EPA-600/R-93-064C
April 1993
<&EFyV Research and
Development
PROCEEDINGS:
1991 S02 CONTROL SYMPOSIUM
Volume 3. Sessions 5B and 6
United Slates
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
-------
ORDERING INFORMATION
Requests for copies of this report should be directed to the EPRI Distribution Center. 207 Coggins Drive.
P.O. 8ox 23205. Pleasant Hill, CA 94523, (510) 934-4212. There is no charge lor reports requested by
EPRI member utilities and affiliates. *
Copyright © 1992, EPRI TR-101054. "Proceedings: 1991 SO£ Control
Symposium, Volumes 1, 2, and 3." Since this work was, in part,
funded by the XJ. S. Government, the Government is vested with a royalty-
free, non-exclusive, and irrevocable license to publish, translate, re-
produce, and deliver that information and to authorize others to do so.
EleOric Power Research Institute and EPRI are registered service marts o( Electric Power Research Institute. Inc.
NOTICE
This report was prepared by the Qecuic Power Research tnstiMi lot (EPfl). Nether EPM. members oC EPW.
nor any person acting on Mr behaft (a) males any wananty, express or inpEefl. win respect to Die use d any
information, apparatus. meOiod. «r process dedosed in this report or that such use may re* infrkige privafety
owned righB; or
-------
TECHNICAL REPORT DATA
(Tircse read Jusirvciujns on the reverse Itefore com picf
l. HGPORT no. ?. |3.
EPA-600/R-93-0G4C j
c.. TITLE AMD SUBTITLE
Proceedings: 1991 SO2 Control Symposium, Volume 3.
Sessions 5B and 6
«j. nCPORT OATG
April 1993
6. PERFORMING ORGANIZATION COOE
7. AUTHOH(S)
Mis cellaneous
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
1 a. SPONSORING AGENCY COOE
EPA/600/13
is.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. 2 is Sessions 4 and 5A. Vol. 4 is Session 7, and Vol. 5 is Session 8.
16. abstractproceedings 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 commercialisation 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.IDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Pollution
Sulfur Dioxide
Nitrogen Oxides
Flue Gases
Desulfurization
Coal
Pollution Control
Stationary Sources
13 B
07B
21B
07A, 07D
2 ID
IB. DISTRIBUTION STATEMENT
Release to Public
19. SECURITY CLASS (ThisReport)
Unclassified
21. NO. OF PAGES
502
30. SECURITY CLASS (Thispage)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73) 6B-143
-------
12PA-600 /R-93-0 64 c
April 1993
Proceedings: 1991S02 Control Symposium
Volume 3. Sessions 5B and 6
For Sponsors:
Electric Power Research Institute
B. Toole O'Neil
3412 Billview Avenue
Palo Alto. CA 94304
U.S. Department of Energy
Charles J. Dnusmood
Pittsburgh Energy
Technology Center
P.O. Box 10940
¦Pittsburgh. PA 15236
lis. Environmental Protection Agency
Brian K. Gollett
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
US. 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 US. 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 (S02/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 Add Additives
A Briefing Paper for the Status of the Flue Gas Desulfuiization
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 die 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, EPRTs 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 S02 Control by Diy 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
LEFAC 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 ADVACATE 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 Humidifiestion
High SO2 Removal Dry FGD Systems
4B-151
4B-173
4B-189
4B-205
Session 5A - Wet Full Scale FGD Operations
FGD System Retrofit for Dalhousie Station Units 1 & 2
Ziminer FGD System: Design, Construction, Start-Up
and Operation
Results of an Investigation to Improve the Performance and
Reliability of HL&Fs 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 Desulfuiization 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
5A-1
5A-17
5A-37
5A-59
5AS1
5A-101
5A-127
5A-143
VOLUME 3
Session 5B - Combined SOx/NOx Technologies
Simultaneous SOx/NOx Removal Employing Absorbent Prepared
from Fly Ash 5B-1
Furnace Slurry Injection for Simultaneous SO2/NOX Removal 5B-21
Combined SO2/NOX Abatement by Sodium Bicarbonate
Dry Injection 5B-41
vx
-------
SO2 and NOx Control by Combined Dry Injection of Hydrated
Lime and Sodium Bicarbonate 5E-67
Engineering Evaluation of Combined N0x/S02 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 Add 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 Held and Laboratory Testing 6A-139
Sulcis FGD Demonstration Plant Limestone-Gypsum Process:
Performance, Materials, Waste Water Treatment 6A-163
Session 6B - dean Goal 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 15-MWe Magnesia Dry-Scrubbing Demonstration
Project 6B-79
Application of DOW Chemical's Regenerable Hue Gas
Desulfuxization 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 SQx/NOx Control Via Soxal™, A Regenerative Sodium
Based Scrubbing System 7-61
Hie 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 7-199
Initial Operating Experience of the SNOX Process 7-221
Progress Report of the NTPSCO - Pure Air - DOE Clean Coal II
Project 7-241
Development of a Post Combustion Dry SO2 Control Reactor
for Small Scale Combustion Systems 7-253
Scrubber Reagent Additives for Oxidation Inhibited Scrubbing 7-269
Recovery of Sulfur from Calcium Sulfite and Sulfate
Scrubber Sludges 7-277
Magnesite and Dolomite FGD Technologies 7-285
SO2 and Particulate Emissions Reduction, in a Pulverized Coal
Utility Boiler through Natural Gas Cofifing 7-293
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 7-321
VOLUME 5
Session 8A - Commercial FGD Designs
Mitsui-BF Dry Desulfurization and Derdtrification Process
Using Activated Coke 8A-1
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.
Ut3iti.es for Add Rain Remediation 8A-21
Incorporating Full-Scale Experience into Advanced Limestone
Wet FGD Designs 8A-43
Design and Operation of Single Train Spray Tower FGD Systems 8A-69
Selecting the FGD Process and Six Years of Operating Experience
in Unit 5 of the Altbach-Deizisau Neckarwerke Power Station 8A-93
Development and Operating Experience of FGD-Technique at the
Vcelklingen Power Station 8A-121
Advantages of the CT-121 Process as a Throwaway FGD System 8A-135
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 £>n 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
-------
I
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 US. Environmental Protection Agency (EPA), and the US- 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 EPRIJEPA 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 Rdler - Assistant,
Policy subcommittee on Energy and Power, U. S. Congress,
Jack . . S. Siegel - Deputy Assistant Secretary, Office of Coal Technology, US.
Department of Energy, and Michael Shapiro - Deputy Assistant Adminstrator,
Office of Air and Radiation, U. S. Environmental Protection Agency.
The assistance o£ Steve Hoffman, independent, in preparing the
manuscript is gratefully acknowledged.
The following persons organized this symposium:
• Barbara Toole CNefl - Co-Chair, Electric Power Research Institute
• Charles Drummond - Co-Chair, US. Department of Energy
• Brian K. Gullett - Co-Chair, US- 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 EPRI Perspective - S-M. Dalton
1-2 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 t
Technology, U.S. Department of Energy
Michael Shapiro - Deputy Assistant Adminstrator, Office of Air
and Radiation, U. S. Environmental Protection Agency
Session 1 - Clean Air Act Compliance Issues/Panel
Session Moderator: S. Jenkins, Tampa Electric Co.
Comments by:
Alice LeBlanc - Environmental Defense Fund
Karl Moor, Esq., Balch & Bingham
John Palmisano AER*X
Craig A. Glazer - Chair, Ohio Public Utilities Commission
xii
-------
Session 2 -Clean Air Act Compliance Strategies
Session Chair: 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, WS. 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, CP. 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 Chair: David R. Owens - EPRI
Overview on the Use of Additives in Wet FGD Systems, R.E. Moser,
EPRI
Results of High SO2 Removal Efficiency Tests at EPRTs HSTC, G.
Stevens, Radian
Results of Formate Additive Tests at EPRTs HSTC, M. Stohs, Radian
Corp.
FGDPRISM, EPRTS FGD Process Model-Recent Applications, J.G.
Noblett, Radian Corp.
Additive Enhanced Desulfurization for FGD Scrubbers, G. Juip,
Northern States Power
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 EPRTs
HSTC, J. Wilhelm, Codan Associates
xiii
I I
3A-1
3A-2
3A-3
3A-4
3A-5
3A-6
-------
Session 3B - Furnace Sorbent Injection
Session Chain Brian GuIIett - 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 CaO-SC>2 Reaction,
L Bjerle, University of Lund
3B-3 Status of the Tangentially Fired LIMB Demonstration Program at
Yorktown Unit No. 2: An Update, J J". 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 Flue Gas
Desulfurization Plants in UK, L 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. do Vale, Wheelabrator Air
Pollution Control
4A-6 Guidelines for FGD Materials Selection and Corrosion Protection, FLS.
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 FJIP. 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^. Enwald, Tampella Power Ltd.
4B-2 1.7 MW Pilot Results for Duct Injection FGD Process Using Hydra ted
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, US. EPA
4B-6 Suitability of Available Fly Ashes in AD VAC ATE Sorbents, C. Singer,
US. 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 TV A 10-MW
Spray Dryer/ESP Pilot, T. Burnett, TV A
4R-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
FGD System Retrofit for Dalhousie Station Units 1 & 2, F.W. Campbell,
Burns & McDonnell Engineering Co.
Zimmer FGD System, W. Brockman, Cincinnati Gas & Electric
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
Full-Scale Demonstration of EDTA and Sulfur Addition to Control
Sulfite Oxidation, G. Blythe, Radian
xv
5A-1
5A-2
5A-3
5A-4
-------
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 SQ2/NOx Removal, BJC
Gullett, U.S. EPA
5B-3 Combined SC>2/NOx Abatement by Sodium Bicarbonate Dry Injection,
J. Verlaeten, Solvay Technologies, Inc (124)
5B-4 S02 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 NOx/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 S02/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 HQ, J.
Peterson, Radian Corp.
6A-2 Control of Acid Mist Emissions from FGD Systems, RS. 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 J. 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 FIvls hi ess Mechanical Seals in Wet FGD
Plants through Field and Laboratory Testing, F.E. Manning, BW/IP
International Inc.
6A-8 Sulds 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, LJi. 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 Humidifies tion for
Enhanced SO2 Removal, D J*. 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
S02 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, AS. 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
-------
7-17 Scrubber Reagent Additives for Oxidation Inhibited Scrubbing, J-
Thompson, Process Calx, Inc.
7-18 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 Chain 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 5B
COMBINED SOx/NOx TECHNOLOGIES
SIMULTANEOUS SOX, NOX REMOVAL EMPLOYING ABSORBENT
PREPARED FROM FLY ASH
H.Nakamura
Y.Katsuki
Hokkaido Electric Power Corporation
461-6, Satozuka, Toyohira-ku
Sapporo, Japan
S.Kotake
M. Kagami
Mitsubishi Heavy Industries Ltd.
5-1, Marunouchi 2-chome, Chiyoda-ku
Tokyo, Japan
ABSTRACT
Hokkaido Electric Power Co. and Mitsubishi Heavy Industries Ltd.
have jointly started a program for the development of a
simplified, effective and economical flue gas treatment system
since the beginning of last year. This system employs absorbent
of a Lively Intensified Lime-Ash Compound (LILAC), which is
produced by fly ash, lime and gypsum through a hot water curing
process, and we therefore have named our system as the "Lilac
Process".
This absorbent is highly reactive with SO2 and NOx in flue gas and
is sprayed into the flue gas in the form of either slurry or
powder.
The resultant solids are collected in the dust collector (Bag
filter or Electrostatic precipitator) installed downstream of the
absorbent spraying.
First, the Lilac Process with spraying absorbent slurry has been
established mainly for SO2 removal from flue gas through a series
of bench scale tests on reactivity of absorbent, absorbent
production process, SO2 removal efficiency, quantity of the
absorbent sprayed, approach temperature in the spray dryer or the
bag filter, and salt addition together with a series of model
tests of a rotary atomizer which MHI has recently developed.
5B-1
-------
employing a high frequency induction motor.
In the bench scale test more than 95% SO2 removal efficiency was
attained when the absorbent was sprayed at Ca/S of 1.2 with 5wt$6
of chloride (CI). In order to establish the details of process
and mechanical design for commercial plants, a pilot plant having
10,000m3N/h capacity is being constructed in the Tomato Atsuma
Power Station of Hokkaido Electric which will be commissioned in
October 1991.
Second, the Lilac Process with spraying absorbent powder has been
investigated by series of bench scale test at 80m3N/h capacity,
and simultaneous removal of 90% SO2 and 10% NOx has been
confirmed at Ca/S of 2.7 without any additives.
The duct injection nozzle and production facility of the powder
reagent are also being installed in the pilot plant mentioned
above and the demonstration of this process is planned to start
in 1 993 .
Waste disposal from flue gas treatment is one of the big concerns
in view of environmental protection and the potential for
effective use of disposal material from the Lilac Process has
been confirmed.
5B-2
-------
INTRODUCTION
Protection of global environment has been internationally
recognized to be an important issue which should be immediately
dealt with. Japanese government has taken a positive attitude
toward solving global environmental problems such as acid-rain
and global warming by offering technologies and finance. Our
companies have also formulated a policy to participate in the
international technical cooperation for protection of global
environment based on our technology for the flue gas
desulfurization system.
Acid rain is one of the crucial environmental problems caused by
SOx and NOx emitted from various types of boiler and engines. We
have already established a technology for flue gas
desulfurization system in which fly ash is utilized as one of the
components of absorbent to adapt coal burning boiler such as for
electric power generation. The commercial plant of the dry type
flue gas desulfurizat ion system has been running for more than
six months at the Tomato Atsuma Power Station.
Based on the established FGD system, we have extended R&D to
develop more simplified,efficient and economical flue gas
treatment system for more versatile demands. In the present
paper, the newly developed flue gas treatment system is
presented.
LILAC PROCESS WITH SPRAYING ABSORBENT SLURRY
Bench Scale Test Facility and Procedures
The test facility consists of two units, a gas treatment unit and
an absorbent preparation unit (in Figure 1).
The gas treatment unit (Figure 2) consists of a gas generator
which produced test gas simulated as flue gas from a coal-fired
5B-3
-------
boiler at temperatures of up to 14 Ot, a spray dryer with air
atomized spray nozzle, and a bag filter which can treat up to 20
m3N/h of gas flow.
Gas sampling was taken from the inlet/outlet of the spray dryer
and outlet of the bag filter.
The absorbent preparation unit (Figure 3) consists of a mixing
tank and a curing tank made of glass having a 3 liters capacity.
Fly ash, slaked lime and gypsum were mixed with water weighing 5
times the total solids in the mixing tank and this mixed solution
was kept agitated at 95^C for 3, 6 or 12 hours in the curing tank
corresponding to each test condition. The hot water cured
solutions, called absorbent slurry, was sprayed into the prepared
gas stream in the spray dryer. The slaked lime slurry was also
tested in this facility as a base line test for corresponding
conditions of various parameters such as inlet gas conditions,
Ca/S, approach temperature, etc.
The effect of salt addition into the absorbent slurry was also
confirmed by this test facility. The quantity of salt added was
controlled so the disposal solid material contained 5 percent in
weight of chlorine.
Bench Scale Test Results
The measured desulfurization efficiency in the spray dryer and
the spray dryer plus the bag filter is shown in Figure 4 for
various Ca/S.
Around 30% higher desulfurization efficiency was obtained by the
hot water cured absorbent (12 hours) as compared to the slaked
lime, which proves higher reactivity of amorphous compound of
Si02, AI2O3, Ca(OH)2 and CaSCM formed by a hot water curing.
The effect of the duration on hot water curing in absorbent
preparation is shown in Figure 5. Comparing desulfurization
efficiencies of various absorbent prepared by different curing
time, the test results advise us that curing time should be
around 6 hours. The microstructure of various absorbents are
shown in Figure 6. The existence of amorphous compound showed
higher reactivity of absorbent.
The effects of salt addition into the absorbent slurry is shown
in Figure 7. Comparing desulfurization efficiencies of absorbent
with/without seawater addition, the test results indicate about
5B-4
-------
20% improvement in the desulfurization efficiency by the addition
of seawater. This improvement would be understood as longer
existence of water in the absorbent resulting from the effect of
boiling point raise of salted water.
Pilot Plant
In order to demonstrate performance of the Lilac Process and to
establish design parameters necessary for constructing a
commercial scale plant, a pilo' plant having the following
specifications is being constructed in the adjacent area of the
No.2 boiler unit of Hokkaido Electric's Tomato Atsuma Power
Station.
Capacity
Inlet SO2
NO
Ely ash
Temperature
Spray Dryer
Rotary Atomizer
Dry Electrostatic Precipitator
Bag Filter
Curing Tank
10,00 0 m^/h
37 0 - 2,00 0 ppm
150 - 500 ppm
10 - 15 g/m3!!
130 - 150 "C
2. 8 m* X 12 '.9 mH
7.5 KW, 17,500 rpm
94 m2
87 m*
1.11 m3 X 4 sets
The pilot plant has been so designed as to confirm process
performances of ;
1. Hot water cured absorbent slurry production.
2. Hot water cured absorbent powder production.
3. Desulfurization/denitration by combination of a spray
dryer and a dry electrostatic precipitator or a bag
filter.
4. Various combinations of SO2 and NO in the inlet gas by
supplemental addition of these gases into the flue gas
from the commercial 600MW coal-fired boiler.
The flow diagram of the pilot plant is shown in Figure 8 and a
photograph of the side view of the plant in Figure 9.
The rotary atomizer is the key piece of equipment for the plant
and is specially designed to meet requirements for spraying the
absorbent slurry.
The construction of the rotary atomizer using a high frequency
induction motor is illustrated in Figure 10. Before installing it
in the pilot plant, the rotary atomizer has been tested in the
5B-5
-------
shop with absorbent slurry and ;.ts estimated performance was
confirmed both in sprayed droplet size and power consumption, as
shown in Figure 11.
LILAC PROCESS WITH SPRAYING ABSORBENT POWDER
Bench Scale Test Results
Bench Scale Test Facility and Procedures. The test facility for
gas treatment unit was constructed at the Tomato Atsuma Power
Station. The absorbent powder was prepared at a different place,
and brought to the test facility. A part of flue gas from the
coal-fired boiler of the commercial plant was diverted to the
test facility as an inlet gas. The inlet gas volume was 80m3N/h.
Effect of SQ2/NOX ratio on the removal of SQ2 and NOX. The effect
of S02/N0x ratio on the simultaneous removal of SO2 and NOx was
examined. The SO2/NOX ratio was varied by injecting of SO2 and/or
NOx to the flue gas from the commercial boiler.
Figure 13 shows the effects of the SO2/NOX ratio on the removal of
SO2 and NOx. As the SO2/NOX ratio increased, the NOx removal rate
drastically increased, but the SO2 removal rate gradually
decreased.
Effect of reaction temperature on the removal of SQ2 and NOX. As
shown in Figure 14, SO2 removal rate is constant in the
temperature range 7 0 - 130NOx removal rate, however, is
drastically increased between 70 - 90"C, and become constant above
90t:.
Effect of moisture on the removal of SQ2 and NOX. As shown in
Figure 15, as moisture of flue gas increased, both SO2 and NOx
removal rate increased.
Reaction Mechanism of SQ2. NOx removal
The tracer study in which N180 and 10O2 were used demonstrates that
the main species to oxidize SO2 is the NO2- adsorbed on the
surface of the absorbent. We propose the desulfurization
5B-6
-------
reaction mechanisms as shown below:
Overall
2CaO + 2S02 + 02 = 2CaS0i
Elementary process
2N0 +02 = 2N02
SO2 + NO2 = S03 + NO
CaO + SO3 = CaS04
In contrast to the desulfurization mechanisms, the denitration
mechanisms are not definite yet. However, it is known that NOx
is fixed in the form of CaN03 and that SO2 is associated with the
oxidation of NO, because NOx removal rate increased with an
increase in the SO2/NOX ratio, as shown in Figure 16
In addition, the NOx removal rate increases linearly with the Si02
content in the absorbent, suggesting that Si02 plays an important
role in the denitration. mechanism (Figure 17)
POTENTIAL UTILIZATION OF SPENT ABSORBENT
Waste disposal material from the LILAC Process named Spent
Absorbent (SA) is a neutral and a stable material. The leaching
value examination for its harmful components was within the
acceptable limit of the Japanese quality standard for landfill
materials .
SA can be used as one of the raw materials for preparation of the
absorbent in place of gypsum, because SA contains a high
percentage of gypsum. For the same reason, SA is expected to be
used as a construction material.
Other utilizations of SA are now examined for use in the
following.
Treatment for sludge
Since SA is a porous material and has ability to coagulate and
deodorize, it can be used for sludge treatment and cleaning of
muddy water.
Deodorizing agent
The ability of SA to absorb NH3 and H2S makes it a possible
deodorizing agent.
5B-7
-------
CAPITAL COST AND LEVELIZED COST
In order to evaluate economic aspect of the Lilac process FGD
plant, investment costs and operation costs of the following
three FGD plants were estimated under the same process design
criteria and economic criteria.
1. Lilac Process (with Bag Filter)
2. Spray Dry Process with lime absorbent (with Bag Filter)
3. Wet Limestone Gypsum Process (with Dry Electrostatic
Precipitator)
For the scope of complete FGD plant (including desulfurization
towers and dust collectors, absorbent storage and preparation
facility, by-product /waste material storage and loading facility)
for a 250MW pulverized coal-fired generating plant, the
investment cost of the Lilac Process is found to be 80 while that
of the Spray Dry Process is 80 and that of the Wet Limestone
Gypsum Process is 100.
The operation costs (total cost of raw materials, utilities,
operation and maintenance, finance and management) levelized as
expense per ton of S02 on the other hand is 95 for the Lilac
Process, 110 for the Spray Dry Process and 100 for the Wet
Limestone Gypsum Process .
CONCLDSION
The Lilac Process is featured for its higher SO2 removal with
absorbent slurry spraying and for its simultaneous SO2 and NOx
removal with absorbent powder spraying, which the existing flue
gas treatment systems are unable to achieve costwise as well it
is competitive with the presently available systems. In addition
its spent absorbents are valuable utilizable resources for
deodorization agent and construction material production. From
all considerations, Lilac Process is a promising comprehensive
flue gas treatment system of high performance.
5B-8
-------
I 1
S02
NO
HC1
02
C02
^>-{X]
^£^>-Kx>
¦&y
kx—*2
-£*3-
¦£*3-
1 1
-[X]—*¦ *1
-ix-
&
-0
/FT
3
Gas preparation
Humidifier
9 Absorber supply unit
a
¦-en
—&
1
rffiij. ;l I
' 1 ]
/
Nozzle
©
\
-©
-©
g>-
-©
©¦
~£
~-~b
¦©
Heater Spary dryer
Bag filter
Figure 1 Flow Olagram of Bench scale Test Facility
Note: ® : Gas Sampling
-------
Figure 2 Bench Scale Test Facility for Gas Treatment Unit
Figure 3 Bench Scale Test Facility for Absorbent Preparation Unit
5B-10
-------
100
Hot Water Cured Absorbent
SD+BF
SD+BF
u
Slaked Line
LU
C
o
SD=Spray Dryer
BF=Bag Filter
Approach Temp=12°C
1.5
0.5
1.0
Ca/S (mol/mol)
Figure 4 Desulfurization Efficiency vs Ca/S
ICO
U
<4.
<4-
LU
C
o
4->
«0
fM
Approach Temp.=15°C
3
CO
a;
o
Ca/S (mol/mol)
Figure 5 Activity of Absorbent by Curing Time
5B-11
-------
Curing Time = 3 hours Curing Time = 12 hours
Figure 6 Microstructure of Absorbent (*50,000)
100
o
C
O)
Approach Temp. = 10°C
c_>
«4-
«4-
LU
C
o
Salt
Addition
Key
3
<*-
3
lo
CD
Yes
o
Ca/S (mol/mol)
Figure 7 Effect of Salt Addition into Absorbent
5B-12
-------
Ul
V
M
W
MMS
Mix Tank
Supply Tank
Atritor
Curing Taks
Figure 8 Process Flow Diagram of Pilot Plant
-------
Figure 9
Pilot Plant
Magnetic
Thrust Searing
Bearing
Magnetic Bearing
Control Unit
~c
High Frequency
Induction Motor
Stator
Rotor
Magnetic
Radial Bearing
Bearing
Cartridge Type
Rotary Atomizer Unit
Disc
Figure 10 Construction of Rotary Atomizer
5B-14
-------
10
o 6
o.
E
O
w
2
o
CL.
—
1
m 20,000rpm _
-
y*
17,500rpm _
- /
12,000rpm
i
500
1000
Spray Quantity (0/h)
Figure 11 Power Consumption of Rotary Atomizer
Duct
by-pass
absorbent
powder
coiplessing
¦ air
blower
heater
cyclone
bag filter
NO cyIinder
S02 cylinder
injection nozzle
Figure 12 Flow Diagram of Bench Scale Test Facility
(Absorbent Powder)
5B-15
-------
100
80
2
73
cc
75 60 h
5
E
CC
x
O
• S02 Removal
o NOx Removal
40 -
CM
o
CO 20
o ' ' I I I 1 I L.
0 12 3 4
-1 L.
S02/NOx Ratio
Figure 13 Effect of S02/N0x Ratio on SO2. NOx Removal Rate
100
• SO2 Removal
o NOx Removal
so .
o
E
-------
100
80
o
"5 60
DC
"5
>
o
o
E
\ -
1 2
SO2/NOX Ratio
Figure 16 Effect of SO2/NOX Ratio (Laboratory Test)
5B-17
-------
120
O : S02 Removal
~ : NOx Removal
100
CO
O
6> 80
\
c
E
>
40
o
<
20
80
60
20
40
0
Si02 (%)
Figure 17 Effect of S102 Content on the activity
(Laboratory Test) (Activity: Defined to be a period to keep removal %
above 80%)
5B-18
-------
Furnace Slurry Injection for Simultaneous SO2/NOX
Removal
5B-19
-------
Intentionally Blank Page
5B-20
-------
Brian K. Gullett
U.S. Environmental Protection Agency
Air and Energy Engineering Research Laboratory
Research Triangle Park, NC 27711
Kevin R. Bruce and Halter F. Hansen
Acurex Corporation
Research Triangle Park, NC 27709
John E. Hofmann
Nalco Fuel Tech
Naperville, XL 60566
ABSTRACT
This paper discusses the results of a cooperative research venture between
the U.S. Environmental Protection Agency and Nalco Fuel Tech investigating furnace
urea/sorbent slurry injection for joint SOj/NO, removal. This emission reduction
technology has been developed as a low capital cost option for electric utilities
and industrial sources in response to the 1990 Clean Air Act Amendments. The
slurry was composed of a urea-based solution and various Ca-based sorbents,
totalling 30% solids by weight. Testing on a natural gas pilot scale reactor
achieved 80% reduction of SOj and NO, at reactant/pollutant stoichiometric ratios of
2/1 and 1/1, respectively. SO2 emission reductions from slurry injection were
enhanced compared with dry Ca(OB}2 sorbent injection methods possibly due to
sorbent fracturing to smaller, more reactive particles. Further, the addition of
the urea-based solution for NOz reduction had a synergistic effect upon SO,
reduction. The effect of injection temperature and stoichiometric ratio upon SCu,
NO,, NHj, and N40 was determined for the combined sorbent and urea-based solution.
Emissions of NH3 and NjO when using a modified urea-based formulation were found to
be significantly lower than previously reported data. The results of this pilot
scale study have shown high reduction of both SO- and NO,, suggesting the need for
full scale studies to further assess this combined sorbent/urea-based slurry
injection technology.
This paper has been reviewed in accordance with the U.S. Environmental Protection
Agency's peer and administrative review policies and approved for 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.
5B-21
-------
INTRODUCTION
Passage of the 1990 Clean Air Act: Amendments has initiated extensive evaluation and
planning for strategies to meet these stricter emission requirements. In a two
phase approach, the Clean Air Act requires reduction of SQj emissions at 265 units
by about 40% to 1075 ng SOj/J (2.5 lb SO^/million Btu) (based on 19S5 to 1987
emissions) by January 1, 1995. By 2000 all SOj sources are affected and must
reduce emissions to 516 ng SOj/J (1.2 lb SO^/million Btu). In addition, the second
phase brings a cap on emissions at 8.08 million Mg SO- (8.9 million tons), or about
9.08 million Mg (10 million tons) less than SO^ emissions in 1980. Thus, new
sources must be offset by further reduction in emissions from existing sources.
The regulations call for reduction in NO, emissions consistent with capabilities of
low-NO, burner technology, which is, as of yet, undefined. Additionally, the
possibility of trading NO, for SO, emissions is under consideration [1].
Installation of wet scrubbers or fuel switching/modification is projected to
account for up to 65% of the first phase compliance strategies [2]. However,
utilities have an option for earning emission credits by adopting early compliance
strategies or further reducing emission levels below those required. Among the
early compliance options available to utilities is furnace sorbent injection
technology. This retrofittable, lower capital cost technology may also be a likely
candidate for long term compliance on older or smaller boilers, plants that are
limited by physical space, or utilities that opt for low capital technologies.
Furnace sorbent injection is a technology that has been field tested on a
number of units, achieving, for example, 63% removal at a Ca/s « 2/1 with a calcium
hydroxide [Ca(OB)2] sorbent and 72% with a surfactant-modified Ca(OB)z sorbent on a
105 MW(e) wall-fired unit [3]. A field demonstration on a 180 MW(e) tangentially
fired unit should produce preliminary results around mid-1992 [4].
The anticipated NO, regulations are likely to be met by a number of varied
technologies including low NO, burners, gas returning, and selective or non-
selective catalytic reduction. These technologies represent a range of removal
efficiencies and costs. One of the more low cost, retrofittable options is
selective non-catalytic reduction (SNCR) which has been shown to achieve 63% NO,
reduction on a 150 MW(e) coal boiler at a reductant/NO, stoichiometric ratio (NSR)
of *2/1 [5] and has been the subject of numerous laboratory or pilot scale studies
[6,7]. SNCR involves high temperature (about 800 to 1100°C) furnace injection of a
N-based reducing agent such as urea (NH:CONH:) or ammonia (NB,) which converts NO, to
n2.
Most concerns with use of SNCR center around NB, slip resulting from
incomplete reaction and production of nitrous oxide (N/)) due to incomplete
reduction. NH, slip can result in formation of ammonium bisulfate (NB^S04) and
ammonium sulfate [(NH^)-SO«] which readily deposit upon air preheater surfaces
causing reduced heat transfer, increased pressure drop, and formation of NH^Cl
which causes a visible white plume in the stack emissions. NjO has been implicated
as a contributor to stratospheric ozone depletion [8] and global warming, the
latter due to its ability to absorb infrared radiation [9]. Research has
demonstrated that levelB of NjO and NH} emissions from various SNCR compounds are
extremely sensitive to injection temperature [10,11]. Efforts to widen the
applicable temperature injection window and control NH} slip and NjO production
through use of additives [12] have brought about some success, yet these remain
concerns that need to be addressed on any SNCR-type process.
The O.S. EPA has conducted research at its Air and Energy Engineering
Research Laboratory (AEERL) with Nalco Fuel Tech to investigate a combination of
5B-22
-------
the furnace sorbent Injection and SNCR technologies for simultaneous SO^/NO,
control. The mechanism for this industry/Government research was the Federal
Technology Transfer Act's (FTTA) Cooperative Research and Development Agreement
(CRDA), an agreement whereby EPA can conduct research with private industry at
EPA's research facilities. Research on a similar SO^/NO. control process has shown
considerable merit [13], yet significant questions still remain in the industry
concerning NH, emissions and HjO by-product formation.
The objective of this research was to develop the technology of simultaneous
SO, and NO, removal by injection of a Ca- and urea-based slurry while minimizing
emissions of tifi and NH,. Variables of operation included injection temperature,
stoichiometric ratio, sorbent type, and urea-based solution composition. Emissions
monitoring results for SOj, NO,, NjO, CO, and NH3 are reported.
EXPERIMENTAL
Furnace
Dry sorbent and slurry injection tests were run on a pilot scale 14.7 kw
(50,000 Btu/hr), refractory lined, down-fired cylindrical furnace capable of firing
natural gas or coal. The furnace, termed the "Innovative Furnace Reactor" (IFR),
has an inner diameter of 15.2 cm and an overall length of about 4 m (see Fig. 1).
View and injection/probe ports traverse the length of the furnace for testing
flexibility. The furnace is used to simulate the gaseous combustion environment
and quench rate conditions anticipated in utility and industrial boilers. During
natural gas firing, this is accomplished by doping the fuel with NHj (which is
oxidized to form NO,) and SOj. Typical operating concentrations were 600 ppm NO,
and 2500 ppm SO-. The furnace was operated with tangential and axial air totalling
0.39 m3/min (13.72 ft3/min) STP, including an excess air of 50%.
Emissions Sampling
Gas emissions are sampled in the horizontal arm section of the IFR (see Fig.
1) and pass through heated sample lines to continuous emission monitors (CEMs).
Analysis of SO2 concentration by an ultraviolet analyzer follows particle traps and
a heated sample line («350°C). SO? removal percentages reported in this work are
typically determined by running at least six tests between Ca/S = 1/1 and Ca/S =
3/1. These values are then curve fit by a regression technique and interpolated to
the reported removal at Ca/S = 2/1. NO, is analyzed by a chemiluminescent method.
This method reports NO, concentrations that do not include NO,; earlier tests
showed that the NOj concentrations were below 5% of the total NO, concentration.
All gas emission results are corrected to 0% O, levels.
Gases analyzed for CO,, O,, and CO were first passed through a gas dryer and a
desiccant canister of anhydrous CaS04. All of the above on-line CEMs are zeroed
and spanned with gases of known concentration both before and after each daily
trial.
NjO concentrations were monitored by both on line gas chromatography (GC) and
tunable diode laser infrared (TDIR) spectroscopy methods. The GC was used for
analysis of grab samples taken before and during testing using procedures in
Reference [14]. The TDIR was used to monitor real time stack NjO emissions. The
TDIR compares the infrared absorption of the gas sample to a known concentration of
NjO span gas at the wavelength of NjO. This method and apparatus, which are
detailed further in Reference [15], were calibrated for this work over the 20 to 80
ppm range, with an accuracy of +/- 0.75 ppn. The two methods' results were
comparable. Tests conducted at six varying conditions showed a linear correlation
coefficient exceeding 0.99 between the two methods (for further comparison of NjO
analytical methods see Reference [16]).
The analysis of stack gas NH, concentration was completed by wet methods
using a Fisher Accumat ion selective electrode. The stack gas was drawn through an
impinger system containing 0.02N H^SO^. Prior to measurement, the pH was adjusted
with 10M NaOH solution. The ion selective electrode, coupled with a pH meter,
determines the NH, concentration. The meter and electrode were calibrated prior to
analyses with known standards and checked throughout the testing to ensure that the
values fell within the manufacturer's limits.
5B-23
-------
Sorbent/Urea Solution Injection
Testing during this work included dry sorbent injection, slurry sorbent
injection, injection of two Nalco Fuel Tech-supplied urea-based solutions termed
NO.OUT A and NO.OtJT A+, and simultaneous injection of a NOxOUT A/sorbent slurry.
The aorbents tested consisted of CaO, Ca(OH)2, and CaCO,, all supplied by the Tenn
Luttrell Company. Dry sorbent was fed by a K-Tron loss-in-weight, twin screw
feeder which was calibrated prior to and after each run. Slurried sorbent at 30%
Bolide by weight was continually mixed in a tank and metered into a Turbotak nozzle
by a calibrated peristalic pump. Baseline emission values prior to testing slurry
injection were monitored while injecting an equivalent amount of deionized water
(H^O). NOxOUT (A or A+) was metered into the water or slurry injection by means of
a calibrated dual syringe pump. A typical test scenario involved baseline emission
monitoring during H2Q injection without NOxOUT (A or A+) flow, then addition of
NOxOUT {A or A+) to the slurry flow, and a final return to HjO-only injection to
ensure return to baseline emission concentrations.
Both dry Borbent and slurry were injected through water-cooled probes that
inject coaxially to the process gas. The dry sorbent probe injects 15.7% of the
total IFR air flow to effect sorbent conveyance and dispersion. The Turbotak
slurry probe uses air (18% of the total IFR air flow) to effect droplet
atomization.
The slurry droplet size distribution exiting the Turbotak nozzle was
determined through use of a Hunhall particle size analyzer which determines droplet
size by measuring diffraction of laser light. These droplet sizes were measured
outside of the IFR using H^O flow rates and pressures identical to in-furnace
operation. The nozzle had a droplet size distribution with a of 13 fim and a Dm
of 88 fim. Prior to IFR testing, analysis of the slurry droplet size distribution
with a spray trajectory model [17] ensured that the large droplets would not
impinge on the furnace walls or remain unvaporized. Sorbent particle sizes were
determined in a bench top measurement using a Micromeritics Sedigraph Model 5100.
Solid sampling
IFR solid samples were collected isokinetically with a water-cooled sample
probe. Gases passed through a particle filter and ice bath impingers then into a
dry gas meter with flow rate control to ensure isokinetic sampling. These solid
samples were analyzed by x-ray diffraction to identify compounds of reaction.
Diffraction analyses were run on a Siemens diffractometer with a copper Ka target
source running at 50 kV and 40 mA.
Temperature Profiles
The temperature profiles through the IFR firing natural gas were determined
by using a suction pyrometer and a type R thermocouple. Temperatures were
determined during injection of air or air/H.O to mimic the conditions expected
during dry sorbent and slurry/NOxOUT {A or A+) injection, respectively. The
temperature at the point of the injection nozzle was calculated by extrapolation of
the temperature values from downstream ports. The quench rate for natural gas
firing with slurry and dry sorbent injection was nominally 240 and 293°C/s,
respectively, over the range of injection ports.
RESULTS
Sulfur Dioxide Tests
Initial tests compared the SO. removal of slurry versus dry injection modes
for both Ca(OH)z and CaCOj,. Figure 2 shows the effect of varying injection
temperature upon the S02 removal by CaC03 at a Ca/S ratio of 2/1. The S02 removal
during both dry and slurry CaCO-, injection was fairly independent of injection
temperature, given the relative error in the plotted values. Both dry and slurry
injection appear to have relative maxima in SO? removal, about 50%, around the
1000°C injection temperature. The addition of NOxOUT A solution to the slurry
water (replacing an equal volume of water) may have caused a slight increase in SO.
capture, but insufficient runs were completed for statistical certainty.
The same tests for dry injection of Ca(OH)3 (Fig. 3) indicate that SOj
removal, with a around 1200°C, was relatively independent of injection
temperature. The slurry injection curve is similar to the dry sorbent injection
5B-24
-------
curve except for a significant maximum in SOj removal around 1000°C, where SO,
capture increases to about 73%. Tests with NOxODT A addition to the slurry water
mimic the temperature response of the sorbent-alone slurry, but indicate
significantly higher SO, removal (about 10%, absolute) up to a maximum around 85%
capture.
Limited tests were also done with commercially available Tenn Luttrell CaO
(lime). In these tests, as-received Ca(OH): was tested against a CaO slaked with
the slurry injection water prior to injection. The results (also shown on Fig. 3)
indicate that injection of a CaO slaked under non—optimized hydration conditions
yields equal SO- capture to the as-received Ca(OH),. Similarly, injection of the
slaked CaO slurry with NOxODT A resulted in similar capture to the as-received
Ca(OH)2 with NOxOUT A, about 85% at Ca/S = 2/1.
Tests varying the particle size of the CaCOj sorbent were conducted for dry,
slurry, and slurry with NOxOUT A injection conditions. Results at the optimum
injection temperature for SO. removal and at a Ca/S ratio of 2/1 are compared
against the Ca(OH)2 results (Fig. 4). smaller particles react more quickly,
whether they are Ca(OH)2 or CaCO,. The likely enhancement of dry sorbent SO.
capture from either slurry injection or NOxOUT A addition is maintained independent
of particle size. Results for the smallest particle sizes tested show that, while
grinding CaCO^ to sizes comparable to Ca(OH). results in equal reactivity through
dry sorbent injection, the same is not true for slurry injection and (especially)
slurry injection with NOxOUT A.
Nitrooen Oxide Tests
Tests were conducted over a range of temperatures to measure the temperature
sensitivity of two urea-based NO, reductants, NOxOUT A and NOxOUT A+. NOxOUT A is
a concentrated solution of urea in water with an antiscalant and dispersant
formulation. Tests varied from about 821 to 1170°C with an NSR of 1/1. The
results of testing with NOxOUT A, including NO,, NH}, N,o, and CO, are shown in Fig.
5. For reference, S02 removal results from slurry injection are superimposed on
this figure, although these results were not obtained simultaneously (other results
showed that the effect of concurrent sorbent injection upon NO, removal is
unnoticeable; tests with and without sorbent in the slurry did not prove to affect
NO, removals). For NOxOUT A, a peak NO, reduction of 82% is achieved at the optimum
temperature of about 1100°C, while NO, reductions greater than 70% were obtained
between injection temperatures of about 980 and 1140°C.
In comparison to these results. Fig. 6 shows the results of NOxOUT A+, which
includes a proprietary chemical modification formulated to reduce NH}, NjO, and CO
emissions while expanding the temperature range to lower temperatures. The maximum
NO, reduction was 81% at the optimum injection temperature of around 1100*0 and an
NSR of 1/1. However, NO, removals of greater than 70% were achieved at injection
temperatures ranging from 930 to 1110°C, about 50°C lower than with NOxOUT A.
NSR Variation. Results of varying the NSR near the optimum NO, removal injection
temperature for both NOxOUT A and A+ are shown in Fig. 7. Increases in the NSR (to
2/1) result in greater NO, removal to about 87%, although above an NSR of 1.5/1 (NO,
removal of «=80%), little additional NO, reduction is noted. For NSR values below
1.5/1, NO, reduction with NOxOUT A is about 10% (absolute) higher than that for
NOxOUT A+, while above an NSR of 1.5/1, little distinction in NO, reduction is
seen. NO, removal results at NSR = 1/1 are slightly lower than in Figs. 5 and 6,
likely due to injection at non-optimal temperatures and/or normal variation in
system performance.
N-O. NjO emission levels (Fig. 5) for NOxOUT A generally appear to follow NO,
removal levels; peak N^0 emission (90 ppcn) occurs at the same temperature as peak
NO, removal, about 1100°C. Peak N-O emissions using NOxOUT A+ (Fig. 6) appear to
occur about 50°C higher than the optimum injection temperature for NO, removal. For
NOxOUT A and A+, N-O emissions follow a similar temperature response, although
levels for the latter (peak value of 34 ppm) are consistently about one-third of
the former.
For tests conducted near the optimum injection temperature for NO, removal
(1087°C), increasing NSR values results in greater NjO emissions for both NOxOUT A
and A+ (Figs. 8 and 9, respectively). NjO concentration ranges from 29 to 91 ppm
for an NSR of 0.5/1 to 2/1, respectively, for NOxOUT A. NOxOUT A+, appears to be
5B-25
-------
less sensitive to NSR increases, ranging from 10 to 50 ppm for NSR values of 0.5/1
to 2/1. For both NOxOUT chemicals, N-O emissions are only slightly affected by
changing NSR values between 1/1 and 1.5/1.
NH,. NHj concentrations for NOxOOT A injection (Fig. 5) reach a maximum of 88 ppm
at 821°C. Increases in injection temperature show declining concentrations with
increases in injection temperature. Peak NB, levels of 83 ppm for NOxOUT A+ (Fig.
6) at 821°C are reduced below 5 ppm at injection temperatures of 887°C and higher.
Changes in NSR values affect NH, emissions, as seen in Figs. 8 and 9-
Increases in NSR for both NOxOUT A and A+ result in higher levels of NH3. As with
Vfi, NH, levels with NOxOUT A are only a weak function of NSR changes from 1/1 to
1.5/1.
DISCUSSION
Sulfur Dioxide Tests
Comparison. The SO; removals (*40 to 50%) reported in Fig. 2 for dry CaCO;
particles (all reported SO; values are at Ca/S = 2/1) somewhat exceed previous
results (**40%) for testing in this furnace [18] and others [19]. The SO; removal
results for dry Ca(OH); sorbent injection, *62%, are consistent with earlier
testing in this reactor [1,20] and numerous tests by others [21]. While it is
difficult to compare results between dissimilar furnaces, fuels, initial SO;
concentrations (SO^), and sorbents, the results for CaCO, slurry injection (*50 to
60% at Ca/S = 2/1) are consistent with results from Reference [22] of about 40 to
55% at Ca/S * 2/1 and four different coal/sorbent combinations. Later work [13]
indicates SO; removals with a Ca(OH)2 slurry (Ca/S * 2/1) of 78%, comparable to our
peak value of *74%.
Temperature. The results for both dry and slurry Ca(OH)2 injection (Fig. 3) are
similar to those found for dry and slurry CaCO, injection in that they are, with
one significant exception, relatively insensitive to temperature. While greater
sensitivity to injection temperature for dry sorbent injection may be observed in
other facilities (see, for example. References [19] and [21]), this phenomenon is a
strong function of reactor quench rate; the temperature response profile of SO?
capture becomes flatter for lower quench rates. The IFR has a fairly moderate
quench rate of about 250°C/s. Results from a pilot facility [22] operating at a
quench rate of 500°C/s did show greater temperature sensitivity of SO; capture with
slurry injection. As expected with this higher quench rate, the optimum slurry
injection temperature (*1200°C) was determined to be about 150°C higher than in our
work (=1050°C).
Dry Versus Slurry Injection. The equal or greater capture by CaCO, slurry versus
dry injection has been attributed to particle fragmentation or delayed sintering
[23]. However, the range of data on these tests is insufficient to be conclusive -
- certainly there is not a significant effect of slurry injection with CaCOj.
The levels of SO; removal from the upper 50% to about 70% (excluding the
NOxOUT addition results) are typiccl for dry Ca(OH): sorbents. Significantly
greater SO; removals (about 10%, absolute) with slurry versus dry injection result
at one temperature (1000°C). Unfortunately, further definition of this temperature
peak was impossible due to injection port limitations. The mechanism for this
enhanced removal during slurry (versus dry) injection remains speculative.
Effect of NOxOUT A. Tests with NOxOUT A added to the sorbent slurry show
significant improvement over the slurry alone or dry tests. Improvements in SO.
capture exceeding 10% absolute occur throughout the 880 to 1170°C injection range.
This phenomenon was also observed [6] when testing a hydrated lime-urea mixture and
comparing it with the hydrated lime alone. It was speculated that the enhancement
was due to either increased sorbent surface area and porosity from urea
decomposition in the sorbent crystal structure or reactions between SO; and urea
decomposition products in the sampling system. Our results suggest, however, that
the mechanism of enhancement of the sorbent's ability to capture SO, is likely the
reaction of the sorbent and urea-based compound with SO;. X-ray diffraction
results from IFR solid sampling during NOxOUT A injection indicate, along with the
5B-26
-------
expected CaSO*, the significant presence of (NH*) JZa.(S04)-• FkO (koktaite). It is
clear at these high temperatures, that CaO, SOj, and the urea breakdown product
(MB,) react together to increase SOn removals beyond that expected simply from the
presence of CaO [from Ca(OB)2 or CaCO,] alone.
Particle Size Effects. The effect of particle size for the CaO/SO, reaction has
been well documented [24,25]. Thus, the trends observed in Fig. 4 of increasing
reactivity with decreasing particle size were anticipated. The effect of sorbent
diameter is inversely related to particle size to the 0.14 power.
For CaCO, injection, the results for three particle size classes show that SO-
capture increases from dry to slurry to slurry with NOxOUT A injection. While the
magnitude of the increase during NOxOUT A injection is within the error range of
the furnace results, the consistency of this rank across the particle sizes implies
that more exhaustive testing might verify this enhancement for CaCO,, however
slight. Comparison with the Ca(OH)2 results indicates that equal capture of SO2 can
be expected for dry injection of CaCO, particles ground to similar sizes. An
explanation for the enhancement of SO2 capture with Ca(OH)2 is beyond the scope of
this effort; however, the higher initial porosity of Ca(OH)2 than CaCO, may allow
for greater infiltration of the slurry water into the particle matrix and, when
injected into the hot furnace, may allow for appreciable particle fracturing due to
water vaporization.
CafOHl-. Versus CaO Slurry. The inability to distinguish between the SOj reactivity
of the slurries from commercially available Ca(OH)2 versus laboratory-slaked CaO
suggests the simplicity of the hydration process towards production of reactive
sorbents. Purchase costs of hydration and transportion of the added weight of Hp
in Ca(OH)2 to the site can be avoided if Cao is mixed at the boiler site. While it
is likely that improved methods of CaO slaking will increase the sorbent
reactivity, our rudimentary methods of sorbent &laking were sufficient to match the
results of manufacturer-supplied Ca(0B)2.
Nitrogen Oxides Tests
Comparison. XFR test results show NO, removals (*75%) with NOxOUT A at 1000°C and
an NSR of 1/1 that are virtually identical to those demonstrated in Reference [13]
under similar conditions with injection of a urea-based solution. Other similar
results have been reported by References [11] and [12] with urea injection, given
consideration for different NSR and NO,, values.
NOxOUT A Versus NOxOUT A-*-. Use of NOxOUT A+ in this work improved the NO, removal
values at lower temperatures. Changing NSR values also yields NO, removal
responses similar to those reported by Reference [11]. Thus, NO, removals effected
by changes in both injection temperature and NSR are consistent with pilot and
field results, indicating the technical success of SNCR.
The ability of NOxOUT A+ to perform well at lower injection temperatures than
NOxOUT A raises the possibility of staged injection of NOxOUT A and A+ at high and
low temperatures, respectively. This has the additional benefit of reducing the
local "load" of the nitrogen reductant injected into the flue gases, thereby
minimizing potential NB, slip problems.
NO. Values of NjO production as a function of NO„ reduction (plotted as aNO/aNO,
in Fig. 10) for both the NOxOUT A and A-*- urea-based solutions were almost
exclusively less than those of References [11] and [26] with pure urea. Work
reported in Reference [11] was done on a pilot scale, natural-gas-fired combustor
(described more fully in Reference [12]) doped with NB, to produce NO,, and
Reference [12] used a pilot scale 2 MW(t) coal-fired circulating fluidized bed.
This suggests that technical improvements to the pure urea solution, represented
here by the NOxOUT A and A+- formulations, can have an impact upon N.O emissions in
SNCR processes.
NH.. Levels of NH, emissions, usually termed NH, "slip" in reference to the
unreacted N-based reductant, for both NOxOUT A and A+ show trends of reduction with
increases in temperature consistent with results of others [12]. For purposes of
comparison, Fig. 11 replots the NB, slip emissions during both NOxOUT A and A+
injection with those from Reference [12]. Despite an initial NO, level over twice
5B-27
-------
that of Reference [ 12 ], NH3 slip values In our work are significantly less
throughout the full temperature range. This may likely be due to differences in
the experimental combustors combined with the increased reactivity of the NOxOUT A+
formulation at lower temperatures.
CONCLUSIONS
This work has demonstrated on the pilot scale the successful coupling of Ca-
based sorbent injection and SNCR technologies in a slurry injection process. SO;
and NO„ removals of about 70 to 80% at Ca/S = 2/1 and an NSR = 1/1, respectively,
have been observed.
Different formulations of the tested modified urea solutions result in
varying sensitivity and effectiveness with temperature, acting to broaden the
applicable injection region of the combined process.
SO; emission control is enhanced by the combined technologies;
identification of NH,/Ca/S04 compounds suggest that the urea-based solutions react
with Ca and SOj to effect additional SOj removal. Some evidence exists for the
enhancement of SOj capture during slurry versus dry injection of sorbents, albeit
over & narrow temperature range.
Levels of NH3 and N4D are significantly reduced below levels previously
reported for urea injection through use of modified urea-based solutions. Near the
peak NO, removal levels for NOxODT A+ of 80% (NO,, = 600 ppm, NSR = 1/1), emission
levels of NH, and NJD were below 5 and 20 ppm, respectively.
ACKNOWLEDGEMENTS
The authors are grateful to George R. Gillis and Frank E. Briden, both of the
EPA's Air and Energy Engineering Research Laboratory for their mechanical and
analytical expertise, respectively, and to Charles B. Courtney of Acurex
Corporation for his invaluable facilities operation. Special thanks to Tenn
Luttrell Company for supplying the sorbents.
REFERENCES
1. R. Smock. "Utilities Struggle with Acid Rain Control Compliance Decisions."
Power Engineering. Vol. 95, No. 8, August 1991, 17-22.
2. T. Armor and G. Poe. "Coal-fired Power Plants in the U.S.: Future Trends." In
Proceedings: 3rd International Conference on Improved Coal-Fired Power
Plants. Electric Power Research Institute, April 1991.
3. P. S. Nolan, T. J. Purdon, H.E. Peruski, H. T. Santucci, M. J. DePero, R. V.
Hendriks, and D. G. Lachapelle. "Results of the EPA LIMB Demonstration at
Edgewater." In: Proceedings: 1990 SO. Control Symposium. Vol. 1, U.S. EPA,
EPA-600/9-91-015a (NTIS PB91-197210), May 1991, 3A-83 to -102.
4. J. P. Clark, M. R. Gogineni, R. W. Koucky, E. Gootzait, and D. G. Lachapelle.
"Status of the Tangentially Fired LIMB Demonstration Program at Yorktown Unit
No. 2." In: Proceedings: 1990 SO, Control Symposium. Vol. 1, U.S. EPA, EPA-
600/9-9l-015a (NTIS PB91-197210), May 1991, 3A-169 to -188.
5. J. E. Hofmann, J. von Bergmann, D. Bokenbrink, and K. Hein. "NO. Control in
a Brown Coal-Fired Utility Boiler." In Proceedings: 1989 Joint SvmpOBium on
Stationary Combustion NO- Control. U.S. EPA, EPA-600/9-89-062b (NTIS PB89-
220537), June 1989, 7A-53 to -66.
6. L. J. Muzio, R. M. Himes, and R. E. Thompson. Simultaneous NO./SO. Removal by i
the Dry Infection of Lime-Urea/Hydrate. U.S. Department of Energy
DOE/PC/88860-T1 (DE90009566), March 1990.
5B-28
-------
7. R- Mangal, M. Mozes, R. Gaikwad, R. Thampi, and D. MacDonald. "Additive
Injection for NO, Control." In proceedings: 1989 Intern. Svnvo. on Combustion
in Industrial Furnaces and Boilera. sponsored by the American Flame Research
Committee, Short Hills, NJ, September 25-27, 1989.
8. P. J. Crutzen and C. J. Howard. "The Effect of the HO, + NO Reaction Rate
Constant on the One Dimensional Model Calculations of Stratospheric Ozone
Perturbations." Pure Applied Geophysical. Vol. 116, 1978, pp. 497-510.
9. J. C. Kramlich, R. K. Lyon, and W. S. Lanier. EPA/NOAA/NASA/USDA M.O
Workshop. Volume I; Measurement Studies and Combustion Sources. U.S. EPA,
EPA—600/8-88-079 (NTIS PB88-214911), May 1988.
10. W. R. Epperly and J. E. Hofmann. "Control of Ammonia and Carbon Monoxide
Emissions in SNCR Technologies." In Proceedings: AIChE Summer National
Meeting. AIChE, Philadelphia, PA, August 1989.
11. L. J. Muzio, T. A. Montgomery, C. C. Quartucy, J. A. Cole, and J. C.
Kramlich. "N^O Formation in Selective Non-catalytic NO, Reduction Processes."
In Proceedings: 1991 Joint Symposium on Stationary Combustion NO, Control —
U.S. EPA/EPRI. Vol. 2, U.S. EPA, Research Triangle Park, NC, March 25-28,
1991.
12. D. P. Teixeira, L. J. Muzio, T. A. Montgomery, G. C. Quartucy, and T. D.
Martz. "Widening the Urea Temperature Window." In Proceedings: 1991 Joint
Symposium on Stationary Combustion NO. Control — U.S. EPA/EPRI. Vol. 2,
U.S. EPA, Research Triangle Park, NC, March 25-28, 1991.
13. R. Mangal, M. S. Mozes, P. L. Feldman, and K. S. Kumar- "Ontario Hydro's
Sonox Process for Controlling Acid Gas Emissions." In Proceedings: 1991 Joint
Symposium on Stationary Combustion NO. Control — U.S. EPA/EPRI. Vol. 2,
U.S. EPA, Research Triangle Park, NC, March 25-28, 1991.
14. J. S. Ford. Recommended Operating Proce-' re No. 45: Analysis of Nitrous
Oxide from Combustion Sources. U.S. EPA, "\-600/8-90-053 (NTIS PB90-238502),
June 1990.
15. F. E. Briden, D. F. Natschke, and R. B. Snoddy. "The Practical Application of
Tunable Diode Laser Infrared Spectroscopy to the Monitoring of Nitrous Oxide
and Other Combustion Process Stream Gases." In Proceedings: 1991 Joint
Symposium on Stationary Combustion NO, Control — EPA/EPRI. Vol. 2, U.S. EPA,
Research Triangle Park, Much 25-28, 1991.
16. R. A. Montgomery, G. S. Samuelsen, and L. J. Muzio. "Continuous Infrared
Analysis of N;Q in Combustion Products." JAPCA, Vol. 39, 1989, 721-726.
17. W. F. Michels, G. Gnaedig, and J. R. Comparato. "The Application of
Computational Fluid Dynamics in the NOxOUT Process for Reducing NOx Emissions
from Stationary Combustion Sources." In Proceedings: American Flame Research
Committee International Symposium. San Francisco, CA, October 8-10, 1990.
18. G. C. Snow and J. M. Lorrain. Evaluation of Innovative Combustion Technology
for Simultaneous Control of SO, and NO.. U.S. EPA, EPA-600/2-87-032 (NTIS
PB87-188926), April 1987.
19. G. H. Newton, D. K. Moyeda, G. Kindt, J. M. McCarthy, S. L. Chen, J. A. Cole,
and J. C. Kramlich. Fundamental Studies of Dry Injection of Calcium-Based
Sorbents for SO, Control in Utility Boilers. U.S. EPA, EPA-600/2-88-069
(NTIS PB89-134142), December 1988.
20. B. K. Gullett and F. E. Briden. Furnace Sorbent Reactivity Testing for
Control of SO* Emissions from Illinois Coals. U.S. EPA, EPA-600/2-89-065
(NTIS PB90-150830), December 1989.
21. B. J. Overmoe, J. M. McCarthy, S. L. Chen, w. R. seeker, and D. W. Pershing.
Reactivity Study of SO. Control with Atmospheric and Pressure Hvdrated
Sorbents. U.S. EPA, EPA-600/7-86-040 (NTIS PB87-129250), November 1986.
5B-29
7
-------
22. R. Manga1, H. Mozes, R. Thampi, and D. MacDonald. "In-Furnace Sorbent Slurry
Injection for SCk Control." In Proceedings: Sixth Annual Int. Pittsburgh
Coal Conference. Univ. of Pittsburgh, School of Engineering, Pittsburgh, PA,
September 25-29, 1989.
23. T. L. Torbov, G. R- Offen, and S. K. Demike, Method for Reduction of Sulfur
Products in the Exhaust Gases of a Combustion Chamber. D.S. Patent Number
4,555,996, issued Dec. 3, 1985.
24. B. K. Gullett and J. A. Blom. "Calcium Hydroxide and Calcium Carbonate
Particle Size Effects on Reactivity with Sulfur Dioxide." Reactivity of
Solids. Vol. 3, 1987, 337-350.
25. C. R. Milne and D. W. Pershing. "Time Resolved Sulfation Rate Measurements
for Sized Sorbents." In Proceedings: Fourth Annual Pittsburgh Coal
Conference. Univ. of Pittsburgh, School of Engineering, Pittsburgh, PA,
September 28 to October 2, 1987, 109-118.
26. T. Hulgaard. Nitrous Oxide Emissions from Danish Power Plants. CHEC Report
No. 9002, Department of Chemical Engineering, Technical University of
Denmark, DK-2800 Lyngby, Denmark, February 1990.
5B-30
-------
Combustion Feed Air
Natural Gas
CO
t:
s.
o
o
0
01
2nd ng
3rd right
4th upper
4th lower
5th
Coal Feeder
Solid Sampling
Port
S02 Sampling
Port ^
CEM and GC
Sampling Port
W
NH3 Sampling
Port
N20 Sampling
Port
Baghouse
Figure 1. Schematic of the Innovative Furnace Reactor.
5B-31
-------
l^I^^j^^j^n^^Slur^^^ection^^Slur^wTtt^jpxOUTA
800
1300
900 1000 1100 1200
Injection Temperature (°C)
Figure 2. Effect of varying injection temperature on SO 2
removal for CaC03.
100
90
<0
I 80
o
CC
CM
O
to 70
60
50
Dry Injection
Slurry Injection
Slurry with NOxOUT A
¦ •
Slaked CaO Slaked CaO with NOxOUT A
O ~
A
y / \
800 900 1000 1100 1200
Injection Temperature (°C)
100
90
80
70
60
50
1300
Figure 3. Effect of varying injection temperature on SO 2
removal for Ca(OH) .
5B-32
-------
90
90
CaCO, (slurry)
80
80
CaC03 (slurry with NOxOUT A)
70
70
Ca(OH) (slurry)
Ca(OH) (slurry with NOxOUT A)
CC 60
cT
CO
60
50
50
40
40
30
30
0
10
20
30
40
Particle Size(^m)
Figure 4. Effect of sorbent parti rJe size on S02 removal.
5B-33
-------
100
100
CO NH NO NO Removal SO Romoval
* 2 x a
to
>
o
E
-------
100
100
NOxOUTA NOxOUT A +
80
80
> 60
O
£
©
cc
x
60
40
40
20
20
0
0.5
1
1.5
2
2.5
3
NSR
Figure 7. Effect of NSR on NO xremoval for NOxOUT A and NOxOUT A+.
5B-35
-------
200
100
j CO NH, N,0 NO.Removal^
O 150
~ 100
cs
>
o
E
©
DC
x
o
z
cS
Figure 8. Effect of NSR on emissions using NOxOUT A.
200
E
o.
Q.
J 150
«-»
u
C
•C 100
3
©
>
©
CO NH, NaO NO,Removal
c
o
co
JO
E
LU
50
moval I
•
/
/
—I I ^0*
#¦
^
0.5
1.5
NSR
2.5
100
80
CO
>
60 |
©
oc
X
O
40 2
- 20
Figure 9. Effect of NSR on emissions using NOxOUT A +.
5B-36
-------
NOxOUT A NOxOUT A+
NQ. -600 pgn. NSR-1/1 NQ -600 pgn, NSH-1/1
Urea (Muzio, oral., 1991) Urea (Hulgaard. 1990)
NQ -700 pom. NSR=1/1 N^-400 ppm, NSR-1/1.3
0.4
O 0.3
0.2
0.1
800
1000
1100
900
1200
Injection Temperature(°C)
Rgure 10. N20 production from NOx at varying injection temperatures.
200
200
NOxOUT A
NQ.-600 ppm NC^ —600 ppm
NOxOUT A+ UreaCTeixeira et al., 1991)
N£(i=2^0ppm
150
100
100
LU
- 50
1200
1100
800
1000
Injection Temperature(°C)
Rgure 11. Effect of injection temperature on ammonia emissions.
5B-37
-------
Intentionally Blank Page
5B-38
-------
Combined SO2/NOX Abatement by Sodium Bicarbonate
Dry Injection
5B-39
-------
Intentionally Blank Page
5B-40
-------
J. Verlaeten
Solvay
1050 Brussels, Belgium
G.G. De Soete
Institut Frail(yais du P6trole
92506 Rueil-Malmaison, France
L. Ninane & J.M. Blondel
Solvay
54110 Dombasle, France
ABSTRACT
Following laboratory tests at the CER, SOLVAY's Centre d'Etudes
et de Recherches in Dombasle (France), which confirmed
suitability of Sodium Bicarbonate for the simultaneous
elimination of S02 and NOx present in flue gas, SOLVAY has
started demonstration trials in actual power plants.
Laboratory tests have shown that S02 abatement with Sodium
Bicarbonate also entails a significant decrease of NOx in
certain cases, depending on the ratio of S02 to NOx in the flue
gas, the temperature, the granulometry of Bicarbonate, the
residence time, etc...
The first confirmation trial was performed in Rosignano, Italy,
on flue gas generated by a 20 MW steam generator fed with a
3 % S fuel. 1750 Nm3/hr of flue gas containing 4500 mg SOz and
300 mg NOx/No3 were treated at 116-124*C with dry injection of
ground Bicarbonate (mean particle size 25
5B-41
-------
through a bag house filter. At stoichiometric Bicarbonate
addition the resulting abatement was 60 % S02 and 90 % NOx.
The second confirmation trial was performed at Geilenkirchen,
Germany, on the flue gas of a 9.7 MM steam generator fed with
1 % sulfur coal. 11.000 Nm3/h flue gas containing 1000 mg SO2
and 200 mg NOx/Nm3 were treated at 110-120"c with dry injection
of coarse or ground Bicarbonate (120, resp. 7.5 Jtfm mean
particle size) before passing through a bag house filter with em
excess of 20 % Bicarbonate. In the case of the ground reagent
the abatement was 98 % S02 and 64 % NOx, and in the case of the
coarse material 42 % S02 and 19 % NOx.
Consequently of these good results, it was decided to study the
mechanism of the reactions during the simultaneous abatement of
SOz and NOx. SOLVAY asked IFP, "Institut Fran^ais du P6trolen,
to do this study in its specially equiped laboratory.
The IFP laboratory study gives confirmation of the ability of
Sodium Bicarbonate to abate simultaneously so2 and NOx in flue
gas. Moreover the principal reaction mechanism of the abatement
seems to be as follows :
- Sodium bicarbonate sulfitation :
NaHC03 + S02 > NaHS03 + C02
- Sodium bisulfite dehydratation :
2 NaHS03 ~ Na2S205 + H20
- Sodium pyrosulfite nitration :
Na2S2Os + 2 NO + Oz >. NaN02 + NaN03 + 2S02
- Sodium bisulfite nitration :
2 NaHS03 + 2 NO + 02 > NaN02 + NaN03 + 2 S02 + H20
All these reactions begin to become important in the temperature
range 370 to 450 K.
5B-42
-------
INTRODUCTION
Earlv Informations
In recent years, international pressures to decrease NOx
emissions are becoming as strong as those regarding S02- Till
now the technology of choice to achieve this requirement is SCR,
selective catalytic reduction.
Due to the high investment and operation costs of the SCR
process, the interest in the development of lower-cost NOX/SO2
abatement processes has been renewed (1,2).
Now, it is well established that Sodium Bicarbonate is a very
effective reagent for S02 removal in flue gas by a low
investment cost dry injection process (3,4,5). More than 15
years ago, during trials at the Mercer station of New Jersey's
Public Service Electric and Gas Co., it was observed that NOx
removal of up to 40 % occured during the S02 abatement by Sodium
Bicarbonate dry injection (7).
Solvay was interested to confirm the ability of Sodium
Bicarbonate for the simultaneous S02 and NOx abatement by dry
injection in the flue gas and to determine the best conditions
of operation range.
5B-43
-------
Laboratory Trials at CER-Dombasle
The Solvay "Centre d'Etudes et de Recherches" (CER) in Dombasle
(France) made some preliminary laboratory trials to verify the
simultaneous S02/NOx abatement in a synthetic flue gas. The
purification of this gas composed of air, S02 and NO was
obtained by passing it through a little Sodium Bicarbonate
fluidized bed.
The conclusions of these trials were :
- in all cases, high SO2 abatement (about 100 %) ;
- NO abatement increased from 35 to 50 % with the ratio S02/N0
in the inlet gas (0 to 3 ppm S02/ppm NO) ;
- NO abatement increased from 45 to 85 % with the residence time
of the gas in the fluidized bed (0,5 to 2,5 s) ;
- very low influence of the temperature on the NO abatement in
the range 90 to 200"c.
The NO abatement was about 30 to 40 % in this temperature range
for 0,5 s residence time and 2 ppm S02/ppm NO in the inlet gas.
After this good results, it was decided to carry on the study
doing confirmation trials in actual power plants.
CONFIRMATION TRIALS IN ACTUAL PLANTS
CER Pilot mobile Equipment
To confirm the ability of Sodium Bicarbonate for S02 abatement,
the Solvay CER of Dombasle has constructed a pilot mobile
equipment.
5B-44
-------
This device includes (see figures 1 and 2) :
. a Bicarbonate hopper filled with 50 kg bags,
. a screw conveyor feeder,
. a Bicarbonate screw flowmeter,
. a pin mill Alpine 315 UPZ (9000 revs/min, 22 KW) (figure 3).
The CER pilot equipment injects crushed Bicarbonate (mean
particle size smaller than 10 J^m) at the rate of 50 to 250 kg/h
in flue gas stream. It has been used for demonstration trials :
- on a municipal waste combustor in Antwerp (Belgium) for HC1
and S02 abatement (test period : May 30 to June 3, 1988) (3) ;
- on a power plant in Heilbronn (Germany),
(test period : September 6 to October 27, 1988) (4) ;
- on a thermal power station in Gardanne (France),
(test period : February 27 to March 3, 1989) (5) ;
- on a municipal waste combustor in Padua (Italy) for HC1
abatement,
(test period : May 21 to 24, 1990) (6).
It was decided to utilize the CER pilot equipment to confirm the
ability of Sodium Bicarbonate for simultaneous SO2/NOX abatement
in flue gas of an actual thermal power plant.
Demonstration Trial at Geilenkirchen
A demonstration trial of S02/N0x simultaneous abatement has been
done with the CER pilot mobile equipment on the steam generating
station flue gas of the NATO air base in Geilenkirchen.
