were made over a grid of sample points on a cross section of the spray
pattern. Individual measurements were then weighted by the volumetric flux through the
sample volume in order to obtain characteristic values for a fixed set of operating conditions.
Two characteristics of the spray were used to evaluate the relative performance of candidate
nozzle configurations. These were the Sauter Mean Diameter of the droplets and the mass
fraction of droplets greater than 100 microns in diameter. The Sauter Mean Diameter is
defined as the diameter of a droplet with a volume to surface area ratio equal to that of the
entire sampled population.
4B-190
A
A
-------
LINEAR VGA NOZZLE CONCEPT
A schematic cross-section of the concept for the linear VGA nozzle is shown in Figure 1.
Water is pumped to the center of the nozzle, where it is contained by two cantilevered divider
walls, which touch at their tips. The walls separate the water flow passage from the air flow
passage. During operation of the nozzle, the air pressure is slightly less than the water
pressure; this difference causes the divider walls to deflect slightly at the tips. A very thin but
uniform water sheet is discharged from the tips of the divider walls, where it is immediately
sheared by the air flow. Because the air is much less dense than the water, its flow velocity
is much greater. The difference in velocity creates a shearing action that breaks the water
sheet into droplets, which are then discharged through the throat of the nozzle into the
surrounding gas stream. The droplets generated in this manner are on the order of the
dimension of the thickness of the water sheet; this is why the thickness of the liquid sheet is
critical to the performance of the nozzle.
The key difference between the linear VGA nozzle and current commercial devices is the
geometry of the discharge orifice. Conventional nozzles use an axisymmetric design for the
exit, where the diameter of the water injection orifice controls the size of the droplets
produced; when a larger flow is needed, the orifice diameter is increased. There is an
unwanted side effect to the increase, however: the larger diameter orifice produces larger
diameter droplets. In contrast, the linear VGA nozzle is scaled up by simply increasing the
distance along which the nozzle discharges. The thickness of the liquid sheet remains
constant, so that an increase in flow causes no increase in the size of the droplets produced
in the nozzle.
DESIGN OPTIMIZATION
There were four technical aspects of nozzle design addressed in the design optimization
task. These included the following:
• Thickness of the divider walls that separate the water and gas flow
passages in the nozzle;
• Cross section and length of the discharge throat;
• Flange seals; and
• Water flowrate, and the corresponding air-to-liquid mass flow ratio.
Two of these elements are associated in that the required discharge throat cross section is
dictated by the total flow (water plus air) that must be discharged from the nozzle. As
flowrates increase, either the air pressure must increase or the cross sectional area is
increased, since flow in the nozzle throat is sonic and therefore choked. There was great
interest in maximizing the water flowrate, since any increase above baseline levels would
result in a reduction in the total length of nozzle needed to meet a fixed water injection rate.
Divider walls of two different thicknesses were tested in both the aluminum and stainless
steel subscale nozzles. No effect on the droplet size distribution was detected for the
change in thickness. A divider wall thickness of 0.200 inch was selected for the prototype
nozzle design for the simple reason that it was easier and cheaper to fabricate than the
thinner, 0.120 inch size.
4B-191
-------
AIR
WATER
THROAT
LENGTH
AIR
r THROAT
Z WIDTH
Figure 1. Cross-Section of Linear VGA Nozzle.
Flange seals were a more difficult design issue; the original design specified a fine ground
finish on the end flanges as well as on the corresponding sealing surfaces of the assembled
nozzle body halves/divider walls. This appeared to be adequate only for the first few times
that the nozzle was assembled; in the course of testing, changes were made to the
configuration and the hardware was taken apart and reassembled numerous times. Normal
wear and scuffing of the sealing surfaces made leaks inevitable. Elastomeric gaskets and
cure-in-place sealing materials were tested. Teflon, polyethylene, viton, and silicone rubber
sheet gaskets were investigated, with viton sheet supplemented by spot applications of a
form-in-place liquid found to be the most effective. Concerns remain, however, over the
survivability of the viton material in the high temperature flue gas environment.
Several form-in-place gasket materials were also tested. The most effective sealant was
found to be Loctite Superflex, applied as a thin film and allowed to cure overnight after
assembly of the nozzle. This material has the added advantage of an operating temperature
to 600°F. Further tests will be run with this material, particularly when the prototype nozzle is
tested in an actual flue gas environment.
One of the first design changes to be investigated in the subscale nozzles was the length of
the throat in the discharge section, as sketched in the magnified view of Figure 1. The
original subscale aluminum nozzle had a throat length of about 0.4 inch, which was reduced
to about 0.1 inch to more closely match the stainless steel subscale nozzle design. An
immediate reduction in Sauter Mean Diameter of about 10 microns was realized upon testing
of the modified throat on the aluminum nozzle. Even better results were found with a
4B-192
-------
modification to the aluminum subscale nozzle, where the discharge throat length was
reduced to 0.040 in; a similar modification was then made to the stainless steel subscale
nozzle, resulting in a reduction in the SMD of the spray.
Further improvement was a little more elusive, however. In particular, close inspection
revealed that a thin water film was forming on the front face of the nozzle, where droplets
were periodically torn off by the exiting spray. Since these droplets were not formed from the
thin sheet exiting the divider walls in the nozzle body, they were believed to be much larger in
size and therefore causing an increase in the Sauter Mean Diameter of the overall nozzle
flow. Recall that a single 100 micron diameter droplet has the same volume as 1000 droplets
of 10 micron diameter. Thus a few droplets over 100 microns in size can dominate the
volume distribution of a spray that has thousands of very small droplets. The key is to realize
that elimination of these few large droplets or reduction of their size can dramatically affect
the Sauter Mean Diameter of the spray.
Additional measures to more tightly seal the flanges of the aluminum subscale nozzle
resulted in the elimination of the water film on the front face, and a further reduction of the
SMD of the spray.
As the testing moved from the ADA laboratory fixture into the High Flow Test Bed, it was
possible to quantify the impact of increases in water flow and changes in air-to-liquid mass
flow ratios on droplet size distributions from the subscale nozzles. This was a prime design
variable for the prototype nozzle, since the dimension of the discharge throat was directly
proportional to the mass flow of air through the nozzle.
Air is a compressible gas, and the maximum flow in the discharge throat is controlled
through the presence of a sonic flow condition. Once sonic velocity is reached, an increase
in mass flow can only be accomplished only by a proportional increase in the air density (as
indicated by the static pressure) in the nozzle body upstream of the throat. Since the design
goal is to operate the nozzle at a fixed air-to-liquid mass flow ratio, any increase in water
flowrate must be matched by a proportional increase in the air flowrate.
A second goal of the prototype nozzle design was to limit the air operating pressure to less
than 40 psig. If this condition was to be satisfied, the discharge cross section of the nozzle
must be dimensioned to move the required air mass flow at the specified static pressure at
sonic velocity. Thus the final width of the discharge throat in the prototype nozzle were not
established until a water flowrate was selected, so that the corresponding mass flow of air
could be specified. Figure 1 indicates the location of the width dimension with respect to the
overall nozzle configuration.
HIGH FLOW TEST BED
A High Flow Test Bed (HFTB) has been designed, outfitted and set up at the Arapahoe
Steam Electric Generating Station of Public Service of Colorado. The HFTB was planned as
a test fixture that would be capable of higher air and water flow rates than the fixture in the
ADA laboratories in which the initial Phase II tests for the aluminum and stainless steel
subscale nozzles were conducted. Because of air compressor limitations, the practical
4B-193
-------
upper limit in the laboratory was about 40 ACFM. The completed HFTB has an air flow
capacity of over 200 ACFM. The HFTB is shown schematically in Figure 2.
The HFTB is housed in a construction trailer that has been reworked to accommodate the
test fixture, instrumentation and flow control equipment for operation of two-fluid atomizer
nozzles. The test fixture is fabricated from Unistrut structural steel tubing, and features a
two-axis stepper motor drive system that translates the nozzle in the X and Y planes in order
to characterize the spray pattern across the face of the spray. Since the alignment of the
Phase Doppler Particle Analyzer (PDPA) is critical to its proper operation, it is simpler to
move the nozzle and to keep the PDPA stationary to maintain this alignment. The PDPA
mounting assembly is installed in a sliding track so that the droplet size and velocity profiles
can be measured over a range of distances from the nozzle discharge.
CDMPUTER/CDNTRDL
ROOM
—VGA NOZZLE
AND TRAVERSING
- PDPA
SYSTEM
OVERHEAD DOOR
PLANT AIR AND WATER
Figure 2. Sketch of High Flow Test Bed.
The HFTB is equipped with instrumentation to control the flow of water and air to the nozzle
under test, to measure the droplet size and velocity distribution in the spray, and to monitor
the flow conditions in the nozzle. Nozzle tests in the HFTB are controlled completely from
the forward room through two personal computers and several switches. One computer
controls the operation and data acquisition for the PDPA. The second monitors the
instrumentation installed in the nozzle, including water and air flow rates, pressures of both
fluids and the temperature of the air. The data acquisition system in the second PC also
controls the rate of air flow to the nozzle through a digital air regulator. Data from all
instruments is converted into engineering units, averaged and recorded every minute during
a test. A switch box in the forward room controls the stepper motors that position the nozzle
in the X and Y planes relative to the PDPA sample volume.
In the Phase I effort it was recognized that an automated control scheme for the air and
water supply would be critical to any commercial installation. To investigate the
requirements and operation of such a control circuit, plans were made to build and test a
4B-194
A
A
-------
first-generation flow control system for the HFTB test fixture. A more sophisticated system
would then be designed for use in the in-duct testing of the prototype nozzle, to be
conducted at the Arapahoe Station.
The first-generation system was designed to control the air flow to the VGA nozzle at a user-
defined ALR. A digitally-controlled air pressure regulator was integrated into the existing data
acquisition hardware and software. A control algorithm was programmed using a full
proportional/integral/derivative (PID) equation. The user-defined ALR set point and the
actual measured water and air flow rates are input to the algorithm which in turn calculates
an adjusted set point command for the air pressure regulator.
Piping and instrumentation for the air and water flow to the nozzle under test were mounted
on a movable Unistrut rack. This was done so that when the prototype nozzle is operated in-
situ in the duct at Arapahoe, the entire air and water plumbing tree could be moved and
installed easily with a minimum of rework. Air and water lines are equipped with filters and
isolation solenoid valves so that if problems are detected in the operation of the nozzle, the
system can be automatically and positively shut down. Inlets and outlets for both air and
water are equipped with quick-connect fittings so that the plumbing may be moved or
reconfigured easily. Also included on the unistrut rack are the power supplies for the
instrumentation transducers and a patch panel in an environmentally sealed electrical box.
Nozzle tests in the HFTB can be conducted by a single person because of the automation
incorporated in the new test fixture. In a typical test, the conditions of water flow rate and
ALR are held constant, and a series of PDPA runs are made on a matrix of locations across
the face of the nozzle spray pattern. The standard matrix includes five horizontal scans at
the centerline and at one and two inches above and below the centerline. PDPA samples are
taken in half-inch increments until more than 30 seconds are needed to acquire a 10,000
droplet sample; at the center of the spray this size sample is acquired in well under one
second. A typical matrix to characterize a single spray condition encompasses 75 to 90
measurements, and is completed in about 90 minutes with the automated setup of the HFTB.
ADA has also developed a computer program to access and reduce the information from the
individual PDPA files for each measurement location. This allows the large data files
produced by the PDPA to be analyzed and summarized on a single page table in a few
minutes.
SUBSCALE TEST RESULTS
The two subscale nozzles, one aluminum and one stainless steel, were tested in the ADA
laboratory test fixture and in the HFTB. The laboratory tests were run to investigate end
flange sealing problems, divider wall thickness, and dimensions of the discharge throat.
Tests in the HFTB were conducted to quantify nozzle performance at higher flowrates. As
noted earlier, the cross-section of the discharge throat is directly related to the design water
flowrate and associated ALR. Thus the final design dimensions of the prototype nozzle could
not be completed until a design flow condition was fixed.
A summary of spray droplet Sauter Mean Diameters for the aluminum subscale nozzle is
presented in Figure 3. The data include nozzle configurations with both the thin and thick
divider walls, and a range of nozzle throat widths, from 0.024 to 0.045 in. The graph shows
the overall weighted Sauter Mean Diameter as a function of the water flowrate; all tests were
L
4B-195
-------
run at an ALR of 0.4. There is a definite trend toward smaller SMDs as water flow increases,
with a minimum somewhere around 1.5 gallons per minute (gpm). The distribution of data
points shows the variance that is introduced by the difference in divider wall thickness and in
operating air pressures. The air pressure is directly affected by the discharge throat
dimension, such that higher air pressures are needed for smaller throat openings (for a fixed
water flowrate).
Divider Wall Throat Width
30
25-1
0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2
Water Flow (gpm)
Figure 3. Aluminum Subscale Nozzle Characterization Results.
Similar data for the stainless steel subscale nozzle is presented in Figure 4. Data from the
stainless steel nozzle indicate generally higher SMDs, except for the last few data points at
high flowrates (1.2 and 1.5 gpm). For these last tests, the nozzle was modified in that the
discharge throat was machined to shorten its length (in the direction of flow) by 0.060 in.
There was a dramatic change in the measured droplet SMD size, which dropped about 10
microns. Since the throat width was unchanged from earlier runs, the operating air pressure
for these cases was greater than the target value of 40 psig. When the throat of the nozzle
was machined to increase the width, another test at 1.5 gpm yielded a SMD of 35 microns for
the spray.
Also significant to the successful operation of a humidification nozzle is the mass fraction of
water droplets greater than 100 microns in diameter. Additional software was written to
extract this information from the PDPA measurements, and is presented in Figure 5 for both
the aluminum and stainless steel subscale nozzles. For most of the test cases included in
the figure, the total fraction over 100 microns is less than 6%. For the lowest SMD values
achieved with the stainless steel subscale nozzle, the fraction greater than 100 microns was
less than 1%, which is considered ideal performance for humidification atomizers (1). There
is also an easily identified trend toward a minimum in the large droplet fraction at a flowrate of
about 1.5 gpm in the subscale nozzles.
4B-196
A
A
-------
(/>
c
o
&
'E,
a)
"S
E
(0
o
c
(0
a)
aj
Z3
(0
CO
Divider Wall Throat Width Throat Depth
0.12" 0.025" 0.100"
0.20" 0.025" 0.100"
o 2Q. 0.025" 0.040"
0.20" 0.038" 0.040"
TIC
*
"H
1 1.2 1.4 1.6
Water Flow (gpm)
1.8
2.2
Figure 4. Stainless Steel Subscale Nozzle Characterization Results.
1 I
¦
> 100 microns
~
> 120 microns
—
1
I
1
i
¦¦
II
II
1
1
¦
¦
i
Ai
i
1
1
1
t
~
1
1
1
d
1
1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8
Liquid Flowrate (gpm)
1.9
Figure 5. Large Droplet Fraction for Subscale Nozzles.
4B-197
-------
PROTOTYPE DESIGN AND TESTING
A final prototype design of the linear VGA nozzle was prepared for fabrication. The design
was based on results from the subscale testing, and included changes in the throat
dimensions and flow paths for air and water supplies to the nozzle. The selected design
water flowrate was 5 gpm for the nine inch active discharge length of the prototype; this
corresponded to a subscale nozzle flow of 1.67 gpm. At an ALR of 0.4, the air flow rate was
220 standard cubic feet per minute. To keep the operating air pressure less than 40 psig,
the calculated discharge throat width was 0.038 in.
The prototype design was detailed on a series of drawings, and was presented to several
precision machine shops for bid. The selected fabricator used electric discharge machining
(EDM) to cut the tight tolerance discharge throat and contour for the mounting of the divider
walls that separate the water and air sections of the nozzle. The nine inch long module of the
prototype consisted of nozzle body halves, divider walls, and end flanges. A separate
fabrication package was prepared for the mounting plate/manifold that supplied air and
water to the module. The assembled prototype was designed to fit inside a three and one
half inch standard pipe that provided a separate low pressure air purge to keep the nozzle
face clean during operation in a flue gas duct. Figure 6 presents an exploded view of the
prototype nozzle and shroud assembly, illustrating the components named above.
The assembled prototype VGA nozzle was installed in the HFTB for initial performance tests.
The SMD of the spray was found to be in the range of 40-43 microns; this was somewhat
larger than the SMD of the subscale nozzles upon which the design of the prototype was
based. The flow along the discharge line of the nozzle was checked, and a considerable
variation was discovered. The nozzle was disassembled, and the divider walls examined.
Some deviations in the dimensions of the divider walls were measured, so the nozzle body
halves were returned to the fabricator and the divider walls were reground to within the
manufacturing tolerance. The reworked nozzle was returned to the Arapahoe Station,
reassembled, and prepared for in-situ testing in the Unit #4 duct.
The prototype linear VGA nozzle was then inserted in an actual flue gas duct to test in-situ.
Arrangements were made with Public Service of Colorado to install the prototype nozzle in a
sampling port on the ductwork of Unit #4 at the Arapahoe Steam Generating Station in
Denver. A scheduled outage of the #4 unit in October, 1991 presented an unique
opportunity to operate the nozzle in the duct and subsequently inspect the nozzle assembly
and duct interior before removal of the nozzle from the sampling port. Plans and
preparations for installation of the nozzle in the duct were accelerated, in order to permit
operation of the nozzle before shutdown of the unit.
The nozzle was installed at a sampling port near the wall of the Unit #4 duct such that the
distance to the side wall was 18 inches, and the distance from the top of the spray discharge
to the ceiling of the duct was 24 inches. The prototype nozzle was operated at a water flow
rate of 5 gpm, and an ALR of 0.4 lb/lb for a total of twelve hours. Approximately 3,600 gal of
water was injected into the flue gas stream. The flow in the nozzle was not such that any
measurable humidification took place, since the size of Unit #4 is 110 MW; however, the
nozzle was exposed to flue gas at a temperature of 285°F and a typical fly ash mass loading
of about 2.5 grains per dry standard cubic foot.
4B-198
A
A
-------
LENGTHS OF SHRDUD AND BACKPLANE/MANIFDLD
ARE SIZED FDR SPECIFIC SITE APPLICATION.
1-5 NOZZLE MODULES (5-25 GPM) PER LANCE.
SHROUD
NOZZLE BACKPLANE
WITH INTEGRAL AIR/H20
^1 MANIFDLD
(]P MMMIK)
air feed-^;;:
END CAP
A
WATER FEED
9 INCH
MODULE
END CAP
NOZZLE BODY HALVES SHDWN
WITH DIVIDER WALLS IN PLACE
Figure 6. Exploded View of Prototype Nozzle and Shroud Assembly.
-------
After shutdown of the unif, the duct was entered through an inspection port and the
condition of the nozzle and the adjacent ductwork was documented. There was no evidence
of any spray droplet impact on walls or ceiling of the ductwork. Some flyash buildup was
observed on the discharge face on the downstream side of the nozzle assembly. The purge
air from the nozzle shroud was not operated during the test, in order to determine the need
for this additional feature. As a result of the build-up found in the one day test, purge air will
be employed on future tests.
The test plan calls for a short duration (one week) test in the ductwork, after which the
prototype will be removed from the duct and returned to the test fixture in the High Flow Test
Bed to check the droplet size distribution. If there is any significant change in the nozzle
performance, the nozzle will be disassembled and examined in detail to identify the source of
the change. The second test period will be of two weeks duration, with a similar
characterization test upon completion. A third, month-long test will be run to complete the in-
situ testing of the prototype.
The prototype will remain instrumented during the in-situ tests. The sensors will be wired in
parallel to the microprocessor flow controller and an intelligent data logger. The flow
controller will provide real time control of both air and water flow, with provision for shut
down of the system if parameters are out of the operating range. The data logger will record
operating flows, temperatures and pressures at regular intervals, which can be used to
investigate any problems that may arise during the in-situ testing.
PRELIMINARY DESIGN FOR A BASELINE INSTALLATION
A survey performed by Burns and Roe for the US Department of Energy identified candidate
sites for the retrofit of duct injection technology to satisfy the potential Federal-mandated
requirements for S02 control technology (2). One output of the survey was a Lotus
spreadsheet database which included information on 340 plants. Items such as duct sizes,
run lengths, shapes, flow velocities, and temperatures were included in the database. The
information was reviewed and a baseline commercial configuration was selected to be used
in the design of a commercial package for duct humidification. Table 1 presents the
commercial baseline configuration dimensions. The information will be used to size a
commercial installation and to outline the elements of a control system to operate the
humidification system. Of particular interest is the ability of the system to follow load
changes and to interface effectively with a plant control system. To address control issues, a
list of control parameters has been developed.
For the above defined baseline installation, additional design work will be carried out.
Manifolds will be sized to supply air and water to an array of nozzle modules installed on a
grid in the baseline duct cross-section. A block diagram that describes the components of
the control system will be created, and a preliminary economic analysis of the capital and
operating costs of a humidification system will be conducted for the baseline installation. Air
compressors and water pumps for the installation will also be sized. Special requirements
for materials, installation and interlocks with the plant operating system will be identified.
4B-200
A
A
-------
TABLE 1. VGA HUMIDIFICATION NOZZLE PHASE II DEVELOPMENT
COMMERCIAL BASELINE CONFIGURATION
PARAMETER
SURVEY RESULTS
(AVERAGE)
BASELINE SPEC
Shape:
Cross Sectional Area:
Total Run Length:
Fluegas Temperature:
Saturation Temperature
Approach to Saturation
Superficial Gas Velocity:
Volumetric Flowrate:
Longest Straight Run:
Max Droplet Evap Time
47.2 ft/sec
618,080 acfm
53.7 ft (typ horizontal)
218 ft2
128.5 ft
316°F
Rectangular
Rectangular
220 ft2 (18X12.2)
120 ft
320°F
130°F (typical)
20°F
50 ft/sec
660,000 acfm
50 ft horizontal
1.0 sec
(Longest Run/Superf. Gas Vel)
Total heat removed from flue gas stream: 1,365,000 BTU/min
Required water flow from humidification system: 165 gpm
CONCLUSIONS AND RECOMMENDATIONS
A subscale linear VGA nozzle has demonstrated the ability to generate droplet size
distributions with a Sauter Mean Diameter in the range of 30 to 40 microns, with less than 2%
of the droplet volume greater than 100 microns diameter. Based on the subscale
performance, a prototype nozzle has been designed and is now being tested. This
prototype nozzle is nine inches in length, and has a design flowrate of five gpm. The design
ALR is 0.4, with a static pressure at the nozzle under 40 psig. A worst case air pressure
requirement of 40 psig represents a 44% reduction in air compressor power consumption
by the VGA nozzle as compared to a typical commercially available 150 psig nozzle
operating at the same ALR.
Assembly of the prototype nozzle was completed in September and the unit is undergoing
initial characterization of its spray. A 12 hour test in a power plant flue gas duct was
performed at the end of September, where the nozzle was installed and operated at design
flows of air and water. The nozzle was positioned approximately 18 inches from the side wall
of the duct and 24 inches below the ceiling. The generating unit in which the nozzle was
installed was shut down for a scheduled outage at the end of the twelve hour test, and the
duct was entered through an inspection port. The condition of the nozzle and the adjacent
duct walls was thoroughly documented in photos and a narrative. There was no evidence of
wall wetting on the side wall and ceiling nearest the nozzle.
4B-201
-------
There was, however, a deposit of fly ash on the discharge face of the nozzle, downstream of
the flow in the duct. This buildup was about one-half inch thick, and was easily broken off
the nozzle by hand. While there is a provision in the design of the prototype nozzle to supply
a stream of purge air to the discharge face of the nozzle, the purge air system was not in
operation for the twelve hour test.
A microcontroller-operated control system for both the air and water supply to the prototype
nozzle has also been demonstrated. The system measures air and water flows and
maintains both via control valves that are operated from the microcontroller. The
programming features an error trapping routine that shuts down the humidification system if
problems are detected in the sensors or supplies of air and water to the nozzle.
In the remaining six months of the development project, testing of the prototype will be
completed, and a task will be undertaken to transfer the linear nozzle technology to a
commercial partner, in order to bring the linear VGA nozzle to market.
REFERENCES
1. Ilan, Ruth. "Nozzle Development", Presentation to the DOE Duct Injection Technology
Working Group. Denver, CO. Oct 1,1991.
2. Sarkus, T.A. and D.S. Henzel, (1988). "Duct Injection Survey," Fourth Annual Coal
Preparation Utilization, and Environmental Control Contractors' Conference,
Pittsburgh, PA, August 8-11.
4B-202
-------
HIGH S02 REMOVAL DRY FGD SYSTEMS
Bert Brown
Joy Environmental Technologies Inc.
404 East Huntington Drive
Monrovia, California 91016-3633
Karsten Felsvang
Niro Inc.
9165 Rumsey Road
Columbia, Maryland 21045
4B-203
-------
Intentionally Blank Page
4B-204
-------
ABSTRACT
Spray dryer absorption is an accepted technology for flue gas desulfurization from
coal-fired boilers. S02 removal levels over 90% can be achieved reliably on low
and high sulfur coals. The S02 absorption process is explained with emphasis given
to alkalinity and reagent enhancers. Specific plants in the United States and
Europe that are operating at over 90% S02 removal are discussed. Availability data
from existing dry scrubber installations is presented. Additional data describing
the effectiveness of spray dryer absorbers on mercury removal is presented.
Preceding page blank
4B-205
-------
INTRODUCTION
Spray dryer absorption is an accepted technology for flue gas desulfurization from
coal-fired boilers. This process was originally developed in the late 1970*s and
has now progressed to the point where there is greater than 16,000 MW of planned
and installed capacity in the United States and Europe.
The process gained acceptance after pilot and full-size test work in conjunction
with Basin Electric Power Cooperative and Northern States Power Company. Much of
this effort concentrated on low sulfur coal applications involving western United
States fuels and moderate S02 removal levels. However, significant test work was
done in order to quantify performance levels with higher sulfur fuels such as
those coals found in the eastern United States and particularly in Illinois. This
work included artificially spiking the inlet gases with S02 at the Joy/Niro Hoot
Lake pilot plant in 1978, a trial burn of Illinois coal at the Northern States
Power, Riverside demonstration plant in Minneapolis in 1981 and a similar EPA-
sponsored program at Riverside in 1983. Commercial installations burning high
sulfur coal included Argonne National Laboratory which started in 1981.
This work showed that dry FGD utilizing a SDA and baghouse as a particulate
collector was capable of achieving S02 removal levels in excess of 90%, even with
high sulfur fuels.
Much of the information presented to date on the performance of spray dryer
absorption has concentrated on lower sulfur coals. Although much work on high
sulfur coals has been performed by Joy/Niro, the public perception is that the SDA
process is appropriate only for low sulfur coals at moderate efficiency levels.
This paper will address the mechanisms behind the high removal levels, discuss
case histories of high S02 removal efficiency installations, present availability
data and conclude with some new data on mercury control by the SDA process.
SYSTEM DESCRIPTION
The spray drying absorption process is shown in a simplified flow diagram in
Figure 1. The reagent for S02 removal, typically lime, is introduced into the
spray dryer absorber by a single rotary atomizer. The fine droplets produced by
the atomizer are evaporated in an absorption chamber utilizing either compound or
single roof gas dispersers. Residence time is typically 10 seconds. In order to
ensure adequate control over the drying process, a single atomizer per chamber is
used, instead of a more cumbersome multiple atomizer process that requires the
liquid feed to the atomizers be balanced to the individual gas flows. As the
evaporation process takes place, acid gases react with the calcium in the feed to
form salts such as calcium sulfite or calcium chloride. The dry powder is
transported to the particulate collector, which can be either a precipitator or a
fabric filter. Further S02 removal takes place in the collector due to the
unreacted calcium present from the spray dryer. The fly ash and reaction products
4B-206
-------
captured in the particulate collector will be recirculated back to the reagent
feed system to reduce lime consumption and improve drying characteristics.
Special consideration must be given to atomizer design to handle the abrasive fly
ash. This is handled in the Joy/Niro process with a Niro patented abrasion-
resistant type KN wheel outfitted with silicon carbide wear plates and nozzles.
High sulfur performance was first contemplated in 1978, when the performance
enhancements of the recirculation process were discovered. This work was
translated into full-scale 100 MW performance at the Northern States Power
Riverside Station when 3.5% sulfur Illinois fuel was trial burned in 1981.
Another installation at the Argonne National Laboratory was started up in December
of 1981. This unit also burned 3.5% sulfur Illinois coal. In April of 1983,
Argonne, working in conjunction with Consolidated Coal Company, successfully
demonstrated 95% S02 removal on both 4.2% and 3.5% sulfur coal.
In Europe, experience has been gained with medium sulfur coal at the 410 MWe power
plant at Ensdorf in Germany. This plant utilizes an SDA with a precipitator and
operates at 95% S02 removal with 1200 ppm inlet.
These plants operate with S02 removal levels typically specified for today's wet
FGD systems under the new Clean Air Act Amendments. In order to understand why,
we need to examine the chemical reactions and the drying process in the spray
dryer absorber.
HIGH S02 REMOVAL CAPABILITIES OF DRY SCRUBBING
The dry scrubbing system consists of two reactors in series; the spray dryer
absorber and the downstream filter. The removal efficiency achievable in each of
these two reactors is a direct function of how much alkalinity is fed to them. As
can be seen from Figure 2, the alkalinity fed to the spray dryer is the fresh lime
alkalinity combined with the alkalinity from recycle solids. At the inlet to the
filter, a new alkalinity ratio can be defined based on the absorption in the spray
dryer absorber. Figures 3 and 4 show the individual S02 removal efficiencies
achievable in the spray dryer absorber and fabric filter as a function of the
alkalinity ratio. This data was generated at the 100 MWe Riverside demonstration
plant in the early 1980's. With sufficient reagent fed to the system, more than
90% S02 removal can be maintained in the spray dryer absorber and more than 90% of
the remaining S02 can be removed in the downstream filter, thereby achieving a 99%
overall removal efficiency. Thus the spray dryer absorber system can, in
principle, achieve extremely high S02 removal levels.
In reality the flue gas that can be scrubbed and the S02 removal that can be
achieved is a function of how much lime can be fed into the system. Therefore,
the water evaporation capacity of the spray dryer, defined by the spray dryer
inlet temperature, and further utilization of the lime fed to the spray dryer is
of utmost importance.
4B-207
-------
Figure 5 shows the basic drying mechanisms in the spray drying absorber process.
The major portion of the absorption is accomplished in the spray dryer absorber.
The figures show the temperature profile of the flue gas and the atomized droplets
during the nominal 10 seconds residence time in the spray dryer absorber. The
temperature of the flue gas drops within a fraction of a second to the value of
the outlet temperature. In the same time period, the droplet temperature is
increased to the adiabatic saturation temperature of the flue gas. The droplet
temperature stays at that level until diffusion of water within the droplet
becomes the limiting factor in the evaporation. In this important first phase of
the drying process, the drying rate is proportional to the aT adsat (outlet
temperature minus adiabatic saturation temperature). The greater the aT adsat,
the greater is the driving force for the drying, or in other words, the rate of
drying.
Figure 5 shows how a high aT reduces the time that the droplet remains wet and
thereby the time available for absorption. In the example, it is shown that going
from a 20° approach temperature to a 50° approach temperature will decrease the
time available for absorption from 5 seconds to 2 seconds.
Figure 6 shows that regardless of the aT, a high S02 removal efficiency can be
achieved; however, lime consumption has to be sacrificed. When a limited amount
of lime can be fed to the system it is therefore important to operate at the
lowest possible aT to reduce the consumption as much as possible.
Lime consumption can be further reduced by the use of enhancers. Lime can be
enhanced by the presence of chloride and other deliquescent materials. The
improvements in dry scrubbing by using chloride enhancement were first recognized
by Niro (1). Some of the basic research work in characterizing the influence of
deliquescent materials on the performance of dry scrubbing was done by Klingspor
(2). The first full-scale demonstration of the chloride effect was done by
EPA/EPRI at the Riverside Demonstration Plant (3). Throughout the 1980's, full-
scale dry scrubbing systems came on line in Europe utilizing the effect of
chloride to enhance the performance (4). In the U.S. much research has been done
to characterize the influence of chloride on high sulfur coal. A recent study by
Barton (5) has documented the influence of various operating parameters when using
chloride rich coal on the performance of the spray dryer absorber and downstream
electrostatic precipitator. Results of these investigations show the clear
correlation of parameters such as S02 concentration, product solids chloride
content, inlet temperature, lime feed rate and approach to saturation temperature.
Figure 6 also shows how chloride enhancement improves the performance of the dry
scrubbing system. Chloride addition has the same effect as lowering the aT adsat;
i.e., the same S02 removal can be achieved at a lower lime consumption.
It is interesting to evaluate performance on some specific coals, such as 3%
sulfur and 4% sulfur utilizing the chloride effect. The lime stoichiometries
shown in Figure 7 will result when using chloride additive such that the dry
product contains between 4 to 5 weight percent chloride and the spray dryer outlet
temperature is controlled to an approach temperature of 25° F.
When using chloride in a spray dryer absorption system, special consideration must
be given to the outlet temperature. High chloride product as normally seen in SDA
systems on municipal waste incinerators rely on very high approach temperatures in
order to ensure a dry free-flowing product. The reason for this is, of course,
the lengthening of the drying time caused by the presence of the deliquescent
chloride. With typical coal-fired boiler SDA systems, the outlet temperature can
be typically increased from 25 to 45° F above the adiabatic saturation temperature
with little or no sacrifice in lime consumption.