This trial was jointly undertaken with Wulff GmbH which has
manufactured the flue gas treatment equipment designed with
hydrated lime injection. During these tests, the lime injection
was replaced by Sodium Bicarbonate injection.
5B-45
-------
The amount of SO2 and NOx contained in the flue gas was measured
both before and after purification by the CER and, at the same
time, monitored by Wulff.
The Sodium Bicarbonate used i*? of hydrophobic Venale "Fein"
quality (a treatment which makes Sodium Bicarbonate free
flowing), provided by Rheinberg Solvay plant (Germany).
The effect of following parameters on the S02 and NOx abatement
effectiveness has been studied :
- size of the Bicarbonate (injected without or after crushing),
- quantity of Bicarbonate injected,
- influence of recycling abatement residues *-iv\ch still contain
reagent.
Venale Fein Bicarbonate has a mean diameter of 120 ; after
crushing, this diameter is reduced to 7.6 JJlm.
The steam-generating station at Geilerikirchen is equiped with
two boilers. The tests were carried out downstream one of the
boilers, the characteristics of which were as follows :
- thermal power : 9.7 MW,
- vapour production at 10 bars : 14 metric tonnes/hour,
- fuel : Ruhr coal with 1 % S ; PCI = 7500 kcal/kg,
- flue gas flow rate : 11,000 Nm3 dry/h (+/- 15 %).
Details of the purification equipment installed by Wulff are
given below (figure 4). The flue gas leaving the boiler may be
sent directly to the chimney using a bypass. After cooling down
to about 130°C in a heat exchanger, it passes through a
cylindrical reactor, into the base of which the Sodium
Bicarbonate has been injected. The Bicarbonate comes out of the"
silo and flows into the CER mobile crushing device.
The flue gas passed through a bag house filter, at the bottom of
which a certain quantity of solid residues remains in a fluid
5B-46
-------
state (a quantity which amounts to around 5 to 6 tonnes) with a
view to possible recycling so as to consume the reagent. The
residence tine between the point of injection of the Bicarbonate
in the reactor and the inlet of the bag house filter is
approximately 3 to 4 seconds.
The S02 and NOx content was measured by bubbling a part of the
flue gas through H202 and potassium dichromate solutions. The
analyses were carried out using a chromatograph with DIONEX QIC
anions. The flow rate of the flue gas and their Oz content were
also measured.
At the injection point of the Bicarbonate, the average content
of S02 and NOx respectively reaches 1650 and 325 mg/Nm3 dry with
7 % 02. The average temperature of the flue gas during
purification was between 112 and 122*C in the case of uncrushed
Bicarbonate and 110*C in the case of crushed Bicarbonate. The
quantity of reagent injected in relation to the stoichiometry of
the simultaneous purification reactions of S02 and NOx varied
between 1.54 and 2.28 for the uncrushed Bicarbonate and between
0.93 and 1.68 for crushed Bicarbonate.
Figure 6 shows the results obtained during purification tests.
The rate at which impurities were removed are given in relation
to the quantities of Bicarbonate injected, indicated on figure 7
for S02 and on figure 8 for NOx respectively.
The results obtained through the injection of crushed
Bicarbonate were highly satisfactory : 95 % of the S02 was
removed using a stoichiometric quantity of Bicarbonate whilst
60 % of the NOx was eliminated through the injection of a
quantity of Bicarbonate in excess of 45 % of the stoichiometry.
The injection of uncrushed Bicarbonate did not achieve such good
results.
5B-47
-------
However, during the course of two tests carried out by injecting
uncrushed Bicarbonate and recycling the residues separated in
the bag house filter, similar results to those obtained when
using crushed Bicarbonate were achieved.
Confirmation Trial at Rosicmano
Another S02/N0x simultaneous abatement confirmation trial was
made during the period October 2 to November 9, 1989 on the flue
gas of a steam generator at the Solvay Rosignano plant (Italy).
This trial was jointly undertaken with Termomeccanica, an
Italian equipment company.
The crushed Sodium Bicarbonate dry injection was made with a
pilot equipment manufactured by Termomeccanica with a subsidy of
ENEA (Ente Nazionale Energie Alternative). The pilot equipment
is composed of :
. a tubular reactor (length : 28,5 m, 14" diameter),
. a flue gas fan, flowrate : 3000 Nm3/h,
. a bag house filter, 72 m2,
. a pin mill Danioni in order to crush the Sodium Bicarbonate.
The trial was made on a by-pass of the flue gas of a 20 MW steam
generator fed with a 3 % sulfur fuel. A flowrate of 1750 Nm3/hr
of flue gas containing 4500 mg SC2 and 300 mg NOx/Nm3 were
treated at 116-124°C.
The measurement of S02 and NOx content in the flue gas was made
by bubbling a gas sample respectively in a H202 solution and in
a sulfochromic solution_
Figure 6 gives the results obtained during the purification
tests. The rate at which impurities were removed in relation to
5B-48
-------
the amount of Bicarbonate injected are given on figure 7 for S02
arid on figure 8 for NOx.
These results show that :
. the percentage of S02 removal is comprised between the results
obtained at GeilenKirchen. In fact, it depends of the size of
the Bicarbonate injected ;
. the NOx removal is very high. This very good result is
probably due to the high S02/N0x inlet ratio.
STUDY OF SO2/NOx ABATEMENT REACTION MECHANISM
IFP Laboratory Equipment
The "Institut Francais du P6trole", IFP, in Rueil Malmaison,
near Paris (France) has a specially equiped laboratory able to
study the SOz/NOx abatement reaction mechanism with Sodium
Bicarbonate dry injection. It was agreed between Solvay and IFP
to do this study during the beginning of this year 1991.
IFP Laboratory is equiped with a little fluidized bed reactor
(figure 9), inside diameter 22 mm, fed with gas at the rate of
40 cm3/s. The bed is filled with 6 g of solid matter. It is
heated in an electric furnace with temperature programme. The
gas injected is a mixture of argon with S02, 02, NOz, NO and N2o
as the case may be.
The gas composition at the outlet of the bed is determined by
continuous measurements and the data are stored in a PC 386.
5B-49
-------
Results of IFP study
IFP made two kinds of trials :
- trials with increased temperature : usually increasing 2 to
6 K/min ;
- trials with constant temperature, eventually putting the solid
matter in the fluidized bed reactor after reaching the
temperature chosen for the trial.
It appears that Sodium Bicarbonate gives rise to NO abatement in
the temperature range of 400 to 500 K (figure 11). On the
contrary Sodium Carbonate has no effect on NO abatement (figure
10).
After this, IFP made trials with different compounds apt to
occur during the S02 abatement with Sodium Bicarbonate dry
injection.
Whereas Na2S03, produced by sulfitation of Na2C03, has no action
on NO abatement (figure 12), NaHS03, produced by the sulfitation
of NaHC03 and subsequently transformed into Na2S205 f induces a
considerable NO abatement in the temperature range 400 to 550 K
(figure 13).
A comprehensive view of the whole set of trials has led IFP to
propose the SOz/NOx abatement mechanism described on figure 15.
The principal reactions leading to the S02/N0x simultaneous
abatement seem to be :
- Sodium bicarbonate sulfitation :
NaHC03 + SOz » NaHS03 + COz
- Sodium bisulfite dehydratation :
2 NaHS03 } Na2S2Og + H20
- Sodium pyrosulfite nitration :
Na2s205 + 2 NO + 02 *NaNOz + NaNOs + 2 SOz
5B-50
-------
- Sodium bisulfite nitration :
2 NaHSQ3 + 2 NO + 02 ?.NaNQ2 + NaNQ3 + 2 S02 + H20
Finally, the maximum value of NOx abatement measured during the
IFP study is given on figure 14 as a function of the ratio of
NOx/S02 content in the inlet gas. The NOx abatement achieved at
Rosignano and at Geilenkirchen during the industrial trials are
also given on figure 14.
SUMMARY
Solvay trials in actual power plants at Geilenkirchen (Germany)
and at Rosignano (Italy) have once more demonstrated the ability
of sodium Bicarbonate dry injection for S02/N0x simultaneous
abatement in flue gas. These trials have shown the necessary
conditions to reach a high yield of NOx abatement :
. high ratio of S02/N0x content in the flue gas
. suitable temperature range
. utilization of a bag house filter ensuring the S02/N0x
abatement continuation during the flue gas flow through the
solid deposit on the sleeves.
For S02 abatement only, it is known that the investment cost of
the Sodium Bicarbonate dry injection is about the half of the
investment cost of a spray dryer system (bag house filter
included). But the price of the reagent handicapes the economy
of the process (8).
At the contrary, for S02 and NOx abatement the economy of the
Sodium Bicarbonate dry injection seems to be very competitive
comparatively with wet FGD for S02 abatement + SCR for NOx
abatement, particulary for medium capacity power plants (9).
5B-51
-------
The IFP laboratory Study has shown the principal reaction
mechanism occuring during S02/N0x abatement by Sodium
Bicarbonate dry injection.
Particulary, it appears from this study that an high HOx
abatement occures by contact of the flue gas with Sodium
Pyrosulfite and Sodium Bisulfite formed by the SOz abatement.
Sodium Pyrosulfite and Sodium Bisulfite can also be produced
industrially for this application.
5B-52
-------
REFERENCES
1. Proceedings : 1990 SOo Control Symposium - Session 6 B
combined SOx/NOx Technologies. New Orleans, Louisiana,
Electric Power Research Institute, Hay 1990.
2. Earl D. Oliver - NOx Removal. SRI International Report No
200. Menlo Park, California, May 1989.
3. Sodium Bicarbonate - Purification of Flue Gases - Results of
the Purification of Flue Gases from the Edegem CAntwerp1
Refuse Incineration Plant - Test Period : May 30 to June 3.
1988 - Solvay Leaflet Tr 895/2C-B-1-0989, Brussels, Belgium.
4. Sodium Bicarbonate - Purification of Flue Gases - Results of
Trials on the Purification of Flue Gases at the Solvav
Factory Power Station at Heilbronn (West Germany1 - Test
Period : Seot^h«>r e> to October 27. 1988 - Solvay Leaflet
Tr. 895/3c-B—1-0989, Brussels, Belgium.
5. Sodium Bicarbonate - Purification of Flue Gases - Results of
Trials on the Purification of Flue Gases from the Citv of
Gardanne Power Station fBouches du Rhone - France carried
out in collaboration with the Firm Svprim Air Industrie
Environment - Test Period ? February 27 to March 3, 1989 -
Solvay Leaflet Tr. 895/4C-B-1-0989, Brussels, Belgium.
6. Sodium Bicarbonate - Purification of Flue Gases - Results of
Tests concerning the Removal of HC1 from Flue Gases emitted
by the Padua Refuse Incineration Plant fItaly) - The Tests
were carried out from 21st to 24th May 1990 - Solvay Leaflet
Tr. 895/5c—B—1—1290, Brussels, Belgium.
7. NOx Removal as a By-product of SO2 control - EPRI Journal -
November 1986, pp. 44 and 45.
8. Sodium Bicarbonate - Purification of Flue Gases - The
Results of Purification Tests simultaneously carried out on
SOn and NOx on the Fumes produced bv the Steam generating
Station at the NATO Air Base in GeilenKirchen fGermany1 - The
Tests were carried out in collaboration with the Firm Wulff
from November 12 to 23. 1990 - Solvay Leaflet Tr. 895/6c-B-l-
0791, Brussels, Belgium.
9. Economic Evaluation of Drv-Iniection Flue Gas
Desulfurization Technology. CS-4373, Final Report, Electric
Power Research Institute, Palo Alto, California, January
1986.
10. Ralf L. Lindbauer and Franz Mair. The 3S-Process :
Simultaneous fAcid Gas and NOxt Sodium-based sorption of Flue
Gases. in Proceedings of the First Combined FGD and Dry S02
Control Symposium, October 25-28, 1988, St-Louis, Missouri.
5B-53
-------
1
- SOLVAY Pilot Mobile Equipment
General View
2 - SOLVAY Pilot Mobile Equipment
Details
5B-54
-------
Fig. 3 - Pin Mill ALPINE 315 UPZ
Fig. 4 - Flue Gas Purification Equipment
at Geilenkirchen
5B-55
-------
Trial
39
40
41
42-1
42-2
43-1
43-2
44
Flue g88 floHrate. Nm3/h dry
1600
1600
1600
1750
1750
1750
1750
1750
S02 Inlet, mg/Nm3 at 7 X 02
4730
4610
4760
4730
4750
4660
4B10
4810
NOx Inlet. mg/Nm3 at 7 X 02
196
200
230
163
300
300
352
394
Temperature Inlet, °C
178
178
178
131
132
135
138
192
02 inlet, vol. X
3.2
3.2
3.2
3.9
4.0
4, 1
4.2
4,0
S02 outlet, mg/Nm3 at 7 X 02
2365
19B0
1620
2030
2000
1540
1600
1870
NOx outlet, mg/NmS at 7 X 02
45
18
0
18
75
75
95
71
Temperature outlet, °C
122
122
123
96
98
101
102
129
02 outlet, vol. X
5. 2
5.2
5. 2
5.9
5,9
6,0
6. 1
5.9
Sodium bicerbonate :
- f.lowratB. kg/h
- mean diameter, fi m
- 90 X emaller than, ^ m
17.3
24
83
24.0
24
83
29,6
24
83
17.7
21
67
17,6
21
67
23.9
21
67
23.9
21
67
23,6
16
44
NSR (SOS + NOx]
0. 84
1,20
1. 44
0.81
0, 80
1. 12
1,09
1,04
SOS removed, X
50
57
66
57
58
67
66
61
NOx removed, X
77
91
100
89
75
75
73
82
Fig. 5 - SO2/NOX Removal - Results of the Trials at Rosignano
-------
Ul
f
Ul
V]
Trial
4
5
6
7
12
13
15
17
Flue qbb flowrate, Nm3/h dry
10100
9650
10100
10100
10600
12000
10200
11000
S02 inlet, mg/NmS at 7 X 02
1650
1570
1569
1552
1624
1573
1926
1393
NOx Inlet, mg/Nm3 at 7 X 02
306
300
254
310
353
269
339
356
Temperature Inlet, °C
125
125
116
115
120
120
121
120
02 inlet, vol. X
11.2
12,0
11,9
11,0
11,0
10,3
11,0
10.5
S02 outlet, mg/NmS at 7 X 02
835
826
979
040
37
43
14
37
NOx outlet, mg/Nm3 ut 7 X 02
190
250
211
257
140
105
115
219
Temperature outlet, °C
119
119
110
115
105
105
104
105
02 outlet, vol. X
12,2
12,0
11,0
12.5
13,0
13, 1
11,5
11,6
Sodium bicarbonate :
- flowrate, kg/h
- mean diameter, /t m
- 90 X emaller than,^ m
74,2
121
195
48,9
121
195
40. 9
121
195
74.2
121
195
60,0
7.6
13,0
62,0
7,6
13. 8
50,0
7.6
13.0
36,0
7.6
13,0
NSR (S02 + NOx)
2,01
1.59
1,54
2. 28
1. 60
1.41
1,24
0, 93
S02 removed, X
49, 7
47.4
37,6
45. 9
97,7
97,3
99.3
97.5
NOx removed, X
37,0
16.5
16,9
17, 1
60,4
61,0
66. 1
38,4
Fig. 6 - S02/N0x Removal - Results of the Trials at Geilenkirchen
-------
BICARBONATE
dfi&Jlm
dM. <"m
+
*
ROSJQNANO
24
21
8
D
10
44
•
A
GELEMtRCHEN
•
%
'ffi.
Ol
f
cn
00
*
i
=
...M' ;
//A,'
m ^ :
1 2
N8R (802 ~ NOx)
i
k
r *
0
"9
A
A
A
1
»
•
•
•
1
N8R (802 * NOx)
Fig. 7 - S02 Removal as a Function of the
Amount of Bicarbonate Injected
Fig. 8 - NOx Removal as a Function of the
Amount of Bicarbonate Injected
-------
Q: Qu Flowrate : " 40 om3/a
N20
—i
R : Quartz Reactor 0 ¦ 22
F : Eteotrks Furnace
TP
i. N02
T : Thermocouple CrNI/NI
OA : Data Acquisition
S02
—l
TP: Temperature Program
FB : Ftuklzed Bod : 6 g of
aolkl matter
02
302/Ar
02/Ar
002
DA
NO/Ar
00
Fig. 9 - Fluidized Bed Reactor for Laboratory Trials
\
\
-------
U1
t
I
BOO
700
eoo
000
400
soo
flOO
100
NO ppm
****** N02 ppm
mtn N20 ppm (x 60)
mm 02 ppm
8.4
¦ 2,as
.2.8
2.2
.2.15
BOO-
700-
aoo-
BOO-
s»
400-
# 1
800-
200-
100-
o-
800
NO ppm
MM** N02 ppm
N20 ppm (X 00)
~~~~~~ 02 ppm
2.B
-2.4B
- 2,4
-2.SB
- 2,8
- 2.2B
8
Id
¦ 2,2
•2.1B
¦ 2,1
- 2.OB
- 2,0
Fig. 10 - 420 ppm NO + 675 ppm S02 + 2.3 % 02
In Ar ON Na2C03 (Trial 316)
Fig. 11 - 412 ppm NO + 675 S02 + 2.3 % 02
In AR ON NaHC03 (Trial 293)
-------
800
700
800-
jj 600
I*0°*
800
200
100
0
¦¦¦NO ppm
MMM N02 ppm
N20 ppm (X 80)
)'6*-iAr ~ ih #i ^ i* A b jt
800 400 500 600 700
000
T (K)
Fig. 12 - 393 ppm NO + 0.29 % 02 In Ar ON
Na2S03 (Trial 469)
NO ppm
****** N02 ppm
N20 ppm (x 20)
~~~~~» 02 ppm
IMMIM
8
*
800 400 6
600 700 BOO BOO 1000
T (K)
Fig. 13 - 410 ppm NO + 2.3 % 02 In Ar ON
Na2S205 (Trial 34 6)
-------
Roslonano
ln(N0X/S02)ln
/
Fig. 14 - NO + SO2 + O2 In Ar ON NAHCO3 - Maximal Value of NOX
AKafamAnf
-------
~ N02
130 °C
~ NO ~ Q2
125 °C
110 «C
NaH
COS
+ 802
100
N«HS03
*°t>
115 °Ct
N0H8O4
N*28205
100 °C
Nt2003
~ NO + 02
110 °C
G
N>28207
+ 02
116 «C
VV
I
I
Nt2803
~ 02^
Na2804
NaN02
~ 802
jLU
I
NaNOS
140 "C
Fig. 15 - Mechanism of SO2/NOX Removal by Sodium Bicarbonate
Injection
-------
Intentionally Blank Page
-------
2 AND N0X CONTROL BY COMBINED DRY INJECTION
OF HYDRATED LIME AND SODIUM BICARBONATE
Dennis Helfritch & Steven Bortz
Research-Cottre 11
PO Box 1500
Somerville, NJ 08876
Roderick Beittel
Riley Stoker Corp.
45 McKeon Road
Worcester, MA 07610
Perry Bergman
U. S. Department of Energy
Pittsburgh Energy Technology Center
PO Box 10940
Pittsburgh, PA 15236
Barbara Toole-O'Neil
Electric Power Research Institute
3412 Hillview Ave.
Palo Alto, CA 94303
5B-65
-------
Intentionally Blank Page
5B-66
-------
ABSTRACT
The dry sorbent injection process for S02 and NOx removal from coal-
fired boiler flue gas consists of "the use of low NOx burner -technology
for primary NOx reduction, injection of hydrated lime at economizer
temperatures for primary capture of S02, and injection of sodium
bicarbonate at the air heater exit for additional SOz and NOx removal.
This concept has been separately tested at the .25 and 50 million
Btu/hour scales, utilizing test systems that duplicate the flue gas
time-temperature profile found in commercial boiler systems. The
testing procedures and results, including the effects of the sorbent
injection on particle control devices, are described in this paper.
5B-67
-------
INTRODUCTION
The team of Research-Cottrell Environmental Services and Technologies
and Riley Stoker is conducting a proof of concept demonstration of an
Integrated Dry Injection process for coal-fired boiler S02 and NOx
control, under a U.S. Department of Energy (DOE) contract with co-
funding by the Electric Power Research Institute (EPRI). The process
consists of combustion modification using low N0X burners to reduce NOx
emissions, dry injection of hydrated lime at the economizer for primary
capture of S02, dry injection of a commercial grade sodium bicarbonate
at the air heater exit for additional S02 and NOx removal, and flue gas
humidification for precipitator conditioning. This concept is
illustrated in Figure 1. The Integrated Dry Injection Process offers
the potential for simultaneously achieving 90+% SO, removal and 75+% NOx
removal from flue gas. The process is well suited for new or retrofit
applications since it can be incorporated within existing economizer
and downstream ductwork. In addition, capital costs are kept to a
minimum since no large system components such as catalytic beds, spray
dryers, or scrubbers are required.
The SOz and NOx removal technologies, which are combined in this
demonstration test, have been independently evaluated but have not yet
been tested as an integrated system. The integrated tests are
important to determine and characterize any interactions between the
technologies, either positive or negative. Some conditions that favor
S02 removal inhibit NO^ removal. For example, high levels of SOz
removal by economizer injection of calcium hydroxide will adversely
affect NOx removal by sodium bicarbonate, which depends on high S02
concentration.
SORBENT EVALUATION
To identify the best calcium and sodium sorbents to use for the proof-
of-concept demonstration, subscale tests were performed that involved
the injection of calcium hydroxide and sodium sorbents at various
points of the flue gas system downstream of a 0.25X106 Btu/hr coal
fired combustor. The subscale system is shown in Figure 2. The flue
gas flow from the furnace was approximately 56 scfm, and the gas
residence times, cooling rates, and temperatures were comparable to
those found for full-scale utility boilers. Sorbents were injected by
means of a compressed air-driven eductor. Water injection could be
performed upstream or downstream of the heat exchanger. The water
injection position upstream of the heat exchanger was used for lowering
the temperature of the flue gas stream while the position downstream of
the heat exchanger was used to inject a urea solution for NOz
suppression.
5B-68
-------
Calcium Hydroxide Sorbents
The ability of hydrated limes to react rapidly with S02 at temperatures
below 1200°F was first noted in the EPRI-sponsored Dry Sorbent Emission
Control program. Tests conducted under this program showed that SOz
capture levels with pressure-hydrated dolomite decreased as the
injection temperature was gradually reduced from 1800°F to
approximately 1200°F and then again increased with further reductions
in injection temperature. This prompted a short study at the Southern
Research Institute where pressure-hydrated dolomite was injected at
temperatures ranging from 2400°F to 200°F, and a second sulfation
window was verified between 1200°F and 800°F.1
Unlike the higher temperature sulfation window level of 2000°F, where
CaS04 is the only thermodynamically stable compound, the temperature
range 1200°F - 800°F also contains the stable species CaC03. Thus the
amount of S02 capture in the lower temperature range will initially
depend on the rate of three competing reactions,2 which are shown
below.
Ca (OH) 2 + S02 ~ CaS03 + H20
Ca (OH) 2 + C02 ~ CaC03 + H20
Ca (OH) 2 CaO + H20
(1)
(2)
(3)
Maximizing S02 removal by hydrates injected at the 1000°F level
requires optimization of both sorbent and process parameters and
requires that reaction (1) is favored over reaction (2). Two
parameters control the hydrate's ability to remove S02. The initial
sorbent porosity (or surface area) is a good indicator of the hydrate's
ability to react with either S02 or C02. The second important hydrate
characteristic for enhancing S02 removal is particle size. Because the
chemical reaction rate is so fast, bulk diffusion of S02 (but not C02,
which has a concentration 50 times greater than S02) to the particle
can be a controlling factor. The diffusion of S02 to a particle is
inversely proportional to particle diameter, and unless the particles
have a mass mean diameter of less than 5 microns,2 most of the hydrate
will react with C02 via reaction (2).
The lime hydrates that were evaluated in the subscale tests are given
in Table 1. The first six hydrates are commercially available and were
produced by conventional dry hydration of lime. The alcohol hydrate
was produced by hydrating lime with a water/methanol mixture, and the
lignosite hydrate is produced by hydration with a calcium
lignosulfanate solution.
The hydrates were injected into the convective section of the pilot-
scale combustor as indicated in Figure 2. The injection points
correspond to gas temperatures of 1100, 1000, and 900°F, with constant
quench rates of 1000°F/sec. The S02 inlet concentration for all
hydrate injection tests was 2600 ppm, and the Ca/S mole ratio was 2 for
all tests.
5B-69
-------
The results are shown in Figure 3, which gives S02 removal in the
convective section as a function of injection temperature. The peak
effectiveness of all hydrates is achieved at about 1000°F, and the
clear superiority of the alcohol hydrate is evident. The single most
important hydrate chairacteristic for good S02 removal is surface area,
and it was found that utilization is almost directly proportional to
surface area.
Sodium Sorbents
When sodium bicarbonate is injected into a flue gas between 200°F and
400°F, the following reactions can occur.
2NaHC03 Na2C03 + C02 + HzO (4)
Na2C03 + S02 ~ Na2S03 + C02 (5)
Na2C03 + S02 + 1/2 02 ~ Na2S04 + C02 (6)
2Na2C03 + S02 + 2N0 + 202 ~ Na2S04 + 2NaNOa + 2C02 (7)
4NaHC03 + S02 + 202 + 2NO ~ Na2S04 + 2NaNOa + 4COz + 2HzO (8)
2NaN03 + S02 ~ Na2S04 + 2N02 (9)
At higher temperatures, the bicarbonate decomposes to sodium carbonate
before reacting with S02 (reactions 4-6). This decomposition results
in a sodium carbonate product with a large surface area, thus enhancing
reaction with S02. Nitrogen oxide cam react with sodium carbonate or,
at lower temperatures, with sodium bicarbonate, but only in the
presence of S02. The sodium nitrate product can react with S02 to yield
N02, which cam cause a brown stack plume under certain circumstances4
when its concentration in ppm exceeds a value equal to 200 divided by
the stack diameter in feet. To suppress the N02 production, urea can
be added to the sodium bicarbonate5, or the flue gas can be humidified.
The sodium compounds evaluated in the subscale tests were sodium
bicarbonate, NaHC03, and sodium sesquicarbonate, Na2C03>NaHC03>2H20.
The mass mean diameter particle sizes were 12.9 microns for the
bicarbonate and 12.2 microns for the sesquicarbonate. For all tests,
the inlet S02 concentration was 2600 ppm and the inlet NO concentration
was 350 ppm. The alcohol hydrate was injected at 1000°F for all tests
and the sodium compounds at temperatures between 250°F and 500°F. For
some tests, 5% urea (relative to the sodium sorbent on a weight basis)
was injected as a solution downstream of the final heat exchanger. The
baghouse temperature was lower than the injection temperature, due to
heat loss through the walls. The injected sodium compounds were
entrained in flue gas at the injection temperature for about one
second, after which they entered the baghouse, where they remained
until removed from the bags. The reactions between the sodium
compounds and S02 and NOx therefore took place initially at the
injection temperature (one second) and subsequently at the baghouse
temperature (minutes).
Figure 4 shows the S02/NOx removal as a function of sodium injection
temperature. Alcohol hydrate was used at an injection temperature of
1000°F for the data of this figure, and accounted for 60-70% S02
removal. Overall S02 removal remained at about 90% for the full
5B-70
-------
injection temperature range. NOx removal improved with increasing
temperature when the urea additive was used, but NOx removal remained
relatively constant at injection temperatures or as injection
temperatures fell below 350°F. A fifty percent N0X reduction was the
assumed contribution of a low NOx burner for the purpose of this
Figure. It can be seen that the sodium bicarbonate gave slightly
better removal of S02 and NOx than did sodium sesquicarbonate.
PROOF-OF-CONCEPT TESTING
The proof-of-concept demonstration is being conducted on the large
combustor at the Riley Stoker Research Facility in Worcester, HA. As
a result of the subscale tests described above, an alcohol water
hydrate is used as the principal calcium sorbent in the proof-of-
concept tests. Sodium bicarbonate will be the principal sodium
sorbent, because of its availability relative to sesquicarbonate.
Figure 5 shows a plan view of the equipment arrangement. The
demonstration plant begins with a combustor equipped with a low NOx
burner firing at a rate of 50 million Btu/hr. A slip stream of 7,000
scfm of flue gas from the combustor is routed to a heat exchanger to
reduce the temperature of the gas entering the simulated economizer to
900-1100°F. A 6-ft long section of open duct separates this heat
exchanger from the economizer tube banks to accommodate hydrate sorbent
injectors. The economizer is simulated with two air-cooled heat
exchangers.
The gas exits the economizer section at 650-750°F and is then cooled in
a simulated air heater to 300°F. The duct is straight except for one
180-degree turn to bring the flue gas back to the particulate control
devices. Dry sodium bicarbonate is injected into the flue gas exiting
the air preheater. Subsequent humidification of the flue gas with a
water spray is expected to enhance precipitator performance. The 30-
inch duct continues into a pulse-jet baghouse, and a 15-inch duct takes
a portion of the flow into em electrostatic precipitator. Separate
Venturis and dampers are used to control flow through the baghouse and
ESP. The gas streams are combined and returned to the scrubber using
a booster fan.
The parametric test program consists of a series of tests for the
purpose of demonstrating S02 and NOx reductions. The program is
designed to allow a determination of the influence of each parameter on
S02 and NOx removal. The parameters that will be investigated are given
in Table 2, along with the range of each.
The selected program lime hydrate is an alcohol hydrate, with a surface
area of 35 m^/gm, and a mass median particle size of 2.3 microns, and
the selected sodium bicarbonate has a mmd of 15 microns.
It is expected that flue gas humidif ication will allow the precipitator
to control outlet particulate loading and opacity to baseline levels
50-71
-------
when subjected to increased inlet loads due to hydrate injection.
Evaporative cooling to about 200 deg. F upstream of the precipitator
results in decreased gas volumetric flow and conditions the collected
fly ash layer yielding lower resistivity.
The testing described here is expected to define the operating limits
of the technology. Curves of removal efficiency versus sorbent to S02
and NOx mole ratios will be generated. The effects of the parameters
of Table 2 on these removal efficiency curves will be quantified.
Finally, the effects of these injected sorbents on the downstream
precipitator will be determined and mitigation techniques, such as
humidification, will be evaluated.
ACKNOWLEDGEMENTS
Funding and materials for this program have been provided by the U.S.
Department of Energy, the Electric Power Research Institute, the
Illinois Department of Energy and Natural Resources, the New England
Power Company, and the Church & Dwight Company.
REFERENCES
1. R. Beittel, et.al., "Effects of Injection Temperature and Quench
Rate on Sorbent Utilization," Proceedings of the 1986 Joint
Svmposv™ Drv SP2 and Simultaneous SO-. /NO]c Control Technologies,
June 2-6, 1986.
2. S. J. Bortz, et.al., "Dry Hydroxide Injection at Economizer
Temperatures for Improved S02 Control," Proceedings of the 1986
Joint Svmposi"™ «-m Drv SP2 and Simultaneous SO^/NO^ Control
Technologies. June 2-6, 1986.
3. S. J. Bortz, V. Roman, and G.R. Off en, "Hydrate and Process
Parameters Controlling S02 Removal During Hydroxide Injection Near
1000°F," Proceeding of the First CoTnT-iinpri FGD and Drv SP2 Control
Symposium, October 25-28, 1988.
4. L. C. Hardison, "Techniques for Controlling the Oxides of
Nitrogen," JAPCA, Vol. 20, No. 6, June, 1970.
5. V. Bland, J.J. Hammond, and R.G. Rhudy, "Full Scale Demonstration
of Additive NOz Reduction with Dry Sodium Desulfurization,"
Proceedings of the Joint Symposium on Stationary Co™tine; 1-ion NO][;
control. March 6-9, 1989.
5B-72
-------
Ca(0H)2
Ca/S = 2
NaHCOs Water
Na2/S = 1 -i Spray
Boiler
tHEcqi
nomizer
Air
1 '
Heater
I
I
SO2 = 2500ppm
NOk » 240ppm
I
SO2 = 525ppm
NOx - 240ppm
600ppm NOx Without Low NOx Burner (LNB)
60% Reduction With LNB
ESP
vw 1
I
I
SOz = 210ppm
NOx » 140ppm
Figure 1 Integrated Dry Injection Concept
Coal
0-25*10*Blu/h
18*
enrM-
Gas
Sampling
Sorbent Injectors
or Sampling Points
) Convective Bank
Gas
Sampling
4-
J Convective Bank
Sodium/Additive
Injection -*¦
- Water -*•
Injection
etxrF
Gas Sampling
Heat
Exchanger
350*F
Baghouse
/
Figure 2 Subscale System
5B-73
-------
80
70
- 60
a
§
E
a>
0C
6
CO
50
40
30
_L
I
X
1100 1000 900
Lime Hydrate Injection "temperature, °F
Hydrate
Ca/S = 2
Figure 3 SOz Removal For the Hydrates of Table l
a
>
0
E
01
GC
T3
C
CO
N
o
CO
200
250
300
100
90
80
70
60
50
40
30
20
10
O
3k
i—r
H r
Baghouse Temperature, "F
A*
200
NOx
Without Additive
SOz
300
400
Urea Additive
500
600
A Sesquicarbonate
A Sesqui w / Additive
O Bicarbonate
• Bicarb w / Additive
Ca/S = 1.8-2.5
Na2/S = £ -1.6
Sodium Injection Temperature. °F
(Hydrate Injection Temperature - 1000°F)
Figure 4 Combined S02 and NOx Removal
5B-74
-------
Equipment List
1 S8p Sfrtjam Tafcfr-Ofl T200*F 38*0
2 Heat Exchanger; Air-Cooled Temp. Control to 900-1100*F
3 Empty Duct lor Sottoert injectors
4 Economizer SaeL ll[" Air Cooled lube Banks In 3"-5' Ouct
5 Economizer SecL 2j|_ Wooty - 28 Fl Far Sec.
6 Air Heater System
7 Duct 30*0 - 100 FL Length
B Bag House
Plan - View
9 EleiJioatallc Precipitator Thhe-OW. 15*0
10 Humtdiflcatkxi Chamber
11 Electrostatic Precipitator
12 Baghouse Mentor!
13 ESP Uanturi
14 Booster Fan
Figure 5 Proof of Concept System
5B-75
-------
Table X'
Test Hydrates
Surface
Area MMD
Hydrate m2/g (/im)
1. Mississippi
23.5
2.2
2. Marblehead Lime
16.0
3.4
3. Beliefonte Lime
20.5
2.8
4. Tenn-Luttrel1
19.0
2.7
5. Chemical Lime
19.1
3.4
6. Colton Lime
19.0
2.6
7. Alcohol Hydrate
38.0
1.7
8. Lignosite Additive
15.1
2.6
Table 2,
Test Parameters
Parameter Range
Economizer Inlet Temperature 900 - 1100°F
Air Heater Exit Temperature 250 - 350°F
Precipitator Inlet Temperature 160 - 350°F
Ca(OH)2/S02 Mole Ratio 1.5 - 2.5
2NaHC03/(S02 + 2NO) Mole Ratio 0.5 - 2.5
Inlet S02 600 - 3100ppm
Inlet NOx 240 - 600 ppm
5B-76
-------
Engineering Evaluation of Combined NOX/SO2 Controls
for Utility Application
5B-77
-------
\
\
\
Intentionally Blank Page
-------
ty
J. E. Cichanowicz and C. E. Dene
Electric Power Research Institute
Win. DePriest and R. Gaikwad
Sargent & Lundy Engineers
J. Jarvis
Radian Corporation
ABSTRACT
This project evaluated the potential for developing combined N0x/S02 technologies to
provide attractive alternatives to conventional flue gas desulfurization (FGD) and
selective catalytic reduction (SCR). The technical feasibility of candidate processes was
determined through a specially-developed process evaluation methodology, that rated
candidate processes according to over 20 criteria. This analysis identified several
processes that appear favorable to conventional FGD/SCR, as well as potential
improvements for additional processes that elevate their rating to be equivalent or
preferable to FGD/SCR. For new plants, the NOXSO, Copper Oxide, and Zinc Oxide
absorption/regeneration processes, and the WSA-SNO catalytic reduction/oxidation
process were rated equivalent or preferable to FGD/SCR. For retrofits, the electron beam
process rates equivalent or preferable to FGD/SCR. Other processes - such as activated
carbon, wet chemical additives, and SNRB - appear favorable pending certain process
improvements, and under selected conditions.
Cost estimates for NOXSO, WSA-SNOx, and electron beam show all require higher
capital cost than conventional FGD/SCR ($200/kW), a consequence of the more complex
and numerous components to regenerate S02 and NOx into reusable byproducts, recover
heat for use in plant, etc. All processes potentially can implement improvements to
reduce capital cost; however only NOXSO has potential to be equivalent to FGD/SCR.
For levelized costs, candidate S02/N0x processes required higher levelized cost than
equivalent to FGD/SCR (8.8 mills/kWh). Similar to capital cost, all candidates have
potential improvements which if successfully implemented could reduce levelized costs
to be competitive with FGD/SCR.
The results of this project are not intended to endorse any specific processes. Accordingly,
additional processes will be analyzed to evaluate process improvements to elevate the
technical feasibility rating. Where appropriate, capital and levelized costs will be
determined.
5B-79
-------
INTRODUCTION
The 1990 Clean Air Act Amendments (CAAA) clarify for future decades the requirements
for S02 and NOx control. The creation of the SO2 allowance - and the ability to transfer,
trade, and/or sell such allowances - provides the industry with significant flexibility in
defining the least cost SO2 compliance strategy. For many utilities, an attractive option
is to obtain extremely high SO2 removal efficiency (>95%) at one site - maximizing S02
control for a given capital investment. Regarding NOx, two factors offer potential
incentive for control beyond that capable with low NOx burners. First, the prospects of
a N0x/S02 emissions trade - to be evaluated in terms of economic and environmental
impacts in a report to Congress due January 1, 1994 - may promote significant NOx
reduction at one site, similar to that for S02- Second, the ability to comply by averaging
NOx emissions - as proposed by Section 407 of the CAAA - provides similar incentive.
The conventional technologies used to provide significant SO2 and NOx removal on low
sulfur coals in Japan and Europe are wet flue gas desulfurization (FGD) and selective
catalytic reduction (SCR). The use of SCR, although not proven for high sulfur coal due
to issues relating to byproduct SO3 emissions, balance-of-plant impacts, and catalyst
lifetime, is being considered by state and local regulators for new plants. Alternative
technologies with less complexity, cost, and heat rate penalty could provide industry with
additional compliance options and cost savings. This project supports EPRI's effort to
maximize the technologies available to the industry to meet S02 and NOx control goals,
as well as waste water and solids management requirements.
PROJECT DESCRIPTION
The objective of this project is to evaluate candidate processes for combined SO2 and NOx
removal. The premise of the project is that processes that combine SO2 and NOx into
one step, or that employ separate steps with synergistic interations, are preferable to
FGD/SCR in terms of cost, reliability, and environmental effects.
The results of this project are not intended to endorse any specific N0x/S02 processes,
but rather to maximize EPRI research investment to develop, commercialize, and deploy
such technologies. The results will direct investment in selected candidate processes
either (1) as presently envisioned, or (2) with modifications to better meet utility needs.
Results will define three possible actions:
• Full-scale process demonstration (at nominally >50 MW), based on proven
performance at 1-5 MW pilot plant scale, including fullv integrated and
continuous operation.
• Further process development at 1-5 MW pilot plant scale, including fully
integrated and continuous operation, based on bench-scale results (at several
hundred acfm) that clarify the underlying chemical/physical concepts, or
5B-80
-------
• Additional bench-scale process development, addressing unresolved
fundamental issues that question the process technical basis or applicability
to utility systems.
To address these and other issues a technical feasibility analysis was conducted. This
analysis evaluated process features in terms of the potential to meet utility needs.
Economic evaluations were conducted for a limited number of processes ranked
according to the technical feasibility analysis as equal or preferable to FGD/SCR. Processes
not ranked equivalent to FGD/SCR were analyzed to identify improvements to increase
their ranking.
PROJECT APPROACH
A process evaluation methodology was developed to rate candidate N0x/S02 processes
by a quantitative scale in terms of potential to fulfill utility industry needs.
Candidate processes were identified by a literature search conducted in 1988 for EPRI by
Battelle (1); other processes were added as identified. The initial 70 processes identified
were reduced to approximately 25; in many instances the process developers had
abandoned development work after identifying a key shortcoming. Also, many
developers focused not on a complete process but on one step - for example gas phase S02
or NOx removal without consideration of practical waste water and solid management
needs. Thus, in many instances a complete process as necessary for commercialization
did not exist.
Developers of the processes that survived the initial screening were solicited for detailed
process information. As many processes are similar in concept, several could represented
by one type, to simplify the analysis. A complete technical feasibility assessment was
completed for 15 of the processes.
Cost estimates have been, or are being, prepared for eight of these processes for both new
plant application and for retrofit to existing plants. In addition, costs were developed for
conventional and advanced versions of FGD/SCR according to the EPRI Technical
Assessment Guide (3). Capital and total levelized costs were developed for a 500 MW
unit firing bituminus high sulfur coal; specific premises are presented later in this paper.
EVALUATION METHODOLOGY
The purpose of the evaluation methodology is to rate each process by quantitative scores
for key criteria. The evaluation methodology provides a broad-based systems perspective
for evaluating technologies, rather than focusing on one or a selected number of
characteristics or features. Although quantitative scores are derived, the results are by
definition subjective, due to the nature of quantifying the value of a process feature.
Table 1 presents the process evaluation methodology criteria, described as follows:
5B-81
-------
Retrofittabilitv (for existing plants only). The features of a process that determine the
advantages/disadvantages for retrofit into existing plant sites are considered. These
include process conditions at the point of access (flue gas temperature, gas composition),
the "footprint" required by the process and the subsequent area for installation, the land
requirements for waste disposal, and the use of existing equipment. Also, processes for
retrofit were evaluated according to two cases of S02 and NOx removal. "ITiese were
(a) 90% and 80% for S02 and NOx removal, respectively, (both new and retrofit), and
(b) a second retrofit case of 50% S02 and NOx removal. The latter was included to reflect
the potential need for moderate control applications.
Environmental Risk. This criterion addresses relative risk posed by either air, water, or
solids emissions; and risk to worker health/safety. The process features considered, are
(a) the fate/composition of high-volume waste from S02 and NOx removal (either
regenerated for commercial use or treated for landfill), (b) the composition/nature of
low-volume wastes or byproducts, (c) secondary gaseous emissions, and (d) risk induced
by process upsets.
Process Reliability. Process features proposed to define reliability issues are (a) chemical
complexity (number of significant chemical process steps), (b) mechanical complexity
(number of significant mechanical steps), (c) sensitivity of process equipment to upsets in
boiler operation, reagent feed, temperature control, etc., and (d) presence of corrosive
environments (requiring exotic/costly materials of construction).
Energy And Resource Requirements. The energy and resource requirements, estimated
based on a simplified process flow sheet developed for each process, allowed an estimate
for (a) auxiliary power use (or additional power generation potential), (b) consumable
reagents (lime, limestone, ammonia, etc.), (c) catalyst/sorbent consumption rates, and
(d) byproduct or energy credit. These quantitative estimates used fuel, auxiliary power,
and chemical cost per die EPRI TAG.
Table 2 provides an example of the manner in which points were awarded for process
features, by presenting the indices for determining scores for selected Retrofittability and
Environmental Risk criteria.
For each process, a total score was derived by summing the points awarded for each of the
preceding criteria, according to weighting factors. The baseline assessment presumed that
all criteria were equally important - and thus each category received equal weight. In
addition to the baseline case, sensitivity analyses were conducted to determine if the
process rating significantly changed when either Retrofittability, Environmental Risk,
Process Reliability, or Energy & Resource Requirements received additional weighting.
Thus, a total of 5 scores provided the basis for comparing candidate processes.
CATEGORIZATION OF N0X/S02 PROCESSES
Six categories can be defined into which almost all combined N0x/S02 processes can be
assigned. Although most processes are unique, many share similar chemical processes
5B-82
-------
and physical equipment, and thus are similar in how they integrate into the balance-of-
plant. Table 3 summarizes the processes considered for evaluation, and the six categories
defined. The categories are described as follows:
Adsorption /Regeneration. This type of process (Figure la), representing the largest
number of candidates identified, employs contacting a physical sorbent or catalyst with
flue gas, which adsorbs or reacts with S02 and NOx- This sorbent/catalyst is physically
removed from the flue gas to a regeneration reactor, where the sulfur or nitrogen species
are liberated from the sorbent. Generally, the regeneration step(s) require high
temperature or reducing gas (e.g., CO, H2, methane) at sufficient residence time to create
a byproduct stream of concentrated S02 or NOx. Each absorption/regeneration process
differs in regard to the type and quantity of sorbent/catalyst that must be recirculated, the
location of the flue gas adsorber (before or sifter the air heater), the requirement for an
additional particulate collector, the regeneration temperature and quantity of reducing
gas, and fate of NOx (e.g., regenerated or reduced selectively in the flue gas).
Flue Gas Irradiation. This category requires exposing flue gas to a high energy flux, most
commonly an electron beam to generate particulates (ammonium sulfates or nitrates) for
collection by an ESP or baghouse. Processes differ in terms of the method of exposing flue
gas to the energy flux, and the control equipment employed to form and collect
particulate.
The key features of the electron beam process are illustrated in Figure lb. Hue gas leaves
the existing particulate collector, proceeds to an evaporative spray cooler and electron
beam chamber, where irradiation generates hydroxyl radicals and oxygen atoms, which
react with S02 and NOx to form sulfuric and nitric acids. These acids react with injected
ammonia to form sulfates and nitrates of ammonia, which ar>2 subsequently collected in a
two-phase particulate control device. Collected solids are granulated and prepared for use
as feedstock for fertilizer manufacture.
Catalytic/Oxidation Reduction. This process type employs two sequential catalysts to
(a) remove NOx by SCR, and (b) oxidize S02 to S03, condensing the latter as sulfuric acid
for byproduct sale. Processes differ in the temperature at which each reactor operates, the
location of the particulate control device, and the mechanism for acid condensation.
Figure 1c presents a schematic of one version of this process designed to follow a
conventional particulate collector. A conventional SCR reactor reduces NOx, followed by
a reheating system (using auxiliary fuel) to elevate flue gas temperature and improve S02
to S03 oxidation reactor performance. After SO3 formation, a condenser is employed to
produce a high quality sulfuric acid for resale. The heat released by the oxidation of S02
to SO3 and the condensation of the sulfuric acid is partially recovered with a heat
exchanger to reduce auxiliary fuel consumption.
Wet Scrubber Additives. This category employs additives in wet scrubbers (most notably
lime, limestone, or dual alkali) to remove NOx. The principle additives are iron chelate
based compounds, which employ chemical properties to dissolve NO in solution,
5B-83
-------
removing the resulting compound as hydroxlyaminedisulfonates (HADS, and associated
similar compounds). Processes differ in terms of the specific additive employed, the
techniques for spent additive regeneration and/or recovery, and waste management
practices.
Dry Alkai Additives. Additives have been developed that can be injected into alkali-
based dry injection systems (e.g., spray dryers, in-duct processes) to effect NOx removal.
For example, sodium hydroxide has been employed with conventional spray dryers to
provide modest NOx removal that varies with the process conditions in both the dryer
vessel and the particulate collector. Similar to wet scrubbing, the specific fate of NOx
removed appears to be a form of HADS. Both the HADS and sodium species in the waste
present special waste management issues.
Electrochemical Catalysts. Catalysts that employ electrically induced polarity to
accomplish electrochemical reduction of S02 and NOx have been developed. The fate of
S02 and NOx is reduction to elemental sulfur and nitrogen; the former is condensed in
low temperature heat exchangers, similar to sulfuric acid. Processes differ with respect to
the form and material in which the catalyst is manufactured.
TECHNICAL FEASIBILITY EVALUATION RESULTS
The process evaluation methodology was applied to 15 of the processes in Table 3.
Analyses were conducted for both new and retrofit applications (differentiated by
including the Retrofittability criterion and the Case 2 [50%/50%] S02/N0x control goals
in addition to Case 1). Toted process scores were compared to those for FGD/SCR for
a "baseline" case - where the criteria of Retrofittability, Environmental Risk, Process
Reliability, and Energy and Resource Requirements were assigned equal weighting.
In addition, four scores were derived for cases where each criterion was assigned a
dominant (80% weighting) role.
The results identified five processes that consistently rated higher than FGD/SCR for the
conditions cited. These processes were three absorption/regeneration (NOXSO, Copper
Oxide, Zinc Oxide), one flue gas irradiation (E-beam, for retrofit only), and one catalytic
reduction/oxidation (WSA-SNOx). The features of these processes that contribute to
their selection are discussed in the following.
Absorption /Regeneration
The NOXSO, Copper Oxide, and Zinc Oxide processes each share common features and
thus reasons for receiving a relatively high score. All three processes scored at or near the
top in Environmental Risk due to eliminating high volume waste by regenerating S02
into a byproduct and reducing NOx to molecular nitrogen, without producing significant
secondary emissions. The combination of these two features without significant off-
setting penalties in other criteria (e.g.. Process Reliability) promoted a high rating.
5B-84
-------
Among these three absorption/regeneration processes, the NOXSO process received
simultaneous high scores for both new and retrofit applications. A significant factor
promoting a high ranking for NOXSO is the Retrofittability score, as the process requires
low temperature flue gas. The Copper Oxide process requires flue gas temperature access
prior to the air heater, and thus suffers in Retrofittability compared to NOXSO. However,
the NOx reduction and S02 regeneration steps are less complex for Copper Oxide than
NOXSO or Zinc Oxide; thus the Copper Oxide process derives relatively favorable scores
for Chemical and Mechanical Complexity. The Zinc Oxide process scores well in Energy
& Resource Requirements, as the combination of sorbent attrition rate, unit cost, and
mass recirculation rate minimizes sorbent make-up costs.
Catalytic Reduction/Oxidation
The process evaluated to represent this category, the WSA-SNOx process, similarly
eliminates high volume waste products and produces a commodity for resale. However,
the Environmental Risk score is penalized by the need to manage/dispose two catalysts,
the potential for secondary emissions (e.g., SO3, N02), and possible worker exposure to
sulfuric acid in the plant. The strengths of this process are the relatively few significant
chemical and mechanical steps, leading to a favorable Chemical and Mechanical
Complexity Score, and a favorable Energy & Resource Requirement score.
Electron Beam Process
The electron beam process scored above advanced FGD/SCR for retrofit plants only.
The ease of retrofit allowed by access to flue gas after the particulate collector contributed
to the Retrofittability score; in Energy & Resource Requirements production of a saleable
byproduct offset a significant auxiliary power penalty. (The lower NOx removal
requirement for Retrofit Case 2 [50%] reduced auxiliary power relative to an 80%
requirement.) This process did not score very high in any one category, but received
moderate to good scores among all categories. The process did not rank above
conventional FGD/SCR for new installations, as the retrofit score was not included and
the auxiliary power penalty for achieving high NOx removal (80%) assumed required for
new plaints is high.
ECONOMIC EVALUATION RESULTS
Preliminary capital and operating costs for these processes have been determined. Table 4
provides the specific design and economic assumptions employed for this analysis.
Complete process flowsheets were prepared, allowing equipment lists to be developed,
and costs assigned based on budgetary bids from several equipment vendors. The
uncertain development state for candidate processes necessitated that the cost analysis
define the sensitivity to changes in key design variables. In this manner, costs were
developed for a "baseline" design, that incorporates the best estimate for design variables,
and a "sensitivity" analysis to determine the influence of uncertainty for these variables
on process cost.
5B-85
-------
The premise of the sensitivity studies is that the lack of operating experience allows a
significant uncertainty in process design, which may not provide commercially acceptable
conditions. Historically, design premises for developing technologies have been
"optimistic," in that a lack of realistic experience leads to inadequate design for factors
such as contactor residence time, mass transfer, reagent reactivity, and equipment
sparing. Thus, costs for the first several full-scale versions of developing processes are
frequently greater than estimates derived from pilot-scale data. However, experience can
lead to improved designs, which eventually can lower cost and/or improve performance.
This trend has been witnessed with wet lime/limestone scrubbers, as recent design
concepts are lower in cost than the earlier generation designs applied in last decade.
Combined NOx/SQ2 Process Cost Discussion
Costs are presented in this paper only for the NOXSO, WSA-SNOx, and E-beam processes;
as well as for a conventional and advanced version of FGD/SCR. An advanced version of
FGD/SCR credited this approach with deriving the same improvements in process
control, materials of construction, and catalyst activity/performance that are assumed for
candidate N0x/S02 processes. For example, the advanced process version assumed
developing catalysts would be available in 5-7 years that provided control of the NOx and
residual ammonia available presently with 2/3 the amount of catalyst, thereby requiring a
smaller reactor. The advanced FGD process employed reduced sparing and other process
developments. Additional specifics of the conventional and advanced process versions
are detailed in reference (2).
Figures 2, 3, and 4 present capital and operating costs for these processes, including the
results of sensitivity analyses.
NOXSO. Figure 2a shows the NOXSO capital cost estimate of approximately S257/kW can
vary based on the design premises for particulate control, solids handling, absorber
residence time, sorbent utilization, and sorbent unit costs. Specifically, the capital cost
increase (or decrease) is shown for changes to (a) particulate collector design, requiring
changes in specific collecting area (SCA) by +15 or -10%, (b) spare solids recirculation
capacity (changes from +100 to -25% from baseline sparing assumptions), (c) flue gas
absorber residence time (+ 20% or -20% changes from baseline, respectively), (d) sorbent
utilization (+20% and -20% changes, respectively), and finally (e) sorbent unit cost (+20%
or -20% from baseline cost). These results show the design premises for particulate
control, adsorber residence time, and sorbent utilization have the most significant effects
on NOXSO process costs.
Figure 2b presents the results for NOXSO levelized costs, including a sensitivity analysis
for the previously discussed design premises, and operating cost factors such as natural
gas and sorbent attrition rate. Figure 2b shows the baseline cost estimate of approximately
11.7 mills/kWh can change by 0.5 mills/kWh or more due to each of the following;
particulate control, sorbent utilization, sorbent cost, waste disposal, and natural gas
consumption. The results show the influence of sorbent attrition is most significant.
5B-86
-------
and if doubled or halved from the baseline values affects levelized cost by 2.8 or
1.3 mills/kWh, respectively.
WSA-SNOx. Similar results for the WSA-SNOx process are presented in Figure 3.
Figure 3a shows the WSA-SNOx capital cost of approximately $375/kW can vary based on
the design premises for particulate control, space velocity of the SCR reactor and the S02
oxidation catalyst reactor, and the acid condensing tower spare capacity. Specifically, the
capital cost increase (or decrease) is shown for changes in (a) air/clcth ratio for the fabric
filter (increase to 1.5, versus decrease to 5.5 ft/s), (b) SCR catalyst space velocity (5000 hr-1
versus 7000 hr-1), (c) S02 oxidation catalyst space velocity (1500 vs. 1900 hr-1), and
(d) increasing (by 10%) or eliminating condensing tower spare capacity. Figure 3b shows
the baseline cost estimate of approximately 10.5 mills/kWh can vary by approximately
0.25 - 0.5 mills/kWh for each of the following: SCR catalyst space velocity, S02 oxidation
catalyst space velocity, S02 oxidation catalyst life, sulfuric acid condensing tower sparing,
and revenue from recovered sulfuric acid.
Electron Beam. Results for the electron beam process are shown in Figure 4. Figure 4a
shows capital cost estimates approach 5400/kW, and show the influence of changes in
evaporator residence time, capital cost of the electron beam generator, the successful
development of an advanced two-stage low power consuming electron generator, and
particulate control. For the levelized cost presented in Figure 4b, the electron beam
approach requires almost 13 mills/kWh. The influence of the preceding design variables
is shown, as is the effect of market value of the byproduct material.
Comparison to FGD/SCR
Figures 5 and 6 compare combined NOx/S02 process results with conventional and
advanced FGD/SCR The results compare baseline costs as well as a minimum and
maximum range, suggested by the previous sensitivity studies. The minimum and
maximum costs are based on the scenario of all described design premises changing
simultaneously to the maximum/minimum range. This occurrence is not anticipated,
but the range is reported to indicate the cost estimate uncertainty.
Results for the "baseline" assumptions indicate all NOx/S02 process candidates require
greater capital and levelized cost than either conventional or advanced FGD/SCR. This is
attributable to the more extensive equipment required for byproduct generation
equipment, heat exchangers, solid sorbent materials handling, etc., necessary to eliminate
high volume waste for disposal. Only the NOXSO process has the potential to be
competitive with conventional and advanced FGD/SCR on a capital cost basis.
For levelized cost, baseline estimates are not competitive with conventional and
advanced FGD/SCR- However, all three combined NOx/S02 processes potentially can
provide competitive alternatives, depending on the validity of the design assumptions.
5B-87
-------
RESULTS APPLICATION
These results have been and will continue to be used by EPRI to direct research to
maximize the number of viable combined N0x/S02 control technologies. The results of
this study indicate that some processes are ready for additional development at 1-5 MW
pilot plant scale, while others could benefit most from from additional bench-scale
development prior to more costly pilot plant work.
Pilot Plant Development
Early results from this project prompted EPRI participation in several key pilot plant
activities to develop the NOXSO and SNRB processes, with Ohio Edison and other.
cosponsors. These activities are described in a companion paper in this Symposium (3).
Additional Processes
These results can be used to evaluate process improvements to increase technical
feasibility scores of candidate processes that, as presently envisioned do not rate
equivalent or preferable to FGD/SCR Two examples are:
Wet Chemical Scrubber Additives. This category received low scores for (a) Energy &
Resource Requirements, as the excessive loss and subsequent makeup required for
chelating agents contributed to a significant operating cost, and (b) Environmental Risk,
as the contamination of conventional scrubber waste with both the chelating agent and
the nitrogen-containing waste (possibly as a form of HADS) could complicate disposal
and management of the scrubber high volume waste. A variation of this process has
been evaluated with enhanced methods for recovery of chelating agent, and treatment of
scrubber slurry for HADS which increases the process score to be equivalent to FGD/SCR
The economics of the wet scrubber additive combined N0x/S02 process with these
improvements will be evaluated.
Activated Carbon (Absorption/Regeneration). This process received a low score initially
due to the (a) high attrition rate and makeup required for activated carbon (penalizing
the Energy & Resource Requirements score), and (b) large number of chemical and
mechanical individual process steps (penalizing the Chemical and Mechanical
complexity score). Reducing the char consumption rate and simplifying the regeneration
steps increases the total process score to be competitive with conventional and advanced
FGD/SCR. A version of this process is presently being evaluated that is capable of lower
char replacement costs, and with simplified regeneration or sulfur compound disposal.
At present economic evaluations are being developed for three additional absorption/
regeneration processes (Copper Oxide, Activated Carbon, and Zinc Oxide), and for the wet
scrubber additive process (iron chelate with electrochemical regeneration). An economic
evaluation is being conducted for the SNRB process for new plants. The SNRB process,
although not scoring equivalent to FGD/SCR in the baseline case, demonstrated a
favorable score when Process Reliability was emphasized (due to the small number of
5B-88
-------
chemical and mechanical steps). Also, two advanced processes with limited bench-scale
data exhibit potential to derive a high score (Lehigh absorption /regeneration, and IGR
electrochemical); their further development will be monitored and possibly supported.
Additional processes that did not initially compare well with FGD/SCR wUl be evaluated
to define process improvements .
SUMMARY
This project evaluated the technical feasibility of initially 70 combined N0x/S02
processes, for the purpose of identifying alternatives to FGD/SCR. The technical
feasibility was determined through a specially-developed process evaluation
methodology that evaluated candidate N0x/S02 processes using vendor-supplied
information. This analysis, conducted for both new and retrofit applications, identified
several processes that ranked favorable to FGD/SCR. For new plants, processes in the
sorbent absorption/regeneration category (NOXSO, Copper Oxide, Zinc Oxide), one from
the catalytic reduction/oxidation category (WSA-SNOx), were rated preferable to
conventional and advanced FGD/SCR for new applications. For retrofits, the electron
beam process scores equivalent or preferable to FGD/SCR Other processes - such as
activated carbon, wet chemical additives, and SNRB - have been evaluated to identify
conditions for which their rating increases relative to FGD/SCR
Cost estimates show all processes require higher capital cost than FGD/SCR,
($160-200/kW) a consequence of more complex and numerous components to regenerate
S02 and NOx into reusable byproducts, recover heat for use in plant, etc Capital cost for
combined S02/N0x processes were greater than FGD/SCR; sensitivity analysis showed
each processes had potential for lower capital cost, equivalent to FGD/SCR for NOXSO.
Regarding levelized costs, candidate S02/N0x processes required higher levelized cost
than FGD/SCR (8.8 mills/kWh). Similar to capital cost, each candidate process has
potential for significantly lower levelized cost. Depending on the design assumptions
and research results, each process could be economically competitive with FGD/SCR
Processes that did not score favorable compared to FGD/SCR can implement process
modifications to improve their rating compared to FGD/SCR.
REFERENCES
1. "Survey of Combined N0x/S02 Technologies," EPRI internal report, March 1988.
2. Technical And Economic Feasibility of Combined N0x/S02 Controls, Volumes I
and n," draft report for EPRI RP3004-1, November 1991.
3. "EPRI Technical Assessment Guide - Electricity Supply," Special Report EPRI P-6587-L,
Volume 1: Revision 6; September 1989.
4. Bolli, R., "Ohio Edison's Clean Coal Projects: Circa 1991," technical paper presented at
the 1991 S02 Control Symposium, Washington, DC, December 1991.
5B-89
-------
TABLE 1
CRITERIA FOR THE EVALUATION METHODOLOGY
Retrofitabilitv
- Required Point of Access to Flue Gas Stream (20%)
- Process Land/Space Requirements (20%)
- Land/Space Requirements for Waste Disposal (20%)
- Use of Existing Equipment (40%)
Environmental Risk
- High Volume Waste or Byproduct Materials (60%)
- Low Volume Waste or Byproduct Materials (10%)
- Secondary Gaseous Emissions (20%)
- Risk Induced by Process Upsets (10%)
Process Reliability
- Chemical Complexity (25%)
- Mechanical Complexity (25%)
- Sensitivity to Process Upsets (25%)
- Corrosive Environments (25%)
Energy and Resource Requirements
- Quantity of Energy Required
- Reagent Consumption Rates
- Catalyst/Sorbent Consumption Rates
- Byproduct or Energy Credit
5B-90
-------
TABLE 2
EXAMPLES OF THE SCORING PROCEDURE/POINTS AWARDED
Retrofitabilitv: Example: Required point of access to the flue gas stream
Points Description
10 Access required following a cold side ESP or baghouse
5 Access required between the air heater and the particulate
collection device
3 Access required between the air heater and economizer
1 Access required upstream of the economizer
Environmental Risk: Example: High volume waste or byproduct materials
Points Description
10 Process produces a byproduct of high purity which is always
marketable
8 Process produces a byproduct for which market may be limited
by seasonal or geographic factors
7 Process produces a byproduct which could potentially be sold but
for which a market has not been established
6 Process produces a benign waste which is easy to handle and
presents no disposal problems
5 Process produces benign waste more difficult to handle
3 Process waste contains soluble materials which could be leached
from the waste
0 Process waste stream is potentially hazardous
Example: Low volume waste or by product materials
Points Description
10 Process produces no low volume wastes
8 Process produces low volume wastes which are saleable
7 Process produces low volume wastes which are easily disposed of
(e.g., co-disposal with other waste streams)
6 Process produces low volume wastes which are treatable
5 Process produces low volume wastes which can be reprocessed
(e.g., catalysts with valuable metals)
4 Process produces low volume waste which may present some
disposal problems due to chemical/physical properties
2 Process produces a low volume waste which is hazardous
o Process produces multiple hazardous low volume wastes
5B-91
-------
TABLE3
PROCESS IDENTIFIED FOR TECHNICAL EVALUATION
Overall Process Category
Solid Adsorption/Regeneration
Irradiation of the Flue Gas
Wet Scrubbing
Gas/Solid Catalytic Operations
Dry Injection Additives
Electrochemical
Process Name
UOP/PETC Fluidized-Bed Copper Oxide
Rockwell Moving-Bed Copper Oxide
NOXSO
Mitsui/BF Activated Coke
Sumitomo/EPDC Activated Char
Sanitech Nelsorbent SOx/NOx Control
Lehigh University Low Temperature
Battelle ZnO Spray Dryer
Ebara E-Beam
ENEL Pulse-Energization
Argonne/Dravo ARGONNOX
Dow Electrochemical Regeneration
Haldor Topsoe WSA-SNOx
Degussa Catalytic
B&W SOx/NOx/ROx/BOx (SNRB)
Parsons Flue Gas Cleanup
Argonne High-Temperature Spray Dryer
PETC Mixed Alkali Spray Dryer Studies
IGR/Helipump
5B-92
-------
TABLE 4
GENERAL DESIGN PREMISES FOR ENGINEERING EVALUATION
Design Premises
Application:
Location:
Boiler Size:
Boiler Type:
Plant Life:
Fuel Analysis:
— C:
— H:
-- N:
-- S:
-- 0:
-- CI:
-- Ash:
-- Moisture:
Firing Rate:
Gross Heating Value:
SO, Emission:
NO, Emission:
S02 Reduction:
N0X Reduction:
Economic Premises
Commercial Operating
Date:
Construction Period:
Discount rate:
AFUDC rate:
Levelized fixed
charge Rate*:
Base Inflation Rate:
Real Escalation Rate:
Natural Gas:
Nonfuel Items:
30-year Levalization
rate:
Utility Boiler
Kenosha, VII
500 MWe
Pulverized coal
30 years
67.0
4.6
1.2
3.0
4.7
0.1
16.4
3.0
196 tph
12,360 Btu/lb
4.8 lb/MBtu
0.4 lb/MBtu, (0.6 lb/MBtu for retrofit)
New - 90%, for retrofit - 90% & 50%
New - 80%, for retrofit - 80% & 50%
January 1995
3 years
6.2%/year
6.2%/year
10.6%/year
0%/year
4.3%/year
0.0%/year
1.0%/year
Based on 30-year book life, 20-year tax life, 38% composite federal and
state tax, and 2.0% for property taxes and insurance.
5B-93
-------
Fkm
Gas
Adsoqation
(ZnO.NOXSO)
Aah (NOXSO, CuO)
(NHglorCuO)
Figure 1 a: Adsorption/Regeneration Processes
HjO
NHj
Ru»
Gas
Ash
vw
Air
PartieuWB
Ru0 Gjs
Gas
By-product
M-r
Colirtion
Coofing
lniadi*ion
Coltodion
By-producJ-
By-product
Trmtaant
, Y Y Y
Rgure 1 b: Rue Gas Irradiation Processes
•HgSOt
Hue
Gas
Air
High DTSciaocy
G«Xaas
Haster
ParDeUbta
Cefaclor
Exetiangw
wv. ^ t
SCR
Cjtaiyit
t
St>2
Olilitiiiii
Flinor
AuxBaiy FimI
Figure 1 c: Catalytic Reduction/Oxidation Processes
5B-94
-------
300
290
280
270
260
250
240
230
220
210
200
Figure 2a: Sensitivity of Capital Cost on Selected NOXSO Design Variables
New Plants, 500 MW (Premises in Table 4)
$/kW
Sorbent
Particulate
Utilization
Adaorbar
Control
Solid*
Handling
Spar*
Raaldance
Sorbent
Cost
Tim*
-20%
mills/kWh
14.5 F
14
13.5 |-
13
12.5
12
11.5
11
10.5
10
Sorbent Attrition
9.5
Baa*
Coat
Sorbant Sorbent
Coat
Disposal
•25%
•20*
-20%
Particulate Adsorber
Control Reaidenc
Figure 2b: Sensitivity of Levelized Cost on Selected NOXSO Design Variables
New Plants, 500 MW (Premises in Table 4)
5B-95
-------
430
420
410
400
390
380
370
360
350
340
S/kW
S02-S03
Catalyst
Basa Cost
330
A/C Ratio
SCR
Catalyst
1600
1900
£000
7000
Condenser
Spare
Spare
I No Spare
Figure 3a: Sensitivity of Capital Cost on WSA-SNOX Design Variables
New Plants, 500 MW (Premises in Table 4)
12
11.5
11
10.5
10
9.5
mills/kWh
Base
Coat
A/C Ratio
S02-S03
Catalyst
Catalyst
S02-S03
Condenser
Ll,l# Spare
1600
1900
7 yr».
12 yrs.
Cost
10% Spai
o Spare
25%
Figure 3b: Sensitivity of Levelized Cost on WSA-SNOX Design Variables
New Plants. 500 MW (Premises in Table 4)
5B-96
-------
$/kW
435
425
415
405
395
385
375
365
355
345
335
325
Figure 4a: Sensitivity of Capital Cost on E-BEAM Design Variables
New Plants, 500 MW (Premises in Table 4)
E-beam generator
Const. Power
By Product
Removal
Larger/
Single
Baghouse
Cooler
Residence
Time
E-beam
Two-stage
Generator
Base Cost
(Decreased Power)
20%
-30*
1.2 Mrad
Base
1.8 Mrad
mills/kWh
By Product
Sale
Generator
Constsnt
By Product
Removal
Larger/
Single
Bsghouse
E-Beam
Cooler
Reaidence
Generator
(Decreased
6 8ec
5 8ec
-25%
-26*
11.8
~100*
-50*
1.2 Mrad
Base
1.B Mrad
Figure 4b: Sensitivity of Levelized Cost on E-BEAM Design Variables
New Plants, 500 MW (Premises in Table 4)
5B-97
-------
CAPITAL COST (S/KW)
450
Cost $/kW in January 1990 Dollars
CONV.