4B-208
A
A
-------
CASE HISTORIES
Studstrup
The Studstrup FGD system in Jutland, Denmark has been described in previous papers
(6). The FGD system treats flue gas originating from burning a mixture of West
Virginia high sulfur coal and Colombian low sulfur coal. The design sulfur
content of the coal blend is 1.6%. The FGD system consists of a spray dryer
absorber followed by a pulse-jet fabric filter. The operation of the system is
unique by the fact that sea water is used for recycled slurry preparation. Total
chloride content in the end product is controlled in the FGD system to obtain
maximum benefit of the chloride effect described earlier in this paper. The
Studstrup plant is a peaking plant and operates between 20 and 100% load on a
daily basis. The FGD system performs at 92% S02 removal efficiency and is forced to
follow the boiler load swings.
Unit 3 and Unit 4 initially went through performance tests in mid-1989 and mid-
1990. The performance tests were successfully passed. After the initial
performance testing, the plant was further optimized. During this optimization it
was discovered that certain operating conditions caused corrosion problems due to
the high content of chloride. A major program was undertaken with involvement of
the Danish Corrosion Institute. Results of the study showed that the dry FGD
system can be operated under conditions where no corrosion will occur. The remedy
is to operate the spray dryer absorber system at a certain level above the
adiabatic saturation point.
In May 1991 a new set of performance tests were performed on Unit 3. The results
of the previous performance tests and the performance of May 1990/1991 are shown
in Figure 8. As can be seen, the FGD system was operated with and without the use
of sea water as make-up water for recycle slurry preparation. Comparing the June
1989 tests and the May 1991 tests with sea water, a 10% increase in lime resulted
when the approach temperature was increased from 36° F to 43° F. However, the S02
removal efficiency increased from 92 to 93.5%. Without the use of sea water,
approximately 90% removal efficiency would have been achieved. From the table it
can be seen that when the chloride concentration of the product is decreased, it
is possible to operate at a somewhat lower aT under corrosion-free conditions.
However, the lower chloride concentration leads to a higher lime consumption.
Figure 9 shows the waste product composition from the performance tests, running
with and without sea water as make-up. Sea water contains primarily sodium
chloride as the chloride source and from the composition it appears that the
performance run without sea water has only a negligible sodium chloride content.
The calcium chloride content in both products is similar as the calcium chloride
arises from absorption of HC1 from the inlet flue gas. It is interesting to note
the lower concentration of hydrated lime in the ash when sea water was used. This
is indicative of the high lime utilization with chlorine enhancement. Notice the
low ash content is caused by precollection of fly ash in an electrostatic
precipitator.
Walheim
The Utility of Neckarwerke in Germany is operating a dry FGD system at its Walheim
Station, Units 1 and 2. The two boilers have a rating of 110 mw and 150 mw
respectively. Two spray dryer absorbers are treating the flue gas from the two
boilers followed by a pulse-jet fabric filter. The dry FGD system was designed
for the German regulations which require half-hour emission averages of maximum 70
ppm. Thus the FGD system was designed for 95% removal efficiency on a half-hour
average with a maximum outlet emission of 70 ppm.
L
k.
4B-209
-------
The dry FGD system has been successfully in operation for three years. It has met
the emission requirements as a half-hour average during these years and has been
operating with a very high availability as presented elsewhere in this paper.
Figure 10 shows results from the performance test in 1989. During the performance
tests an average S02 removal in excess of 97% was reliably achieved at a low inlet
S02 concentration of only 456 ppm. This FGD system has successfully proven the
capability of a dry scrubber to achieve very high S02 removal at low inlet S02
concentrations.
Salzburg
The Salzburg, Austria dry FGD system removes S02 from the burning of high sulfur
coal. In a previous paper (6) it has been reported that this system successfully
achieves up to 97% removal with an inlet concentration of 2,500 ppm. In recent
years, the use of a lower sulfur content coal has led to a drop in the inlet S02
concentration to around 1,600 ppm. This requires 95% removal efficiency to
achieve the allowable outlet S02 emission of 70 ppm. The system has successfully
achieved this removal efficiency on a half-hour average since 1987. Figure 11
shows a computer print from a typical operation during a day. The half-hour inlet
and outlet S02 concentrations, as well as the half-hour S02 removal efficiency, can
be seen. Again the successful operation of this plant shows the capability of the
dry scrubbing process to achieve the high S02 removal efficiency on a steady and
reliable basis for a high sulfur coal application.
Coaentrix
Cogentrix is a 110 MWe generating facility, located in Virginia. It is laid out as
two boilers per 55 mw turbine with four boilers and two turbine generators total.
A single spray dryer absorber/baghouse train serves each boiler unit for a total
of four trains. The boilers are spreader stoker units supplied by Combustion
Engineering.
Each train is served by a single absorber 27.9' diameter, designed to operate
between 25-100% of design load. The particulate collector is a PULSEFLO® pulse-
jet baghouse consisting of six modules. Reagent utilization is enhanced by the
use of recycle and the chlorine found naturally in the coals. Design S02 removal
efficiency is 93%.
The boilers started up in August 1990 and have been running continuously since
then, with the exception of a scheduled outage in September of 1991. The unit
consistently runs at or above its design removal efficiency.
Chambers Works
This is a new generating facility that will be located next to the DuPont Chambers
Works facility in New Jersey; hence the name Chambers Works. Steam will be
supplied to DuPont and electricity supplied to the grid. The unit is nominally
rated at approximately 250 MWe.
The pulverized coal-fired boiler will be served by a high dust selective catalytic
reduction unit for N0X control and a Joy/Niro dry scrubbing system. The Joy/Niro
system will employ two spray dryer absorbers and a single reverse-air baghouse.
The facility has been designed to handle 2.1% sulfur coal with 93% S02 removal on a
one-hour rolling average. It is interesting to note that early dry FGD systems
were designed for 70% removal on a 30-day rolling average. The design and
4B-210
A
-------
reliability have progressed to the point where the spray dryer absorber system can
provide required high removal efficiency on a continuous one-hour basis.
Design data is presented in Figure 12.
AVAILABILITY OF JOY/NIRO DRY SCRUBBERS
The first generation of Joy/Niro dry scrubbers were brought on lie in the 1980's.
In spite of some initial operating problems, dry scrubbing technology proved to be
highly reliable and became the preferred gas cleaning equipment for western coal
applications.
The first generation dry scrubbers were designed with a spare absorber module and
a 100% redundant slaking and feed system including feed pumps. Now due to
increased rating, the Joy/Niro installations at Sunflower Electric, Tucson
Electric and Basin Electric are operating with all spray dryer modules on-line at
full load, which means that no spare module is available.
All the European dry scrubbing systems installed by Niro have been designed
without spare modules. Furthermore, only one spare feed pump is used to serve two
absorber modules. The slaking and feed systems, however, are still designed with
100% redundancy.
Figures 13a and 13b show the reported availabilities for the Joy/Niro systems in
the United States, as well as European installations. As can be seen, the
availabilities are all extremely high, ranging from approximately 97% to 100%.
The operators of the Joy/Niro systems both in the U.S. and in Europe have formed
users' groups where they have taken time to discuss common and unique problems and
learn from the successes and failures of each other. This has proven very
valuable in assisting the maintenance of high system availabilities.
MERCURY CONTROL
The spray drying process has been well documented for removing acids such as S02,
HC1 and HF from the combustion gases emanating from the burning of municipal solid
waste or hazardous waste (7). In this field, much effort has been devoted to
identifying the removal efficiencies of heavy metals and dioxins. One element of
particular interest has been the control of mercury. While it occurs in higher
concentrations in waste incinerator gas streams, the large volume of gases emitted
by coal burning utilities still result in appreciable emissions levels.
Significant test work on mercury removal was done on several European incinerators
because they were interested in meeting local mercury emissions standards with the
use of a spray dryer absorber. Initial results proved that some control could be
achieved with a lime-based reagent, but the desired emission limits could not be
met. During this time, Niro developed and patented an activated carbon injection
system to work in conjunction with the normal SDA system that resulted in very
high mercury removal levels. These results have been covered in a previous paper
(7). The success of these installations has raised questions on the effectiveness
of a SDA system on mercury removal from coal-fired boilers.
4B-211
-------
Tests have recently been completed on a number of spray dryer systems serving
boilers burning western and eastern U.S. coals, and on an additional installation
burning Polish coal. Unit D uses a spreader stoker boiler that has a moderately
high level of unburned carbon. Furthermore, activated carbon injection upstream
of the SDA was tested at a plant burning eastern U.S. coal to test the
effectiveness of the injection process. Results have shown some interesting and
unexpected trends. Figure 14 shows the average removal efficiencies and outlet
mercury concentrations obtained.
All these SDA systems use slaked lime as a reagent and use recycle. The levels of
mercury measured are approximately an order of magnitude lower than those
concentrations found in municipal solid waste incinerator gas streams. Of
particular note is the grouping of the outlet concentrations. The western fuels
have levels between 3 and 9 //g/Nm\ while the two units burning eastern coal have
approximate outlet concentrations between .2 and 1.3 //g/Nm3. Inlet mercury
concentrations varied between 4 to 11 //g/Nm3 corrected to 5% 02.
The use of activated carbon resulted in average outlet concentrations six times
lower than the cases with no injection. This data needs care in interpreting,
however, because two of the four stack tests were below the detection limit of the
test method. Further work is needed to evaluate the effectiveness of carbon
injection on various types of fuels; however, it is obvious that activated carbon
results in an improvement in the collection efficiency of mercury.
CONCLUSION
Despite the perception that spray dryer absorbers are effective only on low sulfur
coal installations with moderate removal requirements, data has been presented to
show that high S02 removal can be achieved with a very high degree of availability.
Work in the area of mercury control has shown that under certain circumstances
spray dryer absorbers can be extremely effective.
REFERENCES
1. S.K. Hansen, K.S. Felsvang, et.al. "Status of the Joy/Niro Dry FGD System and
Its Future Application for the Removal of High Sulfur, High Chloride and N0X
from Flue Gases." 1983 Joint Power Generation Conference, Indianapolis,
Indiana, September 27-28, 1983.
2. J. Klingspor. Kinetics and Engineering Aspects on the Wet-Dry FGD Process.
Sweden: Department of Chemical Engineering IL, Lund's Institute of
Technology, June 1983.
3. G.M. Blythe, et.al. "Field Evaluation of Utility Dry Scrubbing Systems."
EPA/EPRI FGD Symposium, New Orleans, Louisiana, November 1983.
4. Karsten Felsvang, et.al. "Update on Spray Dryer FGD Experience in Europe and
in the People's Republic of China." First Combined FGD and Dry S02 Control
Symposium, St. Louis, Missouri, October 25-28, 1988.
5. Robert A. Barton, et.al. "S02 Removal Performance Improvements by Chloride
Addition at the TVA 10-MW Spray Dryer/ESP Pilot Plant.", EPA/EPRI 1990 S02
Control Symposium, New Orleans, Louisiana, May 8-11, 1990.
6. Karsten Felsvang, et.al. "Dry Scrubbing Experience with Spray Dryer Absorbers
in Medium to High Sulfur Service." National Lime Conference, Philadelphia,
Pennsylvania, January 9-10, 1991.
4B-212
A
A
-------
7. Bert Brown, et.al. "Control of Mercury and Dioxin Emissions from United
States and European Municipal Solid Waste Incinerators by Spray Dryer
Absorption Systems." Second International Conference on Municipal Waste
Combustion, Tampa, Florida, April 16-19, 1991.
4B-213
-------
FIGURE 1. SPRAY DRYER ABSORBER PROCESS
Clean flue gas
Spray Dryer Absorber
Reagent
Water
Slake
Flue gas
Dust Collector
Recircu-
,atl0n ^ End-product
FIGURE 2. DRY SCRUBBER STOICHIOMETRIC RATIO AND ALKALINITY RATIO
FRESH
MAKE-UP
LIME
(Ca(OH),)
SR
AR1
AR2
FLUE GAS
TO DISPOSAL
moles Ca, make-up
moles SO,, Inlet
moles alkalinity
moles SO,, Inlet
FEED TANK
4B-214
-------
FIGURE 3. RIVERSIDE DRY SCRUBBER RESULTS
INFLUENCE OF INLET PPM S02
(SPRAY DRYER ABSORBER)
100
00
80
70
60
60
40
30
20
10
0
% SO, Absorption SDA
DELTA T
INLET SO ,
•d. tat.
ppm
d»g. F
1:
800
16
II;
2000
16
v
0.0 0.2 0.4 0.6 0.6 1.0 1.2 1.4 1.6 1.6 2.0 2.2 2.4
Alkalinity Ratio AR1
FIGURE 4. RIVERSIDE DRY SCRUBBER RESULTS
INFLUENCE OF INLET PPM SO2
(BAGHOUSE)
% SO, Absorption Baghouae
100 r
90 -
80 -
70 "
60 "
60 -
40 "
30 -
20 -
10 -
0
0 1 2 3 4 6 6
Alkalinity Ratio AR2
DELTA T
INLET 80 ,
•d. ««t.
ppm
d*e. F
I:
200
18
II:
600
18
4B-215
-------
FIGURE 5. DRYING MECHANISMS SPRAY DRYER ABSORBER
Temp. "F
910
Flue Gas
260
190 _
145'F
20'F
T out
125
Droplet
62
5 sec.
o
o.i
1.0
10
0.01
Time (sec.)
Temp. • F
310
Flue Gas
250
Tout — 175*F
190
Tad»«t • 50' F
125 _
Droplet
62
2 sec.
o
0.1
0.01
1.0
10
Time (sec.)
4B-216
i
A
-------
FIGURE 6. SDA/BAGHOUSE PERFORMANCE
% 80, ABSORPTION
100
do
LIME CONSUMPTION
FIGURE 7. SDA/BAGHOUSE PERFORMANCE
% SO. ABSORPTION
3* S 4% 8
1.00 1.26 1.60
STOICHIOMETRIC RATIO
4B-217
-------
Figure 8. Performance Test
Unit 3
Time 6/89
Sulfur Content % 1.6
HC1 in Flue Gas ppm
S02 Removal % 92.0
Stoichiometry 1.12
Delta T Adsat °F 36
Sea Water Used +
Chloride in Product % 4.5
Results, Studstrup FGD System
4 3 3
5/90 5/91 5/91
1.6 1.6 1.6
38 37
97.1 93.5 87.7
1.25 1.23 1.35
27 43 39
+ + -
4.0 2.5 0.8
Figure 9. Waste Product Composition, Studstrup FGD System
Sea Water Used +
CaS03 0 1/2 H20 55.1 56.3
CaS04 0 2 H20 9.4 2.9
CaC03 9.5 9.2
NaCl 2.7 0.2
CaCl2 1.3 1.0
Ca (0H)2 15.6 22.4
Moisture 1.0 1.3
Ash, etc. 5.4 6.7
Figure 10. Performance Test Results, Walheim Units 1 & 2
Test Number
1
2
5
6
7
8
9
12
13
AVG
S02 in (ppm)
506
459
507
546
506
445
446
384
305
456
HC1 in (ppm)
29
37
37
16
34
27
28
18
15
26
S02 Removal % 97.4 96.9 97.2 96.8 95.6 96.7 96.2 98.5 98.6 97.1
4B-218
A
A
-------
Figure 11. Salzburg FGD System, 1/2 Hour Emission Values on October 31, 1988.
S02
S02
o2
o2
Inlet
Outlet
% S02
Inlet
Outlet
Ad sat
No.
Time
ma/Nm3
ma/Nm3
Removal
ma/Nm3
ma/Nm3
TemD.
1
00:30
4327.5
186.6
95.7
8.4
8.3
55.9
2
01:00
4356.7
177.8
95.9
8.4
8.3
55.9
3
01:30
4337.6
184.3
95.8
8.7
8.6
55.7
4
02:00
4323.0
189.7
95.6
8.3
0.2
55.9
5
02:30
4174.5
187.0
95.5
8.1
9.2
55.9
6
03:00
4227.3
183.1
95.7
7.6
7.5
56.2
7
03:30
4248.9
181.1
95.7
7.7
7.9
56.3
8
04:00
4262.2
182.1
95.7
7.5
7.7
56.4
9
04:30
4579.3
191.7
95.8
7.4
7.3
56.7
10
05:00
4699.0
180.3
96.2
7.8
7.7
56.7
11
05:30
4685.0
214.1
95.4
7.3
7.2
57.2
12
06:00
4647.7
182.7
96.1
6.2
6.1
58.1
13
06:30
4428.2
197.5
95.5
5.6
5.7
58.4
14
07:00
4381.7
182.0
95.8
5.6
5.8
58.4
15
07:30
4395.7
203.1
95.4
5.8
6.9
58.3
16
08:00
4384.2
180.5
95.9
5.5
6.0
58.4
17
08:30
4568.7
198.1
93.7
6.1
6.0
58.3
18
09:00
4508.6
182.3
96.0
6.0
5.9
58.4
19
09:30
4515.9
212.5
95.3
6.0
5.9
55.4
20
10:00
4512.6
180.7
96.0
6.1
6.0
58.4
21
10:30
4348.8
202.2
95.4
5.9
6.0
58.5
22
11:00
4336.1
179.5
95.9
5.8
6.0
58.4
23
11:30
4333.5
197.0
95.5
6.0
6.2
58.4
24
12:00
4342.5
170.2
96.1
5.7
5.9
58.5
25
12:30
4630.7
220.3
95.3
6.5
6.4
58.2
26
13:00
4658.3
197.6
95.8
6.5
6.4
58.2
27
13:30
4629.9
205.6
95.6
6.4
6.3
58.2
28
14:00
4138.7
210.6
94.9
5.1
6.1
58.3
29
14:30
4419.5
191.5
95.7
6.1
6.2
58.2
30
15:00
4364.5
189.1
95.7
6.5
6.1
58.5
31
15:30
4343.2
184.0
95.8
6.0
6.2
58.2
32
16:00
4329.4
196.6
95.5
6.0
6.1
58.3
33
16:30
4471.5
185.6
95.8
6.3
6.2
58.3
34
17:00
4483.5
202.8
95.5
6.4
6.3
58.2
35
17:30
4469.7
184.1
95.9
6.3
6.2
58.2
36
18:00
4459.5
188.9
95.8
6.4
6.3
58.2
37
18:30
4382.6
187.4
95.7
6.2
6.2
58.3
38
19:00
4354.7
189.9
95.6
6.1
6.3
58.2
39
19:30
4333.8
194.9
95.5
6.2
6.4
58.2
40
20:00
4347.7
178.5
95.9
6.1
6.3
58.3
41
20:30
4469.0
'204.8
95.4
6.4
6.3
58.3
42
21:00
4675.7
194.8
95.8
6.9
6.8
58.1
43
21:30
4730.0
216.0
95.5
7.2
7.1
57.9
44
22:00
4768.0
198.1
95.8
7.8
7.7
57.4
45
22:30
4582.9
207.7
95.5
8.4
8.4
56.5
46
23:00
4238.7
147.1
96.5
9.2
9.3
55.5
47
23:30
4206.0
248.0
94.1
8.1
8.2
56.3
48
24:00
4399.9
178.3
95.9
7.8
8.0
56.5
4B-219
-------
Plant
MWe Rating
Fuel
Sulfur %
Chlorine %
Gas Flow, acfm
Chloride in Product
S02 Inlet ppm
S02 % Removal
Measurements
% Sulfur in Coal
SOj Inlet ppm
SOj Removal
SR
CI in Product (%)
Figure 12. Design and Performance Data
Salzburg Studstrup Coaentrix
70
Coal/Oil
3.6% (equiv)
0
108,000
0
2,500
97.5
2,270
96.9
1.55
0
350
Coal/Oil
Up to 2.5
0.1
1,070,000
5
1,400
92
Max.
2.5
91.1
1.03
4.6
110
Coal
0.75
.01-0.3
431,200
2-3%
Up to 1,100
93
.8
90
1.
4-5
26
Chambers
250
Coal/Oil
2.1
0.1
1,086,000
1
1,400
93
Figure 13a. Availability Data for Joy/Niro Dry FGD Systems, European Plants
Plant Name/Year _87_ 88 89 90
Aroskraft (S) 98.2 98.1 98.7 99.4
Durnrohr, EVN (A) 100 100 100
Durnrohr, VKG (A) 100 100 100
Salzburg (A) 96.9 99.0 99.6
Walheim (D) 98.1 99.6 99.5
Main Kraftwerke Hoechst (D) 97 97
Studstrup #3 (DK) 99.2 99.3
Studstrup #4 (DK) 99.2 99.3
(A) Austria
(D) Germany
(DK) Denmark
(S) Sweden
Figure 13b. Availability Data for Joy/Niro Dry FGD Systems, United States Plants
Plant Name/
Year 83 84 85 86 87 88 89 90 91
Sunflower #1 99 99 -- -- 98.7 99.1 98.6 98.7 98.2
Tucson #1 -- -- 97.5 97.9 99.1 99.3 98.6
Sherco #3 100 100 100
Wyodak #1 96.7
Rawhide #1 100 100 100 100 100 100 100 100
AVS #1 95 95 -100 -100 -100
AVS #2 -100 -100 -100
Cogentrix 99+
4B-220
A
A
-------
Figure 14.
SDA/BH or SDA/EP Performance on Mercury
Coal
Plant Type
A Western U.S.
B Western U.S.
C Western U.S.
D Eastern U.S.
D Eastern U.S.
E Colombian and
Eastern U.S.
F Polish
Mercury
Particulate Emissions
Collector Type aa/Nm3 0 5% 0-
Baghouse
9.1
Precipitator
3.3
Baghouse
7.9
Baghouse
.18
Baghouse
.03
Baghouse
1.3
Precipitator
0.31
Mercury
Collection
Efficiency. %
15
24
7
96.5
99.4 w/Carbon
Injection
56
89
4B-221
-------
Intentionally Blank Page
4B-222
A
-------
Session 5A
WET FULL SCALE FGD OPERATIONS
FGD SYSTEM RETROFIT FOR DALHOUSIE STATION UNITS 1 AND 2
Fred W. Campbell,
John P. Phillips
Burns & McDonnell Engineering Company
4800 E. 63rd Street
Kansas City, Missouri 64130-4696
Frank Sainz,
New Brunswick Electric Power Commission
515 King Street
Fredericton, New Brunswick E3B 4x1
Canada
ABSTRACT
The Dalhousie Station is an electric generating station owned and operated by New
Brunswick Electric Power Commission. The station is located on the north shore of New
Brunswick, Canada, and is comprised of two units totaling 315 MW. In excess of one
million barrels of an emulsified bitumen (Orimulsion) from South America have been
test-fired in Unit 1. Based on this extensive experience, both units are being
modified to burn the Orimulsion. The station is being retrofitted with a single module
FGD system to serve both units. In addition, a marketable grade gypsum product is
desired.
This paper addresses the design considerations made for the FGD system. It discusses
some of the philosophy used to help ensure a high level of reliability and
availability. It also discusses some of the problems encountered due to the fuel.
5A-1
k.
-------
INTRODUCTION
The Dalhousie Station is an electric generating station owned and operated by the New
Brunswick Electric Power Association (NB Power). The station is located on the north
shore of New Brunswick, Canada and is comprised of two units totaling 315 MW. Unit 1
has a gross rating of 100 MW and was placed into operation in 1969; Unit 2 has a gross
rating of 215 MW and was placed into operation in 1979.
Currently Unit 1 is oil fired and Unit 2 can fire coal or oil. (Over 1 million barrels
of an emulsified bitumen (Orimulsion) from South America has been successfully test
fired in Unit 1. The Orimulsion is a heavy asphalt-like material which is emulsified
in water. The end product can be handled similar to No. 6 oil and contains about 30%
moisture. The sulfur content and higher heating value of the Orimulsion ar 2.9% and
12,700 Btu/lb, respectively. Due to an economic advantage NB Power plans to modify
both units for using Orimulsion.
The Canadian National Government through Environment Canada requires that NB Power
reduce their sulfur emission by 1994. Also, as a condition for permitting the unit
modifications, the Provincial government required a reduction in the sulfur dioxide
emissions. According to Environment Canada's National Guidelines for New Stationary
Sources, units emitting between 0.6 lb/MMBTU and 6.0 lb/MMBTU should be controlled so
that the final emission does not exceed .6 lb/MMBTU. Units emitting more that 6.0
lb/MMBTU should be controlled so that a minimum of 90% of the uncontrolled emission is
captured.
The uncontrolled sulfur dioxide emission rate from firing Orimulsion is about 4.6
lb/MMBTU, which is higher than from the oil currently being used. Therefore, NB Power
elected to investigate the addition of a FGD system.
A preliminary study was undertaken in the fall of 1990 to determine the feasibility of
adding the FGD system to the Dalhousie Station. In late 1990 NB Power decided to
proceed with the FGD system addition. A schedule was established which required that
5A-2
A
A
-------
the FGD system be placed in service by September 1994. The FGD system design was then
started in early 1991.
PROCESS SELECTION STUDY
A Process Selection Study had been previously completed for NB Power's Belledune
Station. The Belledune Station is a new 480 MW coal fired generating station currently
under construction. It is also on the north shore of New Brunswick, about 30 miles
east of Dalhousie, and has similar environmental requirements. It was decided that the
results of the process selection study which was completed in January of 1990 for
Belledune would be applied to the Dalhousie project.
A total of 17 FGD processes were investigated. The capabilities and experience of each
process were compared to the minimum requirements for performance, reliability,
flexibility, and environmental acceptability established for the project by NB Power.
These requirements include the ability to achieve 90% removal of sulfur dioxide from
the flue gas produced by a wide range of fuels, the ability to produce a stable waste
product and proven commercial operating experience at similar plants.
Three of the 17 processes were found to have capabilities and experience records which
met all of the minimum requirements. These three FGD processes were: conventional
limestone, limestone with forced oxidation, and conventional lime. For these
processes, a more detailed evaluation was performed which included estimated capital
and operating costs.
Of the three FGD processes which met all of the minimum requirements, the wet limestone
process with forced oxidation was the preferred alternative. The primary
considerations for this selection were the reagent availability, operating experience,
useable waste product and the evaluated life cycle cost.
FGD SYSTEM PRELIMINARY STUDY
The Preliminary Study was done in September and October 1990 to provide NB Power with
FGD system descriptions, layouts, cost estimates, and operating and maintenance
requirements. The study was based on a wet limestone FGD system with forced oxidation.
Because of the optional fuels under consideration for Dalhousie, the study was done for
two cases. In one case both Units 1 and 2 fired Orimulsion. In the other case, Unit
1 fired Orimulsion and Unit 2 fired a local coal. The design basis for these two cases
is shown in Tables 1 and 2.
L
L.
5A-3
-------
The coal is a New Brunswick coal with a sulfur content of about 6 to 9% and a heating
value of about 12,000 BTU/lb. The Orimulsion has a maximum sulfur content of 2.9% and
a heating value of about 12,700 BTU/lb.
For the study, NB Power directed that the FGD systems be designed for a maximum sulfur
dioxide emission rate of 0.6 lb/MMBTU, but not less than 90% removal. This resulted
in a removal efficiency of about 95.8% for Unit 2 when firing the New Brunswick coal.
There is no operating experience on removing 95.8% of the sulfur dioxide emissions from
a 9% sulfur fuel. To estimate the FGD design parameters equations were used which are
based on operating systems. These equations were extrapolated to the predicted
conditions. The maximum sulfur dioxide inlet concentration was predicted at about
7,000 ppm when only Unit 2 is in service. As a result of preliminary calculations, the
study assumed that the required performance could be achieved in a spray tower with a
liquid to gas ratio (L/G) of about 170 and organic acid promoting agents.
While firing Orimulsion, the FGD system must achieve 90% removal. Using the same
equations, the design for the study was based on a spray tower with a L/G ratio of
about 100 and no promoting agents.
FGD SYSTEM DESIGN
Design Basis
As noted above the initial intent was to design the FGD system on the basis that Unit
2 would maintain the capability to fire coal. However, after careful consideration NB
Power decided to eliminate the coal firing capability at Dalhousie. The FGD system
design was therefore based on firing Orimulsion in both units.
A wet limestone FGD system was the basis for the design. NB Power had investigated
packed towers vs spray towers for the Belledune project. Based on experience from
several operating plants they had determined that the spray towers would best meet
their needs at Belledune. The spray tower was also the basis for the Dalhousie FGD
system.
The Canadian government stresses the production of a useable by-product. In trying to
eliminate one problem, they do not want to create another. To achieve this goal, NB
Power elected to design the FGD system for internal forced oxidation to produce a
marketable grade gypsum product. However, provisions for landfill disposal are also
being designed for periods when a market is not available and/or periods when the
gypsum quality is inadequate for sale.
5A-4
-------
The design operating conditions for the FGD system are provided in Table 3, and the
characteristics of the Orimulsion are provided in Table 4. NB Power has not selected
the limestone supply, but preliminary investigations indicated that a supply could be
found with a minimum purity of 90% calcium carbonate.
The performance requirements of the FGD system included a minimum sulfur dioxide
removal efficiency of 90% while producing a marketable grade gypsum product. The
dewatered gypsum must contain no more that 10% water and the solids must have a minimum
gypsum content of 92%.
To achieve this performance minimum design requirements were established for the FGD
system. These requirements included the following:
Minimum L/G ratio - 105 gpm/ 1000 acfm
Minimum reaction tank hold time - 10 minutes
Minimum limestone utilization - 98%
Minimum slurry oxidation - 99%
Number of Modules
During the preliminary studies and the early stages of design, considerable attention
was given to the number of absorber modules to be installed. There are no regulations
that require the installation of spare modules. Due to the high cost of absorber
modules, it was decided that spare modules would not be provided.
Three options on the number of modules were initially considered:
Three Modules: As stated earlier, Unit 1 is 100 MW and Unit 2 is 215 MW. This
option provided three modules of nearly equal size with one module installed on
Unit 1 and two modules on Unit 2.
Two Modules: This option provided for one module to be installed on each unit
with the module on Unit 2 being about twice the size of the module on Unit 1.
One Module: This option provided for the flue gas from both units to be
combined into a common duct and treated in one absorber module.
The economics favored the single module. The potential capital cost savings was about
13% over the cost of three modules and about 8% over the cost of two modules.
5A-5
-------
A single module on two units brings up other considerations. The obvious concern is
that both units are affected if the module is forced out of service. The other concern
is that at least one unit is in operation at all times, which does not allow any time
for internal inspections and maintenance on the module.
NB Power elected to proceed with the project based on one module. It was decided that
both units could be taken out of service for about 10 days each year for internal
inspections and maintenance of the absorber module. The options for periods of forced
outages are still being investigated.
Sparing Philosophy
With the decision to go with only one absorber module, the sparing philosophy becomes
much more important. All of the auxiliary equipment includes installed spares to
increase the reliability and availability of the FGD system.
It was decided to include one spare recycle pump and spray header. The spare pump can
be maintained with the module in service. If a problem is detected with one of the
headers, the spare header will enable the module to remain in service until a
maintenance outage can be arranged.
The design required installed spare capacity for the auxiliary equipment generally in
the form of two 100% or three 50% capacity trains. Typical installed spare capacity
includes the following:
Two 100% ball mill systems for limestone grinding
Three 50% vacuum filters for gypsum dewatering
Three 50% oxidation air compressors
Two 100% bleed pumps for removing slurry from the reaction tank
Two 100% mist eliminator wash pumps
Materials of Construction
A combination of low pH and dissolved ionic species inside a wet limestone FGD system
makes the environment very corrosive. Rubber lining and coatings of various types have
been used to protect carbon steel in these applications. Stainless steels and nickel
based alloys have also been used and new materials and alloys are still being produced
and tested.
The corrosiveness of the FGD environment has been increased at Dalhousie because of the
closed loop operation and the production of marketable grade gypsum. These two items
5A-6
A
A
-------
act to increase the level of dissolved ionic species (including chlorides and
fluorides) inside the absorber tower.
At Dalhousie all of the water within the system will be reused and will stay within the
system to the extent possible. Most of the water in the gypsum slurry is reclaimed
during the dewatering process. The gypsum cake leaving the system is 10% water and
this water takes some chlorides with it. This discharge could limit the chloride
concentration to about 6,000 ppm in the FGD system. However, the gypsum is washed to
make it more suitable for commercial use. This washing returns most of the dissolved
ions, including chlorides, to the FGD process.
To limit the chloride concentration in the absorber tower, a small amount of water must
be blown down from the system. This blowdown stream must be kept to a minimum since
it must be treated before it can be discharged. The blowdown amount has been based on
limiting the chloride concentration to 20,000 ppm.
The selection of construction materials for the absorber tower was given considerable
attention. Stainless steel and alloy materials were selected over the lining and
coating options.
The rubber linings and coatings are entirely corrosion resistant in this environment.
However, the installation of these materials can be difficult and is critical for
success. The rubber linings are combustible and one accident can result in major
delays, not only during construction, but during any maintenance outage. Linings and
coating normally require frequent inspections and repairs to keep them in good
condition. That is not possible with a scheduled outage of only once per year. Also,
as linings and coatings fail (and eventually they will) pieces of the material fall
into the reaction tank. These pieces are pumped to the spray headers whfere they will
potentially plug the spray nozzles. If more than one spray header has pluggage
problems, the absorber tower is forced into an outage.
Properly selected stainless steel and alloy materials are virtually maintenance free.
The proper selection, however, must also consider cost. The higher grade alloys
exhibit greater corrosion resistance, but also demand a higher price. The following
materials of construction were selected for the Dalhousie project:
Absorber inlet duct from inlet damper to module - alloy C276
5A-7
-------
Absorber tower, reaction tank and outlet duct from module to outlet damper -
317L stainless steel (4.25% minimum molybdenum and 0.15% minimum nitrogen)
Outlet duct from absorber outlet damper and from bypass damper to chimney -
alloy C276
Marketable Grade Gypsum
Investigation of possible gypsum markets led to the gypsum requirements shown in
Table 5. To produce gypsum the FGD system was designed for forced oxidation in the
reaction tank.