FGD/SCR
NOXSO
ADV.
Figure 5: Sensitivity of Cumulative Process Parameters on Total Capital Requirement
Levelized Cost (mills/kWh)
24
22
20
18
16
14
12
10
8
6
4
2
0
I
SNOX
ADV.
CONV.
E-BEAM
F00/8CR
Figure 6: Sensitivity of Cumulative Process Parameters on Levelized Cost
5B-98
-------
ADVANCED FLUE GAS TREATMENT USING ACTIVATED CHAR PROCESS
COMBINED WITH FBC
Hitoshi Murayama
Thermal Power Department
Electric Power Development Co., Ltd.
Tokyo, Japan 104
5B-99
-------
Intentionally Blank Page
5B-100
-------
ABSTRACT
There is a process combined with the use of a fluidized bed combustion boiler (FBC)
as a flue gas treatment process using activated char. This technology utilizes
both features of the activated char flue gas treatment system and the fluidized bed
combustion boiler. The activated char flue gas treatment system shows a high
denitrification (DeNOx) performance with the injection of ammonia. This is similar
to the NOx removal system (SCR) using a metal catalyst that is often used to remove
low SOx concentration flue gas. The fluidized bed combustion boiler makes low SOx
concentration flue gas by the in-furnace desulfurization
(DeSOx).
Furthermore, the activated char flue gas treatment system has DeSOx and de-dusting
performances. Combination use with the above DeNOx technology enables high-level
treatment of flue gas.
The Electric Power Development Company has been entrusted by Agency of Natural
Resources & Energy, Ministry of International Trade & Industry, to perform the
pilot (demonstration) test of the activated char flue gas treatment system, for
which laboratory testing has already been performed. This test plant has a
treatment gas amount of 10,000 m3 N/h and is attached to the FBC boiler
demonstration test plant (50 MW) located in our company's Wakamatsu Coal
Utilization Research Center. Testing of the plant has been carried out since last
year.
The results show that the test plant satisfies the targeted performance of 80
percent NOx removal efficiency, 90 percent SOx removal efficiency, and 30 mg/m3N
outlet dust concentration.
Me also understand that the system has several other features and are now
collecting data for commercialization of the system.
5B-101
-------
INTRODUCTION
Dry type flue gas treatment technology using activated char in the thermal power
plant has been under development for more than 20 years in Japan. Its
demonstration test was started 10 years ago. At present, the technology is being
developed for its commercialization.
This technology has the following advantages: the amount of water used is very low
compared to the conventional flue gas treatment technology (when the system is used
as desulfurization (DeSOx) equipment); high-level flue gas treatment is possible;
and it is not necessary to consider the influence of flue gas on the downstream
equipment because the system can be installed right before the stack.
The flue gas treatment process using the activated char includes the following
three processes: the dry type DeSOx process for pulverized coal combustion; the
dry type DeSOx and denitrification (DeNOx) process for pulverized coal combustion;
and the DeNOx process using activated char (AC-DeNOx) for the fluidized bed
combustion boiler (FBC).
The Electric Power Development Company (EPDC) was entrusted by the Ministry of
International Trade and Industry (MITI) to perform tests for putting these three
processes to practical use. This paper reports the results of tests made so far
with respect to the AC-DeNOx process used for the fluidized bed combustion boiler.
ACTIVATED CHAR FEATURES
Since the activated char has a very large specific surface, it has been used widely
as an air cleaning agent and waste water treatment agent since the second half of
the nineteenth century.
The activated char has various performances depending on the raw materials or
manufacturing method. Activated char (activated coke) used for the treatment of
flue gas has the following features.
5B-102
-------
• SOx is adsorbed.
• It has catalytic capability to decompose NOx under the existence of
NH3.
In low-temperature areas (less than 100 °C), NOx adsorption
reaction becomes remarkable in place of catalytic reaction.
SOx adsorption level is higher than NOx removal reaction level in
reactivity.
• Recycle use available.
• The performance improves by repetitive recycle use.
In the DeSOx reaction, SOx is oxidized and adsorbed on the activated char surface
in a form of sulfuric acid as shown in Table 1. If ammonia exists in the flue gas,
it is adsorbed by the ammonium hydrogen sulfate or the ammonium sulfate.
The DeNOx reaction includes the same catalytic reaction as the SCR reaction,
oxidation and adsorption reaction, and the reaction with reducing material on the
surface of activated char. (Refer to Table 2.) In the flue gas treatment around
140 °C in the coal fired power plant, the main reaction is the SCR reaction.
The DeSOx reaction and the DeNOx reaction hardly occur at the same time on the
surface of activated char. The DeSOx reaction has priority to its occurrence.
That is, in flue gas having a high SOx concentration, the activated char performs
DeSOx reaction. In flue gas having a low SOx concentration, the DeNOx reaction
becomes remarkable. Figure 1 shows the relationship between the SOx concentration
at the entrance of the reaction tower filled with activated char and the NOx
removal efficiency. It is understood that the lower the SOx concentration is, the
higher the NOx removal efficiency.
The DeSOx and DeNOx performances of the activated char lower with its adsorption of
SOx, etc. It therefore becomes necessary to remove the activated char with the
lowered performance and to add high-performance activated char. Normally, the
reaction tower structure will be the moving bed type. Furthermore, the DeSOx and
DeNOx performances of activated char are regenerated for reuse because of the high
cost of production. The desorption process of SOx is mainly performed in the 400
°C reducing atmosphere. Table 3 shows the reaction configuration. The SOx which
is oxidized and adsorbed on the activated char is reduced to SO2. If ammonia is
present, it functions as a reducing material. If there is no ammonia, the carbon
of the activated char functions as a reducing material. In the latter case, the
activated char becomes depleted. This will be described later.
5B-103
-------
OUTLINE OF AC-DeNOx SYSTEM
The FBC boiler has better environmental features than PCF because of its in-furnace
desulferization and low temperature burning. Generally, the FBC boiler is equip
with the SCR system to satisfy the emission
regulations in Japan.
The Wakamatsu Coal Utilization Research Center of EPDC also has an FBC
demonstration plant (50 MW) with SCR.
Fig. 2 shows the comparison between the SCR system and AC-DeNOx system". SCR is
placed in front of the AH (air preheater) because it is used in about 350 °C flue
gas. In this case, the compound of NH3, SO3 and fly ash often causes plugging of
the AH and vibration in the fan. However, as the AC-DeNOx system is able to
denitrate in 140 °C flue gas, it is placed just before the stack. Furthermore, SO3
is easily caught by activated char and cannot be detected in the flue gas of the
AC-DeNOx outlet. It is therefore not necessary to consider the troubles of
downstream equipment.
When the flue gas temperature is low at boiler start up, in case of the SCR system,
NH3 has not been injected because the catalyst surface is covered with an amnion i«|
sulfate compound. NH3 injection can be started when the temperature of the flu^
gas is sufficiently high. In the case of the AC-DeNOx system, activated char also
has DeNOx performance in low temperature flue gas, as already described.
This AC-DeNOx system roughly consists of the DeNOx tower and the regenerator. The
type of the DeNOx tower is a moving bed type because the activated char bed catches
dust and SOx. The relation between gas flow and activated char is a crossflow. In
the regenerator, SOx which is adsorbed in the activated char is desorbed, and
condensed SO2 gas which contains HC1 etc. is generated. The generated gas is
washed with water and then blown into the bottom of the FBC furnace with compressed'
air. The SOx in the furnace reacts with the calcium in the FBC bed material and is
removed. Therefore, the generated SO2 gas treatment equipment does not need in
AC-DeNOx system.
As stated above, the AC-DeNOx tower works as a dust collector. The method is the
as same as a granular bed dust collector. The dust which was collected by the
activated char bed is discharged with activated char from the AC-DeNOx tower and
separated from the activated char with a vibrating sifter through the regenerate
5B-104
-------
The features of the AC-DeNOx system are summed up as follows:
• Denitrification can be expected from boiler start up.
• It has a secondary desulfurization effect and high performance
desulfurization by combination with the FBC boiler.
• It has a dust collecting function.
It can achieve high performance flue gas treatment.
• It is not necessary to consider this influence on downstream system
such as AH, as with the SCR system.
PILOT (DEMONSTRATION) TEST FACILITY
The AC-DeNOx test facility is installed in the 50 MW FBC boiler demonstration test
plant at the EPDC Wakamatsu Coal Utilization Research Center. This FBC boiler is
a bubbling type and the demonstration test has been continuing since 1987.
Figure 3 shows the system of the AC-DeNOx test facilities. The test flue gas is
taken out in front of the SCR in the FBC plant. The temperature and the dust
concentration of its gas are controlled by the gas cooler and the bag house
respectively. After the gas pressure is increased by fan, and NH3 is injected, the
flue gas enters the AC-DeNOx tower.
As for activated char flow, activated char which goes in the DeNOx tower moves down
slowly and is discharged out of the tower by the activated char discharging
conveyor. The activated char is then sent to the regenerator by the bucket
conveyor. The char reactivated in the regenerator is put into the vibrating sifter
to remove activated char powder and fly ash from the char. The char is again sent
to the DeNOx tower. Because a small amount of activated char is lost, the lost
amount is replaced.
The condensed SO2 gas which is generated in the regenerator is washed for removing
HC1 etc. and then goes into the wind box of the FBC boiler.
Table 4 shows the outline of the DeNOx tower. The tower is rather large compared
with the SCR reactor.
Figure 4 shows the structure of the DeNOx tower. It is divided three layers along
the char flow. The first layer is the flue gas inlet layer, that is the louver
layer, for the desulfurization and dust removal. The second layer is for dust
removal and denitrification. The third one is for the DeNOx. The moving rate of
5B-105
-------
the three layers is different for each. The rate of the first layer is the fastest
and the rates are slow in order of the second and third layers. These rates are
determined according to the degree of dust collection. The average retention time
of the activated char is 120 ~ 150 hrs.
The structure of the regenerator is shown in Figure 5. As shown in the figure, it
is divided into two parts; the heating part and cooling part. The methods of
heating and cooling are indirect for each. The middle of the regenerator between
the heating part and the cooling part is for collecting generated gas.
TEST ITEMS AND TEST CONDITIONS
The first test item is to achieve the target performance shown in Table 5. The
DeNOx efficiency is more than 80%, the DeSOx efficiency is more than 90%, and the
dust concentration at the AC-DeNOx tower is less than 30 mg/m^N. The other test
items are as follows:
• Stability of operation over a long period time
• Loss ratio of activated char
• Boiler type characteristics
• Optimal amount of moved activated char
• Others
For the test conditions, the inlet gas condition is shown in Table 6. The actual
NOx and SOx concentrations are shown in Figures 6 and 7. Maximum NOx concentration
is about 250 ppm and maximum SOx concentration is less than 100 ppm. In most cases
these are 200 ppm and 50 ppm, respectively.
In regard to NH3 injection, NH3 reacts with SOx before reacting with NOx on
activated char. Therefore, an amount of NH3 injection is also needed for the
reaction with NOx and SOx. The amount of NH3 reacting with SOx is 1.2 ~ 1.7 as
mole ratio which has been obtained in laboratory tests. In these tests, NH3 is
injected at a mole ratio of 1.5 NH3/SOX. For removing NOx, NH3 and NOx reacts one
to one. NH3 is injected by set up mole ratio.
Though other compounds may react with NH3, they are ignored. The amount of NH3
injection is determined by the following equation.
5B-106
-------
UNH3 = {CS0x * 1-5 + CnOx * [set up mole ratio]} x 10-6
x [flue gas volume]
UNH3; NH3 injection volume (m3N/h)
CsOx: Concentration (ppm)
CnOx: NOx Concentration (ppm)
TEST RESULTS
The test equipment has worked for about 3,000 hours as of this September. It has
been confirmed that the target performance of this test is achieved. An example of
the performance test results is shown in Table 7. Also, there has not been any
serious mechanical trouble up to now.
Some factors which give influence to DeNOx efficiency were found. These factors
are the partial accumulation of dust in the DeNOx tower, oxygen concentration and
moisture content in flue gas.
The detailed test results are as follows.
DeNOx and DeSOx Performance
DeNOx efficiency was over the target value, 80%, in the early test run, but went
down gradually. At last, the efficiency became less than 80% after a few months.
There are some reasons for this phenomenon. The most important reason is the
non-uniform flue gas flow in the DeNOx tower because of the partial accumulation of
dust.
The reason of the partial accumulation of dust is that dust goes in the activated
char layer deeply, contrary to our expectations. In the early stage, the DeNOx
tower had two layers of activated char, which were the louver layer with a width of
about 100 mm, and the downstream layer, 1,700 mm. The char flow rate of the front
layer was fast, but the back layer was slow. Part of the dust went into the back
layer so that dust accumulated in the lower level of the layer. As a result, the
pressure drop of the DeNOx tower went up, flue gas distribution became bad and
DeNOx efficiency dropped.
In our study, the DeNOx tower was made over into the three layers by the division
of the back layer into two layers. The char flow rate of the second layer has been
made faster compared with the former rate to ensure smooth discharge of dust from
the DeNOx tower. The modification made improvement of the pressure drop of the
5B-107
-------
DeNOx tower and DeNOx efficiency- Figure 8 shows the trend of the pressure drop
before and after modification. The DeNOx efficiency has been confirmed to maintain
ever 80%.
The relations between oxygen concentration, moisture content and the DeNOx
efficiency are shown in Figure 9 and Figure 10. respectively. Figure 9 shows that
less than 5% oxygen concentration remarkably gives influence to DeNOx efficiency.
Moisture content is higher and the DeNOx efficiency is lower. By rough estimation
the 1% increase of moisture content makes the DeNOx efficiency about 2% down.
Figure 11 shows the trend of DeNOx activity of activated char itself. The white
circle indicates the trend of regenerated char activity. It cannot be seen as a
large change, and the activity goes up only gradually. The solid circle indicates
the char activity at the outlet of the DeNOx tower. There is no big difference
between the regenerated char activity and the outlet of the DeNOx tower char.
Though the char at the outlet of DeNOx tower is covered with fly ash, like dirty
snow ball, the results indicate that it does not cause any great decrease in DeNOx
efficiency. Also, much S02. which makes char activity decrease, was not
adsorbed in the char.
There are no problems in regard to the DeSOx performance. Until now, the DeSOx '
efficiency has been maintained at more than 90% which is the target value.
Dust Removal Performance
Before this test began, the target of dust removal efficiency was less than 30
mg/m^N. According to the test results, dust concentration at the outlet of the
DeNOx tower has been less than 10 mg/m^N at all times. The inlet dust
concentration was 100 ~ 240 mg/m3N. The reason why the results were much lower
than the target value is that the char rate of the back layer, the flue gas outlet
side layer, is set slow, so that the generation of activated char powder is
reduced. In the case of other flue gas desulfurization plants using activated
char, outlet dust concentration is 12 ~ 20 mg/m3N. In these plants, the char rates
are over four times faster than our test plant.
The analysis of the amount of unburned carbon in dust indicates that there is not
such a great difference between the inlet dust and the outlet dust. That is, not
so much activated char powder is generated in the back layer.
5B-108
-------
Activated Char Loss
In this system there is a certain amount of loss of activated char. There are two
types of loss. The first is the mechanical loss which occurs due to abrasion
between the char and the structure or of the char against itself as it circulates
through the system.
The second type is chemical loss, as mentioned before, which is the carbon
consumption in char at the time of deoxidizing SO3 in the regeneration.
The trend of activated char loss during testing is shown in Figure 12. This figure
indicates that the activated char loss rate is less than 1% in all tests. The
activated char loss rate is calculated as the percentage of weight (kg) of char
supply loss against the weight (kg) of char circulation per an hour. The results
are lower than our expectations. The reason is that char is not consumed so much
as deoxidization material to supply sufficient ammonia in regeneration.
In our DeSOx demonstration test using activated char, commissioned by MITI, the
loss rate was about 1.8%. In order for the flue gas to have more than 500 ppm SOx,
there was not sufficient ammonia to deoxidize SO3. Therefore, more char was
consumed. However, each plant has individual features, and it is not easy to make
comparisons among different plants. All that can be said is that the AC-DeNOx
plant has less char loss than the AC-DeSOx plant.
Others
The response of the DeNOx system to boiler start up and shut down was tested.
Figure 13 shows the result of response test at boiler start up. The results
indicate that the AC-DeNOx system can work early from boiler start up. The DeNOx
efficiency is near 100% in the early stage, and gradually goes down with time and
with the increase of flue gas volume. The minimum efficiency is 40 ~ 50%. After
that, with the DeNOx tower temperature going up, DeNOx efficiency increases.
But as activated char has high heat capacity, the temperature in the DeNOx tower
does not rise easily. Therefore, the DeNOx efficiency goes up slowly. This is a
weak point of this system. But operation methods can cover this weak point.
There are no problems at boiler shut down. When flue gas stops, the temperature in
the DeNOx tower rises easily because of char oxidization. The counter-measure of
this phenomenon has been confirmed.
5B-109
-------
In regard to removal of heavy metal and halogen, some materials were measured at
the inlet and outlet of the DeNOx tower. The results show that mercury is removed
effectively by this system, hydrogen chloride (HC1) is removed by 50 ~ 70% at 140
°C. Other materials are also removed, but as their inlet concentration level is
low, the effectiveness of removal is not clearly confirmed.
CONCLUSION
The pilot (demonstration) test of AC-DeNOx technology for FBC has been good,
according to expectations. As a result, this technology will be applied to the 350
MW FBC being replaced, instead of an oil-fired boiler, at Takehara Thermal Power
Station No. 2 Unit.
During the remaining test period, the confirmation of long time stable operation,
the performance of following boiler load change and the like are to be tested and
the reliability of this technology will be confirmed.
In this paper, activated char flue gas treatment technology has been mentioned as
an FBC flue gas cleaning method with high DeSOx, DeNOx and dust removal efficiency.
Also, activated char is useful for removing hydrocarbon-like dioxin, heavy metal |
and other toxic materials. In fact, in Germany, the activated flue gas treatment
system is used to remove dioxin from incinerator flue gas.
Thus, activated char has several kinds of performance. There are some choices to
adopt activated char flue gas system to fit the purpose of gas treatment.
To promote the use of activated char systems, the activity of activated char must
be increased, and its price should be lower. Research and development is currently
progressing regarding these issues.
5B-110
-------
Inlet NOx ; 300ppn
Space Velocity ; 800H-'
>»
O
c
o
LU
200
600
400
Inlet S02 Concentration (ppm)
800
1000
Figure 1. Relation between Inlet SOx and NOx Removal Efficiency
SCR - DeNOx
Stock
SCR-
Boiler M/C DeNO* A/h B/house IDF
\
AC - De NO*
AC-
Oe-NO* Siqck
Boiler M/C A/h EP IDF
Generated Gas
Figure 2. Comparison between SCR and Activated Char DeNOx System in FBC Plant
5B-111
-------
Ol
?
M
M
N)
MC
c.c f >
A-ior i-iof n
>-.vJ I
wmm
AC Conveyer
AC reed lloppcr
Regenerator
'J* VoshIns Tower
'a
Blower
Regenerator
llealer
AC Conveyer
SO. Gas Blower
AC Feed lloppcr
AC Cooling BlOK^r
Mfi Dilution
Gas Blower
AC Discharger
DcNm Towr
AC Rotary Valve
Q..P "+©*C Discharger
ttIF
Dust Collector
i\w
Hake-up AC lloppcr
Cos Cooler I
Big Filler
Fine AC Separator
Dust Blower
Fine AC
lloppcr
Fine AC Bag Filter
Flno AC
lloppcr
Make-up AC Conveyer
X. A. X J
JL£1JL4-j
AC Bog Filter a
Figure 3. Activated Char DeNOx Test Facilities
-------
AC TnlcC
ID 700
Hot All Oullel
Cone. SO] Oullel
Hot Air Inlet
Cold Air Oullel
Cold Air Lnlei
IQ7QQ
Figure 5. Structure of Regenerator
AC lnlei
Gas Inlet Side Oas Outlet Side
3,56$
I. 300
Gas Outlet
Screen
Louver
AC Outlet
Figure 4. Structure of DeNOx Tower
-------
200
Bed Temperature:
850 *— 860°C
S2 150
100
N
50
w"
CO
1.05
.10
1.20
1.25
1.30
In-Bed Air Ratio (-)
Figure 6. SOx Emission from FBC Boiler
//i
A
& j
yf,
JK'
y
.o
/
y
Bed Ten
850
riperature
~ 860°C
:
1.05
1.10 1.15 1.20
In-Bed Air Ratio (—)
Figure 7. NOx Emission from FBC Boiler
.25
1.30
5B-114
-------
200
150
100
before modification
after modification
a
Time (days)
tf)
Figure 8. Trend of Pressure Drop at DeNOx Tower
100
_©
o
Ul
X
O
z
o
o
—O—Pi lot Test
—A—Lab. Test
Oxygen Content (%)
Figure 9. Influence of Oxygen Content in Flue Gas to DeNOx Efficiency
5B-115
-------
90r
SV = 400h"1
T = 140"C
'
' 80-
' A
o
z
—O— Pi lot Test
—A—Lab. Test
Moisture Content(%)
Figure 10. Influence of Moisture Content in Flue Gas to DeNOx Efficiency
10
[test condition] [gas condition]
AC : 300 cc NO = NH. =200ppm
S V : 400 h"
Temp : 140 12
Slo'-HV
(S.S) [0.4]
o qQj-qqI^S.
^ iia [o. 2] ^
10.5]
O
1.7)
ttl.T)(^6) (13.1)
(10.5)
O : regenerated activated char
• : used activated char
8 9 10 11 12 1 2 3 4 5 6
"90 '91
Date of sampling
Figure 11. DeNOx Performance Laboratory Test Results of Activated Char
5B-116
-------
1000
cumulative loss weight
500
average loss rate
9 0.5
1000 1500 2000 2500
Operation time (hours)
Figure 12. Trend of Activated Char Loss
Loss rateisupplied char weight / moving char weight per a hour
Ore. , '990
Outlet Mb (as 0, 6X) Emission lin
/j
V
///i
4
^—7=—
»
i?
"///
¦
/
¦ r^r—'
i /
Boiler Firing(Gn Amission}
I T7 I
Parallel In
IOCS Load of Generator
Z— „ ! -
Surtino Condition
Boiler cold
Turbine ni I
H Gas Flc* Rite
— C0-1ZXO>,N/h)
| Wi Flew Rite
— (Q.OO-LOte'N/h)
3 Inlet IMn BO,)
„ (O-SOOdoO
Q (inlet KKa 6* Oi)
— (0-250 co>)
B Inlet Ut
— (0-500 oci)
I Outlet Kk
(0-250 «¦)
13 Inlet S0>
. (0-100 oo«)
9 Oe-Kk Efficiency
— (0-100*)
TaBereture of Reeew
Q Tod
(0-160 T)
f IboBr oert
— (o-ifio r)
Q Kiddie oert
— (0-160 t)
9 L»er pert
— (0-160 t)
I Bonos
~ (0-160 "C)
I 16:00 12.02 20:00 12.Q5 00:00 12.Q5 0&:00 12.05 08:00 12.05 12:00 12.05 16:00 12.05 20:00'22=00
Figure 13. Start up Characteristics of AC-DeNOx Pilot Plant
5B-117
-------
Table 1
DeSOx REACTION ON AC SURFACE
1/2 02 (g) --> 0 (ad)
502 (g) + 0 (ad) —> SO3 (ad)
503 (ad) + H2O (g) —> H2SO4 (ad)
NH3 (g) + H2SO4 (ad) —> NH4HSO4 (ad)
NH3 (g) + NH4HSO4 (ad) — > (NH4)2S04 (ad)
Note: AC: Activated Char
g: gas
ad: adsorption
Table 2
DeNOx REACTION ON AC SURFACE
Catalytic Reaction with NH3
NO (g) + NH3 (g) + 1/4 02 (g) —> N2 (g) + 3/2 H20 (g)
• Adsorption Reaction
1/2 O2 (g) —> 0 (ad)
NO (g) + 0 (ad) —> NO2 (ad)
• Reaction with Nitrogen Functional Group
NO (g) + 0 (ad) --> NO2 (ad)
NO2 (ad) + NH = C (surface) —> N2 (g) + OH (ad) + 0 - C (surface)
Note: AC: Activated Char
g: gas
ad: adsorption
5B-118
-------
Table 3
REGENERATION REACTION ON AC SURFACE
(high temp. & reduction condition)
(NH4)2S04 (ad) --> 2NH3 (g) + H2SO4 (ad)
H2SO4 (ad) —> SO3 (ad) + H2O (g)
3SO3 (ad) + 2NH3 (g) --> 3S02 (g) + N2 (g) + 3H20 (g)
(. 2SO3 (ad) + C —> 2S02 (g) + CO2 (g))
• 2NH3 (g) + 6 0 - C (surface) --> H2 (g) + OH (ad)
NH3 (g) + 0 - C (surface) --> NH ¦= C (surface) + H2O
Note: AC: Activated Char
g: gas
ad: adsorption
Table 4
Specifications of AC-DeNOx Test Plant
Item
Specification
Type
Reactor Volume
(effective)
Space Velocity
Gas Volume
Gas Temperature
Cross flow moving bed
25 m3
400 1/h
10.000m3N/h
140 C
5B-119
-------
Tables
Target Value of Pilot Test
.Items
Target Value
DeNOx Efficiency
DeSOx Efficiency
Outlet Dust Concentration
Pressure Drop (DeNOx Tower)
More than 80%
More than 90%
Less than 30 mg/m3N
Less than 150mmAq
Table 6
Inlet Rue Gas Condition
item
Value
Rue Gas Volume
Gas Temperature
NOx Cone.
SOx Cone.
Dust Cone.
HCI Cone.
HF Cone.
10.000m3N/h
140 C
200ppm
50ppm
200mg/m3N
10ppm
5ppm
5B-120
-------
Table 7
Performance Test Results of AC DeNOx System
Items Target Values Test Results
Inlet NOx 200 153
(ppm)
OutletNOx 40 26
(ppm)
NOx efficiency 80 83
(%)
Slip NH3 30 or less 12
(ppm)
Inlet SOx 50 12
(ppm)
Outlet SOx 5 or less < 5
(ppm)
SOx efficiency 90 >90
(%)
Inlet dust 200 119
(mg/m3N)
Outlet Dust 30 or less 5
(mg/m3N)
5B-121
-------
Intentionally Blank Page
5B-122
-------
COMBINED SC>2/NOx CONTROL USING FERROUS»EDTA AND A SECONDARY
ADDITIVE IN A LIME-BASED AQUEOUS SCRUBBER SYSTEM
M. H. Mendelsohn
C. D. Livengood
J. B. L. Harkness
Energy Systems Division
Argonne National Laboratory
9700 S. Cass Avenue
Argonne, Illinois 60439
5B-123
-------
Intentionally Blank Page
5B-124
-------
ABSTRACT
Integration of N0X control into existing flue-gas desulfurization (FGD)
systems addresses site—specific control requirements while minimizing
retrofit difficulties. Argonne has studied the use of metal-chelate
additives, such as ferrous*EDTA in various wet FGD chemistries, to
promote combined S02/N0X scrubbing. A major process problem is
oxidation of the iron to the ferric species, leading to a significant
decrease in NOx-removal capability. Argonne discovered a class of
organic compounds that, when used with ferrous«EDTA in a sodium
carbonate chemistry, could maintain high levels of N0X removal.
However, those antioxidant/reducing agents are not effective in a lime-
based chemistry, and a broader investigation of antioxidants was
initiated. This paper discusses results of that investigation, which
found a practical antioxidant/reducing agent capable of maintaining N0X
removals of about 50% (compared with about 15% without the agent) in a
lime-based FGD chemistry with Fe(II)*EDTA.
5B-125
-------
INTRODUCTION
A number of technologies are available to remove either sulfur dioxide
(SO2) or nitrogen oxides (N0X) from flue gas. However, integrated
technologies that can simultaneously control both species could offer
significant advantages, such as lower capital and operating costs,
better system operability and reliability, and possibly lower resource
consumption and waste volumes. The construction of complete integrated
systems will be of interest for new utility plants and industrial
applications, as well as for existing sites that currently have minimal
pollution control. On the other hand, opportunities to incorporate
integrated pollution-control measures into existing flue-gas cleanup
(FGC) systems will be particularly important for operators of SO2
scrubbing systems who are faced with the need to add additional control
of N0X.
Argonne National Laboratory (ANL) has been conducting research on
combined S02/N0X control technologies for the U.S. Department of Energy
(DOE) since 1981. Much of that work has emphasized techniques for the
retrofit of N0X control to both wet and dry (spray drying) scrubber
systems, particularly in high-sulfur coal applications. This paper
reports the results of recent work with combinations of chemical
additives designed to promote the economic removal of N0X in wet flue-
gas desulfurization (FGD) systems using a lime-based chemistry.
Some metal chelates, such as ferrous ethylenediaminetetraacetate
[Fe (II)*EDTA], promote N0X removal because they quickly react with
dissolved nitric oxide (NO), forming the complex Fe(II)«EDTA»NO. The
coordinated NO can react with sulfite and bisulfite ions, freeing the
ferrous chelate for further reaction with NO. This synergism makes
separate regeneration of the Fe(II)«EDTA to release the NO unnecessary.
A significant process problem is oxidation of the iron in the additive
to the inactive, ferric state. This oxidation occurs both by direct
reaction with dissolved oxygen and by reaction with species produced
from decomposition of the Fe(II)«EDTA«NO complex. In some cases,
addition of another chemical, specifically an antioxidant and/or
reducing agent, has been effective in counteracting the harmful effects
of ferrous oxidation. Recently (1), we have published our first studies
performed with Fe(II)*EDTA combined with an antioxidant/reducing agent
in a sodium-carbonate chemistry. However, these antioxidant/reducing
agents were not as effective in a lime-based chemistry, and a broader
investigation of antioxidants was initiated.
5B-126
-------
In this paper, we compare results obtained for Fe(II)*EDTA alone in
sodium carbonate with results obtained in a lime—based scrubber
chemistry. We also present results obtained in a lime-based chemistry
with Fe(II)*EDTA and various antioxidant/reducing agents singly and in
combination. We have found an effective antioxidant/reducing agent
capable of maintaining N0X removals of about 50% (compared with about
15% without the agent) in a lime-based FGD chemistry with Fe(II)*EDTA.
The control of SO2 is not impaired, and may in fact be improved, by the
additive.
EXPERIMENTAL PROCEDURE
A flow diagram of the laboratory-scale scrubber used for this research
is shown in Figure 1. The scrubber vessel itself is approximately 3 in.
in diameter and uses a "disk and donut** design internally to promote
gas-liquid contacting. In addition,, a sieve plate is placed at the
bottom of the scrubber to provide some liquid holdup in the column. By
adjusting the fractional open area in the plate, one can adjust
pollutant removals for a given liquid recirculation rate. That rate can
be varied between about 300 and 1400 mL/min, with typical values for the
experiments reported here ranging from 300 to 500 mL/min. The "flue
gas" (blended from bottled gases) feed rate was approximately 100
standard liters per minute, yielding liquid to gas ratios (L/G) of about
20-40 (gal/min)/(1000 ft3/min). Instrumentation from Beckman is used to
continuously monitor the following flue-gas components: oxygen (O2),
carbon dioxide (CO2), NO, nitrogen dioxide (NO2), and SO2.
Recently, several modifications were made to the scrubber system that
are not included in earlier descriptions (1, 2.) - The previously used
glass humidifier was replaced with an all-metal steam generator. A
precision metering pump feeds water to a metal coil immersed in an oil
bath at 125"C at a rate sufficient to humidify the feed-gas stream to a
water vapor content of about 8%. Two sampling points are used for feed-
gas analysis. One is placed upstream of the water-vapor addition to the
feed stream and is used primarily to set up the feed-gas mixture. The
other is placed immediately upstream of the feed-gas injection point
into the scrubber column and is the primary feed-gas sampling location
after a scrubbing experiment is started.
The initial experiments in hydrated lime, Ca(0H)2, consisted of a
statistically designed series of runs to test the effects of variations
in feed gas O2 concentration from 2-6%, feed gas SO2 concentration from
1000-3000 ppm, and Fe(II)*EDTA (additive) concentration from 0 to
0.067 M. We set the baseline conditions without Fe(II)*EDTA so that the
SO2 removal would be about 90%. This reeuired the use of a sieve plate
at the bottom of the scrubber having an open area of 9.8%. A liquid
recirculation rate of 500 mL/min was needed for 90% SO2 removal and
resulted in a liquid level of 33-36 cm in the scrubber column. An
initial amount of lime equivalent to 0.08 moles/L was added to all
scrubbing solutions. The initial pH varied from about 10 without
Fe(II)*EDTA to about 7.5 with Fe(II)*EDTA. During the experiments, the
pH was controlled with a 10 wt% lime slurry at 6.5 after dropping to
that level. Temperature in the scrubbing solution holding tank was
maintained at 50°C. A 20% excess of EDTA was used in all formulations
5B-127
-------
of the Fe (II) *EDTA additive. The flue gas was prepared by first
preparing a base mixture of gases containing 14.5% CO2, the desired O2
concentration, and the balance as nitrogen (N2). Pollutant gases were
added as follows: NO2 was set first at 50 ppm, NO was added at a
previously set level of 450 ppm, and, finally, the prescribed amount of
SO2 was added.
Subsequent experiments that were conducted to investigate secondary
additives (antioxidant/reducing agents) used similar conditions so that
they could be compared with the previous series and earlier work with
sodium carbonate. The pollutant gas concentrations were 3000 ppm for
SO2, 450 ppm fyr NO, and 50 ppm for NO2. The other constituents were
8% moisture, 14.5% CO2, 6% O2, and the balance N2. The liquid
circulation rate was varied from about 250 mL/min to 400 mL/min in order
to maintain the liquid level in the scrubber at the same 33-36 cm above
the sieve plate. The different rates reflected changes in the
properties of the scrubber liquor. Concentrations of 0.067 M for
Fe(II)*EDTA and 0.033 M for the secondary additives were used. Initial
pH values varied from about 5 to 7.5 for different additive
combinations. All tests were batch runs that lasted from one to two
hours.
RESULTS AND DISCUSSION
The statistically designed experiments showed that the only variable
that had a significant effect on both NO and N0X removal in a Ca(OH>2
chemistry was the additive, Fe(II)»EDTA. The magnitude of this effect
over baseline conditions was found to steadily decline, from zero time,
at 5-, 10-, 30-, and 60-min intervals from +2 6% to +2%. The SO2
variable had a small positive effect, while the O2 variable had a small
negative effect on both NO and N0X removal at those same intervals (<±5%
in all cases).
In both chemistries studied to date, NOx removal with Fe(II)*EDTA alone
can be characterized generally by an initial decline to a minimum value,
which then either rises slightly or stabilizes at an apparent
equilibrium value. We believe that this behavior is due predominantly
to initial oxidation of ferrous»EDTA to ferric*EDTA and removal of free
Fe(II)»EDTA by formation of the complex Fe(II)«EDTA*N0, followed by a
slight increase in regeneration rate of Fe(II)«EDTA, and finally, a
stabilization of the ferrous*EDTA concentration. Overall, the initial
set of experiments with Fe(II)»EDTA in Ca(OH)2 gave N0X removals that
degraded much more rapidly than had been the case in sodium carbonate.
The reason for this different behavior in the two chemistries is
explored later in this paper. However, given the well-known problem of
ferrous ion oxidation in such systems, it was decided to investigate
combinations of Fe(II)*EDTA and antioxidant/reducing agents.
Experiments reported previously (1) demonstrated improved N0X removal in
a sodium-carbonate chemistry with Fe(II)*EDTA when a secondary organic
additive was present. When one of the same compounds (pyrogallol) was
tried in a lime chemistry with Fe(II)*EDTA, however, little if any
improvement was observed. This is shown in Figure 2, where performance
5B-128
-------
with the secondary additive was actually worse early in the run and only
marginally better at longer times. In contrast to this behavior, in
sodium carbonate it was found that after 90 min of scrubbing, NOx
removal had dropped to 32% with Fe(II)*EDTA alone; with Fe(II)*EDTA and
pyrogallol, after 90 min, N0X removal had increased slightly to 64%.(1).
To help understand why the pyrogallol was not effective in a lime
chemistry, we decided to conduct a more detail^: comparison oi the
performance of Fe(II)*EDTA for combined S02/N0X removal in both lime and
sodium carbonate chemistries. We first noted a significant difference
in SO2 removal behavior when comparing performance with and without the
additive Fe (II)-EDTA. In sodium carbonate, SO2 removal without
Fe(II) «EDTA was set at about 90% by adjusting the liquid level in the
column and was found to decrease slightly, to about 87-8J%, with
Fe (II) "EDTA added. In contrast, SO2 removal in lime war. again set to
about 90% without Fe(II)*EDTA, but it was found to increase
dramatically, to about 99%, with the additive.
Next, we compared NOx removals under similar conditions for both of
these chemistries, as shown in Figure 3. Differences in the conditions
for these two experiments were as follows (sodium carbonate versus
lime) : feed gas nitrogen dioxide, 75 ppm versus 50 ppm; feed gas O2,
5.4% versus 6.0%; and excess EDTA, 1% versus 20%. Note that previous
work in sodium carbonate showed little difference in NOx removal when
unmixed nitrogen dioxide levels of 0-150 ppm were used (2.) . We also
found that O2 concentration variations in the range of 4-8% had little
effect on NOx removal (1)-
Finally, a 20% excess of EDTA should have had a beneficial effect on N0X
removal, because a 20% excess of EDTA has been found to decrease the
rate of oxidation of Fe(II) by dissolved oxygen, compared with the rate
for a stoichiometric Fe (II) *EDTA solution (3.). Important conditions
that were identical were the concentration of SO2 in the feed gas,
scrubbing solution pH, liquid level in the scrubber column (33-36 cm),
and initial concentration of Fe(II). The important feature to note in
comparing the two curves in Figure 3 is that the N0X removal declines
much more rapidly in the lime chemistry than in the sodium-carbonate
chemistry, reaching its minimum value after about 45 min.
To assess the role of oxygen in this different behavior, experiments
were performed with the additive Fe(II)»EDTA in both chemistries without
O2 in the feed gas. The results for NOx removal from these tests are
shown in Figure 4. Although a comparison between the runs with and
without O2 for each chemistry shows considerable differences, it is
noteworthy that N0X removal is still significantly worse in the lime
chemistry than in the sodium-carbonate chemistry. The obvious
conclusion is that another effect, besides oxidation from flue-gas O2,
is responsible for the lower NOx removal with Fe(II)*EDTA in lime as
compared with the removal in sodium carbonate. A plausible explanation
for this effect is the much reduced solubility of sulfite ions in a lime
environment as compared with a sodium environment. In fact, sodium
sulfite is about 10,000 times more &-.-luble than calcium sulfite. The
importance of this fact lies in the proposed scrubbing mechanism for NO
5B-129
-------
by Fe(II)»EDTA. As described in the literature (A), the principal
reaction is presumed to be the equilibrium reaction shown below:
Fe(II)-EDTA2- + NO < > Fe(II)-EDTA-NO2-
The regeneration of the complex Fe(II)»EDTA from Fe(II)*EDTA*NO has been
postulated to occur by complex reactions with sulfite (SC>32~) and/or
bisulfite (HS03-) (5.) . The kinetics of these reactions have been shown
to be dependent on the total S(IV) concentration (i.e., concentration of
SO32- + HSO3-) . Hence, the rate at which "fresh" Fe(II)*EDTA can be
regenerated in order to maintain a high NO removal depends on the
concentration of sulfite plus bisulfite. From this point of view,
therefore, the simple fact of the lower solubility of sulfite and
bisulfite in a lime system would lead to a lower N0X removal than in a
pure sodium-containing system.
The investigation of antioxidant/reducing agents was resumed with the
study of several new candidate compounds combined with Fe(II)•EDTA in
lime. Initial screening results with several different antioxidant
candidates are shown in Figure 5. As can be noted from the figure, one
antioxidant had little effect on NOx removal, some gave improved NO*
removal, and one actually gave worse NOx removal. Figure 6 shows
results for N0X removal obtained with the best candidate investigated
thus far, sodium ascorbate, with various Fe:ascorbate ratios. with an
Fe:ascorbate ratio of 2:3, N0X removal after one hour is about three
times higher than with Fe(II)»EDTA alone (49% versus 15%). This level
of removal could be sustained for about 30 min. In order to understand
the mechanism by which the ascorbate species improved NOx removal,
performed an experiment with Fe(II)*EDTA and sodium ascorbate in lime,
but without O2 in the feed gas stream. The result for this test is
compared in Figure 7 with the test for Fe(II)*EDTA alone without oxygen
in lime. As can be seen from Figure 7, even without O2, addition of
sodium ascorbate leads to a great improvement in NOx removal when
compared with Fe(II)*EDTA alone. This result implies that ascorbate is
performing another role besides that of simple antioxidant. It could be
that ascorbate ions may be involved in regeneration of Fe(II)«EDTA from
the complex Fe (II)»EDTA»N0 and/or that ascorbate can reduce ferric*EDTA
to ferrous*EDTA. Further work on the role of ascorbate ions in the
overall mechanism is in progress.
Before we describe the results obtained on some combined antioxidant
systems (i.e., two added chemicals) with Fe(II)»EDTA in a lime-based
scrubbing system, it is of interest to note a change we made in the
experimental system. During the course of the combined chemical work,
we noticed that in some systems, we had to lower our recirculation rate
to unrealistically low levels in order to maintain the liquid level in
the scrubber column at 33-36 cm above the sieve plate. Because of this,
we changed the sieve plate to one having 10.3% open area and found that
we could now achieve 90% SO2 removal with a circulation rate of
510 mL/min and a liquid level 28-31 cm above the sieve plate. One
interesting effect we found when making this change in liquid level can
be seen in Figure 8, which compares N0X removal for scrubbing solutions 1
having an Fe:ascorbate ratio of 1:1 for the liquid levels of 33-36 cm
and 28-31 cm. Overall, the differences are rather small. However, at
the lower liquid level, we were able to maintain a constant N0X removal
5B-130
-------
of 42% for 50 min with no decline observed after a total running time of
100 min. This may indicate a situation where the oxidizing and reducing
effects have been balanced. This potentially significant result needs
to be confirmed by repeating these conditions and running the experiment
for a longer period.
From an economic perspective, the most interesting combination of
secondary additives was sodium ascorbate and urea. Urea itself had no
noticeable effect on NOx removal when used with Fe(II)*EDTA alone.
However, as shown in Figure 9, when urea is added to the scrubbing
solution along with sodium ascorbate, NOx removal improves over that
obtained with the same amount of ascorbate alone. Also, Figure 10 shows
that urea with ascorbate can even perform better than a greater amount
of ascorbate alone (i.e., some ascorbate can be replaced with urea to
obtain a NOx-removal performance comparable with that of the original
amount of the more expensive ascorbate). The optimum ratio of urea to
ascorbate is yet to be determined.
CONCLUSIONS
In this paper, we have shown the following important results in regard
to combined scrubbing of S02/N0X in an aqueous scrubber system:
• Fe(II)«EDTA additive alone improves SO2 removal from
about 90% to 99% in a lime-based scrubber chemistry.
• NOx removal in a lime-based chemistry declines much more
rapidly than in a sodium chemistry either with or
without oxygen in the feed gas.
• Ascorbate ions can markedly improve NOx-removal
performance in a lime-based chemistry either with or
without oxygen in the feed gas.
• Partial replacement of ascorbate with less expensive
chemicals (such as urea) appears to be possible.
ACKNOWLEDGMENTS
This work is supported by the U.S. Department of Energy, Assistant
Secretary for Fossil Energy, under contract W-31-109-ENG-38, through the
Pittsburgh Energy Technology Center (PETC). The authors wish to
acknowledge the support provided by Perry Bergman and Charles Drummond
of the PETC. In addition, the authors express their appreciation to
Sherman Smith for his contributions to the modification and maintenance
of the experimental apparatus, which enhanced the performance of the
tests described herein.
REFERENCES
1. M. H. Mendelsohn and J. B. L. Harkness. "Enhanced Flue-Gas
Denitrification Using Ferrous EDTA and a Polyphenolic
Compound in an Aqueous Scrubber System." Energy & Fuels,
vol. 5, 1991, pp. 244-248.
5B-131
-------
2. J. B. L. Harkness and R. D. Doctor. Development: of Combined
Nitrogen Oxide/Sulfur Oxide Environmpnral-Contrn] Technology.
Argonne, IL: Argonne National Laboratory Report ANL/ECT-14,
August 23, 1985.
3. E.Sada, H. Kumazawa, and H. Machida. "Oxidation Kinetics of
Fell-edta and Fell-nta Chelates by Dissolved Oxygen." Ind.
Rno. Chem. Res., vol. 26, 1987, pp. 1468-72.
4. R. J. Walker and M. B. Perry. "Simplified Wet Scrubbing
Process For SO2, NOx, and Particulate Removal From Flue Gas."
Paper No.47e, American Institute of Chemical Engineers Spring
Meeting, Houston, Texas (1989) .
5. D. Littlejohn and S. G. Chang. "Reaction of Ferrous Chelate
Nitrosyl Complexes with Sulfite and Bisulfite Ions." Ind.
Enp. Chem. Res., vol. 29, 1990, pp. 10-14.
5B-132
-------
To Hood
Vera
TC
Gas Sample
tt> Analytical
Instruments
Deionized HjO
Gas Sample
to Analytical
Instruments
Stirrer
Flue-Gas
Inlet
Chemical
Feed Tank
Fill Port
H-o-H-
Pump
Controller
Variac
I—Holding Tank
_l j HeatTape
h
Sample
Stirrer
TC
Pump
cu-
Pump
Controller
•1 o I
Pump
Controller
Holding
Tank
Circulation
Pump
Sample
Port
Pump
Controller
To Drain
Rinse
HjO
Fill Tank
Temperature,,
Controller
Controller
Figure 1. Flow diagram of laboratory aqueous-
scrubber system
5B-133
-------
O FaflO'EDTA alone
~ Fe(lQ*EDTA with pyrogaSol
50
20
30
40
50
Time (min)
90
Figure 2. NOx removal in hydrated lime with Fe(II)«EDTA
alone or Fe(II)*EDTA and pyrogallol
O Fe(ll)-EDTA in hydrated lime
~ Fe(ll)-EDTA in sodium carbonate
50
40
30
10
20
30
40
50
60
70
80
90
0
Time (min)
Figure 3. NOx removal with the additive Fe(II)»EDTA in
both sodium carbonate and hydrated lime chemistries
5B-134
-------
70-
60
50.
CO
>
40-
cr
cf
z
30-
O Fe(ll)»EDTA in hydraied lime
~ Fe(ll)«EDTA in sodium carbonate
20-
80
100
120
40
60
0
20
Time (min)
Figure 4. N0X removal for Fe(II)*EDTA without oxygen in
both sodium carbonate and hydrated lime
O Fe(ll)«EDTA alone
O DEHA
~ 2-Butanone oxime
+ Hydroquinone
A Isoascorbtc acid
X Sodium ascorbate
04
0 10 20 30 40 50 60 70 80 90
Time (min)
Figure 5. Screening test results for several antioxidant
candidates in a lime-based scrubber chemistry
5B-135
-------
~ Fe:ascorbale -1 £.5
O Feiascorbale -1:1 JS
O Fe(ll)-EDTA aione
A Feascoibate - 1 ri
o
>
§
®
cc
30
100
90
80
30
40
10
Time (min)
Figure 6. N0X removal for various initial ratios of
Fe(II):sodium ascorbate in a lime-based scrubber chemistry
~ Fe(ll)»EDTA alone
A Fe(ll)-EDTA with ascorbate
60-
50'
>
§ 40-
30-
20-
100
20
40
60
80
120
0
Time (min)
Figure 7. N0X removal with Fe(II)*EDTA and with or without
sodium ascorbate and no oxygen in the simulated flue gas
5B-136
-------
70-
65-
O sodium ascorbata. 28-31 cm liquid
~ sodium ascorbata. 33-36 cm liquid
60-
e- 55-
a
1 50-
O
DC
^45-
40-
35-
30-1
0 10 20 30 40 50 60 70 80 90 100
Time (min)
Figure 8. NOx removal for Fe(II)»EDTA with sodium
ascorbate at two different scrubber liquid levels
65
O ascorbata alone; Fa ascorbata - 1:1
~ ascorbata with urea; Fa:ascorbata - 1:1
60-
>
50.
-------
70-
~ ascoibate alone; Fe:ascorbate -1:1
A ascoibate with urea; Fe:ascorbate -1:0.5
65-
60-
45
40-
35
80
90
100
0
10
40
50
60
70
20
30
Time (min)
Figure 10. NOx removal for Fe(II)»EDTA with sodium
ascorbate and with or without urea and with a scrubber
liquid level of 28-31 cm
5B-138
-------
RECENT DEVELOPMENTS IN THE PARSONS FGC PROCESS
FOR SIMULTANEOUS REMOVAL OF SOx AND NOx
K. V. Kwong
R. E. Meissner, III
The Ralph M. Parsons Company
100 West Walnut Street
Pasadena, California 91124
C. C. Hong
Columbia Gas System Service Corporation
1600 Dublin Road
P.O. Box 2318
Columbus, Ohio 43216
5B-139
-------
Intentionally Blank Page
-------
ABSTRACT
The Parsons Flue Gas Cleanup (FGC) Process was developed Co remove 99-plus percent
of both S0X and N0X from coal-fired boiler flue gas.
The Parsons process consists of three key process steps. Two of the steps, H2S
recovery and sulfur production, use commercially proven technologies; the third,
hydrogenation, is an adaptation of commercial experience to permit processing
particulate-containing flue gases dilute in S0X and NC^.
Bench scale and pilot scale units have been built for testing of the key SOx-NOx
hydrogenation step. Bench scale results confirm the ability to remove 99-plus
percent of both S0X and N0X. Recent pilot plant tests have demonstrated that the
catalytic hydrogenation reactor is capable of removing 99-plus percent of S0X and
92 to 96 percent of NC^ from coal-fired boiler flue gas.
5B-141
-------
INTRODUCTION
The objective of the Parsons Flue Gas Cleanup (FGC) Process is simultaneous
removal of 99-plus percent S0X and N0X from coal-fired boiler flue gases. The
Parsons FGC Process is unique for Che following reasons:
¦ It is capable of simultaneously removing 99-plus percent of S0X and N0X.
¦ It is a regenerable process.
¦ It produces salable elemental sulfur.
¦ It is an adaptation of technology that has been successfully used in some
70 commercial plants treating sulfur plant tail gases.
¦ The process economics are essentially inr=nsitive to the amount of sulfur
in the coal.
The technology has under gone continuing development for the past few years. This
paper will describe the process configuration, and the latest results of the bench
scale and the pilot scale test programs.
PROCESS DESCRIPTION
The Parsons FGC Process includes the following process steps:
¦ Simultaneous catalytic reduction of sulfur oxide (S0X) to hydrogen
sulfide (H2S), nitrogen oxides (N0X) to elemental nitrogen (N2), and
residual oxygen to water in a single reduction step.
¦ Recovery of H2S from the hydrogenation reactor effluent gas.
¦ Production of elemental sulfur from ^S-rich gas.
A process block flow diagram and a process flow sketch for a typical Parsons FGC
plant are presented in Figures 1 and 2.
Boiler operation is controlled to produce a flue gas with low residual oxygen
content. The controlled-oxygen content flue gas feed to the FGC plant exits the
boiler's economizer, passes through a multicyclone assembly where large ash
particles are removed, and is then mixed with steam-methane reformer gas and
sulfur plant recycle tail gas to form the feed to the catalytic hydrogenation
reaction module where the S0X, N0X, and residual oxygen are reduced. A
proprietary honeycomb catalyst is mounted in the flue gas duct to permit passage
5B-142
-------
of flue gas particulates with low pressure drop and nil fouling tendency. The
hydrogenation step is an extension of the Beavon Sulfur Removal (BSR) process,
developed jointly by Parsons and Unocal in the early 1970's, to treat particulate
containing flue gas. Seventy BSR plants have demonstrated the commercial
reliability of the hydrogenation, H2S recovery, and sulfur production steps in
tail gas plants.
The hot hydrogenator effluent is used to preheat the boiler combustion air in a
"nil-leak" heat pipe, or equivalent, air preheater. Essentially all of the
remaining ash is then removed from the cooled flue gas in the electrostatic
precipitator and the ash-free flue gas is fed to a direct-contact desuperheater.
Here, the flue gas is cooled and a small amount of water is removed by
condensation. Blovdown from the desuperheater circulating aqueous stream is
filtered to remove traces of fly ash and is subsequently steam stripped to remove
dissolved H2S; it can then be disposed of by severing.
The cooled effluent flue gas from the desuperheater enters an absorption column
containing an l^S-selective solvent. The process (FLEXSORB) is licensed by EXXON.
Essentially all of the H2S and a portion of the CO2 in the flue gas is absorbed by
the solvent. The absorber effluent gas, containing less than 10 ppmv ^S, is
vented to the atmosphere through a stack mounted atop the absorber. The effluent
stack gas is saturated with water vapor; reheat, as required, added to the
absorber effluent to reduce the length and frequency of occurrence of a steam
plume.
The H2S- enriched solvent leaves the bottom of the absorber and enters the
regenerator where it is heated and steam stripped to release the acid gases from
solution. The H2S-containing off gas exiting the top of the regenerator is sent
to a Recycle Selectox sulfur plant which converts the H2S to elemental sulfur;
this process is licensed jointly by Parsons and Unocal. The salable bright yellow
elemental sulfur is collected as a liquid product and the tail gas is recycled to
the hydrogenation reactor for further sulfur recovery.
5B-143
-------
PROCESS DEVELOPMENT
The development of the Parsons FGC Process has included both the bench scale and
the pilot scale programs. Because the H2S recovery via selective amine and
Recycle Selectox sulfur production technologies are both commercial proven, the
objective of the development project was to confirm the applicability of the
hydrogenation process in treating coal-fired boiler flue gas.
Bench Scale Development
The objective of the bench scale work was to develop a catalyst system and to
establish reaction conditions to meet defined performance criteria. From 1987
till 1990, the University of Delaware (UOD), Center for Catalytic Science and
Technology, conducted more than 150 bench scale test runs. Two of the tested
catalyst systems achieved 100 percent conversion of both S0X and N0X, and the
third catalyst system achieved 100 percent conversion of S0X and 983! conversion of
N0X, at reactor temperature of 600°F and space velocity up to 5000 hr"^. These
three preferred proprietary catalyst active ingredients were prepared on ceramic
honeycomb substrates for the 1990 pilot test.
Pilot Scale Development
Pilot Plant Design. The performance of the catalytic hydrogenation reactor has
been tested in a pilot plant designed to process a flue gas slipstream from boiler
No. 6 of the St. Marys Municipal Power Plant located in St. Marys, Ohio. Boiler
No. 6 has a nameplate capacity of 10 MU and burns high sulfur, eastern Ohio
bituminous coal. The objective of the pilot plant is to confirm the high
percentage S0x-N0x reduction capability reported by the bench scale work. The
coal characteristic and flue gas composition for the St. Marys pilot plant design
are given in Table 1 and 2. A pilot plant flow diagram is presented in Figure 3.
Figure U is a photograph of the pilot plant installed at the St. Marys Municipal
Power Plant.
In the pilot plant, a 10,000-scfh (maximum) power plant boiler slipstream is fed
to a cyclone for large particulate removal and then to a fabric filter for further
removal of particulates. The flue gas is preheated and intimately mixed with a
reducing gas produced by controlled sub-stoichiometric combustion of natural gas
in oxygen. The oxygen content of the flue gas is reduced in the reaction furnace.
5B-144
-------
The reaction effluent gas is cooled in a waste heat boiler. The system permits
bypassing a portion of the reaction furnace product to permit accurate control of
the feed temperature to the hydrogenator. Two hydrogenator vessels are installed,
which can be operated singly, in parallel, or in series. Instrumentation was
provided to determine the hydrogenator product composition using effluent
analyzers to measure concentrations of H2S, SO2, N0X, CO, H2, and O2. The pilot
plant product is incinerated in the power plant boiler combustion zone prior to
returning to the power plant stack.
Pilot Plant Operation and Results. The pilot plant was commissioned during the
1989 Phase I campaign. In its initial operations, it successfully achieved 90-
plus percent S0X and 85-plus percent N0X reductions using proprietary honeycomb
substrate catalysts. Equipment performance and control instrumentation problems
were defined during the Phase I campaign.
The pilot plant was modified in 1990. The modifications were aimed at improving
equipment and instrument performance. Also, operating and analytical procedures
were refined. Subsequent Phase II test results showed that the system achieved
SOjj reduction of 99-plus percent and N0X reduction of 92 to 96 percent.
The University of Delaware prepared three different proprietary honeycomb
substrate catalyst systems for the pilot plant test. One modified commercial
catalyst system from another supplier was also tested at the pilot plant.
Catalyst charge consisted of two cubic feet of ceramic honeycomb substrate
catalyst inserted into a catalyst rack with dimensions of 12" X 12" X 26" in
height. The St. Marys pilot plant was operated 24 hours a day, 7 days a week.
A 6-week test was performed using a proprietary honeycomb substrate catalyst A.
The S0X and N0X conversions remained high and steady throughout the entire test
period. The S0X and N0X conversions as a function of the average reactor bed
temperature operating at a space velocity of 2,500 hr~^ are plotted in Figure 5.1
and 5.2, respectively.
Another important independent variable studied during the pilot test runs was
excess hydrogen and its effect in S0X and N0X conversions. Figure 5.3 and 5.4 are
plots showing the S0X and NO]^ conversions as a function of excess hydrogen at the
reactor outlet. As it is shown in Figure 5.3, the high S0X conversion (99-plus
percent) is essentially independent of the amount of excess hydrogen provided to
5B-145
-------
the reaccor. The high NOx conversion is somewhat: dependent: on the amount of
excess hydrogen. As shown in Figure 5.4, the NO^ conversion decreased from 97-
plus percent at about 1.0 mole percent hydrogen at the reactor outlet to 94-plus
percent at about 0 mole percent hydrogen.
A summary of the key results for catalyst A follows:
Pilot Plant Test Results - Catalyst A
SOjj reactor inlet, ppmv 2,000-3,000
SOjj reactor outlet, ppmv 0-20
SOjj conversion, Z 99+
N0X reactor inlet, ppmv 300-600
NOx reactor outlet, ppmv 10-30
NOx conversion, Z 92-96
Space velocity, Hr up to 3,000
Average reactor temperature, °F up to 730
When the test was completed, the catalyst blocks were examined for the effects of
particulate on honeycomb openings. All catalysts blocks were found to be
essentially clean and free of particulate plugging.
Another test was also completed using a different catalyst system designated as
catalyst B. The test results of catalyst B were very similar to catalyst A in S0X
and N0X conversions.
The conversions of S0X and N0X remained high and steady throughout the entire test
period. Figure 6.1 and 6.2 are plots showing S0X and N0X conversions versus
average reactor temperature, respectively, operating at a space velocity of 3,000
hr"^". The slope of the S0X conversion curve is much steeper than the N0X
conversion curve. This implies that the S0X reduction is more temperature
dependent than the N0X reduction.
Correlations between excess hydrogen at the reactor outlet and the reduction of
S0X and N0X are. shown in Figure 6.3 and 6.4, respectively. As in the case of
catalyst A, the S0X conversion for catalyst B is essentially independent of the
amount of excess hydrogen at the reactor outlet. However, the N0X conversion is
somewhat dependent on the amount of excess hydrogen. The N0X conversion decreased
from 95-plus percent at about 1.0 mole percent hydrogen at the reactor outlet to
90-plus percent at about 0 mole percent hydrogen.
5B-146
-------
A summary of Che key results for catalyst B follows:
Pilot Plant Test Results - Catalyst B
SOjj reactor inlet, ppmv 1,800-2,800
S0X reactor outlet, ppmv 0-20
S0X conversion, Z 99+
NOx reactor inlet, ppmv 300-600
NOx reactor outlet, ppmv 10-30
N0X conversion, Z 92-96
Space velocity, hr up to 4,000
Average reactor temperature, °F up to 810
The effect of high particulate loading in the flue gas on the performance of
ceramic honeycomb substrate catalyst was examined during the last 2 days of test
runs. During the high dust loading test, the Gore-tex membrane/Teflon B
fiberglass fabric filter bag house was completely bypassed. No changes in the
performance of catalytic S0X and N0X reductions were observed during the high dust
loading test. At the end of the catalyst B test run, the reactor was open for
inspection. The honeycomb catalyst blocks were examined and the openings were
found to be free of particulate plugging.
Operating parameters were also defined during the pilot test runs to convert 100
percent of the incoming S0X to H£S in the catalytic reactor with zero elemental
sulfur formation, since the presence of elemental sulfur at the reactor effluent
would cause solids deposition and potential plugging problems in the transfer
lines and equipment.
CONCLUSION
Based on the bench scale and the most recent Phase 11 pilot plant test results, it
is concluded that:
¦ The bench scale tests confirm that 99-plus percent conversion of both
S0X and N0X is possible for properly controlled conditions and preferred
catalyst selection.
¦ Two different preferred proprietary honeycomb substrate catalyst systems
produced 99-plus percent conversion for S0X and 92 to 96 percent
conversion for N0X in the pilot test.
a The honeycomb substrate catalyst system provided Tn-tnimmTt pressure drop
and was capable of allowing passage for particulates without plugging
the openings during the high dust loading test.
5B-147
-------
Pilot plane tests demonstrated that very low excess reductant (H2) was
required to achieve high conversions of S0X and N0X. This can
significantly reduce the hydrogen gas requirement and subsequently the
capital and operating costs of the FGC plant.
Operating parameters were defined to achieve 100 percent reduction of
SO2 to H2S. No elemental sulfur was formed in the reactor.
The pilot plant operating experience gained regarding system chemistry,
equipment and instrument performance will provide the basis for the
demonstration and commercial scale plant design and operation.
5B-148
-------
ACKNOWLEDGEMENTS
The bench scale and pilot scale projects reported here were co-funded by the Ohio
Coal Development Office, Columbia Gas System Service Corporation, Delmarva Fover
and Light Company, Consolidated Natural Gas Service Company, and The Ralph M.
Parsons Company. Gratitude is extended to the City of St. Marys, Ohio, for
providing the pilot plant test site at its Municipal Power Plant. In addition, we
would like to thank V. L. Gore & Associates, Inc. for supplying the fiberglass
fabric filters.
5B-149
-------
TABLE 1
Sc. Marys Pilot Plant
Coal Characteristic - Ultimate Analysis
Comoonent
WtZ
Moisture
12.0
Carbon
57.5
Hydrogen
3.7
Nitrogen
0.9
Sulfur
4.0
Chloride
0.1
Ash
16.0
Oxygen
5,8
100.0
Btu/lb HHV (wet)
10,100
Sulfur content,
lbs/10 Btu
3.96
TABLE 2
St. Marys Pilot Plant
Flue Gas Composition
Comoonent
Wrt *
Nitrogen
73.4
Oxygen
3.25
Water Vapor
9.2
Carbon Dioxide
13.8
Sulfur Dioxide
0.35
Nitrogen Oxides
600
Ash
10,615
5B-150
-------
Flue Ga9
Recycle Tall Ga9
H2S
Rich
Stream
Elemental
Sulfur
Coal Boiler
Flue Gas
Syngas
Steam
Natural
Gas
Selective
H2S Recovery
Unit
Sulfur
Recovery
Unit
SOx/NO,,
Hydrogenatlon
Unit
Reformer
Figure 1. Parsons FGC Process Block Flow Diagram
-------
SULFUR PLANT TA1LGAS RECYCLE
' HYOROCENATION
REACTIOH
MODULE
ECONOMIZER
TREATED FLUE CAS
TO ATMOSPHERE
AIR
^ PREHEATER
I ' t^]l~
SULFUR
PLANT
SULFUR
SOUR
WATER
STRIPPER
ABSORBER
BOILER
COAL
ESP
STM
LARGE ASH
PARTICLES
STM
ASH
ASH
AIR
STRIPPED
WATER
SYNGAS
STEAU
REFORMER
Figure 2. Parsons FGC Process Flow Sketch
-------
Combustion RmcUoh EffliMnl HydrtfgtnaHon
Cydeoi Bm Houm Bumw Chtntbw Furmo Coolw Hmcioh
Coil bollw
(Iim gat
Niturtl 0i»
CZ5—
•I >
Tinted
flU# flit
|M
Figure 3. Parsons F6C Process Pilot Plant Flow Diagram
-------
Figure 4 - Photo of Pilot Plant Installed at St. Marys, Ohio
5B-154
-------
CATALYST A
REACTOR TEMP. VS SOX CONV. (S.V.-2500)
100
97
96
. 93
92
90
630
650
590
610
670
690
AVG. REACTOR TEMP.. DEG F
Figure 5.1 - Catalyst A - Reactor Temperature vs. S0X Conversion
590
CATALYST A
REACTOR TEMP. VS NOX CONV. (S.V.-2500)
610 630 650
AVG. REACTOR TEMP.. OEC F
670
690
Figure 5.2 - Catalyst A - Reactor Temperature vs. N0X Conversion
5B-155
-------
CATALYST A
EXCESS H2 VS SOX CONV. (S.V.-2500)
100
m
99
96
97
96
95
94
93
92
90
0
0.2
0.6
0.6
1.2
1.6
1
1.8
2
1.4
H2 CONC. © REACTOR OUTLET. MOL 3 DRY
Figure 5.3 - Catalyst A - Excess H£ vs. S0X Conversion
CATALYST A
EXCESS H2 VS NOX CONV. (S.V.-2500)
100
99
96
97
96
C O
On
95
94
93
92
91
90
0
2
3
1
4
H2 CONC. O REACTOR OUTLET. MOL % (DRY)
Figure 5.4 - Catalyst A - Excess H2 vs. N0X Conversion
5B-156
-------
CATALYST B
REACTOR TEMP, vs SOX CONV.(S.V.-3000)
D DOB
100
to
99
CD
98
97
95
93
92
90
770
710
730
750
650
670
690
AVC. REACTOR TEMP.. DEC F
Figure 6.1 - Catalyst B - Reactor Temperature vs. S0X Conversion
CATALYST B
REACTOR TEMP. vs NOX CONV.(S.V.-3O00)
650 670 690 710 730
AVC. REACTOR TEUP.. DEC F
750
770
Figure 6.2 - Catalyst B - Reactor Temperature vs. N0X Conversion
5B-157
-------
100
99
98
97
K
96
>
Z
o
95
u
X
o
in
94
93
92
91
90
CATALYST B
EXCESS H2 vs SOX CONV.(S.V.-JOOO)
0.2 OA 0.6 0.6 1 1.2
H2 CONC. e REACTOR OUTLET. MOL % (DRY)
Figure 6.3 - Catalyst B - Excess H£ vs. S0X Conversion
o
z
100
99
98
97
96
95
94
93
92
91
90
CATALYST B
EXCESS H2 VS NOX CONV.(S.V.-3000)
0.2 0.4 0.6 0.8
1.2 1.4 1.6
H2 CONC. © REACTOR OUTLET, MOL X (DRY)
Figure 6,4 - Catalyst B - Excess H£ vs. N0X Conversion
5B-158
-------
Session 6A
WET FGD OPERATING ISSUES
PILOT-SCALE EVALUATION OF SORBENT INJECTION
TO REMOVE SO, AND HC1
Radian Corporation
8501 MoPac Boulevard
Austin, Texas 78759
Joseph R- Peterson
Andrew Burnette
Gordon Mailer
Richard G. Rhudy
Electric Power Research Institute
3412 Hillview Avenue
Palo Alto, California 94303
ABSTRACT
This paper presents the results of a pilot plant test program conducted at the
Electric Power Research Institute's {EPRI) High Sulfur Test Center (HSTC) as part
of EPRI Research Project 2250-3 to investigate the feasibility of injecting dry
alkaline materials into flue gas upstream of the ESP for removal of gaseous SO, and
Four sorbents were tested: commercial hydrated lime; high-surface-area hydrated
lime; commercial-grade sodium bicarbonate (NaHCO,); and activated alumina. Condi-
tions which were varied during the test program included the sorbent injection
rates, flue gas flow rate, temperature, ESP specific collection area, and SO, and
HC1 concentrations.
Test results showed that the SO, removal was greater than the HC1 removal for all
sorbents and process conditions evaluated. For a given sarbent, the most important
parameter for SO, removal was the sorbent injection rate, which agrees well with
the predictions from a simple mathematical model. For SO, removal, the commercial-
grade NaHCOi, and the regular and high-surface-area hydrated limes performed about
the same when compared on a weight basis. However, at high injection rates, the
hydrated limes degraded the operation of the ESP, causing both the outlet opacity
and outlet mass loading to increase. The operation of the ESP improved when NaHCO,
was injected compared to baseline operation. The Injection of activated alumina
did not appear to affect the operation of the ESP, but the sorbent was relatively
unreactive towards SO, and HCl.
HC1.
6A-1
-------
INTRODUCTION
The results of a test program to evaluate the technical feasibility of removing
sulfuric acid vapor (HjSO^) and hydrochloric acid (HC1) vapor from flue gas by
injection of dry sorbents are presented in this paper. The testing was performed
at the Electric Power Research Institute's (EPRI) High Sulfur Test Center (HSTC)
located at the New York State Electric and Gas' Kintigh Station near Barker, New
York. The testing was sponsored under EPRI Research Project 2250-3. Additional
funding for the sorbent injection study presented in this paper was provided by
Kansas City Power & Light Co. and Louisville Gas & Electric.