Dewatering of the gypsum is a two stage process. The first stage dewatering is
accomplished by hydroclones. The hydroclones are expected to dewater the slurry
leaving the reaction tank from about 15% solids to about 50% solids. The second stage
dewatering system will use horizontal bed vacuum filters to produce a gypsum cake of
90% solids minimum. The gypsum is washed on the vacuum filter to reduce chlorides to
the specified level. An intermediate storage tank between stages acts as a surge tank
and allows the vacuum filters to be operated on a batch basis.
Limestone Handling
As noted earlier, it was decided to discontinue the firing of coal at the Dalhousie
Station. After careful investigation, NB Power decided to use the Unit 2 coal handling
facilities, including the coal bunkers, for limestone.
Modifications will be made to the bunkers, including adjustments to the level controls.
Because limestone is much heavier than coal, the bunkers cannot be filled to capacity.
The existing feeders will be rearranged to feed limestone to the new ball mills.
Chimney
A new chimney will be constructed to exhaust the wet flue gas from the FGD system.
This chimney will also be used for bypass conditions. The chimney design and selection
of materials of construction were not finalized at the time this paper was prepared.
The Unit 1 chimney is only about 150 ft tall and will be abandoned once the FGD system
is placed in service. The Unit 2 chimney is about 550 ft tall with a carbon steel
liner. This chimney will also be removed from service. The chimney diameter is not
sufficient to serve both units, and any modifications would require an unacceptable
outage period for Unit 2.
5A-8
-------
CONTRACT SCHEDULE
NB Power elected to use a turnkey contract to purchase the FGD system. A specification
was prepared to include the following:
Ball mill grinding circuit
Limestone slurry storage and feed to the absorbers
Absorber tower with wet loop, forced oxidation and mist eliminator wash system
All alloy ducts including dampers
First and second stages of gypsum dewatering
The schedule for the FGD contract is shown in Figure 1.
NB Power will furnish or contract separately for the following:
Foundations including site preparation work
FGD system enclosure including elevator
Chimney
All carbon steel ducts including supports
Booster fans including boiler investigations and modifications
Gypsum handling and disposal after it leaves the vacuum filters
Electrical
Controls (DCS system)
To accomplish a fall of 1994 operation date, the construction of the FGD system must
start in January of 1993. Winter conditions on the north shore of New Brunswick are
not favorable to construction. The high quality required for construction of alloy
materials only compounds the problem.
To achieve this construction schedule, the FGD supplier must have all foundation loads,
steel loads and access requirements no later than January 1992. NB Power then plans
to design and construct the foundations and FGD enclosure for the absorber tower by
January of 1993. This will allow the erection of the absorber tower to take place
inside.
PROJECT STATUS
Boiler modification work on Unit 1 is nearly complete. Additional test firing of
Orimulsion is planned for the last quarter of 1991. Boiler modifications for Unit 2 are
planned for the 1992 outage.
The bids for the FGD contract have been received and evaluated. A letter of intent was
issued October 1, 1991.
l.
5A-9
-------
1SS0 mi m2 1993 1SS4
Preliminary Design Studies M
Prepare Specifications
Bid Period
Evaluate Bids - Award Contract M
Vendor Engineering and Fabrication ¦¦¦¦¦¦¦¦
Construction >¦¦¦¦¦¦¦¦¦¦¦
Startup and Trial Operation
Completed
Figure 1. Contract Schedule - FGD System Addition, Dalhousie Station Units 1 and
5A-10
-------
Table 1
DESIGN BASIS
(Units 1 & 2 - Orimulsion)
Unit Conditions
Unit 1 Unit 2
Fuel rate (1000 #/hr) 79 160
Gas flow (1000 #/hr) 1,101 2,143
Gas temperature (deg. F) 419 419
FGD System
S02 Removal Efficiency 90% 90%
Inlet gas temperature (deg. F) 419 419
Inlet gas flow (acfm) 406,211 790,915
Outlet gas temperature (deg. F) 145 146
Outlet gas flow (acfm) 314,558 614,691
Table 2
DESIGN BASIS
(Unit 1 - Orimulsion; Unit 2 - Minto Coal)
Unit Conditions
Fuel rate (1000 #/hr)
Gas flow (1000 # hr)
Gas temperature (deg. F)
FGD System
S02 Removal Efficiency
Inlet gas temperature (deg. F)
Inlet gas flow (acfm)
Outlet gas temperature (deg. F)
Outlet gas flow (acfm)
Unit 1 Unit 2 (Minto)
79 174
1,101 2,362
419 357
90% 95.8%
419 357
406,211 785,489
145 132
314,558 639,012
5A-11
-------
Table 3
DESIGN OPERATING CONDITIONS
Item
Unit 1 Fuel
Unit 2 Fuel
Fuel flow rate
Dry gas
Water vapor
Total gas flow
Gas temperature
Gas density
S02 loading
S02 loading
Fly ash loading
Range:
Item
Total gas flow
Total gas flow
Gas temperature
S02 loading
S02 loading
Orimulsion
Orimulsion
239,000 lb/hr
2,880,500 lb/hr
292,900 lb/hr
1,078,400 acfm
350°F
0.0490 lb/ft3
13,862 lb/hr
4.57 lb/MMBtu
304 lb/hr
Maximum
1,078,300 acfm
3,173,400 lb/hr
425°F
13,862 lb/hr
4.57 lb/MMBtu
Minimum
102,700 acfm
300,200 lb/hr
300°F
5A-12
A
-------
Table 4
ORIMULSION CHARACTERISTICS
Gross heat of combustion
12,700 Btu/lb HHV
11,700 Btu/lb LHV
Metals (ppm by Wt.)
Vanadium
Nickel
Iron
Sodium
Magnesium
Asphaltenes (Wt%)
Ash (Wt%)
Carbon Conradson (Wt%)
Max.
Max.
Max.
Max.
Max.
Max.
Max.
Max.
360
86
18
80
500
9.50
0.25
11.16
Ultimate Analysis (Wt%)
Carbon
Hydrogen
Sulfur
Oxygen
Nitrogen
H20
Ash
Chlorine
Typical
58.50
7.50
2.90
0.40
0.45
30.00
0.25
0.001
Range
55.00-61.90
7.00- 7.80
Max. 2.90
0.18- 0.61
0.42- 0.50
28.00-32.00
Max. .25
Max. .01
Chemical analysis of Orimulsion fly ash from the test program (100 MW) at Dalhousie
G.S. :
% C
3.1
% s
13.8
% Fe -
2.5
% Mg -
12.5
% Na -
2.5
% Ni -
2.2
% V
12.8
DENSITY = 5 ¦
¦ 10 lb/ft:
NOTE these elements exist in a complex form such as oxides and salts.
Therefore, the total does not add to 100 percent.
5A-13
-------
Table 5
GYPSUM REQUIREMENTS
Component
Gypsum CaS04 *2 H20
Calcium Sulphite
CaS03 * 1/2 H20
Water Soluble Salts:
-Chloride Cl
-Magnesium Mg
-Potassium K
-Sodium Na
-Total Water Soluble Salts
PH
Whiteness
Odor
Particle Size (Microns)
Blaine Surface Areas
Free Water (% of Total Weight)
Composition Specification
(Based on dry weight)
Minimum
(wt/wt)
92%
6.0
80%
Neutral
16
Maximum
(wt/wt)
1.5%
100 ppm
75 ppm
75 ppm
75 ppm
600 ppm
8.0
63
(95% 16 to 63 micron)
35,000 in2/lbs 246,000 in2/lbs
10%
5A-14
-------
ZIMMER FGD SYSTEM:
DESIGN, CONSTRUCTION, START-UP AND OPERATION
WM. D. BROCKMAN,
THE CINCINNATI GAS & ELECTRIC COMPANY
WM. H. ZIMMER STATION
1781 U.S. ROUTE 52
MOSCOW, OHIO 45153
R. W. TELESZ
THE BABCOCK & WILCOX COMPANY
2 0 S. VAN BUREN AVENUE
BARBERTON, OHIO 442 03
DON STOWE
DRAVO LIME CORPORATION
3 600 ONE OLIVER PLAZA
PITTSBURGH, PENNSYLVANIA 15222-2682
k.
5A-15
-------
Intentionally Blank Page
5A-16
-------
BACKGROUND
The Wm. H. Zimmer Generating Station is jointly owned by The
Cincinnati Gas & Electric Company (CG&E), Columbus Southern Power
Company (CSP-a subsidiary of the American Electric Power Company,
Inc.)/ and Dayton Power & Light (DP&L). The 1400 MW gross unit
is located on the Ohio River near Moscow, Ohio, (Figure 1)
approximately 3 0 miles southeast of Cincinnati, Ohio. This unit
is unique since it is the world's first nuclear-to-coal
conversion.
Construction activities on the nuclear unit were halted in 1982
by the Nuclear Regulatory Commission because of safety related
concerns. A decision was made by the owners in January, 1984 to
use their best efforts to convert Zirnmer to a coal-fired station.
American Electric Power Service Corporation (AEPSC) was estab-
lished as the project manager, responsible for licensing,
engineering and design, procurement, and construction. CG&E was
responsible for start-up and operation. AEPSC engineering began
air modeling, engineering, and design in August, 1984 to deter-
mine specification requirements for the project. In February,
1986, the flue gas desulfurization system (FGD) contract was
awarded to Babcock & Wilcox (B&W). In March, 1987, FGD earthwork
construction began at the site. On December 31, 1990, the unit
generated its first megawatts and was declared commercial on
March 31, 1991.
The B&W FGD system at Zimmer, the largest single unit FGD system
in the world, has been designed based on the following
parameters:
The FGDS is required to remove a minimum of 91% of the sulfur
dioxide from the flue gas and not exceed an emission cap of 0.548
lbs. of S02 per million Btu's based on 30 day rolling averages.
DESIGN
Inlet Gas Temperature
Mass Inlet Flow Rate
Volumetric Inlet Flow Rate
Inlet S0„ Loading
Coal Sulfur Content
Lime Stoichiometry
Liquor-to-gas Ratio (L/G)
Absorber Gas Velocity
345 * F
14,800,000 lb/hr
4,900,000 ACFM
2.5 to 8.6 lb/10 Btu
1.5 to 4.5%
1. 03
21
10 fps
Preceding page blank
5A-17
-------
The process chemistry utilizes magnesium enhanced lime (MEL) as
the active scrubbing reagent. This process was selected on the
Owners' previous experience at CG&E's East Bend Station and CSP's
Conesville Units 5 and 6. Dravo Lime Corporation supplies the
MEL up the Ohio River via barge from either its Maysville or
Black River, Kentucky facilities, located approximately 40 miles
and 1 mile respectively from the plant. The greater solubility
of magnesium salts in this process allows the liquor to contain
10 to 15 times more dissolved alkalinity than calcium based
process liquors. Therefore, S02 removal in the MEL process is
only limited by the relatively fast diffusion of gaseous SO. into
the liquor as opposed to calcium based systems depending on the
much slower dissolution of calcium. The following is a
simplified summary of the dominant reactions of equilibrium
conditions:
CHEMISTRY
Neutralization Reaction
2H+(1) + S03"(1) + MgS03(l)
Dissociated Magnesium
Sulfurous Acid Sulfite
—>
Mg(HS03)2(l)
Magnesium
Bisulfite
Regeneration Reactions
Mj(HS03)2(l) +Ca(CH)2(l) —> HgSO^l)
Magnesium
Bisulfite
Calcium
Hydroxide
Magnesium
Sulfite
Mg(HS03)2(l) + Mg(GH)2(l)
Magnesium
Bisulfite
Magnesium
Hydroxide
CaS03 • ^0(5)
Calcium
Sulfite
2M3S03(1)
Magnesium
Sulfite
^0(1)
Water
2^0(1)
Water
Because magnesium hydroxide represents five or less percent
of the lime slurry, the last equation represents a
relatively minor reaction in the process.
Based on the high liquid phase alkalinity achieved in this
system, the equilibrium SO- concentration at the gas/liquid
interface is essentially zero. Therefore, this process achieves
high SO removal efficiencies (90 to 98%) on high sulfur coal
with low tray pressure drop, low L/G, and low nozzle pressure
compared to calcium based systems.
5A-18
A
-------
EQUIPMENT DESIGN
There are four r.ajor areas of equipment comprising the Zimmer FGD
system (Figure 2):
Lime Unloader and Preparation
Absorber System
Primary Dewatering
Secondary Dewatering and Waste Stabilization
Since previous papers have described the equipment in detail, we
will only point out the highlights of each area.
Limestone Preparation
Two, 100% wet ball mill slaking systems, each located beneath a
15,000 ton lime storage silo, are used to slake the lime. Ball
mills are preferred over slakers to ensure that the grits
(insoiubles) are ground ana do not accumulate in the system.
Absorber System
The six stainless steel absorbers (five operating, one spare) are
of the tray tower design (Figure 3). A single tray provides even
gas distribution across the tower, ensuring complete gas/slurry
contact, plus providing a surface of intimate contact between the
gas and slurry. The even gas distribution provided by the tray
is also intended to relieve the mist eliminators of localized
high gas velocity upsets. The reactivity of the MEL system, and
resultant lower L/G, allows the absorbers to be 4 5% smaller than
a comparable-performing limestone based absorber (Figure 4).
One of the unique features of this plant is the arrangement of
the six absorbers in an ISO degree arc around the stack (Figure
5). This arrangement minimizes the initial cost, and future
maintenance cost, of outlet duct work. The close coupled
enclosure between the absorbers and the stack serves as a pump-
house, offers a maintenance aisle, and supports the outlet
fluework to the stack.
Primary Dewatering
Three, 50% thickeners, each 2 00 feet in diameter, are provided to
dewater spent slurry to 2 5 to 3 0% suspended solids concentration.
The thickeners are of the cable-torque design to allow the rake
to move both parallel to the thickener floor, and move vertically
as a result of thickened bed solids minimizing over-torque
potential on the rakes.
Secondary Dewatering and Waste Stabilization
Six rotary drum filters (five operating, one spare) further
dewater the spent slurry to better than 40% solids cake. The
cake is then fixated and stabilized with flyash and lime for
k.
k
5A-19
-------
final disposal in a landfill. The 40 year landfill, located 3.5
miles from the plant, requires 400 acres for disposal, plus 200
acres as a buffer. The owners have contracted with a local firm
for hauling and placement of the FGD stabilized end product for
$2.19 per ton firm for three years with options for the fourth
and fifth years of the contract. This hauling and placement cost
includes the services of all contractor personnel (25) associated
with what is required for the disposal portion of the operation.
To date, 1.4 millions tons of material have been hauled to and
placed in the landfill.
Fabrication/Construction
Due to the tight site constraints at this former nearly complete
nuclear facility (300 acres vs. 1000 acres for other 1400 MW
units) and available barge shipment, heavy emphasis was placed on
maximum shop fabrication (modularization) of all components.
This approach presented a number of advantages, including:
Minimized laydown space
Cost savings
Improved quality control
Minimized field labor requirements with levelized
manpower
Shortened construction schedule
The FGD system absorber towers were modularized (Figure 6) . The
40 foot 6 inch diameter towers were split in half, all internals
shop installed, and transported to the site via barge. Each half
section, weighing approximately 100 tons, required about one day
to be transported from barge to the absorber foundations on the
site. The circumferential weld tying together upper and lower
halves of the modules required about two weeks for each absorber.
SYSTEMATIC APPROACH TO START-UP
A systematic approach to start-up was undertaken. Included in
this approach were the following phases:
Personnel selection and training
Formalized punchlist program
Electrical verification, instrument calibrations and
check-out
Detailed pre-operational test program
Problem resolution and start-up of trains for steam
blow
Completion of systems start-up
5A-20
A
-------
Personnel
Management, professional and supervisory personnel were selected
in the early stages of the Zimmer conversion. Shortly there-
after, hourly personnel selections began from within Cincinnati
Gas and Electric with the rounding out of the 260 person station
complement from outside the company. The scrubber section is
comprised of a Superintendent, an FGD Engineer, five FGD
supervisors and 20 hourly personnel. Extensive advance training
was undertaken during control system configuration via system
specific contracted book training followed by hands-on control
system simulation training. The same trained operators then
helped conduct and coordinate the formalized punchlist program.
Punchlist Program
During construction, the FGD operating group was heavily involved
in system punchlisting under the direction of the FGD Engineer.
This was highly beneficial in complementing the classroom
training via hands-on field familiarization of FGD systems.
Detailed punchlist reports (Figure 7) were prepared noting
deficiencies, submittal date, expected completion date, priority,
responsible group and disposition. These were submitted to
construction and updated monthly.
Electrical Verification. Instrument Calibrations and Checkout
As various systems were completing, electrical wiring
verifications were being performed. During this phase, the
Bailey Net 9 0 touchscreen distributive control system consoles
were installed and, when ready, control/system interface was
checked. This was done utilizing the FGD operators at the
control consoles under the direction of the electrical
verification group of construction. This further enhanced the
familiarity of the operators with the system, at the same time
building a confidence level on the part of these same people.
Simultaneously, instrument calibrations and check-outs were being
performed and completed.
Pre-operational Testing
As systems completed mechanically and electrically, they were
turned over to the operating group via official release sheets
(Figure 8). Previously developed formalized pre-operational test
procedures were used in full functional testing of the systems
per the FGD Engineer direction and pre-op schedule. All inter-
locks, permissives, setpoints, logic, manual and auto sequences
were checked consistent with the procedures and signed off.
Equipment was first tested with belts dry or fluid circuits with
water to verify readiness. First slurry production occurred on
October 2, 1990.
k
5A-21
-------
Steam Blow
Two of six absorber modules, one of tvo ball mills, cne of ^nree
thickeners, and two of six vacuum filters were readied for --earn
blow. As a result, Zimmer scrubbed every bit of flue gas aver
generated. In preparation for this phase, as actual oper-zcing
difficulties manifested themselves, problem resolution was -ple-
mented to the point that the necessary equipment for steair. blow
of the main plant was proven ready in advance and poised for the
call to duty. Steam blow occurred in November, 1990.
Completion of Initial Start-up
Following steam blow, the main plant was reconfigured and readied
for initial turbine roll and eventual synchronization. 'During
this time, the remaining FGD systems/equipment were pre-op tested
and prepared for anticipated full load operation. The Zimmer
Station produced its first megawatts on December 31, 1990,
reached full load on January 16, 1991 and began commercial
operation on March 31, 1991. The first material hauled to the
landfill occurred on January 9, 1991.
OPERATIONS
The Zimmer FGD waste stabilization system commenced operation on
November 11, 1990. A timetable outlining major milestones is
depicted in Figure 9. The unit passed its compliance test,
required by the State of Ohio, on March 7, 1991, with the FGD
system achieving an average of 96.4% S02 removal (See Figure 10).
The FGD system is operated around-the-clock by five crews, each
comprised of a supervisor and four operators (scrubber, slaker,
fixation, and assistant fixation). At least one instrument and
control person, sometimes more, is routinely assigned to
instrumentation. Maintenance is provided five days per week
during the day shift by a crew comprised of one foreman, eight
craft persons, and one planner. At all other times, and on an
emergency basis, main plant maintenance personnel provide
support.
This nominal 1300 MW unit operates at 1400 MW during the day,
1100 to 1400 MW at night, and maintains a 93% output factor. The
FGD system has not been perfect and there are a few "lessons
learned" to report:
1. Ball Mills - The ability to precisely measure and
control lime flow to the ball mills would enhance
operation.
2. A need for flexible tubing and vibration isolated or
damped equipment, instruments and actuators in the
absorber area proper required post start-up attention.
5A-22
-------
3. Waste Stabilization - This equipment works well except
that frequent cleanings of the dust collection system
are required. A heating system to maintain moisture
laden flows above dewpoint will likely be installed.
4. Thickener rakes originally were too buoyant and
required the addition of ballast.
5. The absorber mist eliminators have experienced
deteriorization of the material of construction. New
317L stainless ME's have been purchased for
installation in January, 1992.
Some areas have performed beyond expectation, such as:
1. Thickener performance has been remarkable. Excellent
settling occurs while never establishing a bed.
2. Installed spares have kept the system on line and in
compliance without a single S02 emission exceedance to
date.
3. Reliable ball mill operation with no significant signs
of liner wear after one year of operation and 100,000
tons of lime slaked per mill.
4. Excellent vacuum filter performance with an average 43%
solids cake.
ADVANCED SCRUBBER
The Zimmer FGD system meets the Electric Power Research
Institute's (EPRI) three criteria for an "advanced" FGD system.
These criteria are:
1. Capable of achieving greater than 95% SC>2 removal
2. Reliability greater than 99%
3. Energy consumption less than 2% of station output
That the first criteria is met is evidenced by the fact that the
system tested at 96% SO_ removal during the compliance test with
only one of two pumps (per absorber) operating at an L/G of 21.
If the spare pump is operated, the L/G would rise to 28, and the
supplier predicts 99% SO removal. A second contract performance
test was conducted and removal efficiency was 94%+. And, Zimmer
has already been awarded the EPA Region V "Award of Excellence in
Sulfur Dioxide Control" for exemplary design and operation of the
FGD system.
5A-23
-------
Although Zimmer has not operated long enough to prove that
reliability is 99%, it is designed to be reliable. The absorber
vessels are fabricated of 317L stainless steel with C-376 alloy
at the absorber inlet. The short absorber outlet duct
incorporates a Pennguard (registered trademark), borosilicate
block lining system. In addition to proper attention to material
selection (based on past operating experience), the system also
features redundancy of critical components, as well as auxiliary
equipment, to assure high reliability. Spare components include:
20% capacity absorber
50% capacity thickener
20% capacity vacuum filter
100% capacity slaking system
Lime storage silo
All pumps (except the vacuum filter feed) have a 100% installed
spare.
The third criteria for an "advanced" FGD system is met by the
fact that the instantaneous maximum FGDS demand (including
heating, sump operation, etc.) is estimated to be 16,285 KW.
Based on a full load gross generation of 1,394,186 KW the
consumption is 1.1% of the unit's gross generation.
CONCLUSION
It is not easy to design, construct, and operate the largest
single FGD system in the world, a 1400 MW behemoth. A team
effort and systematic approach from design through construction
and start-up were required of all parties - the owners, FGD
system contractor, and lime supplier.
From Zimmer's past difficult nuclear problems,
conversion to coal firing has been accomplished,
receives visitors from all over the world and is
example of what can be accomplished.
a successful
Today, Zimmer
held up as an
5A-24
-------
BIBLIOGRAPHY
"Magnesium Enhanced Flue Gas Desulfurization at the Zimmer
Station", T. L. Hart, American Electric Power Service Corp.
"Zimmer FGD System and Start-Up" presented to the Lime User Group
Meeting, Wheeling, West Virginia, May, 1991, W. D. Brockman,
Cincinnati Gas & Electric Company.
"Zimmer Station Thiosorbic FGD System Process Chemistry", Dravo
Lime Company, Pittsburgh, Pennsylvania.
5A-25
-------
ZIMMER COAL CONVERSION PROJECT
COAL STORAGE Pt-E
JUU IMS
-------
FIGURE
2
5A-27
-------
FIGURE 3
MST EUHMATOR BLADES
CLEAN FLUE
GAS OUTLET
STAINLESS MST ELIMINATOR
SPRAY NOZZLES
SILICON CARBIDE
SPRAY NOZZLE
EXPANSION
MIXING /JOINT BYpASs
OEVICE / ^DAMPER
TURNING
VANES
YPASS DUCT
OUTLET
HEAO
.EXPANSION
JOINT
FLUE GAS
INLET
OVERFLOW
BOX
ABSORBER INLET
RJUE CONNECTION
REACTION
TANK
SECONOARARY MIST.
ELIMINATOR
MIST ELIMINATOR _
OVERSPRAY HEADER
PRIMARY MIST.
ELIMINATOR
MIST ELIMINATOR
UNOERSPRAY HEAOER
ABSORBER SPRAY
HEAOER
ABSORBER TRAY-
QUENCH SPRAY-
HEADER
STAINLESS PERFORATED TRAY
FDGS
ABSORBER SYSTEM
Abaorbar Croas-Sactional Vl«w
5A-28
-------
Absorber Towers - 250 MW
Magnesium Enhanced Limestone
Lime
-------
FIGURE 5
ABSORBER
AREA
ELECTRICAL
BUILDING
STACK
FROM
PRECIPITATOR
ABSORBER
(6-TYP)
ABSORBER
AREA
ENCLOSURE
FIGURE 6
ins
5A-30
A
-------
r
Piot No.
10/02/90
<_n
>
do
IIEN COHPONENT SESCR1PII0N PR10RIIV RE5P
NO. CODE GRP.
MJ- ABS. RECIRC. PUHPS 2 CSP
131- HT0RAUL1C DISCHARGE VALVES 2 CSP
136- OUTLET DUCT DRAIN PENETRATION I CSP
137- AIS. HOOF 2 CSP
131- STEPS G0IN6 UP STACK I AROUND 2 CSP
STACK
139- AIS 16. ESN FLON SNITCHES I HECK
160- ALL SSN FLON DETERS 1 HECH
161- NISI ELIltlNATOR BOOSTER PUHPS I NECH
I I 2
162- AIS. AREA IASIET STRAINERS II 1 NECH
7
•63- ELECTRIC JUNCTION IOIES - 7 ELEC
CLIPS
164- AIS II HIST El IN. OUAD. VLV I NECH
CV1J136
163- AIS 13, HIST ELIH. OUAI. VLV I HECH
CV3JJ34
•66- AIS 16, HIST ELIH. HAIRHU. ISO I HECH
«.¥.
167- MS 12, DENSITY HETER BCRAJ230 1 ELEC
161- AIS 16, DENSITY HETER 0CRAJ630 I ELEC
•69- AIS It, SLURRY FEED II CONT I HECH
W.V CVIJIOO
•70- ALL AIS" SLURRY FEED LOOPS' I ELEC
FLON NETER
SUBNIT E1PECTED COMPLETE
DATE COMPLETE DATE
DATE
02/26/90 II II
06/21/90 II II
07/03/90 II II
07/03/90 II li
07/05/90 II II
00/31/90 II II
00/31/90 .11 I i
08/31/90 ! I I !
01/31/90 II II
08/31/90 II II
08/31/90 II II
0B/31/90 II II
08/31/90 II II
08/31/90 II II
08/31/90 II II
08/31/90 II II
08/31/90 II II
NH. H. IIHNER STATION
ABSORBER NODULES SVSTEH PUNCHIIST
AS OF 9/28/90
PRDSLEI1 CORRECTIVE ACTION REHARKS
PLEASE PROVIDE HA1NIENANCE AND ASSEMBLY HAHUALS ON
THE ABS. RECIRC. PUHPS.
HYDRAULIC DISCHARGE VALVES ARE INSTALLED BUT NOT
LABELED VALVE IL06I927 ANI IL06I9270.
NOT SEALED AT PENETRATION INTO AIS. BUILDING ROOF.
DEBRIS, NEEDS CLEANED UP. ROOF RUBBER LOOSE IN
AREAS.
SAPS IN HANDRAILS, NEEDS HANDRAIL.
FLON SNITCH IFISHM60 FOR AGIT. II HAS « 0-13 6PH
RANGE I FLON SNITCH IFISHNU3 FOR ARP II HAS A
0-10 6PN SCALE RAN6E. THESE RAN6E3 ARE REVERSE OF
ALL THE OTHER AGIT'S I ARP'S.
SCALES FOR FLON METERS ON HEBP'S I A6ITAT0RS ARE
TOO NIDE
NEBP'S HAVE 0-20 SPH SCALE FOR 1.2 GPH FLON, AND
ART ASIT'S HAVE 0-10 GPH SCALE FOR 1.0 SPN FLON.
BOTH PUHPS ARE HISSING PACKING I PACtING GLAND
FOLLONER.
STRAINER ASSEHILIES HAVE BEEN PHYSICALLY RENOVED.
HOLD-DOW CLIPS ON JUST ABOUT ALL JUNCTION BOIES
ARE LOOSE OR HISSING - TOO HANY TO SPECIFY.
REHOVEI SINCE HYDRO.
BEHOVED SINCE HYDRO.
•I2FN962I REHOVEI SINCE HYDRO.
DENSITY HETER ON INSTRUMENT RAC( A8-2 HAS A LOOSE
CONNECTION TO MAID.
SHALL REI HIRE INSIDE UNIT IS IEIN6 PINCHED IY
HINGE SCREN.
AIR LEAK OH TUBING CONNECTION INTO DIAPHRAGI1 AREA.
NOTHING BEEN PROVIDED.
NEEDS TO IE INSULAIEI AND SEALED.
INSTALL PIPE SLEEVES.
0-10 GPH SCALE FLON HETER SHOULD IE
FISHH460 ON AGIT. II I 0-13 GPH FLON
HETER SHOULD IE ON (FISHM6J) - ARP II.
CHANGE FLON HETERS TO NARRONER SCALE
REAOINGS (SAY 0-3 GPHI.
REPACK GLAND AND INSTALL GLAHD FOLLONER.
REINSTALL IN PROPER LOCATION AND CHECK
FOR LEAKS.
POSITION PROPERLY I TIGHTEN.
REINSTALL VALVE ANI CHECK FOR LEAKS.
REINSTALL VALVE AND CHECK FOR LEAKS.
REINSTALL VALVE ANI CHECK FOR LEAKS.
TIGHTEN.
FII AND REPOSITION HIRE.
REPAIR LEAK - HAY NEED NEN TUIING.
NO CHAN6E AS OF 9/28/90.
n
G>
c
3)
m
ALL HETERS, 2 TO A HODULE. HAVE TNO ALUM NUB
GROUNDING STRAPS TO ATTACH TO 'PANCAKE' HANDLES AT
CONNECTING FLAN6ES.
ATTACH STRAPS.
-------
FIGURE 3
lha Chotmdl fia A BaeMc Cameo*
Cakanbw SoUhan tanr Conpo*
VaOovtonPoMrottl&t Company
/WiluuiOw1** >11— l«»lu«CauuUten.p»if-'""*<;
COAL CONVERSION PROJECT
December 13, 1990
Release No. 321
SUBJECT: Equipment Releases
FROM:
R. Miller
TO:
D. N. McNeal
Attached you will find equipment release for the following item(s):
A) ABS-163-01 Absorber Recirc. & Blowdown - Complete
The Final Acceptance Letter or signed System Release Form should be returned to
T. R. Adams, Construction Manager.
If you have any questions or comments, please contact me at extension 576.
RFM/esf
cc: J. A. Howard
T. R. Adams/R. S. Kern w/o attachments
R. Life
Civil w/o attachments
Electrical
Mechanical
K. Loving/M. Hopper/R. Palmer
MK-Ferguson
Babcock & Wilcox
L.K. Comstock
J . Bartles w/o attachments
P. King (CG&E)
D. Drengler
Chief Construction Coordinator
1781 U.S. 52 • P.O. BOX 7901 • MOSCOW. OHIO 45153
5A-32
A
-------
FIGURE 9
ZIMMER CHRONOLOGY
START NUCLEAR CONSTRUCTION
CEASE WORK
DECISION TO CONVERT
SCRUBBER CONTRACT AWARD
START CONVERSION
START SCRUBBER ERECTION
FIRST SLURRY PRODUCTION
STEAM BLOWS & FIRST WASTE PRODUCTION
INITIAL TURBINE ROLL
FIRST MEGAWATTS
FIRST MATERIAL TO LANDFILL
FULL LOAD OPERATION
SCRUBBER COMPLIANCE TEST
COMMERCIAL OPERATION
90 DAY RELIABILITY RUN (SCHEDULED)
1969
1982
1984
2/26/86
1987
8/1/88
10/2/90
NOVEMBER, 90
12/19/91
12/31/91
1/9/91
1/13/91
MARCH, 1991
3/31/91
MARCH, 1992
5A-33
-------
FIGURE 10
COMPLIANCE TEST RESULTS
Inlet S02 Concentration
L/G
S02 Removal Efficiency
Flow to Stack
System Differential Pressure
Average Module Velocity
Module pH
Module Density
Stoichiometry
Coal: Btu per lb.
% S
% Moisture
% Ash
5.54 lb. per million Btu
21 gal per 1000 ACFM
96.4%
4.4 million ACFM
3.3 inches water
11.4 ft. per sec.
6.2
10%
<1.03
11,836
3.01
8.62
9.50
5A-34
-------
RESULTS OF AN INVESTIGATION TO IMPROVE THE PERFORMANCE AND RELIABILITY OF
HL&P'S LIMESTONE ELECTRIC GENERATING STATION FGD SYSTEM
M. D. Bailey
Houston Lighting & Power
P.O. Box 610
Jewett, Texas 75846
0. W. Hargrove, Jr.
J. B. Jarvis
M. Stohs
G. E. Stevens
Radian Corporation
8501 MoPac Boulevard
Austin, Texas 78759
R. E. Moser
Electric Power Research Institute
3412 Hillview Avenue
Palo Alto, California 94303
k.
5A-35
-------
Intentionally Blank Page
5A-36
-------
ABSTRACT
This paper presents the results of a three-phase investigation designed to improve
the performance and reliability of the Houston Lighting & Power Company's (HL&P)
Limestone Electric Generating Station (LEGS) FGD system. The investigation centers
on identifying additives and additive combinations which are cost effective, and at
the same time, improve performance and reliability. The test program was co-funded
by HL&P and the Electric Power Research Institute (EPRI) under RP 2248-1, FGD
Chemical Process Problems.
The first phase of the test program involved bench-scale testing to simulate the
effects of various additive combinations on the operation of LEGS. This work was
conducted using EPRI's 5-acfm bench-scale FGD system located at Radian Corporation
in Austin, Texas. The second phase of the test program involved testing the more
promising additive combinations on the mini-pilot FGD system at EPRI's High Sulfur
Test Center (HSTC). The third phase of the test program consisted of full-scale
testing of DBA in combination with sulfur at LEGS to demonstrate the feasibility of
this additive combination.
HL&P was using dibasic acid (DBA) alone to enhance S02 removal in the LEGS system.
Based on the bench-scale test results, HL&P switched to the combination of DBA and
thiosulfate (sulfur) addition in July 1989. This has resulted in a modest decrease
in DBA consumption, less scaling of the mist eliminators and other tower internals,
and a significant reduction in operating costs.
Preceding page blank
5A-37
-------
INTRODUCTION
The HL&P Limestone Electric Generating Station (LEGS) is a two-unit, 1560-MW,
lignite-fired power station located in Limestone County outside Jewett, Texas. The
sulfur content in the lignite can vary but averages about 1 wt.%. Sulfur dioxide
emissions from each of two units are controlled using a conventional limestone FGD
system which consists of five open spray towers (four in operation at full load
with one spare) per unit. The S02 emissions are regulated pursuant to the 1977
NSPS.
HL&P has used dibasic acid (DBA) to enhance S02 removal at LEGS since 1986. An
enhancement additive such as DBA is needed at LEGS because there is insufficient
gas/liquid contact area and liquid-phase alkalinity for S02 removal unless an S02
removal enhancement additive is employed. However, HL&P is concerned about the
long-term availability and cost of DBA. DBA is produced as a byproduct in the
manufacture of adipic acid, and its supply is limited.
The availability and cost of DBA was one factor which prompted HL&P to investigate
alternative additives. A second motivating factor involved reliability problems at
LEGS; chiefly, scaling of the mist eliminator system. The FGD system is a natural
oxidation system, and gypsum scale formation on tower internals has been a trouble-
some problem. As a result, HL&P was interested in the use of thiosulfate to in-
hibit sulfite oxidation and minimize the potential for gypsum scaling. Thiosul-
fate, which can be produced in-situ by adding elemental sulfur to the reaction
tank, has been shown to be effective in preventing gypsum scale formation in sev-
eral utility FGD systems (I). However, previous research (2) and testing (3) sug-
gested that thiosulfate could increase the consumption of DBA. Bench-scale tests
were undertaken at HL&P's request to verify this effect on DBA usage.
5A-38
i
-------
HL&P initiated a study of potential alternative additives which could be used in
the limestone system. Since the research results would be applicable to the util-
ity industry in general, EPRI contributed to the effort under RP 2248-1, FGD
Chemical Process Problems. This program is designed to assist the utility industry
in solving chemically related problems associated with wet FGD systems.
The research effort consisted of three phases:
• Laboratory testing of additives and additive combinations using
EPRI's 5-acfm bench-scale FGD system;
• Testing of selected additive combinations in the mini-pilot system
at the HSTC; and
• Demonstration of the most promising additive combination at LEGS.
A total of 15 tests were conducted in a laboratory bench-scale FGD system. The
additives tested included adipic acid, DBA, sodium formate, sulfur (thiosulfate),
and magnesium. In addition to tests with the individual additives, a number of
additive combinations were evaluated which took advantage of synergism among the
different compounds. Based on the bench-scale test results, recommendations were
made for testing the more promising combinations on the mini-pilot system at the
HSTC.
A total of 14 tests were conducted at EPRI's HSTC on the mini-pilot unit which was
configured to simulate operation of the FGD system at HL&P's LEGS. The objective
of the mini-pilot test program was to verify the results of the bench-scale tests
and provide additional information (e.g., the solids dewatering characteristics)
which could not be adequately characterized at the bench-scale level. Based on the
mini-pilot results, an economic evaluation was performed to determine the most
cost-effective additive combinations. HL&P can use these results as a guide for
further testing and/or to revise operating conditions at LEGS.
The remainder of the paper describes test results from all three phases of the test
program.
5A-39
-------
LABORATORY TEST PROGRAM
This section presents a description of the bench-scale test program. The objec-
tives of the program included:
1. Determining the effect of each additive or additive combination on
important performance indicators such as S02 removal, solids oxi-
dation level, operating pH, limestone utilization, and solution
composition; and
2. Measuring the consumption rates of the additives for each additive
combination under conditions representative of LEGS.
The test results described in this report were obtained using EPRI's 5-acfm bench-
scale system located at Radian Corporation in Austin (a similar system is located
at the HSTC). The bench-scale system is very small, but is useful in evaluating
the effects of process variables on FGD system chemistry. The system uses a syn-
thetic gas mixture and can be controlled at stable operating conditions.
The bench-scale tests performed in this study were batch tests (no solids blowdown)
in which the system was spiked with the additive(s) being tested. Then, the con-
centration of the additive(s) was monitored with time. In general, the additive
concentration decayed with time due to both chemical degradation of the additive
and coprecipitation of the additive in the calcium sulfite crystal matrix. In a
typical test, the additive concentration was monitored for a period of about 12
hours. Material balance calculations were employed to estimate steady-state con-
sumption rates from the batch test results.
S02 removal was monitored throughout the test. In addition, samples were obtained
at the end of each test to characterize solids oxidation, limestone utilization,
and changes in system chemistry.
Ground limestone, obtained from LEGS, was added to the system to maintain a pH of
5.7 for most tests. In some cases, it was evident that limestone utilization had
improved as a result of changes in system chemistry. When possible, the system pH
was increased at the end of a test to characterize S02 removal at the baseline
limestone utilization level (around 90% to 95%).
5A-40
i
-------
The laboratory tests were conducted at conditions which simulated the operation of
LEGS. These conditions included operation at 60°C (140°F) to simulate the adia-
batic saturation temperature of a lignite system, the use of cooling tower blowdown
from LEGS as makeup water, an adjustment in the flue gas oxygen content to achieve
about 25% oxidation in the solids, and the addition of a small amount of fly ash
obtained from LEGS. The inlet S02 concentration was controlled at 1400 ppm.
Some of the important system performance results are shown in Figures 1 through 3.
Figure 1 summarizes the effect of the additives on sulfite oxidation. Compared to
the baseline oxidation of 27%, the addition of additives to concentrations required
for 90% S02 removal reduced the sulfite oxidation. Magnesium and DBA reduced the
oxidation to about 17% to 18%. This reduction was caused by increased absorption
of S02 while absorption of 02 from the flue gas remained unchanged. If all other
conditions are constant, an increased S02/02 ratio in the liquid results in a
decreased fraction of sulfite being oxidized. Formate, on the other hand, reduced
the sulfite oxidation to a significantly lower level than magnesium or DBA (10%).
This is consistent with previous test results which have shown that formate is an
oxidation inhibitor. Figure 1 also shows that addition of thiosulfate in combi-
nation with the other additives reduced the oxidation to below 10%. Thiosulfate
inhibits oxidation by reacting with free radicals generated in the initiation and
propagation steps of sulfite oxidation. Since up to 15 mole percent sulfate is
included as a solid solution in the calcium sulfite hemihydrate crystal lattice,
inhibition of sulfite oxidation to a level significantly below 15% eliminates the
formation of gypsum solids and gypsum scaling conditions.
Figure 2 shows the effect that magnesium concentration had on S02 removal at the
bench scale. Magnesium increases liquid-phase alkalinity (and therefore the S02
removal efficiency) by increasing the amount of dissolved sulfite ions in solution.
Figure 2 shows that 89% removal was achieved with about 3,400 ppm magnesium added
in solution. Figure 2 also shows that 91% removal was achieved with about 2,000
ppm magnesium with the addition of 13 mmol/liter (1,500 ppm) of thiosulfate. Thio-
sulfate's inhibition of sulfite oxidation results in more sulfite and less sulfate
in solution and, thus, an increase in liquid-phase alkalinity. In addition, Figure
2 shows that less magnesium is required if organic buffers are added. This
illustrates that there are several combinations of additives that will achieve
greater than 90% S02 removal.
5A-41
-------
Figure 3 shows the effect that the organic buffers, DBA and formate, have on S02
removal efficiency. Because it buffers at a higher pH than formate, DBA is more
effective in increasing S02 removal efficiency. At a concentration of 30 milli-
equivalents of acid (about 2,000 ppm DBA and 1,350 ppm formate), DBA achieved 94%
S02 removal compared to 88% with formate. To achieve 94% removal on the bench sys-
tem, about 2,700 ppm formate was required. Figure 3 also shows that a combination
of Mg, thiosulfate, and either formate or DBA can achieve a high S02 removal effi-
ciency with a lower concentration of organic buffer.
Organic acid consumption rates were also measured during the laboratory experi-
ments. Losses of organic acids such as DBA and formate occur primarily through
four mechanisms:
• Chemical degradation;
• Coprecipitation of the organic acids in the calcium sulfite crystal
matrix;
• Vaporization (formate only, not DBA); and
• Solution losses in the liquid blowdown streams, mainly in the liquid
associated with the waste solids filter cake.
The first three of these mechanisms are termed non-solution losses and are
independent of the dewatering efficiency.
Figures 4 through 6 summarize the organic acid consumption results from the HL&P
laboratory study. Figure 4 shows the effect that thiosulfate had on DBA non-
solution losses. The DBA loss rate decreased at 500 ppm thiosulfate and then
increased at 1,500 ppm thiosulfate. Results from two mini-pilot tests are also
plotted in Figure 4 and show a decrease in the DBA non-solution loss rate with the
addition of thiosulfate at the HSTC. However, the effect of a higher thiosulfate
concentration on the DBA consumption was not tested.
Prior to the bench-scale tests, it was anticipated that the reduced oxidation (more
calcium sulfite produced) resulting from operation with thiosulfate would increase
the amount of DBA coprecipitated in the solids and increase the overall DBA con-
sumption rate. These bench results illustrate an important competing factor in
moderate-calcium systems. As the concentration of thiosulfate increases and the
5A-42
-------
sulfite oxidation decreases in a moderate-calcium system such as at LEGS, the cal-
cium concentration decreases, reaches a minimum value, and then begins to increase
again. This parallels the consumption rates measured for DBA with increasing thio-
sulfate. Therefore, the calcium concentration appears to be a major factor influ-
encing DBA consumption in moderate-calcium, inhibited-oxidation systems. (Copre-
cipitation of calcium formate, adipate, glutarate, and succinate is a function of
the calcium-organic salt activity product.)
Figure 5 shows that increasing the magnesium concentration decreased both DBA and
formate non-solution loss rates. This again can be attributed to a reduction in
calcium concentration. As the magnesium concentration is increased, the dissolved
sulfite and sulfate concentrations also increase, and the calcium concentration
decreases due to solubility constraints. As discussed above, a reduced calcium
concentration should reduce the calcium-organic salt activity product and the
amount of organic buffer coprecipitated in the calcium sulfite solids.
Other than the effect of magnesium, formate non-solution losses were affected only
by the formate concentration in the bench-scale tests. This is shown in Figure 6
which includes runs with and without thiosulfate. Thiosulfate would not be ex-
pected to show the same effect on formate consumption that it had on DBA consump-
tion because of the oxidation inhibition of formate alone. The addition of 2,700
ppm formate reduced the oxidation to 10%, and the addition of thiosulfate did not
reduce the oxidation significantly below this value. Thus, the addition of thio-
sulfate under these conditions did not lower the dissolved calcium concentration
appreciably, and the amount of formate being coprecipitated would not be expected
to change. However, it should be noted that the mini-pilot test results did indi-
cate that the presence of thiosulfate lowered the formate non-solution loss rate at
a constant dissolved calcium concentration.
In summary, the bench-scale work showed that there are additives or combinations of
additives other than DBA alone which might achieve the desired S02 removal effi-
ciency at LEGS. There were also some obvious tradeoffs identified. Further test-
ing at a larger scale was required to quantify the economics associated with the
additive combinations. These results prompted the testing of DBA and thiosulfate
at HL&P's LEGS.
5A-43
-------
MINI-PILOT TEST PROGRAM
Following completion of the laboratory testing, additional simulation of the LEGS
FGD system was performed at the HSTC mini-pilot plant. The specific test objec-
tives were to:
# Confirm additive combinations which could provide the required S02
removal efficiency;
# Accurately measure the additive consumption rates;
# Determine the effect of different additive combinations on process
operation and performance; and
# Develop cost estimates for each of the additive combinations tested
which were specific to HL&P's LEGS.
The mini-pilot system at the HSTC consists of an 18-inch diameter absorber with a
single spray nozzle in combination with a counterflow tray. Because of the small
diameter of the absorber, a tray is required to obtain an S02 removal level repre-
sentative of the full-scale system. The unit treats flue gas from the 650-MW New
York State Electric and Gas' Kintigh Station.
Two modifications to the mini-pilot system were required to allow a simulation of
HL&P's LEGS. The primary modification was the installation of a 300 psi (1000
lb/hr) steam boiler. Since LEGS fires lignite coal, steam injection is required to
raise the adiabatic saturation temperature of the flue gas from about 50 to 60°C.
A second modification consisted of fabricating an additive feed system which would
allow the simultaneous addition of up to three additives.
In most cases, the test procedure involved adjusting the pH and additive concentra-
tions to achieve 89% to 91% S02 removal and 94% to 96% limestone utilization. The
test was started once the system stabilized at the desired operating conditions.
The system was then operated for three solid-phase residence times (typically 90 to
100 hours) to produce solids representative of the conditions in the system. Three
liquid- and solid-phase samples were then collected from the reaction tank during
the next 24 hours, resulting in a total test time of about 120 hours.
Based on the bench-scale test results, a test program was designed for the mini-
pilot system at the HSTC. This consisted of 14 tests with different combinations
5A-44
A
-------
of DBA, sodium formate, sodium thiosulfate, and magnesium. The operating condi-
tions employed in each test included an inlet S02 concentration of 1400 ppm, an L/G
ratio of 54.5 gal/1000 acf, and 15% solids in the recirculating slurry. The pro-
cess performance results of primary concern included limestone utilization, sulfite
oxidation level, and solids dewatering properties. Table 1 summarizes the additive
concentrations required to achieve 90% S02 removal during the mini-pilot HL&P simu-
lation. The following conclusions were drawn from the mini-pilot test results:
• The mini-pilot system provided a fairly accurate simulation of the
LEGS F6D system under conditions of DBA addition alone and the com-
bined use of DBA and thiosulfate. Significant deviations included
lower sulfite oxidation (with thiosulfate present in the slurry) and
slightly higher limestone utilization in the mini-pilot tests.
• Magnesium addition increased S02 removal efficiency and reduced
limestone utilization at a constant pH. The S02 removal efficiency
increased because magnesium increases the solubility of sulfite (a
buffer which boosts S02 removal) and because of the increased con-
centration of solid limestone. Since the buffer capacity of sulfite
decreases as the pH is reduced, it was not possible to lower the pH
to maintain a constant limestone utilization while still obtaining
the required S0? removal efficiency when magnesium was added to the
FGD system. This result confirmed the trend observed in the bench-
scale testing.
• The LEGS FGD system will require higher magnesium concentrations
than those used in the mini-pilot tests. This is because the
sulfite-to-sulfate ratio will be lower at the higher oxidation rates
measured in the LEGS system.
• The increase in the dissolved magnesium and sulfite concentrations
associated with magnesium addition resulted in a significant de-
crease in the dewatering properties of the waste solids produced in
the FGD system. Filter leaf tests showed that the cake solids con-
centration decreased from 71 wt.% during the DBA/thiosulfate test to
43 wt.% during the magnesium/thiosulfate test. This decrease in
solids content increases both the additive solution loss rate and
the sludge production rate.
• As was noted during bench-scale testing, formate was a less effec-
tive buffer than DBA. Therefore, a higher concentration of formate
(on both weight and molar bases) will be required to achieve the
same S02 removal efficiency as the DBA concentration currently
maintained in the LEGS system.
• The addition of sodium formate, thiosulfate, or magnesium to the
LEGS system will cause a decrease in the dissolved calcium concen-
tration (relative to system operation with DBA only). This will
result in lower coprecipitation losses of additives. In the case of
5A-45
-------
sodium formate and thiosulfate addition, this will also allow oper-
ation at a higher reaction tank pH while maintaining a constant
limestone utilization.
• Thiosulfate was a more effective oxidation inhibitor than sodium
formate at comparable concentrations. To ensure scale-free opera-
tion and to maximize the effectiveness of sodium formate, it should
be used in combination with thiosulfate.
• The driest solids were produced in tests which simulated the LEGS
system using DBA before and after sulfur addition. (Table 2 summa-
rizes the dewatering test results.) Therefore, additive solution
losses and waste production rates will be higher for all of the
alternate additive combinations examined during the mini-pilot
tests. Because of this, the characteristics of the fly ash/sludge
mixture should be evaluated during any full-scale additive tests in
the LEGS system to determine the stability of solids sent to the
landfill.
• The presence of both formate and DBA in solution can result in un-
predictable solids dewatering properties. Thus, the solids charac-
teristics should be monitored closely if a switch is made from DBA
to formate.
Organic acid consumption rates were closely monitored during the mini-pilot test
series. Important observations pertaining to the DBA and formate consumption rates
are as follows:
• As in the LEGS FGD system and at the bench scale, a reduction in the
DBA consumption rate occurred when thiosulfate was added to the
mini-pilot system. This decrease was primarily due to the decrease
in the calcium concentration which occurred because the system
became subsaturated with respect to gypsum. (Additive coprecipita-
tion rates are a function of the calcium-additive activity product.)
However, a lower chemical degradation rate, as a result of the lower
oxidation level, may have also contributed to the decrease.
• The addition of sulfur (thiosulfate) will reduce the overall cost of
additives at LEGS by reducing the DBA consumption rate. (Sulfur is
a very inexpensive additive.) This result was observed at the
bench-scale and ultimately at LEGS so DBA/thiosulfate was used as
the basis for comparison.
• The addition of magnesium will also reduce organic additive copre-
cipitation losses because of a reduction in the dissolved calcium
concentration.
• In the presence of a higher thiosulfate concentration (2000 ppm),
the DBA and formate chemical consumption rates were lower than
expected based on changes in the additive and calcium concentrations
5A-46
A
-------
alone. This is a significant result, since maintaining a higher
thiosulfate concentration could result in a net savings by lowering
the DBA or formate consumption rate.
• When compared on a mass basis, the formate chemical loss rate (i.e.,
not including solution losses with the waste liquor) was consis-
tently higher than the DBA loss rate under similar operating condi-
tions. This was partially due to the higher formate concentration
required to achieve the same S02 removal efficiency as 1500 ppm DBA.
Based on the mini-pilot test results, the operating costs for six different addi-
tive combinations were estimated using a design basis specific to HL&P's LEGS FGD
system. The estimates were made for the annual additive and limestone reagent
costs only and did not include capital costs or estimated changes in maintenance
costs. The results of the economic analysis are presented in Table 3 as the cost
of the option relative to the cost of the current mode of operation, i.e., using
DBA in combination with thiosulfate. The following conclusions can be drawn from
the results of the economic evaluation:
• The current mode of operation using DBA and thiosulfate appears to
be the most cost-effective option available to HL&P at this time.
However, if the cost of DBA rises by 50% relative to the cost of
formate, then sodium formate addition will become more attractive.
• From an additive cost standpoint, the most attractive alternative to
current operation is replacing DBA addition with sodium formate
addition and operating with a higher thiosulfate concentration (2000
ppm). If a test using sodium formate is conducted, the effect of
changes in the filter cake moisture on the stability of the fly
ash/sludge mixture must be determined.
• The costs associated with the additive combinations involving dolo-
mitic lime addition are subject to some uncertainty because the
exact magnesium concentrations that would be required in the LEGS
system are difficult to predict. In addition, the effect of a
higher magnesium concentration on limestone utilization cannot be
accurately predicted, and this can have a significant impact on the
costs of these options.
• Dolomitic lime addition will cause the filter cake moisture to
increase, and this change may affect the stability of the fly
ash/sludge mixture. Other important considerations for magnesium
addition are the capital and operating costs for lime storage and
handling equipment and the complexity of operating with three
additives if used in combination with DBA or formate.
5A-47
-------
In summary, based on the results of the mini-pilot tests, it appears that HL&P is
currently operating with the most cost-effective combination of additives avail-
able. The most attractive alternative identified in these tests is the use of
sodium formate in combination with 2000 ppm thiosulfate.
FULL-SCALE TEST PROGRAM
HL&P has been adding DBA to their FGD system since April 1986 to assist in achiev-
ing S02 removal efficiency requirements. However, in 1989, they were still experi-
encing considerable gypsum scaling in the mist eliminator, in the absorber spray
nozzles near the gas inlet, and at the wet/dry interface in the flue gas inlet. To
investigate this problem, EPRI provided engineering and technical support to HL&P
under RP 2248-1, FGD Chemical Process Problems. The focus of the work at LEGS was
to evaluate the addition of emulsified sulfur to produce thiosulfate in the FGD
system. Thiosulfate inhibits sulfite oxidation and can be very effective in mini-
mizing problems associated with gypsum scaling.
Problem Description
The FGD system at LEGS was designed as a natural oxidation system. Prior to thio-
sulfate addition, typical solids oxidation levels were on the order of 25% to 35%,
depending on the sulfur content of the fuel. At this oxidation level, the scrub-
bing solution was slightly supersaturated with respect to gypsum. This increased
the tendency for gypsum scale formation in the mist eliminators and on tower
internals.
Prior to the addition of thiosulfate, one of the modules was opened for inspection
which had been in service for about six weeks. The mist eliminators were estimated
to be about 50% to 60% plugged, and numerous nozzles were severely scaled. Previ-
ous inspections had routinely found nozzles to be completely plugged due to the
gypsum scaling.
The gypsum scaling problem in the mist eliminators periodically caused liquid and
slurry carryover. Due to the arrangement of the ductwork, this liquid could reach
the steam reheat coils where it produced a thick scale on the heat exchange sur-
faces. Damage to the reheat coils has been extensive in the past, requiring
replacement of entire banks of heat exchangers.
5A-48
A
-------
It is likely that gypsum scaling also had an adverse effect on the S02 removal
efficiency. Maldistribution of gas, caused by mist eliminator plugging, results in
lower liquid-to-gas ratios in the cleaner towers because they are forced to treat a
higher fraction of the flue gas. Plugged nozzles result in an uneven spray pattern
in the absorbers, causing poor gas/liquid contacting. A reduction in S02 removal
efficiency requires higher DBA concentrations to meet S02 removal requirements,
which increases DBA consumption. It may also result in higher L/G requirements and
ultimately allow less gas bypass for reheat.
The annual cost associated with gypsum scaling in the mist eliminators alone was
estimated by the plant to be $1.4 million. This includes costs for mist eliminator
replacement parts, replacement of steam coils, and labor required to replace the
equipment and clean the ducts, nozzles, and other surfaces.
The elimination of gypsum scaling reduced these costs dramatically. However, it
should be noted that plugging of the mist eliminators still occurs due to occa-
sional malfunctions in the mist eliminator wash system. Numerous instances of this
have been reported by the plant. Therefore, the ultimate cost reductions which can
be achieved depend on optimizing this system as well.
Sulfur Addition Testing
The objectives of the sulfur addition test at LEGS included the following:
• Determine the thiosulfate concentration necessary to lower the
solids oxidation level and reduce the scaling potential within the
system;
• Determine the effect of thiosulfate on DBA consumption; and
• Determine the effect of thiosulfate on important system performance
parameters.
Sulfur was added to one absorber module beginning in June 1989. There were no
adverse affects noted such as increased DBA consumption or major changes in the
settling properties of the solids. Consequently, sulfur was added to the remainder
of Unit 1 at the end of June and to Unit 2 in early July.
5A-49
-------
The test module was taken off line in mid-July after about six weeks of operation
with sulfur addition. Upon inspection, the test module was found to be exception-
ally clean and free of scale. According to plant personnel, the mist eliminators
appeared cleaner than they were at the beginning of the test.
There continue to be some problems with mist eliminator plugging. These appear to
be related to inadequate mist eliminator washing, particularly during occasional
periods of poor limestone utilization. However, the material plugging the mist
eliminators is much softer than that observed before sulfur addition. Conse-
quently, the affected mist eliminator sections can be cleaned more easily. Pre-
viously, vigorous cleaning of the plugged mist eliminator sections was required,
resulting in considerable damage and high replacement rates.
Test results comparing important operating parameters before and after sulfur addi-
tion are presented in Table 4. These parameters include solution composition,
operating pH, relative saturations of gypsum and calcium sulfite, solids oxidation
level, and limestone utilization. These results are discussed and compared to the
bench-scale and mini-pilot test results below.
There was initially some concern that thiosulfate addition might cause an increase
in DBA consumption. This possibility was suggested based on previous research
results (2) and previous tests on the mini-pilot system (3). (However, the bench
and mini-pilot simulations under HL&P conditions indicated that DBA consumption
would be reduced.) During previous mini-pilot tests under high-calcium conditions,
a two-fold increase in adipic acid consumption was noted with the addition of
thiosulfate. The same effect would be expected for DBA. In high-calcium systems,
however, thiosulfate does not significantly depress the calcium concentration.
Apparently, the increased coprecipitation rate resulting from lower solids
oxidation was the primary cause of the increased adipic acid consumption in tests
prior to the HL&P simulation tests.
For the moderate-calcium chemistry of the LEGS FGD system, however, thiosulfate
causes the calcium concentration to drop. This offsets the tendency for increased
coprecipitation due to the reduction in the sulfite oxidation level. At the
moderate-calcium chemistry of LEGS, there appears to be a modest decrease in DBA
5A-50
-------
consumption. This effect was first noted during the bench-scale tests and was
verified on the mini-pilot system.
Based on the bench-scale results, HL&P converted to operation with DBA and thiosul-
fate in August 1989 and has since enjoyed a reduction in DBA consumption. The cost
savings resulting from lower DBA consumption are offset somewhat by the cost for
adding sulfur to produce thiosulfate. (HL&P is currently paying $192 per ton for
elemental sulfur. To maintain a thiosulfate concentration around 1,000 ppm, the
estimated yearly cost for sulfur addition is about $55,000.) However, the primary
motivation for adding sulfur was to obtain more reliable mist eliminator perfor-
mance, rather than to decrease DBA consumption. The fact that operation with thio-
sulfate is possible, and does not increase DBA consumption, is one of the major
results of the study.
HL&P could have added thiosulfate without benefit of the bench-scale test results.
However, had DBA consumption increased significantly (as was initially expected),
the cost to perform the test could have been very high since DBA consumption would
have remained high until the thiosulfate was purged from the system. The bench-
scale test results provided some assurance that DBA consumption would not increase.
CONCLUSIONS
• LEGS has seen a significant improvement in the operation of their
system with the addition of sulfur (thiosulfate). This includes
less scaling of the mist eliminator system, less plugging of spray
nozzles, and lower maintenance requirements for cleaning the
absorber vessels. With modifications to the mist eliminator wash
system, this should translate into significant cost savings relative
to operation with DBA alone.
• The use of thiosulfate has resulted in a modest reduction in DBA
consumption at Limestone. This is significant because a substantial
increase in consumption had initially been expected based on the
results of previous studies. HL&P is therefore able to enjoy the
benefits associated with thiosulfate without increased DBA costs.
• Thiosulfate is not as effective in reducing solids oxidation at LEGS
as was expected based on the bench-scale and mini-pilot tests. How-
ever, the reduction in solids oxidation at LEGS was sufficient to
allow subsaturated operation and to obtain the benefits in scale
reduction associated with thiosulfate.
5A-51
-------
• A number of alternative additive combinations were evaluated in
the bench-scale and mini-pilot test programs. The most promising
alternative additive combination was thiosulfate/sodium formate. A
preliminary economic evaluation suggests that none of the additive
combinations will provide a significant reduction in total costs
over DBA and thiosulfate. However, this alternative may be of
interest to HL&P if the price of DBA increases.#
• The applicability of additives and additive combinations depends on
a number of factors. These include the type of action required
(i.e., for SO, removal enhancement or scale control) and the chemis-
try of the FGD system. These tradeoffs mean that additive combina-
tions suitable for one FGD system may be unsuitable for another. In
addition, secondary effects, such as the influence of additives on
the solids settling and dewatering properties, are important. All
of these factors must be taken into account in determining the opti-
mum additive combination for a particular FGD system.
• The general success of the alternative additive and additive combi-
nations in favorably affecting the performance of limestone scrub-
bing is an important result in itself. This means that there are
numerous types of additives and suppliers from which the utility
industry can choose for improving the S02 removal efficiency, reli-
ability, and overall economics of wet limestone FGD.
ACKNOWLEDGMENTS
The work reported in this paper is the result of research carried out in part at
EPRI's High Sulfur Test Center (HSTC) located near Barker, New York. We wish to
acknowledge the support of the HSTC cosponsors: New York State Electric and Gas,
Empire State Electric Energy Research Corporation, Electric Power Development
Company, Ltd., and the U.S. Department of Energy. The cosponsors provide valuable
technical review of the work in progress as well as funding test center operations.
REFERENCES
1. Moser, R.E., et al. "Control and Reduction of Gypsum Scale in Wet Lime/Lime-
stone FGD Systems by Addition of Thiosulfate: Summary of Field Experience,"
presented at the First Combined FGD and Dry S02 Control Symposium, St. Louis,
Missouri, October 1988.
2. Jarvis, J.B., J.C. Terry, S.A. Schubert, and D.L. Utley. Effect of Trace
Metals and Sulfite Oxidation on Adipic Acid Degradation. EPA 68-02-3171,
April 1982.
3. Burke, J.M., et al. "HSTC Pilot and Mini-Pilot Wet Scrubber Progress Reports
(November and December 1988)," prepared for EPRI.
5A-52
-------
Table 1
SUMMARY OF CONDITIONS REQUIRED TO ACHIEVE 90% S02 REMOVAL ON MINI-PILOT SYSTEM
DBA Formate Added Mg Thiosulfate
EiL (ppm) (PPm) (ppm) (ppm)
5.7 1,500
5.9 1,500 830
5.9 2,090 2,100
5.9 1,800
6.0 1,800 700
6.0 2,000 2,100
5.9 760 1,070 830
6.0 820 510 1,900
6.0 880 510 870
Table 2
EFFECTS OF ADDITIVE COMBINATIONS ON DEWATERING EFFICIENCY
Oxidation Filter Cake
Additive Combinations (%) (wt. %)
DBA 35 72
DBA/Thiosulfate 8 71
DBA/Thiosulfate/Magnesium 8 49
Thiosulfate/Magnesium 8 43
Formate 9 59
Formate/Thiosulfate 9 57
Formate/Thiosulfate/Magnesium 6 44
DBA/Formate/Thiosulfate 8 43
5A-53
-------
Table 3
OPERATING COST COMPARISON OF DIFFERENT COMBINATIONS
Additive Combination
DBA
Thiosulfate
Formate
Thiosulfate
DBA
Magnesium
Thiosulfate
Formate
Magnesium
Thiosulfate
Magnesium
Thiosulfate
Relative Annual
Limestone and Additive Cost
1.00
1.08
1.10
1.26
1.32
Table 4
FULL-SCALE TEST RESULTS - SUMMARY OF KEY PERFORMANCE INDICATORS
Operating Period
Additives
PH
S02 Removal, %
Limestone Utilization, %
Sulfite Oxidation, %
Gypsum Relative Saturation
April-May 1989
1,450 ppm DBA
5.7
85
94
35
0.96
August-September 1989
1,450 ppm DBA
1,370 ppm Thiosulfate
6.0
84
85a
13
0.79
August = 90%, September = 80%; limestone utilization upset in September.
5A-54
-------
ftlg, 3400-3000 ppm
M
o
£
3
o>
DBA. 2000
Form**, 2700 ppm
0
200
400
600
•00
1000 1200 1400 1600
THoeultate (ppm)
Figure 1. Effect of Thlosutfate on Sulfite Oxidation
100
DBA, 800 pm
TNoetJfato, 1000
TNoiuKati, 1450 ppm
Formate, 540 ppm * a
ThioeuHAB, lOOOppm
MgOnly
0
300
1000 1500 2000
2500 3000
3300 4000
Added Uagneekan Concentration (ppm)
Figure 2. Effect of Magnesium on SOj Removal
DBA or Formate +
Mg IThloaUfate
20 30 40
Organic Acid Concentration (meq)
Figure 3. Effect of Organic Acid Concentration on SOj Removal
DBA, 2000 ppm
Mlnmiot DBA-'
1500 ppm
0 400 600 100 1000 1 200
TNoeJfate ConcantraUon (ppm)
Figure 4. Effect of Thioeulfate on
DBA Non-Solution Losee*
Formats, 1350 ppm
DBA, 2000 ppm
500 1 000 1 500 2000 2500 3000
Added Ug Concentration (ppm)
Figure 5. Effect of Added Magnesium on
Organic Acid Non-Solution Loese*
3500 4000
1000 1500 2000
Formate Concentration (ppm)
Figure 6. Effect of Formate Concentration on
Formate Non-Solution Loeeee
5A-55
-------
Intentionally Blank Page
/
5A-56
-------
FULL-SCALE DEMONSTRATION OF EDTA AND SULFUR ADDITION
TO CONTROL SULFITE OXIDATION
G. M. Blythe
Radian Corporation
8501 Mopac Boulevard
Austin, Texas 78759
T. J. Slater
Southwestern Electric Power Company
Hallsville, Texas 75650
R. E. Moser
Electric Power Research Institute
3412 Hillview Avenue
Palo Alto, California 94303
Preceding page blank
5A-57
-------
Intentionally Blank Page
5^-58
-------
ABSTRACT
In 1985, Southwestern Electric Power Co. (SWEPCO) began operation of their Henry W.