Sorbent injection technology involves the injection of a dry alkaline sorbent into
a flue gas duct upstream of a particulate control device (e.g., ESP, baghouse, or
particulate scrubber). The application of this technology has the potential to
reduce stack plume opacity resulting from condensed sulfuric acid droplets in the
stack exit gas. The presence of these very fine droplets (ranging from about 0.1
to 0.5 micron) can significantly affect visual opacity.
The application of sorbent injection technology can also remove vapor-phase HC1.
For utilities operating wet FGO systems, removing HC1 upstream of the scrubber can
reduce the soluble chloride concentration of the scrubber recirculation liquor.
This has the potential to improve scrubber performance (e.g., removal efficiency,
limestone utilization) and to reduce the corrosion tendencies of scrubber materials
of construction. If this technology could be successfully applied to new FGD sys-
tems, the use of less expensive materials of construction may be possible.
The current study was a follow-on to an earlier evaluation of this technology by
EPRI, performed at the HSTC. Although very limited in scope, the earlier study
suggested that removal of HjSO^ and HC1 was feasible and that some of the operating
6A-2
-------
variables that may affect removal efficiency included the type of sorbent, the
addition rate of the sorbent, the sorbent duct residence time, and the gas tempera-
ture at the point of Injection. The primary objective of the current study was to
perfora a more exhaustive evaluation of sorbent injection technology 1n an attempt
to more fully characterize the process.
CHEMICAL REACTIONS
Although the SO, in the flue gas at the process conditions existing at the HSTC is
actually present primarily as sulfuric acid vapor (HjSOJ, 1t 1s more convenient to
discuss the chemical reactions as if the SOj is the true chemical species. There-
fore, the term SO, is used throughout this paper in place of HjSO^.
SO, (g) + Ca(0H)2 (s) —-> CaS0< (s) + ^0 (g) (1)
2HC1 (g) + Ca(0H)2 (s) —-> CaCl2 (s) + 2^0 (g) (2)
SOj (g) + 2NaHC0j (s) > Na^ (s) + J^O (g) + 2C02 (g) (3)
HC1 (g) + NaHCOj (s) > NaCl (s) + fy) (g) + C02 (g) (4)
3SO3 (g) + A120j (s) —-> A12(S0J3 (s) (5)
6HC1 (g) + A120j (s) > 2A1C13 (s) + 3H20 (g) (6)
In addition to the above reactions, all cf the sorbents have the potential to react
with S02 and C02, both of which are present in flue gas. These reactions are not
expected to have a significant effect on the tests at the temperatures evaluated.
Therefore, they are not addressed in this paper.
While the above reactions are known to proceed and produce relatively stable prod-
ucts, 1t was not known whether the overall rates would be sufficient to remove SO3
and/or HC1 in a cost-effective manner at typical flue gas conditions.
6A-3
-------
TEST METHODOLOGY
Process Diagram/Description
A simplified process flow diagram for the pilot unit showing the configuration used '
for the current test program 1s presented in Figure 1. Flue gas was extracted
approximately Isokinetically from the outlet duct on the Kintigh Station boiler for
use at the HSTC. This test program was conducted on the HSTC 4-MW spray dryer/ESP
pilot unit flow path. The flue gas passed through the spray dryer vessel (which
was not in operation during this program) and then proceeded to the outlet duct
where sorbent injection occurred. The flue gas then passed through a five-field
ESP for sorbent and fly ash removal. For most of the tests, only the first three
fields were energized for an SCA of about 300 ft2/kacfm. A few tests were also
conducted with two fields for an SCA of 200 ft2/kacfm. After the ESP, the flue gas
was returned to the Kintigh Station ductwork.
The normal sulfur content of the coal fired at Kintigh (2.8%) produces a flue gas
S02 concentration of about 1600 to 1800 ppmv and a SO, concentration of about 10 to
15 ppmv.
The chloride content of the Kintigh Station's coal (0.1%) produces a flue gas HC1 1
concentration of about 50 to 55 ppmv. For most of the current tests, the inlet HC1
concentration remained at the baseline level, but for a few tests, it was increased
to approximately 100 ppmv by spiking the flue gas with anhydrous HC1.
The flue gas SO, concentration was varied for many of the tests by spiking with
SO,. The SO, was produced by passing an SO^air mixture over a vanadium catalyst
at 800*F. The S02 content of the SO^air mixture was changed to alter the amount
of SO, that was injected into the gas stream. The SO, was injected into the flue
gas just upstream of the spray dryer vessel.
The gas flow rate and temperature at the outlet of the spray dryer vessel were
controlled to their desired setpoints using a variable-speed fan and an electric
heater. The flue gas S02 and 02 concentrations were measured at the spray dryer
inlet, the spray dryer outlet, and the outlet of the ESP to determine if measurable
S02 removal occurred and to correct the measured concentrations and calculated
removals for air inleakage into the system.
6A-4
-------
Sorbent was graviaetrlcally fed Into small hoppers and then pneumatically conveyed
Into the flue gas downstream of the spray dryer vessel. A small weight loss feeder
with a self-contained hopper was used for the low sorbent flow rates (3 to 50
lbs/hr). A 4-inch weigh belt and a sorbent silo were used for the higher flow
rates (32 to 160 lbs/hr) and for overnight tests.
Gas Sampling
The SO, concentration in the flue gas was determined by a controlled condensation
technique. This technique involved pulling a sample of flue gas through a heated
filter, then through a glass condenser which was maintained at 140*F. This temper-
ature was below the SO, dewpoint but above the water dewpoint. As a result, SOj
and not water condensed on the walls of the glass condenser. Condensation appeared
as a visible "fog" In the condenser. The gas sample then entered a set of impin-
gers designed to remove gaseous HC1 and water vapor. The sample then exited
through a pump and a dry gas meter.
For this study, the process inlet and outlet flue gas streams were sampled simulta-
neously for at least 30 minutes, which was more than adequate to observe the con-
densation of SO, in the condenser. At the end of the sampling time, condensed SO,
was recovered by rinsing the condenser with about 60 mL of distilled water into
previously weighed sample bottles. The SO, concentrations in the flue gas streams
were determined by analyzing the samples for sulfate (by ion chromatography) and by
recording the amount of gas sampled (i.e., from the dry gas meter readings).
The HC1 concentrations in the inlet and outlet flue gas streams were determined by
two methods. For most tests, the impinger solutions from the inlet and outlet flue
gas samples were analyzed for chloride by ion chromatography. An infrared HC1
monitor was also used to continuously measure the HC1 concentration in the flue gas
at the outlet of the ESP. This monitor was checked with span gases and found to be
quite accurate over the concentration range of interest (less than 150 ppmv). The
HC1 concentrations measured by the impingers did not agree well with those measured
by the monitor. However, from past experience with the monitor on the spray dryer
system, it 1s believed that the HC1 concentrations determined by the monitor better
represent the true HC1 concentrations in the flue gas. This monitor has shown that
the HC1 concentration in the flue gas at the inlet to the HSTC is normally about 52
ppmv when the Kintigh power plant is near full load. Since all of the current
6A-5
-------
tests were performed when the power plant was close to full load, the removal of
HC1 across the pilot system was deteralned using the ESP outlet concentration mea-
sured by the HC1 Monitor and an assumed Inlet concentration of 52 ppmv.
The spiked flue gas was sampled for SO, and HC1 just downstream of the spray dryer
vessel but upstream of any sorbent Injection. Sampling at this location supplied
the inlet SO, and HC1 concentrations (i.e., before any sorbent injection). The
outlet concentrations were measured by sampling at two locations: imediately up-
stream of the ESP; and downstream of the ESP and induced draft fan. By sampling
simultaneously at all three locations, which was done for a selected number of
tests, one could determine the SO, and HC1 removal occurring across the flue gas
ductwork and across the combination of the ductwork and the ESP.
Reagent Properties
Four reagents were tested: commercial hydrated lime; a special high-surface-area
hydrated lime; commercial-grade NaHCO,; and activated alumina. Samples of each
reagent were taken twice each day when that particular sorbent was being injected
into the ductwork. Selected samples were analyzed for specific surface area (using
a one-point BET method) and for sorbent particle size. A summary of the reagent
properties is presented in Table 1.
Experimental Conditions
The experimental conditions for the current study are summarized in Table 2. Host
of the tests were conducted at an ESP inlet temperature of 315"F and an inlet flue
gas flow rate of 13,600 acfm. For almost all of the tests, only the first three
fields of the five-field ESP were energized. At that flue gas rate, operation with
three fields yielded a specific collection area of about 300 ft2/kacfm. Throughout
the program, the first field was rapped every 5 minutes, the second every 10 min-
utes, and the third every 20 minutes. The last two fields, which were not ener-
gized, were rapped every 20 minutes.
Host of the tests lasted less than 2 hours. For these tests, the system was
allowed to equilibrate for about 15 minutes after the sorbent flow was initiated
prior to beginning data collection. The equilibration time period was chosen based
on data from the continuous HC1 analyzer which sampled the gas exiting the ESP.
6A-6
-------
These data showed that the HC1 concentration stabilized about 15 minutes after the
sorbent injection rate was changed.
DISCUSSION OF RESULTS
This section discusses the results from the current program. First, a theoretic-
ally based model developed to aid in data Interpretation is discussed, then the
measured results are discussed in light of the model.
Mathematical Model Development
The results of this program are best interpreted by a theoretically based model
which was developed for predicting SOj removal as a function of operating condi-
tions. With this model, SOj removal can be predicted for other locations and other
operating conditions.
At the high reagent ratios tested in this program [e.g., Ca(0H)2-to-S03 molar
ratios ranging from 2 to over 40], it is likely that the rate-controlling step for
SOj removal was the diffusion of SO, from the bulk gas to the sorbent particles.
Therefore, a gas diffusion model was developed to compare the measured SO, removal
to that predicted by the model and to see if any knowledge could be gained by exer-
cising the model for a variety of conditions.
The development of the model assumed:
• A large excess of reagent was present, relative to the amount of SOj
removal;
• All of the resistance to mass transfer occurred in a thin film sur-
rounding the particle;
• The competing reactions of HC1, S02, and C02 with the sorbent parti-
cles were not important;
• The sorbent particles were spherical with smooth external surfaces
(I.e., internal or pore surface areas did not contribute to the
overall reaction rates at the relatively low sorbent conversion
efficiencies);
• The average particle diameter accurately approximated the true dis-
tribution of sorbent particle diameters;
• The particles were well dispersed in the flue gas at all times;
6A-7
-------
• There was no net velocity between the particles and the flue gas;
and
• Constant temperature and pressure were maintained.
Since the particles were assumed to be well dispersed in the flue gas, the problem
could be reduced to a single particle associated with some amount of flue gas.
Therefore, the modeling process involved calculating the volume of flue gas per
particle, then calculating the rate of diffusion, or flux, of SO3 to that particle.
In its general form, the flux of SO, to the sorbent particle is given by:
"so3 " kg • A • Ct • (y^ - y^jj %uM) (7)
where: Njqj - the flux of SO, to the sorbent particle (gram moles SOj/sec);
kg - gas-phase mass transfer coefficient (cm/sec);
A - external surface area of the particle (cm2);
Ct - concentration of flue gas (gram moles total gas/cm3);
^sq3 buu " ¦l0^e faction of SO3 in the bulk gas (moles SOj/total moles
gas); and
^sa surf m mole faction of SO3 at the sorbent's surface
Since the model assumes that the rate-controlling step for the SO, removal process
is the diffusion of SOj through a thin film surrounding the sorbent particle, the
concentration of SO3 at the surface of the particle must be zero and the flux
expression is reduced to:
"sta " kg • A • Cjaj bulk " ^g * * * * ^*S03 bulk
where: dp - diameter of sorbent particle (cm); and
csou bulk " concentration of SO3 in the bulk flue gas (gram moles SOj/cm3).
The above equation can be rearranged and solved analytically to give:
6A-8
-------
Percent SQj Removal « 100-
1 - exp
-2* • D,
KB
(9)
Pp ' dp ' G
where: D^j - (v) diffusion coefficient (cmfysec);
N - sorbent injection rate (g/sec);
pp - particle density (g/cm1);
G - total gas flow rate (actual cmVsec);
t - reaction tine (sec).
A very important parameter in this equation Is the diameter of the sorbent parti-
cle. The isodel assumes a single particle size, but all of the sorbents showed a
distribution of particle sizes. To correctly model the SOj removal data, the model
would have to integrate the removal occurring for each of the particle sizes.
Since this was beyond the scope of this study, an average value for the particle
diameter was used. Furthermore, the aerodynamic particle size (i.e., the actual
agglomerated particle size in the ductwork) is more important for modeling the SOj
removal process. The SOj removal data seemed to closely fit the diffusion model if
a particle diameter of 10 microns was assumed. As shown in Table 1, this assumed
diameter does not differ greatly from the average diameter determined from the par-
ticle size distribution data.
The time for the reaction between the sorbent and the flue gas is also an important
parameter in the modeling equation. Since the flue gas flow rate, duct length, and
duct diameter were well known for the current study, it was possible to accurately
determine the reaction time for the sorbent in the ductwork. However, since most
of the particles were removed in the first field of the ESP, it is difficult to
predict the total reaction time of the particles with the flue gas. For the model-
ing results presented in this study, it was assumed that the particles continued to
react with the flue gas in the ESP for a time equal to one-half of the flue gas
residence time 1n the first field of the ESP (i.e., 1.32 or 2.31 seconds, depending
upon the flue gas flow rate).
Another important parameter in the above equation is the diffusion coefficient for
SO3. Since the SO, is present as gaseous HjSO^ under typical flue gas temperatures
6A-9
-------
and flue gas moisture levels, the diffusion coefficient was estimated by calculat-
ing the diffusion coefficient of HjSO^ in nitrogen at the temperature and pressure
of the flue gas stream. The diffusion coefficient was calculated to be 0.169
cmVsec at 315*F using the method of Fuller, et al. (4). The diffusion coefficient
would increase with the absolute temperature of the flue gas (e.g., the diffusion
coefficient was calculated to be 0.183 cm2/sec at 350*F), so the model predicts
higher SO, removal at higher temperatures.
Sorbent Addition Rate - S03 Removal
For a given particle size, flue gas flow rate, and reaction time, the diffusion
model predicted that the only other important parameter for SO, removal is the
sorbent addition rate. This rate determines the number of particles injected into
the flue gas, and therefore, the volume of total gas surrounding each particle.
The model assumed that there was no effect of sorbent type. The experimental
results from the current test program agree fairly well with this model as shown in
Figure 2.
The data for the commercial-grade NaHCO, and the regular and high-surface-area
hydrated limes (Figure 2) show that the observed SO, removal agrees fairly well
with the model predictions at moderate sorbent injection rates. However, the model
tends to overpredict SO, removal at very high sorbent injection rates and to under-
predict SO, removal at low sorbent injection rates. For activated alumina, the
model seems to overpredict SO, removal at nearly all injection rates. There are
plausible explanations for the deviations from the model.
At very high sorbent injection rates, it is likely that the assumption of well-
dispersed sorbent particles is much less valid than at low injection rates. There-
fore, the model will tend to overpredict SO, removal. This is also supported by
the fact that the measured SO, removal never reached 100%, even at very high
sorbent injection rates.
At lower sorbent injection rates, the diffusion model tends to underpredict SO,
removal across the system. We speculate that this may result from some SO, removal
being caused by condensation of H2S04 at cold spots in the ESP. Since the ESP is a
pilot-scale unit, it has more external surface area per unit volume than a full-
sized unit. As a result, cold spots in the ESP are much more important on a
6A-10
-------
pilot-scale unit than on a full-scale unit. The background SO3 removal across the
pilot ESP was quantified by simultaneously measuring the SOj concentrations at the
system inlet and outlet when no sorbent was Injected Into the ductwork. The back-
ground removal ranged from 10X to 30% and seemed to increase with the inlet SO,
concentration. This observation 1s consistent with the background SO, removal
since higher inlet SO, concentrations create higher dew point temperatures, result-
ing 1n higher SO, removal.
Activated alumina is known to readily agglomerate, which, as for high injection
rates with the other sorbents, tends to increase the effective particle diameters
and lower the actual SO, removal. Thus, the model's overpredictions of the SOj
removal for activated alumina could be rationalized but not proven.
Sorbent Addition Rate - HC1 Removal
Figures 3 and 4 show the HC1 removal data for the four sorbents at a gas temper-
ature of 315*F. These data are quite different from the SOj removal data for
several reasons:
• The magnitude of the HC1 removal was less than that for the SOj
removal.
• The effect of sorbent type on HC1 removal was more pronounced than
for SOj removal when the sorbents were compared on a mass basis
(Figure 3). Comparing the data on a reagent ratio basis caused most
of the data to collapsed onto one curve (Figure 4).
• The shape of the removal versus sorbent injection rate curve was
more linear.
These data suggest that HC1 removal was not limited by gas-phase diffusion. Some
other mechanism evidently controlled HC1 removal. The data were not sufficient to
prove which mechanism controlled the overall reaction, but it is easy to fit the
data if one assumes that the overall reaction was controlled by the kinetics of a
first-order reaction between the sorbent and HC1. The curves shown in Figures 3
and 4 were the results of fitting a first-order reaction rate expression to the
data.
6A-11
-------
Inlet SOj Concentration
For a given sorbent muss Injection rate, the diffusion model predicted that there
was no effect of inlet SO, concentration on the percent SO, removal for the sorbent
injection process. The data taken in this test program appear to agree with this
prediction. However, for a given reagent ratio (i.e., moles sorbent/moles SO,),
the data show that the percent SO, removal was higher for higher inlet SO, concen-
trations (Figure 5). This trend was also predicted by the diffusion model and can
be explained by noting that, for the same reagent ratio, more particles must be
injected into the flue gas for the higher inlet SO, concentrations than for the
lower concentrations. Therefore, less gas volume is associated with each particle
at high inlet SO, concentrations, and the distance that the SO, has to diffuse to
reach the sorbent particles is reduced.
Flue Gas Temperature
Another objective of this test program was to evaluate the effect of flue gas tem-
perature on SO, and HC1 removals for the sorbent injection process. Host of the
experiments were completed at a flue gas temperature of 315*F (ESP inlet tempera-
ture). Additional tests were performed at 350'F at the same sorbent residence
time.
The data from the tests showed no significant effect of flue gas temperature on SO,
and HC1 removal levels. The diffusion model predicted a slight increase in the SO,
removal when the temperature was increased to 350*F because the diffusion coeffi-
cient of H2S04 increases with temperature. It is likely, however, that in this
pilot-scale system, the predicted increase in SO, removal was negated by the
decrease in SO, removal due to cold spots in the ESP. The cold spots become less
effective as the gas temperature is increased.
Sorbent Residence Time
The effect of sorbent residence time on the SO, removal level was investigated by
injecting the sorbent into the flue gas at a location closer to the ESP inlet
(Figure 1). Injecting at this point decreased the duct residence time from 2.0 to
1.3 seconds and the total (duct plus ESP) estimated residence time from 3.3 to 2.6
seconds. Figure 6 shows the effect of changing the sorbent injection location on
the SO, removal obtained with the commercial hydrated lime. No significant effect
6A-12
-------
of changing the total residence tine from 3.3 to 2.6 seconds was observed. Any
change in SO, reaoval with residence tlae was apparently within the ability to
¦easure the SOj reaoval, which was estlaated to be ±5%.
Sorbent Surface Area
The data froa the current study show that the high-surface-area hydrated llae
performed no better than the commercial-grade hydrated line, even though the high-
surface-area hydrated llae had almost twice the surface area (35 vs. 20 uP/g). In
addition, the activated alwnina had a very high surface area (170 nVg) but was
auch less reactive towards SO, and HC1. Sodium bicarbonate had the lowest surface
area of all the sorbents tested (3 mVg), but It performed as well as the hydrated
Haes. However, the sodium bicarbonate thermally decomposes 1n the flue gas to
form higher-surface-area sodium carbonate; the measured value of about 3 mVg
probably understates the actual reactive surface area of the reagent after It Is
Injected Into the flue gas. Even so, data from the literature (1) Indicate that
the surface area of the thermally decomposed NaHCOj Is probably much less than
those of the other sorbents.
The apparent lack of dependance upon sorbent specific surface area agrees with the
predictions of the diffusion model. The model states that the Internal surface
area of a sorbent particle Is not Important since the SO, removal process Is
assumed to be limited by the diffusion of SO, from the bulk gas to the external
surface of a sorbent particle.
EFFECT ON ESP OPERATION
One goal of this test program was to determine the effects of sorbent Injection on
ESP operation. The Injection of alkaline sorbents Into the flue gas upstream of
the ESP can affect the operation of the ESP due to the Increased mass loading,
changes In the overall particle size and resistivity, and the removal of SO, which
1s a known ESP conditioning agent.
The following measurements were made for baseline (I.e., fly ash only) and sorbent
Injection conditions:
• Voltage-current relationships for each field of the ESP;
6A-13
-------
• Continuous flue gas opacity measurements at the outlet of the ESP;
and
• Flue gas mass loadings at the outlet of the ESP.
Host of these Measurements were conducted when three of the ESP fields were ener-
gized, corresponding to a specific collection area (SCA) of 300 ft2/kacfn. Some
measurements were performed when only two ESP fields were energized (SCA of 200
ft2/kacfm) to simulate a smaller ESP. The results of these measurements are dis-
cussed below.
Results of ESP Testing
Early In the program, it was observed that injecting hydrated lime at high flow
rates (greater than 2.2 lb/hr/1000 acfm) had adverse effects on the operation of
the ESP. This was first evidenced by strong sparking in the first field of the
ESP. If the power to the first field was turned off to stop particle collection in
the first field, sparking immediately started in the second field of the ESP.
At lower hydrated lime injection rates (less than 2.2 lb/hr/1000 acfm), the sever-
ity of the sparking was diminished, but the voltage-current relationships in the
first field were still altered. The corona current in the first field was much
lower when the hydrated lime was injected than the during fly-ash-only conditions.
Since the lower corona current probably indicated a low particulate collection
efficiency in the first field, it was speculated that the second and third fields
would exhibit the same behavior as the first field if the sorbent injection con-
tinued for an extended period of time.
Several overnight tests were performed to investigate the effect of hydrated lime
Injection on the ESP for a longer time period. The ESP outlet opacity and outlet
mass loading are summarized in Table 3 and illustrated in Figure 7. Figure 7 shows
that the ESP outlet opacity increased when hydrated lime was injected at high flow
rates (greater than 2.2 lb/hr/1000 acfm). Figure 8 compares the voltage-current
relationships 1n the absence of sorbent injection with those for the injection of a
large amount of hydrated lime. These data were taken after the hydrated lime had
been injected continuously for about 36 hours. The data show that the hydrated
lime drastically reduced the operating current in the first two fields and produced
6A-14
-------
back-corona in the third field. The current nay be reduced to very low levels in
all fields if the sorbent were to be injected for a longer time.
The results from the outlet Bass loading tests (Table 3) tend to agree with the
outlet opacity measurements and appear to support the observation that ESP perfor-
mance deteriorates over time when large amounts of hydrated lime are injected into
the flue gas. For example, the outlet mass loading increased from 0.025 lb/MBtu
under fly-ash-only conditions to 0.068 lb/MBtu after hydrated lime was injected at
50 lb/hr/1000 acfm for approximately 48 hours. However, for the same injection
rate of hydrated lime, another outlet mass loading test showed a lower-than-
baseline outlet mass loading of 0.019 lb/MBtu. Data collection for this test began
approximately 1 hour after the start of the hydrated lime injection and lasted for
approximately 10 hours. As shown in Figure 7, the ESP outlet opacity during this
tiae period was relatively low until the very end of the mass loading test. About
11 hours after the start of the hydrated lime injection, the outlet opacity
increased to the relatively high level. The outlet opacity remained at this level
while the other mass loading test was conducted (the 0.068 lb/MBtu test). These
data indicate that the performance of the ESP degrades with time when hydrated lime
is injected at high injection rates.
The ESP outlet opacity returned to the baseline level soon after the sorbent injec-
tion was turned off. When the sorbent injection was restarted at a lower rate (1.0
lb/hr/1000 acfm), the opacity did not increase. However, as shown in Table 3, the
mass loading at the ESP outlet appeared to increase even at this low sorbent injec-
tion rate.
The sparking problems and the drastic altering of the voltage-current relationships
were not apparent when either activated alumina or NaHCOj were injected into the
flue gas stream. The outlet opacity also remained fairly constant while these sor-
bents were injected into the flue gas. In fact, the ESP outlet mass loading test
which was performed while NaHCO, was injected showed that the efficiency of the ESP
improved compared to that for fly-ash-only conditions. This result was somewhat
expected because sodium compounds are known conditioning agents for ESP's due to
their relatively low resistivity.
6A-15
-------
FULL-SCALE IMPLEMENTATION
The data from this study suggest that it is possible to reduce SO, levels and plune
opacity by injecting either hydrated lime or sodium bicarbonate into the flue gas.
These sorbents were equivalent for SO, removal when compared on a mass basis.
Costs for injecting these sorbents for SO, removal and plume opacity reduction were
estimated for a 300-MU, base-loaded power plant. To achieve an 80% reduction in
flue gas SO, levels from a moderate initial level (e.g., 20 ppm), about 7500 tons
per year of either sorbent would have to be injected. For hydrated lime reagent at
$65/ton, this would result in an annual sorbent cost of about $500,000, which is
equivalent to about 0.2 to 0.3 mil/kWh. For sodium bicarbonate reagent at
$200/ton, the annual sorbent cost would rise to about $1.5 million, which is equi-
valent to 0.7 to 0.8 mil/kWh. For either sorbent, a permanent sorbent storage and
injection system would be estimated to cost between $500,000 and $750,000.
The injection of these sorbents would slightly increase the volume of solid waste
produced by the plant. For a case with a 2.8% sulfur content and 8% ash content in
the coal, the sorbent injected would represent about 4% to 5% of the dry weight of
the combined ash and FGD sludge stream produced. Note that this represents about
twice the amount of hydrated lime generally used in cases where it is added to the
combined ash/FGD sludge stream for stabilization. In such cases, it may be pos-
sible to eliminate lime addition to the sludge if hydrated lime is used for SO,
control. Thus, for this circumstance, the net reagent cost for SO, control by
hydrated lime injection would be about half that of the estimate above, or only
about $250,000 per year for the example case.
The only drawback to using hydrated lime may be the potential adverse effects on
ESP performance. The magnitude of these effects will likely be site specific and
will depend greatly on the hydrated lime injection rate required. While sodium bi-
carbonate could be used instead to avoid any potential adverse effects on the ESP,
for the example case described above, the sodium bicarbonate reagent would be at
least three times more expensive than hydrated lime reagent. Also, the addition of
highly water-soluble sodium salts to the solid waste stream from the plant may be
undesirable.
6A-16
-------
CONCLUSIONS
Based on the results from the current test program, the following conclusions can
be drawn:
• The injection of alkaline sorbents will remove SO, and, to a lesser
extent., HC1 from power plant flue gas streams. However, care must
be taken to avoid ESP outlet particulate emission problems caused
by certain sorbents.
• For all of the sorbents tested, the removal of HC1 was much less
than the removal of SO,.
• For SO, removal, the commercial-grade NaHCO,, the commercial-grade
hydrated lime, and the high-surface-area hydrated lime all performed
about the same when compared on a weight basis. The activated
alumina was less reactive than these sorbents, even though it had a
much higher specific surface area.
• The SO, removal results from the current study agree fairly well
with the predictions of a simple gas-phase diffusion mathematical
model. This model predicted that the most important parameters for
SO, removal were the particle size of the sorbent, the sorbent
injection rate, and the sorbent residence time in the flue gas.
• The injection of large amounts of hydrated lime caused the ESP
outlet opacity and mass loading to increase. The voltage-current
relationships for the ESP were also significantly altered.
• A permanent sorbent storage and injection system would cost between
$500,000 and $750,000 for a 300-MW, base-loaded power plant. The
annual sorbent costs for obtaining 80% removal of a 20 ppm SO,
concentration in this plant would be about $500,000 (0.2 to 0.3
mil/kWh) and $1.5 million (0.7 to 0.8 mil/kWh) for hydrated lime and
sodium bicarbonate, respectively.
ACKNOWLEDGEMENTS
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 & Gas,
Empire State Electric Energy Research Corporation, Electric Power Development Cor-
poration, and the U.S. Department of Energy. In addition, partial funding of this
project was provided by Kansas City Power & Light and by Louisville Gas & Electric.
We also wish to acknowledge the donations of the regular and high-surface-area
hydrated lime reagents by the Chemical Lime Group of Fort Worth, Texas.
6A-17
-------
REFERENCES
1. Fellows, K.T. and H.J. Pflat. J. Air Waste Manage. Assoc.. 40 (6), pp. 887-
893, 1990.
2. Karlsson, H.T., J. Klingspor, and I. Bjerle. APCA Journal. 31 (11), pp.
1177-1180, 1981.
3. Uchida, S., S. Kageyama, M. Nogi, and H. Karakida. J. Chinese Institute of
Chem. Engineers. 10, pp. 45-49, 1979.
4. Reid, R.C., J.H. Prausnitz, and B.E. Poling. The Properties of Gases and
Liquids. Fourth Edition, McGraw-Hill, New York, New York, 1987.
Table 1
SUMMARY OF REAGENT PROPERTIES
Sorbent
Grade
Source
Average Surface Area
("2/g)
Avg. Particle Diameter
(pin)
cafom.
Coomercial
Chemical
Lime
20
14
CafOm.
High Surface
Chemical
Lime
35
NaHCO.
12
Commercial
Kerr McGee
11
—^1^3
Activated
Alcoa
170
6A-18
-------
Table 2
SUMMARY OF EXPERIMENTAL CONDITIONS
Lowest Value
Base-Case Value
Hiahest Value
Gas Flow Rate (acfai) 7,700
13,600
13,600
Gas Tenperature (*F) 305
315
350
Inlet SO, (ppav) 10
24
40
Inlet HC1 (ppmv) 52
52
100
Duct Residence Tine (sec) 1.0
2.0
3.5
Total Residence Tine (see)" 2.3
3.3
5.7
"Assuming particles are collected in the first
half of the first ESP field.
Table 3
SUTWARY DATA FROM ESP OUTLET MASS LOAD TESTS
Sorbent Tvoe
Sorbent Flow
flb/hr/lOOO acfm)
ESP Outlet
Loading
flb/MBtu)
Fly Ash Only
--
0.027
Fly Ash Only
--
0.025
Fly Ash Only
--
0.048"
Ca(0H)2 (1 hr after sorbent flow initiated)
3.7
0.019
Ca(0H)2 (48 hrs after sorbent flow initiated)
3.7
0.068
Ca(0H)2
2.3
0.084"
Ca(OH)2
1.0
0.065"
NaHCOj
3.7
0.009
'Performed with only two fields energized.
6A-19
-------
Sample Aittnutt Sorbent Sorbent
Port
Igjetiou LootioB
Gas
Outlet
Cu
Heater
Sample
Port
Sample
Port
Pemmtidc Precipitator
(ESP)
Spray
Drjer
Absorber
(noticed)
Sorbent
SDo
Weigh Belt Feeder
Meier
Blower
Figure 1. Pilot Unit Configuration for Sorbent Injection Experiments
100
ao
60
CO
>
e
E
o
E
O
CO
20
rtntloml Ca(OH).
Surfeca Ca
-------
100
80
m
>
o
E
o
E
40
u
X
20
12
0
10
2
8
4
6
Sorbent Infection Rat* (lbmr/1000 aeftn)
Figure 3. Experimental and model results for HC1 removal by injection of NaHCOj,
conventional hydrated lime, high-surface-area hydrated lime, and activated alumina.
Sorbents are compared on a mass basis. All data are at a gas temperature of 315"F.
100
60
>
o
E
CE
U
X
40
20
SO
60
0
40
10
20
30
Reagent Ratio to HCI (equhr. mol/mol HCQ
Figure 4. Experimental and model results for HCI removal by injection of NaHCOj,
conventional hydrated lime, high-surface-area hydrated lime, and activated alumina.
Sorbents are compared on a reagent ratio basis. All data are at a gas temperature
of 315'F.
6A-21
-------
100
• - O
80
60
a
>
o
E
o
E
o" 40
a>
mult* tor IS ppm
multi tor 8 ppm
20
50
40
30
20
10
0
Raagant Ratio (mol Ca/mol SOj)
Figure 5. Effect of inlet S0j concentration on SOj removal for conventional
hydrated lime. The model assumes 10 pm particles and a 3.28-second reaction time.
All data.are at a gas temperature of 315*F.
100
80
60
o
E
o
c
O* 40
CO
Ma tor U ne
Dan tor 2£ aac
20
4
5
6
7
8
0
2
3
1
Sorbant Injection Rat* (lb/hr/1000 actm)
Figure 6. Effect of sorbent residence time on S0j removal for commercial-grade
hydrated lime. The model assumes 10 pm particles. All data are at a gas tempera-
ture of 315*F.
6A-22
-------
1 2
1 0
8
6
4
2
3.7 Ib/ltrnooo
0
40
60
80
0
20
100
Elapsed Tim* (hours)
Figure 7. Effect of hydrated lime injection on flue gas opacity as measured at the
ESP outlet.
80
70
3rd FlokJ
with Urn* ln|*ction
3rd Flald Baaallna
60
50
40
30
20
2nd Flald with Urn* Infection
-to
1« Flald - No Currant
Lira* Injection
0
35
40
45
50
60
65
55
Voltage (kV)
Figure 8. ESP current-voltage relationships for baseline and hydrated lime injec-
tion conditions. Data for hydrated lime conditions were taken after hydrated lime
injection had been in progress for 48 hours at 3.7 lb/hr/1000 acfin.
6A-23
-------
Intentionally Blank Page
6 A-24
-------
CONTROL OF ACID HIST EMISSIONS FROM FGD SYSTEMS
Robert S. Dahlin
Southern Research Institute
P.O. Box 55305
Birmingham, AL 35255-5305
Thomas D. Brown
U.S. DOE/Pittsburgh Energy
Technology Center
P.O. Box 10940
Pittsburgh, PA 15236-0940
6A-25
-------
Intentionally Blank Page
6 A-26
-------
ABSTRACT
Acid mist emissions can be a significant problem at power plants burning high-
sulfur coal and using wet scrubbers for flue gas desulfurization (F6D). The
acid mist, which is formed by condensation of sulfuric acid vapor within the
scrubber system, can be a major contributor to particulate emissions. Since
the acid mist is predominantly submicron in size, it avoids capture in
conventional mist eliminator systems, and it scatters light very effectively.
This can result in excessive visible emissions in some cases.
Improved control of acid mist emissions can be achieved by replacing or
augmenting the conventional mist eliminators with a wet electrostatic
precipitator (WESP). This paper describes a two-phased study performed to
determine the degree of control that can be achieved with this approach.
Phase I was a study of the electrical operation of a lab-scale WESP collecting
an acid mist from a coal combustion pilot plant equipped with a spray chamber.
The results of this study were used to develop and validate a computer model
of the WESP. In Phase II, measurements were made at two utility scrubber
installations to determine the loadings of acid mist, fly ash, and scrubber
carryover. These measurements were used as input to the model to project the
performance of a retrofitted WESP.
6A-27
-------
INTRODUCTION
Acid mists can be a major source of corrosion problems and visible emissions
at power plants that burn high-sulfur coals and are equipped with wet flue gas
desulfurization (FGD) systems. When flue gas is rapidly cooled in an FGD
system, the S03 is condensed along with water vapor to form an ultrafine mist
of sulfuric acid. The mist droplets are so small that they escape collection
in the scrubber and the mist eliminators (HEs) (1). When discharged into the
atmosphere, these fine droplets scatter and absorb light very effectively,
sometimes resulting in excessive visible emissions. The presence of the acid
mist in the flue gas can also be a contributing factor in excessive corrosion
of the ducting and the stack liner downstream from the HEs. If a wet electro-
static precipitator (WESP) is used to replace or augment the HEs, the acid
mist loading can be substantially reduced, along with the associated corrosion
problems.
Under contract to the Department of Energy/Pittsburgh Energy Technology Center
(DOE/PETC), Southern Research Institute investigated the use of a compact UESP
to control acid mist emissions. The project was primarily directed toward
acid mist emissions from wet FGD systems, although other sources of acid mist
could be controlled by this approach. The goal of this investigation was to
assess the improvement in acid mist control that was possible by using a UESP
to replace or augment the existing HEs 1n an FGD system. The project was
organized in two Phases. Phase I was initiated in August 1988 and completed
in November 1989. It involved laboratory and pilot-scale studies of the WESP
concept, along with the development of a WESP computer model. Phase II was
completed in April 1991 and involved field measurements at utility FGD
installations, projections of WESP performance, and development of a WESP
demonstration plan.
6A-28
-------
PHASE I OBJECTIVES
The primary objectives of the Phase I effort were to (1) determine the
ability of a compact WESP to collect the fine acid mist, (2) determine the
effect of fly ash loading on mist collection efficiency, and (3) develop and
validate a computer model of the WESP to help interpret test results and
extrapolate results to full-scale applications. The second objective was
included because the fly ash loading leaving the scrubber can vary widely,
depending upon the performance of the upstream particulate control device.
Also, a very high loading of submicron fly ash could adversely affect WESP
performance by space charge suppression of the corona current.
PHASE I APPROACH
The approach used in Phase I of this project was to first fabricate a
laboratory-scale WESP that could be used to determine the expected WESP
fractional collection efficiency and provide data for validating a computer
model of the WESP. Since it was anticipated that the volatile acid mist could
present sampling difficulties, initial testing was done with a nonvolatile
simulant oil, di-2-ethylhexyl sebacate (DES). A sketch of the WESP setup used
for these tests is shown in Figure 1. After successful completion of these
tests, the WESP was modified and connected to a pilot-scale combustion system
to allow testing on an actual acid mist. The acid vapor was generated by
firing either S02-doped natural gas or a combination of S02-doped natural gas
with coal. This was done to allow testing of the WESP on the mist alone and
the mist in combination with a fly ash loading typically encountered down-
stream from a scrubber. The acid mist was formed by passing the flue gas
through a spray humidification chamber to simulate condensation in the
scrubber system. A sketch of the modified WESP setup used in the pilot
combustor tests is shown in Figure 2.
The data obtained from the tests with the DES and the actual acid mist were
used to validate the computer model after each series of tests. The validated
computer model was then used to make projections of WESP performance in a
utility retrofit situation.
6A-29
-------
LABORATORY TESTS WITH SIMULANT OIL
TTie WESP setup used in these tests consisted of a tubular WESP made from a
1/8-inch diameter wire suspended along the axis of a 8-1nch diameter galva-
nized metal tube. The energized length of wire was 3.5 feet. For the
laboratory tests, air was drawn through the WESP system at a nominal flowrate
of 100 cfm, resulting in a WESP specific collecting area (SCA) of about 74
ft2/kacfm. This may be compared to a typical fly ash precipitator having an
SCA of 250 to 350 ft2/kacfm. Thus, these tests were designed to simulate a
very compact WESP design.
The DES oil was atomized using a Sonic Development ST-47 nozzle operated at an
air pressure of 88 psig and an oil pressure of 12 psig. This typically pro-
duced an aerosol with a mass mean diameter of about 4 /on. Although this is
somewhat coarse for an acid mist, it provided an adequate concentration of
submicron particles for reliable size-resolved efficiency measurements.
Collection efficiency as a function of particle size was determined from
cascade impactor measurements made at the inlet and outlet of the WESP.
Typical results obtained from these measurements are compared with the results
of two alternate computer models, a current-specific model (2) and a current-
seeking model (3), in Figure 3. For the particle size range resolved in these
tests, the collection efficiency varied from about 97% for submicron particles
to 99.8% for 10 fan particles. These results were extremely encouraging and
showed good agreement with one of the two models initially considered.
The current-specific model was found to give better agreement with the WESP
performance data, because it allowed input of both the applied voltage and the
operating current. The current-seeking model predicts the current based on
the applied voltage and the particulate space charge. The equation that is
used for this is valid only in the region near corona onset (i.e., at rela-
tively low voltage and current). The current-seeking model does not do a good
job of predicting performance in this case since the actual voltage and
current (60 kV and 270 /xA/ft2) are far from the region of corona onset (about
30 kV and near-zero current).
6A-30
-------
PILOT COHBUSTOR TESTS WITH ACTUAL ACID MIST
For these tests, the WESP was connected to a pilot-scale coal combustion
facility equipped with a spray humidification chamber to serve as a source for
the condensed acid mist. The acid mist was generated by firing S02-doped
natural gas in the coobustor and condensing the resulting acid vapor into a
fine mist in the spray humidification chamber. Since this resulted in a
saturated flue gas entering the WESP, it was necessary to make extensive
modifications to the WESP setup to avoid electrical tracking along the high-
voltage insulator. As shown in Figure 2, a hot-air purge system was installed
to keep the high-voltage insulator dry, and a mist eliminator was added to
prevent the carryover of large unevaporated droplets into the WESP. The hot
purge air typically accounted for about half of the total gas flow through the
WESP. Since the mist eliminator would collect mostly large particles, it had
little effect on the acid mist fraction.
As in the laboratory tests, the size-dependent efficiency of the WESP was
determined by cascade impactor measurements at the inlet and outlet of the
WESP. Since the hot purge air was added downstream from the inlet sampling
location, the inlet loadings had to be corrected for this dilution. Blank
impactor runs were performed with each set of runs to ensure that no artifi-
cial weight gains resulted from flue gas interaction with the impactor
substrate material. The impactor substrates were also acid washed to neutral-
ize any alkaline sites that might adsorb S02 and cause a spurious weight gain.
Prior to each set of impactor runs, a measurement of the gas-phase S03 level
by the controlled condensation method was made to assure constant conditions.
To cover a range of acid mist concentration, two series of tests were conduct-
ed at nominal S03 levels of 25 ppm and 47 ppm. For these two series of tests,
the average inlet mass loadings of acid mist were 8.6 mg/acm (0.0038 gr/acf)
and 16.3 mg/acm (0.0071 gr/acf). These loadings were lower than expected for
complete condensation of the acid, possibly due to removal of some of the acid
vapor in the spray chamber. Nevertheless, the loadings showed the expected
variation with S03 level. A sunmary of the test results is given below.
6A-31
-------
Test Parameter
Low SOj
Hioh SCL
Initial SO, concentration, ppm
Inlet mass loading, ng/acm (gr/acf)
Inlet mass median diameter, pa
WESP applied voltage, IcV
ESP current density, pA/ft
Specific collecting area, ft /kacfm
Collection efficiency, %
25
8.59 (0.0038)
1.9
68
251
48.9
47
16.3 (0.0071)
1.5
68
196
40.3
Smaller than 1 ism
Smaller than 5 pn
87.4 to 92.9
88.5 to 93.0
62.1 to 83.0
71.4 to 91.8
These results show that the WESP is capable of good control efficiencies at an
SO, level of 25 ppm. However, the WESP performance degrades at the higher SO,
level of 47 ppm. Although this is partly attributable to a slight difference
in the specific collecting area, the primary factor is the reduction in cur-
rent density from 251 to 196 pA/ft2, a reduction of 22%. This results from
the increased particulate space charge and the concomitant suppression of
corona current.
The space charge effect could be seen very dramatically in the voltage-current
characteristics of the WESP, as illustrated in Figure 4. At an applied volt-
age of 50 kV, the operating current with no acid mist present was about 1.1
mA, compared to a current of about 0.4 mA with 25 ppm of SO, (8.6 mg/acm of
acid mist). With 47 ppm of SO, (16.3 mg/acm of acid mist), the current was
further reduced to about 0.35 mA at 50 IcV. In actual practice, it may be
possible to compensate for this effect to some degree by increasing the ap-
plied voltage. As shown in Figure 4, the voltage was actually increased to
over 80 kV without sparkover, but this was not considered to be a realistic
operating point for a commercial WESP.
It should be noted that all of the WESP testing with an actual acid mist was
done with a much lower SCA than that used in the laboratory tests with the DES
aerosol (40 to 49 versus 74 ft2/kacfm). This was done to provide a more real-
istic simulation of a very compact WESP that could be retrofitted onto a
scrubber. This difference in SCA, combined with the reduced current densities
(196 to 251 versus 270 /iA/ft2), account for the lower collection efficiencies
with the acid mist. The reduced current densities are a result of the space
charge effect, which is more pronounced with acid mist due to the larger
number of fine particles (1.5 to 1.9 versus 4 pa mass median diameter).
6A-32
-------
PILOT COMBUSTOR TESTS WITH HIST AND FLY ASH
Since a dramatic space charge effect was evident in the mist-only results,
additional tests were conducted to examine any further degradation in WESP
performance that might be caused by fine fly ash particles. This was accom-
plished by co-firing the S02-doped natural gas with a small quantity of coal
to simulate the mass loading of fly ash in flue gas leaving a scrubber. The
total thermal input into the combustor was maintained constant so there would
not be a significant change in the temperature profile. With coal burned at a
rate of 2 lb/hr and S02-doped natural gas fired at a rate of 970 cfh, the flue
gas entering the spray chamber contained about 45 ppm of SOj, and the inlet
mass loading averaged about 27.6 mg/acm (0.012 gr/acf). This is comparable to
recent measurements made by Flakt, Inc., at a scrubber installation of
Seminole Electric, where an average loading of 28.8 mg/acm (0.0126 gr/acf) was
reported (1).
Assuming that the ratio of the acid mist mass loading to the S03 level was the
same as in the two mist-only tests, the inlet loading of acid mist may be
estimated to be 15.6 mg/acm (0.0068 gr/acf). By difference, the inlet loading
of fly ash is about 12.0 mg/acm (0.0052 gr/acf). With a coal containing 10%
ash, this loading of fly ash would correspond to an upstream control efficien-
cy (in the primary ESP or baghouse and scrubber) of about 99.7%, yielding a
mass emission rate of about 0.013 lb/MMBtu, based on fly ash only. The total
mass emission rate, including acid mist, would be about 0.03 lb/MMBtu. The
total particulate mass would be composed of about 57% acid mist and 43% fly
ash. Based on the measured mass median diameters (mods) of the mist (1.5 fan)
and the mist/fly ash combination (2.2 /on), the mod of the fly ash is estimated
to be 3.1 /an. This case is believed to be a reasonable simulation of a
precipitator/scrubber installation operating in compliance with the 1979 NSPS
(4). The results of this test are summarized below; the results of the high-
S03 mist case are also included for comparison.
6A-33
-------
Test Parameter
Mist Only
High SOj
Mist Plus
Flv Ash
Initial S03 concentration, ppm
Inlet mass loading, ng/acn (gr/acf)
Inlet mass median diameter, im
HESP applied voltage, kV
HESP current density, [A/ft
Specific collecting area, ft/kacfm
Collection efficiency, %
\ 40.3
47
16.3 (0.0071)
1.5
68
196
45
27.6 (0.012)
2.2
45
64
42.5
Smaller than 1 pm
Smaller than 5 /an
62.1 to 83.0
71.4 to 91.8
63.4 to 77.1
65.1 to 77.6
Comparison of the mist-plus-ash case and the mist-only case shows a striking
degradation of the electrical operating conditions with fly ash present. With
fly ash in the system, it was not possible to maintain the same applied volt-
age that was used in the mist-only case. Intermittent sparking resulted in
excessive tripping of the power supply and limited the applied voltage to
about 45 kV. It may have been possible to operate at a higher voltage, but
this would have required frequent resetting of the power supply, which may
have compromised the outlet impactor data. In actual practice, the use of a
spark-rate controller may partially alleviate this problem.
The presence of the fly ash appears to produce a larger performance degrada-
tion in the 1 to 5 im size range than in the submicron size range. Since a
very small mass fraction of the fly ash is submicron (typically less than 1 to
2%), it would not be surprising to see similar submicron collection efficien-
cies for the two cases, if the electrical operating conditions were similar.
However, the degraded electrical conditions apparently limited the maximum
submicron collection efficiency to 77.1%, compared to 83.0% for the mist only.
For all particles smaller than 5 faa, the cumulative collection efficiency was
reduced from a maximum of 91.8% to 77.6% with fly ash present.
6A-34
-------
PHASE II OBJECTIVES
The primary goal of the Phase II work was to refine the projections of WESP
performance by using data from two utility FGD installations. This required
S03 and particle size measurements, along with chemical analyses, to determine
the loadings of acid mist, fly ash, and scrubber solids downstream from the
two scrubbers. These measurements provided the data needed to project WESP
performance using the computer model developed in Phase I. A secondary
objective was to estimate the effect of the WESP on opacity relative to the
baseline configuration (scrubber with HE only). Another major objective of
Phase II was to develop plans for a follow-on demonstration of the WESP
concept at utility sites, if warranted.
SITE SELECTION
The first task under Phase II was to select the test sites for the field
measurements. It was preferred that the two sites have substantially differ-
ent levels of S03 in the flue gas, so that the effect of acid mist loading
could be examined. A high S03 concentration was desirable at one of the
sites, so that the effects of space charge corona suppression could be further
analyzed using the computer model. It was also considered desirable to avoid
the selection of two sites with the same types of boilers, scrubbers, and HEs,
so that the results would be applicable to a wider range of equipment types.
Based on these considerations, two sites were ultimately selected.
Site 1 was a 700-MW, cyclone-fired unit equipped with a combination venturi/
spray tower FGD system. The FGD system consisted of six scrubber modules,
five of which were normally in operation. At the exit of each tower, the gas
was discharged at a right angle and flowed horizontally through two vertical
mist eliminators. Because of severe corrosion of the reheater tubes, the
reheater had been removed, and the unit had been converted to wet-stack
operation. The combination of a high sulfur content (3.2% nominal) and high
iron content in the ash, along with the high-temperature cyclone firing, was
reported to produce a very high S03 concentration at this site (5).
6A-35
-------
Site 2 was a 575-HW, pulverized coal-fired unit equipped with a venturi
rod/spray tower FGO system. Each of the four scrubber nodules was equipped
with a horizontal mist eliminator mounted directly above the spray tower, so
that the gas flow through the mist eliminator was vertically upward. Just
above the mist eliminator was an in-line tubular reheater. Previous data
obtained at this site (6) suggested that the S03 concentration would be lower
than at Site 1, although the nominal coal sulfur content was about the same as
at Site 1. The lower conversion of S02 to S03 at Site 2 is presumably
associated with the different boiler type and the lower iron content of the
ash.
S02/S03 DATA
Table 1 gives a summary of the S02 and S03 concentrations measured ahead of
the scrubber at both sites. There is essentially no difference in the S02
concentrations measured at the two sites, but the S03 concentration is
significantly higher at Site 1, for the reasons already discussed. If the
reported amounts of S03 were completely condensed in the form of sulfuric acid
(H^Ot), this material alone could account for a mass loading of about 0.03
gr/dscf at Site 1 or about 0.02 gr/dscf at Site 2. At Site 1, this would be
sufficient to account for over 85% of the particulate mass measured at the HE
outlet by the cascade impactors. At Site 2, it would be sufficient to account
for about 70% of this mass. However, chemical analyses of the impactor
samples (discussed later) reveal that the HjSO^ actually accounts for only 40
to 45% of the particulate mass at Site 1 and about 57 to 62% of the particu-
late mass at Site 2. This suggests that some of the S03 or sulfuric acid is
removed in the scrubber and HE system.
To investigate the question of S03 removal across the scrubber, outlet S03
measurements were made at Site 2. To determine the amount of S03 removed, the
sampling probe was heated to convert all of the HjSO^ back to SOj. Measure-
ments were also made with the probe at flue gas temperature (about 160*F after
reheat) to verify that all of the S03 was condensed. These measurements
confirmed that all of the S03 was condensed at this point (residual below the
6A-36
-------
detection limit of 0.3 ppm) and that the amount of condensed HjSO^ was
equivalent to an SO3 concentration of 6 ppo. After correcting the inlet and
outlet S03 concentrations to the same 02 levels, the removal of SO3 across the
scrubber was calculated to be 28%. Allowing for this loss, the acid mist
would be expected to account for about 61% of the outlet particulate mass at
Site 1 and about 50% of the outlet particulate mass at Site 2. Chemical
analyses of the impactor samples revealed 40 to 45% l^SOt at Site 1 (about 16
to 20% less than calculated from the gas-phase S03 concentration) and 57 to
62% H2S04 at Site 2 (about 7 to 12% more than calculated). The lower H^O^
recovery at Site 1 may indicate that the S03 removal was higher than at Site 2
(removal measurements were made at Site 2 only). The slightly higher recovery
at Site 2 could be attributable to other sulfates in the ash.
TOTAL AND SUBMICRON MASS LOADINGS
Particle size and mass loading measurements were made at both sites using
University of Washington Nark V cascade impactors that were heated to avoid
condensation within the impactor. At Site 1, these measurements were made at
the NE inlet and outlet with either one or two HEs in place. This provided an
analysis of the size-dependent collection efficiency of the NEs to compare to
the projected performance of the WESP. Normally, the FGD system at Site 1
operates with two NEs in series. However, a WESP supplier (ABB Flakt, Inc.)
recommended that one of the NEs be removed if a WESP were to be retrofitted.
Therefore, measurements were made with both one and two NEs in place.
Surprisingly, there was very little difference in the cumulative mass loadings
measured with either one or both NEs in service. Therefore, only a single
value is reported for the outlet mass loading.
At Site 2, measurements were made at the KE outlet and the reheater outlet.
Only the HE outlet data are of interest for a WESP retrofit. It would not
make sense to retrofit a WESP after the reheater, because the evaporation
across the reheater would make the droplets finer and possibly more difficult
to collect. These measurements were made at the request of the host utility
to assist them in correlating the measured emissions w.i.n opacity.
6A-37
-------
Table 2 presents the average total and submicron mass loadings obtained at
both sites at each sampling location. As expected, the mass loading was very
large ahead of the HEs (13.7 gr/acf), and this mass was dominated by particles
larger than 1 /an. The mass mean diameter (HMD) of this material was estimated
to be 44 /xm. Downstream from the HEs, the mass loading was much lower, and
the particulate mass was predominantly submicron in size. The cumulative
submicron mass loading was slightly higher at Site 1 than Site 2 (0.022 versus
0.021 gr/acf), although the cumulative submicron percentage was lower at Site
1 than Site 2 (87% versus 95%). Thus, Site 2 appears to have a finer distri-
bution on the basis of submicron mass percentage, but it actually presents
less challenging conditions for a WESP retrofit than does Site 1, because the
absolute loading of submicron particles is lower at Site 2. This small
difference in submicron mass translates into a large difference in the
number concentration of submicron particles, which is critical in terms of
space charge effects.
CHEMICAL COMPOSITION
The cascade impactor samples were analyzed to determine the weight percent of
H^O^, fly ash, and scrubber solids as a function of particle size. The
analytical methods and procedures for calculating the weight percent of each
component are detailed in the Phase II final report (7). To provide a
sufficient quantity of sample for analysis, selected impactor stages were
combined, yielding four size fractions: (1) larger than 8 /on, (2) 1 to 8 pm,
(3) 0.1 to 1 [m, and (4) smaller than 0.1 ion. Figure 5 shows the H^O^
content of the various size fractions from Site 1. As expected, content
increases with decreasing particle size. At Site 2, this same trend was evi-
dent down to the 0.1 to 1 m fraction, but the fraction smaller than 0.1 [aa
contained slightly less acid than the 0.1 to 1 pm fraction, as indicated
bel ow.
Scrubber Fly
Size fraction, m H^SO,. Wt % Solids. Wt% Ash. Wt%
Less than 0.1 56.9 0.4 42.7
0.1 to 1 61.8 2.2 36.0
1.0 to 8 47.1 11.2 41.7
Larger than 8 27.3 72.7 0
6A-38
-------
WESP PERFORMANCE PROJECTIONS
The WESP computer model developed 1n Phase I was used to make performance
projections for WESPs retrofitted at the two test sites discussed above. For
the Site 1 projections, the mass loading and size distribution measured with
one ME in place were used, since that configuration was recomnended by a WESP
supplier. For the Site 2 projections, the measurements made at the ME outlet
were used. In each case, projections were made with two sets of electrical
conditions, corresponding to the best and the worst conditions achieved in the
Phase I pilot testing.
Predicted collection efficiency. %
Electrical conditions Site 1 Site 2
45 kV and 72 nA/cm2 84.9 97.5
60 kV and 114 nA/cm2 9C.2 98.7
The lower efficiencies for Site 1 are entirely attributable to the higher
loading of fine particles and the resulting space charge effects. For both
cases, the SCA was 50 ft2/kacfm; the gas velocity was 20 ft/sec; and the tube
length was 10 ft.
For Site 1, the baseline opacity, with both MEs in place, was 42 to 60% (as
determined by a trained smoke reader after dissipation of the steam plume).
The projected opacity after the WESP retrofit described above was 11 to 19%
with the worst electrical conditions and 8 to 14% with the best electrical
conditions. For Site 2, baseline opacity data were not available, but the
projected opacity (after the WESP retrofit) was substantially lower than at
Site 1 (1.5 to 3% with the worst electrical conditions and less than 1.5% with
the best electrical conditions). These results strongly suggest that the WESP
retrofit could maintain opacity below 20% at both of the sites investigated.
FOLLOW-ON DEMONSTRATION
Based on the results of Phase I and Phase II, a follow-on demonstration of the
WESP concept appears to be justified. Therefore, preliminary planning for a
follow-on demonstration has been undertaken. Two approaches have been consid-
ered: (1) Installation of a prototype WESP on a full-scale scrubber module
and (2) construction of a smaller, mobile WESP demonstration unit that could
6A-39
-------
be tested at various utility sites. Although the full-scale unit would be
preferable in some respects, the ability to test a wide range of conditions is
a key factor favoring the mobile unit. One possible embodiment of a mobile
unit is shown in Figure 6. This unit would handle a gas flow of about 10,000
acfm at 125*F (about 5-HW equivalent) and would fit on a standard 40 to 50 ft
trailer. The design would include provisions to test different types of
discharge electrodes as well as different types of collecting tubes. Provi-
sions would also be made for real-time monitoring of the gas flow, tempera-
ture, voltage, current, and opacity. Test parameters for the WESP demonstra-
tion program would include: coal type, SOj/acid mist loading, fly ash and
scrubber solids loadings, size distribution, scrubber type, NE type, electrode
types, and cleaning methods and frequencies.
There is no definite source of funding for the WESP demonstration unit at this
time. However, the Department of Energy, the Electric Power Research Insti-
tute, several utilities, and a major WESP supplier have expressed considerable
interest in this concept. Recognizing the potential benefits to the utility
industry and the potential market for WESP technology, the utilities and the
WESP supplier have agreed in principal to share a portion of the project
costs. During the initial portion of the proposed follow-on demonstration, an
economic analysis of the WESP technology would also be done. This analysis
would address existing WESP technology as well as various advanced concepts in
discharge electrodes and materials of construction.
6A-40
-------
Bakke, E. and 0. Sarniento. Performance Impact of Hist Eliminators and
Wet Electrostatic Precipitators on Particulate Emissions and Opacity.
Presented at the Combined F6D and Dry SO, Control Symposium, St. Louis,
Missouri, October 25-28, 1988.
Faulkner, M.6. and J.L. DuBard. A Mathematical Model of Electrostatic
Precipitation. Revision 3. KTIS PB84-212-679, Environmental Protection
Agency, Research Triangle Park, North Carolina, May 1982.
Pontius, D.H. Private Conmunication to M.G. Faulkner, Southern Research
Institute, Birmingham, Alabama, October 1988.
New Source Performance Standards: Electric Utility Steam Generating
Units. Federal Register: 44 (113) 33580 to 33624 (June 11, 1979).
Dickson, W. R. Internal Memorandum, Southern Research Institute,
Birmingham, Alabama, August 5, 1986.
Dismukes, E. B. A Study of Resistivity and Conditioning of Flv Ash. EPA
Report No. EPA-R2-72-087 (NTIS PB 212607), Environmental Protection
Agency, Research Triangle Park, North Carolina, December 1972, p. 75.
Dahlin, R.S. Electrostatic Precipitation of Condensed Acid Mist. Phase II
Final Report to U.S. Department of Energy/Pittsburgh Energy Technology
Center, Pittsburgh, Pennsylvania, April 1991.
Table 1.
Summary of SOj/SOj Measurements
Site 1 Site 2
Average SO, Concentration, ppm 2100 2200
S02 Concentration Range, ppm 2000 to 2260 2190 to 2210
Average SO, Concentration, ppm 19 11
S0s Concentration Range, ppm 13 to 25 9 to 13
Average SO,-to-SO, Ratio 0.009 0.005
S03-to-S02 Ratio Range 0.0065 to 0.011 0.004 to 0.006
Table 2.
Total and Submicron Mass Loadings
Mass Loading, gr/acf
Total Submicron
Site 1 - ME Inlet
Site 1 - HE Outlet
Site 2 - ME Outlet
Site 2 - Reheater Outlet
13.7 0.026 (0.2%)
0.025 0.022 (87%)
0.022 0.021 (95%)
0.011 0.010 (91%)
6A-41
-------
si >?*.~
ID FAN
LOCATION
INLET SAMPUNG*
LOCATION
FLOW
DAAIN
Figure 1. Sketch of WESP Setup Used in Laboratory Tests
with Simulant Oil.
~ Q
S S
SP
Vqo
Figure 2. Sketch of Modified WESP Setup Used in Pilot
Combustor Tests with Actual Acid Mist
6A-42
-------
10'
10® —
<
er
1CT2
\ 5 lab DATA
I • • •• CURRENT-SPECtFIC MOOEL
.• ____ CURRENT-SEEKING MODEL
W
\
o*.
\ 3\
\ t
1 1
90.0
99.0
99.9
10°
101
10*
99.99
PARTICLE DIAMETER, micrometers
Figure 3. Comparison of WESP Fractional Collection Efficiency
Measured Using Simulant Oil with Predictions of Two
WESP Computer Models.
2.0r
MAX OUTPUT
B2-4 kV
w/o SPARK
1.6 —
CONDITIONS
Natwai gas firing "IOOOCFh
SO2 concentration ¦ 2900 ppm
Get volume treeted -ilSecfm
<
E
u
o
<
CE
IU
a.
O
1.2 —
0.8 —
0.6 —
0.4 —
0.2 —
20
WITH 26 ppm
OFSO3
NO SO3 PRESENT /
30 40 60 60 70
APPLIED VOLTAGE, hV
Figure 4. Comparison of WESP Voltage-Current Curves
Obtained With and Without Acid Mist.
6A-43
-------
o
o
<
g
cc
Z3
CO
1-
z
UJ
o
cn
UJ
£L
1-
X
o
UJ
5
BOTH MEs
40
ONE ME
BOTH MEs
ME INLET
30
ME INLET
BOTH MEs
ME INLET
ONE ME
20
ONE ME
ONE ME
10
—
BOTH MEs
0
ME INLET
< 0.1 jim 0.1 to 1 (im 1 to 8 jim
PARTICLE SIZE FRACTION
>8(im
Figure 5. Sulfuric Add Content of Particle Size Fractions at Site 1
Piping
above line
installed
on site
Rotated into position
using winch
Nob: AI ptplTI B 12-ln. schodJe 5S
saintess steel (317 LM or bensr)
Figure 6. Sketch of Mobile WESP Pilot Unit for Demonstration Program
6A-44
-------
MANAGING AIR TOXICS:
STATUS OF EPRI's PISCES PROJECT
by
Winston Chow
Leonard Levin
Michael J. Miller
Electric Power Research Institute
P.O. Box 10412
Palo Alto, CA 94303
6A-45
-------
Intentionally Blank Page
6^'
-------
INTRODUCTION
The US Environmental Protection Agency (EPA) has historically regulated air toxics
(hazardous air pollutants) under Section 112 of the Clean Air Act To date, EPA has
established emission standards for 8 hazardous air pollutants (arsenic, asbestos,
benzene, beryllium, mercury, radionuclides, coke oven emissions and vinyl
chloride). The US electric utility industry was not determined to be a source
category requiring regulation for any of the eight chemicals. Of the eight,
radionuclides were the last species for which EPA established hazardous emissions
standards. In this instance, EPA determined that the risks associated with electric
utility fossil fuel power plant emissions were sufficiently low that they should not
be regulated. However, the 1990 Clean Air Act Amendments require a new
evaluation of the electric utility industry emissions of hazardous air pollutants (1).
This paper summarizes the key features of the air toxics provisions of the Clean Air
Act Amendments, describes EPRI's activities on the subject, and provides some
preliminary insights from EPRI's research to date.
1990 CLEAN AIR ACT AMENDMENTS
The Clean Air Act Amendments of 1990 greatly expanded EPAs rulemaking
authority over hazardous (toxic) air pollutants. The Act contains a list of 190
chemicals (Table 1) that would be subject to control. Other substances may be added
to the list by the EPA Administrator if they present adverse environmental effects.
It requires sources, with exception of utility sources, that emit 10 tons or more per
year of any one pollutant, or 25 tons or more per year of any combination, to apply
Maximum Achievable Control Technology (MACT). Although not clearly defined,
MACT is the maximum degree of reduction of hazardous pollutants that the
Administrator determines is achievable. Consideration would be given to the cost
and feasibility of control, energy impacts, and environmental factors. For existing
sources, MACT may not be less stringent than the average emission limit achieved
by the best performing 12% of existing sources in categories containing 30 sources or
more. After applying MACT, a residual risk analysis will need to be performed to
determine if additional controls are warranted.
Five studies which affect electric utilities are mandated: a 3-year study to address the
hazards to public health associated with emissions from fossil-fuel power plants
(after compliance with the acid rain provisions of the Act); a 4 year evaluation of
mercury emissions, their effect on human health and the environment, and the
availability and cost of potential control technology; a 3 year mercury study
conducted by the National Institute of Environmental Health Sciences to define
health and environmental thresholds for mercury; and, a 3-year study of
atmospheric deposition rates, impacts on public and environmental health and
water quality effects of air toxics on the Great Lakes and coastal water bodies; and a
study on residual risk methods. EPA is directed to regulate the utility industry for
6A-47
-------
air toxics only if its 3 year study indicates that such regulation is necessary and
appropriate.
Although the electric utility industry is not the primary focus of the air toxics
provisions in the Clean Air Act Amendments, the potential financial implications
are enormous. The Edison Electric Institute has estimated that compliance costs
could be as high as $7.8 billion per year (2). This is the case despite recent EPA studies
which have shown that emissions of potential cancer-causing substances from
electric utility boilers pose insignificant risks — less than 1 excess cancer per year in a
population of over 200 million (3). EPA's study included arsenic, selenium,
hexavalent chromium, cadmium and formaldehyde emitted from both coal and oil-
fired boilers.
As the basis for this risk assessment, EPA used available air toxic emissions factors
from the literature (4). However, the values used are of uncertain quality with
many acknowledged limitations. Quality assurance procedures were not performed,
nor do the authors of the EPA report endorse the emission factors as representative.
Variations in trace element levels in coal, the design and operating parameters of
boilers and control devices, and uncertainty in sampling and analytical
methodologies for detecting trace pollutants all contribute to the uncertainty.
More recent analyses of these data and data gathered since completion of the EPA
study generally support lower emission factors than those recommended in the EPA
report (especially for chromium, manganese, and nickel). Also, the EPA study only
focused on those chemicals classified as potential carcinogens. The list of 190
hazardous air pollutants also includes chemicals that are noncarcinogenic such as
hydrochloric acid (HC1).
EPRI AIR TOXICS ACTIVITIES
To help the electric utility industry better understand emissions of potentially toxic
chemicals from fossil fuel power plants, EPRI initiated the PISCES (Power Plant
Integrated Systems: Chemical Emissions Study) project in mid-1988. PISCES is
multi-media in perspective; that is, the study evaluates the presence and fate of
chemicals in air, water and solid waste discharges (Figure 1). TTiis approach is being
taken so that the effects of controls on air emissions, for example, can be assessed
with full knowledge of the impacts on other plant process streams (i.e., solids and
wastewaters).
The project involves the collection and review of existing data regarding the source,
distribution, and fate of chemicals in both conventional and advanced fossil-fuel
fired power systems. It consists of several major products and activities including: a
relational database of information gathered from the literature and other sources; a
computerized power plant systems model to track the pathways of chemical
substances and quantify emissions; a field monitoring program to measure
emissions of 24 chemicals in utility flue gas at plants employing a variety of
emission control technologies; an emission control technology engineering
reference manual; an analytical methods guideline for measuring trace chemicals in
utility process and discharge streams; and comprehensive, multimedia risk
assessment (Figure 2). Other EPRI air toxics research currently underway or
6A-48
-------
planned, although not described in this paper, includes an evaluation of potential
health effects (e.g., arsenic, nickel) associated with utility emissions; a
comprehensive study of mercury cycling, analytical techniques, and ecosystem
impacts; and an investigation of the atmospheric transformations of selected
chemical species such as mercury.
The PISCES Database
The project to date has focused on information collection for conventional coal-, oil-
, and gas-fired power plants. Over 500 chemicals have been identified in power
plant process streams. Approximately 80 of these 500 were selected for additional
data search on regulatory limits and health effects. In general, more and better data
tend to be available for inorganic species in liquid and solid process streams than for
air toxics. Further, lesser amounts of data are available for inorganic species in
gaseous process streams and organic species in all media.
To date, the PISCES database contains more than 150 megabytes of information,
including 80,000 records of reported quantity data. Detailed descriptions of the
database have been reported elsewhere (5, 6). The PISCES database allows users to
assemble data from a largely fragmented body of open literature and other sources
on chemicals in power plant process streams. With this vehicle, one can organize
the data to explore relationships between chemicals, process streams, and one or
more systems or plant components.
An example of information derived from the interim database is shown in Figure 3,
depicting the concentration of nickel in various fly ashes. These curves represent
the probability of finding nickel less than a specific concentration for four fuels. The
highest concentration is in oil-fired power plant ash. Figure 4 compares the
concentration of mercury in various fuels. Based on data in the PISCES database,
the fuel with the highest variability is oil.