Pirkey Power Plant, located in northeast Texas. The minemouth, lignite-fired plant
is rated at 720 gross MW and is equipped with an electrostatic precipitator for
particulate control, followed by a dual-loop limestone FGD system for S02 control.
After start-up, the natural-oxidation FGD system was troubled with significant
gypsum scaling, particularly in the lower slurry loops and in the mist eliminators.
In 1987, SWEPCO began adding sodium thiosulfate to operate in an inhibited sulfite
oxidation mode, and in 1988, switched to adding elemental sulfur emulsion as a
lower-cost method of introducing the thiosulfate ion. Although sulfur addition was
effective at lowering sulfite oxidation levels and reducing gypsum scaling ten-
dencies, there remained times when sulfite oxidation exceeded 15%. This level is
typically the maximum at which subsaturated operation with respect to gypsum can be
maintained. Consequently, minor problems with gypsum scaling remained.
In Hay 1991, SWEPCO began a full-scale evaluation of an EPRI-patented concept of
adding EDTA as well as sulfur to control oxidation at lower percentages than could
be achieved with sulfur addition alone. While sulfite oxidation percentages were
not observed to drop dramatically after EDTA addition began, the concept was shown
to continuously maintain oxidation percentages below 15% over a four-month period.
Previous operation at similar plant operating conditions suggested that, with sul-
fur addition alone, sulfite oxidation levels of 20% and greater would have been
encountered intermittently over this period. As a result of the four-month oper-
ating period where the system continuously operated subsaturated with respect to
gypsum, remaining areas of gypsum scale deposits in the absorbers appear to be
dissolving. SWEPCO plans to continue to evaluate EDTA and sulfur addition at least
through the next station outage period, which is planned for Hay 1992.
Preceding page blank
k
5A-59
-------
INTRODUCTION
The Henry W. Pirkey Power Plant is owned and operated by SWEPCO of Shreveport,
Louisiana. The power plant is a 720-MW (gross), mine-mouth, lignite-fired unit
located in northeast Texas, and is equipped with a weighted-wire electrostatic
precipitator (ESP) for particulate control and a limestone flue gas desulfurization
(FGD) system for S02 emissions control.
At start-up in 1985, the FGD system was plagued by a number of operational prob-
lems. SWEPCO, with support from EPRI, the FGD system vendor, and Radian Corpora-
tion as a contractor to both EPRI and SWEPCO, has worked diligently to improve FGD
system performance. The most recent FGD system improvement has been the addition
of EDTA to compliment the effects of emulsified sulfur additive to maintain sulfite
oxidation levels below 15%. This paper discusses the improvements implemented to
the Pirkey FGD system, with emphasis on the most recent EDTA addition tests.
SYSTEM DESCRIPTION
Figure 1 is a simplified flow diagram for the Pirkey FGD system. Flue gas exits
the boiler and passes through a weighted-wire ESP for particulate control. From
the ESP, the flue gas passes through an axial-flow ID fan and into the FGD system.
A small portion of the flue gas is bypassed to provide reheat for the saturated gas
treated in the absorbers. At the flue gas entrance to each absorber, there is a
quench spray header, then the gas is contacted by a recirculating slurry of lime-
stone and calcium sulfite and sulfate salts. Gas leaving each absorber enters two
stages of chevron-style mist eliminators, then exits the absorber to be combined
with bypassed gas and discharged from the stack.
5A-60
A
A
-------
5A-61
The FGD system has four absorber modules. The original design was for only three
absorbers to be in service at full load, but due to a higher-than-expected lignite
sulfur content, all four modules are normally operated. The individual absorbers
are dual-loop towers with trap-out trays separating the upper and lower loop slur-
ries. The upper loops of each absorber share a reaction tank with a second absorb-
er (i.e., there are two upper loop reaction tanks for four absorbers), while each
absorber has an integral reaction tank in its base for its lower loop. In each
absorber, the upper loop recycle slurry is fed through three spray headers, result-
ing in a liquid-to-gas ratio (L/G) of 35 gal/1000 acf (design conditions). The
upper loop absorber sections were originally designed to be open spray towers, but
perforated plate trays have been retrofitted below the bottom spray header level to
improve S02 removal performance. The lower loop sections of each absorber operate
as open spray towers, with two spray header levels providing a design L/G of 17
gal/1000 acf. Both loops were originally designed to operate in a natural sulfite
oxidation mode, but have since been converted to inhibited oxidation.
Limestone reagent is received sized to 1/4 x 1/4 inch, and is ground in two Koppers
ball mills. The product from each ball mill is fed to a bank of four Krebs hydro-
clones. The underflow from the hydroclones is returned to the ball mill, while the
overflow is sent to the limestone slurry storage tank. Makeup reagent slurry is
fed to the absorber loops on pH control.
Slurry blowdown from the lower absorber loops is by gravity overflow to a waste
slurry sump. Slurry from this sump is fed to one of two 110-foot thickeners, both
of which normally operate. The overflow from the thickeners flows to a reclaim
water tank, while the underflow sludge is pumped to a storage tank and then to one
of three rotary-type vacuum filters for secondary dewatering. The filtrate is sent
to the reclaim water tank, and the filter cake, at about 70 wt.% solids, is sent to
pug mills and blended with fly ash for mine disposal.
Reclaimed water from the thickeners and vacuum filters is used for limestone grind-
ing, mist eliminator wash, and general plant wash down. Ash pond water is used to
supplement reclaimed water levels as needed.
The system was designed to achieve 90% S02 removal across the absorbers, and 85%
overall removal after the bypassed gas is recombined with the scrubbed gas. For
5A-61
-------
the expected worst-case lignite sulfur content of 6.67 lb S02 per 106 Btu, this
removal level would be adequate to maintain S02 emissions below the limit of 1.2
lb/106 Btu. However, actual lignite quality has been quite variable, with sulfur
levels reaching as high as 10 lb/106 Btu.
REVIEW OF FGD SYSTEM OPERATING HISTORY
Upon start-up in 1985, SWEPCO encountered a number of operating problems with the
FGD system, including:
• Ball mills operating below design capacity,
• Inadequate S02 removal in the absorbers,
• Gypsum scaling in the absorbers,
• Gypsum scaling in the mist eliminators,
• Positive water balance conditions,
• Variable limestone utilization, and
• Poor control of reaction tank weight percent solids levels.
Obviously, several of these problems were related. For example, low weight percent
solids levels in the reaction tanks lower solid-phase residence times, which can
result in gypsum relative saturation levels increasing into scaling regimes and low
limestone utilization values.
The most pervasive problem was the gypsum scaling in the absorbers and mist elim-
inators. A crew of 20 laborers was required to clean the absorbers on a full-time
basis. When one absorber was cleaned, it would be put back in service, and the
next absorber in line would be taken out of service for cleaning. This resulted in
an annual maintenance cost of about $400,000 per year for absorber cleaning and for
replacement of mist eliminator blades damaged during scale removal.
During the first five years of operation of the system, a number of modifications
were implemented to improve system operation. The first significant improvement
was to improve the control of weight percent solids levels in the absorber reaction
tanks. Leakage of slurry from the upper to the lower loops past the trap-out trays
had the effect of causing the level of slurry in the upper loop reaction tanks to
5A-62
A
-------
drop. To maintain level, recovered mist eliminator wash water had to be added to
the upper loop tanks. This had the effect of lowering the weight percent solids
levels there, and due to the cascading slurry flow, lowered solids levels in the
lower loops as well.
At the suggestion of the FGD system vendor, piping was added to allow a return of
slurry from the lower loop reaction tanks to the upper loop tanks to allow the
slurry level to be maintained without excessive dilution with mist eliminator wash
water. Better control of slurry densities has resulted in improvements in lime-
stone utilization and reductions in gypsum scaling tendencies in the absorbers as
well as in the mist eliminators.
Other changes implemented included retrofitting perforated plate trays in the upper
loop portion of the spray tower and changing spray nozzle type in both the upper
and lower slurry loops. The trays were retrofitted to improve S02 removal perfor-
mance. These trays were installed below the lowest spray header in the upper loop,
just above the trap-out trays between the two loops. The trays increase the liquid
hold-up in the absorbers to improve gas/liquid contacting and enhance limestone
dissolution within the absorbers. The nozzles were replaced to improve droplet
atomization, also for improved S02 removal performance. The new nozzle type has
proven to be less prone to plugging, and has demonstrated an increased service life
over the original style.
The gypsum scaling tendencies in the absorbers and mist eliminators were largely
due to the natural sulfite oxidation percentages encountered in the FGD system.
Natural oxidation levels averaged about 45%, which represents a worst case with
respect to gypsum scaling tendencies: too high for all of the sulfate formed to
coprecipitate with calcium sulfite, yet not high enough to produce a substantial
surface area of gypsum particles in the slurry to allow rapid gypsum crystal growth
at low relative saturation values.
In 1987, sodium thiosulfate addition testing was conducted in an attempt to reduce
gypsum scaling tendencies by inhibiting sulfite oxidation to low levels. SWEPCO
was assisted in conducting these tests by EPRI as part of Research Project 2248-1,
"Chemical Process Problems," with Radian as the prime contractor. By reducing the
sulfite oxidation percentage from the natural level of about 45% to levels below
5A-63
-------
approximately 15%, all of the sulfate formed is coprecipitated with calcium sul-
fite, and no gypsum solids are formed. At Pirkey, lowering the oxidation fraction
below 15% was observed to lower gypsum relative saturation levels in the reaction
tanks from approximately 1.2 to levels below 1.0. Operating in a subsaturated
regime with respect to gypsum greatly reduced tendencies for scale formation
throughout the absorbers.
In May 1988, SWEPCO switched from adding sodium thiosulfate to adding elemental
sulfur. Dissolved sulfur reacts with liquid-phase sulfite to generate thiosulfate
ion in a more cost-effective manner than by adding sodium thiosulfate directly.
The switch to elemental sulfur emulsion reduced additive costs from approximately
$150,000 per year to less than $40,000 per year. Both additives were effective at
eliminating the requirement for outside contract labor for continual absorber and
mist eliminator cleaning, which had cost approximately $400,000 per year.
However, with either source of thiosulfate ion, sulfite oxidation could not be con-
trolled below 15% under all operating conditions. Even at thiosulfate ion concen-
trations as high as 4000 ppm, oxidation levels as high as 25% were encountered on
occasion. During operation at 25% oxidation levels, gypsum relative saturation
values exceeded 1.0, indicating conditions that were conducive for gypsum scale
formation.
There are several factors that appear to promote higher oxidation percentages under
some conditions, even in the presence of relatively high thiosulfate ion concentra-
tions. These include:
• Reduced load operation. During portions of the year, the Pirkey
Plant operates at reduced megawatt output rates, particularly during
off-peak time periods in the day. Low-load operation generally
requires higher boiler excess air levels, which can promote sulfite
oxidation in the F6D system. Also, during days where the load
cycles to lower levels for only portions of the day, the Pirkey
Plant often retains all four absorber modules in service. The spray
headers for the upper and lower loops are each manifolded together,
so it is not possible to turn off individual upper or lower loop
spray header levels to reduce L/G ratios. The net effect is that,
at reduced loads, the S02 pickup rate is reduced while the oxygen
pickup rate, which is directly related to L/G ratio, remains high.
This also promotes higher oxidation percentages.
5A-64
A
-------
• Higher slurry temperatures. The Texas lignite fired at this plant
has a relatively high moisture content, such that the flue gas
moisture content is also relatively high. The flue gas adiabatic
saturation temperature is typically about 135*F. A wet scrubber
recirculating slurry temperature typically remains near the flue gas
adiabatic saturation temperature, so slurry temperatures at the
Pirkey FGD system are typically 5 to 10*F greater than in FGD sys-
tems on coal-fired plants. Oxidation rates have in general been
observed to increase at elevated slurry temperatures, presumably due
to increases in the intrinsic reaction rate for one or more of the
oxidation reactions. Thus, higher oxidation percentages would be
expected in a lignite-fired plant FGD system such as at Pirkey than
in a similar system on a bituminous or subbituminous coal-fired
unit.
• Variable coal sulfur levels. At Pirkey, the FGD system inlet S02
concentrations can range from as low as equivalent to 3 lb/106 Btu
to as high as 10 lb/10 Btu. In general, for a given absorber and
with most other operating variables being fixed, the oxidation rate,
in terms of moles of sulfite oxidized per unit time, remains rela-
tively constant. If the absorber inlet S02 concentration is reduced
due to lower lignite sulfur content, and no other operating vari-
ables change, the SO, pickup rate (in moles per unit time) will be
reduced while the sulfite oxidation rate will remain relatively
constant. Hence, the percent of sulfite which is oxidized will be
increased. While it may be possible to control oxidation percent-
ages to 15% or less when the inlet S02 concentrations at Pirkey are
at the high end of this range, at inlet concentrations equivalent to
3 to 4 lb/106 Btu, such low oxidation percentages become difficult
to maintain.
• High transition metals content in the scrubber liouor. Transition
metals such as iron and manganese are thought to catalyze sulfite
oxidation by reacting with sulfite ion to produce sulfite free
radicals. Additional chain reactions occur to produce sulfate and
more free radicals and to regenerate the transition metal catalyst.
The lignite fired at Pirkey produces a high ash loading at the inlet
to the ESP upstream of the FGD system. Particularly towards the end
of the six-month periods between scheduled outages, ESP performance
begins to degrade to the point that a significant quantity of fly
ash is carried into the wet FGD system. Scrubber solids inert
contents are commonly measured between 2 and 3 wt.%, and to a large
part, consist of fly ash removed in the FGD system. Fly ash is a
common source of soluble transition metals in FGD liquors, so in
such incidents of degrading ESP performance, there should be ample
supplies of oxidation reaction catalysts in the Pirkey FGD liquor.
These aspects of the Pirkey FGD system design and operation make it a difficult
case for maintaining sulfite oxidation percentages below 15% at all times, and
hence, for avoiding operation in a gypsum scaling regime. The ability to operate
in a subsaturated mode most of the time has resulted in greatly improved FGD system
5A-65
-------
operation and in elimination of the need for a full-time laborer crew for cleaning
absorber modules. However, as long as excursions to 25% or greater sulfite oxida-
tion occur, the absorbers will never operate in a completely scale-free mode.
From the 1987/1988 time period until early 1991, the Pirkey FGD system operated in
the inhibited-oxidation mode with these occasional excursions to 25% sulfite oxida-
tion or greater. Clean-up of scale in absorbers has for the most part been limited
to the semi-annual power plant scheduled outages, and has been accomplished by FGD
operating and maintenance crews with little support from outside contract labor.
However, clean-up activities during these periods have remained substantial. Gyp-
sum scale formation is most prevalent just under the trap-out trays between the
upper and lower loops, around the circumference of the absorber. These are rela-
tively quiescent zones with respect to slurry and flue gas flow, so that once
gypsum scale forms in these areas, it does not readily redissolved once oxidation
levels drop and the slurry liquor returns to subsaturated conditions. Other clean-
up activities include scale removal from portions of the lower mist eliminators and
cleaning of plugged spray nozzles.
Consequently, there is still an incentive for further improvement of the operatii
of the Pirkey FGD system. The biggest improvement in FGD system operation would
likely come from changes that would allow sulfite oxidation to be controlled at
less than 15% under all operating conditions. Continuous operation at low oxida-
tion percentages, and hence, at subsaturated conditions with respect to gypsum,
would accelerate dissolution of gypsum scale that remains in the tower and would
limit formation of additional scale. The current project involves the addition of
EDTA to the scrubber liquor in an attempt to control the sulfite oxidation below
15% at all times.
CURRENT PROJECT
Background
EPRI-funded laboratory- and bench-scale research conducted by Radian Corporation
identified EDTA as an additive that would enhance the effectiveness of thiosulfate
in inhibiting sulfite oxidation in wet limestone FGD systems (!). A mechanism pro-
posed to describe sulfite oxidation in wet limestone FGD systems involves transi
tion metals such as iron or manganese, which react with sulfite ions in solution
5A-66
A
-------
produce sulfite free radicals (2). These sulfite free radicals react with oxygen
and other sulfite ions in a series of chain reactions to produce sulfate and other
free radicals, and to regenerate the transition metal catalyst. Thiosulfate limits
sulfite oxidation by reacting with these free radicals to break the chain reac-
tions. EDTA is thought to limit sulfite oxidation by chelating transition metals
in solution in the FGD system liquor. These chelated metals are not available to
react with sulfite ions, and hence, sulfite free radicals are less likely to be
formed. EDTA and thiosulfate together appear to be synergistic because EDTA
inhibits the formation of the free radicals which begin the oxidation chain reac-
tions, while thiosulfate reacts with free radicals formed to terminate the chain
reactions. The EPRI-funded laboratory- and bench-scale tests determined that a
combination of EDTA at relatively low levels (15 to 100 ppm) and thiosulfate could
be employed to achieve lower sulfite oxidation rates than could be achieved with
either additive alone. As a result of this testing, EPRI applied for and received
a patent for the concept of employing EDTA as an additive to lower oxidation per-
centages in inhibited oxidation FGD systems (3).
The first full-scale evaluation of this concept was conducted at the Associated
Electric Cooperative's Thomas Hill Unit 3 FGD system. These tests were conducted
in December 1990 and March 1991. However, at that site, conditions were such that
oxidation levels were quite low before the EDTA was added. That system employs
both thiosulfate and sodium formate additives, a combination that has been pre-
viously demonstrated to control oxidation at lower levels than can be achieved
through thiosulfate addition alone (4). Also, the coal fired at Thomas Hill Unit 3
has a rather high sulfur content (equivalent to 8 to 10 lb S02/106 Btu), which also
tends to result in low oxidation percentages. In these full-scale tests, baseline
oxidation percentages were found to be very low, in the range of 3% to 5%. The
addition of EDTA to levels as high as 50 ppm in the FGD liquor was not found to
measurably lower oxidation percentages from that extremely low level.
The Thomas Hill FGD system did not represent an ideal case for testing the concept
of EDTA addition, from the perspective that there was little room for further
reductions from what proved to be extremely low oxidation fractions. The Pirkey
system was seen as a particularly challenging case for the concept in that, at
times, the conditions at Pirkey were not conducive to operating in an inhibited
oxidation mode.
5A-67
-------
SWEPCO was interested in exploring alternatives that would allow operation of the
FGD system at Pirkey at oxidation percentages below 15% at all times, and EPRI was
interested in demonstrating their EDTA additive patent at full-scale sites. Conse-
quently, a full-scale test of EDTA addition was conducted at the Pirkey FGD system
with EPRI support starting in May 1991.
Short-Term Test
SWEPCO was planning an outage for the Pirkey plant for mid-May of 1991. The period
just before an outage usually represents a worst case for controlling sulfite
oxidation percentages: ESP performance has usually degraded to the point where
significant quantities of fly ash are carried into the FGD system, and system loads
must be reduced to remain within compliance with respect to opacity regulations.
Approximately two weeks before the outage was scheduled to begin, a full-scale EDTA
addition test was begun.
EDTA was purchased as a sodium salt in a 38 wt.% aqueous solution. At the begin-
ning of the test, EDTA was metered into one of the two upper loop reaction tanks at
a rate calculated to achieve a 100 ppm (as the sodium salt) EDTA level, while thio-
sulfate levels were maintained at the normal level of 1200 ppm through the addition
of elemental sulfur. EDTA levels in the lower loops of the two affected modules
were allowed to increase to 100 ppm by the normal displacement of slurry from the
upper to the lower loops.
The initial tests were not conducted on the entire system because of the potential
for problems in the thickeners. As sulfite oxidation percentages drop below
approximately 5%, the relatively pure calcium sulfite crystals formed tend to grow
to relatively large particle sizes which settle much more quickly than typical
calcium sulfite solids (5). There was a concern that, if the initial tests were
conducted on the entire system, and sulfite oxidation percentages dropped suddenly
below 5%, the thickener rakes might become overloaded before the normal inventory
of sludge could be withdrawn. From this perspective, it would have been desirable
to conduct this first test on only one module. However, two absorber modules had
to be tested with EDTA addition because one upper loop reaction tank serves two
absorbers.
5A-68
A
-------
Analytical results for FGD system slurry samples collected during this time period
are summarized in Table 1. When EDTA addition began, sulfite oxidation levels did
not drop markedly, and solids settling rates were not measurably affected. How-
ever, these results are confounded by the fact that the unit load was dropping
throughout the test due to degrading ESP performance and the need to remain within
stack flue gas opacity compliance. The dropping load over this period had two
effects which tended to promote higher sulfite oxidation percentages: increasing
flue gas oxygen content and decreasing S02 pickup per absorber module.
The data in the table show that the oxidation percentage in one of the modules to
which EDTA was added dropped by about 2 percentage points from the baseline (no
EDTA) value. By the time the last sample was taken, the oxidation percentage had
returned to the baseline value of 12%. However, under previous circumstances (i.e.
no EDTA addition), the sulfite oxidation percentage under such low-load conditions
would have climbed well above 15%. For example, during first tests with elemental
sulfur addition in the spring of 1988, when the unit load dropped to 300 MW, sul-
fite oxidation levels ranged from 19% to 26% in spite of thiosulfate concentrations
in the range of 3000 to 4000 ppm (6).
Note that the oxidation percentages in another module to which EDTA was not
directly added were very similar to those for the module to which EDTA was added.
At first glance this might suggest that EDTA addition was not the cause of the
lowering of oxidation fractions on May 9 and 10. However, because of the reuse of
water from the thickeners, the EDTA level in the "control" modules had already
climbed to 15 ppm by the time the May 9 samples were taken. At this point, it is
not well documented what levels of EDTA are required to control oxidation percent-
ages at lower levels than can be achieved with thiosulfate alone; a level of 15 ppm
may well be sufficient to produce these effects. Consequently, we do not have
"control" data from the same sample times for modules with no EDTA in the liquor.
The EDTA test period was cut short at one week due to the rapidly deteriorating ESP
performance. After the system was off line, an inspection of the modules revealed
that the amount of scale remaining in the lower portions of the reaction tanks was
noticeably reduced from that observed after clean-up during the previous outage.
5A-69
-------
No significant amount of scale was cleaned out of the absorbers during the May out-
age, whereas scale clean-up had been conducted during all of the previous outages
since the unit began operating in the inhibited oxidation mode in 1987. After a
cleanout of the slurry headers and nozzles to clear pluggages, the absorbers were
put back into service essentially as found.
It is speculated that this one week of continuous operation at oxidation fractions
below 15% and gypsum relative saturation levels below 1.0 allowed a portion of the
scale present in the absorbers to dissolve. Such dissolution no doubt occurs under
normal operation with thiosulfate addition alone. However, in the past, excursions
of high sulfite oxidation and supersaturated gypsum relative saturation levels,
particularly just before outages, probably negated much of those effects. It is
also quite likely that, during the one-week EDTA addition test, gypsum scale dis-
solving and coprecipitating with calcium sulfite raised the solid-phase oxidation
percentages measured.
Although the results of this one-week EDTA addition test were not dramatic, SWEPCO
felt that they were positive enough to warrant a longer-term evaluation of EDTA
addition to the Pirkey F6D system. Consequently, F6D liquor was maintained at EDTA
levels of approximately 15 to 25 ppm for most of the four-month period between the
May 1991 and October 1991 outages.
Long-Term Tests
By the time the spring outage began on May 13, EDTA levels in the entire F6D system
liquor were in the range of 55 to 65 ppm. Because most of the liquor was retained
in the system during the outage, it was decided to monitor oxidation levels as the
EDTA levels decayed due to solution losses with waste solids and to add more EDTA
when oxidation levels began to rise.
The results of sample analyses conducted during this long-term EDTA test period are
summarized in Table 2. Through mid-July 1991, sulfite oxidation values of 6% to 7%
were commonly measured. These values represent as low an oxidation percentage as
has ever been measured in this system during inhibited oxidation operation. Near
the end of July, oxidation percentages climbed to approximately 14%, although
liquid-phase gypsum relative saturation levels remained below 0.5.
5A-70
A
-------
t
At this point, no EDTA had been added to the FGD system since before the May out-
age, and EDTA levels were estimated to have dropped into the range of 5 to 10 ppm.
On August 5, 55 gallons of 38 wt.% EDTA solution were added to the system (by add-
ing it to the reagent slurry storage tank). This was a sufficient amount of EDTA
to raise the EDTA level in the system to about 15 ppm. The next sample after the
EDTA addition dropped to approximately 7% to 8% sulfite oxidation. From this time
through the beginning of the fall outage, EDTA was added to the system as required
to maintain EDTA levels between approximately 15 and 25 ppm. Oxidation fractions
measured in the remainder of August and September remained at 11% or lower.
For comparison, Table 3 summarizes analytical data from system operation during
September and October 1990, just before the fall outage that year. With thiosul-
fate levels maintained at 1200 ppm and no EDTA addition, sulfite oxidation levels
were measured ranging from 19% to 28%. While the unit operated at slightly lower
boiler load and lignite sulfur levels in 1990, these differences from the 1991
levels are not thought to markedly affect the sulfite oxidation levels measured.
When the unit was brought off line in early October of this year, the absorbers
were cleaner still than during the May outage. A thin layer of hard scale remains
on the lower loop reaction tank walls, below the slurry operating level, and on the
recirculation pump suction screens. The integral reaction tank walls above the
slurry level are now completely scale free, whereas these areas previously had sig-
nificant scale buildup. Scale remains in the quiescent zones just under the slurry
trap-out trays between the upper and lower loops, around the circumference of the
absorbers. However, there are areas where large pieces of this scale have dropped
from the wall into the reaction tank, and much of the remaining deposits show evi-
dence of eroding away. The upper loops and 95% or more of the mist eliminators
remain virtually scale-free, although these areas have been relatively scale-free
ever since the unit has operated in the inhibited oxidation mode.
It should be noted that the clean-up of scale from the walls of the lower loop,
just below the trap-out trays, cannot be completely attributed to the addition of
EDTA. During the May 1991 outage, small slurry nozzles were added to the upper
slurry headers in the lower loops to spray slurry into these otherwise quiescent
areas. These nozzles have clearly aided in dissolving the hard gypsum scale pre-
sent in these areas. However, the addition of EDTA to maintain sulfite oxidation
5A-71
-------
levels below 15% and completely eliminating operation at supersaturation with
respect to gypsum has no doubt greatly enhanced the effectiveness of these nozzles
in dissolving this scale.
The one prevalent operating problem that remains is frequent plugging of nozzles in
the absorber spray headers. When the absorbers were inspected during the October
1991 outage, as many as half of the nozzles were plugged. The nozzles plug with
pieces of rubber that break loose from rubber-lined slurry feed headers. During
the October 1991 outage, the last of the rubber-lined slurry headers have been
replaced with stainless steel headers. It is hoped that this replacement will vir-
tually eliminate such widespread incidents of nozzle plugging. With better slurry
spray coverage in the lower loop, it is hoped that the dissolution of remaining
gypsum scale will be further enhanced.
Because sulfite oxidation levels appear to be controllable below 15%, and because
the conditions of the absorber internals have continued to improve since beginning
EDTA addition, SWEPCO plans to continue EDTA addition at least until the Spring
1992 outage. This will allow further documentation of the ability to control oxi-
dation levels below 15% and determination of whether continued operation at sub-
saturated conditions with respect to gypsum will completely dissolve the remaining
gypsum scale.
Although tetrasodium EDTA is a relatively expensive chemical at about $1.30/1b (dry
weight basis), very low levels are required. Also, the chemical is very stable,
such that makeup should be required only to replace solution losses with the solid
waste stream leaving the system. Annual EDTA costs for the Pirkey system are esti-
mated at about $4,000 per year, which is about l/10th the estimated costs for ele-
mental sulfur. It is also possible that elemental sulfur addition rates can be
reduced somewhat from present levels due to the apparent tempering effects of EDTA
addition on sulfite oxidation percentages.
The biggest remaining issue regarding EDTA addition to the Pirkey FGD system is how
to measure liquor EDTA concentrations to determine when more EDTA should be added
to the reagent storage tank. Standard colorimetric techniques for determining EDTA
concentrations and/or chelating capacity are subject to interferences in most FGD
liquor matrices, and provide no useful information. Some success has been achieved
5A-72
A
-------
using an ion chromatograph technique for determining EDTA concentrations, but this
technique has a detection limit of about 15 ppm.
For the testing described in this paper, EDTA addition rates were based on concen-
tration values estimated by material balance around the FGD system. For the short-
term tests conducted in May 1991, the ion chromatograph measurement technique
yielded EDTA concentration values near those estimated by material balance. For
the longer-term tests in June through September 1991, target EDTA levels were
lower, approaching the detection limit for the ion chromatograph method. Also, the
Pirkey laboratory does not have an ion chromatograph to allow on-site determination
of actual EDTA levels. It is expected that, during the November 1991 through
Spring 1992 time period, EDTA addition rates will be continue to be added on the
basis of material balance calculations. Occasional off-site ion chromatograph
analyses will be conducted to verify actual levels.
SUMMARY
The results of this four-month evaluation indicate that EDTA addition to the 25 ppm
level in the FGD liquor, along with maintaining thiosulfate levels at 1200 ppm, can
prevent excursions of sulfite oxidation above 15% in the Pirkey FGD system. Avoid-
ing such excursions appears to be instrumental in dissolving existing gypsum scale
in the absorbers and avoiding additional scale formation. Based on the positive
results of this four-month evaluation, SWEPCO plans to continue to evaluate EDTA
addition until the next scheduled outage in May 1992. Over this time period, EDTA
levels will be maintained near 25 ppm in the FGD system liquor, and sulfite oxida-
tion percentages will be measured periodically. During the May 1992 outage, the
absorbers will be inspected to determine if remaining areas of gypsum scale in the
absorbers are being eliminated.
If the May 1992 results continue to look as promising as the current results, it is
likely that EDTA addition will continue. Future optimization work will likely
attempt to offset the minor expense of the EDTA added through reduced sulfur addi-
tion rates.
The concept of combining EDTA and sulfur addition appears to be ready to evaluate
at other sites. Candidate sites would be FGD systems that, like Pirkey, occa-
sionally have excursions of sulfite oxidation above the 15% level, sites that would
k.
5A-73
-------
like to evaluate EDTA addition as a means of reducing sulfur additive costs, and/or
sites that currently employ sulfur addition to maintain sulfite oxidation percent-
ages at low levels (5% to 10%) and would like to evaluate the potential of EDTA
addition to improve solids dewatering properties by lowering sulfite oxidation
below 5%.
REFERENCES
1. Mailer, G., F.B. Heserole, and R.E. Moser. "Use of EDTA and Thiosulfate to
Inhibit Sulfite Oxidation in Wet Limestone Flue Gas Desulfurization Processes:
Results of Laboratory-Scale and Bench-Scale Testing." Presented at the EPRI/
EPA 1990 S02 Control Symposium, New Orleans, LA, May 8-11, 1990.
2. Mailer, G., et al. "Inhibited Sulfite Oxidation by Thiosulfate in Wet Lime/
Limestone FGD Processes: Results of Laboratory Studies and Testing at EPRI's
High Sulfur Test Center Mini-Pilot." Presented at the EPA/EPRI First Combined
FGD and Dry S02 Control Symposium, St. Louis, MO, October 25-28, 1988.
3. U.S. Patent No. 4,994,246.
4. Burke, J.M., M. Stohs, T.J. Price, and R.E. Moser. "Results of Sodium Formate
Addition Tests at EPRI's High Sulfur Test Center and Associated Electric Coop-
erative's Thomas Hill Unit 3 FGD System." Presented at the EPRI/EPA 1990 S02
Control Symposium, New Orleans, LA, May 8-11, 1990.
5. Moser, R.E., J.D. Col ley, J.G. Noblett, and A.F. Jones. "Control and Reduc-
tion of Gypsum Scale in Wet Lime/Limestone FGD Systems by Addition of Thiosul-
fate: Summary of Field Experiences." Presented at the EPA/EPRI First Com-
bined FGD and Dry S02 Control Symposium, St. Louis, M0, October 25-28, 1988.
6. Colley, J.D., D.R. Owens, and E.S. Roothaan. "Results of Thiosulfate Addition
to Southwestern Electric Power Company's Henry W. Pirkey Power Plant FGD
System." Technical Note, Radian Corporation, Austin, TX, July 30, 1988.
5A-74
A
A
-------
Table 1
SUMMARY OF RESULTS FROM SHORT-TERM EDTA ADDITION TESTS AT THE PIRKEY POWER PLANT FGD SYSTEM
Date: 5/7/91 5/7/91 5/7/91 5/7/91 5/9/91 5/9/91 5/9/91 5/9/91 5/10/91 5/10/91 5/10/91 5/10/91 5/13/91 5/13/91 5/13/91 5/13/91
Sample Location": AU AL BU BL AU AL BU BL AU AL BU BL AU AL BU PL
Daily Average Unit 490 490 490 490 467 467 467 467 394 394 394 394 232° 232° 232° 232°
Load, MW
Lignite S Content, 5.7 5.7 5.7 5.7 6.5 6.5 6.5 6.5 5.6 5.6 5.6 5.6 5.6 5.6 5.6 5.6
lb S02/106 Btu
12.5 11.6
75 87
0.8 0.9
33 ' 31
"Absorber A had EDTA spiked directly into its upper loop reaction tank, while Absorbers B and D had EDTA entering only from the reclaim water system.
AU = Upper loop reaction tank for Absorbers A and C.
BU = Upper loop reaction tank for Absorbers B and D.