Using other information in the database, one can determine the fate of certain
classes of chemical species within the power plant. For example, comparing the
concentration of chromium in coal with that found in the fly ash indicates that a
large proportion of chromium is captured with the particulate matter (Figure 5).
This would suggest that highly efficient particulate control devices, such as
electrostatic precipitators (ESPs) and baghouses, would remove chromium from
power plant flue gas streams quite efficiently. In fact, EPRI field studies have shown
that chromium concentrations in the stack are quite low.
Conversely, available data for mercury indicates that most of this volatile element
remains in the flue gas following an ESP (Figure 6).
The PISCES database is currently available only through EPRL In late 1991, a subset
of the database on emissions and plant parameters will be available to EPRI member
companies on diskettes. The large database is expected to be placed on a CD-ROM
system in 1992.
6A-49
-------
Monitoring Chemical Species
Evaluating potentially toxic emissions is critically dependent on the ability to
sample and measure these chemical species, especially when a vast majority of those
listed in the new Clean Air Act Amendments only appear in trace amounts in plant
process streams. Without the requisite understanding of a method's capabilities and
limitations, misleading results are not only possible, but highly probable.
To furnish utilities with interim guidance, EPRI has produced a compendium of
available methods for measuring trace substances in a variety of process streams,
including flue gas. The document will contain information on precision and
detection quantification limits, where available. This information should help
utilities establish and conduct sampling programs based on the most up-to-date
methods, and assist them in understanding the limitations of the various
measurement methods. This compendium is currently in draft form and should be
published in early 1992.
Future PISCES efforts will involve both laboratory development as well as field
evaluation studies of specific methods for measuring important chemicals in fuels
and flue gas. Of particular interest is improved sampling techniques for mercury
and benzene and speciation of important trace elements such as arsenic and
chromium. A long range goal (1995-1997) is development of a continuous air toxics
monitoring system for key chemical species.
Control Technology Engineering Reference Manual
Based on information in the PISCES database, a Control Technology Engineering
Reference Manual will be prepared to assist utilities in determining the
performance of various emission control devices in removing trace chemicals, if
risk assessment supports the need for controls. These documents will not only
provide emission factor estimates but also insights into the mechanisms involved
in chemical removal. For example, the role of temperature, ESP size and
performance, and the concentration and form of trace elements in the coal will be
discussed (if known) in terms of their impact on emissions and removal. An initial
state-of-the-knowledge document is planned for mid-1993.
The major issue in predicting control technology performance for air toxics removal
is lack of data. Although the number of available data points for plant emissions of
various chemical species is quite large, the number of paired data sets — inlet and
outlet — on any given control device is sparse. For example, the PISCES database has
no performance data for nickel or chloride removed by fabric filters. Figure 7
illustrates this same point for chromium. There are 51 data points for high dust gas
(inlet to ESP) but only 5 data points on emitted gas. This data paucity has been a
critical factor in EPRI initiating the Field Chemical Emission Monitoring (FCEM)
program in association with EPRI member companies.
Field Monitoring
The PISCES Field Chemical Emission Monitoring (FCEM) program began in May
1990 with the collection of data on 24 chemicals (Table 2). Emissions and discharges
6A-50
-------
are being measured from several control technologies, including cold-side ESPs,
fabric filters (conventional and pulse-jet), low-NOx burners, postcombustion NOx
systems, spray dry FGD, and wet lime/limestone FGD. Plant mass balances are being
performed for each metal and inorganic chemical to define their sources, pathways,
and the way they partition in the plant system. Therefore, all appropriate liquid and
solid waste streams are sampled in addition to the flue gas. A variety of coal-types
and combustion systems are included in the program. Bituminous,
subbituminous, and lignite coals are included. Combustion sources include wall-,
tangential-, and cyclone-fired boilers and fluidized bed systems. Also, oil- and gas-
fired power plant emissions have been examined for a smaller subset of air toxics.
The acquisition of new, high quality data from current utility operations, using
improved sampling and analytical procedures, will upgrade the database and
provide performance information for the Control Technology Engineering
Reference Manual.
Preliminary data from EPRI's field monitoring study are just becoming available.
One facility sampled was a midwestem U.S. power plant equipped with an ESP and
wet limestone scrubber burning a western subbituminous coal. The FGD system at
the time was operating with 24% flue gas bypass. The data indicate that, with the
exception of mercury and chloride, over 90% of each chemical was removed with
most showing over 95% removal. Mercury removal was difficult to accurately
determine since it is present in such low concentrations in the dean flue gas Gess
than 0.0002 mg/Nm^). EPRI is currently working on an improved sampling and
analytical procedure for mercury for use at future test sites.
To date, EPRI has sampled at six power plant sites. Approximately 10 more sites will
be sampled through 1993. In addition, the US Department of Energy (DOE)
Pittsburgh Energy Technology Center (PETC) will begin a complementary program
in 1992 at approximately 10 more locations. DOE PETC will sample for th3 same 24
chemicals as the EPRI FCEM program and will also use the same sampling and
analytical protocol.
Systems Model
The power systems model, just released for limited utility testing, provides either
deterministic or probabilistic estimates of chemical emissions in the gaseous, liquid,
and solid waste process streams from a specified power plant configuration. Stream
conditions for coal-fired plaints are characterized for fifteen plaint subsections (Table
3) which are used to configure a plant for an analysis. Major plant flow rates are
quantified based on internal mass and energy balance calculations for a specified
plant size, equipment design, and fuel choice. To operate the model, users must
specify inputs such as power system design parameters, performance characteristics,
emission constraints, fuel properties, and pollution control performance measures.
The pollution control performance measures can be acquired in one of two ways.
Utilities may have site specific performance data on environmental control devices
based on operating experience. Or, from the chemical composition data contained
in the PISCES database, partitioning factors for various chemical species between
solid, liquid, and gaseous streams can be derived for a device if sufficient data exist.
6A-51
-------
The development of this model is motivated by utilities' desire to better understand
the relationship between their power plant designs/operating factors, choice of fuels
and emissions of potentially hazardous or toxic species. An important application
of this model will be in characterizing the chemical composition of various power
plant emission streams such as collected fly ash, bottom ash, FGD by-product, ash
pond effluents, and stack gases.
The probabilistic feature of the model allows incorporation of uncertainty in
calculating emissions of various chemicals. This uncertainty can stem from the
variability of specific chemicals in the fuel, plant operations, and sampling and
analytical variability and uncertainty (e.g., precision, accuracy, bias). This feature of
the model allows utilities to assess the likelihood of emitting a substance at a specific
rate with a given confidence level. In other words, the model estimates cumulative
probabilities, depicting the median likelihood of observing a given emissions rate.
Such estimates may be used to evaluate the overall emissions of a specific plant.
Further discussion of the model and examples of its results in case studies are
reported elsewhere (7, S, 9, 10). An interim version of the model for conventional
coal-, oil-, and gas-fired power plants will be available for EPRI utility member
testing in late 1991.
MANAGING HC1 AND MERCURY
As debate on the air toxics provisions of the Clean Air Act Amendments has
progressed, two chemicals have received significant attention — hydrochloric acid
and mercury. Due to inorganic chlorides in coal, hydrochloric acid emissions will
generally exceed 10 tons per year for most power plants in the US. Based on an
average (0.12%) chloride content in bituminous coals, a 500 MW power plant
without an FGD (flue gas disulfurization) system would emit about 1400 tons per
year of HC1. Plants equipped with FGD systems would have substantially lower
emissions (over 90% removal).
Mercury, on the other hand, is emitted in relatively small quantities. Uncontrolled
emissions from a typical 500 MW plant would be about 500 pounds per year. Actual
emissions are less given that current environmental control technology does
remove some mercury. Utility emissions of mercury are relatively small; that is,
the annual contribution from U.S. fossil-fuel fired electric utility boilers represents
roughly 2 percent of the 6 million kilograms global mercury budget and less than 4
percent of global anthropogenic emissions (11, 12). The following discussion is a
summary of the state-of-knowledge regarding the emissions and control of HQ and
mercury.
Hydrochloric Acid
Chloride concentrations vary widely in US coals, from virtually immeasurable
quantities to over 0.5% (13). Generally, eastern high-sulfur coals have higher
chloride concentrations than western subbituminous and lignite coals. During
combustion in the furnace, over 95% of the chloride in the coal is initially released,
primarily (90%) in the form of gaseous HQ. There is little interaction between the
gaseous HC1 and the ash. HC1 will deposit onto the fly ash only below 60 degrees
Celsius (140 degrees Fahrenheit), the acid dewpoint for HC1. This is true regardless
6A-52
-------
of the pH of the fly ash. Data indicate extremely low to nondetectable levels of
chloride in fly ash from lignite, bituminous, and subbituminous coals. HQ reacts
quickly in the atmosphere with ammonia and calcium and is generally not detected
beyond 10 kilometers (several miles) from the stack.
HQ emissions are also not a major health concern. For a power plant emitting 200
tons of HQ per year with a stack height at GEP (good engineering practice), ground
level concentrations over a one-hour maximum average would be less than 1
microgram/cubic meter under adverse meteorological conditions. This is negligible
compared to the threshold limit value for occupational health effects of 7000
micrograms/cubic meter.
Mercury
Mercury levels in U.S. coals vary from about 0.01 to 8 ppm (14, 15). Typical values,
however, are about 0.24 ppm for Appalachian coals, 0.14 ppm for Interior Eastern
coals, and 0.21 ppm for Illinois Basin coals (16). Mercury is probably emitted
primarily in its elemental form, but it could also be in one of its many combined
forms.
The literature is quite confusing regarding mercury emissions from coal-fired power
plants. First, it is not always dear whether the measured mercury was total, vapor
phase, or that condensed on particulates. Secondly, the sampling and analytical
methods used may have been unreliable. Finally, most data available on mercury
removal are from the municipal waste incinerator industry where mercury
concentrations are higher by several orders of magnitude than in utility flue gas.
A literature review by Smith (17) showed mercury removals ranging from 10% to
50% through fabric filters or ESP's. This reference also reported that FGD systems
removal spans a large range, from 20 to 95%. In contrast, the current EPRI PISCES
database indicates about 20 to 90% removal for cold-side ESPs (5 data points) and 85
to 90% for fabric filters (3 data points). The primary reason for these large ranges is
the sampling and analytical variability discussed earlier.
A recent study of a coal-fired power plant in Japan showed approximately 33%
mercury removal in the particulate control system (cold-side ESP), 36% by the FGD
system (wet lime), while the remainder was vented up the stack (18). The same
study cited another coal-fired plant with a hot-side ESP and an FGD system with 25%
flue gas bypass around the scrubber. The data in this situation showed virtually no
removal in the ESP, 26% removal in the FGD system, with the remainder vented up
the stack. It appears from these data that temperature plays an important role in
mercury emissions. The likely explanation is that mercury is condensing on coal
ash particles at the lower temperatures and remaining volatile at the higher
temperatures common in a hot-side ESP.
Several recent papers have reported that mercury can be removed from municipal
waste incinerator flue gas through use of chemical additives. Joy Technologies (19)
reported that use of an additive in a spray dryer system improved mercury removal
as did operation at lower exit gas temperatures. Joy's data show that a spray
dry/baghouse combination operating on a municipal waste incinerator removed
6A-53
-------
69% of the total mercury without the additive and from 91% to 95% with the
additive. The spray dry/ESP combination removed from 27% to 66% of total
mercury without the additive and from 78% to 86% with the additive. The higher
removals were observed at the lower exit gas temperatures. Although the additive
was not specified, it is assumed to be activated carbon. Use of activated carbon has
been reported by others with similar results (20,21, 22, 23).
Another paper by several Japanese authors reported 95% to 100% mercury removal
through use of a wet lime FGD system on flue gas of a municipal waste incinerator
(24). Here, an oxidizing agent (sodium hypochlorite) was added to the scrubber to
solubilize the mercury. The mercury was then removed from the scrubber waste
water through a reduction, volatilization, condensation, and mercury separation
process.
Another chemical removal method for mercury is sodium sulfide scrubbing (25).
The sodium sulfide combines with both vaporous mercury and mercuric chloride
to form mercuric sulfide. Mercuric sulfide is a stable compound. Reported removal
efficiencies for a municipal waste incineration flue gas are between 73% and 88%.
No cost data were reported.
Condensing wet scrubbers may also achieve over 90% mercury removal (26).
However, to achieve this level of mercury removal, the mercury must be in the
form of mercuric chloride (which is soluble) rather than vaporous mercury which
will require use of an additive (such as sodium sulfide) for removal.
Another adsorption mechanism for mercury removal has been reported from
Germany (27). In this instance, a reactor designed for NOx removal following an
FGD system also indicated removals of virtually all of the mercury present in the
flue gas. These tests were conducted at pilot scale on a municipal waste incinerator
plant using lignite coke as the absorbent material.
The foregoing discussion indicates some of the uncertainties regarding mercury
emissions and control. It should be emphasized that the highest removals reported
have been accomplished on municipal waste incinerator flue gas, not flue gas from
the coal-fired power plants where mercury concentrations are lower by several
orders of magnitude. The ultimate fate of mercury is also undefined. That is, the
form of the mercury in the solid or liquid by-product is not known, nor is whether
the mercury revolatilizes once the solid by-products are landfilled. One author
reported that 10 to 15% of the mercury in fly ash evaporated at room temperature
over a period of 14 days (28).
Most of the older mercury emissions data reported in the literature are suspect
given the difficulties in mercury sampling and analysis. Since mercury
amalgamates with many metals, it is ubiquitous in many laboratories and thus
contaminates samples. It does appear that the more recently reported data using
better sampling techniques and analytical methods are reducing some of this
uncertainty. However, the EPRI FCEM program is pointing to the need for further
improvements in mercury sampling and analysis in utility flue gas streams.
6A-54
-------
RISK ASSESSMENT
EPRI is also conducting studies aimed at improving our understanding of what
happens to air toxics after they leave the power plant stack. The goal is to develop
knowledge aimed at helping answer the central question around which the whole
air toxics issue revolves: What risk does the emission of these substances from
power plants pose to public health and to the environment?
A key step toward answering that question is finding out which substances are of
most concern. As stated previously, 190 substances have been designated as
hazardous air pollutants in the new Clean Air Act Amendments. However, since
most of these substances are not emitted by utility power plants or are emitted only
in extremely small quantities, EPRI is trying to substantially narrow the list to a few
priority substances that warrant detailed risk analysis.
Assessing the potential health risks of toxic substances in the environment is a
demanding task. There is a big difference between a large, direct exposure in a short
time — such as could occur, for example, if a tank containing a toxic substance
ruptured near people — and exposure to a minute, diluted amount of the same
substance over many years. Also, humans can be exposed to substances by different
routes: inhalation, absorption through the skin, or ingestion of food and water
containing the substances. Exposure to a single substance may result in a number of
different physiological responses. To further complicate matters, the substances
emitted from power plant stacks may be chemically transformed in the atmosphere
by exposure to sunlight and water vapor or may be transformed by their interaction
with the ecosystem. These transformed substances may be either more or less toxic
than what was originally released from the stack. All of this must eventually be
taken into account in risk assessment.
EPRI has developed a set of methodologies and is applying a series of computer
models using data developed in other EPRI research programs, including the
PISCES project, to determine human health risks from air toxics emissions.
The first of the models, the Air Emissions Risk Assessment Model (AERAM), is
used to represent individual sources of air toxics. It uses a set of modules to
calculate plant emissions, the transport and dispersion of emissions in the
atmosphere, human exposures, and, ultimately, the human health risks from a
particular power plant. By varying input data on fuel characteristics and the
efficiency of pollution control technologies, the user can evaluate the impact of
various control options on potential health risks.
Another model, called AirTox, expands on the capabilities of AERAM. It permits
multiple decisions on controls to be analyzed and provides information on a range
of outcomes, including cost. AirTox also allows utilities to explicitly incorporate
uncertainties in such factors as ambient concentrations of substances, utility
emissions, control efficiency, and the relationship between exposure and health
effects. The model can help a utility put in perspective its contribution to air toxics
emissions and evaluate the implications of changes in emission levels over time.
6A-55
-------
EPRI and individual utilities have used these tools to conduct risk assessments for
several carcinogens emitted from a variety of utility sources to determine the
relative contribution from each chemical to the risk posed by each source. To date
these analyses have considered only the inhalation exposure route to air emissions.
A comparison benchmark for risk was used in these analyses: an incremental
lifetime cancer probability CLCP) of one in one million. This value is one of several
criteria used in regulatory reviews nationwide, representing the LCP experienced by
a "theoretical" individual exposed for seventy years to the maximum air
concentration of all toxins under consideration. Such an exposure defines the
Maximum Exposed Individual, or MEI.
Test versions of AERAM and AirTox are available to EPRI member utilities; a new
methodology is being developed that will expand on these models to allow
consideration of exposure routes besides inhalation. Called RiskPISCES, this
multimedia risk evaluation model will link existing models for multiple exposure
pathways and will perform a screening evaluation of multiple chemical species
under a common framework to identify significant species; these species will then
be subjected to detailed risk analyses.
The results of EPRI's risk assessment studies will be used in the compilation of a
Comprehensive Risk Evaluation (CORE) to be completed in early 1993. The CORE
effort will provide utilities and decision-makers with EPRI's best assessment of the
human health and environmental risks posed by fossil fuel-fired power plants.
CONCLUSIONS
Electric utility flue gas emissions are generally well controlled and will be even
more so after complying with the acid rain provisions in the 1990 Clean Air Act
Amendments. However, the new Clean Air Act Amendments require several
detailed studies of the risks associated with the combustion of fossil fuels. Based on
these studies, the U.S. EPA will make a determination whether further controls
beyond the acid rain provisions are necessary.
Ongoing studies by EPRI, U.S. DOE and others will provide information to assist in
this evaluation of air toxics. These studies, including emissions characterization
and risk assessment, will provide valuable input to EPA's studies of air toxics. With
these efforts to acquire better quality data, the electric utility industry will be in an
improved position to evaluate EPA's conclusions on hazardous air pollutants from
fossil fuel-fired power plants.
6A-56
-------
REFERENCES
1. Public Law 101-549, November 15,1990
2. "Economic Impact of S. 1630, The Clean Air Act Amendments of 1989, Air
Toxics, NOx and S02 Provisions for Electric Utilities." Study prepared for the
Edison Electric Institute by Temple, Barker and Sloane, January 22,1990.
3. "Cancer Risk From Outdoor Exposure to Air Toxics," U.S. Environmental
Protection Agency, External Review Draft, September 1989.
4. "Estimating Air Toxics Emissions from Coal and Oil Combustion Sources,"
Radian Corporation, Report 450/2-89-001, April 1989.
5. Balfour, D.W., et al, 'PISCES: A Utility Database for Assessing the Pathways of
Power Plant Chemical Substances," Presented at the Air & Waste
Management Association Annual Meeting, Anaheim, Ca., June 1989.
6. Behrens, G.P. and Chow, W., "Use of A Multi-Media Database for Chemical
Emission Studies of Conventional Power Systems," Presented at the Air &
Waste Management Association Annual Meeting, Pittsburgh, Pa., June 1990.
7. Rubin, E.S., et al, "A Probabilistic Assessment Model For Power Plant
Chemical Substances," Paper presented at the American Power Conference,
Chicago, IL., April 1989.
8. Rubin, E.S., et al, "Chemical Characterization of Power Plant Waste Streams,"
Paper presented at the Air and Waste Management Association Meeting,
Pittsburgh, Pa., June 1990.
9. Rubin, E.S., et al, "A Probabilistic Approach to Multi-Media Environmental
Management," Paper presented at the Air and Waste Management
Association Meeting, Pittsburgh, Pa., June 1990.
10. Rubin, E.S., et al, "Evaluations Power Plant Control Strategies For Air Toxics,"
paper presented at the Air and Water Management Association Meeting,
Vancouver, B.C., June 1991.
11. Porcella, Donald, EPRI, private communications.
12. Nriagu, J.O. and Pacyna, J.M., "Quantitative Assessment of Worldwide
Contamination of Air, Water and Soils by Trace Metals," Nature. Vol. 333,
May 12,1988, Pages 134-139.
13. Coal Conversion Systems Technical Data Book, March 1982.
14. Estimating Toxic Air Emissions From Coal and Oil Combustion Sources,
Draft Final Report, DCN No. 88-203-080-19-04, Radian Corp., June 1988.
6A-57
-------
15. Smith, L, "Trace Elements From Coal Combustion: Emissions," IEA Coal
Research, IEACR/011, June 1987.
16. Coal Conversion Systems Technical Data Book, Op. Cit
17. Smith, L, Op. Cit
18. Yokoyama, T. "Investigation of the Behaviours of Trace Substances in Flue
Gas From Coal-Fired Thermal Power Plants," Komae Research Laboratory,
Central Research Institute of Electric Power Industry.
19. Donnelly, JR and Felsvang, KS, "Joy/Niro SDA MSW Gas Cleaning System:
New Developments," paper presented at the Air & Waste Management
Association Annual Meeting, Anaheim, Ca., June 1989.
20. Teller, Aaron and Quimby, Jay, "Mercury Removal from Incineration Flue
Gas," paper presented at the 84th meeting of the Air & Waste Management
Association, Vancouver, BC. June 16-21, 1991
21. Volland, Craig S., "Mercury Emissions From Municipal Solid Waste
Combustion," paper presented at the 84th meeting of the Air & Waste
Management Association, Vancouver, B.C. June 16-21,1991.
22. Guest, Terrence L., and Knizak, Ota, "Mercury Control at Burnaby's
Municipal Waste Incinerator," paper presented at the 84th meeting of the Air
& Waste Management Association, Vancouver, B.C. June 16-21,1991.
23. Riley, et. al., "Removal of Heavy Metals and Dioxin in Flue Gas Cleaning
After Waste Incineration," paper presented at the 84th meeting of the Air &
Waste Management Association, Vancouver, BC, June 16-21,1991.
24 Fujisawa, Y., et. al., "Mercury Removal From Flue Gas For Municipal Refuse
Incineration Plants," NKK Technical Report, No. 123, September 1988.
25 Volland, et- al.. Op Cit
26. Guest, et. aL, Op Cit
27. Marnet, C., et. al., "Use of Lignite Coke for Reduction of NOx After Flue Gas
Desulfurization," paper presented at the Fourteenth Biennial Lignite
Symposium on the Technology and Utilization of Low Rank coal, Dallas, TX.,
May 1987 and at the Fourth Symposium on Integrated Environmental
Control, Washington, D.C., March 1988.
28. Bergstrom, Jan G.T., "Mercury Behavior in Flue Gases," Waste Management
and Research, 4:57,1986.
6A-58
-------
PISCES: A Multimedia. Chemical Assessment Project
Rue Gas
Cleaning
Chemicals
T~;
tZSSSS.
1
:-?o«££ga
p 1 I "
Chemical
Additives
Solid Waste.
Water
.^.Sludge
Leachate
.Wastewater
FIGURE 1
6A-59
-------
POWER PLANT INTEGRATED SYSTEMS:
CHEMICAL-EMISSION STUDIES
Chemicals
Database
Monitoring
Guidelines
\f Risk
Assessment
InJHw.'.ia
Sii-'iSig I1
r""WS|i1™a'l
System
Model
Control Technology
Guidelines
figure 2
6A-60
-------
Nickel Concentration in Fly Ash
100%
80%
CD
CO
a 60%
(O
(O
40%
u
Q_
20%
0%
100,000
100
1,000
Concentration (ppmw)
10.000
10
Bituminous Lignite Oil Subbituminous
FIGURE 3
6A-61
-------
o\
t
ro
100%
0)
3 80%
ra
>
c
ra
jz 60%
(0
(0
0
40%
c
0)
o
Q) 20%
Q.
0%
0.01
Mercury Concentration in Fuels
^ /
/
' ¦ i
/ /' " /
si .¦
.///"
~
/ i
/'
y . ¦ y
i
/ «*
LZ.LillI.l-i. 1—I... i_.I —i—I—I 1 _i I I i
0.02 0.05 0.1 0.2 0.5 1
ppmw
Oil Bituminous Lignite Subbllumlnous
FIGURE 4
-------
Amount of Chromium Retained in Ashes
Coal Chromium, Measured (ppm)
FIGURE 5
6A-63
-------
Amount of Mercury Retained in Ashes
-y
. >
9 •
•
•
0.05 0.1 0.15
Coal Mercury, Measured (ppm)
02
025
FIGURE 6
6A-64
-------
PISCES Database Records
Chromium - ESP/FGD
High dust gas
BOILER
ESP
_l FGD
Co;
STACK
505
1 Limestone
FGDSoBds 43
Collected particulate
Bottom ash
358
570
FIGURE 7
6A-65
-------
TABLE 1
1990 CLEAN AIR ACT AMENDMENTS
LIST OF CHEMICALS
Acetaldehyde
Acetamide
Acetonitrile
Acetophenone
2-Acetylaminofiuorene
Acrolein
Acrylamide
Acrylic acid
Acrylonitrile
Allyl chloride
4-Aminobiphenyl
Aniline
O-Anisidine
Asbestos
Benzene
(including benzene from gasoline)
Benzidine
Benzotrichloride
Benzyl chioride
Biphenyl
Bis(2-ethylhexyl) phthalate (DEHP)
Bis(chloromethyl)ether
Bromoform
1,3-Butadiene
Calcium cyanamide
Caprolactam
Captan
Carbaryl
Carbon disulfide
Carbon tetrachloride
Carbonyl sulfide
Catechol
Chloramben
Chlordane
Chlorine
Chloroacetic acid
2-Chloracetophenenone
Chlorobenzene
Chlorobenzilate
Chloroform
Chloromethyl methyl ether
Chloroprene
Cresols/Cresylic acid
(isomers and mixture)
o-Cresol
m-Cresol
p-Cresol
Cumene
2,4-D. salts and esters
DOE
Diazomethane
Dibenzofurans
1.2-Dibromo-
3-chloropropane
Dibutylphthalate
1,4-Dichlorobenzene(p)
3.3-Dichlorobenzidene
Dichloroethyl ether
(Bis(2-chloroethyl) ether)
1,3-Dichloropropene
Dichlorvos
Diethanolamine
N.N-Diethyl aniline
(N ,N-Dimethylaniline)
Diethyl sulfate
3.3-Dimethyl benzidine
Dimethyl carbamoyl chloride
Dimethyl formamide
1.12-Dimethyl hydrazine
Dimethyl phthalate
Dimethyl sulfate
4,6-Dinitro-o-cresol, and
salts
2.4-Dinitrophenol
1,4-Dioxane (1,4-Diethyleneoxide)
1,2-Diphenylhydrazine
Epichlorohydrin
(1-Chioro-2,3 epoxypropane)
1,2-Epoxybutane
Ethyl acrylate
Ethyl benzene
N-Nitroso-N-methylurea
6A-66
-------
TABLE 1
CONTINUED
Ethyl carbamate (Urethane)
Ethyl chloride (Chloroethane)
Ethylene dibromide (Dibromoelhane)
Ethylene dichloride
(1,2-Dichlorethane)
Ethylene glycol
Ethylene imine (Aziridine)
Ethylene oxide
Ethylene thiourea
Ethylidene dichloride
(1,1 -Dichlorethane)
Formaldehyde
Heptachlor
Hexamethylphosphoramide
Hexane
Hydrazine
Hydrochloric acid
Hydrogen fluoride
(Hydrofluoric acid)
Hydrogen sulfide
Hydroquinone
Isophorone
Lindane (all isomers)
Maleic anhydride
Methanol
Methoxychlor
Methyl bromide
(Bromomethane)
Methy chloride (Chloromethane)
Methyl chloroform
(1,1.1 -T richloroethane)
Methyl ethyl ketone (2-Butanone)
Methyl hydrazine
Methyl iodide (lodomethane)
Methyl isobutyl ketone (Hexone)
Methyl isocyanate
Methyl methacrylate
Methyl tert butyl ether
4.4-Methylenedianiiine
Naphthalene
Nitrobenzene
4-Nitrophenol
2-Nitropropane
N-Nitrosodimethylamine
N-Nitrosomorpholine
Parathion
Pentachloronitrc^Kfsne
(Quintobenzefle;
Pentachlorophenol
Phenol
p-Phenylenediamine
Phosgene
Phosphine
Phosphorus
Phthalic anhydride
Polychlorinated biphenyls
(Aroclors)
1.3-Propane sultone
beta-Propiolactone
Propionaldehyde
Propoxur (Baygon)
Propylene dichloride
(1,2-Dichloropropane)
Propylene Oxide
1,2-Propylenimine
(2-Methyl aziridine)
QuinolineO
Quinone
Styrene
Styrene Oxide
2.3.7,8-Tetrachlorodibenzo
p-dioxin
1.1,2,2-Tetrachloroethane
T etrachlorethy lene
(Perchlorethylene)
Toluene
Titanium tetrachloride
2.4-Toluene diamine
o-Toluidine
Toxaphene
(chlorinated campene)
1.2.4-Trichlorobenzene
1,1,1-Trichloroethane
Trichloroethylene
2.4.5-T richlorophenol
2.4.6-Trichloro phenol
Triethylamine
T rif luralin
2,2.4-T rimethylpentane
Vinyl acetate
6A-67
-------
TABLE 1
CONTINUED
Vinyl bromide
Vinyl chloride
Vinylidene chloride (1,1-Dichtoroethylene)
Xylenes (isomers and mixture)
o-Xylenes
m-Xylenes
p-Xylenes
Antimony Compounds
Arsenic Compounds (inorganic including arsine)
Beryllium Compounds
Cadmium Compounds
Chromium Compounds
Cobalt Compounds
Coke Oven Emissions
Cyanide Compounds
Glycol ethers
Lead Compounds
Manganese Compounds
Mercury Compounds
Fine mineral fibers
Nickel Compounds
Polycylic Organic Matter
Radionuclides (including radon)
Selenium Compounds
6A-68
-------
TABLE 2
CHEMICALS FOR PISCES FIELD EMISSIONS MONITORING
INORGANICS
Ammonia
Arsenic
Barium
Beryllium
Cadmium
Chlorine/Hydrochloric Acid
Chromium
Cobalt
Copper
Cyanide
ORGAMCS
Benzene
Formaldehyde
Lead
Manganese
Mercury
Molybdenum
Nickel
Phosphorus / Phosphate
Radionuclides (U-238
Selenium
Vanadium
Fluorine/Hydrofluoric acid
Polynudear Aromatics
Toluene
TABLE 3
POWER PLANT SUBSECTIONS FOR THE POWER SYSTEMS
EMISSIONS MODEL
1. Coal Handling and Storage System
2. Boilder and Steam Cycle System
3. Spray Dryer FGD System
4. Particulate Collection System
5. Wet FGD System
6. Ash Pond System
7. Landfill/Sludge Disposal System
8. Wastewater Treatment System
9. Main Condenser System
10. Recirculating Cooling Tower System
11. Recirculating Cooling Pond System
12. Auxiliary Cooling System
13. Plant Makeup Water System
14. Plant Service Water Syslem
15. Miscellaneous Plant Systems
6A-69
-------
Intentionally Blank Page
-------
RESULTS OF HIST ELIMINATOR SYSTEM TESTING
IN AN AIR-WATER PILOT FACILITY
A. Frank Jones
Kenneth E. Mclntush
Radian Corporation
8501 North MoPac Blvd.
Austin, Texas 78759
Richard G. Rhudy
Electric Power Research Institute
3412 Hi 11 view Avenue
Palo Alto, California 94303
C. F. P. Bowen
NELS Consulting Services, Inc.
1084 Lakeshore Rd., West
St. Catharines, Ontario L2R 6P9 CANADA
6 A-71
-------
Intentionally Blank Page
6 A-72
-------
ABSTRACT
Hist eVminator system (MES) problems are a major cause of FGD system outages,
resulting in additional operating and maintenance costs. The Electric Power
Research Institute (EPRI) is sponsoring an ongoing research program under RP 2250-3
to determine the cause of MES problems and to evaluate potential solutions. The
program is currently focused on testing commercial MES configurations in a spe-
cially designed air-water pilot facility. The facility has been designed to test
with either vertical or horizontal gas flow over a range of mist loadings and gas
velocities.
This paper presents test results that relate the effects of gas velocity and mist
loading on carryover. The effect of washing on carryover is also discussed. To
date, eight commercial mist elimination systems for vertical gas flow have been
tested: two single-stage designs, four two-stage designs, and two three-stage
designs. Horizontal gas flow testing was conducted with a two-stage design.
Five of the eight vertical-flow MES designs had no measurable carryover at a
mist loading of 1.5 gpm/ft2 and a gas velocity of 12.5 ft/sec. Multiple-stage,
vertical-flow designs that had one or more stages of peaked chevron ME's were found
to operate better at extremely high velocities (roughly 16 to 19 ft/sec). The MES
design with horizontal gas flow also operated very well at high mist loadings (up
to 3.0 gpm/ft2) and gas velocities (up to 28 ft/sec).
6A-73
-------
INTRODUCTION
Hist eliminator system (HES) problems are a major cause of FGO system outages and
frequently result in additional operating and maintenance costs as well as duct
buildup and particulate emission problems. The Electric Power Research Institute
(EPRI) is sponsoring an ongoing research program (RP 2250-3) to determine the
causes of MES problems and to evaluate potential solutions.
Past work performed under this program has focused on characterizing MES problems
at full-scale FGD systems and identifying potential solutions. As a result of that
effort, a mist eliminator (HE) troubleshooting guide was developed and published by
EPRI to assist utilities in solving HE problems. However, full-scale testing is
often difficult and expensive. Also, the potential solutions are not always appli-
cable to other FGD systems because of site-specific factors. The full-scale test-
ing did identify two areas for further research, and a test program was created to
investigate these research needs in an air-water pilot facility. The first area
was the accuracy and suitability of different measurement methods which have been
used in the field to determine mist loading to the HES, the amount of carryover,
and the source of carryover. Results of research in this area were presented in a
paper at the 1990 S02 Control Symposium (New Orleans).
The second research area identified was definition of the operating limits of vari-
ous HES configurations. Determining the operating limits of these configurations
in relation to gas velocity, mist loading, and wash intensity is needed to identify
and develop solutions to HES problems and to design HE systems for new and retrofit
applications.
This paper describes the.air-water pilot facility and discusses the results from
testing the operating limits of several HES designs. The test work included an
evaluation of nine pilot-scale HES designs intended to simulate commercial full-
scale HE systems. The designs differ in the number of stages used and the direc-
tion of the gas flow (eight vertical and one horizontal gas flow systems). All of
the systems simulated are currently employed in existing FGD systems.
The following sections discuss the air-water pilot facility, HES configurations
simulated, tests performed, and test results. The conclusions at the end of this
paper summarize significant findings and discuss the impact of the results on the
operation of HE systems in utility FGD systems.
6A-74
-------
PILOT TEST FACILITY DESCRIPTION
The air-water pilot facility was constructed by NELS Consulting and is located at
their offices in St. Catharines, Ontario, Canada. The facility, illustrated in
Figure 1, is very versatile in the number and type of mist elimination systems
which can be tested. The entire facility is constructed out of Plexiglass®; this
allows complete observation of the tests and simplifies the installation of mist
eliminator stages and test ports. The facility can be used to test single or mul-
tiple mist eliminators with either vertical or horizontal gas flow. The mist elim-
inators can be tested over a range of gas flow rates and mist loadings. Hist elim-
inator washing can also be simulated. The pilot facility can be used to determine
the effect of liquid and/or gas maldistribution on MES performance. Direct mea-
surements which can be made in the test facility include the carryover rate and
pressure drop across the mist eliminator stages. This information, when coupled
with visual observations, can be used to determine the operating limits of various
MES configurations. The test capabilities of the pilot facility are presented in
Table 1, and a physical description of the test facility is provided below.
Air flow for testing is provided by one to three fans, depending on the air flow
rate required. From the fans, the air passes through a louver damper for control
of the flow rate. The air then proceeds to the bottom of the vertical gas flow
test section where a set of vanes distribute and orient the gas flow for the
vertical test section. The air passes through the vertical section of the facility
and then through a set of perforated plates for redistribution before entering the
horizontal flow test section. The duct widens out after the horizontal test sec-
tion, and the air proceeds through a final mist eliminator to remove all of the
entrained mist. The air then flows through a venturi for measurement of the flow
rate before re-entering the fans. This flow arrangement recirculates the air and
keeps it saturated, preventing evaporation which could make accurate measurement of
carryover difficult at low carryover rates.
The vertical gas flow test section is approximately 25 ft high and has a 3-ft by
6-ft cross-section. One to three stages of mist elimination devices (e.g., chev-
rons, bulk entrainment separators, impingement trays, etc.) can be tested in the
vertical section. The disengagement zone above the last mist eliminator stage can
be varied from 0 to 4 ft; a zone of 2 ft was used for the tests reported in this
paper. Mist for the vertical test section is generated by an array of nozzles
which spray cocurrently with the gas flow. The nozzle pressure is varied to change
the mist loading. Extensive calibration testing has been done to correlate the
mist loading with the nozzle pressure and gas flow rate. Tests of mist eliminator
washing are performed with a separate set of nozzles spraying directly on the ME
face to be washed. All of the mist eliminator faces can be washed alone or in
combination.
The horizontal gas flow test section is approximately 11 ft long and has a cross-
section that is 8 ft high by 2.25 ft wide. One or two mist eliminator stages can
be tested in this section. Mist generation and mist eliminator washing are both
achieved by spraying directly on the mist eliminator with a nozzle array. All of
the mist eliminator faces can be washed alone or in combination.
Gutters and drains are located in both the vertical and horizontal test sections to
allow the direct measurement of carryover. Pressure transducers are used to mea-
sure the pressure drop across the venturi and the mist eliminator stages. The
6A-75
-------
transducers are connected to a computerized data acquisition system, allowing con-
tinuous monitoring and recording of the gas flow rate and pressure drop across the
mist eliminators. Nozzle pressures for the mist loading and mist eliminator wash
headers are also measured and recorded as a part of the tests.
HES DESIGNS SIMULATED
Vertical-Flow HE Systems
Figure 2 presents the profiles of the various chevron HE blades and the spacing
between the stages for the HES configurations which were simulated in the air-water
pilot facility. The blade and stage spacings are intended to be representative of
full-scale installations of these HES configurations. The B&W three-stage design,
which is not shown in Figure 2, is identical to the B&W two-stage design with the
addition of a perforated plate located 6 ft below the first chevron stage. Each
design is described in more detail below.
The ABB two-stage RC/HP HES is the same design that was formerly offered by Peabody
and is currently in use in a few Peabody FGD systems. Since ABB's acquisition of
Peabody, the RC/HP HES has become ABB's current offering for new FGD systems. The
first stage of the RC/HP HES consists of a two-pass rough cut (RC) chevron HE with
a blade spacing of 2 inches; the blades are oriented horizontally (perpendicular to
the gas flow). The RC HE is followed by a four-pass high performance (HP) chevron
HE that has a blade spacing of 1.5 inches. The HP HE was tested in two orienta-
tions: 1) with the blade axes horizontal; and 2) with the blade axes at a 2.5*
angle from horizontal. The HP HE was tested in both orientations since the slight
incline was originally included by Peabody as part of the design to improve perfor-
mance. The lower face of the HP HE is located 62 inches above the lower face of
the RC HE. Both HE's have a slightly extended (1.5 inches) gas-straightening sec-
tion on the trailing edge of the chevron blades.
The Hunters one-stage HES (T-272) is used in a few General Electric Environmental
Systems, Inc. (GEESI) FGD systems. This HES consists of one horizontal stage of
two-pass chevron HE modules with a blade spacing of 1.75 inches. The T-272 HE
modules have an extended gas-straightening section (2 inches) on the trailing edge
of the chevron blades.
The B&W HE chevrons tested have been used in several B&W FGD systems. However, the
use of a perforated plate for a BES and the stage spacings used are representative
of an early B&W HES design. In FGD systems more recently installed, a perforated
plate is used as a tray for S02 removal (not as a BES), and the HE stages are
spaced farther apart. The B&W HES tested uses two identical stages of chevron
HE's; each stage has three passes and a 3-inch blade spacing. The stages are ori-
ented horizontally. The chevron HE's have a small lip (0.5 inch) at the trailing
edge of the blades. The pilot-scale simulation of the B&W two-stage HES used a
spacing of 48.5 inches between the bottom of the first chevron HE stage and the
bottom of the second chevron HE stage.
The three-stage B&W HES tested is identical to the two-stage B&W HES, except that a
perforated plate (25% open area} located 6 ft below the first chevron HE stage is
used as a BES. As mentioned above, a perforated plate was used as a BES only in
older B&W designs. A perforated plate BES is not currently being installed in
6A-76
-------
newer B&W FGD systems; B&W now selects HE systems based on customer specifications
and competitive bidding between ths various ME manufacturers.
The Koch HE chevrons are relatively recent additions to the HE market and have been
used or proposed for use in a few retrofit and new FGD systems. The Koch one-stage
HES (Flexichevron VII1-3-1.5) has three passes, 1.5-inch blade spacing, and an ex-
tended gas-straightening section (1.5 inches) on the trailing edge of the chevrons.
The blades are oriented horizontally.
The two-stage Koch HE system (Flexichevron VIII-1-2.5 followed by VII1-3-1-5) has
been retrofitted in one commercial FGD system and proposed for use in several
retrofit and new FGD systems. The first stage of this HES, the VIII-1-2.5, has the
same blade profile as the VIII-3-1.5, but the blade spacing is about 2.5 inches.
The spacing between the stages of the Koch two-stage HES in full-scale applications
has varied. The spacing used in the pilot-scale simulation of the Koch two-stage
design was 66 inches from the bottom of the first stage to the bottom of the second
stage. This spacing was chosen based on similar spacings for other two-stage
designs.
The Hunters two-stage HES (T-8B followed by T-271) has been used in many Research
Cottrell (RC) FGD systems. The T-8B is a peaked two-pass chevron HE with an open
construction, a blade spacing of 1.5 inches, and no lip on the trailing edge of the
blade. The blades of the T-8B are oriented at an angle of 45* to horizontal. The
second stage is a T-271 HE identical to the T-272 mentioned above except that the
blade spacing is 0.875 inch.
Figure 3 presents a more detailed drawing of the Hunters two-stage design. The
peaked BES modules are arranged so that the gas exits at opposite angles on each
side of the peak. The pilot-scale simulation of the Hunters two-stage system used
T-8B modules with a span of 36 inches so that the modules would fit in the 3-ft by
6-ft vertical gas flow section of the pilot facility. The full-scale T-8B modules
are normally available in spans of 38 inches. As shown in Figure 3, the T-8B HE
modules have vertical plates spaced about 4.75 inches apart. These plates provide
a support for the chevron blades. They also cause any water which has collected on
the blade above the plate to drain off the blade and back into the absorber tower.
The drainage length down a full-scale T-8B profile is limited to roughly 6.75
inches because of the vertical plates. The specially fabricated, 36-inch span,
T-8B modules used in the pilot-scale simulation did not significantly affect the
drainage of the T-8B HE design and allowed them to be tested in the vertical test
section without modifying the pilot facility.
The ABB three-stage HES is commonly found in Combustion Engineering (CE) FGD sys-
tems and may be otherwise known as the CE HE design. ABB has stopped offering this
HES in new FGD systems in preference to the RC/HP HES mentioned before. This HES
consists of a stage of 45* angled slats used as a bulk entrainment separator (BES),
followed by two stages of peaked, two-pass chevron HE modules. The spacing between
the bottom of each stage and the bottom of the next stage is 2 ft. Both the BES
and the peaked chevron HE modules have 3-inch blade spacings. The peaked chevron
HE modules are angled at approximately 30* from horizontal and have a small lip
(0.5 inch) on the trailing edge of the chevrons.
Figure 3 shows another view of the ABB three-stage design. Because of the peaked
design, there were a few special considerations in the design of the HES testing.
6A-77
-------
As shown in Figures 2 and 3, the ABB three-stage HES uses BES slats and peaked two-
pass chevron HE modules. The gas exits the HE stages at an angle of approximately
45*. The slats and peaked chevron HE modules do not have an extended section at
the trailing edge of the profiles to straighten the gas. Additionally, the drain-
age length for the chevrons is approximately 55 inches on each side of the peak for
full-scale modules. Since drainage is considered important in the evaluation of
carryover, it was necessary to preserve the full drainage length of the ABB peaked
modules. Therefore, they were modified to fit into the 3-ft by 6-ft vertical test
section. Because of these two factors (the exit angle of the gas and the length of
the legs of the peaked HE's), a new angled gas flow section of the pilot facility
was built to test the ABB three-stage HES design. This angled flow section was
designed specifically to follow the path of the gas exiting the BES slats and
peaked chevrons.
Figure 4 shows a schematic of the angled flow section which replaced the vertical
gas flow test section in the pilot facility. As shown, only one side of the peaked
HE modules was used. Pilot facility structural constraints prevented testing of a
full peak (total span of roughly 8 ft, 2 inches). The angled flow section of the
pilot facility had a horizontal cross-section of 16 ft2.
The disengagement zone for the ABB three-stage HES simulation was 2 ft vertically
above the highest point of the second stage of peaked HE modules. At this point,
the tower cross-section decreased drastically, speeding up the velocity of the gas
and entraining any droplets which passed through the disengagement zone. A disen-
gagement height of 2 ft was used to be consistent with the other vertical-flow HE
systems tested.
Although the ABB three-stage HES design was simulated in an angled flow portion of
the pilot facility, it is actually used as a vertical-flow mist eliminator in full-
scale systems. To better simulate the design, angling the pilot facility ducting
was done to prevent extreme maldistribution of the gas and mist caused by the exit
angles of the BES and peaked chevron HE's.
Horizontal HES
Figure 5 shows the Hunters two-stage horizontal-flow HES tested as part of this
program. This HES design is normally found in Kellogg FGD systems and consists of
a T-130 HE followed by a T-125 HE. The chevrons of the two stages have sinusoidal-
shaped profiles. The blades are oriented vertically and have channels for mist
drainage. The two stages are identical except for the blade spacing, 1.2 inches
for the T-130 (first stage) and 1 inch for the T-125 (second stage). In full-scale
systems, the spacing between the two stages is generally determined by the amount
of room needed for wash headers. The same spacing (61 inches) was used in the
pilot-scale simulation. Prefabricated HE modules like those used in full-scale
systems were used for the pilot-scale simulation, and each horizontal-flow HE had
a separate box to drain liquid from the HE blades. The drain boxes allow the HE
loadings to be measured directly with the same system of drains used to measure the
carryover.
Upon inspection at the pilot facility, it was found that the blades of the horizon-
tal HE modules, both the T-130 and T-125, were manufactured incorrectly. They were
slightly shorter than those normally installed in the field, resulting in a small
gap inside the top frame of the modules which was not normally present according to
6A-78
-------
Hunters. The shorter blades did not affect the surface area of the HE presented to
the gas flow. To correct the problem, a Plexiglass* plate was used to block off
the gap at the top of the HE blades. The plate kept gas and nist from entering the
frame space.
TESTING
Heasurement Hethods
For the vertical-flow HE systems tested, carryover is defined as any liquid which
made it through the 2-ft disengagement zone, which is the distance between the top
of the last HE stage and the point where the tower cross-section starts narrowing.
This liquid was physically collected with a system of gutters and drains and
weighed. For the Hunters two-stage, horizontal-flow HES, a disengagement zone was
not applicable. All liquid which made it past the second ME (T-125) was physically
collected and weighed.
Carryover results are presented in this paper as a function of gas velocity. For
the vertical-flow HE systems simulated, the gas velocity was based on the horizon-
tal cross-sectional area taken up by the chevrons. For the horizontal-flow MES
design, the prefabricated frames were slightly different in size. Therefore, the
gas velocity was based on the average facial area of the two chevron stages.
For vertical-flow HE systems, the mist loading to the first stage was calibrated
prior to testing any of the HE systems. This was done by collecting and measuring
the amount of mist which reached the level of the lowes- stage without any HE's in
the tower. A system of gutters and drains was installed especially for the purpose
of collecting the mist.
Mist loading to the Munters two-stage, horizontal-flow MES was generated by nozzles
spraying directly on the HE, evenly covering the chevrons. The mist loading was
measured directly during testing with the system of drains and drain boxes shown in
Figure 5 by physically collecting and measuring all liquid which impacted on the
first ME as well as any that passed through it.
Tests Performed
Dry and wet pressure drop data were collected as part of this program. However,
the pressure drop results are not presented in this paper due to space limitations.
Tests to determine the carryover as a function of the mist loading and gas velocity
were performed on each MES. Gas velocities of approximately 7 to 20 ft/s were used
for the vertical-flow ME tests, and velocities of approximately 7 to 28 ft/s were
used for the horizontal-flow ME test. Host of the HE systems were tested at mist
loadings of 0.6, 1.5, and 3.0 gpm/ft2. However, some systems which performed
better were tested at higher loadings to define their operating limits.
As part of this program, the mist loading that could be expected at the inlet of a
HES in a full-scale FGD system was predicted. The estimate was based on the termi-
nal velocity of droplets produced from two common commercial spray nozzles used in
full-scale FGD systems. The estimate predicted that potential mist loadings to the
inlet of a HES could range from approximately 0.1 to 7.3 gpm/ft2 at velocities from
6 to 14 ft/s, assuming a liquid-to-gas ratio (L/G) of 100 gal/kacf. This estimate
6A-79
-------
did not take into account droplet coalescence or droplet wall impingement, which
could lower the mist loading to the NES. Hist loadings to the entrance of a HES
are also believed to depend on several other site-specific factors (e.g., distance
from last spray header to first HE, L/G, etc.). Hist loadings of about 0.5 gpm/ft2
have been measured at two full-scale FGD systems; however, the measurements were
taken just downstream of structures which could have acted as BES's. A mist
loading range of 0.6 to 3.0 gpm/ft2 was chosen for this test program to cover the
potential range of mist loadings that might be present in a full-scale FGD system.
The nozzles used in the air-water pilot unit were considerably smaller than those
used for FGD slurry in a full-scale plant. However, essentially all of the
droplets from the smaller nozzles were still greater than 50 microns in diameter.
With these size droplets and the velocities used in the pilot unit, all of the HES
designs tested should be capable of extremely high droplet removal efficiencies, •
according to information supplied by the manufacturers. The majority of the carry-
over observed in the air-water pilot facility was due to re-entrainment of droplets
from the HE blades, not droplet penetration.
In the case of the Hunters two-stage, horizontal-f To;? both the overall system
(T-130 followed by T-125) and the individual stages were -zsted to determine the
carryover rate. The individual stages were tested at a mist loading of 1.6 gpm/ft2
so that their relative performance could be compared.
The majority of the carryover tests were repeated two or three times consecutively
without changing conditions to ensure that consistent results were obtained. For
each HES, a number of repeatability tests were also performed in which the condi-
tions were changed and then returned to the desired settings.
Evaporation was not a concern in the test facility because of the closed-loop
design of the system. However, some condensation did occur when testing the
vertical-flow HE systems. Tests to quantify the condensation were performed, and
the effects of condensation have been accounted for in all results presented here.
Condensation was not a problem in the horizontal flow tests because of a lack of
ductwork prior to the carryover collection point.
Wash testing was performed to determine the effect of different wash intensities on
the carryover from the HE systems. Since this program was performed in an air-
water system, the wash tests were designed, only to test the effect of the wash rate
on carryover, not the effect of the wash rate on HE cleanliness. In the majority
of the wash tests, a base mist loading of about" 1.5 gpm/ft2 was maintained through-
out the wash testing. Wash schemes were designed to simulate existing full-scale
systems. Each HE was washed on the face (upstream or downstream) where it is nor-
mally washed in full-scale FGD systems. For two- and three-stage HE systems, the
wash sequencing was also duplicated. For example, in some systems, the first and
second HE's are washed at the same time, while in other systems, they are washed at
separate times.
As with the carryover testing, repeatability tests were performed regularly. Wash
durations were varied depending on the length of time required to see an effect on
the carryover rate. In most cases, the HE's were washed from approximately 1 to 10
minutes with steady state generally achieved very quickly based on visual obser-
vations. Gas velocities for wash tests generally bracketed the normal operating
velocity of each HES. The range of wash rates tested included those typically used
6A-80
-------
in commercial FGD systems. For seven of the eight vertical-flow systems, these
included wash rates of roughly 0.7 gpm/ft2, 1.7 gpm/ft2, and one higher loading
(3.0 to 4.7 gpin/ft2). However, as for the carryover tests, HES designs with higher
capacities were washed at higher-than-normal loadings and velocities to determine
their operating limits.
Considerations for Evaluating Carryover Values
The velocities reported in this paper are based on the cross-sectional area of the
pilot facility occupied by the chevron blades. The superficial velocity based on
the overall cross-section of an FGO tower would be considerably lower (20% to 30%)
than the actual velocity of the gas going through the HES. This is due to the HES
support structure present in the tower which blocks off some of the cross-sectional
area. Therefore, it is important that the effects of support structure on the gas
velocity through the HES be taken into account when applying the results reported
here to full-scale systems.
The point at which carryover becomes significant can depend on particulate emis-
sions and/or solids and slurry buildup in downstream ductwork. The significance of
carryover also depends on the particular HES design. For example, horizontal-flow
HES designs generally operate at a higher velocity than vertical-flow HES designs.
Therefore, a horizontal-flow HES will clean a larger volume of gas per unit of
surface area. The carryover and wash testing results presented in this paper are
in units of gpm/ft2 of HE surface area in the plane perpendicular to the gas flow.
The point at which carryover becomes significant is based on the total volume of
gas treated and the mist eliminator surface area. Since horizontal-flow HES
designs generally have a lower surface area than vertical-flow HES designs, a
horizontal-flow HES could have a higher carryover rate (in gpm/ft2) and still per-
form better than a vertical-flow HES with a lower carryover rate in relation to the
total volume of carryover.
Following are some examples of simplified calculations showing the general method
used to determine the value at which carryover becomes significant for a full-scale
FGD system.
Carryover That Could Cause Significant Particulate Emissions:
Assumptions
• Particulate emissions of 0.015 lb/HHBtu (half of the most
recent NSPS limit) caused by slurry carryover are significant.
• For the purpose of this estimation, all slurry carryover is
assumed to be carried out the stack, and carryover from washing
is assumed to cause insignificant particulate emissions.
• 10% suspended solids and 50,000 ppm dissolved solids in slurry.
• 10 ft/s superficial velocity in tower and 12.5 ft/s actual
velocity through HES (20% blockage by supports)--typical
velocities for vertical-flow HES designs.
6A-81
-------
Based on these assumptions, the point at which carryover becomes significant from
a particulate emissions standpoint is 0.0004 gpm/ft> If the MES was designed to
work at a tower superficial velocity of 20 ft/s (25 ft/s actual velocity through
the HES) as is typical for horizontal-flow HES designs, the value at which slurry
carryover becomes significant from a particulate emissions standpoint would
increase to 0.0009 gpm/ft2.
Carryover That Could Cause Significant Duct Buildup:
Assumptions:
• The same assumptions as in the previous example plus the
fol1owi ng.
• 3 inches of dry solids buildup on the floor of the duct in six
months is significant. This is equivalent to 18.8 lb of dry
solids per square foot of duct floor.
• For the purposes of this estimation, all slurry carryover is
assumed to deposit on the floor of the duct, and carryover from
washing is assumed to cause insignificant duct buildup.
300-MW unit with 0.871 x 106 acfm at 130*F.
• 150 ft of duct with a 15-ft by 15-ft cross-section.
• Solids on duct floor have dry bulk density of 75 lb/ft3.
Based on these assumptions, the point at which carryover becomes significant from a
duct buildup standpoint would be 0.0001 gpm/ft2. Three inches of dry solids puts a
weight load of 18.8 lb/ft2 of duct floor on the ducting.
Again, if the HES worked at a superficial velocity of 20 ft/s (25 ft/s actual
velocity through the HES), twice the slurry carryover would be required before it
becomes significant from a duct buildup standpoint in the example case. The actual
point at which carryover becomes significant depends on the specific FGD system
concerned. For instance, in the above example, if the slurry that landed on the
duct floor was not drained and did not evaporate, 21.5 inches, weighing 125 lb/ft2
of duct floor, would build up in six months with the 0.0001 gpm/ft carryover rate.
This could shift the point at which carryover becomes significant.
The above estimations show only the factors that must be considered when deciding
how much carryover is significant. Clearly, some of the carryover would be emitted
from the stack, and some would land on the ductwork, rather than all of it going
either place. Additionally, carryover from washing may also cause significant par-
ticulate emissions or duct buildup. However, the estimates do show that the velo-
city at which a HES operates affects the magnitude of the value at which carryover
becomes significant.
6A-82
-------
RESULTS OF MES SIMULATIONS
This section presents the carryover results for the MES simulations in the air-
water pilot facility. Results for vertical-flow ME systems are presented first,
followed by results for the horizontal-flow MES. General results from the wash
tests of the various MES designs are also discussed.
Carryover Results
Each MES design was tested for carryover over a range of velocities at several dif-
ferent mist loadings. For the eight vertical-flow ME systems, carryover results
for a mist loading of 1.5 gpm/ft2 are presented since this is the mid-point of the
range tested. Each of the vertical-flow ME systems was tested at other mist load-
ings. The relative performances of the ME systems at these conditions generally
followed the carryover results for the mist loading tests at 1.5 gpm/ft . The
horizontal-flow MES design that was simulated performed very well. Carryover did
not occur until very high mist loadings and gas velocities were tested. Therefore,
carryover results are presented at loadings of roughly 3.1 and 5.2 gpm/ft2 for the
horizontal-flow MES simulation.
It should be noted that the carryover results are plotted on semi-log scales (for
Figures 6 and 7). This is important since a change in the gas velocity of 1 to 2
ft/sec can result in an order of magnitude change in the carryover rate. It is
also important to note that the results presented in this paper are based on the
actual velocity through the mist eliminator in the pilot facility. To equate the
results presented here to gas velocities in full-scale systems, the superficial
velocity in the full-scale system must be adjusted for blockage of the MES support
structure.
Vertical-Flow ME Systems. Figure 6 shows the carryover rates of the two single-
stage vertical-flow ME systems over a range of velocities from about 8.5 ft/s to
roughly 16.5 ft/s when loaded with mist at 1.5 gpm/ft2. With all the vertical-flow
ME systems simulated, 0.0001 gpm/ft2 was considered to be the lower measurement
limit for the pilot unit due to the small liquid volumes collected and the amount
of condensation occurring in the test facility. Points shown on the 0.0001 gpm/ft2
axis are considered to signify conditions that were tested for which there was no
measurable carryover. The Koch single-stage MES appeared to have less carryover
than the Munters single-stage MES for velocities below approximately 14.5 ft/s for
a mist loading of 1.5 gpm/ft2. Above 14.5 ft/s, the Munters single-stage system
performed better.
Figure 7 gives the carryover results for the multiple-stage vertical-flow ME sys-
tems with a mist loading of 1.5 gpm/ft2. The Munters two-stage MES had no measur-
able carryover over the range of conditions tested, even at higher mist loadings.
The ABB three-stage MES appeared to have the next highest performance with no mea-
surable carryover below roughly 15 ft/s; however, the ABB three-stage MES results
may be low for reasons discussed below. The Koch two-stage MES also had little
carryover below approximately 15.7 ft/s, but the carryover rate increased much more
rapidly than for the ABB three-stage MES at velocities higher than 15 ft/s. The
ABB two-stage MES with the second stage in the horizontal position had no measur-
able carryover at velocities below approximately 12.8 ft/s, and slightly higher
carryover than the Koch two-stage MES above that. The ABB two-stage MES had essen-
tially the same performance when the second stage was tilted at 2.5* as when flat.
6A-83
-------
The B&W MES design had measurable carryover at all conditions tested which was sig-
nificantly above that for the other MES designs tested. Velocity traverses above
the HE stages indicated that some maldistribution of gas occurred during the tests
with the B&W MES. This may have been due to the lack of a gas-straightening sec-
tion on the chevron blades at the HE exit. This caused some of the carry-up exit-
ing the first ME stage to impinge on one 3-ft side of the vertical test section.
The impingement of droplets on the 3-ft side of the tower removed carry-up which
would otherwise have reached the second HE and is not believed to have increased
the overall carryover of the system.
For the ABB three-stage MES, the gas flow exiting the angled test section made a
vertical turn into the disengagement zone due to the modular construction and size
constraints of the pilot facility (see Figure 4). The turn caused some of the
droplets to impinge on the vertical wall of the test facility. If the 45* angled
test section had extended through the disengagement zone, some of the droplets
impinging on the wall may have been collected and measured as carryover. It was
not possible to quantitatively estimate the potential increase in the measured
carryover values.
Horizontal-Flow MES. Table 2 shows the carryover rates measured for the Munters
two-stage horizontal-flow MES. The good performance of the Munters horizontal-flow
MES is suspected to be due to good drainage down the vertically oriented blades
which prevented re-entrainment. Conditions with lower velocities and loadings than
those shown in Table 2 produced no measurable carryover. The measurement limit
when testing in the horizontal-flow configuration is 0.00001 gpm/ft2; this is lovx
than for the vertical-flow configuration due to the lack of ductwork prior to the
carryover collection point which prevents condensation from becoming a significant
factor.
It is important to remember that, in full-scale FGD systems, washing of the MES
generally only occurs on an intermittent basis on a portion of one ME face. For
example, a wash.system may wash 15% to 30% of one ME face for 1 to 5 minutes every
1 to 4 hours. Instantaneous carryover due to washing could be high for the MES
section being washed. However, the instantaneous carryover based on the entire MES
cross-sectional area could be 15% to 30% of the carryover rate of the section being
washed, and the time-averaged carryover due to washing could be 10 to 100 times
less due to the intermittent nature of the wash.
The MES designs were tested over a range of wash loadings and gas velocities; the
ME faces that were washed were intended to simulate operation of the particular
designs in full-scale FGD systems (combinations of first-, second-, and third-stage
wash tests on the upstream and downstream faces were conducted, depending on indus-
try practice). Because of the complexity of the wash test results and the limited
space available in this paper, only general results and conclusions are presented.
In general, MES designs which had lower carryover rates during the carryover tests
(no washing) also had less additional carryover during washing.
Single-stage MES designs (both vertical and horizontal gas flow) are particularly
sensitive to the wash rate used on the upstream side of the ME. When there is only
one stage, this stage has to prevent carryover from washing in addition to all of
the mist loading to the MES. During washing of single-stage systems, carryover was
observed to increase by a factor of 10 or more, particularly at gas velocities in
excess of 13 ft/sec.
6A-84
-------
Washing the downstream side of a single-stage HES design or the downstream side of
the last stage in a multiple-stage HES design can result in high carryover rates.
With this type of wash, carryover rates were observed to increase by 2 to 3 orders
of magnitude, even at gas velocities below what would be typically found during
full-load operation. During one set of tests, about 10% of the wash water was
entrained in the gas and ended up as carryover; this can amount to a significant
amount of liquid even at relatively low wash rates and with intermittent washing.
Wash water generally does not have a significant suspended solids content which
could lead to duct buildup. However, carryover of relatively large volumes of
water can lead to other problems such as stack rainout of an acidic liquid, deteri-
oration of stack linings not designed for wet operation, and a significant increase
in reheat costs among others. These are just a few examples of the problems that
could occur depending on the specific equipment and operating conditions at each
site.
Carryover due to HE washing of two- and three-stage HES designs was found to be
affected most by the wash rate on the upstream side of the last HE stage. Carry-
over from washing the top or bottom of the first stage was generally an order of
magnitude less or below detection limits. Host of the multiple-stage designs could
be washed on the upstream side of the last stage at a rate of about 1.7 gpm/ft2 at
velocities up to about 12 ft/sec without a significant increase in carryover. This
is a fairly high wash rate for the last stage in a multiple-stage design. Host of
the vertical gas flow HES designs had problems preventing carryover when the gas
velocity exceeded about 13 ft/sec; above this velocity, carryover rapidly increased
by an order of magnitude or more. This points out the need for good gas distribu-
tion in absorber towers.
SUHMARY AND CONCLUSIONS
Nine commercial HE systems were tested in the air-water pilot facility, including
two single-stage and six multiple-stage (two and three stages) HES designs for
vertical gas flow, and one two-stage, horizontal-flow HES design. To put the test
results into perspective, the typical operating conditions of commercial FGO sys-
tems need to be considered. The most important conditions include gas velocity
through the mist eliminator and the carryover limit. Typical superficial gas velo-
cities in the mist eliminator area are generally about 10 ft/sec for vertical gas
flow and 20 ft/sec for horizontal gas flow. However, the results presented here
are for actual velocity through the mist eliminator in the air-water pilot facil-
ity. To equate the results presented to gas velocities in full-scale systems, the
superficial velocity in the full-scale system has to be adjusted for the HES sup-
port structure. After an average support structure blockage area is taken into
account, the typical average mist eliminator operating velocities are about 12.5
ft/sec and 25 ft/sec for vertical and horizontal gas flow, respectively.
Carryover is the main concern in evaluating mist eliminator performance since the
carryover limit is important in preventing problems with particulate emissions or
the buildup of slurry or solids in downstream ductwork. As discussed earlier, this
is a site-specific value which will depend on a number of factors. For this dis-
cussion, assume that carryover limits are represented by the values that were cal-
culated earlier to cause particulate emissions problems--0.0004 gpm/ft2 and 0.0009
gpm/ft2 for vertical and horizontal gas flow, respectively.
6A-85
-------
All but the two B&W MES designs that were simulated met these limits at typical
mist eliminator design operating velocities, assuming a mist loading of 1.5
gpm/ft2. It should again be noted that B&W no longer offers this design; they
produce competitive bids based on customer specifications.
Gas velocities have been observed to deviate by well over 50% in full-scale FGD
systems as a result of poor distribution of the flue gas between multiple towers in
an FGD system and/or maldistribution of flue gas within an individual tower. Con-
sidering that carryover increases dramatically with increasing gas velocity (up to
an order of magnitude with 1-2 ft/sec velocity increases), it is important to eval-
uate the potential effect of higher gas velocities.
Gas maldistribution of up to about 20% would not be an unreasonable expectation in
full-scale FGD systems. This would increase the highest gas velocity treated by a
mist eliminator to around 15 ft/sec (about 30 ft/sec for horizontal gas flow). If
the high gas velocity occurred over about 20% of the mist eliminator area, a carry-
over rate exceeding 0.002 gpm/ft2 (0.0045 gpm/ft2 for horizontal gas flow) for the
high gas flow area would cause problems even if the remainder of the mist elimi-
nator did not have any carryover. The MES test results indicate that the single-
stage HES designs as well as the B&W MES designs would clearly have problems meet-
ing the carryover limit specified earlier, and the performance of the ABB two-stage
design would be marginal.
Similar considerations also need to be taken into account with respect to carryover
caused by washing the mist eliminators. However, it is important to remember that
mist eliminators are generally only washed a small percentage of the time. There-
fore, the time-averaged carryover rate needs to be calculated and evaluated, assum-
ing that the instantaneous carryover rates which occur intermittently will not pose
any problems. It is also important to consider which HE face(s) is being washed
and the wash intensity.
Future work planned as part of the EPRI HES program includes testing a "dirty" HES,
testing optimized HES designs, updating the "FGD Mist Eliminator System Trouble-
shooting Hanual," and developing a mist eliminator design handbook. The "dirty"
HES that will be tested will only have a light coating of scale, not excessive
scaling and pluggage. All of the testing to date has been with clean HE's, and the
effect of cleanliness needs to be evaluated. Based on the HES simulations already
performed, future work will also involve trying to -identify and test the optimum
HES design (e.g., number of stages, blade and stage spacing, peaked or flat stage
orientation, etc.) in the pilot facility and in a full-scale FGD system. Based on
all of these results, the existing HES troubleshooting manual will be updated, and
a handbook to assist utilities in designing HE systems will be developed.
ACKNOWLEDGEHENTS
The pilot test facility was designed by NELS Consulting (St. Catharines, Ontario,
Canada) with input from Radian and EPRI. NELS provided the materials and labor for
the initial construction of the test facility free of charge. Special thanks are
extended to NELS for the time, effort, and ingenuity put into the design, construc-
tion, and operation of the test facility.
We would also like to thank the members of the Hist Eliminator Advisory Committee
for their comments and suggestions on the design of the test facility and on the
r
6A-86
-------
test plan for the program. The chairman of the committee is N. N. Dharmarajan
(Central and South West Services), and the committee members are: Rui Afonso (New
England Public Service Company), Bill Horrocks (Penn Power Company), Jeff Jernberg
(Hoosier Energy), George Munson (Tennessee Valley Authority), Jim Russell (Houston
Lighting and Power), John Smigelski (New York State Electric and Gas), Steve
Wolsiffer (Indianapolis Power and Light), Bob Wright (Associated Electric Coopera-
tive, Inc.), and Kent Zarnmit (Los Angeles Department of Water and Power).
Table 1
PILOT FACILITY TEST CAPABILITIES
Operating Conditions
Superficial Gas Velocity, ft/sec
Mist Loading, gpm/ft2
Mist Eliminator Wash Intensity, gpm/ft2
Mist Eliminator Stages
Disengagement Zone, ft
Vertical
Test Section
4 to 201
0 to 5
0 to 5
1 to 3
0 to 4
Horizontal
Test Section
4 to 301
0 to 6
0 to 6
1 to 2
NA2
The actual velocity achievable depends on the ME pressure drop.
2NA - not applicable
Table 2
HORIZONTAL-FLOW MES CARRYOVER RESULTS FOR
MUNTERS TWO-STAGE MES
Loading Velocity Carryover
(gpm/ft*) (ft/sl (gpm/ft2)
5.2 28.4 0.00020
3.0 28.3 0.00002
3.1 27.8 0.00003
6A-87
-------
Final Mill
Ehnlnatoi
8ff High by
4.5' W1d*
r
Horizontal Flow
Tail Soctlon
IT HIqIi by Z WM«
On
*
8
Air DlitrlbuDOn
VarScal Flow
Test Section
FbyJ
Crou-SecHon
Mill QtntraUon
Haadtrtor
Horliontal
Flaw Tasting
Mr Flow
Control
Dftmpai
Final ME
Milt OanaraUon
Itaader for
VaitlcalFlow
Tailing
Soctlon
Water iMdreulation
Pump
Figure 1. Layout of Air-Water Pilot Mist Eliminator Test Facility
-------
Koch
2-Slage
(Ftexlcfisvron
VIIH-2.5/VI1I-3-1.6)
Koch
1-Stage
(Flexlchavron
VIII-3'1.6)
ABB 3-Slago
Munters
2Slage
(T-8B/T-271)
B&W
2-Stage
ABB
2-Staga
(RC/HP)
Munters
1-Stage
(T-272)
1.4"
i.r
HP
T-271
VIII-3-1.5
J.0"
VIII-3-1.5
t »•
M
1.5"
T-8B
' f T-272
BES
io;
RC
Oa9 Flow
Gas Flow
Oas Flow
QasFlow
Gas Flow
Ga9 Flow
Oa9 Flow
Figure 2. Blade and Stage Spaclngs for Vertical Gas Flow HES Configurations Tested
-------
Mxters 2*Staga MES
ABB 3-Stagv MES
i
% 11 | 1 1 | 1 1 1 Slaoe2
F
ffl
1
m
-H K- t
* H m ^
1 WNVv 1 1 1
MM
OaEoMp*
Top View
of T-8B
Ch
CnFto-
SdeV«w
~ {}
GsFlo«
SOe View
Figure 3. Detailed View of the Peaked MES Designs Tested
(Munters Two-Stage MES and ABB Three-Stage MES)
BES
o
Gat Row
Figure 4. Pilot Facility Design Used to Test the ABB Three-Stage MES
6A-90
-------
Top View
1.0"
Blada
Spacing
Blade
Spacing
Gas Flow
Blade
Prolile
Final Carryovor Coltoclkxi Device
MODEL
T-130
MODEL
T-125
Front View
—2 v
Side View
Frame
Gas
Flow
Drain
Bo*
Dialn
Box
CoDeclon Boies
and Drains lor
Determination of
Mbtloadng and
Carryover
Figure 5. Munters Two-Stage Horizontal-Flow MES
-------
CSI
E
Q.
O)
0.01 :
0.001
0.0001
¦ Mumers 1-Stage
~ Koch 1 -Stage
0.01 =
> *
o
co
O
0.0001
8
10
-T
12
14
16
18
20
Gas Velocity (ft/sec)
Figure 6. Carryover Results at 1.5 gpm/ft2 Hist Loading for the Single-Stage,
Vertical-Flow HE Systems
Q 0.001 =
0.0001
ABB 2-Stage
ABB 3-Siage
B&W 2-Stage
B&W 3-Stage
Munters 2-Stage
Koch 2-Stage
10 12 14 16
Gas Velocity (ft/sec)
Figure 7. Carryover Results at 1.5 gpm/ftz Hist Loading for the Hultiple-Stage,
Vertical-Flow HE Systems
6A-92
-------
CEMS VENDOR AND UTILITY SURVEY DATABASES
0. L. Shoemaker
Engineering-Science, Inc.
Two Flint Hill
10521 Rosehaven Street
Fairfax, Virginia 22030-2899
R. Binsol
Electric Power Research Institute
3412 Hillview Avenue
Palo Alto, California 94303
6A-93
-------
Intentionally Blank Page
-------
ABSTRACT
Regulations require most fossil fuel-fired power plants to monitor stack gases
continuously for S02, N0X, and opacity. Electric utilities have installed and are
operating a broad range of Continuous Emission Monitoring Systems (CEMS). The
Clean Air Act Amendments of 1990 have placed even greater emphasis on electric
utility industry's need to select accurate and dependable CEMS to meet the
requirements of environmental regulations.
One of EPRI's current research programs is designed to aid members in obtaining
up-to-date information concerning CEMS. Two separate databases are being
developed. The utility database contains historical data obtained from coal-
fired electric utilities and will reflect the experience of the industry with
specific CEMS. The vendor database contains CEMS specifications and costs from
selected vendors and will summarize the monitoring equipment commercially,
available. Approximately 372 coal-fired plants were selected for the study. The
CEMS vendor survey included fifteen pollutant and opacity equipment vendors and
seven velocity measurement equipment vendors.
This paper will discuss each database's organization and will present summaries
of utility and vendor experience, availability, failure ranked by frequency,
maintenance requirements, etc. of the various CEMS.
6A-95
-------
CEHS VENDOR AND UTILITY SURVEY DATABASES
BACKGROUND
Regulations require most fossil fuel-fired power plants to monitor stack gases
continuously for S02, N0X, and opacity. To meet this need, electric utilities
have installed and are operating a broad range of Continuous Emissions Monitoring
systems (CEMS). The Clean Air Act Amendments of 1990 have placed an even greater
emphasis on the electric utility industry's need to select accurate and
dependable CEMS to meet regulatory requirements.
One of EPRI's current research programs is designed to aid members in obtaining
up-to-date information concerning CEMS. Surveys were conducted to construct two
databases to assist member utilities in the selection of CEMS. The utility
database contains historical data obtained from coal-fired electric utilities and
will reflect the experience of the industry with specific CEMS. The vendor
database contains CEMS equipment descriptions and costs from selected CEMS
vendors and will summarize the monitoring equipment commercially available.
EPRI has contracted Engineering-Science (ES) to assist in the preparation of the
survey questionnaires, to conduct the survey, and to compile the survey data into
databases. EPRI selected 372 coal-fired utility plants from across the
continental United States for the study. EPRI member and non-member utilities
with at least 100 MW generating capacity were included in the initial utility
survey. A total of 254 EPRI member and 118 non-member plants were sent survey
questionnaires and instructions. By September 17, 1991, 158 EPRI member and 40
non-member plants had responded to the survey.
The CEMS vendor survey included 15 pollutant and opacity equipment vendors and 7
velocity equipment vendors. The pollutant and opacity equipment vendors were
selected based on their reputation in the industry and EPRI's familiarity with
each vendor. The velocity vendor survey was sent to those who offered velocity
equipment. A total of 12 pollutant and opacity and 5 velocity equipment vendors
responded to the survey by September 17, 1991. Two additional equipment vendors
6A-96
-------
returned questionnaires; however, they were completed in a manner not compatible
with the questionnaire.
Project progress is shown in Table 1, and future plans for the project include
those items noted in Table 2. The updates to the initial utility database will
be conducted to keep current with the CEMS purchased during the two years
following the initial survey. This will allow member utilities to have timely
information on the quality of the new CEMS on the market from users in their
industry. A one-time database summary will be provided to all EPRI non-member
utilities participating in the survey. EPRI members will continue to receive
database updates as they become available.
RESULTS
The initial utility survey data required a great deal of manipulation to allow
for the construction of a uniform database. Because of this, the first update to
the survey is being used to quality control the initial utility survey. Quality
control is currently being performed on the utility data received from the
October update surveys and will not be completed until after December 5, 1991.
The vendor database, however, is available to EPRI members in dBase dbf file
format on IBM compatible floppy disks and in ASCII semicolon delineated format on
EPRINET. The utility database should be available for EPRI members by January,
1992, utilizing the same formats. Contact EPRI for the latest information on
availability of the databases.
Vendor Results
The vendor survey results, based on 70 entries, are summarized in Tables 3
through 5. An example summary report for SOg is displayed in Table 6. A
description of the vendor database is available for review in Table 7. Since
survey responses were not uniform, considerable editing was necessary to correct
this problem, and to present a more understandable database. Many manufacturers
used different terms to express the monitor analytical technique or method of
sampling. Cost estimates are inconsistent because some vendors included
different items in the cost categories, but did not specify cost for those
different items. Quality control procedures were followed in compiling the
database including error-indicating data entry screens, database editing using
dBase sort and browse commands, and a successful quality control review of 2.5%
of the records.
6A-97
-------
Utility Results
Preliminary results from the utility survey based on 778 entries are summarized
in Tables 8, 9, 10 and 11. An example SOg summary report is found in Table 12.
The database description for the utility data is available for review in Table
13. A wide variety of answers were found for most survey questions. Many hours
of survey questionnaire editing prior to data entry were necesssary. Survey
results were not uniform for the same analyzer and were at times confusing.
Estimates of staff-hour requirements for preventive and non-preventive
maintenance, accuracy assessment, and zero and span checks were incomplete.
References to error and accuracy assessments were not understood by many of the
respondents. After data entry was completed many more hours of database editing
and quality control work were necessary to provide « database with uniform
responses. Quality control work included performing a successful quality control
review of 2.5 percent of the records.
CONCLUSIONS
Once the databases are available in dBase format the three standard reports, the
dBase query and quick summary reports, and the dBase command line functions will
be available for generating dBase reports. Tables 6 and 12 are examples of two
of the standard dBase reports. The third standard dBase report is a printout of
each record formatted to look like the utility survey questionnaire. Quick
reports allow the user to use any of the record fields in a formal formatted
dBase report by simply selecting the fields and running a quick report.
The command line functions of dBase are more informal and two examples of the
several available commands may serve to show how simple and helpful these
commands can be. For example, the database can be searched to find all the
utilities that use Lear Siegler CEMS equipment or the number of records which
show CEMS maintenance performed by a contractor. If data desired in this first
example included the utility contact person's name and telephone number, the
model number and parameters sampled by the Lear Siegler equipment, the following
command could be entered into the dBase command line at the dot prompt while
using the utility database:
List First_Name. Last_Name. Telephone. Hon_Hodel. Parameters for Honitor_HF ¦= "LEAR SIEGLER" to print
The results of such a command line query is shown in Table 14. The number of
records which show CEMS maintenance performed by a contractor can be performed by
executing the following command while using the utility database:
6A-98
-------
Count for Ha1nt_tfho - "COHTKACTOR"
The dBase response will be a number, 45 in this case, shown after the dot prompt.
Similar commands can be used for any data field of interest plus the entire
database can be sorted based on the contents of one or more data fields. It is
important when using these types of commands that the field titles from the
record are copied exactly as they appear in the database and the quote enclosed
variable be identical to the entries in the database for that specified field or
the command will not work. The "record number" listed on the printout can be
used to locate the complete record in the database. The data field "Cntrl_No" is
also useful for this purpose on the standard reports or if the database has bean
moved to a spreadsheet.
REFERENCE
1. Managers and Plants databases, Utility Data Institute, Inc., Washington, D.C.
6A-99
-------
TABLE 1
SIGNIFICANT PROJECT MILESTONES
Description
Date Completed
Completed questionnaire data entry
8/28/91
Completed questionnaire quality assurance work
8/30/91
Submitted draft utility and vendor databases to EPRI
for review
8/30/91
Submitted final vendor database to EPRI for
distribution
9/30/91
Hail first update of utility surveys to utilities
10/1/91
TABLE 2
FUTURE MILESTONES OF INTEREST
Description
Date Completed
End quality assurance work on utility database
12/5/91
Submit copy of utility database to EPRI for
distribution
12/31/91
Hail second update of utility surveys to utilities
10/1/92
End quality assurance work on utility database
12/5/92
Hail third update of utility surveys to utilities
10/1/93
End quality assurance work on utility database
12/5/93
Submit final copy of utility database to EPRI for
distribution
12/15/93
Complete final report
12/30/93
6A-100
-------
TABLE 3
VENDOR DATABASE POLLUTANTS SAMPLED
Number of
Pollutants*
Monitors
CO
12
CO?
15
Dust
5
NHo
2
NO
4
N02
1
NO'
19
Opacity
8
%
11
1
SOo
THC
25
3
Temperature
1
Velocity
6
~There is one CEMS that purports to monitor S02, NOx,
CO, C02, and NHg
TABLE 4
VENDOR DATABASE TYPES OF SAMPLIN6
Number of
Type
Monitors
Dilution
1
Dilution Extraction
15
Extractive
9
Extract i ve-Heated
1
Extractive-Gas Cooler
1
Extract i ve-Probe
26
In-situ
1
In-situ (path)
10
In-situ (point)
4
In-situ (point, optical)
1
6A-101
-------
TABLE 5
VENDOR DATABASE MONITORING TECHNIQUES
Number of
Technique
Monitors
Second Derivative UV
1
Acoustic
1
Chemi1umi nescence
11
Continuous Gas Purge
1
Flame Ionization
3
Fluorescence
2
Gas Filter Correlation
1
IR
2
IR, GFC
1
NDIR
13
NDIR, GFC
3
NDIR, Paramagnetic
1
Non-pulsed UV Fluorescence
1
Paramagnetic
1
Sonic and Ultrasonic
3
Thermal Dispersion
1
UV
4
UV Fluorescence
2
UV Visible and Electrochemical
3
UV, Visible
4
Visible-Side Scatter
1
Visible Light
4
Zr02
6
6A-102
-------
TABLE 6
EPRICEM VENDOR AND UTILITY DATABASES
EXAMPLE VENDOR S02 SUMMARY
Company
Contact
Telephone
Control
Type
Monitor
Parameters
Monitor
Manufacturer
Preventive Malntenan
Name
Name
No.
No.
Sample
Principle
Sampled
Manufacturer
Model No.
(Staff Hours)
Annual
Daily
Company A
John Doo
(80B) 637-5352
V004
Dilution
Extraction
UV
S02,C02,C0
Company E
160
96
0
Company B
May Jane
(80B) 953 1013
V042
Dilution
NDIR
S02.NOX
Company B
7000
4B
0
Company C
Sam Doe
(808) 552-8997
V041
Extractive
With Gascooler
NDIR
S02,NOx
Company C
100
4B
1
Company D
Tim Smith
(808) 204-1064
V040
Extractive
Heated
NDIR
S02.N0x.NH3
Company D
8100
48
0
-------
Fiel
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
TABLE 7
STRUCTURE FOR VENDOR DATABASE
Field
Name
Type Width
COMPANY
Character
40
STREET1
Character
30
STREET2
Character
28
CITY
Character
15
STATE
Character
2
ZIPCODE
Character
10
PREFIX
Character
2
FIRST_NAME
Character
10
MI
Character
3
LAST NAME
Character
15
TITLE
Character
15
TELEPHONE
Character
14
EXT
Character
4
FAX_NUMBER
Character
14
CNTRL_N0
Character
4
M0NIT0R_MF
Character
30
M0N_M0DEL
Character
15
SYS_SUPPLY
Character
26
COND_MANF
Character
26
COND_MODEL
Character
15
PARAMETERS
Character
20
TYPE_SAMPL
Character
25
M0NTR_PRIN
Character
20
M0N_RANGE
Character
35
CAL_STD
Character
27
MAX_TEMP_F
Character
4
MX_DUST_LD
Character
6
ANNUAL_PM
Character
4
DAILY_PM
Character
4
Z SPAN TCH
Character
30
Z_SPAN_LMT
Character
20
Z_SPAN_FRQ
Character
10
Z_SPAN_H0W
Character
15
C_PR0BE_SA
Character
8
C_C0ND_SYS
Character
8
Company Name
Company Name of Firm Supplying CEMS
Street Address
2* Street Address
City
State
Zipcode
Mr or Ms Designation of Contact Person
First Name of Contact Person
Middle Initial of Contact Person
Last Name of Contact Person
Title of Contact Person
Telephone Number of Contact Person
Telephone Extension of Contact Person
Fax Number of Contact Person
Control Number of Record (unique for each
record)
Analyzer Manufacturer Name
Analyzer Model Number
CEM System Supplier Name
Gas Condi ting System Manufacturer Name
Gas Conditioning System Model Number
Parameter Sampled and Analyzed by CEMS
Location and Technique Used to Obtain Sample
Detection Principle Used by Analyzer
Upper and Lower Concentration Range of
Analyzer
Standard Against Which Analyzer Readings are
Compared
Maximum Temperature Sample Probe Sees in *F.
Minimum Temperature Sample Probe Sees in *F.
Annual Preventive Maintenance Requirement in
Staff Hours
Daily Preventive Maintenance Requirement in
Staff Hours
Technique and Location of Zero-Span Checks
Control Limits for Zero-Span Checks
Frequency of Zero-Span Checks
Automation Status of Zero-Span Checks
Cost of Probe and Sampling System
Cost of Conditioning System
6A-104
-------
TABLE 7
STRUCTURE FOR VENDOR DATABASE (Continued)
Field
Field
Name
Type
Width
Company Name
36
C ENCLOSUR
Character
8
Cost of Enclosure Requirements
37
C_MONITOR
Character
8
Cost of Analyzer
38
C_CNTL_RPT
Character
8
Cost of CEMS Control and Data Reporting
System
39
PART_CNTRL
Character
30
Particulate Control Device
(type and
location)
40
MOIST_REMV
Character
30
Moisture Removal Device Type
41
SLINE_SPEC
Character
30
Sample Line Specifications
42
PUMP_SPEC
Character
30
Pump Specifications
43
MOIST_INDR
Character
20
Method of Reporting Moisture in
Conditioning
System
44
ENCLOS_REQ
Character
30
Enclosure Requirement
45
WARRANTY
Character
40
Summary of Warranty
46
CUST_SERVE
Character
10
Committment to Customer Service
(normally in
hours)
TABLE 8
UTILITY DATABASE POLLUTANTS SAMPLED
Number of
Pollutants
Monitors
SO?
223
CO
27
C02
53
02
106
N0X
123
Opacity
385
Velocity
6
Temperature
3
6A-105
-------
TABLE 9
UTILITY DATABASE TYPES OF SAMPLING
Number of
Type
Monitors
Dilution extraction
27
Extractive
86
Extractive with probe
40
Gas extractive
6
In-situ
60
In-situ (electrolyte)
5
In-situ (point)
80
In-situ (path)
449
In-situ (probe)
2
Sonic transducer
1
TABLE 10
UTILITY DATABASE MONITORING TECHNIQUES
Number of
Technique Monitors
Second derivative spectroscopy
1
Second derivative UV
1
Chemical reactivity
1
Chemi1uminescence
26
Double pass optical
1
Electro-chemical cell
4
Electrolyte sensor
5
Emf electrode
3
Emf electrode, UV
1
Flourescence
3
FTIR
1
IR
33
IR, GFC
2
K?S04 cell
1
Microfuel cell
1
NDIR
20
NDUV
7
Paramagnetic
5
Pitot tube
2
Pulse light
2
Pulsed fluorescence
2
Transmissometry
369
UV
157
UV, second derivative spectroscopy
7
UV, Electro-chemical cell
2
UV, IR
12
UV, NDIR
3
White light, IR
5
Zr02
65
6A-106
-------
TABLE 11
UTILITY DATABASE MONITOR BRANDS
Monitor
Monitor
Number
Manufacturer
Model
of Monitc
ACS Fuji
3300
11
Ametek/Thermox
FCA
2
Ametek/Thermox
III
8
Ametek/Thermox
WDG
3
Ametek/Thermox
WDG III
5
Ametek/Thermox
WDG INS
1
Beckman
951 E
1
Beckman/Rosemount
951 A
1
Columbia :
Scientific
1600
2
Columbia :
Scientific
SA700
3
Combustion Engineering
501
1
Contraves
Goerz
100
5
Contraves
Goerz
100 GEM
2
Contraves
Goerz
400
20
Contraves
Goerz
400-0010
1
Contraves
Goerz
400-0013
2
Contraves
Goerz
500
4
Contraves
Goerz
701,700
1
Contraves
Goerz
GEM 1
3
Contraves
Goerz
GEM 100
1
Contraves
Goerz
GEM 400
1
Contraves
Goerz
TR 4034
1
Datatest
Corp.
90 A
1
Datatest
Corp.
900 A
2
Datatest
Corp.
900 RM
4
Dupont
400
2
Dupont
460
11
Dupont
460/1
1
Dupont
463
10
Durag
280, 281
1
Durag
281
2
Durag
DR 280 AV
2
Durag
DR 281 AV
1
EDC
1000 A
1
EDC
2841
24
EDC
DIGA 1100
2
EDC
DIGA 1200
2
EDC
DIGA 1400
3
EDC
DIGA Series
2
EDC
NA
1
6A-107
-------
TABLE 11 (Continued)
UTILITY DATABASE MONITOR BRANDS
Monitor
Monitor
Number
Manufacturer
Model
of Monitors
Erwin Sic Company
RM 41
2
Hartmann-Braun
URAS-2T
2
Horbia
PIR 2000
4
KVB
531
2
KVB
NA
6
Land Combustion, Inc.
7000
3
Land Combustion, Inc.
9000
2
Lear Siegler
4200
4
Lear Siegler
8100
2
Lear Siegler
CM 50
37
Lear Siegler
CM 60
8
Lear Siegler
CM 70
6
Lear Siegler
EX 4700
2
Lear Siegler
RM 41
204
Lear Siegler
RM 4200
11
Lear Siegler
SM 800
5
Lear Siegler
SM 810
72
Lear Siegler
SM 8100
16
Lear Siegler
SM 812
1
Lear Siegler/Dynatron
1100
3
Lear Siegler/Dynatron
1100 M
42
Lear Siegler/Dynatron
301
1
Lear Siegler/Dynatron
401
2
Meloy Labs
SA700
5
Monitor Labs
0X010
1
Monitor Labs
8830
1
Monitor Labs
8840
2
Monitor Labs
8850
1
Monitor Labs
8850 S
1
Rosemount
260
1
Rosemount
5100
6
Sampling Technology, Inc.
NA
1
Siemens
Oxymat 5E
5
Thermo Electron
10 A/R
22
Thermo Electron
100
2
Thermo Electron
14 B/E
1
Thermo Electron
200
10
Thermo Electron
40
5
6A-108
-------
TABLE 11 (Continued)
UTILITY DATABASE MONITOR BRANDS
Monitor
Monitor
Number
Manufacturer
Model
of Monitors
Thermo
Electron
400
14
Thermo
Electron
400/23205-209
1
Thermo
Electron
43
1
Thermo
Electron
500
2
Thermo
Electron
701
1
Thermo
Electron
703 D
1
Thermo
Electron
DIGA 1400
1
Thermo
Environmental
14 B/E
1
Thermo
Environmental
200
2
Thermo
Environmental
400
15
Thermo
Environmental
400, 500
5
Thermo
Environmental
400, 700
1
Thermo
Environmental
500
2
Thermo
Environmental
DIGA 1300
5
Thermo
Environmental
EDC 1400
2
Thermox
NA
1
United
Sciences, Inc.
500 C
4
United
Sciences, Inc.
Digital 100
3
United
Sciences, Inc.
Ultra Flow 100
1
Western Research
720 AT
1
Western Research
721 A
15
Western Research
721 AT
9
Western Research
721 ATZ
4
Western Research
722 A
2
Westi nghouse/Rosemount
1500
1
Westi nghouse/Rosemount
260
2
Westi nghouse/Rosemount
EC960
7
Westi nghouse/Rosemount
Hagen #218
1
Yokogawa
Land O2 Analyzer
2
6A-109
-------
TABLE 12
EPRI CEM VENDOR AND UTILITY DATABASES
EXAMPLE UTILITY S02 SUMMARY
Company Contact Telephone Control Type Monitor Parameters Monitor Manufacturer Start Up CEM Vendor
Name Name No. No. Sample Principle Sampled Manufacturer Model No. Rating Rating Rating
(10 g Good, 1 " Poor)
Company A Mary Smith (808)534-2363
U747
In Sllu
NDIR
S02,C02,C0
Company X
210
5
7
7
Company B John Doe (808)262-8020
U328
In Situ
NDIR
S02,N0x
Company Y
160
7
3
5
Company C John Smith (808) 563-1711
U301
In Silu(Path)
IR
S02,N0x
Company W
100
4
7
8
Company D Sam Doe (808) 459-6181
U489
In Sltu(Path)
IR
S02.N0x.NH3
Company Z
600
8
6
5
-------
Fiel
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
TABLE 13
STRUCTURE FOR UTILITY DATABASE
Field
Name
Type
Width
Company Name
COMPANY
Character
40
Company Name
PLANT
Character
40
Plant Name
STREET1
Character
35
Street Address of Contact Person
STREET2
Character
35
2* Street Address of Contact Person
CITY
Character
30
City of Contact Person
STATE
Character
2
State of Contact Person
ZIPCODE
Character
10
Zipcode of Contact Person
PREFIX
Character
2
Mr or Ms Designation of Contact Person
FIRST NAME
Character
10
First Name of Contact Person
MI
Character
3
Middle Initial of Contact Person
LAST NAME
Character
20
Last Name of Contact Person
TITLE
Character
15
Title of Contact Person
TELEPHONE
Character
14
Telephone Number of Contact Person
EXT
Character
5
Telephone Extension of Contact Person
FAX NUMBER
Character
14
Fax Number of Contact Person
CNTRL NO
Character
5
Control Number of Record (unique for
record)
UNIT_N0
Character
20
Plant Boiler Number
M0NIT0R_MF
Character
30
Analyzer Manufacturer Name
MONMODEL
Character
15
Analyzer Model Number
SYS_SUPPLY
Character
30
CEM System Supplier Name
C0ND_M0DEL
Character
30
Gas Condition System Supplier
PARAMETERS
Character
20
Parameters Sampled and Analyzed by CEMS
INSER_DATE
Character
8
CEMS In-Service Date
TYPE_SAMPL
Character
25
Location and Technique Used to Obtain Sample
M0NTR_PRIN
Character
18
Detection Principle Used by Analyzer
LOCATION
Character
27
Location of Sampling Probe
MAXTEMPF
Character
4
Maximum Temperature Sampling Probe sees in
•C
MIN_TEMP_F
Character
4
r •
Minimum Temperature Sampling Probe sees in
• c
SATURATION
Character
3
r •
Sampling Location's Status on Saturation
REHEATJJSE
Character
3
Status of Reheat Use
UP_0R_D0WN
Character
4
Status of Reheat Downstream of Probe
BOILER HRS
Character
4
12-Month Total of Boiler Hours of Operation
CEM_HRS
Character
4
12-Month Total of CEM Hours of Availability
ANNUAL_PM
Character
4
Annual Preventive Maintenance Requirement in
Staff Hrs
6A-111
-------
TABLE 13
STRUCTURE FOR UTILITY DATABASE (Continued)
Field
Field
Name
Type
Width
Company Name
35
ANNUAL_NPM
Character
4
Annual Non-Preventive Maintenance
Requirement in Staff Hrs
36
MAINT_WH0
Character
15
Status of Supplier of CEMS Maintenance
37
RATE_S_PRB
Character
2
Rating of Start-Up Problems (1 many, 10 few)
38
DAY_CAL_HR
Character
3
Non-Availabil ity of CEM Due to Daily
Calibration Needs in Hrs
39
QA_AUDT_YR
Character
4
Times/Year QA Audits are Performed
40
QA_HRS_PER
Character
2
Non-Availability of CEM Due to QA Audit
Average Time in Hrs/Time
41
DRIFT_P_YR
Character
4
Times/Year Out-of-Control Due to Drift
Specifictions
42
DRIFT_HR_P
Character
Non-Availability of CEM Due to Drift
Specifications Average Time in Hrs/Time
43
HAND_SYS_P
Character
4
Times/Year Sample Handling System Caused
Non-Availability
44
HAND_HRS_P
Character
3
Non-Availability of CEM Due to Sample
Handling System Average Time in Hrs/Time
45
HON_PER_YR
Character
4
Times/Year Analyzer Caused Non-Availability
46
M0N_HRS_P
Character
3
Non-Avail ability of CEM Due to Analyzer
Average Time in Hrs/Time
47
CNTRL_P_YR
Character
4
Times/Year Control and Data Reporting System
Caused Non-Availability
48
CNTRL_HR_P
Character
3
Non-Availabil ity of CEM Due to Control and
Data Reporting System in-House/Time Average
Time
49
PST_DATE
Character
8
Date of Most Recent PST Certification
50
CERTS_P_YR
Character
15
Times/Year of Performance Certifications
51
Z_SPAN_TCH
Character
30
Technique and Location of Zero-Span Check by
Parameter Sampled
52
Z_SPAN_LMT
Character
20
Control Limits for Zero-Span Check by
Parameter Sampled
53
Z_SPAN_FRQ
Character
10
Frequency of Zero-Span Check by Parameter
Sampled
54
Z_SPAN_H0W
Character
15
Automation Status of Zero-Span Checks
55
WHY_CEM_IN
Character
20
Reason for CEMS Installation
56
ASSESSTYP
Character
20
Type of Gas Accuracy Assessment Used
57
WHY_ASSESS
Character
20
Reason Gas Accuracy Assessments Conducted
6A-112
-------
TABLE 13
STRUCTURE FOR UTILITY DATABASE (Continued)
Field
Field
Name
Type
Width
Company Name
58
WHO_ASSESS
Character
15
Who Accomplishes Gas Accuracy Assessments
59
CAL_ERROR
Character
7
Calibration Error Percentage
60
REL ACCURY
Character
7
Relative Accuracy Percentage
61
G_ASSESS_D
Character
8
Last Date of Gas Accuracy Assessment
62
0_ASSESS_D
Character
8
Last Date of Opacity Precision Assessment
63
ASSESS_LOW
Character
7
Opacity Low Range Precision Assessment
64
ASSESS_MID
Character
7
Opacity Mid Range Precision Assessment
65
ASSESS~HI
Character
7
Opacity High Range Precision Assessment
66
CEMS_CNTRL
Character
15
Control Equipment for CEMS
67
REP_SYS_CN
Character
15
Control Equipment for Data Reporting System
68
MT_FRE_PRB
Character
25
Most Frequent Problem with CEMS
69
MO_FRE_PRB
Character
25
More Frequent Problem with CEMS
70
FRE PROBLM
Character
25
Frequent Problem with CEMS
71
LT_FRE_PRB
Character
25
Less Frequent Problem with CEMS
72
RATE_CEM
Character
2
Overall Rating of CEMS (1 poor, 10
recommend)
73
RATE_VENDR
Character
2
Overall Rating of CEMS Vendor (1 poor, 10
recommend)
TABLE 14
EXAMPLE dBASE LIST COMMAND
Record# First_Name
Last_Name
Telephone
Mon_Model
Parameters
20 David
Smith
(808) 462-9251
SM 810
N0X
51 Ken
Jones
(808) 249-8377
EX 4700
co2
53 Linda
Doe
(808) 330-7633
EX 4700
co2,co
63 Scott
Smith
(808) 393-1408
CM 50
N0X
6A-113
-------
Intentionally Blank Page
6 A-114
-------
Determination of Continuous Emissions Monitoring
Requirements at Electric Energy Inc.
6 A-115
-------
Intentionally Blank Page
6 A-116
-------
V.V. Bland
C.O. Hommel
Stone & Webster Engineering Corporation
7677 East Berry Avenue
Englewood, Colorado 80111-2137
B. Parker
Electric Energy, Inc.
Joppa Steam Electric Station
Box 165
Joppa, Illinois 62953
ABSTRACT
More stringent air pollution emissions monitoring requirements have resulted from
enactment of the U.S. Clean Air Act Amendments of 1990. Utility companies will
be required to retrofit Continuous Emissions Monitoring Systems (CEMS) to steam
generating units that are affected units under Title IV of the Act. Draft
regulations governing the requirements and operation of CEMS have been prepared
by the EPA. Final CEMS regulations are due to be promulgated by May 1992.
This paper discusses these proposed regulations and a range of options with which
the Electric Energy, Inc. Joppa Power Station may respond to the requirements.
A general survey of CEMS instrumentation, system configurations, and related cost
factors applicable to the Joppa Station are also presented.
6A-117
-------
INTRODUCTION
The U.S. government believes that the SO, allowance trading component of the Acid
Rain Program (Clean Air Act Amendments of 1990 (CAAA) - Title IV) is an
innovative, market-based approach to compliance with new SO, emission limitations
set by the Act. In order to allow this system to work, tne government further
believes that complete and accurate emissions data are the keys to implementation
of and confidence in the approach. Reliable Continuous Emissions Monitoring
System (CEMS) data is a critical component to the smoothly operating market that
is envisioned. To this end, EPA has proposed regulations to require that CEMS
be installed to monitor S02 emission concentration, volumetric flow rate, NO
emission concentration,diluent gas fraction, and opacity at each affected
facility. In addition, Section 821 of the Act requires that all affected units
in the Acid Rain Program monitor and report C02 emissions, although not
necessarily by continuous gaseous instrumentation techniques.
The CAAA establishes the requirements for CEMS through section 412 of the Act,
"Monitoring, Reporting, and Recordkeeping Requirements". The title defines a
CEMS as "...the equipment as required...used to sample, analyze, measure, and
provide on a continuous basis a permanent record of emissions and flow...as the
Administrator may require." The applicability of this section extends to any
source subject to Title IV (Acid Deposition control). Since all six steam
generating units at the Electric Energy, Inc. (EEI) Joppa Station have been
identified in Section 404 (Phase I) of the Act as affected units, CEMS must be
installed, operated, and certified by November 15, 1993 (due to a requirement of
the proposed regulations to conduct initial CEMS certification no later than 120
days prior to the above date, the actual installation, operation, and
certification deadline is July 15, 1993).
This paper discusses the proposed CEMS requirements that must be met at the Joppa
Station and presents a range of options that may be implemented to satisfy the
requirements.
PROPOSED REGULATIONS
Regulations that detail specifications and requirements for CEMS are currently
in force for new air pollution sources. In general, these existing regulations
are not as stringent as those proposed for the Acid Rain Program, but they form
the initial basis of the new requirements. Much of the data upon which the new
Acid Rain Program CEMS requirements are based were provided through experience
with the existing instrumentation and systems used for new source monitoring
requirements.
6A-118
-------
The proposed CEMS regulations discussed in this paper were developed for the
Federal Office of Management and Budget (OMB) and the Acid Rain Advisory
Committee (ARAC). These preliminary draft regulations will be reviewed, modified
if necessary, and published by EPA for comment in the Federal Register in the
fall of 1991 (perhaps prior to presentation of this paper). After close of the
conment period, the proposed regulations will once again be modified, then
finalized and promulgated by May of 1992. This process will allow opportunity
for significant changes to occur to the currently proposed regulations.
General CEMS Requirements
Under the proposed rule, the owner or operator of an affected unit (or units)
would be required to install a CEMS on each affected unit unless otherwise
specified in the regulation. The CEMS is defined as including the following
components: (1) an SO, pollutant concentration monitor, (2) a N0X pollutant
concentration monitor, (3) a volumetric flow monitor, (4) an opacity monitor, (5)
a diluent gas monitor, and (6) a data acquisition and handling system (usually
computer-based) for recording and performing calculations with the data.
Measurements of S02 concentration must be combined with measurements of
volumetric gas flow (exhausting from the unit) to obtain estimates of S02 mass
emissions per unit time (in lbs/hr), as required by the Act. Flow monitors
always measure gas flow rate on an actual or "wet" basis. Some SO, pollutant
concentration monitors, however, measure SO^ concentration on a "dry" Dasis. The
measurements used to determine SO, emissions in lbs/hr must be on the same
moisture basis. Accordingly, units that employ "dry" S02 pollutant concentration
monitors must correct their gas flow rate measurements for moisture. Under the
proposed rule, EPA would allow any moisture determination method, including
standard saturation/temperature tables and continuous moisture monitors, provided
the corrected flow rate measurements satisfy the performance standards for
monitor certification (i.e. hourly averages, relative accuracy/bias requirements,
etc.).
Similarly, measurements of N0X concentration must be combined with the
appropriate EPA F or Fc factor and measurements of a diluent gas, either oxygen
(02) or carbon dioxide (CO,), exhausting from the unit to obtain the estimates
of the NO emission rate relative to the heat input of the fuel (in Ibs/MMBtu).
Accordingly, the proposed rule defines a NO CEMS as the combination of a N0X
pollutant concentration monitor and a diluent gas monitor.
Only an opacity monitor is needed for monitoring the obscuration caused by
particulate matter in the gas; actual particulate loadings are not required by
Title IV of the CAAA.
Units that monitor CO, continuously could use a flow monitor to estimate C02
emissions in lbs/hr, which are to be aggregated into daily totals for reporting.
The proposed rule would require only some units (i.e., units that generate C02
emissions by means other than fuel combustion, for example, by wet limestone
scrubbers during the flue gas desulfurization process) to continuously monitor
CO, emissions discharged into the atmosphere. Most units would be allowed to
calculate C02 mass emissions (in lbs/day) using specified methods and procedures
based on the measured carbon content of the fuel and the amount of fuel
combusted.
6A-119
-------
Under the proposed rule, each monitor in the CEMS and the system as a whole must
be installed, and its performance verified and certified by the EPA, before it
can be used in the Acid Rain Program.
Performance Certification Requirements
The following performance certification tests would be required for continuous
emission monitoring systems: (1) calibration error tests for each pollutant
concentration monitor and diluent gas monitor; (2) an electronic stability test
for the flow monitor; (3) relative accuracy and bias tests for the S02 pollutant
concentration monitor, the flow monitor, and the N0X emission monitoring system;
(4) a cycle response time test for the SO, pollutant concentration monitor and
the N0X emission monitoring system; and (5) an orientation sensitivity test and
an interference test for differential pressure flow monitors only.
No later than January 1, 2000, relative accuracy and bias tests would be added
for the combined S02 emission monitoring system (pollutant concentration monitor
and flow monitor). For continuous opacity monitoring systems, performance
certification tests for calibration error, response time, zero drift, and
calibration drift would be conducted according to the requirements in 40 CFR 60,
Appendix B.
For each CEMS, the proposed rule also requires the development and implementation
of a written quality assurance/quality control plan. Daily performance checks
of the monitoring equipment, including gas calibration error tests and visual and
electronic inspections, would be required by the proposed rule. In addition,
test audits and bias tests would be required for the SO^ pollutant concentration
monitor, the flow monitor, and the N0X emission monitoring system. A three-point
calibration error test would also be required quarterly for all pollutant
concentration and diluent monitors.
Alternative Monitoring Systems
No alternative monitoring system has been proposed as a preapproved system
equivalent to a CEMS on the required criteria of precision, reliability,
accessibility, and timeliness.
In order to receive approval to use an alternative monitoring system in lieu of
a CEMS or a component of a CEMS (e.g., S02 pollutant concentration monitor or
flow monitor), the affected unit would be required to submit long-term
statistical evidence and other data that demonstrate the proposed alternative
would provide information equivalent or superior to a CEMS. Under the proposed
rule, EPA would use the performance of certified S02 pollutant concentration
monitors, flow monitors, and NO emission monitoring systems as benchmarks for
approving or rejecting proposals for alternative monitoring systems. The
proposed CEMS regulations specify procedures, analyses, and supporting
documentation that would be required for demonstrating the equivalency of
alternative monitoring systems to CEMS on the required criteria of precision,
reliability, accessibility, and timeliness.
Phase I Qualifying Technology
Affected units which apply for and are granted approval to implement the optional
compliance method using Phase I qualifying technology (e.g., achieves a 90-
6A-120
-------
percent reduction in S02 emissions) would have to employ additional monitoring.
The proposed rule requires that each such unit be equipped with an SO, pollutant
concentration monitor and a flow monitor for measuring SO, emission at the inlet
to the control device in addition to the required monitors for measuring SO,
emissions discharged to the atmosphere. Provisions are included in the proposed
rule for demonstrating achievement of the required 90 percent reduction in SO,
emissions through Phase I qualifying technology, on an annual basis, from 1995
through 1999.
Common Stack
The proposed rule would allow (or perhaps require) EEI to combine S02 allowances
according to the procedures in 40 CFR Part 73 and install one monitoring system
where two or more affected units utilize a common stack.
N0X emissions could be determined in the exhaust from a single unit or measured
in a common stack. It is not currently clear, however, if N0X emissions from
discrete and separate stacks or flues may be averaged to represent multiple units
or a plant wide "bubble".
CEMS Availability
All CEMS would be required to be in continuous operation and to be capable of
sampling, analyzing, and recording at least every 15 minutes. All emissions and
flow data would be reduced to one-hour averages. Four data points would comprise
a valid hour. During calibration or other required quality assurance activity
periods, however, two or more data points would be allowed to comprise a valid
hour. Failure of the system to acquire the required data points would result in
the loss of data for the entire hour. In this event, the utility would be
required to use prescribed procedures for calculating emissions for the missing
data periods.
The proposed rule contains procedures for compiling "information satisfactory to
the Administrator" for substituting data where no valid data have been recorded
for the S02 pollutant concentration monitor, the flow monitor, or the N0X CEMS
(consisting of the N0X pollutant concentration monitor and the diluent gas
monitor).
For the SO, and flow monitors, where valid data have not been recorded for either
monitor, the missing data procedure would apply to each monitor individually.
For the N0X CEMS, if either monitor (NO monitor or diluent monitor) is without
a valid hour of recorded data, the data for both monitors would be deemed
invalid, and substitute data must be provided for both monitors using the
prescribed missing data procedures. Such information establishes preapproved
information satisfactory to the Administrator. The proposed approach establishes
the methods that may be used to "fill in" missing data, following the general
principle that the longer the gap in the recorded data and/or the lower the
annual monitor availability, the more conservative the value to be substituted.
Annual monitor "availability" refers to the number of total hours of valid data
capture per year, expressed as a percentage of total unit operating hours.
6A-121
-------
Table 1 sunmarizes the proposed missing data procedures. Three availability
categories are identified by the EPA:
1) A*95%
2) 90*
-------
AVAILABLE CEMS EQUIPMENT
The need for the development of Continuous Emissions Monitoring technology was
a direct result of the Clean Air Act and the Code of Federal Regulations.
When Congress passed the clean air act in 1970, there were few, if any, true
continuous emissions monitoring systems available. There were a few instrument
manufacturers that packaged some laboratory grade instruments, but there were no
systems designed to operate on a continuous basis while being located in the
utility plant environment. After the clean air act was passed, the EPA and some
states began to require CEMS installations which created a market to which many
analyzer manufacturers and system integrators responded. Unfortunately, due to
a lack of uniform monitoring requirements, when CRF 40, Part 60, Subpart A,
Section 60.13, (Monitoring Requirements) was promulgated in 1975, the majority
of CEMS installed prior to this date could not comply with this performance
standard.
Over the years since 1975, there have been attempts by various manufacturers to
revolutionize CEMS technology, but as of this date, the most successful CEMS
employ analyzer technology that was designed more than a decade ago. Many of
these analyzers have been updated by adding some state-of-the-art electronics,
but the basic designs have not changed.
In addition to providing a permanent record of emissions, most sources with a
CEMS are also required to report their emissions on a periodic basis to the local
and/or state and/or federal air quality organization. To automatically produce
the required reports directly from the analyzer outputs, most CEM suppliers also
provide (at extra cost) a Data Acquisition System (DAS). Some suppliers have
attempted to supply DAS systems with canned software packages which, through menu
driven options, allow an operator to make keyboard selections of the calculations
to be made, the report format and the frequency of the reports. Unfortunately,
there has been no uniformity of reporting requirements by the local districts,
the states or even between EPA regions. Therefore, it is common that capital
costs for a CEMS also include costing for the DAS computer and printer and
substantial computer programmer time for development of the customized software.
This will most likely still be the case in responding to the proposed CEMS
requirements driven by Title IV of the CAAA.
Gaseous monitoring
Current CEMS technology exists which can accurately and reliably measure the
normally permitted gases such as NO^, S02, 02, CO, C02, NHj, and Hydrocarbons.
A variety of techniques are available for the measurement of these gaseous air
pollutant emissions from combustion sources. Among these technologies are: In-
situ, Conventional Extractive, and Dilution Extractive monitoring systems. Each
of these techniques offers advantages and disadvantages, dependent upon the
specific application requirement.
In-Situ. Under the in-situ category, there are two basic types of systems. The
two types are single point (Figure 1) and cross stack (Figure 2) systems.
The single point in-situ system with sensing elements inserted into the gas
stream produces an electrical output signal proportional to the concentration of
the gas being measured. The cross stack system projects a light beam through the
6A-123
-------
sample to be analyzed by a receiver located on the opposite side of the stack.
The single point system has less sensitivity than the cross stack system because
of the significantly shorter measurement path length. However, the single point
system has the ability to be dynamically calibrated with calibration gas which
is a definite requirement of the proposed CEMS regulations. There are two
analytical methods used by the single point system. The first is an
electrocatalytic type analyzer which uses a continuous flow of calibration gas
as a reference across a sample cell. The second is a second derivative
ultraviolet spectroscopy analyzer that inserts a protected mirror into the gas
stream which provides a return path for the ultraviolet light source to measure
the absorption caused by the component of interest.
The advantages of in-situ are:
1. Standard Design
2. Low Purchase Price
3. Low Installation Cost
4. Low Scheduled Maintenance
The disadvantages of in-situ are:
1. Cross stack version would be uncertifiable under proposed
regulations
2. Single point system would have difficulty passing new "bias" test
requirements of the proposed regulations.
3. Limited gas measurement capability
4. Limited ability to measure low concentrations
5. System exposure to hostile environments
6. Limited operating temperature
7. Some types have high operating costs
8. Some are difficult to verify accuracy which makes quarterly audits
very expensive
9. Single analytical technique, not best for all gases
10. At Joppa, equipment would have to be installed in a location that
would make maintenance more difficult.
In-Situ gas monitors are not generally recommended for application at Joppa,
primarily because of disadvantages 1 and 2.
Conventional Extractive. An extractive CEMS (Figure 3) withdraws an unaltered
sample of the flue gas to be processed for analysis at some remote location.
This flue gas is protected by maintaining, or, in some cases, increasing the flue
gas temperature as it is being transported. It is also necessary to prohibit the
flue gas sample from contacting any material that could alter the concentration
of the sample until conditioning is complete. When conditioning is complete,
only the particulate matter and moisture have been removed from the flue gas
leaving all other components unaltered. After conditioning, the gases are
provided to a gas manifold which distributes the flue gas to each analyzer.
Conventional extractive systems can be configured to accomplish gas analysis
prior to removal of moisture in the "hot-wet" approach. This approach, while
generally increasing system costs, would be more consistent with elements of the
proposed CEMS regulations that require determination of S02 emissions on a wet
flue gas basis. Alternatively, if the gas sample is analyzed "dry", a moisture
correction factor will be required, complicating the measurement and reporting
process.
6A-124
-------
The conventional extractive system allows the design to include many different
analytical techniques within one system in order to offer the best analytical
technique for each gas component being analyzed. The typical techniques employed
by an extractive system are NDIR, NDUV, flame-ionization, chemiluminescence,
paramagnetic, and electrochemical sensing cells.
Some conventional extractive systems elect to use a single analyzer to measure
all gaseous components. This approach may reduce overall system costs.
The advantages of conventional extractive systems are:
1. Flexible, usable in most applications
2. Accuracy
3. Verifiable
4. Moderate operating cost
The disadvantages of conventional extractive systems are:
1. High capital cost
2. High installation cost
3. Long runs of expensive, high temperature sample line
4. Sample conditioning system maintenance intensive
5. Negative pressure system creates leak potential
Dilution Extractive. Under the dilution extractive category, there are two types
of systems; (1) a dilution probe system that dilutes the sample within the probe
(Figure 4) and (2) a dilution box which dilutes the sample in a box just down
stream of the sample probe. The dilution systems are designed to extract a small
sample'of flue gas and dilute that flue gas with large amounts of clean and dry
air. The clean and dry air lowers the moisture dew point to an acceptable level
for analysis by an analyzer which has been designed for ambient air monitoring.
The dilution ratios can be as high as necessary to dilute the source
concentration to ambient air levels. The dilution ratio is controlled by the
dilution air pressure and a critical orifice. Dilution ratios as high as 350:1
are common.
The advantages of dilution type systems are:
1. No heat traced sample line required
2. No sample conditioning required downstream of dilution
3. Low maintenance requirements in proper application
4. Lower initial cost than conventional extractive
5. Positive pressure system minimizes leak potential
6. Uses well proven ambient monitoring instrumentation
The disadvantages of dilution type systems are:
1. Some parameters may be diluted below analyzer sensitivity range
2. Requires high purity air for operation
3. Potentially slower response time
Volumetric Flow Honitori no
There are three main types of flow monitors currently being used for the
continuous monitoring of flue gas flow: ultrasonic, differential pressure, and
6A-125
-------
thermal. Although all flow monitors estimate the flue gas flow rate by
multiplying the cross-sectional area inside the flue (stack) by the average gas
velocity, each type employs a different principle to measure average gas
velocity. Ultrasonic flow monitors determine average gas velocity directly by
measuring the time it takes for sound bursts to travel between two transceivers,
one located downstream of the other. Differential pressure flow monitors
determine average gas velocity by measuring the pressure at one or more points
in the flue gas stream, and using the established relationship between gas
pressure, temperature, molecular weight, and velocity. Thermal flow monitors
measure the difference in temperature between a heated and an unheated element
in the flue gas stream.
While flow monitoring is a proven technology, the proposed CEMS regulations
represent the first major air pollution control regulations to require flow
monitors for the continuous monitoring of flue gas flow. Accordingly, utilities
in the U.S. have had limited experience in the installation, operation, and
maintenance of flow monitors for this particular application. This limited
experience has led to some concern regarding the reliability and accuracy of flow
monitors, particularly in wet stack environments.
EPA believes, however, that available knowledge is sufficient to support the
proposed requirement for flow monitors in Phase I as well as Phase II of the
CAAA.
In many cases, single point flow monitoring will not be allowed. For
applications at the Joppa plant, ultrasonic flow monitoring would probably be the
first choice for well behaved flow locations with relatively flat velocity
profiles (e.g., the typical stack test level). For locations that could exhibit
gas flow irregularities and maldistribution (e.g. induced draft fan outlet
manifold flue), a pitot tube array may be preferable.
Opacity Monitoring
Opacity monitors have been in service for a number of years and, in general, have
performed reliably during that time. The opacity monitoring technique acceptable
to EPA is a measurement system based upon the principle of transmissometry.
Light having specific spectral characteristics is projected from a lamp through
the effluent in the stack or duct, and the intensity of the projected light is
measured by a sensor. The projected light is attenuated because of absorption
and scattered by the particulate matter in the effluent; the percentage of
visible light attenuated is defined as the opacity of the emission. Transparent
stack emissions that do not attenuate light will have a transmittance of 100
percent or an opacity of zero percent. Opaque stack emissions that attenuate all
of the visible light will have a transmittance of zero percent or an opacity of
100 percent.
Several opacity monitors based on this principle are commercially available and
would perform well in satisfying Joppa CEMS requirements. One limitation
affecting these opacity monitoring systems is that they generally cannot be
located downstream of a wet FGD system.
JOPPA SITE OPTIONS
There are a number of CEMS alternatives that are created as a result of the
specific requirements and layout of the Joppa plant steam generating units.
6A-126
-------
These alternatives involve options derived from the general arrangement of plant
equipment (e.g. determination of monitoring configurations that best satisfy
proposed regulations) and from other factors that could significantly impact
design choices (e.g. using the NO„/diluent system to fine tune boiler operation
in addition to providing compliance data).
To address this assortment of requirements and goals, the application scenarios
presented in Table 3 have been developed. Additional scenarios exist, but may
generally be considered a variation (or combination) of those presented.
Each of the configurations mentioned in the table are briefly discussed below.
Configuration - One CEMS each stack
Configuration (1) represents the minimum CEMS installation. These three systems
would be installed, one in each stack at the existing stack test level. New
ports would be installed at this level to accommodate the opacity monitor, flow
monitor, and gaseous sampling probe (for SO,, N0X, and C02 or 02 samples).
Existing ports would still be usable to conduct particulate tests and newly
required instrument certification tests (nominally EPA Test Methods 2, 3, 4, 6,
and 7). Sample gas via the sample probe and line would be conveyed to
instrumentation located at grade. Sample handling and conditioning equipment,
gas analyzers, control modules, span gases, etc. would all be housed in a stand
alone enclosure near the stack base, in a suitable environmentally acceptable
existing structure (e.g. within the base of each stack), or contained in an
enclosure placed within an existing structure (many choices are available but
final location selection should be made after vendor recommendations). At this
instrumentation location, gas samples would be further conditioned (if
appropriate) and analyzed. Electronic output from opacity monitors and flow
monitors would be transmitted to the same general location (Figure 5 shows a
typical schematic of this arrangement).
Analyzer signal output would be stored in a data logger and input to the CEMS
data handling and reporting system. The major components of the CEMS data
handling and reporting system could be remotely located at the respective control
room or at a central CEMS data processing center. At a minimum, it would be
desirable to locate strip chart recorders in the appropriate control room to
allow operator tracking of important CEMS data items.
Gaseous monitoring would be by the conventional extractive or dilution extractive
technique. A stack mounted transmissometer would be used to determine opacity
and an ultrasonic system would, most likely, be used to monitor flow. These
general approaches to specific monitoring tasks were described previously in this
paper, as was the rationale for preferring one system to be used at the Joppa
station over another potential system.
Configuration (2) - S0,/f1ow/opacitv in stack, time shared N0X in ducts
Configuration (2) also contains three basic CEMS with some significant
modifications from Configuration (1). For Configuration (2) opacity and flow
would still be monitored in the stack at the stack test level (Figure 6). A gas
sample would also still be withdrawn from this location to provide S02 emissions
data for both units exhausting to the stack. N0X and diluent gas samples (CO,
or 0,), however, would be withdrawn from the induced draft (ID) fan outlet
manifold of each steam generation unit, just prior to commencement of the common
6A-127
-------
duct connecting to the stack. In this manner, a discrete sample from each unit
may be analyzed for N0X concentration. This data may then be used not only for
compliance determination, but as an operational tool for adjusting and optimizing
marginal N0X emissions from each boiler. A composite sample taken from the stack
would not be usable for this latter purpose.
In general, the same amount and description of instrumentation and auxiliary
systems as in configuration (1) will be required.
To accomplish the dual purpose NO monitoring objective with the same amount of
instrumentation as that indicated for configuration (1), time sharing of the
NOx/diluent monitoring system between the first and second steam generation units
would be required. Although not expressly prohibited by the proposed
regulations, various new requirements for instrument certification may
essentially eliminate "time sharing". If this proves to be the case,
Configuration (3) could be applied.
Configuration (3^ - S0,/f1ow/opacitv in stack, dedicated N0JL in ducts
This configuration is the same as (2) except that an additional NOx/diluent
system could be installed in each CEMS. This arrangement would avoid limitations
presented by a time-share system as discussed above.
Configuration - One CEMS each stack with one portable backup CEMS
Configuration (4) would be identical to Configuration (1) or Configuration (3)
except with the addition of a backup CEMS. The purpose of this backup system is
to increase overall CEMS reporting reliability and availability. Because of the
high potential economic impacts of operating systems with low availability,
redundant systems may be highly desirable.
One approach would be to operate a complete backup CEMS in hot stand-by mode.
This system could be fully portable and designed for quick relocation and hookup
(within 1 to 2 hours) to any of the three in-service systems that were
experiencing operational difficulties or were deemed to be out of control under
the proposed regulations.
It would be necessary to certify, maintain, and quality assure this system in the
same manner as the three dedicated systems.
Configuration (51 - 6 CEMS in ducts or stacks
This configuration is presented to further increase system reliability and
availability above that allowed by previous configurations. Six CEMS would be
installed in the I.D. fan outlet ducting and designed so that SO, and N0X
emissions requirements would be met on an individual unit basis. Reliability
would be improved by configuring the systems for time-sharing on sister units in
case of a CEMS failure. Penalties for low availability would also be minimized
due to the fact that an instrument system outage in this configuration would only
require high substitute values to be recorded for one unit's emissions instead
of two.
If time-sharing or other limitations discourage or prohibit duct installation of
the 6 CEMS units, an alternative could be employed. This would be the
6A-128
!
-------
installation of two CEMS in each stack for a total of six CEMS for the plant.
In this manner, each on-line stack CEMS would be continuously backed up by
another complete system held in the hot standby mode.
CEMS Costs
Estimates have been made of relative cost factors associated with each of these
CEMS configurations. These factors are presented to allow relative economic,
technical, and regulatory impact comparisons. Costs related to chimney and/or
test platform modifications to accommodate CEMS and related activities are not
included.
Relative cost factors include the following CEMS specific cost elements:
• Equipment costs
• Installation costs
• CEMS Certification
• Training
• Quality Assurance Plan
• A&E Services
Table 4 presents the data.
6A-129
-------
I
REFERENCES
1. U.S. Clean Air Act Amendments. Titles IV and VIII. Public Law 101-549.
November 15, 1990
2. EPA Draft Proposed Continuous Emission Monitoring Regulations. OMB/ARAC
Draft. June, 1991.
3. Protection of Environment. 40CRF 53 to 60. U.S. Code of Federal
Regulations: Revised as of July 1. 1990. Office of the Federal Register
National Archives and Records Administration.
4. Continuous Emissions Monitoring. R-C EST Report, December 1989.
5. U.S. Environmental Protection Agency. 1979. Continuous Air Pollution
Source Monitoring Systems Handbook. EPA 625/6-79-005. Cincinnati, OH
6A-130
-------
UV LIGHT MOOULfiTED
BT US ABSORPTION
retroreflector
LIGHT
filter
Figure 1. Single Point In-Situ System
STACK
NEUTRAL PUER
SOURCE
Figure 2. Cross Stack In-Situ System
6A-131
-------
SAMPLE PROBE
OAS FLOW
SAMPLE MTERFACE
ENCLOSURE
SAMPLE LINE
T3
a J I
REMOTE MOMTORS
Figure 3. Conventional Extractive System
OOXmON PROBE
OAS FLOW
SAMPLE IMC
J l!
RBIOTEMOMTOR
Figure 4. Dilution Extractive System
6A-132
-------
SIGNALS FROM
FLOW AND OPACITY
MONITORS
STACK
LEVEL
GAS SAMPLE LINE
(SO2.NOx.CO2)
STACK
CaiS SIGNALS
CONTROL ROOM
REMOTE DATA
PROCESSING
CENTER
GAS SAMPLE CONDITIONING
AND ANALYSIS ENCLOSURE
Figure 5. Configuration (1) - Three Dedicated CEMS Installed One in Each Stack
SIGNALS FROM
FLOW AND OPACrTY
MONITORS
FLUE GAS FROM FIRST UNIT
so2sampleune
STACK
TEST
LEVEL
NOx ICO 2
SAMPLE LINES
STACK
OEMS SIGNALS
GAS SAMPLE CONDITIONING
AND ANALYSIS B4CLQSURE
FLUE GAS FROM SECONO UNIT
CONTROL ROOM
REMOTE DATA
PROCESSING
CENTER
Figure 6. Configuration (2) - CEMS Compliance Monitoring
With Unit Specific N0X Emissions Data Option
6A-133
-------
TABLE 1
SUMMARY OF CEMS SUBSTITUTION CRITERIA FOR
ESTIMATING VALUES FOR MISSING DATA PERIODS
Annual availability (%)
of monitor or system
Number of
hours missing
(N)
Value substituted for each missing
hour
Greater than or equal to
95%
N £ 24 hours
N > 24 hours
Average of N hours recorded before
missing data period and N hours
recorded after missing data period
Maximum hourly value recorded in
previous 30 days of service
Less than 95% but greater
than or equal to 90%
N < 3 hours
N > 3 and < 24
hours
N > 24 hours
Average of the hour recorded before
missing data period and the hour
recorded after missing data period
Maximum hourly value recorded in
previous 30 days of service
Maximum hourly value recorded from
previous 365 days of service
Less than 90%
N > 0 hours
Maximum recorded hourly value for
the monitor since initial service
TABLE 2
LOST VALUE OF S02 ALLOWANCES
FOR 1-HOUR OF CEMS DOWNTIME
(for CEMS with <90% reliability and
full load boiler operation with 100% Black Thunder Coal)
Estimated Worth of SO, Allowance ($)
200 400 600 800
FUEL AND CEMS SCENARIO
Lost Value of Allowances ($)
50% bituminous/50% subbituminous burned
at any time since 7/15/93
Shared CEMS
Between two units
420 840 1,260 1,680
100% bituminous burned at any time since
7/15/93
Shared CEMS
between two units
780 1,560 2,340 3,120
6A-134
I
-------
TABLE 3
JOPPA STATION CEMS CONFIGURATIONS
Configuration Description
No. of
CEMS
Comments
(1) One CEMS for each stack,
sample probes and sensors located
at the stack test level
3
Combine units' allowances
acceptable under proposed
regulations
(2) S02/flow/opacity probes and
sensors located in each stack at
the stack test level. NO /Diluent
probes located at each steam
generation unit outlet
3
NOx/Diluent time shared and used
for diagnostics and to fine tune
boiler operation
(3) Same as (2) above except with
a NOx/Diluent system dedicated to
each steam generation unit
3
(+ 3 extra
NOx/Di1uent
systems)
Time sharing of compliance
monitors may seriously challenge
proposed EPA requirements
demanding dedicated monitors.
Dedicated monitors may be
required.
(4) Same as (1) except with
portable backup CEMS
4
Increased reliability
(5) Six CEMS - one per unit,
Opacity Monitors located in
stacks at the stack test level,
all other probes and sensors
located at each steam generation
unit outlet
6
Systems configure to allow time
sharing in case of CEMS failure
if allowed by EPA. Increased
reliability
6A-135
-------
TABLE 4
COST FACTORS FOR JOPPA STA1:ON CEMS CONFIGURATIONS
I Configuration Description
A
Relative Cost Factor
1 Confiauration (1)
1 -1 CEMS installed in each of 3 stacks
1.00
Confiauration (2)
-Same as Configuration (1) except N0x/C02 measured
(time-shared) at each unit outlet
1.03
Confiauration f3)
-Same as Configuration (2) except with N0x/C02
monitors dedicated to each unit
1.12
Confiauration (4)
-Same as Configuration (1) except with a backup CEMS
1.32
Confiauration (5>
-1 CEMS installed in each of 6 units except with
opacity monitors located in 3 stacks
1.81
I
6A-136
-------
Improving Performance of Flushless Mechanical Seals
in Wet FGD Plants through Field and Laboratory
Testing
6A-137
-------
Intentionally Blank Page
6 A-138
-------
F. E. Manning
R. L. Grace
BW/1P International, Inc.
27941 Front Street
Temecnla, CA 92390
ABSTRACT
With the need to control water-balance and process efficiency, mechanical seals requiring
no flush water are acknowledged as an important component of the Flue Gas
Desulfurization (FGD) process.
For mechanical seals to be applicable in FGD systems, the seals must handle a variety of
process conditions and upsets, including situations where high percentages of abrasive
solids such as flyash enter the process stream.
To improve reliability and performance of flushless mechanical seals in centrifugal slurry
pumps in FGD applications, comprehensive laboratory and field testing was undertaken.
This paper reports the results of the laboratory and field testing, explores modes of
failure in these applications, defines ways to improve Mean Time Between Failures and
demonstrates that flushless mechanical seals can operate successfully in highly-abrasive
applications.
6A-139
-------
INTRODUCTION
With the need to control water-balance and process control, mechanical seals requiring
no flush water (flushless) are an important component of the Flue Gas Desulfurization
(FGD) process. Most current FGD scrubber projects specify flushless mechanical slurry
seals in place of packing. In recent literature exploring design and operation of FGD
systems for cycling service, flushless mechanical seals were recommended to reduce
waste water generation (1). Although waste water production varied for the different
types of FGD scrubbers evaluated, production of waste water was thought to be
significant in all cases during load changes and low output. Since flushless mechanical
slurry seals use no flush water, a significant reduction in water consumption is realized.
For example, flushless sluny seals can be used on recycle pumps to reduce water
consumption, to eliminate dilution of lime reactant sluny and reduce water content in
waste slurry transferred to settling ponds. Eliminating flush water usage provides the
opportunity to introduce fresh water into the scrubber process at locations where dilution
is beneficial.
The wet FGD process creates a variety of operating conditions for the flushless slurry
seaL In absorber recycle pumps, solids range from 10-15% by weight In contrast, a
thickener underflow pump may reach 40% solids by weight FGD slurries consisting of
lime or limestone and gypsum also contain varying amounts of flyash. Flyash is very
abrasive because it is largely composed of aluminum oxide and silicon dioxide. Flyash
content will normally increase during load changes and can reach high levels in the event
of a precipitator or prescrubber failure. The slurry seal is also subjected to pH levels
ranging from 2-10 and Chloride levels that may approach 100,000 ppm. To provide long
service life, the flushless slurry seal must provide dependable performance in the above
operating conditions.
Important consideration must be given to the sluny pump when adapting the flushless
slurry seaL Slurry pumps are heavy and rugged by design and do not typically have the
concentricity and precision fits associated with process pumps. Axial and radial
clearances found in certain bearing arrangements allow significant shaft movement The
6A-140
-------
mechanical slurry seal must accept normal misalignment and shaft deflection found in
slurry pumps.
This paper provides practical guidelines to improve the performance and life of flush]ess
mechanical slurry seals in FGD and related applications. Information presented in this
paper is based on years of Field Experience and results of extensive Laboratory Testing.
The recommendations given in this paper are consistent with designs currently supplied
by many slurry pump manufacturers.
MECHANICAL SEAL DESIGN CONSIDERATIONS
Slurry is a mixture of solids suspended in a liquid. Certain criteria must be met for
successful operation of a flushless mechanical seal operating in a slurry. By design,
slurry seals must be different than conventional mechanical seals used in the
petrochemical industry.
Conventional seals normally use a single coil or multiple pocket springs to provide
uniform load on the dynamic seal ring (Fig. #1). These springs normally are exposed to
the product and will pack or scale when used in slurry applications. This prevents
movement of the spring and can lead to seal failure. In addition, conventional seals have
a secondary sliding gasket which seals between the shaft or flange and the rotating or
stationary face. The close clearances between these sliding components is susceptible to
packing with solids. If these close fits become packed with solids, flexibility of the seal
face is limited and can lead to seal failure. Welded diaphragm rotating bellows seals are
considered self cleaning and are used in some slurry services. These bellows designs are
susceptible to abrasive wear, solids buildup, stress corrosion and fatigue, which can limit
effectiveness in high concentration slurries.
Figures #2-4 show three examples of flushless mechanical slurry seal arrangements. All
three slurry seal designs utilize a spring system that is protected from the slurry. This is
done by encapsulating the spring in rubber or using rubber as the spring (rubber in
shear) or placing the springs outside of the slurry. The encapsulated cone spring and the
rubber in shear spring designs are inherently non-clogging designs and have no sliding
fits. In the spring pusher slurry seal (Fig. #4), a specially designed o-ring groove is
utilized to prevent packing of solids and allow movement of the dynamic seal ring.
PUMP SHAFT TO HOUSING ALIGNMENT AND SHAFT DEFLECTION
A good slurry seal design must include self aligning features. Many slurry pumps do not
provide adequate alignment of shaft to stuffing box mounting surfaces for good
6A-141
-------
mechanical seal performance. Figures #5-8 demonstrate the various alignment concerns
that may be encountered in a typical slurry pump.
Figure #5 shows the shaft to stuffing box bore concentricity. Worn fits found on older
pumps will aggravate this type of misalignment. This problem can be overcome by
mounting a floating flange or adapter that allows centering the seal to the pump shaft
Figure #6 shows perpendicularity of pump shaft to stuffing box mounting surface. This
relationship is built into the pump and difficult to overcome. Excessive run-out in this
area causes several problems. If the slurry seal is of the rotating spring design,
misalignment will cause the springs to flex with every revolution. This may cause
leakage and lead to seal failure. In the case of the stationary spring design, face load
will be uneven. It is possible to eliminate these problems by attaching the stationary seal
member to a flexibly mounted flange or to the bearing frame (Fig. #2).
Figure #7 shows axial travel of the shaft This motion is the result of bearing clearances
in most cases, but can be caused by clearances in the impeller adjusting mechanism.
Before installing any mechanical seal, bearing condition should be checked and axial
travel should be limited to the bearing manufacturers recommendations.
Figure #8 shows shaft sag and hydraulic offset This is a function of shaft overhang,
bearing radial clearances and hydraulic loading of the impeller during operation. Slurry
seals are designed to tolerate normal deflections of this type by incorporating wider seal
faces with matching wearing surfaces.
MATERIALS OF CONSTRUCTION
In FGD services, the primary concerns are chemical resistance to slurries with low pH
and high chloride levels and abrasion. The choice of materials for metal components
must be based on a knowledge of the operating conditions of the FGD system. To
provide good seal life in these conditions, materials ranging from 316SS, CD4MCu,
Hastelloy and High Chrome Iron are available.
With the abrasive and chemical nature of FGD slurries, seal face material selection is
critical. Results from field installations and laboratory testing confirm that Silicon
carbide vs Silicon carbide gives the best seal performance. The performance of Silicon
Carbide can be attributed to its high hardness, thermal conductivity, chemical resistance
and excellent sliding properties. Ethylene propylene rubber (EPR) is normally used for
gaskets and rubber components and has provided excellent performance in water based
slurries.
6A-142
-------
SEAL CHAMBER DESIGN CONSIDERATIONS
Seal chamber design is important for successful seal operation. Studies bave been
conducted to evaluate how seal chamber design effects mechanical seal performance in
process pumps (2)(3). In these studies, radial clearance between the mechanical seal and
stuffing box bore were evaluated to determine their influence on liquid behavior around
the seal faces. Mechanical seal feces rely on the product or flush media for lubrication
and cooling. Results of these studies concluded that close clearances between rotating
seal components and the stuffing box bore create heat buildup and limit product
circulation around the seal faces. For flushless or dead-ended seals, large stuffing box
bores and open throat areas provide the best environment by promoting product
circulation in the seal chamber.