AL through DL denote the lower loop reaction tanks for Absorbers A through D, respectively.
bAbsorber B was off line due to low load.
Solid-Phase Analyses:
Sulfite Oxidation, X 11.5 12.2 11.5 11 10 10.4 9.8 10 9.6 10 9.6 9.6 12.2 12.4
Limestone Utiliza- 84 90 85 90 78 85 86 91 85 92 90 94 83 87
•T t i on, %
V]
U1
Liquid-Phase Analyses:
Gypsum Relative 0.4 0.6 0.3 0.4 0.3 0.3 0.3 0.3 0.4 0.5 0.4 0.6 0.9 0.9
Saturation
EDTA Concentration, 0 0 0 0 102 99 15 18 91 97 30 30 41 49
ppm
c0nly three modules in service.
-------
Table 2
SUMMARY OF RESULTS FROM LONGER-TERM EDTA ADDITION TESTS AT THE PIRKEY POWER PLANT FGD SYSTEM
CJl
>
¦
v]
CT\
Date:
Sample Location":
Dally Average Unit Load, MW
Lignite S Content,
lb S02/106 Btu
7/12/91 7/12/91 7/22/91 7/22/91 7/23/91 7/23/91 7/26/91 7/26/91 7/30/91 7/30/91 7/31/91 7/31/91
AU AL AU AL AU AL AU AL AU AL AU AL
534
6.8
534
6.8
532
5.3
532
5.3
617
5.9
617
5.9
639
4.2
639
4.2
598
5.2
598
5.2
564
5.0
Date:
Sample Location":
564
5.0
Solid-Phase Analyses:
Sulfite Oxidation, % 6.5 6.7 8.9 9 9.3 8.5 9.7 10 13.8 13.7 13.9 13.3
Limestone Utilization, % 93 96 85 91 84 88 91 92 80 90 85 93
Liquid-Phase Analyses:
Gypsum Relative Saturation NA NA 0.2 0.2 0.1 0.1 0.5 0.3 0.7 0.4 0.6 0.3
EDTA Cone., ppm (est.) 10 10 5-10 5-10 5-10 5-10 5-10 5-10 5 5 5 5
8/9/91 8/9/91 8/23/91 8/23/91 8/28/91 8/28/91 9/10/91 9/10/91 9/13/91 9/13/91 9/20/91 9/20/91
AU AL AU AL AU AL AU AL AU AL AU AL
Daily Average Unit Load, MW
Lignite S Content,
lb S02/106 Btu
Solid-Phase Analyses:
Sulfite Oxidation, %
Limestone Utilization, X
556
6.1
7.2
91
556
6.1
6.9
94
543
7.6
10
91
543
7.6
10.3
94
564
7.0
8.3
84
564
7.0
7.7
91
591
6.0
9.4
79
591
6.0
9.3
87
525
5.3
7.5
84
525
5.3
7.6
85
417
5.8
11.1
79
417
5.8
11.1
82
Liquid-Phase Analyses:
Gypsum Relative Saturation 0.2 0.2 0.4 0.4 0.2 0.2 0.3 0.3 0.3 0.3 0.9 0.8
EDTA Cone., ppm (est.) 10-15 10-15 25 25 25-30 25-30 10-15 10-15 20-25 20-25 15-20 15-20
"Absorber A had EDTA spiked directly into its upper loop reaction tank, while Absorbers B and D had EDTA entering only from the reclaim water system.
AU = Upper loop reaction tank for Absorbers A and C.
AL = Lower loop reaction tank for Absorber A.
-------
Table 3
RESULTS FROM PREVIOUS OPERATION OF THE PIRKEY POWER PLANT FGD SYSTEM
Date: x
Sample Location":
9/19/90
AU
9/19/90
AL
9/26/90
AU
9/26/90
AL
9/27/90
AU
9/27/90
AL
10/2/90
AU
10/2/90
AL
Daily Average Unit Load, MW
463
463
396
398
439
439
469
469
Lignite S Content, lb SOj/lO^
Btu
na
na
5.0
5.0
5.4
5.4
4.7
4.7
Solid-Phase Analyses:
Sulfite Oxidation, X
27.5
24.5
28.1
28.3
20.8
23.2
19.3
19.7
Limestone Utilization, %
74
69
78
85
79
85
73
82
Liquid-Phase Analyses:
Gypsum Relative Saturation
na
na
na
na
na
na
na
na
EDTA Concentration, ppm
0
0
0
0
0
0
0
0
na - Value not available.
aAU = Upper loop reaction tank
for Absorbers A
and C.
AL = Lower loop reaction tank for Absorber A.
*• Gas Out
Gas
Bypass
Absorber \
(1 Of 4)
xmmvmvw;
Limestone
Grinding
Mist
Eliminators
TrapoutTray
Upper
Loop
Perforated Plate Tray
Trapout Tray
Upper Loop
Slurry
Reaction Tank
dot 2)
Overflow
Lower
Loop
Gas
Lower Loop
Slurry Reaction
Tank
Reclaim Water
Fly Ash
Waste Slurry
Sludge/Ash
*¦ Mixture
To Mine
Jhickenej
Pug Mill
Vacuum Filter
Figure 1. Simplified Flow Diagram for Pirkey FGD System
5A-77
-------
Intentionally Blank Page
5A-78
A
-------
OPTIMIZING THE OPERATIONS IN THE FLUE GAS DESULPHURIZATION PLANTS
OF THE LIGNITE POWER PLANT NEURATH, UNIT D AND E
AND
IMPROVED CONTROL CONCEPTS FOR THIRD GENERATION ADVANCED FGD DESIGN
H. Scherer
R. Widzgowski
Noell-KRC Umwelttechnik GmbH
8700 Wurzburg
Federal Republic of Germany
G. Weiss
Noell, Inc.
2411 Dulles Corner Park, Suite 410
Herndon, VA 22071
Preceding page blank
5A-79
-------
Intentionally Blank Page
5A-80
-------
ABSTRACT
The two-stage FGD process which was developed in the seventies in the United States was brought to
Europe in the early eighties. An advanced Third Generation Design has been developed based on the
experience gained on U.S. FGD systems with a total of 7,800 MW. There are thirteen (13) operating
FGD plants using these improvements in reliability and SO2 removal efficiencies; 10,000 MW are in
various stages of operation, construction, or design with this process in Europe.
The Noell-KRC FGD Systems of Unit D and E (2 x 600 MW) consist of four interconnected Absorber
Towers (each 62 feet in diameter) and two Absorber Feed Tanks (one AFT for two absorber towers).
Powdered limestone is used as reagent and gypsum is the FGD by-product. Since the Power Station
burns lignite from the Rhein district, the FGD System was designed for varying S02 inlet
concentrations, 500 to 5,200 mg/Nm3 dry (170 to 1,800 ppm dry).
The paper covers system modifications and improvements of the instrumentation and control philosophy
made at the RWE Lignite Power Plant Neurath, guaranteeing high availability, and high S02 removal
rates, while maintaining the quality of the FGD endproduct gypsum rigorously. Modifications were
made in the flue gas inlet duct, the level and pH control of the quencher loop, the waste water
discharge, the bowl flushing system, the mist eliminator washing control system, and the limestone
feeding control system. Also optimized and standardized process instrumentation related to integrated
control for boiler and FGD system have been installed.
Preceding page blank
k.
5A-81
-------
INTRODUCTION
This paper will cover some of the aspects of optimizing an advanced FGD design for the Rheinisch
Westfalische Elektrizitatswerke (RWE) Neurath Station, using the Noell-KRC double-loop concept
originally pioneered in the USA by Research-Cottrell.
We will also cover some of the improvements which have been used in the control and instrumentation
of these advanced FGD plants.
The RWE Lignite Power Plant Neurath consists of five power plant units, 3 x 300 MW (Units A-C)
and 2 x 600 MW (Units D and E).
The boiler installations of Units D and E were put into operation June, 1975 and February, 1976.
They were planned for a wide variety of coal. The permissible degree of heat for this lignite ranges
between 6,280 and 10,720 kJ/kg (or 2,700 to 4,600 BTU/lb).
The FGD retrofit of Units D and E have to date a total of over 24,500 hours of operation. The design
of these plants was done for varying SOj concentrations in the flue gas, from 500 to 5,200 mg/rrr* in
standard condition/dry. The equivalent SOj loading to the absorbers is approximately from 170 to
1,800 ppm dry.
The guaranteed level of SOj removal was 95%, and constantly exceeded, while the qualities of the
endproduct were maintained rigorously. The plant availability has been an of average 99.8% with two
years of the operation at 100%.
The construction started in October 1985; Unit D went into operation in October 1987, and Unit E in
January 1988.
The FGD system uses limestone as reagent and gypsum as endproduct and has the following interesting
characteristics:
Two absorber towers are connected to a common absorber feed tank (Figure 1). Two induced draft
blowers take suction from the flue gas ducts and discharge to the absorbers. The outlet flue gas is
discharged to the natural-draft cooling towers.
5A-82
-------
There are four 67% absorber towers for the 2 x 600 MW units, therefore, in an upset condition three
towers are sufficient for the operation. The towers are 62 feet in diameter (Figure 2).
The flue gas of each unit can also be discharged through the new bypass chimney.
The gypsum discharge takes place through hydroclones.
The gypsum dewatering is performed through band filters centrally located for all units on this site in
a separate dewatering building.
The FGD is designed to produce gypsum in wallboard quality, but at the present time there is no
market for wallboard gypsum, therefore, the dewatered gypsum is, after an interim storage, disposed
as stabilizer, mixed with fly ashes from the power plant (dry and wet ashes), into old coal mines.
The waste water removed from the FGD process is used to moisten the dry ash. No additional waste
water is accumulated.
DESCRIPTION OF THE DESULPHURIZATION PROCESS
The FGD system in each of the double-loop Units D and E has a quencher and absorber loop (Figure
3). This process was originally developed by Research-Cottrell in the USA. Noell-KRC optimized
this process due to the higher requirements for availability, higher percentage of desulphurization, and
quality of the endproduct gypsum required by German legislation.
With this process, the flue gas enters the quencher flowing tangentially, the flue gas is cooled down,
limestone dissolution occurs and first-stage S02 removal takes place. The pH-values in this first loop
are controlled between pH 4 and 5, so that the oxidation of sulfites to sulfates is optimized and excess
limestone from the absorber loop is almost completely utilized.
To guarantee complete oxidation, compressed air is discharged in front of the quencher agitators and
distributed into the slurry.
The absorber loop is designed to remove the remaining sulphur dioxide. To achieve this, a pH-value
of approximately 6 is maintained with a high excess limestone concentration. At the maximum removal
design point two spray levels and one Wet Film Contact spray level are continuously being operated.
The Wet Film Contact serves as the final S02 removal stage. The spray levels are supplied with slurry
from the absorber feed tank.
The slurry flows downward through a funnel (bowl) which separates the two loops, and then into the
absorber feed tank. In this tank all the basic reactions between the removed S02 and the limestone
take place. Part of the absorber slurry is discharged through overflow into the quencher. The quantity
of overflow depends on the amount of water used for washing the mist eliminators.
OPTIMIZING THE FGD OPERATION
The design has been improved to prevent deposits and scaling in various areas of the system.
At the beginning of the operation, minor deposits were noted in the flue gas inlet duct to the quencher
5A-83
-------
(Figure 4). Although using a nickel base alloy, some corrosion did occur due to locally concentrated
chloride under the gypsum layers in the wet-dry interface zone.
Experiments with synthetic coatings to decrease wet-dry zone corrosion proved unsuccessful. This
problem was solved through material reinforcement. At this time there is no indication of any further
corrosion, and therefore no further action is contemplated.
Six agitators are arranged around the quencher (Figure 5). Some minor deposits which develop despite
the agitators are considered minimal and do not affect the performance of the process.
During startup, deposits in the flue gas diversion area around the bowl were noticed. This issue had
to be solved since it was critical from an availability viewpoint. Also, if not solved, absorber elements
and the rubber lining could have been damaged if large deposits were broken off. This scale was
mostly gypsum. The formation of gypsum in these areas is explained below.
Acid quencher droplets, carried into the upper absorber area by the flue gas, settle and mix with SC>2
and alkalized absorber slurry. During this process gypsum is rapidly formed.
This gypsum scale was eliminated with the following three steps:
1. The pH-value in the quencher loop was increased to its highest allowable level to slow
down gypsum formation in the upper absorber while maintaining gypsum quality.
2. The bowl flushing system was optimized by replacing the 24 90°-full-cone nozzles around
the bowl with 120°-hollow-cone nozzles (Figure 6 and 7).
3. Periodically water is used instead of absorber slurry to flush the bowl. Therefore, the
cycle intervals of the mist eliminator washing had to be reduced in order to maintain the
water balance.
Despite the reduction of water used for mist eliminator washing, the outlet ducts of the FGD system
remain clean and free of any deposits or scaling. This is due to the good performance of the mist
eliminators. At the upper level of the mist eliminators, there are slight deposits which do not influence
performance. These deposits are so slight that they are easily cleaned by maintenance personnel using
hoses during regular outages (Figure 8 and 9).
The Wet Film Contactor (WFC) is made from polypropylene material. Originally, six layers were
installed, each with a height of 100 mm (4 inches). However, minor scaling occurred which, although
not reducing the gas velocity nor increasing the pressure drop, required cleaning during inspections.
Four layers were removed and then minor deposits occurred only in the outer perimeter area. Tests
performed after these modifications showed no decrease in the removal efficiency performance.
Additional further process optimization dealt with the pH values in the quencher loop. In Units D and
E, each absorber feed tank is connected to two absorbers. Both quenchers must be supplied from the
single absorber feed tank with about the same flowrates of overflow. Otherwise, the pH-values become
different in the two quencher loops.
At the start of operations, it was attempted to provide uniform slurry pH by installing another
overflow. This simplified procedure did not overcome the problem. The problem was solved by
installing internal guides directly in front of the absorber feed tank overflows to the quenchers. Initial
5A-84
A
-------
experience with FGD Unit D indicated that this procedure is sufficient for maintaining equalized pH-
values. FGD Unit E was then also equipped with these internal guides.
Other alternatives which have been considered for additional modifications are:
• Limestone powder feeding directly into the quenchers;
• Cross connecting the two quenchers which results in further mixing of the slurry.
• Eliminating natural overflows by regulating flowrate from the absorber feed tanks to both
quenchers, depending on the liquid level and the pH-values.
The waste water discharge system was also modified so that the only liquid blowdown consists of the
moisture in the ash and the remaining moisture in the gypsum. Material selection allows a chloride
concentration of 30,000 ppm in the quencher. The waste water discharged has to be designed so that
this level is not exceeded. Prior to the modifications, the waste water quantity became unacceptable
when the cloth rinse water began to dilute the vacuum belt filter filtrate.
After modifications, the required chloride levels were obtained using two methods, as follows:
• discharging quencher hydrocyclone overflow directly into the ash moisturizing system;
• discharging vacuum belt filter filtrate separately from the cloth rinse water, and recycling
the cloth rinse water to the process.
Both methods were successful and have been implemented.
IMPROVEMENTS IN CONTROL AND INSTRUMENTATION FOR THE THIRD GENERAL FDG
DESIGN
Because of the stringent requirements of German legislation, FGD plants such as RWE Neurath, as
well as other Noell FGD plants such as Bayernwerke Schwandorf, Units B, C, and D (also lignite
fired) were required to install improved instrumentation and control systems.
These modifications caused improvements in the following main areas: removal efficiency, water
balance, availability and reliability, gypsum quality control, and limestone utilization. To achieve these
objectives, better instrumentations and high levels of standardization were developed. An integrated
control system for boiler, turbine, FGD and DeNOx systems, using the same operating personnel, was
used.
As a result, the following important improvements have been made:
• Improved operating and control system for demister washing
• Modified control of limestone feedrate regulating system
• Optimized and standardized process instrumentation systems
• Integrated controls for the boiler and FGD systems
5A-85
-------
• Provided fully-automatic options, with a single-button control to operate the entire FGD
system
• Flow modulating valves were avoided to the largest extent possible
Previous designs of the demister wash system were based on variable wash time intervals. In addition,
open-loop control used boiler load as an input signal for a modulated timer. The improved design uses
fixed wash periods with adjustable off times. A closed-loop control with flue gas enthalpy used as a
feed-forward signal and quencher level as feedback is now used (Figure 10 and 11).
The advantage of this type of control is that any disturbance in the FGD process system is immediately
detected and corrected. The washing impulse signal is of a guaranteed minimum duration. This
assures the best possible cleaning of the demister and the maintenance of the water balance over the
entire load range (Figure 12 and 13).
With respect to controlling of the limestone feedrate, earlier designs used an open-circuit system with
manual adjustment when required. The improved design utilizes closed-loop control and as mentioned
above, boiler load for a fast-forward signal and quencher pH as a feedback.
For operation with high sulphur coal similar to lignite, or high chloride concentration, a dual limestone
feed is used. With this type of system, any disturbance in the quencher cycle can be safely corrected.
In practice, the system works on the basis of a batch-metered supply of pH-modulated limestone
feeding directly to the quencher.
The primary conditions for the above-mentioned improvements were the development of
instrumentation and control having a greater degree of reliability and availability. We have used better
quality and standardized pressure gauges for the liquid slurry loops (Figure 14 and 15).
Previous designs used for measuring the slurry concentration in the quencher was either an ultrasonic
device or a capacitive probe housed in an extension tank. The improved method uses a AP level sensor
with flushed separating diaphragms.
Improved pH measurement devices have been developed using special-type glass electrodes
(Borosilicate glass, diaphragm thickness, etc.). We have developed an improved pH measuring vessel
which is self-cleaning, non-clogging, and easy to maintain.
The use of density meters with an americium source clamped on the main slurry pipes has been
changed to in-line density meters with the americium source on a slipstream application. This
improvement has resulted in higher accuracy and enables calibration while the FGD system is in
operation.
With the evolution of these advanced designs, we developed a proprietary "quality station" with a
sophisticated setup to measure the pH, the slurry density, and other parameters.
Many of the advanced third generation plants have selected the fully-automatic single-button control
option. This allows hot startup and shutdown to be initiated via a single command. The various
functional groups are not changed, however, they are controlled via an additional control program.
The advantages of an integrated control room are obvious. They reduce the operating requirements
and the number of staff required in the control room. This concept assures minimal inadvertent control
5A-86
A
-------
errors and avoids difficulties during startup and shutdown of the system (Figure 16).
In many of our advanced FGD plants, we designed and installed (or cooperated in the application of)
a Distributed Control System (DCS), and/or PLC based systems. These can generally be retrofitted
into the existing boiler control room.
SUMMARY
The wet FGD Units D and E at Neurath have now been in operation for 24,500 hours and have
rigorously complied with requirements regarding removal efficiency, endproduct (gypsum) quality, and
availability.
To improve operation of these FGD units, the following procedures were successfully introduced:
1. Installation and optimizing a washing system to prevent deposits in the bowl ring area.
2. Elimination of several layers of Wet Film Contact without any noticeable decrease in SC>2
removal efficiency.
3. Installation of internal guide vanes in the absorber in front of the overflows to prevent
uneven pH-values in the quencher recirculation loops.
4. Deducing the waste water discharge and maintaining the chloride concentration within
required limits, using direct hydroclone overflow discharge to the ash system, recycle of
cloth rinse water, and separate discharge of vacuum belt filter filtrate.
5. Improving the instrumentation and control systems in the areas of demister washing,
limestone feeding, and instrumentation for measuring pressure, pH, and slurry density.
5A-87
-------
mmmm
iMfaMHf
Figure 1. Neurath Double Loop Process Concept
~ ^,
Ml
Figure 2. Detailed Neurath Absorber Section
5A-88
A
-------
1UJ 9
11
Figure 3. Double-Loop Absorber Schematic
1. Absorber Tower
2. Bowl
3. Wet Film Contactor (W.F.C.)
4. Mist Eliminators
5. Flue Gas Inlet Duct
6. Flue Gas Outlet Duct
7. Quencher Pump
8. Absorber Feed Tank (A.F.T.)
9. Absorber Pumps
10. Limestone Day Silo
11. Oxidation Air Compressor
5A-89
-------
Intentionally Blank Page
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
-------
Figure 4. Flue Gas Inlet Duct
(Dry-Wet Interface Zone)
Figure 5. Neurath Quencher Details
(Agitators, Oxidation, Suction Screens)
5A-90
-------
Intentionally Blank Page
-------
Figure 6. Bowl Flushing System Nozzles (Before Optimization)
Figure 7. Bowl Flushing System Nozzles (After Optimization)
5A-91
-------
. y j_)_lciiijv ray^
-------
Figure 8. Mist Eliminator Details (Before Optimization)
Figure 9. Mist Eliminator Details (After Optimization)
5A-92
-------
Intentionally Blank Page
I
I
I
I
I
I
I
I
I
I
I
I
-------
volume
measurement
flue gas
pH-quencher
slurry
flue gas volume
Iscfm)
-0-
SO, concentration
"ippml
pH-quencher
'pH
variables:
stoichiometric
factor, purity
calculation
required
limestone
llb/min)
density
limestone Af)
slurry ^-'density
u-
Limestone
slurry tank
dosine control
logic (time
proportion
control)
pH
absorber
slurry
valve control
signals
©
L-
Absorber
feed tank
Note: Inputs for automatic start sequence
an{j manual actions are not shown
Figure 10. Limestone Control Logic Diagram
¦j3 M X-
QUENCHER LEVEL
MAX
WIN -
MM
till
BREAK TIME
X: DUTY TIME CONSTANT
e.g. 120 SEC.
-FL R_
'• f*i : m
MIST ELIMINATOR
LEVEL
TEMP. IN
TEMP OUT
NEED OF
MAKE-UP
WATER
EVAPORATION
CAPACITY
DENSITY
UPPER LOOP
—o-
ETC.
Max.
O—
QUENCHER LEVEL
M1N.
Figure 11. Control Logic For Limestone Feed
5A-93
-------
A 1 START SIGNAL
IT/, RESET SIGNAL
W FOR INTEGRATOR
T INTEGRATION
~ REQUIRED
LIMESTONE
^D) SLURRY VALVE
W OPEN
TIMEsCONSTANT
V E ; INTEGATION
LIMESTONE
FLOW
T: TRIGGER SIGNAL
~ FOR STORE
LIMESTONE FLOW
INTO MEMORY
INTERVAL t,
Qa —
Qa '
REQUIRMENT: ACTUAL LIMESTONE FLOW HAS TO
BE CREATED THAN THE REQUIRED
LIMESTONE FLOW
Figure 12. Schematic Limestone Feed
First stage
Uist Valve V2
Eliminators
Second stage
Wst Valve V2
Eliminators
1
1
i a.
Lover level
Third flush cycle
Total flush cycle
1
i
1
Flush cvcle
Upper level
Total cycle: three cycles in lower level
and one in upper level
Example: First cycle-average load (evaporation)
Second cycle: full load
Third cycle: minimum load
Tf: flush time
Tb: break time between flushing
(calculated by computer program)
Figure 13. Schematic Mist Eliminator Wash Cycle
5A-94
-------
U1
>
I
VO
U1
MOUNTING HEIGHT
5-6' ABOVE GRADING
E3 52
EQUIPMENT UST
5 ! i
- I O
DESCRIPTION
J TAP WITH flange RUBBER'JNED
2 ! 1
EXTENDED Diaphragm -jvnge
3 ! '
[ GASKET
4 >
PRESSURE GAUGE
! BOLT WITH -"EXNUT
HZE
MATERIAL ! MANUFACTURE
OR I OR
type I MODEL
r/15D#
I A105/RUBBER'JNED
150#
I
STAINLESS steel
r/300#
31655
3/4-- 2 •,/:'! STAINLESS steel
I
NOTES:
1. MAINTENANCE CLEARANCE 2'
2. DIAPHRAGM TYP. '162/6/33 HST, 316L
3. RANGE 0-90LBS. ECTEFE COATED
a. FILLED WITH SILICON OIL. OPERATING TEMP 110°"
5. DESIGN oRESSURE 'SOLSS
LADISH
990
WIKA
233.3a
Figure 14. Pressure Gauges in Slurry Applications
-------
U1
*
ON
1) LEVEL MEASURMENT- ABSORBER TDWER/INSTALLATIDN SURVEY g) TRANSMITTER ' M * — FUNCTIONS
2ERD-ATJUSTHEMT
1
CALI 1
BRA 1
TIDN _ I
CtWTR
B
-IX-
A
INDICATION
PURGING WATER
1000
Figure 15. Level Instrumentation - Absorber Tower
-------
CONTROL LOGIC STRUCTURE
NOELL-KRC
START
AUTOMATIC
h!
STOP
MODE
H
MAIN PROGRAM-FGD
(SEQUENCE CONTROL LOGIC)
gii
X W I
DAMPER
FLUE GAS
ABSORBER
RECYCLE
PUMPS
i SO,LOAD
BOWL
FLUSHING
LEVEL
ABSORBER
DENSITY
I.D. FAN
CONTROL
PUMP
PUMP
; SELECTION
START/
! SELECTION
HIGH
EFFICIENCY
HYDRO
CYCLONES
WAST
DEMISTER
SLURRY
FLUSHING
FLUE GAS
FLOW
FLUE GAS
TEMPERATURE"
CYCLONE
CONTROL
ABSORBER LEVEL
(QUENCHER)
WASTE WATER
TREATMENT
I OXIDATION
| AIR BLOWER
! SELECTION
PUMP
! SELECTION
OXIDATION
AIR
SYSTEM
LIMESTONE
SLURRY
PUMPS
SO,LOAD
ABSORBER
START/ PUMP
STOP SELECTION
MAKE
AGITATORS
WATER
DENSITY
LIMESTONE !
SLURRY
I LEVEL
I LIMESTONE
j SLURRY TANK
LIMESTONE
PREPARATION
TO RECLAIM
"WATER
NOTE:
EACH PUMP. BLOWER AND
AGITATOR CAN BE STARTED/
STOPED MANUALLY ALSO.
Figure 16. Control Logic Structure
-------
Intentionally Blank Page
5A-98
-------
ORGANIC ACID BUFFER TESTING AT
MICHIGAN SOUTH CENTRAL POWER AGENCY'S ENDICOTT STATION
B. J. Jankura
The Babcock & Wilcox Company — Research and Development Division
Alliance, Ohio 44601
M. G. Milobowski and R. U. Hallstrom
The Babcock & Wilcox Company — Environment Equipment Division
Barberton, Ohio 44203
J. P. Novak
Michigan South Central Power Agency
Litchfield, Michigan 49252
Preceding page blank
5A-99
-------
Intentionally Blank Page
5 A-100
-------
ABSTRACT
As the electric utility industry prepares to meet the challenges of the 1990 Clean
Air Act Amendment, one area of considerable interest is the use of organic acid
buffer additives for limestone wet scrubbing. Prior to 1990, organic acid buffers
have been used as a wet scrubber retrofit to achieve performance on existing wet
scrubbers not meeting performance. Some current flue gas desulfurization (FGD)
system specifications are now dictating their inclusion into initial system de-
signs.
Babcock & Wilcox (B&W) has conducted organic acid buffer testing at Michigan South
Central Power Agency's (MSCPA's) 55-MWe James R. Endicott power station in 1991 to
broaden the company's database on FGD additives. This station has a single ab-
sorber, and has a wet scrubber system designed for 90% sulfur dioxide (SO2) re-
moval. The wet scrubber has the B&W in situ forced oxidation process to produce
gypsum, which is currently being utilized in the cement and wallboard industries.
Results are presented from the Endicott Station test program to identify relevant
design criteria for formic and dibasic acid buffers. These results indicate that
at 500 ppm organic acid buffer concentrations, the scrubber's SO2 removal effi-
ciency was significantly improved.
Preceding page blank
5A-101
-------
INTRODUCTION
Passage of the 1990 Clean Air Act Amendment, with its "credit" system for better-
than-legislated sulfur dioxide (S02) removal, has some utility power producers
considering supplementing their base flue gas desulfurization (FGD) system specifi-
cations with requirements for added S02 removal via organic acid buffer addition.
While operating experience with organic acid buffers does exist in the United
States on naturally oxidized FGD systems, little operating experience has been
accrued on acid addition on in situ forced oxidized systems. In 1991, Babcock &
Wilcox (B&W) undertook a test program at a 55-MWe utility power plant to evaluate
the effect of formic acid and dibasic acid buffers on a fully oxidized unit's FGD
performance.
Testing was performed at Michigan South Central Power Agency's (MSCPA's) James R.
Endicott Station. This plant is designed to burn a high-sulfur Ohio bituminous
coal having S02 levels of 7.2 Xb/MMBtu. Actually, FGD system inlet S02 levels of
9.0 Xb/MMBtu are not uncommon. Commercial quality gypsum is routinely produced
following the retrofit installation of a B&W-designed in situ oxidation system that
has resulted in oxidation levels greater than 99%.
The scope of the additive test program at MSCPA's Endicott Station was to establish
the wet FGD system baseline conditions without additives for 90% S02 removal and
99%+ oxidation, and then determine the effect of organic acid buffer addition in
various concentrations (100 through 1500 ppm) on FGD system performance and gypsum
properties. Two organic buffers, formic acid and dibasic acid, were evaluated in
separate test series.
BACKGROUND
The MSCPA Endicott Station's limestone reagent wet scrubber module was retrofitted
with a B&W-designed in situ oxidation system in early 1990 (Feeney, 1990;
Hallstrom, 1991). Tests of organic acid additives at MSCPA was the final phase of
a program to demonstrate the design limits of B&W's new in situ oxidation system.
Formic acid, obtained as a 90% concentration, was the first buffer scheduled to be
tested. The advantage of formic acid is its low cost and ready availability. Di-
basic acid buffer (DBA) was also tested. To date, DBA has been the most frequently
used buffer in utility FGD limestone wet scrubbers.
5A-102
A
-------
PREVIOUS UTILITY DBA EXPERIENCE
Several utility wet FGD systems use DBA on a routine basis (Mobley, 1986) or have
conducted tests to evaluate its use to achieve high levels of SO2 removal (90%+) to
meet SO2 emissions requirements. At the Plains-Escalante Generating Station
Unit 1, DBA was investigated as one method to improve SO2 removal (Hendry, 1990;
Hendry, 1991). Up to 900 ppm DBA provided only marginal improvements in SO2 re-
moval in a natural oxidation system. The system was later upgraded to meet S02
emissions requirements by increasing the liquid-to-gas ratio, improving flue gas
distribution and spray coverage, and reducing spray droplet size. At the Associ-
ated Electric Cooperative Thomas Hill Unit 3 FGD system (Burke, 1990), DBA has been
used routinely in combination with sodium formate to achieve compliance S02 remov-
als. Typical DBA concentrations are 5,000 ppm. This system does not use forced
oxidation. The San Miguel Electric Cooperative, Inc. routinely uses DBA in a natu-
ral oxidation system (Cmiel, 1986) at approximately 600 - 900 ppm to meet S02 emis-
sions requirements while allowing for partial flue gas bypass for reheat purposes.
The Indianapolis Power and Light Co. Petersburg Unit 3 system is also a natural
oxidation system (Guetig, 1985) and has observed significant improvement in S02
removal at approximately 500 - 1000 ppm DBA.
PREVIOUS UTILITY FORMIC ACID EXPERIENCE
There are no utility systems in the United States operating with formic acid. Sev-
eral power stations in Germany (Glasmer, 1989; Schutz, 1986) use formic acid along
with forced oxidation. Reported formic acid buffer concentrations are approxi-
mately 500 - 2000 ppm.
CHEMISTRY
The ability of DBA and formic acid buffers to enhance SO2 removal can be described
as a buffering mechanism with reactions occurring in both the absorber spray zone
and the reaction tank. The theoretical effects of organic acid buffers on mass
transfer in wet FGD systems has been described by Rochelle (1977). Acceptable
organic acids will have a pKa* between the pH of the gas-liquid interface (3-4)
and the recirculating slurry (4.5 - 5.5) (Chang, 1982). The concentrated organic
acid is added to the reaction tank and dissociates in the relatively alkaline
* pKa Log10 [H+] [AD"] / [HAD]
5A-103
-------
slurry (Eq. 1) . In the absorber, gas phase SO2 is absorbed (Eq. 2) at the gas-
liquid interface. As the pH at the gas-liquid interface drops, adipate ions formed
in the reaction tank diffuse from the bulk liquid, react (Eq. 3) with hydrogen ions
to form adipic acid, and suppress the normal drop in pH at the gas-liquid interface
across the absorber. The increased average pH at the buffered gas-liquid interface
improves the overall SO2 absorption per unit volume of recirculating slurry.
Dissociation H-AD-H + 20H- => AD- + 2H20 (1)
(Acidic)
Absorption 2S02 + 2H20 => 2H+ + 2HS03- (2)
Buffer Effect AD- + 2H+ => H-AD-H (3)
ORGANIC ACID BUFFER LOSS MECHANISMS
The loss mechanisms for organic acid buffers have been previously described (Ruiz-
Alsop, 1988) and occur through: 1) chemical degradation in the absorber and reac-
tion tank, 2) organic acid buffer removal with liquid in the waste solids and
liquid blowdown, and 3) coprecipitation with the waste solids. As these three
methods are dependent on site-specific operating conditions, the average organic
acid buffer consumption rate is best determined by on-site testing. In forced
oxidation systems, investigators have measured essentially no coprecipitation in
the waste solids (Jarvis, 1982; Jarvis, 1986). For these systems, the major loss
mechanisms are either chemical breakdown or organic acid buffer removed with liquid
in the waste solids and liquid blowdown. However, forced oxidation may increase
chemical degradation due to volatilization and increased chemical oxidation in the
reaction tank. Chang (1986) reported approximately 9% higher degradation and
coprecipitation losses with forced oxidation.