Similar studies have been conducted on slurry services (4). Results of these studies
report that a large radial clearance is needed to promote circulation and that a tapered
seal chamber will provide additional benefits. The natural swirling flow of slurry in a
tapered seal chamber centrifuges heavy abrasives from the seal rotating parts and
provides better heat removal. In addition, the tapered seal chamber design is self
venting during startup and operation and retains no slurry when the pump is drained. A
self venting tapered or open seal chamber provides the best environment for flushless
mechanical slurry seals. Figures #9 & 10 show tapered and open seal chamber designs,
with back vanes removed from the impeller.
DISCONTINUITIES IN THE SEAL CHAMBER
Discontinuities in a self venting seal chamber upset the uniform flow and create localized
pockets of turbulence. These discontinuities can be in the form of strakes or drilled vent
or drain ports. The effects of this turbulence can produce accelerated wear of pump and
seal components. In cases where a high percentage of solids or large particle abrasives
are found, wear can be heavy. Impeller back vanes that extend into the seal chamber
cause high flow rates and produce excessive turbulence, which accelerate abrasive wear.
A properly configured self venting seal chamber eliminates the need for a vent or flush
port.
IMPELLER BACK VANES
Impeller back vanes are designed to reduce pressure in the stuffing box. Efficient back
vanes can produce a vacuum in the seal chamber under conditions of low suction and
low discharge pressure and high flow (Fig. #11). This condition robs the mechanical
seal of lubrication and cooling and can cause seal failure. This condition is typical at
start-up where little or no pressure is in the discharge pipe. Maintaining static head on
the discharge or throttling the discharge on start-up greatly reduce the
6A-143
-------
potential of this problem. In cases where pump discharge pressure is high, back vanes
reduce seal chamber pressure, which may be beneficial to the slurry seaL
LABORATORY TESTING
Most testing of slurry seal designs has been conducted using water for the test medium.
Although this testing yields some measure of seal performance, it does not address the
concern of abrasive wear to seal faces and adaptive components. To determine effects
of slurry solids on seal performance, requires a test rig capable of circulating slurry.
A unique test rig was constructed with the express purpose of testing mechanical seals in
abrasive sluny(S). Refer to Figures #12 and 13. The design utilizes two slurry pumps in
series. The first stage pump is used to circulate the slurry and provide increased suction
pressure to the second test pump. The second stage pump, the main test pump, is
coupled to a variable speed drive which provides a range of shaft speeds and
corresponding seal chamber pressures. A 700 gallon cone bottom tank fitted with a heat
exchanger provides a source of controlled temperature slurry. The large volume tank
slows the process of slurry breakdown, increasing the effective life of the slurry.
Variable orifice flow control valves are used to regulate flow and pressure in both
pumps. A flow control valve placed in the suction pipe of the second test pump allows
simulation of starved suction operation. The system is designed to provide adequate
flow rates to maintain slurry solids in suspension, yet slow enough to prevent excessive
wear to piping. Test Rig specifications:
1. Accommodate Seal Sizes From 1.875 to 4.5" Diameter
2. Controlled Temperature Range of 100 to 160 F
3. Controlled Seal Chamber Pressure Range of 50-110 PSI
4. Accommodate Most Water Based Slurries
5. Vary Test Pump Suction Pressure
6. Real Time Data Acquisition System
7. Capable of Unmanned Operation
The choice of a slurry is very important, since the key goal of testing is to obtain
meaningful results in a short time. A long test duration would be required, if a soft
slurry was used for the test media (limestone or gypsum, Mohs 3-4). Table #1 lists the
Mobs scale and selected mechanical seal face materials. Softer slurry would also require
frequent replacement, since the softer particles would breakdown quickly. Flyash slurry
6A-144
-------
was chosen because it contains a large percentage of abrasive particles Mohs 7 and
harder. This material was readily available from a power plant in which
flushless slurry seals were being evaluated. This flyash was primarily composed of SO
micron and smaller particles which provide the most aggressive environment for
evaluating seal performance. Refer to Table #2 for flyash slurry description. It is
important to consider the effects of flyash on seal performance since this material is
present in most FGD slurries.
TEST RESULTS
Flyash slurry has proven to be a very effective media for evaluating flushless mechanical
slurry seal design. To date, nearly three years of testing have been conducted. Test
programs have included development of seals for large particle abrasive applications,
increasing slurry seal life and performance, seal face material evaluation, evaluation of
adaptive hardware and seal chamber design.
SEAL DESIGN.
Results of testing confirmed that even small amounts of flyash leakage would lead to
abrasive wear and eventual washout of the seal faces. A close look at failed seal faces,
revealed a pitted or sandblasted appearance. This is caused by micro-spalling or
chipping of the seal face when hard slurry particles enter the sealing gap. Slurry
particles entering the sealing gap, move across the seal faces from stuffing box to
atmospheric pressure. While traveling across the seal faces, particles slide and tumble.
They produce a high localized load when a high spot on the particle is forced between
the seal faces. At this point, one of two things happen, the slurry particle is crushed or
the seal face fractures (micro-fracture producing a very small chip). In the case of high
leakage, this chipping will occur quickly and cause seal face washout The above
mechanism is referred to as three body abrasion (6). Figure #14 contains additional
information on abrasive wear.
An improved design which reduced seal face leakage to "near zero" (no visible leakage)
was developed using analytical tools and results from early testing. This improved design
has undergone extensive testing under a wide range of operating conditions and
consistently provided excellent seal performance. Using the new design in hard abrasive
slurries eliminated seal face washout, in fact after a 1500 hour test, seal faces were in
excellent condition.
Utilizing the improved design, extensive seal face material testing was conducted.
Results confirmed that Silicon carbide vs Silicon carbide provides the best performance
in abrasive applications. In field applications Silicon carbide has given excellent
performance in FGD, mineral and ore processing, and tailings services.
6A-145
-------
ADAPTIVE HARDWARE.
To evaluate the effects of seal chamber design on circulation and heat transfer, testing
was conducted with a quartz based slurry, 25-30% solids by weight. The slurry test pump
used had no back vanes and had an enlarged seal chamber opening to adapt to various
seal chamber configurations.
In tests evaluating a seal chamber with close clearances over the seal, the seal failed
from poor circulation and packing of solids in the chamber. Using a tapered seal
chamber design, no packing of solids was found and the seal performed well.
To monitor seal chamber pressure and temperature, a vent port was drilled into the
tapered seal chamber. Over a period of 400 hours testing in sand slurry, localized
abrasive wear was pronounced in the vent port area. This same type of localized wear
was also seen in the seal chamber used during the flyash testing, Although it took several
thousand hours for the wear to occur. This confirmed the need for uniform
uninterrupted contours in the seal chamber area.
FIELD TEST RESULTS
Although much was learned from laboratory testing, the ultimate test is always success in
the field. Over the past several years slurry seal performance has improved significantly.
The knowledge gained in laboratory testing has provided new opportunities for the
flushless mechanical slurry seal. Slurry seals are now providing extended life not only in
FGD applications, but also in flyash scrubber applications, mine tailings, iron ore
processing, alumina and chemical applications.
Four slurry applications are reviewed in this paper, two typical FGD applications and
two field test applications. Each application covers a different facet of the guidelines
presented in this paper. Results of each application with operating conditions are
presented in Tables #3-6.
ACKNOWLEDGEMENTS
The authors wish to thank J. Larson-Basse of the NSF, B. Nau of the BHR Group, Ltd.,
D. Colley of Radian Corp., and P. Radcliffe of EPRI for providing information and
literature referenced in this paper. We would also like to thank W-A. Meuse and S.
Olson of PEPCO (Dickerson Station) for process information, EJ. Maupin of TMPA
(Gibbons Creek S.E.S.) for process information and Steve Hakala of USX Mintac for
process information. Special thanks to Peter Janku and Salvador Garcia, Mineral & Ore
Processing Design Engineers, BW/IP, Mike Freholm, Tony Sala, Bob Curry and Rick
Williams, Sales Engineers, BW/IP.
6A-146
-------
REFERENCES
1. W. DePriest, W J. Rymarczyk & P. Radcliffe. "Design and Operation Of Flue
Gas Desulburization Systems for Cycling Service." In Proceedings of the 1990 SO,
Control Symposium. voL 1.
2. M. Davison. The Effects of Seal Chamber Design on Seal Performance." May
1989 Pump Symposium. Houston, Tx.
3. William V. Adams. "Seal Chambers Cut Pump Failure Rate." Maintenance
Technology. August 1990, pp 32-36.
4. N. D. Barnes, R_ K. Flitney and B. S. Nan. "Considerations Affecting the Design
of Pump Chambers for Improved Mechanical Seal Reliability". Presented at
BPMA 12th International Pump Technical Conference. April 1991.
5. Frank E. Manning & William E. Key. "A Novel Test Rig for Development of
Mechanical Seals for use in Hard Abrasive Slurries." Lubrication Engineering.
April, 1991, voL 47, no. 4, pp 321-325.
6. J. Larsen-Basse and B. Premaratne. "Effects of Relative Hardness on Transitions
in Abrasive Wear Mechanisms." In Wear of Materials 1983, K. C. Ludema
(editor), ASME 1983, pp. 161-166.
6A-147
-------
FIGURE *1 CONVENTIONAL MULTIPLE SPRING PUSHER
MECHANICAL SEAL
CHARACTERISTICS1
* MULTIPLE COIL SPRINGS EXPOSED TO PRODUCT
* SPRING POCKETS EXPOSED TO PRODUCT
* DYNAMIC GASKET EXPOSED TO PRODUCT
* ROTATING SPRING DESIGN
* CARTRIDGE DESIGN
FIGURE *Z FLUSHLESS RUBBER IN SHEAR
MECHANICAL SLURRY SEAL
CHARACTERISTICS;
* NO DYNAMIC GASKETS OR SLIDING FITS
» RUBBER IN SHEAR SPRING ELEMENT
* STATIONARY SPRING DESIGN
» STATIONARY MEMBER MOUNTED TO BEARING FRAME
6A-148
-------
FIGURE *3 FLUSHLESS ENCAPSULATED CONE SPRING
MECHANICAL SLURRY SEAL
CHARACTERISTICS1
• NO DYNAMIC GASKETS OR SLIDING FITS
• RUBBER ENCAPSULATED CONE SPRING ELEMENT
• ROTATING SPRING DESIGN
» CARTRIDGE DESIGN
%
1 £T."
w®
—
3tB r
=j—
ro
2
— — ¦
FIGURE *A FLUSHLESS PROTECTED MULTIPLE SPRING
MECHANICAL SLURRY SEAL
CHARACTERISTICS:
• SPECIALLY DESIGNED DYNAMIC GASKET EXPOSED TO PRODUCT
• MULTIPLE COIL SPRINGS PROTECTED FROM PRODUCT
» STATIONARY SPRING DESIGN
• CARTRIDGE DESIGN
6A-149
-------
Typical Range1
,010"-.090" T.I.R. .J
FIGURE *5 CONCENTRICITY OF PUMP SHAFT
TO STUFFING BOX BORE
Should be Limited
to .015" T.I.R.
/T
FIGURE '7 PUMP SHAFT AXIAL TRAVEL
Should be LimitedrsSM(//^,\
to .015" T.I.R. -J
I
/f
/
FIGURE *6 PERPENDICULARITY OF PUMP SHAFT
TO STUFFING BOX MOUNTING SURFACE
Typical range :
.015"-.060" under
seal faces.
1^0
FIGURE *8 SHAFT DEFLECTION. COMBINED
HYDRAULIC AND STATIC LOAD
-------
BACK VANES REMOVED OR TRIMMED
ABOVE SEAL COVER OPENING
BEARING HOUSING
SEAL COVER
MPELLER
FIGURE * 9 SLURRY PUMP WITH FLUSHLESS SLURRY SEAL
IN A SELF VENTING TAPERED SEAL CHAMBER
BACK VANES REMOVED OR TRIMMED
ABOVE SEAL CHAMBER OPENING
BEARING HOUSING
SEAL CHAMBER
IMPELLER
FIGURE * 10 SLURRY PUMP WITH FLUSHLESS SLURRY SEAL
IN A SELF VENTING OPEN SEAL CHAMBER
6A-151
-------
PUMP CURVE
FLOW
Seal chamber pressure can be expressed by the following relationship:
(1) P(e = Ps + Pj-Pbv
and Discbarge pressure is equal to
(2) Pd = P, + Pi
hence, Seal Chamber pressure can be expressed as
(3) P« =
Where:
P.
P:
Seal Chamber pressure
Suction pressure
Developed Head
Discharge pressure
Back vane Developed pressure
Figure #11. At higher flow rates the pressure developed by the impeller drops
off quicker than the pressure developed by the backvanes. If suction pressure is
low, the seal chamber will be exposed to a vacuum. This condition may be
experienced during startup or under high flow rates with low suction pressure. It
is important to provide adequate suction pressure to prevent this mode of
operation, as it will shorten seal life.
6A-152
-------
Figure #12 The Slurry Test Rig
FIXED OlAWETER-
FLOW CONTROL
ORIFICES
VARIABLE ORl'iCE
flow cc*trol
VALVES
ATMCS3-ERC
PRESS-RE
TEST PLfcP V
BYPASS VALVE
TEST PUMP
DRAIN KNIFE
gate valve
SLLFRY TEST PUM=
tank Skjtoff
knife gate
valve
SYSTEM drain
knife Gate
valve
SLLRSY RECRGJLATON °LM=
Figure #13 Slurry Test Rig Flow Schematic
6A-153
-------
TWO-BODY ABRASION
HARD ABRASIVE / SOFT MATERIAL
PLASTIC DEFORMATION
THREE-BODY ABRASION
HARD ABRASIVE / HARD MATERIAL
MICRO FRACTURE
WEAR RATE IS
PROPORTIONAL TO
- HARDNESS
- LOAD
- ROTATIONAL
VELOCITY
<
CC.
01
<
1
1
1
SOFT ABRASIVE | >
/ HARD ABRASIVE
NON-DUCTLE /
MATERIAL / [
y i
i
—^ i
DUCTILE MATERIAL 1
'
1.2
HARDNESS OF ABRASVE / HARDNESS OF MATERIAL
Figure #14. The effect of abrasive hardness / material hardness ratio on abrasive wear
rate and wear mechanism. Abrasive wear can be caused from two-body abrasion or
three-body abrasion. Key parameters that affect wear are hardness, load and rotational
velocity (6).
6A-154
-------
TABLE 1
FLYASH SLURRY PROPERTIES AND CONCENTRATION AS TESTED
-Flyash Composition (% by Weight)
-A1203 (Mobs 9 Hardness)
27%
-Si02 (Mohs 7 Hardness)
43
-misc. soft compounds
30%
-Particle size Distribution (%)
particle size
new
400 hrs*
3-5 micron
54%
54%
5-15
32%
29%
15-25
8%
16%
25-50
5%
1%
50-larger
1%
>1%
-Slurry Solids by Weight (%)
11-15%
-Water by Weight (%)
85-89%
-Slurry Specific Gravity
12.-13
•Flyash after 400 hrs testing
6A-155
-------
TABLE 2
MOHS HARDNESS SCALE WITH SELECTED SEAL FACE MATERIALS
Mohs Hardness Scale
Knoop Scale*
10
Diamond
Boron Carbide
Silicon Carbide, Alpha Sintered
Titanium Diboride
Silicon Carbide, Direct Sintered
3000
2800
2700
2500
9
Corundum (A1203)
Silicon Nitride
Tungsten Carbide
2000
1300-2000
1500-1800
8
Topaz
7
Quartz (SiOj)
Taconite
6
Orthoclase
5
Apatite
4
Fluorite
3
Calcite
Limestone
2
Gypsum
1
Talc
•Knoop Scale included for reference
6A-156
-------
TABLE 3
FGD APPLICATION #1 - Louisville Gas & Electric Co.
-Plant Name:
-Plant Location:
-Process Description:
-Pump Location:
-Slurry Description:
-Solids:
-% Solids:
-% Flyash:
-Temperature:
-Seal Chamber Press:
-Disch. Press:
Louisville Gas & Electric Co.
Cane Run Station
Wet Lime Flue Gas Desulfurization
Absorber Recycle, Unit #4-2 pumps
Unit #5-2 pumps
Recycle Slurry
Lime/Gypsum/traces flyash
5-15%
1-2%
120-135 F
25-35 psi
65 psi
-System Operation: Absorber recycle pumps operate around the dock, with
occasional shutdown for standby. Pump is drained for standby condition
and filled and vented prior to startup.
-Pump Description:
-Slurry Seal Size & Type:
-Slurry Seal Construction:
-Total Time in Service:
Warman 550 TUL
RIS-9500
316SS metal parts
EPR Rubber in Shear Element & gaskets
Reaction Bonded SiC Sta. & Rot. Faces
1987-1991, 3 years unit #4 pumps
unit #5 pumps still in operation
-Approximate # of Hours: Unit #4 - about 18,000 operation
Unit #5 - 20,000+ & still in operation
-Seal Performance Evaluation: Slurry seal performed flawlessly for 3 years in
unit #4 with no visible leakage. Both pumps in unit #4 were removed
from service to replace worn liners. One seal was removed for inspection
at this time. A new seal was installed per standard maintenance practices.
After 18,000 hours service this seal had no measurable wear and looked as
new. The two seals in the unit #5 pumps continue to run and have not
required any maintenance for nearly 5 years.
6A-157
-------
TABLE 4
FGD APPLICATION #2 -Texas Municipal Power Agency
-Plant Name:
-Plant Location:
-Process Description:
-Pump Location:
-Slurry Description:
-Solids;
-% Solids By Weight
'% Flyash in sluny solids:
-Temperature
-Seal Chamber Pressure:
-Discharge Pressure:
Texas Municipal Power Agency (TMPA)
Gibbon Creek S.E.S.
Wet Limestone Flue Gas Desulfurization
Absorber Recycle
Limestone Reagent/Gypsum
Limestone/Gypsum/traces Flyash
5-15%
2-5%
120-135 F
20-35 psi
Vertical line, 60 ft.
-System Operation: The absorber recycle pumps operate continuously. They are
shut down for normal inspection and scrubber maintenance. The pump is
normally in operation for 9 1/2 months of the year.
-Pump Description:
-Slurry Seal Size & Type:
-Matl's. of construction:
Warman 450 STL
RIS-7500
316SS metal parts
EPR Rubber in Shear Element & gaskets
Reaction Bonded SiC Sta. & Rot. Faces
-Total Time in Service: 1989-1991, 1 years, 2 months
-Approximate # of Hours: 8,000 hours
-Seal Performance Evaluation: This was new pump with a factory installed
slurry seal. Due to a clerical error, seal was supplied with incorrect seal
faces, Tungsten vs. Silicon carbide. Shortly after startup, seal began to leal
and failed from abrasive washout of seal feces. A replacement seal with
Reaction Bonded Silicon Carbide faces and improved design was installed.
This seal has given flawless performance since installation. Recently, the
pump was removed from service for a routine warranty pump inspection.
At this time, the seal was examined and found to be in excellent condition
with no signs of wear. Prior to reassembling the pump, seal feces were
reconditioned per standard maintenance practices.
6A-158
-------
TABLE #5
FIELD TEST #1 - Potomac Electric Power Company
-Plant Name:
-Plant Location:
-Process Description:
-Pump Location:
-Slurry Description:
-Solids:
-% Solids By Weight
-pH
-Temperature
-Seal Chamber Pressure:
-Discharge Pressure:
Potomac Electric Power Company
Dickerson Station
Flyash Particulate Scrubber
Absorber Recycle
Flyash Slurry
Flyash, 26% Alp* 43% Si02
2-5%
2-4 (Chlorides level is low)
110-120 F
45-55 psi
65-75 psi
-System Operation: Two absorber recycle pumps are used on each unit, one on
line, the other on standby. During normal operations, each pump is
cycled, 8 hours on and 8 hours ofE. While on standby, the pumps remain
filled.
-Pump Description:
-Mechanical Seal Details:
-Matl's. of construction:
ASH DG-9-5
RIS-6500
316SS metal parts
EPR Rubber in Shear Element
Reaction Bonded SiC Sta. & Rot Faces
-Total Time in Service: 1989-1991, 2 years, 4 months
-Approximate # of Hours: Over 6,000 hours run time
-Seal Performance Evaluation: Initial seal installations provided inconsistent
performance. This site was used to test improved seal designs. Results of
testing confirmed that success depended on reducing leakage to zero visibl
leakage. Currently standard O-ring Retained seals with over 2 years
service continue to operate successfully. Plant has converted 4 of the 6
recycle pumps to mechanical slurry seals and plans to convert all pumps.
This plant is also successfully applying mechanical slurry seals in bottom
ash transfer pumps.
6A-159
-------
TABLE #6
FIELD TEST #2 - USX MINTAC
-Plant Name:
-Plant Location:
-Process Description:
-Pump Location:
-Slurry Description:
-Solids:
-% Solids By Weight
-Temperature
-Seal Chamber Pressure:
-Discharge Pressure:
USX Mintac/GPM
Mountain Iron, MN
Taconite Concentrate
Hydro Sizer Underflow
Taconite Slurry Concentrate
Taconite/Quartz (Mohs 63-7)
18-25%
Ambient
20-35 psi
-System Operation: This pump is operated on a continuous basis. The pump is
coupled to a variable speed drive which is used to vary pump speed and
maintain a constant concentration of 18-25% solids by weight. The plant
operates on a 16 week cycle and then shuts the process down for 1 week of
maintenance.
-Pump Description:
-Mechanical Seal Details:
-Matl's. of construction:
Denver Frame 4
RIS-4500
316SS & 416SS(Ht.Treated) metal parts
EPR Rubber in Shear Element & gaskets
Reaction Bonded SiC Sta. & Rot. Faces
-Total Time in Service:
11/90-9/91
-Approximate # of Hours: 6000 hours
-Seal Performance Evaluation: This seal gave excellent performance for nearly
a year. On 9/11/91, the seal foiled when the pump was started dry and
left to run for several minutes before unblocking the suction valve. This
installation utilizes a tapered seal chamber with a vent port (the pump has
no backvanes). A new seal has been installed in this pump and is running
at this time. It should be noted that localized abrasive wear was found in
the area of the vent port after the year in service. The customer is
planning to install seals into the three remaining pumps in this service (the
cover will not have a vent port in these pumps).
6A-160
-------
SULCIS FGD DEMONSTRATION PLANT
LIMESTONE-GYPSUM PROCESS:
PERFORMANCE, MATERIALS, WASTE WATER TREATMENT
E. Marches!
D. Principe
ENEL - Construction Department
Piacenza Central Laboratory
Via Nino Bixio, 39
29100 Piacenza, Italy
6 A-161
-------
Intentionally Blank Page
6 A-162
-------
ABSTRACT
ENEL has installed 3 FGD demonstration plants, each of 40.000 Nmc/h, at the Sul^ia power
plant, Sardinia, to experiment limestone-gypsum, VJellman-Lord, Walter processes.
ENEL chose the limestone-gypsum process for its FGD installations in tb> building stage
on a first series of new multi-fuel power units (coal, oil, natural gae) and on in-
service coal units. The other two processes are taken into consideration as
perspectives, both in local socio-economic situations of a particular nature, aad with
a view to diversifying the resultant by-products.
The first phase of limestone-gypsum process experimenting, which lasted about 3200
hours, was completed in 1990 with the following results:
• plant performance is assured (with 1 4 S coal and 3 % S fuel oil) to
be well within the legally required limit of 400 mg/Nmc of SO2 in
emission.
The validity of ENEL's plant choice for "Sulcis Project" was also
confirmed (use of Sulcis-basin coal having 7-8 % S) which envisages 2
absorption towers in series;
• The on plant study of materials and equipment, conducted through
periodical NOT inspections, along with the installation of a series of
specimens of alternative materials, supplied a wide-ranging view of
useful information; the behaviour of a wide range of metal materials,
organic coatings and equipment was defined in the different environmental
conditions typical of a wet FGD plant;
• The waste-water process treatment designed by ENEL for its commercial
power plants was verified and optimized on a 200 1/h pilot plant. The
treatment allows to meet the stringent italian laws on water effluents.
6A-163
-------
1. INTRODUCTION
The need to gradually reduce in Italy too the sulphur oxide emissions from the various
industrial sources was recognized at the end of the seventies. Early in the eighties,
the Board of ENEL decided that a demonstration scale experiment should have been set
up in order to acquire direct experience with a new and complex technology and to
investigate the problems connected with its application in the particular Italian
context. A Working Croup was organized by ENEL Construction Department who examined
about ten processes which had already reached a sufficient degree of development, taking
into account the cost, the complexity of the plant, the technological maturity and the
marketabilility of the end products.
The final proposal approved by the Board relates to experiments with the following
processes:
• limestone-gypsum process, capable of producing commercial quality
gypsum;
• Wei 1 man-Lord process, of the regenerative type with separation of
pure SO2 which can be converted to sulphur or sulphuric acid;
• Walther process, which uses ammonia to produce ammonium-sulphate usable
as a fertiliser.
The construction of the three desulphurization systems was entrusted to Italian
companies which had received licences (Idreco with a Bishoff licence for the limestone-
gypsum process, CIFA with a Davy-McKee licence for the Wellman-Lord process, and
Termokimik with a Walther-Krupp Koppers licence for the ammonia/ammonium sulphate
process), whilst Ansaldo was given the task of constructing the common works (civil
engineering and interface systems with the power station).
The project for the construction of the experimental complex was the responsibility
of the Milan office of ENEL Construction Department and the most suitable site for
constructing this plant was identified in the Sardinia island at Sulcis power station
in view of the proposed use of local coal which has a high sulphur content.
Following the Decree N° 105 of 10.3.87 of the Ministry of the Environment relating to
thermal power station emissions that took up the commitment contained din the Helsinki
protocol for the percentage reduction of SO2 emissions, bringing it forward to 1990,
ENEL made up the decision to install FGD systems for all the new multi-fuel power
stations and for coal-fired existing power stations. The technology chosen for the first
series of FGD plants to be installed (for a total of Z 8,000 MWe) was the limestone-
gypsum process.
6A-164
-------
This decision, together with the new emission limit of 400 mg/Nmc set up by the Ministry
of the Environment for individual power stations, made it all the more urgent to have
some results from the experiments on the demonstration scale, in particular with regard
to the limestone-gypsum process, but also for the other processes which are potentially
valid alternatives either in particular local situations from the economic and social
point of view, or from the point of view of diversifying the end-products.
Desulphurization plants involve the use of specific liquid effluent treatment
processes. It was therefore considered desirable to also validate the design of this
treatment by means of tests on an "ad hoc' ' pilot plant in order to guarantee a discharge
in accordance with Italian regulation.
2. OBJECTS OF EXPERIMENTATION
The object of the experimentation is to verify:
• plant performance for all the three processes with imported coal
(S about 1%), Sulcis coal (S up to 8%), fuel oil (S about 3%);
• end products characteristics (gypsum and ammonium sulphate);
• emissions characteristics;
• waste water treatment performance;
• construction materials and linings behaviour;
• emissions monitoring instrumentation operation.
It is also planned to train operating personnel for future commercial FGD plants.
The entire experimentation is divided into four steps:
1. limestone-gypsum tests 1st phase
2. Wellman-Lord tests
3. Walther tests
4. limestone-gypsum tests 2n^ phase.
So far only step 1 has been accomplished and it is foreseen to end up with the other
three by the end of 1992.
3. DESCRIPTION OF EXPERIMENTAL PLANT
3.1 Desulphurization systems
3.1.1 Limestone-gypsum system. The schematics of the plant for the limestone-gypsum
system is illustrated in Fig 1.
The plant contains a bypassable prescrubber, which is useful to obtain a high purity
gypsum, particularly in case of high levels of impurities (especially hydrochloric and
hydrofluoric acid and ash) in the flue gases.
6A-165
-------
Absorption and oxidation take place in a counter-flow tower (scrubber). The Injection
of sorbent is provided by 9 spray nozzles on 9 different levels connected three by three
to a recirculation pump. In the lower part (oxidation zone), the gypsum suspension is
extracted and is sent to the filtration system.
The limestone is supplied in coarse particle size and is stored in this form; it is
then grounded and suspended in water.
The make-up water maybe industrial water or sea water; the latter possibility is of
great advantage for the installation of desulphurization plants at coastal sites where
there is little fresh water available.
3.1.2 Wellman-Lord system. In the Well man-Lord system, illustrated in Fig 2, a high-
efficiency prescrubber is used in which hydrochloric and hydrofluoric acid and ash,
which would create problems in the absorption circuit, are removed.
The absorption is carried out in a tower with trays and packs, in which a solution of
sodium salts sprayed in counterflow to the flue gas absorbs the SO2. The exhausted
solution is regenerated thermally in an evaporator and is then re-used. The circuit
is only theoretically closed, since purging takes place to limit the accumulation of
non-regenerable by-products such as sodium sulphate; the make-up consists of sodium
carbonate supplied in powder form and stored in solution.
The SO2 released by the evaporator in commercial plants is converted to sulphur or H2SO4;
in the Sulcis plant, the conversion system was not constructed. The SO2 is therefore
sent, together with the desulphurized gas, to the gas duct of the thermal unit.
The plant is equipped with tanks for storing the fresh and exhausted absorbent solutions
so as to make independent the absorption and regeneration systems , in order to be able
to operate the former system for about 24 hours with the second system out of operation.
3.1.3 Walther system. The Walther system, illustrated in Fig. 3, uses ammonia in an
aqueous solution as an absorber of SO2, producing a dilute ammonium sulphite solution
which is then oxydized to sulphate with air.
The plant contains a prescrubber for the partial removal of aerosols.
The absorption occurs in two spray towers in series, followed by two types of filters
for the removal of solid and liquid particles. The first type is a coalescent filter,
whilst the second is a wet E.P.
The ammonium hydrate is supplied in liquid form and is stored in tanks. The ammonium
sulphate solution is treated in a production unit consisting of a crystallizer and
a granulator. The plant is fitted with tanks for accumulating the ammonium sulphate
solution in order to make the operation of the absorption and oxidation system
independent of that of the production unit.
3.1.4 Design Criteria. The design criteria for the entire demonstration plant are as
follows:
• gas flow rate 40,000 Nmc/h (10
MWe)
• S02 conc. in gas 0.22% (equivalent to the
combustion of coal with 3.5% S)
6A-166
-------
• reagent storage
(CaC03, NH40H,Na2C03)
(sulphur)
1 month
2 months
• Intermediate storage of products in silo
(gypsum and ammonium sulphate) 3 days
• storage of products in bags
(under cover) 3 months-1 year
The demonstration plant can operate, even though on reduced load, when supplied wi
gases having characteristics deriving from the combustion of Sulcis coal (7-8% S
The deei9n gas flow rate can be taken from Sulcis power station units 1 or 2 and
distributed to the plants as shown in the schematics of Fig. 4.
It is possible to supply two plants simultaneously, with the total gas flow rate lio
of 40,000 Nmc/h remaining the same.
The gases can receive additions of SO2 or BC1 from suitable systems in order to simult
the desired chloride and sulphur content.
Before entering the plaint, the raw gases pass through a Ljungstroem type regenerate
heat exchanger (GAVO) in which the desulphurized gases are heated in counterflow
The gas is circulated through the plant by means of two fans (one in reserve) situat
between the desulphurization systems and the GAVO, so that inside the latter t
desulphurized gas has a higher pressure
than that of the raw gas, thus avoiding the ingress of the latter into the desulphuri:
gas which would result in a reduction in the desulphurization efficiency. The ammoni
sulphate and gypsum produced are placed in silos and then packed in bags througi
bagging system.
3.1.5 Operating auxiliaries. The auxiliary fluids required for the operation of t
plant are:
• steam obtained from the auxiliary header on Sulcis units 1-2
• demineralized, potable, industrial water taken from the power station
main headers
• compressed air produced from am independent system
• sea water taken in by means of dedicated pumps installed in the intake
works of the Sulcis power station
The electrical supply is provided by a main 6 kv switchboard which supplies the 1
blowers directly and, via two transformers (6 kV/380 V), two power switchboards. <
switchboard supplies the control panels of the Walther and the Hellman-Liord plani
the common plant and the flue gas additive system; the second switchboard is entir<
dedicated to the electricity requirements of the limestone-gypsum plant.
The control and instrumentation equipment is located in the control room inside 1
General Services Building of the demonstration plant with the exception of the amnion:
sulphate production unit which has its own control room located near the product:
unit itself.
In addition to being indicated and recorded on control and monitoring panels in 1
control room, the process data flow to a data acquisition and processing sysi
installed inside the General Services Building for data storage, real time calculat:
of performance and material balances.
6A-167
-------
A chemical laboratory has been constructed inside the General Services Building in order
to carry out the analyses relating to the process control.
3.2 Waste water treatment
The fluid waste to be treated consists of gas wash water, coming from limestone-gypsum
and Wellman-Lord processes.
Fig. 5 shows the schematics of the three stage process.
The design flow rate of the pilot plant is 200 1/h.
The first stage of treatment involves an initial addition of lime to neutralise the
acidity, followed by a second addition to increase the pH to 8.7:9, which allows the
precipitation of a large proportion of metals as hydroxides. It is also planned to add
sulphide in order to remove Hg and Cd, FeCl2 to precipitate the excess sulphide, and
polyelectrolyte which has a flocculating effect. This stage includes a circular
clarifier.
The 2nd stage is designed to remove the residual Se by coprecipitation with ferric
hydroxide to pH 6:7 and subsequent sedimentation by means of lamellar packing. Dosing
with ferric chloride and acid and/or soda is provided to regulate the pH.
Finally, the purpose of the 3rd stage is to oxidise, by the use of oxygenated water,
the residual sulphites and sulphides, the nitrites and, more generally, all oxidizable
substances (COD).
4. RESULTS OF THE EXPERIMENTS ON THE LIMESTONE—GYPSDM PROCESS (1st phase)
4.1 Process
The tests performed up to now were carried out in the period January-March 1990 in order
to verify the design assumptions adopted by ENEL during the procurement specification
phase for the flue gas desulphurisation systems for 660 and 320 MW multifuel units and
for the 240 MW units at the Sulcis power station. In this station, where local high
sulphur content coal (7:8% by weight) will be used, two absorption towers in series
are foreseen.
In particular, a determination was made of the overall desulphurisation efficiency and
of the SO2 concentration in the emissions when the liquid/gas ratio of the absorber
was varied over a range of S02 concentration at the plant inlet from 2,000 to 16,000
mg/Nmc (equivalent to a coal sulphur content between 1% and 8%).
In all, 20 tests were carried out with various plant configurations.
The emission values have been measured downstream of the regenerative heat exchanger.
With reference to the standard design of desulphurization plants for 320 and 660 MW
units the global plant performance is confirmed. The full compliance with the SO2
regulatory limit of 400 mg/Nmc in emission with am inlet concentration range varying
from 2,000 to 4,000 mg/Nmc and a L/C ratio equal to 15 and 20 respectively, has been
ascertained.
The max. inlet SO2 concentration compatible with the 400 mg/Nmc emission limit at a
L/G ratio « 20 is about 7,000 mg/Nmc (3.5% sulphur in coal).
6A-168
-------
Also when simulating the Sulcis coal, the hypotheses concerning the performance of the
first absorption tower of the Sulcis FGD plants (expected outlet about 4,500 mg/Nmc,
with 16,000 mg/Nmc at the inlet and a L/C ratio * 14) are confirmed. As a consequence
and according to the above data, the second absorption tower will reduce the S02 below
the regulatory limits (tab 1).
The above considerations are better visualized in figs. 6a and 6b. The first one shows
the S02 concentration in the emissions as a function of the L/C ratio for two significant
values of S02 concentration at the inlet (2,000 mg/Nmc and 4,000 mg/Nmc, corresponding
to 1% S imported coal concentration and to 3% S fuel oil concentration, respectively).
It can be seen that, in order not to exceed the 400 mg/Nmc legal limit at the outlet,
the "critical'* L/C ratios are around 8 and between 13 and 14 with a S02 inlet
concentration of 2,000 mg/Nmc and 4,000 mg/Nmc respectively.
Fig. 6b is in a certain way the reverse of fig. 6a, since it shows in the ordinates
again the SO2 concentration in the emissions but as a function of the inlet SO2
concentration for three L/G ratios (10,15,20). From this figure it is particularly
interesting to note that with a L/G ratio equal to 20 one must go to an inlet S02
concentration as high as 6,000 mg/Nmc in order to exceed the 400 mg/Nmc legal limit;
but also that even with a relatively low L/G ratio of 10 it is still possible to operate
the plant with an inlet S02 concentration of around 3,000 mg/Nmc.
The characteristics of the gypsum produced during the tests comply with the
specifications (tab 2) and, according to its composition, for the Italian regulation
can be classified as a non hazardous waste and therefore can be utilized in industrial
and civil activities.
To be more precise, the soluble chlorides concentration exceeds the specification limit
but that was very well expected due to the high level of chrorides in the water used
to wash gypsum (from 700 to 1,000 mg/1): this type of water was actually the only one
available at the site for the experimentation.
High quality limestone has been used for the tests (fig 7).
The particulate emissions were about 10 mgNmc (legal limit: 50 mg/Nmc) with about 20
mg/Nmc at the plant inlet.
The chloride emissions were about 1 mg/Nmc with 50 mg/Nmc at the inlet (legal limit:
100 mg/Nmc).
The fluorides emissions were 1 mg/Nmc with about 4 mg/Nmc at the inlet (legal limit:
5 mg/Nmc).
Measurements of SO3 were taken at GAVO inlet, at the prescrubber inlet and outlet as
well as in reheated desulphurized gases (GAVO outlet).
The first indications show a total SO3 reduction in the plant by at least 50%.
4.2 Waste Water Treatment
The tests carried out concerned the treatment of the prescrubber blow-down using coal
as a fuel.
The results of the tests were positive since the legal limits are already met in the
1st stage of treatment (fig. 8).
It is worth while mentioning that the Italian regulation sets a double limitation: the
first one on the concentration of individual microelements; the second one requires
that the sum of the ratios of the actual concentration of each microelement to its limit
legal concentration be less than one.
Going back to the obtained results, the first stage of treatment is able to drastically
reduce the concentration even for the most difficult elements to separate (such as Cd
and Hg) ; for the Se too the efficiency (60:80%) is still sufficient to permit to remain
within the legal limit at the second stage exit.
The final oxidative treatment has not been defined yet, but it seems to give no problems.
In order to improve the fluorides removal, lab tests are being run.
The process sludges are easy dewatered due to their high content in gypsum; they belong
to "non hazardous*' waste class.
6A-169
-------
4.3 Plant: materials and test specimens
In the design phase, materials were chosen on the baBis of technical solutions already
adopted abroad for such type of plant, also taking economic aspects into account. Due
to the experimental nature of the plant, different alternative materials and linings
have been installed and tested.
The evaluation has then been based on the following two criteria:
* Checking of the behaviour of plant materials, linings and components;
• Checking of the behaviour of test specimens installed in positions of
particular interest and on specially prepared secondary loops.
As for specimens, lined and unlined metallic materials have been tested (tab. 3). For
the unlined specimens, the following metallic classes have been considered: carbon
steel, stainless steel type 18-8, high Cr-Mo stainless steels, nickel alloys, titanium.
As for the lined specimens, carbon steel with glass flake vinylesters, fluoroelasto-
mers, GFRP (composite materials) have been employed.
The location of the specimens in the plant (prescrubber, scrubber, GAVO) are shown in
fig. 9.
Visual, telecamera and photographic documentation was acquired during the baBe line
inspection before the start-up of the plant.
Periodical inspections were made afterwards.
After about 3000 hours of operation the main observations obtained are as follows:
• the rubber applied to the prescrubber and scrubber towers gave
satisfactory esults with the exception of some little damage on pipes
edges which project inside the two towers. The rubber applied to the
pipes proved not able to stand high turbolence, such as that occurring
near the throttled valves;
• the glass flake vynilester in the flue gas ducts and dampers turned out
to behave satisfactoryly in the cold raw flue gas section. On the
contrary, in the cold desulphurized flue gas duct the upper layer of
this liner showed exfoliation, whereas in the hot desulphurized flue
gas duct it showed cracks and poor resistance to the environment;
• a fluoroelastomer in the cold raw flue gas duct showed small blistering
phenomena ;
• as for metals, the following were observed: active pit corrosion signs
on the pipes in superaustenitic alloy used for air distribution in the
scrubber; slight pitting corrosion on the fans in austenitic steel;
pitting corrosion on the dampers in 316 L steel placed on hot and
cold desulphurized flue gas ducts;
• generally no serious problems were observed on the pumps, specially for
those in continous operation; some pitting and crevice corrosion was
viceversa observed in those pumps which were out of operation for long
periods;
• the GAVO showed the detachment of plastic elements from the cold layers.
The enamelled plates of the hot layer were still in good conditions
and only a few points of rust were present.
6A-170
-------
As far as the specimens installed on the plant are concerned, it is worth while to
underline that they enabled to identify the problems connected with the typical
environments of a desulphurization plant and to evaluate the behaviour of a large number
of metals and linings.
The prescrubber environment is extremely corrosive for all metallic materials: even
titanium suffered very severe corrosion, while nickel alloys gave better results. The
linings gave satisfactory results.
In the scrubber, mainly pit and crevice corrosion were found: the classes of metallic
materials that gave good results include some nickel alloys, while the behaviour of
some superaustenitic and austeno-ferritic steels with a high molybdenum content were
not always been satisfactory also related to the long outage periods of the plant.
The glass flake vynilesters showed erosion problems only near the spray nozzles.
Fluoroelastomers and GFRP (composite materials) were still in good conditions.
Pit and crevice phenomena were also found both in the hot and cold raw flue gas ducts
and in the cold desulphurized ducts: the less resistant materials is the 316L stainless
steel.
Particularly interesting is the environment of the hot desulphurized flue gas duct:
it caused mainly generalized corrosion and it turned out to be more aggressive than
expected even to high-quality alloys due to the relatively high temperature (in any
case below the dew point).
The corrosion rates of the different metallic materials classes fall within a rather
limited range. However, the best materials are nickel alloys, titanium and superau-
stenitic alloys as shown in table 4.
The test specimens of organic liners installed in the ducts gave results in accordance
with those observed for the liners applied on the ducts except for the glass flake
vynilester in contact with the cold desulphurized flue gas: the specimens gave good
results showing no surface exfoliation.
5. CONCLUSIONS AND FUTURE DEVELOPMENTS
Preliminary conclusions can be drawn on the basis of limestone-gypsum process start-
up, operation and first phase experimentation.
The evaluation of the performance of the process has made it poseible to verify the
correctness of the design assumptions adopted by ENEL when drawing up the specifications
for the procurement of the flue gas desulphurisation systems for multifuel 660 and 320
MW units and for the 240 MW units of the Sulcis power station, in which it is intended
to burn local coal with a high sulphur content (7:8% by weight).
Particularly, it is confirmed that the guaranteed 400 tng/Nmc S02 outlet concentration
for multifuel plants (burning 1% S imported coal and 3% S fuel oil) is met with a wide
margin. As far as Sulcis power station FGD system design is concerned, it is also
confirmed that the design choice of two absorbing towers in series is a valid one.
The characteristics of the gypsum produced during the test comply with the specification
required for its use in industrial and civil activities.
The results of the test of the waste water treatment system were also positive.
The test considered only the treatment of the prescrubber blowdown, with coal as a fuel,
which represents the most critical fluid for the presence of major concentrations of
metallic contaminants.
The microelement concentrations lie within the legal limits even in the 1st stage of
the waste water treatment.
The behaviour of a wide range of metal materials, organic coatings and equipment was
defined in the different environmental conditions typical of a wet FGD plant.
Metallic materials, even the most resistent ones, showed signs of active pitting
corrosion and their choice must be carefully made.
6A-171
-------
Rubber linings gave satisfactory results but it is necessary in the detailed design
phase to pay special attention to the suspension transportation pipes in order to
prevent the occurrence of the contrary problems of erosion and fouling at local points,
due to high or low velocities.
Flake glass vinilesters and fluoroelastomers gave different results depending on the
ambient conditions and the geometry of the surfaces to be protected.
As feu: as future developments are concerned, activities similar to those accomplished
in the limestone-gypsum first phase experimentation will be run on both Wellman-Lord
and Walther processes. Actually the first ammonium sulphate production has been
accomplished and presently quality analysis are being done.
Besides, it is foreseen to run a second phase for the limestone-gypsum process in which
the following aspects will be more deeply examined:
• material balances (water and solids);
• validation and determination of the parameters which affect the
sulphate and sulphite saturation index;
• evaluation of the hydrocyclone separation system as a function of the
particle size of the separated solid phases;
• characterisation of the gypsum with variations in the residence times,
the solids content in the recirculation suspension and the pH;
• evaluation by means of dedicated measurement campaigns of the
performance of the prescrubber with regard to the removal of dust, Cl~
and F-
• chemico-physical characterisation of the emitted dust;
• evaluation of the instrumentation in the field for the measurement of
emissions;
• construction materials and linings behaviour, particularly in the hot
desulphurized duct were a section lined with three nickel alloys and
titanium (through wall papering technique), boro-silicate bricks and
fluoroelastomers have been installed.
ACKNOWLEDGMENTS
The authors wish to thank Mrs I. Binecchio for her very valuable contribution to the
content of the present paper.
6A-172
-------
OR
T.TWILllim
»CX
inc
TO
STACK
tTlUE
tCSUBBER
TO
IE*
TO
VACUUM
TAPE m.T1
Figure 1. Schematics of limestone-gypsum plant
TO
STACK
o
SO- TO STACK
TO
OXXDATXOB
Figure 2. Schema-tics of Wellman-Lord plant
6A-173
-------
TO
COAZJESCEBCC
tTIUS
xr
BH
TO
ukzt:
OUBATXOZC
(KR.>, SO.
Figure 3. Schematics of Walther plant
HtLLXAX-lO»
TO
KBIT X
TO
jtr<
BisHorr
ADPfxiOH
Figure 4. Schematics of desulphurization demonstration plant
6A-174
-------
SCRUBBER PRESCRUB8ER
WASTE WATER WASTE WATER
I
IRON(ll) CLORDE
SULFIDE
LIME
I>1
tnt Tnf
1>1
NEUTRALIZATION
ALKAUNIZATtON
CCASUUTION
DRAIN.
FILTRATION
PCLYELECIROUTE
IRON [110
ACID OR SODA CLDRIDE
I SEDIMENTATION
TL
JF
V
4-i
uH
HYDROGEN
PEROXIDE
lf°SEDIMENTATION
l>1
COPULATION
NEUTRALIZATION
w
' L_ni
uh
TO DECHMJGE
SLUDGE
.TO
DISCHARGE
Figure 5. Schematics of waste water treatment pilot plant
6A-175
-------
mc'Hmc Sq2 enlsslons
KO->inls« concMrtpatli
4000 mc/Hhc 2000 M£/Hnc
eoo
REGULATORYVLIMIT
IS
Figure 6a. SO2 emissions as a function of L/G ratio for
two inlet SOj concentrations
h{/Hhc S02 >Hissions
20
15
aoo
RECULATORY
400
1
2
O
sa2 inlet concwtpations
Figure 6b. S02 emissions as a function of the inlet S02
concentration
6A-176
-------
LIMESTONE COMPOSITION
MACROCOMPONENTS
Date
CaC03
KgO
Inert:
S102
Moisture
<%>
<%)
<%>
(%)
(%)
25/01/90
98.9
<0.05
0.44
0.20
0.17
29/01/90
98.9
<0.05
0.50
0.21
0.21
05/02/90
99.1
<0.05
0.30
0.15
0.43
13/02/90
99.3
<0.05
0.40
0.85
0.26
19/02/90
99.1
<0.05
0.61
0.16
0.16
26/02/90
99.1
<0.05
-
0.16
0.16
13/03/90
98.9
<0.05
0.32
0.17
0.28
19/03/90
99.0
<0.05
-
0.14
0.15
28/03/90
98.9
<0.05
0.45
0.18
0.16
MICROCOMPONENTS
rameter
Units
Value
As
ug/g
<1.0
se
ug/g
0.3
Cu
ug/g
2.6
Pb
ug/g
0.9
Cr
ug/g
2.0
Ni
ug/g
8.0
T1
ug/g
<1.0
Te
ug/g
<1.0
Cd
ug/g
0.1
Be
ug/g
<0.1
Hg
ug/g
<0.2
Sb
ug/g
<0.5
Mn
ug/g
13.0
Zn
ug/g
97.0
UMESTOHE REACTIVITY PARTICLE SIZE PXSTRZBUIXOH
Inactivity CO F«romta(t by voluwt CO
1QO
ao
00
OO
90
40
40
20
' I ¦ ' ¦ I1
20 00
6
0 « 20 90
00 70 80 00
o
10
F ARTICLE SIZE
TIME (ninutvs)
Figure 7. Limestone characteristics
6A-177
-------
MMXHI
0-
1
3-f
I
1
Zn
Ni
' •*»£"""
Cu
thi
Cr
BM1III1
0.6
0.2
0.003
1.6
0.5
0.02
3.22
2
0.001
0.58
0.1
0.05
0.7
2
0.05
2.6
2
0.01
u^/Ke
600
400 4 '.-
200
=^H
0 -
c—1—
As
'—1
ca
e—1
'—1
Se
III11IIH
90
500
10
17
20
2
90
5
0.1
80
30
35
I in Lin
RECULATORY LDOT
OUTLET
Figure 8a. Waste water treatment 1st stage reduction efficiencies
ufXt
600
400 -t'\.
200
0
w=:=fil
Cd
68
20
3
Hg
200
5
0.2
Se
n* ST AD 10
I INLET
REGULATORY LDOT
OUTLET
Figure 8b. Waste water treatment reduction efficiency for speci-
fic microelements (inlet concentration increased by dosing)
6A-178
-------
L 2
L 1
¦©"
asms
muiM
EHI
EKCHAHGHt
•@1
PI
@
©
0
SAMPLES
posmoH
Q™
I CIRCUITS
"©7""\
@
/
\
hydrqciclqhe|
TAPE
FILTER
SHLLNjU!
TASK
id
CUFSUM
Figure 9. Schematics of the materials position in the
demonstration plant
6A-179
-------
Table 1
DESULPHURIZATION EFFICIENCY
(comparison between design data and experimental results)
Design data for 320 and
660 MWe power stations
1% S coal
L/G = 15
S02 emissions: 400 mg/Nmc
3% S fuel oil
L/G = 19
S02 emissions: 400 mg/Nmc
8% S Sulcis coal
L/G = 14
1" tower SO, outlet: 4500 mg/Nmc
L/G = 19
2nd tower SOz outlet: 400 mg/Nmc
Experimental results
SO, inlet c.a 2000 mg/Nmc
L/G = 15
S02 emission: <400 mg/Nmc
SO, inlet c.a 4000 mg/Nmc
L/G = 19
S02 emission: <400 mg/Nmc
SO, inlet c.a 16000 mg/Nmc
L/G = 15
1" tower SO, outlet: c.a 5000 mg/Nm
L/G = 19
2nd tower S02 outlet: <400 mg/Nmc
Table 2
GYPSUM CHARACTERISTICS
Unit of
Parameter mesaurement
CaSO. x 2KLO %
CaSO, x 1/2H,0 %
daCO, %
a soi. %
Mg Sol. %
Na Sol. %
Moisture %
Impurity %
pH pH Unit
Si %
A1 %
Ti %
P %
Ba %
K %
Mn %
V %
Zn %
Fe %
Ni %
Typical
value
Technical
specification value
98.9
95
min
0.03
0.25
max
0.84
1.5
max
0.005
0.01
max
0.002
0.01
max
0.01
0.006
max
8.3
10
max
0.23
7.4
0.05
0.03
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
0.07
<0.01
6A-180
-------
Table 3
SPECIMENS INSTALLED ON THE PLANT - MATERIALS USED
Materials
Cr
<%>
Ni
<%)
Fe
67
Ma
<»)
MO
(%>
Ti
(%)
c
<%)
Si
(%)
M
<%)
P
<%>
Aisi 316 L
17.45
11.45
1.08
2.2
_
.026
.53
_
.025
904 L Avesta
19.9
24.8
48.85
1.43
4.3
-
.019
.46
-
.02
31803 Austeno ferr.
22.01
5.77
66.82
1.66
2.9
-
.02
.59
.15
.025
312S4 6Mo AusteniL
19.84
17.5
55
.8
6
-
.013
.4
.19
.016
Hastelloy C 276
16
61.9
5
1
16
-
.02
.01
-
-
Hastelloy C22 (2C224)
22
54
5
5
13
-
.015
.8
-
-
Hastelloy H9M (3H1)
22
47
19
1
9
-
.03
1
-
-
Cronifer 1925 LCN
20.55
24.85
48
1.29
4.7
-
.011
.3
.18
.018
Cionifer 1925 hMo
20.75
25.1
46.57
.82
6.2
-
.004
.33
.21
.019
Nicrofer 4823 hMo
23.0
47
19
.52
6.9
-
.008
.1
-
.017
Nicrofer 6020 hMo
22
63.45
1.76
.06
8.78
.19
.013
.08
-
.003
Nicrofer 5716 hMoW
15.30
59.1
5.69
.25
15.7
-
.005
.04
-
.009
Nicrofer 5621 hMoW
21.45
57.1
3.89
.16
13.7
-
.008
.07
-
.002
Nicrofer 6616 hMo
16.0
67.28
.23
.05
15.9
.28
.005
.03
-
.004
Nicrofer 5923 hMo
24.0
57.3
1.5
.5
16.5
-
.010
.1
-
.015
Uranus 52
24.82
6.37
61.43
1.02
3
-
.018
.45
-
.018
Titanium grade 7
-
-
.3
-
-
98.8
.10
-
.03
-
Titanium grade 2
-
-
.3
-
-
99
.10
-
.03
-
Flakeline 282
Lining
- Flack glass Vinilester
Fuji Flake
Lining
- Flack glass Vinilester
Keiaflake 6H
Lining
- Flack glass Vinilester
Keraflake 6R
Lining
- Flack glass Vinilester
Table 4
CORROSION RATE IN HOT DESULPHURIZED GAS OUTLET REFERRED TO THE FIRST 500 AND THE
FOLLOWING 2500 OPERATING HOURS
500 h 2500 h
Material mm/year mm/year
Carbon steell 0.241 0.459
Aisi 316 L 0.022 0.136
904 L Avesta 0.009 0.049
31803 (austeno ferritic) 0.068 0.161
Hastdloy C 276 0.054 0.073
Hastelloy C 22 - 0.056
Titanium grade 2 — 0.050
6A-181
-------
Session 6B
CLEAN COAL DEMONSTRATIONS
RECOVERY SCRUBBER - CEMENT APPLICATION OPERATING RESULTS
Garrett L. Morrison
Passamaguoddy Technology, L.P.
P.O. Box 350, Route 1
Thomaston, ME 04861
ABSTRACT
The first full scale installation of the Recovery Scrubber, a cost
effective flue gas scrubbing process and a DOE ICCT Program project,
began operation at the Dragon Cement plant in Thomaston, Maine on
December 20, 1990. Waste cement kiln dust containing limestone,
alkali, and calcium sulfate was utilized as flue gas scrubbing reagen
and high efficiency sulfur dioxide removal was achieved. Processed
waste cement kiln dust was chemically altered by the process to make
it totally acceptable as raw material feed for the cement kiln,
allowing use of the waste and elimination of the need for landfill
disposal. Chemical modification of the waste included conversion of
gypsum to limestone, carbonation of CaO to CaCOj, and dissolution of
alkali salts. By-product potassium sulfate was recovered from solutio
by use of waste exhaust gas heat energy for evaporation and
concentration of dissolved salts to form crystalline solids as high
valued, marketable by-product. System description, operating
experience, flue gas scrubbing data, and input/output material
analyses, and other potential applications are discussed.
6B-1
-------
INTRODUCTION
The system, demonstrated as part of the U.S. Department of Energy
Innovative Clean Coal Technology Program, at the Dragon Products
Company Inc. cement plant in Thomaston, Maine has been described here
in previous meetings. A brief overview of the process will provide
introduction to the technology. Emphasis in this discussion will be
on operating experience and results achieved.
This application of the Recovery Scrubber" addresses flue gas and
solid waste pollution problems at New England's only Portland Cement
producing plant (not a concrete or "Ready Mix" batch plant). The
process, through use of fly ash, biomass ash, cement kiln dust, and
other alkali rich materials is applicable to utility boilers, pulp and
paper mills, waste incinerators, waste to energy plants, and a variety
of industrial boilers and furnaces.
PROCESS DESCRIPTION
The Recovery Scrubber uses alkaline potassium and/or sodium containing
wastes as reagent for sulfur dioxide removal from flue gas. These
wastes allow production of marketable by-product and generate a
tipping fee by their use. There have been questions raised whether
sufficient alkali waste is generally available for widespread use of
this process. Drawing on the mass of fly ash, biomass ash, and waste
cement kiln dust produced annually in the U.S. it is estimated that
in excess of 75,000 MW of installed generating capacity, or its
equivalent in industrial boiler output, can be efficiently scrubbed.
Sulfur dioxide is hydrated, oxidized^_ arid_remov
-------
carbonate.
Figure 1 shows the essential process flows. The flows of: flue gas;
solids in slurry; heat; and crystalline precipitate are discussed
separately as follows.
Flnp Ras
Flue gas flows through the RECOOPERATOR (A) where heat is extracted,
to the REACTION TANK (b) where it is scrubbed, and exits the stack.
Solids in Slurry
Waste cement kiln dust is added to the STORAGE TANK (D) through vortex
mixer (C). Slurry is recirculated through the REACTION TANK (B) where
it reacts with flue gas. Reacted material is then settled from
solution in the FIRST SETTLING TANK (E), rinsed with distilled water,
settled from solution again in the SECOND SETTLING TANK (F), and
delivered to the cement plant raw material preparation system.
Heat
Heat is recovered from the flue gas stream in the RECOOPERATOR (A),
conveyed to HEAT EXCHANGER (G) and to CRYSTALLIZER (H) where it is the
energy source for evaporation of water from the potassium sulfate
solution.
Crystalline Precipitate
The supernatant liquid (potassium sulfate solution) from the FIRST
SETTLING TANK (E) is conveyed to the CRYSTALLIZER (H) and evaporated
using heat recovered from the flue gas. Crystalline potassium sulfate
is recovered by centrifugation.
INITIAL OPERATION
Weather
Operation began in December 1990, a time which should be avoided for
start-up of anything in Maine. All fluid piping, whether for process
flows, cooling water, or seal water in slurry pumps, must be heat
traced or otherwise protected from freezing. All inadequately heated
pumps and pipes were quickly identified.
Seal water
Construction continued during start-up operation in order to complete
detail work. The work force became a valuable asset for correcting
problems that were quickly apparent. The first malfunction was loss of
seal water flow to a slurry pump. Because of the extremely abrasive
nature of the slurry produced from cement kiln dust, a momentary
interruption of seal water flow caused immediate failure of the shaft
seal and shut down of the pump. Two seals were lost before constancy
of seal water pressure was established.
6B-3
-------
Motor Bearings
A forced draft fan moving flue gas from the cement plant through the
scrubbing system is the largest power consumer in the process. During
the second week of operation a motor bearing failed causing failure
also in one fan bearing. The project owner had instructed the
contractor to provide the lowest cost (used) motor available. The
unit supplied did not meet design specifications but was installed by
the contractor as a cost saving measure. The ultimate cost of the
failure, in equipment, lost time, and manpower could have covered the
cost of a new motor. Replacement required four weeks.
General Piping and Pumping
Other minor problems were encountered in piping and pumping equipment
and solved with little difficulty. Operation continued and
adjustments to the process flows were made to optimize conditions. In
June 1991 corrosion of a heat exchanger shell required replacement of
the shell material with a more resistant alloy. Steel pipes leading
to and from the shell were changed to high density polyethylene at
the same time.
CQplj.ng w?ter
Cooling water is required to condense water vapor generated in the
evaporator, thereby maintaining vacuum. During June and July it
became apparent that the cooling pond was not adequate. It was
replaced with a gravity flow spray cooling system which has performed
well since that time.
Slurry Preparation
Although the process converts gypsum to calcite (limestone), the
conversion does not take place until the waste, which contains
gypsum, or more accurately calcium sulfate, comes in contact with the
flue gas.
When the waste cement kiln dust is first mixed with water the
hydration of calcium sulfate already present in the waste forms
gypsum crystals. Material precipitates onto pipe interior surfaces,
gradually reducing flow until the pipe must be cleaned to allow
continued operation.
The initial slurry mixing and transport system consisted of a mix
tank where dust and watci were introduced and agitation was provided.
Slurry was then pumped to the reaction tank for use in scrubbing.
Build-up of gypsum in the pipes necessitated a change.
The new system is called a vortex mixer. A cylindrical tank with a
conical bottom is used. Slurry, which has been previously reacted,
and therefore has no calcium sulfate left to be precipitated, is
pumped into the top of the tank tangential to the tank walls. It
flows around the tank circumference and spirals down toward the
conical bottom, accelerating as it flows out. The reacted slurry
serves to coat the tank with a layer of fluid that effectively
prevents fresh dust, which is added at the center of the tank top,
from coming in contact with the walls of the mixer. Mixing is
thorough because of the high flow and turbulence in the tank. The
6B-4
-------
mixture exits the vortex mixer into the reaction tank where it joins
a 9,000 GPM flow passing into a 72,000 gallon reservoir of reacted
slurry. There have been no slurry handling problems since the change
to this system.
Gas Pistribution
A major problem in gas handling has existed throughout much of the
operating period. Operation of the reaction tank, which is the system
providing contact between gas and scrubbing fluid, depends on there
being even distribution of gas throughout the plenum under the
bubbling tray reactor. Any zones or pockets of low gas pressure under
the tray, caused by inadequate gas distribution, result in downward
flow of scrubbing liquid through the tray and eventual plugging of
the tray holes. This causes a rise in operating pressure and,
therefore, operating cost and is not acceptable for long term use.
Specifications in the process design called for appropriate duct and
plenum design to assure gas distribution to within a set tolerance,
measured in inches of water pressure, at any point on the under side
of the tray. It is presumed that the initial design, which was
provided by the design engineers engaged for the overall design work,
would have achieved that distribution. Unfortunately, we will never
know. The initial design was changed, as a cost saving measure, by
the construction contractor. The change was not requested or approved
by, or reported to either the owner, who acted as project manager, or
to Passamaquoddy Technology, the process technology provider. As a
consequence, the components of the 24 foot by 48 foot reaction tank
were shop fabricated, shop coated with $15 per square foot corrosion
protection lining, and delivered to the construction site for
erection before the changes were discovered. It was decided at that
time to proceed with erection and to correct the inadequacies by
retrofit changes within the ducting and plenum, rather than undergo
the high cost to redesign, fabricate, and coat all new components.
Operation of the system has continued through testing and measurement
of the gas flow characteristics and its interaction with the slurry
flow on top of the tray. Plugging of the tray has been a continuing
problem, causing repeated stoppages for cleaning. Continued operation
has provided the necessary data for design of the retrofit fix, and
has also allowed assessment of the performance of the rest of the
scrubbing system, that is, scrubbing efficiency, adequacy of waste
kiln dust renovation, heat recovery and evaporator function, and
by-product potassium sulfate quality.
Gas flow distribution was corrected by addition of turning vanes
(part of the original design) in the duct leading to the plenum.
These turning vanes have distributed the gas over the entire six foot
height of the entry duct and reduced inlet velocity by a factor of
twelve. Also inserted into the duct were straightening vanes
downstream of the turning vanes. These conduct gas toward two
additional vane sets which complete the gas redistribution. Operation
is now not complicated by flow through the tray and hole plugging.
Tray Flatness
Proper performance of the bubbling tray is not only dependant on
proper gas distribution. It is also necessary to have equal slurry
6B-5
-------
depth over the entire tray surface. The tray must be flat. Initial
specification of flatness to within 1/8 inch was not met by the
construction contractor. Mapping of tray topography resulted in
corrective work that produced an even worse condition. The tray was
made flat on the fourth try and problems resulting from unequal fluid
depth have been eliminated.
Heat Recovery
Waste heat is recovered from Hue gas for use in evaporation of water
in the potassium sulfate recovery system. There have been no problems
with the heat recovery and reuse system other than the previously
referenced change in heat exchanger shell material.
Evaporation
Production of potassium sulfate crystals from the alkali sulfate
extracted from the waste cement kiln dust depends on evaporation of
solution by use of recovered waste heat. Recovery and use of waste
heat has worked well. The energy supply is more than sufficient for
the evaporation needs.
£,££2, Crystal-Liquid Separation
Crystals of potassium sulfate form in the evaporation system as the
potassium sulfate solution becomes saturated by evaporation of water.
Crystals of any other dissolved solids present will also form. One
constituent present in low concentration is calcium sulfate. Because
potassium sulfate concentration in the liquid during the early
operation was low, large volumes of water had to be evaporated to
bring the solution to saturation. The crystals formed, therefore,
included an accumulation of calcium sulfate.
Product crystals are removed by centrifugation of the liquid/crystal
slurry. A moisture content of 12 percent in the product is desirable.
The calcium sulfate crystals included in the suspended solids,
however, are very fine and do not dewater well. Fifteen percent
moisture in the centrifuge output was the lowest achieved. Therefore,
pelletization of the final product was not performed. During later
operation the initial concentration of potassium sulfate solution was
increased, thereby increasing the relative concentration of potassium
sulfate crystals that will be produced.
CONTINUED OPERATION
Operation has continued, with interruptions, since December 1990.
Operation was interrupted during equipment changes as noted above.
Operation was also interrupted by several extended periods of
down-time on the cement kiln for required maintenance. Total
operating time from January 1991 to October 1991 was 6,100 hours. The
kiln will not operate during much of November, 1991 and, therefore,
total operating time for 1991 is expected to be 6,800 hours.
Operating time in 1992 is expected to be 6,000 to 7,000 hours.
6B-6
-------
OPERATING RESULTS
Flue Gas Scrubbinc-Sulfur Dioxide
Average scrubbing efficiency is 92 percent. There have been periods
during which the scrubbing efficiency has been 95 to 98 percent. The
reason for the difference is the magnitude of the S02 input
concentration. If the S0: concentration in the output scrubbed gas is
5 parts per million when the unscrubbed input gas contains 250 parts
per million the removal efficiency is 98 percent. For the same
emission, if the input S02 concentration is 25 parts per million the
removal efficiency is only 80 percent. In either case, however,
removal of all but 5 parts per million, when burning 11 tons per hour
of 2.6 percent sulphur coal, is effective control. A coal fired
cement kiln exhibits significant variation in flue gas composition
because of process related variables. Normal unscrubbed S02output
concentrations may range from 10 parts per million to as high as 800
or 1,000 parts per million. The concentration in the scrubbed output
stream from the Recovery Scrubber is frequently in the range of 1
part per million to 10 parts per million.
Figure 2, Typical Scrubber Performance, compares the records from
continuous NO, and SOj monitors for the "before" scrubbing and "after"
scrubbing emissions data taken during October 1991. Note the system
purge and self calibration that occurs daily at midnight.
Flue Gas Scrubbino-Nitroaen Oxides
Removal efficiency for nitrogen oxides (NO.) has been 5 to 25 percent
(see Figure 2 for comparison of NO. emissions before and after
scrubbing. Variability in NO, removal is probably due to changes in
burning conditions within the kiln and the resulting proportions of
NO] vs. NO in the flue gas stream.
Flue Gas Sorubhina-Carbon Dioxide
Use of carbon dioxide from the flue gas for recarbonating calcined
lime and for carbonating calcium in solution which has been derived
from dissolved gypsum results in scrubbing of carbon dioxide. In this-
cement kiln application the exit carbon dioxide concentration has
been reduced by 3 percent. In boiler applications the Recovery
Scrubber" may remove up to 15 percent of the available carbon dioxide
depending on the chemical nature of the waste used as scrubbing
reagent.
Solid Waste Reeve1ino-Cement Industry
One of the major economic and environmental benefits provided by the
process is the opportunity to recycle solid waste. In most boiler
applications use of waste as reagent will provide a significant
source of income from tipping fees. In the cement application the
process uses waste from the cement making operation, and so no
tipping fee is generated. Use of waste from cement making as
scrubbing reagent in this process, however, provides significant
other savings to the cement plant.
Waste cement kiln dust (CKD) is produced at the Dragon cement plant
6B-7
-------
at the rate of 250 tons per day. The cost in raw material,
landfilling expense, quarry lifetime, environmental controls, and
handling is very large. CKD is wasted because it contains excess
potassium and sulfate. Removal of those materials from the waste
leaves calcium, silica, iron oxide, and alumina, the normal
constituents of cement kiln raw material input. Table 1, Raw Materia 1
and Waste Ana Ivses. shows analyses for a.) normal kiln feed, b.) CKD
(waste cement kiln dust) as produced from the kiln, c.) reacted CKD
after use and chemical modification in the scrubber, and d.) combined
kiln feed produced by adding reacted waste CKD to normal raw feed in
the proportions they are routinely produced.
The essential requirements for renovation and reuse of CKD are that
potassium be reduced to near normal kiln feed levels, and that
sulfate be significantly reduced. Table 1 shows that return of
reacted waste CKD to kiln feed provides excellent raw material.
Solid Waste Recvclinq-Other Industries
Ash from combustion of biomass, fly ash from combustion of coal, and
other caustic wastes may also be used as scrubbing reagent. In each
of these, soluble alkali will be extracted as the sulfate salt while
calcium compounds, for example gypsum, will be dissolved and
reprecipitated as the calcium carbonate plus sulfate in solution. Use
of these wastes as reagent, therefore, provides a spent reagent which
at best will be used as raw material feed to a cement plant, and at
worst will be landfilled as material free of soluble alkalis and
leachable compounds.