FACILITY
The 55-MWe Endicott station, which started up in 1982, has been operated as a cy-
clic unit, with the load dropping off to about 30 MW at night and to about 20 MW on
weekends. In summer, the weekend load increases to 40 MW to accommodate the
greater demand caused by air conditioning. Limestone reagent is prepared in a
tower mill, and fed to the absorber as a 25% solids slurry. The oxidized spent
slurry is processed through a thickener and vacuum drum filter. The Endicott sta-
tion avoids chloride build-up by operating with continuous blowdown. Typically,
chlorides range from 1200 to 1800 ppm in the oxidized slurry.
5A-104
A
-------
This is a single absorber tower system, unlike many existing operating plants which
have a spare module. This meant it was essential that nothing done during testing
adversely affected operations. At no time could the compliance of the station (90%
SO2 removal average per 24 hours) be compromised while the tests were in progress.
A typical fuel analysis, on an as-received basis, is:
Carbon = 60.36% Ash = 11.99%
Hydrogen = 4.11% Moisture = 11.42%
Nitrogen = 1.28% Oxygen = 7.36%
Sulfur = 3.48% Chlorine = 0.11%
HHV = 10,904 Btu/lb
While this FGD unit was designed for 4.3% sulfur coal, actual coal sulfur content
ranges from 3.5% to 5.0%. To meet the required S02 removal levels when firing the
higher sulfur coal, the boiler load is sometimes reduced.
The absorber control pH is typically from 5.4 to 5.6. The calcium stoichiometry
varies from 1.04 to 1.24. A typical limestone analysis is:
Calcium as CaO = 49.3%
Magnesium as MgO = 3.3%
Carbonate as C02 = 43.4%
One of the advantages of testing at the Endicott station is that it is a full-
scale, operating power plant. No scale-up factors are needed to apply the test
results to other full-scale units, although the inlet SO2 and other site-specific
conditions must be taken into account in determining the required additive dosage.
Oxidation Method
Prior to installation of the in situ oxidation system, the unit was originally
designed to operate with semi in situ oxidation, in which some of the scrubber
slurry was diverted by a scoop at a point below the module's gas distribution tray
to an external oxidation tank. On exiting the oxidation tank, all or part of the
oxidized slurry could be returned to the scrubber, or go directly to the thickener.
This method of forced oxidation, shown in Figure 1, gave less than optimum results.
This method is successful at the City of Grand Haven plant, which operates with
lime. In 1990, B&W retrofitted the MSCPA Endicott Station with an in situ forced
5A-105
-------
oxidation system (Figure 2). Among the absorber modifications necessary for this
retrofit were removal of the scoop and installation of sparge pipes for oxidation
air introduction. Post retrofit, Endicott has been oxidizing to 99%+ calcium sul-
fate at the desired 90%+ SO2 removal with vacuum filter cake solids increased from
60% to 85%. Instead of being landfilled, the by-product gypsum is being used as
either a wallboard or concrete additive.
FORMIC ACID CHARACTERISTICS
The 90% formic acid solution used in the Endicott test was purchased from Ashland
Chemical. One of the advantages of formic acid is its low freezing point, allowing
it to be shipped, stored, and used in a higher concentration than other organic
acids such as DBA. The boiling range for the 90% formic acid solution is 221° -
223°F (105* - 106°C). The freezing point is 23°F (~5°C), which would require
freeze protection for storage and injection equipment in some areas in winter. One
disadvantage of formic acid is that, especially in higher concentrations such as a
98% solution, it decomposes slowly to form carbon monoxide and water. Formic acid
is also a combustible liquid, with a flash point of 122° - 136'F (50° - 58°C) as a
90% solution.
Use of any organic acid requires greater safety awareness by plant personnel. For-
mic acid is considered extremely corrosive and may cause skin and eye irritation.
Rubber gloves and boots, goggles, and an oversuit should be worn when handling
formic acid. Adequate ventilation is required because inhalation of vapor or mist
may cause fatal respiratory irritations. Formic acid is harmful if swallowed and
it may be fatal.
DBA CHARACTERISTICS
Buffering agents increase SO2 removal efficiency in limestone scrubbers by limiting
the pH drop at the gas-liquid interface during SO2 absorption. Other potential
benefits are reduced scaling, improved reliability, improved reagent utilization
and reduced sludge volume, and increased generating capacity when previously lim-
ited by scrubber performance. In addition, some plants may be able to increase
flue gas bypass, while others may reduce electric power consumption by reducing the
absorber spray flux.
The DBA used for Michigan South Central testing was supplied by E. I. Dupont.
DuPont DBA is processed to eliminate most of the nitric acid and metal contami-
5A-106
-------
nants. A typical analysis of DuPorvt DBA on a dry basis is as follows:
Glutaric Acid
Succinic Acid
Adipic Acid
52%
Nitric Acid
0.2%
22%
24%
Copper
Vanadium
0.01%
0.02%
Organic N2 Compounds = 1%
Dibasic acid buffer is a co-product in the production of adipic acid, which is
primarily used in the manufacture of nylon. In addition to FGD applications, DBA
is also used to purify other by-product compounds produced in the adipic acid manu-
facturing process, or concentrated and incinerated for energy recovery (Chi, 1989).
Although DBA is normally shipped in tank trucks as a water solution in concentra-
tions of 50% to 94% by weight, its solubility in water is temperature-dependent.
Table 1 and Figure 3 show the solubility of DBA in water at various temperatures.
Specific gravity at 212°F (100°C) is 1.04 for a 50% DBA/water solution, and 1.13
for an 80% DBA/water solution.
Direct contact with DBA causes irritation of eyes and skin. DBA exhibits a low
level of oral toxicity. If DBA is heated above 266°F (130°C), nitrogen oxides may
be released, which may be fatal if inhaled in high concentrations.
One point of interest between the two acids tested, aside from the characteristics
affecting safety and handling, is the relationship between equivalents and molecu-
lar weight. One equivalent of an acid is the quantity that will supply or donate
one mole of protons (H+) . It follows that one equivalent of an acid neutralizes
one equivalent of base. Formic acid has a molecular weight of 46, and has one
equivalent per mole. Dibasic acid buffer has an average molecular weight of 129
and two equivalents per mole. A cursory review might infer that dibasic acid
should be twice as effective as formic acid. However, some 27% of pure DBA is made
of succinic acid, which is inactive as a buffering agent in naturally oxidized
systems because it co-precipitates immediately. Adipic and glutaric acids also may
degrade to succinic acid.
TEST PROGRAM
Objectives of the additive test program at MSCPA's Endicott Station were to:
Establish FGD system baseline conditions for simultaneously achiev-
ing 90% S02 removal, 95% minimum limestone utilization, and 99%+
oxidation
5A-107
-------
• Determine the effects of organic acid addition on FGD system per-
formance and gypsum properties
The test program was approved by MSCPA prior to the start of testing. Provided B&W
did not disrupt normal plant operations, such as maintaining 90% SO2 removal on a
daily basis, B&W test personnel were permitted to make changes to wet FGD scrubber
operation as needed. B&W personnel interacted directly with MSCPA scrubber operat-
ing personnel during testing. Testing was staffed by B&W around the clock. This
involved two crews of three people, each working 12-hour shifts. Test crews con-
sisted of a lead engineer, engineer, and technician.
Table 2 lists the controlled and uncontrolled wet scrubber significant operating
parameters. It should be noted that the most significant uncontrolled operating
parameter was inlet S02 concentration. Although the utility's operating personnel
did their best to use coal from the same shipment, inlet S02 concentrations varied
widely. For example, S02 inlet values ranged from 6.97 to 8.55 lb/MMBtu during DBA
testing, but were fairly steady at 5.40 lb/MMBtu during the abbreviated formic acid
test.
For organic acid buffer testing, it was determined that the most cost-effective
approach would be to initially "spike" the absorber reaction tank to the desired
concentration, then run the acid injection pump regularly thereafter to maintain
setpoint concentration.
During acid addition test planning, it was anticipated (and realized during test-
ing) that at a certain slurry acid concentration, S02 removal would be maximized at
baseline pH and liquid-to-gas ratio. During such cases, it was determined that
testing be continued to broaden the database by increasing slurry acid concentra-
tion to the next higher target setpoint, and decreasing reaction tank pH until S02
removal returned to the previous data set's maximum value.
Formic acid testing at MSCPA began on May 16, 1991. Based on the published data
for both DBA and formic acid testing at naturally oxidized wet FGD units, the ob-
jective was to collect performance data at 500, 1000, and 1500 ppm formic acid.
The tests included one day for baseline data gathering, and six days operation at
three different levels of formic acid addition. In actuality, data for two differ-
ent formic acid concentrations in the reaction tank, both below 500 ppm, were ob-
tained. It should be noted that a target concentration range, rather than
setpoint, was expected since continuous monitoring of the organic acid buffer con-
5A-108
-------
centration was not possible. The organic acid concentration would be determined by
on-site laboratory testing of reaction tank slurry grab sanples rather than by
continuous monitoring.
Prior to the planned resunption of formic acid testing at MSCPA, E. I. DuPont of-
fered to supply DBA for testing on a fully oxidized FGD system. The continuation
of the formic acid testing was tentatively postponed and DBA testing began on
August 19, 1991. Because the acid storage/feed equipment at MSCPA is not heat
traced, and since DBA solubility in water is highly temperature dependent, the DBA
was supplied as a 20% by weight concentration in water. This resulted in ai low-end
temperature use limit of 64'F (18'C).
DBA testing was scheduled round-the-clock for five days, with testing planned at
absorber reaction tank target DBA concentrations of 100, 250, 500, and 1000 ppm.
Acid metering pump capacity limitations resulted in attaining only a 732 ppm DBA
concentration. However, 500 ppm DBA was more than sufficient to achieve signifi-
cant increases in S02 removal efficiency at a constant pH. Greater than 500 ppm
acid addition permitted the use of lower reaction tank pH to attain high S02 re-
moval values.
The organic acid buffer acid tests were run with the unit's boiler operating in its
typical cycling mode — full output (49 - 55 MW) during the day and reduced output
(28 - 35 MW) at night. Previous testing confirmed that 99%+ oxidation could be
maintained with air stoichiometrics below 2.5 moles O/moles S03. For the DBA
tests, it was planned to operate at 2.5 air stoichiometry. MSCPA's oxidation air
blower is a constant output machine, which can provide an air stoichiometry of 4.1
during reduced boiler load. The air stoichiometry is controlled by the manual
operation of an air bleed-off valve in the blower discharge piping upstream of the
reaction tank sparger.
All test data were obtained as either panel board data or chemical analysis on
liquid sanples. Data analysis was performed during the test program to determine
what adjustments were required to operating parameters during the remaining tests.
SAMPLING AND ANALYSES
Flue gas sanples from the absorber inlet and outlet were continuously monitored for
S02 and 02 using Dupont and Teledyne analyzers, respectively. The analyzers were
calibrated before and after each test. Limestone, limestone slurry, recycle water,
5A-109
-------
vacuum drum cake, and coal samples were obtained daily. Appropriate samples were
analyzed based upon the outcome of the absorber performance tests. For each test,
absorber recirculation slurry and filtrate samples were -obtained and analyzed for
oxidation, limestone utilization, and organic acid buffer concentrations. Selected
sairples were sent to B&W s Alliance Research Center's Chemistry Laboratory for
analysis. Wallboard-grade gypsum testing was performed by Westroc Industries Ltd.,
Ontario, Canada. Selected samples of vacuum drum solids and absorber recirculation
filtrate were analyzed by Dupont for DBA and its constituents.
Formic acid and DBA concentrations were determined for reaction tank and recycle
water samples by EPRI Test Method SI (EPRI CS-3612). This method is a buffer ca-
pacity titration which is typically applied to FGD systems using buffering agents
to improve SO2 removal. This titration method determines the unique buffer capac-
ity for the FGD system being tested. It can then be used to estimate how much the
buffer capacity will increase with a known organic acid addition. The amount of
organic acid buffer required to achieve a certain buffer capacity is specific to
each FGD system.
RESULTS
Formic Acid Buffer
On May 16, 1991, a 7-day test of formic acid as an FGD additive was begun. Table 3
contains a summary of selected operating conditions for these tests. The formic
acid test plan involved one day for baseline operation and six days operation at
three levels of formic acid addition, while maintaining the minimum required air
stoichiometry for 99% oxidation. Each concentration was to be tested for 48 hours.
The pH would be controlled to maintain 90% S02 removal. The baseline tests indi-
cated S02 removal was approximately 90% at an absorber pH, limestone utilization,
and inlet S02 of 5.7, 96%, and 2550 ppm, respectively.
Formic acid injection began at 1:14 PM on May 16, 1991. Within 15 minutes, the
reaction tank formic acid concentration was 250 ppm and there was a noticeable drop
in the absorber outlet S02 concentration (Figure 4). Within 45 minutes, the ab-
sorber outlet S02 concentration had dropped from approximately 220 ppm to 50 ppm.
For the next 15 hours, formic acid was periodically injected to maintain 500 ppm.
At 6:00 AM on May 17, 1991, the injection pump inlet check valve failed. As a
result, only a few data points were obtained as the formic acid concentration
dropped. However, the marked improvement at low acid levels was significant.
5A-110
A
-------
Dibasic Acid Buffer
On August 19, 1991, a 5-day test of DBA as an FGD additive was initiated. The
DBA test plan involved one day for baseline, and four days operation at four levels
of DBA addition, while maintaining the minimum required air stoichiometry for 99%
oxidation. The plan involved a step change increase in DBA acid concentrations
starting with 100 ppm, then 250 ppm, 500 ppm, and 1000 ppm DBA. Boiler load during
testing was a cycling operation, with daytime highs of 49 MW to 52 MW. Oxidation
was 99%+ throughout testing.
Table 3 contains a summary of selected operating conditions for these tests. The
baseline tests indicated SO2 removal was approximately 92% at an absorber pH of
5.2, 97% limestone utilization, and 3000 ppm inlet S02. DBA injection began on
August 20 with 100 ppm DBA. The DBA buildup was much slower compared to the formic
acid concentration and is consistent with the fact that the formic buffer concen-
trate contained significantly greater acid equivalents for each gallon injected.
On August 21, the acid concentration was increased to 250 ppm; the 500 ppm DBA test
began on August 22. That evening at 10:00 PM, testing at 1000 ppm DBA was at-
tenuated. However, the maximum acid buffer concentration attained with continuous
acid buffer injection was approximately 700 ppm. This was consistent with the
relatively open loop water balance operation (high purge rate) of the system and
the minimal increase in the recycle water buffer concentration.
Throughout the test period, acceptable waste product oxidation of 99% or better was
maintained. During high load periods, the air stoichiometrics ranged from approxi-
mately 2.2 to 2.6 moles O/moles S absorbed. During low load periods, the air stoi-
chiometry ranged from 2.6 to 4.1. With frequent changes in boiler load, inlet S02,
and ever increasing S02 removal, precise control to the target 2.5 air stoichiom-
etry was difficult. At times, the maximum air stoichiometry possible was only
approximately 2.2 while maintaining fully oxidized conditions.
CHEMICAL AND PHYSICAL ANALYSES
Selected results of chemical analyses of the absorber recirculation slurry are
presented in Table 3. The absorber slurry total solids content varied from 14.7%
to 19.6%. The organic acid buffer concentrations were measured during each test to
adjust injection rates. Waste solid and absorber liquor samples from the DBA tests
were analyzed for the presence of glutaric, succinic, and adipic acid by
5A-111
-------
E. I. Dupont (Table 4). The total concentrations of buffer for the selected
samples compared well to those measured by the field buffer capacity method.
Gypsum samples were obtained during the DBA tests for scanning electron microscope
(SEM) (Figures 5 and 6) analyses and chemical evaluations to determine acceptabil-
ity as a cement additive, and for wallboard manufacture.
DISCUSSION
Formic Acid Buffer
While several factors such as inlet SO2 varied between the baseline system perfor-
mance and the formic acid tests, the short-term test results indicated that low
concentrations of formic acid can significantly improve S02 removal (Figure 7). For
example, the inlet S02 varied from approximately 2500 ppm to 2900 ppm during the
baseline (no formic acid) tests and was well above the levels of the formic acid
tests (2100 ppm) . This reduction in S02 would contribute to greater S02 removal.
Considering these variations, the system S02 removal increased from 87% to 97% with
approximately 400 ppm formic acid.
Dibasic Acid Buffer
The short-term test results with DBA confirmed that low concentrations can signifi-
cantly improve S02 removal (Figure 8). While test conditions were relatively
steady during the baseline tests, the inlet S02 concentration increased from ap-
proximately 3000 ppm to 3400 ppm between the baseline and 100 ppm DBA tests.
At 95 ppm DBA, no improvement in S02 removal was observed. However, this is con-
sistent with operation at a higher inlet S02 concentration and lower absorber
recirculation pH. At approximately 250 ppm DBA, S02 removal improved from 92%
(baseline) to 95%. The absorber pH (5.1) was slightly lower than at the baseline
conditions (5.2). When the DBA concentration was increased to the 500 ppm level,
the S02 removal continued to improve (98%). At these conditions, the outlet S02
concentration was less than 100 ppm.
Testing at the highest planned DBA concentration (1000 ppm) was not conducted for
two reasons: First, it would have been difficult to measure the lower absorber
outlet S02 concentrations with the existing S02 analyzers. Secondly, due to the
high loss rates, the maximum slurry acid buffer levels achievable were approxi-
5A-112
-------
mately 700 ppm. The last test condition instead addressed lowering the absorber
recirculation pH at approximately 500 ppm DBA to obtain 90% S02 removal (Figure 8).
The absorber pH was lowered by 0.1 increments from approximately 5.2 to reduce the
system S02 removal. After reaching a pH of 4.8 without any significant reduction
in performance, the pH was lowered by 0.2 increments until reaching 4.0. At this
pH, plant personnel expressed concern that there could be localized zones in the
absorber where acid corrosion may be unacceptably high. To minimize these con-
cerns, the absorber pH was raised to 4.2 for the last test condition. Here, the
S02 removal remained as high as 95.7% with 512 ppm DBA. Operation at lower pH
would be expected to improve limestone utilization.
CHEMICAL AMD PHYSICAL ANALYSES
During the May 1991 tests, limited data were obtained on the loss rate of formic
acid (Figure 9). The formic acid concentration was periodically monitored in the
absorber recirculation slurry to adjust the injection interval. As this was the
first use of the injection system, several hours of operation were necessary to
understand the systems response to periodic formic acid spiking. Soon after stabi-
lizing the formic acid concentration at 500 ppm in the absorber recirculation
slurry, the injection pump failed. Within approximately six hours of stopping
injection, the formic acid concentration fell from 580 to 280 ppm. In a forced
oxidation system, the significant loss mechanisms are assumed to be chemical oxida-
tion and liquid blowdown losses. Due to the short duration of these tests, there
was no formic acid in the recycle water. Therefore, if the losses were solely due
to system purge, then the measured loss rate indicates that the reaction tank resi-
dence time is approximately 14 hours. This agrees well with the 15-hour reaction
tank residence time estimated from the known solids purge rate.
DBA concentrations in the absorber and recycle water were measured throughout the
August 1991 DBA tests and for one week after conpleting the tests (Figure 10). The
recycle water concentration increased very slowly and, as expected, lagged behind
the absorber. The maximum concentration measured in the recycle water was approxi-
mately 200 ppm. These concentrations were monitored after conpletion of the test
program. Sufficient concentrations remained throughout the week to enhance SO2
removal.
Selected samples of the dewatered gypsum were also analyzed for glutaric, succinic,
and adipic species (Table 4). The solids analyses indicate that there was essen-
tially no co-precipitation. This was anticipated for the August 20 waste solids as
5A-113
-------
the recycle water buffer concentration was unchanged from the baseline tests. The
August 22 recycle water buffer concentration was approximately 200 ppm. However,
no co-precipitation was detected. Similar results have been reported for adipic
acid (Jarvis, 1982), and for DBA (Jarvis, 1986). It is believed that there is
essentially no co-precipitation in forced oxidation systems where the major pre-
cipitating species is gypsum. For these systems, the major loss mechanisms are
thought to be chemical oxidation and liquid losses.
Scanning electron microscope (SEM) analyses of the gypsum produced during the DBA
tests (Figures 5 and 6) show large, rhombic shaped crystals that provide for easy
dewatering. The presence of DBA does not appear to alter the crystal size or
shape.
The results of gypsum testing were not available at the time of this paper's publi-
cation. Holnam Incorporated, a cement producer, has reported no deleterious ef-
fects in the processing stages or in the final product when using the by-product
gypsum of an organic acid enhanced forced oxidation system.
CONCLUSIONS
Organic acid additive testing has been successfully performed on a 55-MWe fully
oxidized wet limestone reagent FGD scrubber equipped with a patented Babcock &
Wilcox perforated tray. From the test results, it is concluded that:
• With 500 ppm reaction tank acid concentrations of either formic or
dibasic acid buffers, the baseline S02 removal efficiency can be
increased from approximately 90% to 95%.
• Gypsum resulting from this process has been found to be acceptable
for use in cement production.
• The use of organic acid buffer should allow MSCPA to meet compli-
ance during periods when the coal's sulfur content is above 4.3%.
While test results of the effects of acid addition on wallboard-grade gypsum prop-
erties are not yet available, it is believed that the low acid concentrations will
have no detrimental effect on required gypsum quality. The economics of formic
versus dibasic acid buffer use have not been covered in this paper, but should be
evaluated on a site-specific basis when considering organic acid buffering of an
FGD system.
5A-114
-------
ACKNOWLEDGMENTS
B&W thanks Mr. John Novak, Plant Superintendent; Mr. Gary Van Vlerah, Plant chem-
ist; and the plant personnel at the Michigan South Central Power Agency James R.
Endicott power station for on-site support and for providing B&W extensive use of
their chemistry laboratory facilities.
B&W thanks Mr. Alan Jeffs, DuPont's Senior Account Manager for Organic Acid Addi-
tives, for supplying twenty 55-gallon drums of 20% solution DBA and for technical
support analyzing process samples for DBA components.
We acknowledge the support of Westroc Industries, Ltd., Ontario, Canada, for pro-
viding testing services on gypsum samples for wallboard qualities.
We also acknowledge the support of Holnam Incorporated for their feedback on acid
additive effects on FGD gypsum used for cement processing.
REFERENCES
Burke, J.M., Price, T.J., and Moser, R.E., 1990, "Results of Sodium Formate Addi-
tion Tests at EPRI's High Sulfur Test Center and Associated Electric Cooperative's
Thomas Hill Unit 3 FGD System," 11th EPA/EPRI Symposium on Flue Gas Desulfuriza-
tion.
Chang, C-S, and Rochelle, G.T., 1982, "Effect of Organic Acid Additives on S02
Absorption into Ca0/CaC03 Slurries," AIChE Journal, Vol. 28, No. 2.
Chang, C-S, and Brna, T.G., 1986, "Enhancement of Wet Limestone Flue Gas Desulfu-
rization by Organic Acid/Salt Additives," 10th EPA/EPRI Symposium on Flue Gas Des-
ulfurization.
Chi, C.T., and Lester, J.H., 1989 "Utilization of Adipic Acid Byproducts for Energy
Recovery and Enhancement of Flue Gas Desulfurization," Environmental Progress, Vol.
8, No. 4.
Cmiel, R., and Seeman, D., 1986, "Three Years of Organic Acid Use at San Miguel,"
10th EPA/EPRI Symposium on Flue Gas Desulfurization.
FGD Chemistry and Analytical Methods Handbook, 1988, EPRI Report CS-3612, Vol. 2,
Nov., 1988
Feeney, S., Downs, W., and Novak, J., 1990, "In Situ Forced Oxidation Retrofit at
Michigan South Central Power Agency's Endicott Station," Power Generation Confer-
ence, Orlando, Florida.
5A-115
-------
Guetig, D., Ou, S., and Wedig, C.P., 1985, "Dibasic Acid Test and Chemical Process
Evaluation at Petersburg Unit 3 FGD System," 9th EPA/EPRI Symposium on Flue Gas
Desulfurization.
Hallstrom, R.U., and Johnson, D.W., 1991, "From Acid Rain to Wallboard: Reducing
Secondary Pollution in Flue Gas Desulfurization," AIChE Pollution Prevention Con-
ference .
Hendry, D.W., and Baumgardner, D., 1990, "FGD Performance Improvements at Plains-
Escalante Generating Station," 11th EPA/EPRI Symposium on Flue Gas Desulfurization.
Hendry, D.W., and Weis, J.G., 1991, "Scrubber Upgrades Achieves 95% Removal Effi-
ciency, " Power Engineering.
Hicks, N.D., and Fraley, D.M., 1982, "Commercial Application Experience With Or-
ganic Acid Addition at City Utilities of Springfield," 7th EPA/EPRI Symposium on
Flue Gas Desulfurization.
Jarvis, J.B., Terry, J.C., Schubert, S.A., and Utley, D.L., 1982, "Effect of Trace
Metals and Sulfite Oxidation on Adipic Acid Degradation," EPA Report 68-02-3171.
Jarvis, J.B., and Owens, D.R., 1986, "Conparison of the Effectiveness of FGD Addi-
tives for SO2 Removal Enhancement and Additive Consumption," 10th EPA/EPRI Sympo-
sium on Flue Gas Desulfurization.
Mobley, J.D., Cassidy, M., and Dickerman, J., 1986, "Organic Acids Can Enhance Wet
Limestone Flue Gas Scrubbing," Power.
Ruiz-Alsop, R., and Rochelle, G.T., 1988, "Co-precipitation of Organic Acids With
Calcium Sulfite Solids," Ind. Eng. Chem., Res., Vol 27, pp. 2123-2126.
Rochelle, G.T., and King, C.J., 1977, "The Effect of Additives on Mass Transfer in
CaC03 or CaO Slurry Scrubbing of SO2 From Waste Gases," Ind. Eng. Chem., Fundamen-
tals, Vol 16, No. 1.
5A-116
-------
Table 1
SOLUBILITY OF DBA IN HATER
Concentration Temperature
(wt/wt%)
*F
*C
20
64
18
50
93
34
80
181
83
Tabla 2
SYSTEM PROCESS FACTORS
Parameter Controlled Uncontrolled
Absorber
S02 Removal Eff.
X
Inlet S02 Cone.
X
Sieve Tray
X
Spray Nozzle Type
X
Spray Nozzle Pres.
X
Liquid Flux
X
Gas Flow Rate
X
Inlet Gas Temperature
X
Rue Gas Oxygen Cone.
X
Fly Ash Concentration
X
Absorber Tank
PH
X
Sulfite Oxidation
X
Limestone Stoichiometry
X
Residence Time
X
Sparger Depth
X
Sparger Design
X
Air Stoichiometry
X
Mixer Speed
X
Air Pressure
X
Slurry Calcium Cone.
X
Slurry Chlorides Cone.
X
Slurry Magnesium Cone.
X
Solids Concentration
X
Reagent Preparation
Limestone Grind
X
Solids Concentration
X
5A-117
-------
Table 3
SUMMARY OF SELECTED PERFORMANCE RESULTS
Boiler
Organic
Inlet
Inlet
S02
Data
Load
Solids
Acid
Util
S02
02
Removal
Air
Oxidation
Date
Time Set No
MW
%
pH ppm
%
ppmv
%
%
Stoich
%
Formic Acid
Baseline
16-May
1100
1
51.9
15.9
5.79
0
97.05
2930
5.7
86.57
2.56
99.95
16-May
1130
2
52.2
15.82
5.79
0
97.63
2864
5.7
85.34
2.56
99.94
16-May
1200
3
52.2
15.37
5.71
0
95.97
2550
6.2
88.55
2.71
99.93
Formic Acid Addition
17-May
1100
4
49.9
14.72
5.21
373
96.57
2150
6.6
97.67
3.10
99.96
17-May
1400
5
49
19.59
5.1
282
97.76
2075
6.6
96.51
3.39
99.95
Dibasic Acid
Baseline
19-Aug
1545
6
50.06
17.65
5.15
0
96.34
3057
6.06
91.29
2.32
99.98
19-Aug
1730
7
45.4
17.76
5.18
0
97.28
2862
5.88
91.72
2.52
99.98
19-Aug
1930
8
45.28
17.64
5.22
0
97.37
3172
6.26
92.71
2.34
99.96
Dibasic Acid Addition
20-Aug
1245
9
45.96
15.46
5
95
95.33
3387
6.06
92.57
2.37
99.87
21-Aug
1230
10
48.24
16.17
5.11
261
96.81
3337
5.96
96.88
2.29
99.93
21-Aug
1345
11
47.44
15.15
5 02
254
98.04
3312
6.17
95.25
2.37
99.94
21-Aug
1500
12
45.3
15.20
4.96
255
97.94
3217
6.35
95.60
2.44
99.96
21-Aug
1715
13
49.62
15.30
5
209
95.24
3316
6.50
94.66
2.17
99.87
22-Aug
845
14
49.78
16.54
4.94
487
96.06
3120
6.02
97.24
2.25
99.95
22-Aug
1045
15
50.28
15.77
4.87
458
99.56
3181
5.97
98.12
2.18
99.94
22-Aug
1230
16
50.88
16.25
5.16
468
97.18
3106
6.07
98.25
2.23
99.96
22-Aug
2150
17
48
16.03
4.04
696
97.85
3088
6.27
95.11
2.52
99.96
23-Aug
830
18
48.28
16.98
4.19
512
99.70
3072
5.62
95.67
2.55
99.93
Table 4
DIBASIC ACID COMPONENTS ANALYSES IN FILTRATES AND WASTE SOLIDS
Succinic
Glutaric
Adipic
Total
Sample Description
Time
(PPm)
(ppm)
(ppm)
(ppm)
Absorber Liquor
8/20/91
ND
<60
20
<80
Absorber Liquor
8/21/91
37
130
57
224
Absorber Liquor
8/22/91
50
354
149
553
Filter Solids
8/20/91
ND
ND
ND
Filter Solids
8/22/91
ND
ND
ND
5A-118
-------
HYDROCYCLONE
GAS
SCOOP
TO
VACUUM
FILTER
SURGE
TANK
THICKENER
ABSORBER
TOWER
OXIDATION
TANK
AIR
Figure 1. MSCPA's Endicott Station Before In Situ Forced Oxidation
WASTE
WATER
TREATMENT
THICKENER
STACK
TO BYPASS
(OPTIONAL)
~
FLUE •
GAS FROM —
PRECIPITATOR
AND I.D. FAN
SCRUBBER
MODULE
- REAGENT
K3
VACUUM T
FILTER GYPSUM
AMBIENT
AIR
BOOSTER
FAN (OPTIONAL)
Figure 2. MSCPA's Endicott Station After In Situ Forced Oxidation Retrofit
5A-119
-------
100
80
O
. 60
LU
oc
<
oc
LU
Q.
5 40
20
TEMPERATURE AT WHICH FIRST
CRYSTAL OCCURS UPON COOLING
20 40 60 80
WEIGHT PERCENT DBA IN WATER
200
180
160
LU
140 OC
<
OC
120 £
S
UJ
100
80
60
100
Figure 3. Approximate Solubility of DBA in Water
02,%
0 5 10
1800
ABSORBER
OUTLET S02
(0-1000 ppm)
ABSORBER
OUTLET 02
(0-10%)
1700
x 1600
ADD ACID
1500
1400
1000
750
500
S02l ppm
250
Figure 4. Effect of Formic Acid Buffer on Absorber Outlet S02 Concentration
5A-120
-------
100|iIT1
10 (im
Figure 5. SEM Analyses of Absorber Recirculation Slurry
(99.96% Oxidation; Organic Acid = 0)
5A-121
-------
75X
100|im
400X 10(im
Figure 6. SEM Analyses of Absorber Recirculation Slurry
(99.91% Oxidation; Organic Acid = 515 DBA)
5A-122
-------
100
2075
2150
a 90
>
o
E
£
2550
2930
80
Labels = Inlet SO,in ppm
70
100 200 300 400
Formic Acid Concentration (ppm)
Figure 7. Effect of Formic Acid on SO, Removal
100
[4.96
5.00
4.19
4.04
5.00
&18l
i.15
90
80
Line = time sequence
Data labels = absorber recirculation pH
Unit design =90%
70
0
200
400
600
DBA Concentration (ppm)
Figure 8. Effect of DBA on S02 Removal
5A-123
-------
700
600
500
400
Ppm
HCOOH
w 300
200
100
0
Figure 9. Formic Acid Addition and Decay
* *
/*
/
\ No formic addition
' K /
: /V/
—i—i—i—i—i—1—i—i—
4 1 1 1 1
Thursday Friday
Relative Day
1000
800
No DBA addition
600
ppm
400
200
Absorber
liquor B
Clarified recycle
water
0
8-19 8-20 8-21 8-22 8-23 8-24 8-25 8-26 8-27 8-28
Day
Figure 10. DBA Concentration in the Absorber
5A-124
-------
Stack Gas Geaning Optimization Via German Retrofit
Wet FGD Operating Experience
5A-125
I
-------
Intentionally Blank Page
5 A-126
-------
W. Ellison
Ellison Consultants
4966 Tall Oaks Drive
Monrovia, Maryland 21770
H. Weiler
STEAG AG
Huyssenallee 86-88
4300 Essen 1, Germany
ABSTRACT
Extensive electric utility retrofit FGD system design and operating experience in
German high-quality and low-grade bituminous coal service is detailed to assess
initially unforeseen difficulties, required corrective measures, know-how gained
and technological advancements accomplished. This outgrowth of the massive, 1983-
1990 acid rain control construction program in western Germany is further analyzed
to indicate principal areas of relevance and applicability in typical U.S. wet FGD
applications in the 1990s with high-sulfur coal. A further focus is the
coordination/integration of the FGD installation and the advantageous
design/operation of SCR facilities positioned in the cold (tail) end mode
downstream of the S02 removal step. Additional sub-topics include a dissertation
on new field findings on extent of collection of flue-gas trace/heavy metals in
wet scrubber equipped plants; and review of new technological understanding of the
functioning and process relevance of ab-/desorption of flue-gas acids on the
regenerative heating surfaces of the rotary flue gas reheater.
INTRODUCTION
Retrofit application of high-efficiency S02 and NOx removal equipment in Germany
bears significantly on control and management of emissions of diverse secondary
pollutants including trace heavy metals, S03/H2S04 and fluoride emission forms.
Preceding page blank
5A-127
-------
The sophistication of control system components including rotary regenerative
gas/gas heat exchangers commonly applied in Europe has been found to introduce
related problems and complications in system operations, potentially requiring
corrective modifications as outlined in this presentation.
HEAVY METAL EMISSION REDUCTION
Investigations concerning the removal of heavy metals have been carried out for
two types of bituminous-coal fired, electric utility furnaces, slag tap (wet
bottom) and dry bottom, each served by wet systems for control of S02 emissions.
Wet bottom boilers are known to have inherently higher trace metal emissions and
are of special concern. Specifically addressed were the volatile metallic
elements and their efficiency of removal in the diverse flue gas cleaning steps.
These field results may have direct applicability in the U.S. in conjunction with
future air toxics emission control requirements of the new Clean Air Act under
Title III.
Dry Bottom Boiler Service
The fate of trace metals entering the boiler was determined to be as per Table 1
for a 750 mWe dry bottom installation. This unit fires high-grade, low-sulfur
bituminous coal with 6.4% ash and a lower heating value of 28.5 MJ/kg (12,260
Btu/lb), and is served by a prescrubber-loop-equipped, lime/gypsum FGD system with
greater than 95% S02 removal efficiency. Unlike wet bottom boilers, dry bottom
boilers do not have facility or specific purpose for recirculating fly ash to the
boiler furnace. The electrostatic precipitator (ESP) positioned upstream of this
FGD system operates with 99.9% particulate removal efficiency, reducing the
particulate loading at the FGD system inlet to 3 mg/Nm3 (0.0013 grain/DSCF).
Wet Bottom Boiler Service
Trace metal distribution was as per Table 2 for a 350 mWe wet bottom (slag tap)
installation. This unit fires low-grade, medium-sulfur bituminous coal with 29.1%
ash and a lower heating value of 20.6 MJ/kg (8,860 Btu/lb), and is served by a
single-loop, lime/gypsum FGD system with greater than 95% S02 removal efficiency.
85% of the ESP fly ash catch is recirculated to the boiler furnace. The ESP
operates with 99.76% particulate removal efficiency, reducing the particulate
loading at the FGD system inlet to 100 mg/Nm3 (0.044 grain/DSCF). Other
distribution data shows that non-volatile elements such as copper, nickel,
chromium, manganese etc., displayed behavior similar to zinc. Further, with the
high fly ash recirculation rate of 85%, semi-volatile elements such as lead,
5A-128
-------
cadmium and antimony that are volatile at high temperature become enriched in the
fly ash. This is typified in the above table by the element arsenic. The
efficiency of removal of volatile elements such as mercury and selenium, ranging
between 50 and 80%, is not as high as that for the acid gas pollutants. It was
confirmed that mercury is largely in the gaseous elementary state because in the
sampling activity it could only be collected by use of impingers containing
oxidizing liquid, i.e. only approximately 1% of mercury present was captured using
micro-fabric dust filters in the sampling train. On the other hand selenium was
measured in approximately the same amount losing either impingers or micro-fabric
dust filters. Therefore it may be assumed that the element selenium exists in the
flue gas not as a gas but largely as an aerosol. Additional removal of these two
volatile elements would require the installation of an activated carbon adsorber
system downstream of the FGD facility. An alternative remedy is hypothetical use
of an FGD prescrubber system operated in an acid/oxidizing mode.
TRAMP EMISSIONS FROM FLUE GAS REHEATING SYSTEMS IN FGD INSTALLATIONS
To improve the dispersion of the scrubbed flue gas issuing from the stack emission
control systems, regulations require a minimum flue gas temperature of 72°C
(162°F) at the stack outlet. Inline reheating of water saturated flue gas thereby
requires a minimum temperature increment of approximately 25°C (45°F) to achieve
the 72°C stack exit temperature. This gas temperature boost is equivalent to
approximately 1.3* of boiler heat input at a power station. For cost economy
approximately 80* of the wet FGD installations in Germany are equipped with
regenerative gas/gas reheaters, mostly of the rotary Ljungstrom type. This
reheater system, similar in design to the commonly used, combustion air preheater
of this type, has an inherent rate of internal leakage of from 3 to 5* of the
total gas flow. If the flue gas booster fan is positioned on the generally
preferable, raw gas side, the leakage passes from the raw gas to the clean gas
side and gas cleaning is compromised. This by-passing of the scrubber via leakage
in the rotary regenerative flue gas reheater is effectively eliminated by
installing the flue gas booster fan between scrubber and reheater, i.e. on the
clean gas side. Thereby, alternative, leakage-minimizing provisions in the
reheater are not needed. Recent experience has shown, however, that less obvious
mechanisms exist for transfer of pollutants from the dirty side to the clean side.
Even in absence of use of rotary regenerative FGD reheaters as applied in Germany,
these findings can have important meaning to operators of uniquely high sulfur
U.S. boilers that may be retrofitted with high-dust SCR.
5A-129
-------
H2SQ4 Emission
An adsorption-desorption phenomenon in the flue gas reheater operation was found
to create a sulfuric-acid-mist emission opacity problem. Historically, discharge
of this stack gas pollutant is not an unfamiliar circumstance since power stations
using high sulfur oil for start up and shut down operation are known to
intermittently emit such acid particulates to the atmosphere. And, after the
complete installation of FGD and DeNOx (SCR) systems, and irrespective of
positioning of the flue gas booster fan on the clean gas side of the reheater,
many power plants experienced continuous acid emission.
At the same time the effect was limited to power stations firing high grade (low
ash content) coal and equipped with high-dust SCR system installations, and that
experienced frequent boiler load swings and weekend shutdowns. This problem was
also associated with plant operation in which the operating temperature of the
catalyst was at times up to 30°C (54°F) higher than the original design value.
This increment of temperature increase resulted from normal boiler fouling, which
brought about a drastic increase of the S02 to SO3 conversion rate in the high-
dust SCR reactor, and led to a flue gas SO3 concentration of up to 100 mg/Nm^ (29
pprov) •
Passing through the combustion air preheater the flue gas is cooled down to 120-
140°C (248-284°F). This means that the temperature of the heating plate surface
on the cold side of the rotary air preheater as it passes from the combustion air
sector to the flue gas sector is below the sulfuric acid dew point and that the
acid gas forms thin acid films on the heating plate surfaces or generates
entrained aerosol droplets. In this way gaseous S03 forms condensed sulfuric acid
that may be completely or partially neutralized by interaction with alkaline
components of fly ash. In this way flue gas SO3 is removed upstream of the ESP in
an amount that depends on the flue gas temperature at the cold end of the air
preheater. If, as is the case at the flue gas reheater downstream of the ESP
facility, the amount of fly ash is inadequate, or if the fly ash contains only
small or negligible amounts of alkali, the sulfuric acid condensate is only
partially neutralized.
In the rotary reheater the heating plates are contacted alternately by hot raw
flue gas containing S03 and then clean cold flue gas saturated with water vapor.
Accordingly the acid film on the heating plates surface is transferred from the
raw gas side to the clean gas side by the rotative movement of the plate assembly.
Depending on boiler load, varying amounts of acid condensate aire deposited on the
5A-130
-------
heating plates of the reheater, which in turn, varies the amount of acid droplets
entrained in the stack gas.
To avoid the emission of these acid particulates diverse measures as follows can
be taken, the first two of which have been applied commercially in Germany:
• Continuous cleaning of the FGD outlet ducts and periodic washing
of the regenerative reheater to remove the deposited acid from
the heating plates
• Allowing an increase in the dust content downstream of the ESP
to help neutralize and bind the acid. To keep the gas pressure
drop across the reheater at an acceptable level when using this
remedy, additional soot blowing is necessary.
• Decreasing the flue gas outlet temperature at the air preheater
forming a greater amount of acid condensate, which is neutralized
through the presence of high fly ash loading.
• Injection of additional alkali in the furnace or upstream of the
air preheater to prevent the formation of the acid condensate film.
HF Emission
The German emission control regulations require the following emission limits for
halogens (fluoride, chloride) at power stations with a thermal input > 300 MJ/s
(1,024 million Btu/Hr):
O
• Hydrogen chloride: 100 mg/Nm or 63 ppm (dry)
• Hydrogen fluoride: 15 mg/Nm3 or 17 ppm (dry)
It was initially demonstrated by test that this limitation could be met without
difficulty. The measurements indicated that downstream of the scrubber the
fluoride was virtually not detectable and the chloride concentration was a maximum
O
of 5 mg/Nm (3 ppm). Therefore scsne plant owners acceded to substantially more
stringent regulation of HF emission as imposed by local authorities. But,
surprisingly the acceptance tests clearly showed that the HF emission limit could
not be met, especially in partial load operation. See Figures 1 and 2. Similar
difficulty was also reported at lignite fired power facilities using a rotary
reheater.
It was known from the fluorohydrocarbon manufacturing industry that high
solubility of HF in sulfuric acid is the basis for separation of HF from other
compounds. Seme related data about acid deposition in FGD ducts are presented in
5A-131
-------
a 1983 EPA/EPRI-report (3). The tests and investigations mentioned in this report
were checked and study results provided confirmation. The fluoride concentration
in the condensate has a marked dependence on condensing temperature, and above
75°C (167°F) only small amounts of fluoride were found in the condensate. This
validated the following chemical equations and theories, exemplified by Figure 2,
describing the HF absorption process that occurs on the surface of the heating
plates with the rotation of the rotor:
• Sulfuric Acid Deposition:
^(g) + H2°(g) < > H2S04(g) + H2S04< 1)
• Hydrofluoric Acid Absorption:
HF(g) + H2S04(1) < > HS03F(1) + H20(1)
Bypassing of the reheater by HF into the clean gas can be prevented by ensuring
that the plate surface temperature does not drop below the 75°C (167°F) HF dew
point temperature, but this condition cannot be maintained on the cold side of the
rotary reheater. Moreover, basket temperature just before rotary movement into
the raw gas sector is very close to the scrubber outlet temperature, i.e. 50°C
(122°F). Thus the available means for avoiding the desorption of HF on the clean
gas side is to evaporate the HF on the raw gas side of the reheater by achieving
heating plate temperatures above the HF dew point as per Figure 1 before the clean
gas sector of the reheater is reached. This desorbed HF gas is absorbed in the
FGD system forming CaF2. Another measure that can be used is the prevention of
the liquid sulfuric acid film by injecting alkaline compounds as previously
discussed.
COMPLICATIONS EXPERIENCED WITH CATALYTIC NOX-REDUCTION SYSTEMS
Future application in US coal service of any of the modes of SCR system operation
may be broadly guided by substantial German experience, including both wet bottom
and dry bottom boiler plants
Low Dust Service
Due to limited space aid for reasons of economy, the selective catalytic reduction
type NQx removal system for STEAG's Walsum Unit 7 wet bottom (slag tap) boiler was
installed in a unique low-dust mode, positioned between a conventional low-
temperature ESP and the FGD facility. This decision was made out of concern about
the presence of arsenic in the undedusted raw flue gas and to thereby achieve
acceptable, long-term, catalyst service life. For necessary flue gas reheating up
5A-132
-------
to 320°C (608°F) a rotary gas/gas heat exchanger and a high-pressure steam
reheater are used.
2 time of 10,000 hours ammonia slip rose to more than 10 ppm in
Dn to achieve the 86% design NOx removal efficiency. Extensive
ind analysis shewed that deposition of silicon based material was
t for this catalyst surface poisoning. Measurements made at several
s shewed that this boiler type emits significant amounts of gaseous
'ily as SiF4, with concentrations up to 1 mg/Nm3 (0.22 ppm). These
unds, being of gaseous form, cannot be removed by the ESP.
:arried out in other investigations have shown that dry bottom
emit gaseous silicon compounds. In an initial temporary remedy to
: surface fouling of the catalyst, a pure, active Ti02 layer was
. of the catalyst to remove the SiF4 from the flue gas.
!§
observations appeared to confirm that, because of SiF4 removal
R system by the FGD facility, tail-end SCR installations were not
lling by silicon material. But routine checking of samples of the
ilyst elements shewed a major decrease in residual catalyst activity
aerating hours. By monitoring test elements in different layers it
that the silicon concentration at the first catalyst layer and
the catalyst surface at the inflcw section was higher than at the
n. Tests with a bench scale reactor indicated a clear, inverse
;tween level of catalyst activity and the silicon concentration on
irface. The lowest activity was found at the catalyst inlet section
: at the outlet section where the activity was nearly twice that at
Lon. Micro photos with 1,000 to 10,000 magnification showed silicon
es with a diameter of 0.1 micrometer (micron) covering the catalyst
amount of Si02 particles correlated very well with the measured
af Si02 on the catalyst surface, which decreased from 18% (inlet)
rutlet). The depositing of Si02 on the catalyst surface occurred
ous transport mechanism that is also manifested by aluminum oxide
n the fly ash. Testing of catalyst cleaning effectiveness showed
degraded sample increased in activity approximately 20* after
:ountermeasures it was important to identify the reasons for the
ing caused by silicon. It was determined that this catalyst
limited to DeNQx installations that incorporated both an FGD and a
I
5A-133
-------
DeNOx-rotary reheater. But catalyst was impaired only in those cases in which the
FGD flue-gas rotary reheater is operated with high scrubbed-gas temperature boost.
If the amount of reheating is only 30-35°C (54-63°F), or if the flue gas
temperature at the ESP outlet is below 100°C (212°F), only limited catalyst
fouling occurred. Additional measurements were made at an affected FGD/DeNOx
installation using a highly sensitive analytical method. Immediately downstream
of the scrubber no gaseous silicon was found in the flue gas. But, surprisingly,
up to 80 micrograms of gaseous silicon per Nm was measured downstream of the
rotary reheater on the clean gas side. In applying this analytical method at an
FGD/SCR system with a low FGD reheat boost, little or no SiF4 was found. This
result was in direct accord with the measured catalyst activity, which, in this
case, decreased only in keeping with normal aging of the catalyst material. If
the FGD is equipped with a tubular reheat exchanger instead of a regenerative
rotary reheater, or all reheating duty, FGD and SCR, is achieved with a single
rotary heat exchanger, no significant amount of gaseous silicon compounds is
measured in the treated flue gas downstream of the reheater.
The absorption/desorption process on the surface of the heat exchanger /plates is
still not fully understood, but it must be assumed that strong links exist among
condensation of sulfuric acid vapor, solution of hydrogen fluoride and the
presence of gaseous silicon compounds.
It has been determined, as diagramed in Figure 3, that the absorption process in
the raw and clean gas sections of the rotary heat exchanger is described by the
following equations:
• At the raw gas side (Absorption): SiF4 + (HF, H2S04) > H2SiF6
• At the clean gas side (Desorption): H2SiF4 + (H20,T,P) > SiF4 + HF
This theory has been confirmed by the results of different measurements, sis
correlation between HF content and SiF4 content of the clean gas appears extremely
strong. From among the numerous measures that can be taken to avoid or reduce HF
and SiF4 emissions, only simple methods were tested in a first step. This avoided
high cost before gaining assurance that the above theory is correct. As an
initial step, testing examined the binding of the acid condensate by the injection
of alkali into the raw gas duct upstream of the rotary reheater. By optimizing
the amount of injected alkaline sorbent and thereby avoiding pluggage of the
reheater, a test giving good results was sustained for more than 4 weeks.
• The pressure drop across the reheater increased slightly but could be
kept stable.
5A-134
-------
• SiF4 bypassing was reduced by nearly 50-70%. It was determined that
HF emission was similarly decreased.
Another test was carried out seeking to reduce the SiF4 emission by decreasing the
rotational speed of the heat-exchanger rotor. By this means, the temperature of
plates at the cold end of the heat exchanger is above 75°C (167°F); the difference
in temperature between adjoining heating plates is increased and the SiF4 can be
evaporated before it enters the clean gas section. It was found that the rate of
HF and SiF4 bypassing was reduced by 30% by decreasing the rotational speed of the
rotor by 50%.
High Dust Service
The guaranteed and the actual service life of SCR catalyst can be increased 50% or
more without problems by provision for use of surplus (excess) catalyst volume.
However, overall projection of Germany experience indicates that in application of
high dust SCR in U.S. high-sulfur coal service the installed catalyst volume and
SO2/SO3 conversion by the catalyst must be minimized to limit the SO3 problem.
Thus, maximum use of primary combustion methods to limit gross NOx concentration
will be essential. In seme German SCR high-dust applications with particulate
loading of 60,000 mg/Nm^ (26 grain/DSCF) honeycomb type catalyst has had to be
replaced by plate type catalyst to gain higher abrasion resistance and for reduced
susceptibility to fouling by gasborne fly ash.
REFERENCES
1. H. Gutberlet. "Measurement of the Heavy Metal Concentrations in Pure Gas
Downstream a Limestone FGD". Forschungsbericht ENV-492-D, Kommission der
Europaischen Gemeinschaft, Bruessel, Belgium, April 1984.
2. H. Brosch. "Heavy Metal Balance in a Limestone FGD". Internal STEAG Report,
July 1990.
3. D. Froelich, C. Weilert and P. Dyer. "Acid Deposition in FGD Ductwork".
In Proceedings of the EPA/EPRI Symposium on Flue Gas Desulfurization,
New Orleans, Louisiana, November 1983.
5A-135
-------
•c
o
o
tc
V
<
tc
£
<
tc
Ld
Q.
HF—desorption
80 -
75
o
2 70
O
o
<
UJ
X
UJ
tc
60 -
Maximum load
50 -
HF-desorption
HF dew point
temperature
(raw gas)
f.
Partial load
Raw gas sector
Clean gas sector
Raw gas sector
180 360
ROTATIONAL ANGLE (Degrees)
180
Figure 1. HF desorption in
rotary flue gas reheater.
5A-136
J
-------
HF: less than 1 mg/Nm3
Clean gas
Resorption
H2°(,) + HSC!SF<1)-H2S04<1)+ HF(g)
Clean gas
HF: 10 mg/Nm3
HF: 21 mg/Nm3
Raw gas
Absorption ^
+ HF(gr-HS03F(i)+ h2°(i)
Raw gas
HF: 30 mg/Nm3
Figure 2. Mechanism and equations
for HF—transport process.
5A-137
-------
Sify: <10 xig/Nnrf
Clean gas
SiF4: 350 >ug/Nnrf
Raw gas
H2S,F6 +
Desorption
Absorption
Water vapor
Pressure
Temperature
H_S0
H2SIF6
Clean gas
SiF4: 60 ^g/Nrri5
Raw gas
SIF4: 400 xig/Nrrf
Figure 3. Transfer mechanism for SiF4,
5A-138
-------
Table 1
DRY BOTTOM BOILER TRACE METAL DISTRIBUTION (1)
Trace
Bottom
Fly
FGD
Clean
Emission
Metal
Ash
Asn
Catch
Gas
Cone.*
Arsenic
1.8 %
98.1
%
0.1 %
0.01 %
0.042 ug/Nm3
Mercury
<0.1 %
24.6
%
25.7 %
49.6 %
10.0 ug/Nm3
Selenium
<0.1 %
33.8
%
33.6 %
32.5 %
30.0 ug/Nm3
Zinc
4.6 %
95.3
%
0.1 %
0.01 %
0.25 ug/Nm3
* Basis dry gas, 6% 02
Table 2
WET BOTTOM BOILER TRACE METAL DISTRIBUTION (2)
Trace
Metal
Slag
(Molten Ash)
Fly
Asn
FGD
Catch
Clean
Gas
Emission
Cone.*
Arsenic
5.6 %
85.3
%
8.5 %
0.6 %
5.3
ug/Nm3
Mercury
<0.1 %
3.6
%
74.5 %
21.2 %
46.1
ug/Nm3
Selenium
<10.0 %
5.6
%
47.5 %
36.9 %
45.0
ug/Nm3
Zinc
44.4 %
50.9
%
4.7 %
0.02 %
9.0
ug/Nm3
* Basis dry gas, 5% 02
I
5A-139
-------
Intentionally Blank Page
5A-140
-------
OPERATION OF A COMPACT FGD PLANT USING
CT-121 PROCESS
Yoshio Ogawa and Teruo Sugiya
Chiyoda Corporation
P.O. BOX 10, Tsurumi, Yokohama, Japan
Yoshisuke Kawabata
Hokuriku Electric Power Co., Inc.
Shinpo 57-1-6, Mikuni-cho, Sakai-gun, Fukui, Japan
I
Preceding page blank
5A-141
-------
Intentionally Blank Page
5A-142
-------
ABSTRACT
A compact system of the jet bubbling flue gas desulfurization process was built
for Hokuriku Electric Power Company in Fukui, Japan.
The system features:
(1) Elimination of the prescrubber,
(2) A heating medium circulated/multi-tube type gas to gas heat exchanger and
fully automatic operation using a distributed control system (DCS).
The following satisfactory operating results were obtained:
(1) A high desulfurization efficiency of 95% and low outlet particulate loading
of 0.4 mg/Nm3,
(2) Low power consumption of about 1% of the power generated by the electric
generation.
INTRODUCTION
The latest CT-121 installation in Japan began operation in February 1990 and
successfully passed its performance test in March 1990.
The new 350 MW facility is located in Fukui Prefecture, and is owned by the
Hokuriku Electric Power Company. As illustrated in Figures 1 and 2, flue gas is
pressurized by two booster fans, and is cooled by the gas-to-gas heat exchangers
(GGH) and after gas saturation in the transition duct, the gas is introduced into
a JBR and bubbled into the absorbent through vertical sparger pipes where SO2 is
absorbed and particulates are throughly removed. The absorbed SO2 is completely
oxidized to form sulfates by adding air at several times the stoichiometric
requirement. The cleaned flue gas is then passed, through a chevron type mist
eliminator and after reheating by the GGH, the gas is vented into the atmosphere.
Limestone slurry is pumped directly into the JBR to precipitate the sulfates as
gypsum. Stoichiometric amounts of limestone are added to maintain a pH range
between 3.5 and 4.5. Within this range, typical scrubber scaling and plugging
problems are eliminated. The gypsum slurry concentration is kept between 10 and
30 wt% by drawing off a small stream of gypsum into three continuous screw-type
decanters.
Preceding page blank
5A-143
-------
The heart of the CT-121 process is the Jet Bubbling Reactor (JBR), where the flue
gas is dispersed through crystalline slurry, using spargers as shown in Figure 2.
The depth of sparger submergence and the pH level within the gently agitated
slurry jointly determine removal efficiency. Sparger submergence creates a steady
back pressure at the booster fan discharge, accounting for 60 to 80 percent of the
system's total electric consumption. The gypsum slurry is maintained in
suspension by one vertically mounted agitator.
FEATURES OF THE PLANT
Elimination of Prescrubber
Electrostatic precipitators limit particulate loading ahead of the scrubber to
less than a 30 mg/Nm^ basis. The gas is first cooled from 145°C to 108°C in the
GGH. Then it is cooled from 108°C to about 53°C in the inlet duct of the JBR by
injecting water. The system is compact without a prescrubber as shown in Figure
3. Contact of gas with water is made in four spray zone stages. The first
precooling is conducted with make up water. The second to fourth are with gypsum
slurry to remove particulates. The slurry nozzles stay clean as they are
continually exposed to spray carryovers.
Gas to Gas Heat Exchanger
As shown in figure 4, GGH is a combination of heat extractor and reheater. Flue
gas is cooled by the multi-tube type heat extractors from 145°C to 108°C and the
gas treated by the JBR is heated from 53°C to 113°C by the reheater. Heat
exchange between JBR inlet gas and JBR outlet gas is performed via a heating
medium (water). This system is used to permit a leak less heat exchange and it
is not necessary to cross the duct outlet with the duct inlet as opposed to a
Ljungstrom type GGH.
Axial Flow Fan
The need to maintain high energy efficiency under varying load conditions led to
selection of a horizontal axial flow fan with two stages of variable-pitch
rotating blades. Because of their small blade size, these fans have a low
starting torque, which leads to a substantial reduction in the size of the motor
load controllers. Blades and housing are of carbon steel. Ammonia gas injected
into the flue gas ahead of the precipitator reduces the severity of cold side
corrosion by lowering the acid dew point.
5A-144
-------
Full Automatic Operation
Operating the FGD system should satisfy varying gas flow requirements of the
boiler. The FGD panel is adjacent to the boiler control panel. Monitoring,
operation, and control are done entirely from that location by DCS. The CT-121
has two major control loops: (1) flue gas rate control, and (2) JBR pH control.
The flow rate of flue gas is automatically adjusted by controlling the variable
rotating blade orientation on the fan rotor to respond to the differential
pressure controller of the by-pass damper. The feed rate of limestone slurry is
determined by the feed forward signal of the SO2 load and adjusted by the feedback
signal from the JBR pH. The system is monitored and controlled remotely from the
field, making use of a distributed control system. SO2 outlet concentration
response to typical load changes over a range of 100 to 350MW as shown in Figure
5. This figure, a typical strip chart taken from the FGD system control panel,
shows the ability of the system to hold a steady pH over a wide range of load
conditions.
The liquid level of the JBR is controlled to keep a certain submergence depth of
the sparger pipes and to obtain the required SO2 removal efficiency.
OPERATION
A summary of performance test results appears in Table 1. Under typical operating
conditions, the SO2 outlet concentrations registered in Figure 5 are equivalent to
95% SO2 removal efficiency. The outlet SO2 at a given load typically varies by
+/-10 ppmv and the pH holds at 4.0 +/-0.3. The system responds well
automatically, even with load changes. Total power consumption for the system,
with gypsum dewatering, was 3,900 kW under the following conditions:
Boiler at full load, MW 350
Flue gas flow, Nm^/hr 1,018,000
SO2 concentration, ppmv 1,363
SO2 removal efficiency, % 94.7
The power consumption was 1.1 percent of the power generated by the boiler/turbine
plant. The 7.2 mg/Nm^ of the inlet grain loading leaves the precipitator for the
FGD system with a higher proportion of submicron fines. The JBR removed more than
90 percent of the fine particulate.
5A-145
-------
A major reduction in operating manpower has also been realized by installing DCS.
Whereas two operators per shift were originally dedicated to the FGD system, the
job responsibilities are now covered by the boiler operator.
Gypsum from the continuous decanter is sold for Portland cement manufacturing, and
runs 98.3 to 99.2 percent CaC04*2H20, and about 90 percent solids. Table 2 shows
an analysis of the by-product gypsum.
Waste water treatment regulations in Japan require the removal of COD that is
mainly in the form of dithionate (S2O6") from the waste water produced by most
wet FGD processes. Dithionate removal is complicated, requiring ion-exchange
resin adsorption, pH adjustment, and thermal decomposition of the dithionate.
Using the CT-121 process, the COD is low as is shown in Table 3, even without
installing a dithionate removal plant.
RELIABILITY
The reliability record for this installation is shown in Table 4. Updated
reliability parameters for the other CT-121 plants in commercial operation also
appear in Table 4.
SUMMARY
The system has routinely achieved excellent SO2 and particulate removal. The DCS
was successfully applied and allows for a saving in the number of operators.
Precooling of this type instead of using a separate prescrubber was satisfactory.
The CT-121 process continues to prove effective in other areas, including:
• Operating and maintenance simplicity
• Low power consumption
• High limestone utilization of over 99 percent
• High gypsum quality
• Low effluent COD
5A-146
-------
Table 1 Performance Test Results
Item
Unit
Test Results
Flue
Gas condition
(1)
Flow Rate
N/hm3
1,018,000
(2)
Composition
vol%
N2
C02
H20
02
nh3
mg/Nm3
77.8
13.5
10.2
3.3
17.2
(3)
Temperature
°C
138
(4)
Pressure
mmH20
114
(5)
S02
ppm
Nm3/h
1,363
1,246
(6)
Particulate
mg/Nm3
7.2
Guaranteed Item
Test Results
(1)
SO2 Removal
%
94.7
(2)
Temperature at
stack Inlet
°C
114
(3)
Particulate
at FGD Outlet
mg/Nm3
0.4
(4)
SO2 at FGD
Outlet
ppm
Nm3/h
76
67.7
(5)
Gypsum Moisture
%
9.7 - 13.1
(6)
Waste Water
Flow Rate
COD
SS
m3/h
mg/1
mg/1
6.0
7.7 - 8.6
24 - 28
5A-147
-------
Table 2 Gypsum Analysis
CaO (%)
32.0 ~ 32.2
Gypsum Purity based
on CaO (%)
98.3 ~ 99.0
S03 (%)
45.9 ~ 46.2
Gypsum Purity based
on SO3 (%)
98.6 ~ 99.2
HC1 Insoluble (%)
0.22 ~ 0.34
AI2O3 + Fe203 (%)
0.14 ~ 0.24
Moisture (%)
9.7 ~ 13.1
Table 3
Waste Water Analysis
COD (mg/1)
7.4 ~ 8.6
SS (mg/1)
22 ~ 28
PH
6.2 ~ 6.8
5A-148
-------
Table 4 CT-121 Reliability Parameters
Plant Ovner
Mitsubishi Petrochemical
Co., Ltd.
Yokkaichi
Nippon Mining Co., Ltd.
Toyama Kyodo Electric Power
Shinminato, Toyama
Co.<, Ltd.
Kashima Northern Joint
Power Co., Ltd.
Kashima
Chita
UNIT 1
UNIT 2
Period
Hay 11,
December
1982 through
31, 1990
Nov. 10,
December
1983 through
31, 1990
July 9, 1984 through
December 31, 1990
Aug. 23,
December
1984 through
31, 1990
Nov. 15,
December
1985 through
31, 1990
Parameter
l
Value
Z
Total Time
hrs/hrs
Value
Z
Total Time
hrs/hrs
Value
Z
Total Time
hrs/hrs
Value
Z
Total Time
hrs/hrs
Value
Z
Total Time
hrs/hrs
Reliability3
98.8
71043/71929
99.3
57071/57488
100.0
48408/48414
100.0
47572/47572
99.1
39859/40202
Availability''
95.3
72208/75756
95.0
59475/62580
94.0
53367/56794
92.0
51248/55714
95.5
42905/44928
OperabilityC
98.8
71043/71929
99.3
57071/57488
100.0
48408/48408
100.0
47572/47572
98.9
39859/40301
Utilization Factor''
93.8
71043/75756
91.1
57071/62580
85.2
48408/56794
85.4
47572/55714
88.7
39859/44928
Plant Owner
Hokuriku
Power Co,
Kusa j ima,
Electric
., Inc.
, Toyama
Hokuriku
Power Co,
Fukui
Electric
,, Inc.
Period
July 24,
December
1987 through
31, 1990
March 26,
December
, 1990
31, 1990
Reliability
^Avaliability
- hours the FGD system was operated divided by the
hours the FGD system was called upon to operate.
- hours the FGD system was available for operation
Parameter
Value
Z
Total Time
hrs/hrs
Value
Z
Total Time
hrs/hrs
cOperability
(whether operated or not), divided by the hours
in the period.
- hours the FGD system was operated divided by the
Reliability3
Availability'5
100.0
96.7
17906/17906
29142/30150
100.0
96.7
5902/5902
6509/6732
''utilization factor
boiler operating hours in the period.
- hours that the FGD system was operated divided
by the hours in the period.
Operabilityc
100.0
17906/17906
100.0
5902/5902
Utilization Factor"'
59.4
17906/30150
87.7
5902/6732
-------
MIST ELIMINATOR
LIMESTONE
SILO
STACK
GGH
o
CENTRIFUGE
GGH
GYPSUM
FAN
MOTHER
LIQUOR
TANK
JBR
GGH
<3
LIMESTONE
SLURRY
TANK
FAN
f STM
L-ff
< WASTE WATER
OXIDATION AIR
BLOWER
FIGURE 1 CT-121 -PROCESS FLOW DIAGRAM
-------
OUTLET PLENUM AGITATOR
UPPER DECK
PILLAR
INLET DUCT
CAS RISER
INLET PLENUM
SPARGER PIPE ^
OUTLET DUCT
LOWER DECK
JET BUBBLING ZONE
REACTION ZONE
FIGURE 2 JBR CUT-AWAY SKETCH
-------
FIGURE 3 OVERVIEW OF THE CT-121 PLANT
(JBR SIZE : 19.Om0 x 12.5™H)
-------
FROM BOOSTER FANS
»
113 °C
1. 057. 000Nm/h
145 °C
ft
STEAM
i
\
")
!!)
BOOSTER
HEATER
¦Of
;,)
C!
!!)
I
HEAT EXTRACTOR
108 °C
ft
REHEATER
1,119, OOONmVh
53 °C
TO JBR
FROM JBR
FIGURE 4 GAS TO GAS HEAT EXCHANGER
5A-153
-------
2000
Boiler Load
—I ¦ I"IMI I
Boiler
Load
Elapsed Time (Hr)
FIGURE 5 CT-121 SYSTEM RESPONSE TO BOILER LOAD CHANGE
------- |