Bv-Product Production-Potassium Sulfate
Cement kiln dust at the Dragon cement plant contains 3 to 5 percent
K;0 as potassium oxide or potassium sulfate. During the scrubbing
process the pot.'xssiun is combined with sulfate scrubbed from the flue
gas stream. Because the sulfate salt has high solubility it is easily
separated from the various insoluble solids in the CKD. Dse of heat
recovered from the flue gas for evaporation of water allows
economical recovery of solid crystalline potassium sulfate.
Table 2, Potassium Sulfate Ana 1vsis. gives the composition of the
recovered solid. The sample represented here was taken early in the
crystallizer operating history and includes material added to the
crystallizer system as seed crystals needed to provide nucleation
sites and promote growth of a large number of crystals The added
seed crystals were calcium sulfate. Therefore, both the calcium and
the sulfate content of the recovered precipitated solids are higher
than would be expected during normal production. Potassium sulfate
comprises approximately 61 percent of the total sample. Material
produced at a later time, after initial seed crystal material has
been processed out of the system, is expected to be 78 to 80 percent
potassium sulfate.
Bv-Product Production-Distilled Water
Evaporation of potassium sulfate solution for recovery of potassium
sulfate crystals also yields distilled water. In the Dragon cement
plant application the distilled water is returned to the process as
part of the make-up water supply. In the future it will be sold and
6B-8
-------
replaced in the process with other liquid effluent needing treatment.
For other applications the distilled water may be sold or used as
boiler make-up supply.
APPLICATIONS IN OTHER INDUSTRIES
References have been made to the applicability of the process to
other industries throughout this paper. Examples of some specific
applications which are currently under discussion may be helpful.
Pulp and Paper Industry
Many pulp mills now burn their waste biomass in order to avoid its
landfill disposal and create instead an ash disposal problem. Dse of
biomass ash, which typically has significant potassium content, in
the same manner as CKD in the scrubbing system can provide the
scrubbing of flue gas from oil or coal fired boilers and income from
the potassium sulfate produced.
The case for one mill currently being evaluated is instructive. They
will scrub sulfur dioxide from flue gas allowing their continued use
of 2.5 percent sulfur oil; consume ash which now costs $1 million per
year to landfill;
transport all spent ash to a cement plant for use there as kiln feed;
and produce both potassium sulfate and distilled water as
by-products. The process would generate no waste and provide in
excess of $6 million per year in combined savings and income while
eliminating the mill's need to landfill ash.
Waste To Energy Industry
Generally the ash from municipal trash incinerators is deficient in
alkaline material and will not be sufficient for complete reaction
with the acid gas constituents. If biomass ash is added in some
proportion the process works as in either pulp and paper or cement
applications. The recovered soluble salts will be sodium chloride
with some amount of potassium sulfate and potassium chloride rather
than just potassium sulfate. The value of those recovered materials
will, therefore, be lower.
Spent ash from the process will be useful in the manufacture of
cement. It may contain small quantities of various heavy metals, but
will consist primarily of silica, alumina, calcium carbonate, and
iron oxide which are the principal required ingredients for cement
kiln raw material input. If a cement kiln is not available to receive
spent ash the material may be landfilled at relatively low cost. Low
cost disposal is possible because the ash will no longer contain
soluble materials. The leaching of toxic metals into the ground water
table will no longer be a concern.
Dtilitv Industry
A waste to energy plant now being evaluated will use ash from the
waste incinerator, mixed with an equal mass of biomass ash, as
scrubbing reagent. Alkali metal sulfate and chloride aslts will be
produced. Spent ash will be used as raw material input to a cement
kiln.
6B-9
-------
Table 1
RAW MATERIAL
AND WASTE ANALYSES
Oxide
d
Kiln Feed
h
Waste CKD
c
Reacted
Waste CKD
d
Combined Waste
arid K i 1 n Feed
SiOi
! 21"3'
18.7
19.9
21.2
A] 303
j 4.8
3.7
6.0
4.9
Fe.Oa
' ° 1
t - • ±
|
1.8
2.7
2.1
CaO
! 65.6
(
54.5
62.4
65.3
MgO
I
¦ A *>
! 4'"
2.8
4.3
4.2
S03
! 0.46
i
9.7
2.5
0.56 ;
K = 0
t
: i.o7
8.6
1. 6
1.1
Na
; 0.2
0.7
0.2
0.2
Note: The balance of material in each analysis is carbon dioxide-'
as present in calcium carbonate.
6B-10
-------
Table 2.
POTASSIUM SULFATE ANALYSIS
Chemical Species Concentration (Percent.)
K20 33-40
Na20 0.12
CaO 12.12
Si02 1.16
A1203 0.26
MgO 1.63
C03 8.07
S04 44.6
6B-11
-------
Figure 1
6B-12
-------
Figure 2
Typical Scrubber Performance
350
300-
NOx Before
250-
200
NQx After
150
100-
50-
SOo After
Note: Calibration occurs at midnight.
-------
Intentionally Blank Page
6B-14
-------
The NOXSO Clean Coal Technology Demonstration
Project
6B-15
-------
Intentionally Blank Page
6B-16
-------
L.G. Neal
John L. Haslbeck
Hark C. Hoods
NOXSO Corporation
P.O. Box 469
Library, Pennsylvania
ABSTRACT
The NOXSO Clean Coal Technology Project is a 115 MW demonstration unit to be located
at Ohio Edison's Niles Station. The project is co-funded by the U.S. Department of
Energy (DOE) and a consortium of companies assembled by NOXSO including NOXSO
Corporation, W.R. Grace & Co., UK-Ferguson Company, Ohio Edison, the Ohio Coal
Development Office (OCDO), the Electric Power Research Institute (EPRI), the Gas
Research Institute (GRI), and the East Ohio Gas Company. The DOE manages the project
through the Pittsburgh Energy Technology Center (PETC). Both the NOXSO Process and
its application to the Niles Plant are described in this paper. The status of the
NOXSO Proof-of-Concept Pilot Plant is updated, and its impact on the Niles
Demonstration Plant design is described. Finally, the NOx recycle test program that
is being performed concurrently with the pilot plant operations is discussed.
INTRODUCTION
The NOXSO Process is a post-combustion flue gas treatment technology that removes both
sulfur dioxide (SO?) and nitrogen oxides (NO,,) from the flue gas of a coal-fired
utility. Under development since 1979, the process is in the final three stages of
commercialization. The first stage is a 5 MW Proof-of-Concept (POC) pilot plant that
was built at Ohio Edison's Toronto Station in Toronto, Ohio (construction was
completed in July of 1991). -The second stage is a 500 lb/hr (coal feed rate) test
of the NOx recycle concept using a Babcock & Wilcox (B&W) cyclone coobustor that will
be conducted concurrently with the POC pilot tests. The third and final stage is a
115 MW full-scale demonstration plant to be built at Ohio Edison's Niles Station in
Niles, Ohio. The 115 MW Demonstration Project will be cost shared between the
Department of Energy through the third round of the Clean Coal Technology program by
6B-17
-------
a cooperative agreement between DOE and MK-Ferguson. DOE will provide 50% of the
funds necessary to build and operate the plant while the remaining 50% will be covered
by NOXSO, Grace, MK-Ferguson, Ohio Edison, OCDO, EPRI, and GRI. DOE will manage the
demonstration project through the Pittsburgh Energy Technology Center (PETC). In this
paper we describe the NOXSO Process as it will be implemented at Ohio Edison's Piles
Station and the current schedule for design, construction and operation of the 115
MW facility. We also describe the test programs planned for the POC pilot plant that
will provide the final design and scale-up data necessary for the Niles Plant. Also,
NO. recycle data obtained previously during the 3/4 MW test program is described and
the test program planned for the B&W NO, recycle tests is described.
HOST SITE DESCRIPTION
The Niles Station is located on the Mahoning River in northeastern Ohio and is shown
in Figure 1. It has a net demonstrated power production capability of 246 MW. Two
coal-fired units produce 108 MW net each (115 MW gross each) and 30 net MW is obtained
from a combustion turbine which is used for peaking purposes. At full load, the plant
fires 97 tons of bituminous coal per hour. The average annual coal quality analyses
for 1987 and 1988 are shown in Table 1. Of all the coal received at the Niles Plant,
60 percent is typically Ohio coal and 40 percent is non-Ohio (western Pennsylvania).
Both process and cooling water are withdrawn from the Mahoning River at a rate of
140,550,000 gallons per day. Ample water supply is available for the NOXSO Process
requirements which amount to less than 200,000 gallons per day. NOXSO electricity
requirements will be provided by the Niles plant and are estimated to be about 4% (or
4.6 MW) of the gross power output of Unit #1.
NOXSO PROCESS DESCRIPTION
The NOXSO demonstration plant will be retrofitted to Niles unit #1, a pulverized coal-
fired cyclone boiler with a rating of 115 MW (gross) and 108 MW (net). The tie-in
point will be the flue gas ductwork between the existing electrostatic precipitator
(ESP) and the plant stack. The NOXSO Process can operate either upstream or
downstream of the particulate collection device; however, the current tie-in point
was chosen to minimize the effect on ESP performance. The demonstration plant will
occupy an area 120 feet by 150 feet. A description of the process technology is given
below and a process flow diagram is shown in Figure 2.
6B-18
-------
Flue gas from unit /X will be combined with the Claus Plant tail gas stream and ducted
to the NOXSO flue gas booster fan. Downstream of the flue gas fan, the flue gas is
cooled by vaporizing a stream of water sprayed directly into the ductwork. After
being cooled, the flue gas is passed through two parallel fluidized bed adsorbers
where SO? and NO, are simultaneously removed using a high surface area -y-alumina
sorbent impregnated with an alkali material. The cleaned flue gas is returned to the
plant ductwork and exits through the stack.
The sorbent is removed from the adsorbers by an overflow pipe, and is then transported
by one of two dense-phase pneumatic conveyors to the sorbent heater. Fresh make—up
sorbent is added downstream of the adsorbers so that it may be calcined in the sorbent
heater before making its first pass through the adsorbers. The sorbent heater is a
three-stage fluidized bed where a hot air stream is used to raise the sorbent
temperature to 1120°F. During the heating process, NO, and loosely bound SOj are
desorbed and transported away in the heating gas stream. This hot air stream is
recycled back to the cyclone burners and replaces approximately 38% of the required
combustion air. A portion of the recycled HO, is converted to nitrogen (N2) and either
carbon dioxide (CO,) or water (H^O) by reaction with free radicals in the reducing
atmosphere of the combustion chamber. NO, recycle studies were performed during a
previous NOXSO test program (a 3/4 MW pre-pilot scale test) and additional NO, recycle
studies are currently under way. These tests are discussed in more detail below.
Once the sorbent reaches a regeneration temperature of 1120°F, it is transported by
means of a J-valve to a moving bed regenerator. In the regenerator, sorbent is
contacted with reformed natural gas in a countercurrent manner. The reformed natural
gas reduces sulfur compounds on the sorbent (mainly sodium sulfate) to primarily SOj
and hydrogen sulfide (H^S) with some carbonyl sulfide (COS) also formed.
Approximately 20% of the sodium sulfate (NajSOj) is reduced to sodium sulfide (Na2S)
which must subsequently be hydrolyzed in the Bteam treatment vessel. A moving bed
steam treatment reactor follows the regenerator, and a concentrated stream of H^S is
obtained from the reaction of steam with NajS. The off-gases from the regenerator and
steam treater are combined and sent to a Claus Plant. Elemental sulfur is the end
product from the Claus Plant. The tail gas stream from the Claus Plant is passed
6B-19
-------
'through an incinerator to convert all remaining sulfur compounds to SOj, cooled to
about 350°F, and recycled to the flue gas stream prior to entering the adsorbers.
From the steam treatment vessel, the sorbent is transported by means of another J-
valve to the sorbent cooler. The cooler is a three-stage fluidized bed using ambient
air to cool the sorbent. The warm air exiting the cooler is further heated by a
natural gas fired in-duct heater before being used to heat the sorbent in the
fluidized bed sorbent heater. The sorbent temperature is reduced in the cooler to
the adsorber temperature of 250°F. Sorbent from the sorbent cooler overflows into
a surge tank. The surge tank is used as a source and sink for sorbent to maintain
constant bed levels in the other process vessels. From the surge tank, sorbent is
transported to the adsorbers again by means of a J-valve thus completing one full
cycle.
NILES DEMONSTRATION PLANT SCHEDULE
Much of the information required to design the full-scale demonstration plant is
already available through earlier NOXSO test programs. The POC pilot plant will
supply additional design data and scale-up information. Design of the Niles
demonstration plant will begin during the operation period of the pilot plant. Thus,
pilot plant data will be used to refine the Niles design. Preliminary design work
on the Niles plant is scheduled to begin in September of 1991. Detailed design will
be completed in October of 1992, at which time the POC test program will be complete.
Plant construction will then begin in November of 1992 and run through March of 1994.
The operations period will last for a period of 24 months through March of 1996. At
the completion of the operation period, Ohio Edison will have the option to purchase
the unit for continued operation. The schedule is summarized below in Figure 3.
Data for the Niles plant detailed design will come from three sources. The first
source is previous NOXSO test programs. NOXSO Corporation has conducted laboratory-
scale tests, pre-pilot scale, tests (3/4-MW), and a life cycle test of the NOXSO
Process. Each of these test programs have provided data useful in process design,
and the results of each test program have been reported previously (1,2,3). The
second source of design information is the POC pilot plant which will provide
additional design and scale-up data as well as materials selection data through
implementation of a corrosion test program. The third source of design information
6B-20
-------
will be from a NO, recycle test program to be conducted at the Babcock & Wilcox (B&W)
Research Center in Alliance, Ohio. The POC test program and NO. recycle tests are
discussed below.
POC PILOT PLANT TEST PROGRAMS
The POC pilot plant began cold start-up in July of 1991. Cold start—up was the first
of three test series. The second test series is a hot start-up with inert gases and
is currently in progress. The third test program is a set of parametric tests with
the system fully operational, i.e., using flue gas in the adsorber and reactive
(rather than inert) gases in the ri.generator. There are thirty parametric tests
planned and the process parameters being varied are sorbent circulation rate, adsorber
settled bed height, regenerator solids residence time and adsorber gas flow rate.
The parametric tests will be followed by a duration test at optimum process conditions
as defined by the parametric tests. The results from these tests will be included
in the detailed design of the Niles facility.
The first test program, cold start-up, was designed to verify the proper operation
of each piece of equipment in the plant. After initial shakedown tests, sorbent was
circulated through the system continuously for 43 hours. This test revealed the need
to modify vessel internals in the staged fluid beds to achieve the maximum required
sorbent circulation rates. After the modifications were completed, a hot sorbent
circulation test was performed for 38 continuous hours. The hot circulation test
showed that the fluid bed residence time needed to be increased to achieve adequate
heat transfer in the sorbent heater and sorbent cooler. After these additional
modifications were completed, a second hot sorbent circulation test was initiated.
This current test program includes gas tracer studies and operator testing of the
distributed control system trip matrix. Following this final hot inert test, flue
gas will be treated in the NOXSO adsorber and the entire system will be run with
reactive gases.
A corrosion test program is also planned during POC plant operation. Corrosion test
spools containing metal test samples will be installed in seven different locations
to assess corrosion rates in different gas and sorbent environments. Coupon weights
and dimensions are measured before and after exposure, and these values are used to
calculate corrosion rates of each material. Table 2 lists corrosion spool locations
6B-21
-------
at the POC and the process components that will experience the same environment.
Figure 4 is a photograph showing a corrosion teat spool prior to installation at the
pilot plant. The materials to be tested on each corrosion spool of coupons are listed
below in Table 3.
The expected duration of the entire POC test program is about 10 months. In this
relatively short period of time, it may be difficult to distinguish corrosion rates
between some of the materials tested. Therefore, concurrently with the POC test
program, there will be an accelerated corrosion test program will be conducted by an
independent laboratory. The accelerated corrosion tests will consist of exposing
corrosion coupon spools to simulated regenerator environments. A total of six tests
will be conducted. The tests will be at three different temperatures and two
different gas compositions. The reactor tube containing the corrosion test spool will
also be packed with sulfated NOXSO sorbent to simulate the regenerator vessel
environment. The test matrix is listed in Table 4. Each test condition will last
for three weeks of continuous exposure. The results of these two corrosion test
programs will be used to select materials of construction for the Niles demonstration
plant.
The NO, recycle concept will not be tested at the pilot plant because the POC only
uses a slipstream (12,000 SCFK) of flue gas. However, simulated NO. recycle tests
were conducted during the pre-pilot scale tests conducted at the DOE's Pittsburgh
Energy Technology Center. These tests showed that from 65% to 75% of the recycled
NO, was destroyed in the combustion chamber (2). Additional NO, recycle tests are
planned at the BSW Research Center. B&W has a small boiler simulator (SBS) that
mimics the operation of the Niles cyclone burners. A schematic of the B&W SBS is
shown in Figure S, and a comparison of operating parameters with a typical cyclone
fired boiler is shown in Table 5.
The NO, recycle tests are conducted by injecting bottled NO, compounds into the coal
combustion air in concentrations that reproduce the NO, concentration in the sorbent
heater off-gas.
The test program at B&W will consist of shakedown, baseline, simulated NO recycle,
simulated NO2 recycle, and novel concept tests. The shakedown tests are designed to
6B-22
-------
optimise furnace operation including cyclone burner settings, injection system
equipment, and sampling instrumentation. Following the shakedown tests, baseline
tests will be conducted to establish NO, emission levels at three loads and three
excess air levels without NO, reinjection- Once the baseline NO. emissions have been
quantified, NO, recycle tests will begin. First, NO will be injected in multiples of
0.5, 1.0, 1.5, and 2.0 times the baseline NOx production rate. The first injection
point will be in the primary combustion air and tests will be run at the four NO,
recycle rates, with three furnace loads, and three excess air levels for a total of
36 tests. NO will then be injected in the secondary air stream using the same four
NO recycle rates, with two furnace loads and two excess air levels for a total of 16
additional tests. The same set of tests will be repeated (two injection points, four
NO, recycle rates, two loads, and two excess air levels) using NO; injection in place
of NO. An allowance for 15 additional tests has been included in the test plan.
These tests could be used to examine novel ideas for the enhancement of NO,
destruction such as the addition of methane to the NO, recycle stream.
SUMMARY
NOXSO Corporation's Clean Coal technology project is a 115 MW demonstration of the
NOXSO flue gas treatment process. The host site for the project is Ohio Edison's
Niles station located on the Mahoning River in Niles, Ohio. Preliminary design for
the demonstration unit is scheduled to begin in late 1991 with detailed design being
completed in late 1992. Plant construction should then be completed in early 1994
when operation will begin. Much of the necessary design data has been acquired
through previous experimental test programs. The final design data required will be
obtained from NOXSO's POC pilot plant, and the NO, recycle studies will get under way
at B&W's Research Center using their small boiler simulator.
REFERENCES
1. Kaslbeck, J.L., C.J. Waqg, L.G. Neal, H.P. Tseng, and J.D. Tucker, "Evaluation
of the NOXSO Combined NO,/SO; Flue Gas Treatment Process", U.S. Department of
Energy Contract NO. DE-AC22-FE60148, November 1984.
2. Haslbeck, J.L., W.T. Ma, and L.G. Neal, "A Pilot-Scale Test of the NOXSO Flue
Gas Treatment Process", U.S. Department of Energy Contract No. DE-FC22-
85PC81503, June 1988.
3. Ma, W.T., J.L. Haslbeck, and L.G. Neal, "Life Cycle Test of the NOXSO Process",
U.S. Department of Energy Contract NO. DE-FC22-85PC81503, May 1990.
6B-23
-------
Piguro 1. Ohio Ed i Don' o IHIod Station, Nlleo, Otiio.
-------
NO* RECYCLE
SORBENT
HEATER
TO CLAUS
NATURAL
, GAS
AIR
HEATER
flue gas to stack
FLUE
GAS
ADSORBER
CENSE I
phase f]
transport li
water
Figure 2. NOXSO Process Flow Diagram.
6B-25
-------
1991 1992 1993 1994 1995 1996
Design
Construction
Operation & Test
Figure 3. Clean Coal III Project schedule.
Figure 4- Photograph of Corrosion Spool.
6B-26
-------
SU»ERmEATE«
FOULING TutE
DEPOSITION IIDIE
steam
b
R£mE*tEs
DEPOSITION —
PROBE
OvEK »'•£
AIM
MOLTEN SlAG
RETURNING
•URNERl
SECONDARY AiR
flue g«s
RECIRCULATION
SLAC COLLECTOR
AND FURNACE
WATER SEAL
Figure 5. Babcock & Wilcox's Small Boiler Simulator (SBS) Schema-tic
6B-27
-------
Table 1
Annual Coal Quality Analysis for the Niles Plant
1987 and 1988>
1987
1988
Moisture (%)
7.82
7.51
Ash (%)
11.72
11.98
Sulfur
3.17
3.24
Heating Value (Btu/lb)
11,694
11,735
Tab]
Location of POC Corrc
Process and Com
.e 2
>sion Test Spools and
ponents Affected
Spool Location
components
t\, Adsorber Inlet
Ductwork between spray cooler and
adsorber, base of adsorber, and
adsorber qas distributor.
#2, Adsorber Outlet (top of
adsorber)
Adsorber (above distributor),
adsorber cyclone, and ductwork
between adsorber and stack.
#3, Air Heater Outlet
Air heater, duct between air heater
and sorbent heater, bottom gas
distributor in sorbent heater, and
sorbent heater.
X4, Regenerator (gas space)
Regenerator, piping between
regenerator and incinerator, and
control valves on pipinq.
XS, Regenerator (sorbent bed)
Regenerator, sorbent transfer line
from sorbent heater to regenerator,
and transfer line from regenerator
to steam treater.
X6, Steam Treater (gas space)
Steam treater, piping between steam
treater and incinerator, and control
valves on pipinq.
X7, Steam Treater (sorbent bed)
Steam treater, vessel surface in
contact with sorbent.
6B-28
-------
Table 3
Materials -to be Tested During
The POC Corrosion Test Proqram
Materials
Spool No.
Accel-
erated
Tests
1
2
3
4
5
6
7
STAINLESS STEEL
304 SS
X
X
X
X
X
X
X
X
304H SS
X
X
X
X
X
X
316 SS
X
X
X
X
X
X
X
X
446 SS
X
X
X
X
X
X
X
X
1010 CS
X
X
X
X
X
X
X
X
HASTELLOYS
C—276
X
X
C—22
X
X
C-4
X
X
304 SS (Alonized)
X
X
X
X
X
X
304H SS (Alonized)
X
X
X
X
X
X
316 SS (Alonized)
X
X
X
X
X
X
1010 CS (Alonized)
X
X
X
X
X
X
304 SS (Chromized)
X
X
X
X
X
X
1010 CS (Chromized)
X
X
X
X
X
X
OVERLAYS
304 SS with 556 SS
X
X
X
X
X
X
304 HS with HR—160
X
X
X
X
X
X
304 SS with 446 SS
X
X
X
X
X
X
304H SS with 446 SS
X
X
X
X
X
X
SPRAYCOAT, AFTER WELDS
Alonized 304 SS with 446 SS
X
X
X
X
X
X
Alonized 304H SS with 446 SS
X
X
X
X
X
X
304 SS with 446 SS
X
X
X
X
X
X
304H SS with 446 SS
X
X
X
X
X
X
Haynes 556
X
X
Haynes HR-160
X
X
Carpenter 20Cb3
X
X
Jeseop JS276
X
X
Inco C—276
X
X
Inco 625
X
X
Teflon
X
X
6B-29
-------
Table 4
Accelerated Corrosion Test Conditions
Test
No.
Temp.
(eF)
Gas
Environment
1
1200
40%CO, 40%SO,, lOtEjO, lOiCH^
2
1400
40«CO, 40%SO,, 10%H^O, io%cs,
3
1600
40«CO, 40%SCK, lOtHjO, 10%CH«
4
1200
50%HjS, 50%H-O
5
1400
50%HjS, 50%H;O
6
1600
50%HjS, 50%HIO
Table 5
Comparison of Operating Parameters for the B&W SBS
and a Typical Full-Scale Cyclone Burner
Parameter
SBS
Typical
Cyclone Burner
Cyclone Temperature
>3000°F
>3000°F
Residence Time at Full
Load
1.4 sec
0.7-2 sec
Furnace Exit Gas
Temperature
2265°F
2200-2350°F
NO, Level
900-1200 ppm
600-1400 ppm
Ash Retention
80-85%
60-80%
Unturned Carbon
1% in ash
1-20%
Ash Particle Size
6-8 microns
6-11 microns
6B-30
-------
Economic Comparison of Coolside Sorbent Injection
and Wet Limestone FGD Processes
6B-31
-------
Intentionally Blank Page
-------
ECONOMIC COMPARISON OF COOLSIDE SORBENT INJECTION
AND WET LIMESTONE FGD PROCESSES
D. C. McCoy, R. M. Statnick, M. R. Stouffer, and H. Yoon
Consolidation Coal Company
4000 Brownsville Road
Library, Pennsylvania 15129
P. S. Nolan
Babcock and Wilcox
20 S. Van Buren Avenue
Barberton, Ohio 44203
ABSTRACT
The Coolside process is a duct sorbent injection process developed for retrofit S02
control on a coal-fired boiler. The process is attractive for retrofit applications
because of low capital cost, low space requirements, and short procurement-through-
installation time in comparison to wet flue gas scrubbers. The Coolside demonstra-
tion was conducted from July 1989 to February 1990, on the 104 MWe Unit No. 4
Boiler 13 at the Ohio Edison Edgewater Station, Lorain, Ohio, under the partial
sponsorship of the U.S. Department of Energy Clean Coal Technology Program. The
Edgewater demonstration achieved 70 percent S02 removal while burning 3 percent
sulfur coal. Short-term process operability was demonstrated during continuous
process operations under steady state conditions for up to eleven days. The
demonstration provided information on desirable process equipment design improve-
ments which would be required for commercial operation. This paper analyzes the
factors which influence Coolside process economics, i.e., sorbent price, utility
plant capacity factor, book life, and waste disposal cost. The optimized Coolside
process and wet limestone FGD capital and total levelized annual costs are compared
as functions of boiler capacity and coal sulfur content.
6B-33
-------
INTRODUCTION
In 1987, Babcock & Wilcox (B&W) was awarded a Cooperative Agreement under the DOE
Clean Coal Technology Program to demonstrate the Cool side and LIMB sorbent injection
processes at the Ohio Edison Edgewater Station. In addition to DOE, project
financial participants included B&W, Consolidation Coal Company, Ohio Edison Company
(host utility), and the State of Ohio Coal Development Office. This paper describes
the Cool side process results. The Cool side process demonstration was conducted from
late July 1989 to mid February 1990. During that period, Edgewater Unit 4,
Boiler 13 was burning compliance (1.2 to 1.6 percent sulfur) and non-compliance
(2.8-3.2 percent sulfur) coals. The objectives of the full-scale test program were
to verify short-term process operability and SO, removal, to determine factors which
could affect long-term process operability, ana to develop a Coolside process design
and performance data base to establish process economics. The demonstration program
included sorbent once-through and simulated recycle operations. Key process
variables (Ca/S mole ratio, approach to adiabatic saturation temperayure (aTs),
Na/Ca mole ratio) were evaluated in short-term (6-8 hr) parametric tests and longer-
term (1-11 day) process operability tests. Two different commercially available
hydrated limes were tested. Prior to the demonstration, pilot-scale tests were
conducted to select hydrated limes to be tested and to develop Edgewater site-
specific process performance data. The pilot data were used to develop the
demonstration test program and aided in interpreting the full-scale results. This
paper discusses a commercial Coolside process design, the factors which influence
Coolside process economics, and the potential Coolside process market (boiler
capacity and coal sulfur content).
General Process Description
The Coolside desulfurization technology involves pneumatic injection of dry hydrated
lime (sorbent) into the flue gas downstream of the air preheater, followed closely
in distance by flue gas humidification using water sprays. Flue gas S02 is captured
by reaction with the entrained sorbent particles in the humidifier and with the
sorbent collected in the particulate removal system. The humidification water
serves two purposes. First, it activates the sorbent to enhance S02 removal and,
second, it conditions the particulate matter to maintain efficient electrostatic
precipitator (ESP) performance. Spent sorbent is removed from the gas along with
fly ash in the existing particulate collector (ESP or baghouse). The sorbent
activity can be significantly enhanced by dissolving sodium containing additives
such as sodium hydroxide (NaOH) or sodium carbonate (Na2C0^) in the humidification
water.1,2 Because of the relatively short demonstration period, NaOH was used as the
additive in the Edgewater demonstration. Sorbent recycling can be used to improve
the sorbent utilization if the particulate collector can handle the resulting
increased solids loading.
6B-34
-------
Edaewater Cool side Performance
The Edgewater Cool side program demonstrated that the Coolside process can routinely
achieve 70 percent SO, removal at the design conditions of 2.0 Ca/S mole ratio, 0.2
Na/Ca mole ratio of additive, and 20*F approach to adiabatic saturation temperature
using commercially-available hydrated lime. Coolside SO, removal is dependent on
Ca/S mole ratio, Na/Ca mole ratio, approach to adiabatic saturation (a"L), and
physical properties of the hydrated lime. A detailed discussion of the effects of
these variables is presented elsewhere.3'4
Although the data are limited, sorbent recycle showed significant potential to
improve sorbent utilization. The observed S02 removal with recycle sorbent alone
was 22 percent at 0.5 available Ca/S and aTs « 18*F approach. The observed SO,
removal with simultaneous recycle and fresh sorbent feed was 40 percent at 0.8 fresn
Ca/S, 0.2 Na/Ca (fresh), 0.5 available recycle Ca/S (about 7,000 lb/hr recycle feed
rate), and aTs = 18"F.
COOLSIDE AND WET LIMESTONE FGD PROCESS DESCRIPTIONS
Commercial Coolside Process Description
The Coolside process design described here is based on current FGD industry practice
and the results of the Edgewater Coolside process evaluation. The design is based
on mature Coolside technology. The Coolside process consists of four major process
areas: sorbent and additive receiving/storage/preparation/feed; flue gas humidifica-
tion; particulate removal and solids recycle; and waste disposal.
Sorbent and Additive Receiving/Storage/Preparation/Feed. There are two calcium
sorbent preparation options. The first is purchase of quicklime (CaO) which is
hydrated on-site. The second is purchase of hydrated lime (Ca[0H]2). As presented
in Figure 1, for 1.5 percent sulfur coal, the purchase of hydrated lime has a lower
life cycle cost. For other coal sulfur contents, the purchase of hydrated lime is
favored for all cases in which the Coolside process is economically attractive ,
compared to the wet limestone, forced oxidation process. In all cases, bulk
delivery of hydrated lime is by rail. As shown in Figure 2, the hydrated lime is
pneumatically transferred from the rail unloading area to bulk storage. From bulk
storage, the sorbent is conveyed pneumatically to a day bin, and then volumetrically
metered by a powder pump through a distribution bottle to the humidifier duct
injection nozzles.
Because of its lower cost? soda ash (NaJCOj) is used as an additive to increase S02
removal and hydrated lime utilization. Dry soda ash is unloaded pneumatically from
trucks into a 30-day, wet soda ash storage and supply system (shown in Figure 3).
The soda ash feed system is a package unit which maintains a saturated solution of
sodium carbonate. The saturated soda ash solution is metered, dependent upon the
hydrated lime feed rate and the Na/Ca mole ratio setpoint, to an in-line mixer in
the humidification water supply line.
Flue Gas Humidification. As shown in Figure 4, the boiler flue gas is conveyed to
a single vertical downflow humidifier. In the humidifier, water containing the
sodium additive is fed to an atomizer lance array of dual-fluid atomizing nozzles.
Each nozzle is designed to operate at a 0.8-1 gpm throughput at an atomizing air-to-
humidification water ratio of 0.5 lb of 120 psig air/lb of water. The dual-fluid
nozzles produce very fine water droplets (-35 micron Sauter Mean Diameter) with less
6B-35
-------
than 1 weight percent of the droplets greater than 100 microns. The humidifier is
sized to provide three seconds residence time at full boiler load. The humidifica-
tion water feed rate is controlled to maintain an outlet setpoint temperature which
is typically 140 to 145*F.
The hydrated lime and recycle solids are conveyed to distribution bottles where the
feed streams are split into several equal, smaller streams. The hydrated lime and
recycle solids are blown into the flue gas through an array of injector pipes
located in the plane of the humidification nozzles.
Particulate Removal and Solids Recycle. As shown in Figure 4, the humidified flue
gas is conveyed to an ESP. All of the reference plants used in the economic
evaluation are assumed to have ESPs with a specific collection area (SCA) of 400
ft2/1000 acfm, ESP particulate collection efficiencies of 99.6+ percent and a
particulate emission limit of 0.1 lb/106 Btu.
The fly ash, Cool side reaction products, and unreacted hydrated lime are collected
and removed in the ESP. The collected solids are pneumatically conveyed from the
ESP hoppers to a fly ash/spent sorbent silo for intermediate storage. The cleaned
flue gas exits the ESP and passes through the induced draft fan to be discharged
through the plant stack.
Recycle of the solids collected in the ESP is used to increase sorbent utilization.
The solids in the fly ash/spent sorbent silo are metered through a volumetric feeder
and pneumatically conveyed to the recycle solids distribution bottle. The
distribution bottle uniformly splits the solids flow l.^fo several smaller streams
prior to injection into the flue gas.
Waste Disposal. The waste material is a fine powder. To prevent dust emissions
during silo unloading, the wastes are fed through a dustless unloader where water
is added to moisten the solids before discharge into dump trucks. The waste is
trucked to a landfill for disposal. Coolside waste was evaluated using the EP test
procedure5 and was found to be non-hazardous.
Met Limestone FGD Process Description
The wet limestone forced oxidation (LSFG) FGD process was selected for economic
comparison with the Coolside process. The process is shown in Figure 5. Limestone
is delivered by rail to the site and then transferred to a 30-day covered storage
pile. The limestone is then fed to a day bin and, as required, to the grinding
equipment for pulverization to 90 percent minus 325 mesh in a wet ball mill. The
limestone slurry produced is metered to the S02 absorber. The absorber is a single
module, rubber-lined, carbon steel, open spray tower which treats the flue gas from
the boiler. Boiler flue gas passes through a booster fan and enters the absorber
where the gas is contacted with limestone slurry. The SOg reacts with the slurry
forming calcium sulfite, CaS0j*l/2H20, and calcium sulfate (gypsum), CaS04*2H20. The
reacted slurry collects in the absorber sump. Air is blown into the absorber sump
at a stoichiometric ratio of 1.5 mole 0,/mole S0? absorbed to convert all the
sulfite to sulfate. Large slurry pumps take suction from the absorber sump and
recycle the slurry back through nozzle-equipped spray headers in the absorber.
Slurry from the absorber sump is sent to a thickener for partial dewatering. The
thickener overflow returns to the absorber while the underflow is pumped to rotary
drum vacuum filters. Gypsum is dewatered to 80 percent solids and disposed of in
a landfill. The clean flue gas is exhausted through a new wet stack.
6B-36
-------
Economic Premises
Power Plant Parameters. The reference plant performance assumptions are listed in
Table 1. The site plan is assumed to be similar to those in DOE PON DE-PS01-
89FE61825. For the purpose of Cool side and wet FGD process layouts, all plant sizes
(100, 150, 250, and 500 MWe) are assumed to be equipped with two parallel air
preheaters, ID fans, and ESPs. The designs are based on eastern bituminous coals
containing 1.5 (2.2), 2.5 (3.7) , and 3.5 (5.2) weight percent sulfur (lbs SOj/106
Btu). The fuel specifications are listed in Table 2. The nominal flue gas composi-
tion, rate, and temperature are listed in Table 3.
Economic Assumptions. The Consol Cool side and wet FGD models use a combination of
capital equipment cost algorithms and look-up tables to estimate specific plant
costs for individual equipment items or equipment packages. Specific equipment cost
information was developed from internally funded FGD design reports, vendor quotes,
and public literature sources. Installed plant costs (IPC) are determined by:
IPC « Z(EC,- x BFj x RFJ x CI x SF
where: EC,- = individual (denoted by ,-) process equipment costs, BFi - bulk factor
(1.0 to 3.0), RF,- = retrofit factor, CI = process plant cost index adjustment, and
SF - site factor. The retrofit factors were individually assigned. The average
plant retrofit factors for the Cool side and LSFO processes were between 1.28 to
1.30. The southern Ohio site factor was 1.06.
The total capital requirement (TCR) is determined by:
TCR = IPC + IFC + HOC + BIC + PC + IDC + PSC + IC
where: IFC - indirect field costs (0.138 x IPC); HOC « home office costs (0.224 x
IPC); BIC * bond and insurance costs (0.011 x IPC); PC - project contingency (0.18
x [IPC + IFC + HOC + BIC]); IDC = interest during contstruction; PSC - preproduction
start-up costs; IC « inventory capital (working capital).
The calculations for IDC, PSC, and IC follow Electric Power Research Institute
(EPRI) recommendations as outlined in the EPRI TAG1".6
Variable costs are dependent on unit cost, process capacity, maintenance, and on-
stream factor. The unit costs for the economic evaluations are presented in
Table 4. For the Coolside process, incremental operating labor (OL) is 1.33
men/shift (hydrated lime, purchased). For the LSFO FGD, the incremental operating
manpower is 2.33 men/shift. The maintenance cost for labor and materials is
calculated as follows:
AM = 2(1 EC, x HFj/100)
where: AM = annual maintenance cost, IEC( - installed equipment costs in process
area (=); and MFi = maintenance cost percent in area (-). MF= is between 2.5 to 10
percent depending on service severity. Maintenance labor (ML) costs are 40 percent
of annual maintenance cost. Administrative overhead cost is 30 percent of operating
labor plus maintenance labor costs (0.3 x [OL + ML]).
The economic evaluations of process options are presented as capital cost, expressed
at $/kWe (net), and/or levelized revenue requirement, expressed as $/ton of S02
removed. In some cases, the effect of short-term levelization (10 year) is
considered. Short-term economic analysis is used by many utilities to determine if
the long-term benefits of an option show economic advantage in the short-tere when
6B-37
-------
unforeseen factors are less likely to occur. The economic factors needed for the
financial calculations are presented in Table 5.
COOLSIDE PROCESS AND WET LIMESTONE FGD COST COMPARISON
Capital Cost Comparison
The published capital cost estimates for sorbent injection and wet limestone
scrubbing vary over a wide range because of process and economic assumptions used
in each study. This study used an internally consistent set of process and economic
assumptions in developing the capital cost estimates. For this study, the relative
cost comparisons of Cool side and wet FGD economics are valid. Site-specific factors
will determine the absolute costs for "real world" applications.
Throughout this study, the LSFO FGD annual average S02 removal was assumed to be 95
percent and the Cool side annual average SO, removal was assumed to be 70 percent.
The Clean Air Act Amendments of 1990 established a 1995 SO, emission limit of 2.5
lb S0,/106 Btu which decreases in year 2000 to 1.2 lb SO^/lO6 Btu. Assuming the
utility will comply with the 1.2 lb SOj/lO6 Btu limit at each site, the Coolside
process can treat coal containing up to about 4 1b S02/106 Btu or about 2.7 percent
sulfur coal (HHV - 13,400 Btu/lb).
The Coolside and LSFO FGD capital costs are compared in Figure 6 for the three
design coals. In all cases, the LSFO FGD capital costs are higher than the Coolside
process capital costs. The LSFO FGD capital costs are 2.2 to 2.5 times the Coolside
capital costs. In cases where high S02 removal is not required, and remaining plant
life is short, lower capital cost favors the installation of the Coolside process.
Total Levelized Annual Revenue Requirement
The total level ized annual revenue requirements in constant mid-1990 dollars for the
Coolside and LSFO processes are presented in Figures 7, 8, and 9. The base case
plant capacity factor and plant life are 65 percent and 20 years, respectively. For
the 1.5 weight percent sulfur coal case (see Figure 7), the Coolside process is
economically competitive with LSFO up to a crossover point plant size of about 350
MWe. For the 2.5 weight percent case (see Figure 8), the Coolside process is
economically competitive up to about a 130 MWe plant size crossover point. For the
3.5 weight percent case (see Figure 9), the LSFO process is preferred over the
entire reference plant size range. Clearly, coal sulfur content is a critical
parameter in selecting the least-cost compliance technology.
A 65 percent plant capacity factor was assumed for the base case. The effect of
plant capacity factor on the economic crossover point and the levelized annual
revenue requirement at the crossover point between the Coolside and LSFO processes
are presented in Figures 10, 11, and 12 for the design coals. For all reference
coals, the economic crossover plant size increases as the capacity factor decreases.
For the 1.5 percent coal sulfur case, the economic crossover plant size increased
from 350 MWe at 65 percent plant capacity factor to about 450 MWe at 50 percent
capacity factor. The total levelized annual revenue requirement, expressed as S/ton
of S02 removed, increases from about $550 to about $590 at 65 and 50 percent
capacity factor, respectively. Similar results are observed with the 2.5 and 3.5
weight percent coal cases. The average capacity factor over the remaining plant
life is an important process selection consideration.
6B-38
-------
Short-term economic analysis is used by many utilities to determine if the long
range benefits of an option shows an economic advantage in the short-term when
unforeseen factors such as changes in regulatory environment are less likely to
occur. The results of short-term (10 year) cost analyses are presented in
Figures 13, 14, and 15 for the 1.5, 2.5, and 3.5 percent coal cases, respectively.
In the 1.5 and 2.5 percent sulfur coal cases, the economic crossover point is
increased by 50 MWe and 30 MWe, respectively. For the high-sulfur coal case, LSFO
is always favored over the Coolside process.
Variable Operating Costs
Two of the major variable operating costs for the Coolside process are delivered
cost of hydrated lime and waste disposal cost. The effect of delivered hydrated
lime cost on the differential levelized cost, expressed as $/ton of S02 removed, is
presented in Figure 16. The base case assumption is $60/ton delivered cost of
hydrated lime. Decreasing the delivered cost to $50/ton lowers Coolside annual
costs by $25 to J35/ton of S02 removed, depending on coal sulfur content. For a 2.5
weight percent sulfur coal, the base case ($60/ton hydrated lime) levelized SO,
control cost is about S488/ton of S02 removed. If the delivered hydrated lime cost
is $50/ton, then the levelized S02 control cost is lowered to $450/ton of S02
removed--a differential levelized cost of $28/ton of S02 removed.
The effect of changing waste disposal unit cost on Coolside levelized annual revenue
requirement is presented in Figure 17. The base case waste disposal cost is
J8.50/dry ton. If the waste disposal cost is S7.00/dry ton, the levelized cost is
lowered by about SlO/ton of S02 removed from the base case. If the waste disposal
cost is increased to S15/dry ton, the levelized cost is increased by $40/ton of S02
removed for the 1.5 percent sulfur case.
Lowering the LSFO FGD S02 removal requirement from 95% to 70%, then to 50%, reduces
the capital cost but the lower SO, removal increases the levelized SO, control cost
from $426/ton of S02 removed (?5%), to $512/ton of S02 removed (70%), and to
S630/ton of S02 removed (50%). Lowering the Coolside SO, removal from 70% to 50%
reduces the levelized S02 removal cost from $488/ton of 302 removed to $481/ton of
S02 removed. For non-compliance, low- to medium-sulfur coals, the Coolside process
would tend to be economically favored.
CONCLUSIONS
The Coolside process is economically competitive with an LSFO FGD process for base
load boiler operations (65% capacity factor) under the following conditions.
1. For 1.5% sulfur coal, up to 350 MWe (net).
2. For 2.5% sulfur coal, up to 130 MWe (net).
Process sensitivity analyses show that the following factors tend to favor the
Coolside process for S02 control.
1. Lower Boiler Capacity Factors--The Coolside process can be charac-
terized as a low capital cost, high operating cost process. When
compared to high capital cost, low operating cost processes like
LSFO FGD, the economic attractiveness of the Coolside process
increases with decreasing boiler capacity factor.
6B-39
-------
2. Lower Required S02 Percentage Reductions--The base case S02 removals
are 70 percent and 95 percent for the Cool side and LSFO processes,
respectively. As the S02 removal requirement decreases below 70
percent, the Coolside process becomes more economically attractive
relative to the LSFO process.
3. Shorter Remaining Boiler Life.--A shorter remaining boiler life
favors the low capital cost Coolside process.
LEGAL NOTICE/DISCLAIMER
This paper was prepared by Consolidation Coal Company (Consol) and Babcock & Wilcox
(B&U). Consol was acting under a contract with (B&W). This report was prepared in
accordance with a cooperative agreement, partially funded by the U.S. Department of
Energy (DOE), and neither B&U, nor any of its subcontractors, nor the U.S. DOE, nor
any person acting on behalf of either:
a) Hakes any warranty or representation, express or implied, with respect to the
accuracy, completeness, or usefulness of the information contained in this
paper, or that the use of any information, apparatus, method, or process
disclosed in this paper may not infringe privately-owned rights; or
b) Assumes any liabilities with respect to the use of, or for damages resulting
from the use of, any information, apparatus, method, or process disclosed in
this paper.
Reference herein to any specific commercial product, process, or service by trade
name, trademark, manufacture, or otherwise, does not necessarily constitute or imply
its endorsement, recommendation, or favoring by the U.S. DOE. The views and
opinions of authors expressed herein do not necessarily state or reflect those of
the U.S. DOE.
REFERENCES
1. Yoon, H., Stouffer, H. R., Rosenhoover, W. A., Withurn, J. A.; and Burke, F.
P., Environmental Progress. Vol. 7, No. 2, 1985, pp. 104-111.
2. Yoon, H., Stouffer, M. R., Rosenhoover, W. A., and Statnick, R. M., "Laborato-
ry and Field Development of Coolside S02 Abatement Technology," Proceedings:
Second Annual Pittsburgh Coal Conference. Pittsburgh, Pa., September 1985, p.
223.
3. Stouffer, M.R., and Rosenhoover, W. A., "Pilot Support Test for Edgewater
Coolside Demonstration. Part 2: Once-through Process Simulation Tests."
Report to B&W under DOE Cooperative Agreement No. DE-FG22-87PC79798. October
1988.
4. Yoon, H., Statnick, R. M., Withum, J. A., and McCoy, D. C., "Coolside
Desulfurization Process Demonstration at Ohio Edison Edgewater Power Station,"
Presented at 84th Air and Waste Management Association Meeting and Exhibition.
Vancouver, B.C., Canada, June 16-21, 1991.
5. Wu, M. M., Winschel, R. A., Wasson, G. E., and Jageman, T. C., "Properties of
Solid Wastes from the Edgewater Coolside and Limb Process Demonstration,"
Proceedings: 83rd Annual Meeting of the Air and Waste Management Association,
Pittsburgh, Pa., June 24-29, 1990.
6B-40
-------
6. Electric Power Research Institute. "TAG* Technical Assessment Guide, Electric
Supply," EPRI P-6587-L, Vol 1: Rev. 6, Special Report, September, 1986.
1000
900-
800-
700-
600-
500-
400
Optimized Coolside Process
Coal Sulfur As lbs SO2/IO6 Btu - £2
Plant Capacity Factor >65%
Quicklime or Hydrate Cost - S60/Ton
On Site Hydration
Purchased Hydrate
100 200 300 400
PLANT SIZE. MW (net)
500
600
Figure 1. Coolside process levelized life cycle
cost comparison for on-site hydration versus
purchase of hydrated lime.
6&41
-------
Figure 2. Hydrated lime sorbent receiving/storage/
feeding system.
VENT
DRY SODA ASH
WATER
SODA ASH
STORAGE
TO
ADDITIVE
MDCEfi
HEATER
PNEUMATIC TRUCK
Figure 3. Sodium additive (soda ash) storage
and feed system.
6B-42
-------
BOILER FLUE GASES r pv.
FROM AIR HEATERS ~ k \
RECYCLE SOLIOS
EHSTRtSLfTlON
BOTTLE
HYDRATED ume
PARTICULATE COLLECTION
ESP
ID. FAN
IN-LINE
V7VA7
ASH/LIME
DRAIN
SODIUM ADDITIVE
HUMIDIFIER
DEL UM PER
AND BLOWER
AIR COMPRESSOR
Figure 4. Coolside process humidification and recycle solids
feed system.
UMESTONE
GRMDMG BALL
MILL & HYDROCLONE
PACKAGE
JTTTn
UMESTONE STORAGE
UMESTONE RECBVMG
BOIER FLUE GAS
FROM ESP
PARTICULATE
COLLECTORS
MATER TO
MIST ELJMMATORS
BOOSTER FANS
feedtank
LATION
OVERFLOW
AM BLOWER
GYPSUM TO
DISPOSAL
ROTARY DRUM
VACUUM FILTERS
Figure 5. Wet limestone forced oxidation (LSFO) process.
6B-43
-------
350
Optimized Processes
300-
250-
Forced Oxidation
Coal Sutfur As
IDs SO2/106 Btu
——3.7
' S>2
s 150-
100-
Coolside
5 2
3.7
22
50-
600
500
400
300
Plant Size, MW (net)
200
100
Figure 6. Coolside LSFO capital cost comparison.
3
1000-
900-
>
o
e
®
ir
800-
O
CO
c
£
Optimized Processes
Coal Sulfur As lbs SC^/106 Btu - £2
Plant Capacity Factor = 65%
700-
8
O
>
ffl
600-
500-
400-I
Forced Oxidation
100 200 300 400
Plant Size, MW (net)
500
600
Figure 7. Levelized cost comparison Coolside versus
LSFO (low-sulfur coal).
6B-44
-------
700
650-
o
fc
©
600-
0c
CNJ
o
CO
550-
c
£
500-
«»
in
in
O
450-
o
T5
at
N
400-
ID
>
ID
—1
350-
300
Optimized Processes
Coal Sulfur As lbs S0^/106 Btu « 3.7
Plant Capacity Factor = 65 %
Coolside
Forced Oxidation
100 200 300 400
Plant Size, MW (net)
500
600
Figure 8. Levelized cost comparison Coolside versus
LSFO (medium-sulfur coal).
600
§
550-
b
a>
DC
500-
CM
O
CO
450-
c
£
400-
0)
to
s
350-
N
300-
ID
>
—1
250-
200
Optimized Processes
Coal Sulfur As lbs S02/106 Btu .
Plant Capacity Factor >65%
Coolside
52
Forced Oxidation
100 200 300 400
Plant Size, MW (net)
500
600
Figure 9. Levelized cost comparison Coolside versus
LSFO (high-sulfur coal).
6B-45
-------
650-i
600-
| 550-
m
CC
cm 500-
O
CO
o 450-
o 400-
350-
Coal Sulfur = 1.5 wt. %
(Z2 Its S0z/106 Btu)
Coolside Leveltzed Cost
Economic
Crossover Point
Plant Size
300-
250-
40 45 50 55 60 65
Plant Capacity Factor, %
70
75
Figure 10. Effect of plant capacity factor on
Coolside/LSFO economic plant size crossover point
and levelized costs (low-sulfur coal).
600
550
1 500
450
400
350-
!= 300-
£ 250
200
150-
100-
Coolside LeveDzed Cost
Coal Sulfur = 2.5 wL % |
(3.7 lbs S02/106 Btu) |
Economic
.
Crossover Point
Plant Size
40 45 50 55 60 65
Plant Capacity Factor, %
70
75
Figure 11. Effect of plant capacity factor on
Coolside/LSFO economic plant size crossover point
and levelized costs (medium-sulfur coal).
6B-46
-------
700
Coolside Levelized Cost
600-
Coal Sulfur = 3.5 wt %
(52. lbs SCL/106 Btu)
E 500-
CM
400-
w 300-
c 200-
Economic
Crossover Point
Plant Size
100-
75
70
65
55 60
Plant Capacity Factor, %
50
45
40
Figure 12. Effect of plant capacity factor on
Coolside/LSFO economic plant size crossover point
and levelized costs (high-sulfur coal).
Optimized Processes
Coal Sulfur As lbs SG^/106Btu = 22
Plant Capacity Factor = 65 %
Forced Oxidation
Coolside
100 200 300 400
Plant Size, MW (net)
500
600
Figure 13. Short-term (10 yr) cost comparison
Coolside versus LSFO (low-sulfur coal).
6B-47
-------
Optimized Processes
Coal Sulfur As lbs SO2/106Btu - 3.7
Plant Capacity Factor = 65 %
Coolside
Forced Oxidation
100 200 300 400
Plant Size, MW (net)
500
600
Figure 14. Short-term (10 yr) cost comparison
Coolside versus LSFO (medium-sulfur coal).
Optimized Processes
Coal Sulfur As lbs S02/i06Btu = S3
Plant Capacity Factor = 65 %
Coolside
Forced Oxidation
100 200 300 400
Plant Size, MW (net)
500
600
Figure 15. Short-term (10 yr) cost comparison
Coolside versus LSFO (high-sulfur coal).
6B-48
-------
5°1
Q
LU
S
40-
2
LLI
CC
30-
CVI
o
CO
20-
c
o
t.
«
10-
CO
K
8
0-
o
_I
-10-
<
t-
z
LU
-20-
2
LU
CC
-30-
<_)
z
-40-
CoalSuHurAs
tosSO2/10® Btu
45 50 55 60 65 70
SORBENT COSTS, $/Ton
I
75
80
Figure 16. Effect of delivered hydrate cost on Cool side
process 1 eve!ized costs.
o
LLI
>
o
2
LU
cc
eg
0
CO
1
60
40-
30-
£
CO
O
o
z
LLI
2
LU
IT
O
20-
10
-20
Coal Sulfur As
lbs SOj/106 Btu
-22
-
3.7
. X
1 1
8 10 12 14 16
WASTE DISPOSAL COSTS, $/Ton (dry)
18
Figure 17. Effect of waste disposal charge on Cool side
process 1 eve!ized costs.
6B-49
-------
Table 1
REFERENCE PLANT PERFORMANCE ASSUMPTIONS
PM CtM
MW« (nod 100 160 250 600
MWtlgien) lOS 158 262 530
Nwwrtd Pin Haat to
Without FCO. Btu/kWH (n«d
9830 9770 9510 9400
OFm
Cotdaiila
Wet fCD
• daquf
Add Booetv Fwi
ESP
EnNiOA Rata, to/10" Btu
Sovo'ic Co*«ctcr Are*. ftVlOOO at.fin
0,1
«0
SO, Eiiiiiiiu11 La*w
2.5toSO,/10" Bit*. 1995
1 2 to SO/IO" Btu. 2000
s
PlMW Ca*ae*V FMW. %
Plant Location
PUnt fUwM Facte*!
Coab^a InemndD
PGO InemMi)
05
Naar the Otw> Kmt
1.3
1.3
Table 2
DESIGN FUEL SPECIFICATIONS
Coal Sttffw. Wl % (AR)
1.50
250
3.50 U
Higher HmU)«
jj
HHV Idryl
141B0
14190
14190 1
HHV l
|
C
79.99
78.65
77.60 1
H
4.93
5.12
5.19
0
4.03
4.03
4,04
N
1.52
1,51
1.43
5
1.69
2.85
3.70
Aafl
9.04
804
9.04
Total
TOO .00
100.00
100.00
Coal Mortn, Wl. %
6,50
650
5.50
Aah Gonwt. feflO* Btu
5.67
5,07
5.67
SO, ffetarmal. to/10" Btu
2.24
3.74
6.22
Table 3
NOMINAL FLUE GAS COMPOSITION, RATE AND TEMPERATURE
Cwl %
1 5
2.5
35 1
CO*
0,
SO, toeml
6.16
12.11
5.89
912
8.35
11.96
5.86
1523
6.43 j
1180 1
5-85 I
2141 [
Boiler Sin. WWa (nad,
100
150
250
500
Pk
m Cm Rata. MSCFM
T 1
237.3
353.7
574.6
1154.3
238.8
352.9
573.4
1151.8
235.7 |
351-3
570-7
1146 4
fkj* Cm Tamperew*. *f 304
laaa than 1 pmw of coet cartoon tan to cartan in fly ah fee oee CO.
T Ar 60 "F & 1 Atrraphara. |
6B-50
-------
Table 4
UNIT COSTS: MATERIALS, UTILITIES, DISPOSAL AND LABOR
Item
Water
Lime
Hydrated Lime
Limestone
Soda Ash
Replacement Power
Fly Ash Disposal
Gypsum Disposal
Dry Sorbent Waste Disposal
Operating Labor
Unit Cost
$0.65/1000 gallons
$60.00/Ton
$60.00/Ton
$15.00/Ton
$155.00/Ton
S29.00/MW-hr
$7.00/Ton (dry)
$7.00/Ton (dry)
$8.60/Ton (dry)
$22.92/man-hr
Table 5
ECONOMIC FACTORS FOR COST ANALYSES
Base Year of Estimate
Book Life, years
Tax life, years
Discount Rate
Constant Dollar Levelizing Factors
Expenses
Capital
Construction Period, years
Cool side process
LSFO FGD
1990 Mid Year
20
15
6.1
Life Cvcle Short-Term
10 yr
1.000 1.000
0.118 0.134
2
3
6B-51
-------
Intentionally Blank Page
6B-52
-------
OHIO EDISON CLEAN GOAL PROJECTS
CIRCA: 1991
Rica E. Bolli
Ohio Edison Company
76 South Main Street
Akron, OH 44308
6B-53
-------
Intentionally Blank Page
6B-54
-------
Fp.
l
2
3
4
5
6
7
8
9
10
INDEX OF FIGURES
Project
HALT
E-SOX
CC Fuels Management
LIMB
N0XS0
SNRB
SNOX
REBURN
SORBTECH
Battelle
Title
Hydrate Utilization and Removal Efficiencies
Flow Diagram of E-SOx Process
Flow Chart of Computer Model
Flow Diagram of LIMB Process
Flow Diagram of N0XS0 Process
Flow Diagram of SNRB Process
Flow Diagram of SNOX Process
Schematic of REBURN Process
Flow Diagram of SORBTECH Process
Gypsum Recovery Process Schematic
6B-55
-------
ABSTRACT
Ohio Edison is participating in 10 clean coal demonstrations which
are in various stages of development. These projects include LIMB, E-SOx,
NOXSO, SNRB, REBURN, SNOX and SORBTECH. This paper presents current design
features and recent test results from these demonstrations. Content
emphasizes specific technology advantages/problems from the utility's
perspective. The presentation also focuses on the status of the technology
and the important attributes to consider for a utility's specific compliance
strategy. Additional R&D activities and progress summaries are presented to
encompass Ohio Edison's current SO2-related technology advancements including
FGD waste utilization and regeneration.
6B-56
-------
This paper is a review and status of each clean coal project
involving Ohio Edison and has been divided into three major categories:
• COMPLETED PROJECTS
• PROJECTS IN PROGRESS
• PROJECTS ON THE HORIZON
A brief description, technical assessment and utility advantages
have been identified for information applicable to current R&D developments.
The real benefit from investing and participating in these demonstrations is
the detailed examination of the technical and economical operating aspects of
technology not commercially available.
A figure for each technology has also been attached for immediate
reference and specific examination of the process operation and design.
COMPLETED PROJECTS
HALT - Hydrate Addition at Low Temperature
The HALT project was Edison's first clean coal project. It was
installed as a 5 MW pilot at the Toronto (Ohio) Plant.
Basically, the HALT process injects hydrated lime and then moisture
directly into the flue gas ductwork downstream of the furnace and upstream of
the dust collector. SO2 in the flue gas reacts with the hydrated lime to form
a solid, which is collected with the fly ash. Much of the HALT developmental
work led to operating parameters and nozzle design used in a number of current
duct injection processes.
The primary sponsor of the $2.1 million HALT project was the U.S.
Department of Energy. The prime contractor was the Dravo Lime Company.
Operation of the pilot unit started in November 1986 and continued
until August 1987.
Conclusions from the HALT testing confirm the approach temperature
is the single most important variable for optimum hydrate utilization and SO2
removal efficiency. The Ca/S ratio was the second most important variable for
optimizing process efficiencies and economics.
Figure 1 displays the Hydrate Utilization and removal efficiencies
at various approach temperatures. Vith operating parameters optimized, a
consistent removal of 60 percent SOj was achievable. The process also tested
the efficiency of ESP particulate collection vs. baghouse collection
downstream of the process. Both systems were successful in obtaining NSPS
performance; however, there was a slight improvement of SO2 removal with the
baghouse arrangement.
Conclusions
The HALT process is a viable process and an excellent cost
justifiable alternate for moderate SOj reduction in medium to high sulfur
coal-bumlng boilers. Final results indicate that SO2 removal of up to 60
6B-57
-------
percent was achieved with a 3.2 percent sulfur coal at a Ca:S ratio of 2.0
while maintaining optimum operating conditions. The process can be designed
and operated on a continuous basis by proper installation and selection of the
humidification nozzles. Preliminary leachate tests on the pilot disposal
solids indicate that they meet RCRA limits for non-hazardous classification.
E-SOx - Electrostatic Precipitator SOx Removal
E-SOx is a post combustion retrofit technology for controlling
sulfur dioxide emissions. It maximizes the use of existing plant facilities
making the E-SOx process a relatively simplistic retrofit installation for
low-cost, moderate SO2 control.
E-SOx emission control technology represents an enhancement of an
electrostatic precipitator (ESP) that involves reducing sulfur dioxide
emissions and removing particulates. A flow diagram of the E-SOx process is
presented in Figure 2. A lime slurry atomization system is added either
immediately preceding the ESP or in the space created by removing the
electrical internals from the first field of the ESP. Lime slurry is sprayed
concurrently with the flue gas flow and provides for sulfur dioxide removal
by reaction with calcium hydroxide. The flue gas evaporates the water in the
slurry, resulting in a relatively dry solid product at the ESP inlet.
Reaction products, unspent hydrated lime and fly ash are then collected in
the ESP.
A 5 MUe equivalent field-pilot facility was constructed at the
R. E. Burger Station of Ohio Edison to demonstrate the feasibility of the
technology on a slipstream of flue gas from a coal-fired boiler. The facility
was operated over about six months to develop process design parameters and
demonstrate acceptable levels of removal for sulfur dioxide and particulates.
From the beginning, E-SOx technology has been considered a retrofit
rather than a new plant control system. E-SOx does not provide a high level
of sulfur dioxide removal relative to some alternative control processes such
as wet scrubbing. But it does show promise as being extremely competitive
economically on the basis of dollars per ton of sulfur dioxide removed.
The E-SOx emission control concept was originally developed and
patented by the U.S Environmental Protection Agency (Sparks and Plaks, 1989).
Further development and evaluation of the concept was performed under a grant
from the Ohio Coal Development Office (0CD0) by a project team consisting of
Babcock & Wilcox (B&W), U.S. Environmental Protection Agency (EPA), Southern
Research Institute (SRI) and Ohio Edison. This development work included a
pilot scale test for sulfur dioxide removal, ESP performance, gas flow
computer model studies and pilot-scale atomizer evaluation and led to the
design and operation of a demonstration-scale test facility.
The E-SOx 5 MW pilot was constructed in 1988 for a total project
cost of $9.4 million including testing. Flue gas was first introduced through
the facility in April 1989, followed by six months of testing over an approach
temperature range of 30*F to 60*F and a range of stoichiometric ratios from
1.0 to 1.8.
Conclusions
SO2 removals of SO percent were achieved at approach temperatures of
28*F to 30*F and stoichiometric ratios of 1.3.
6B-58
-------
Hov well sulfur dioxide is removed in the E-SOx process is primarily
a function of two control parameters: approach to saturation temperature and
calcium-to-sulfur stoichiometric ratio.
The most influential operating variable is the approach temperature.
This value is a measure of the difference between the flue gas temperature at
the ESP inlet and the adiabatic saturation temperature of the flue gas. In
general, for a fixed stoichiometric ratio, sulfur dioxide removal improves
significantly as the approach temperature is reduced. However, operation
below a practical limit of 25°F to 30°F may result in unacceptable deposition
of unevaporated slurry in the spray chamber.
The economics of the process are enhanced as the stoichiometric
ratio is reduced. Operation at a lower stoichiometric ratio also reduces the
total particulate loading to the ESP. Therefore, it was desirable to achieve
the required sulfur dioxide removal at the lowest possible stoichiometric
ratio by operating at low approach temperatures.
The process design and economics provide a favorable SO2 reduction
for minimal capital costs and moderate (SO percent) SO2 removal and create a
dry product for easy handling and disposal with potential by-product
utilization.
Clean Coal Fuels Management
Ohio Edison Company, Electric Power Research Institute (EPRI),
Bechtel Group, Inc., B&W and OCDO teamed together in a Clean Coal Fuels
Management Project to investigate the technical and economic feasibility of a
novel coal cleaning system that combines existing technologies (conventional
cleaning methods and an atmospheric fluidized bed boiler).
The system investigated utilizing an advanced coal cleaning system
to produce two levels of coal quality: a deep-cleaned, low sulfur, high
quality coal, and a high sulfur, poor quality coal rejected in the deep-
cleaning process. The low sulfur, high quality coal is lower in sulfur than
can be achieved economically in present coal cleaning plants. This fuel was
used by an existing boiler without SO2 emission control equipment. The high
sulfur, poor quality coal was sent to an atmospheric fluidized bed boiler.
One of the products of the $1.1 million research project is a
computer model that may be used by the State of Ohio, electric companies,
coal producers and others to evaluate fuels management for their specific
conditions.
Figure 3 represents a flow chart of the input coal and the three
product possibilities as a result of the cleaning process. The computer
program will calculate the percentages of the three products for any specific
fuel the utility selects to enter, which allows the flexibility of predicting
for current and future coal supplies.
This project was completed in 1990 and the computer program is
available from EPRI. This project was the only Ohio Edison clean coal
technology that is categorized as a pre-combustion process.
LIMB/Coolside - Limestone Injection Multi-Stage Burner
A $47 million, full-scale demonstration of the LIMB technology was
6B-59
-------
conducted an Ohio Edison's Edgevater Plant in Lorain on a 104 MW boiler
burning a nominal 3 percent sulfur coal.
LIMB involves upper furnace injection of hydrated calcitic line,
coupled with the use of low-NOx burners. The injected dry sorbent mixes with
the boiler combustion gases and chemically removes the sulfur dioxide. The
by-product then travels through the remainder of the boiler and is removed
along with fly ash in the unit's existing dust collector.
Developed as a technology aimed at moderate levels of SO2 and NOx
control for relatively low-cost retrofit applications, LIMB operation at a
Ca/S ratio of 2.0 was capable of achieving 55 to 72 percent SO2 removal. The
removal was dependent on the specific sorbent utilized and the degree of
humidification employed. In conjunction with SO2 removal, the performance of
the XCL low-NOx burners was evaluated with an average emission rate of 0.48
lb. per mmBtu observed.
The Base LIMB program began in July 1987. The long-term operation
of the LIMB process with humidification, a second phase of the project, began
in September 1988 and continued for approximately 10 months. A process flow
diagram of the complete LIMB/humidifier system is provided in Figure 4.
The results of this testing included the use of a commercial,
hydrated calcitic lime treated with calcium lignosulfonate, an additive that
appeared to improve SO2 removal. Testing indicates that the modified sorbent
increased SO2 removal from 55 to 63 percent with minimal humidification, and
from 65 to 72 percent at close approach to saturation, when operating at a
Ca/S ratio of 2.0.
Since the SO2 removal was particularly dependent upon the
temperature at the injection point, particular attention was focused on
operating variables that affected parametric optimization. These variables
included injection at different elevations in the furnace, the momentum flux
ratio (injection velocity and furnace penetration at a given load), the angle
of injection (nozzle tilt), and boiler load. The results show that these
parameters have little effect on SO2 removal in the Edgewater boiler over most
of the ranges tested which indicate that the system is insensitive to minor
changes if the initial design parameters enable near-optimum operation.
The project also demonstrated that SO2 removal could be enhanced by
humidification of the flue gas. Humidification increased the SO2 removal from
55 to 65 percent at a Ca/S ratio of 2.0.
For NOx reduction, the existing circular burners were replaced with
B&W's XCL burners as part of the demonstration project. Baseline data
indicated that 0.7 to 0.9 lb. per mmBtu NOx was reduced to an average of 0.48
lb. per mmBtu with unburned carbon averaging 1.54 percent. There appeared to
be no interactive effects between sorbent injection and NOx reduction.
A third phase of the LIMB project was made possible by the U.S. DOE
Clean Coal Technology Program, as a Clean Coal I project. Coolside is a low-
cost retrofit technology similar to HALT. Data from this earlier project was
used in the design.
6B-60
-------
Dry lime is injected into the flue gas duct after the gas leaves the
boiler. Then the flue gas is humidified with a water spray containing
chemical additives to enhance SO2 adsorption and electrostatic precipitator
performance. Coolside testing started in July 1989 and was completed in
February 1990. SO2 removal of up to 70 percent was achieved.
The final phase of the program which involved testing different
combinations of three coals and four sorbents was completed in August of this
year.
This project was sponsored and funded by Ohio Edison, U.S. EPA,
U.S. DOE, OCDO, B&W, Consolidation Coal, Radian and Stone & Vebster.
Conclusions
The LIMB process impacts on boiler and plant operations are related
primarily to the increased quantity of particulate matter that must flow
through the boiler, ESP and ash handling equipment. The need for effective
soot blowing is the single most important requirement when considering
application of the technology.
LIMB technology is a process which requires low capital investment
and is easily retrofitted to existing boilers. Although the removal
efficiencies for SO2 are lower than conventional and advanced scrubbers, the
cost per unit of sulfur removed is much lower than for scrubbers.
PROJECTS IN PROGRESS
The next generation of clean coal projects which are currently in
progress at Ohio Edison represent a generic principle of high removal
efficiencies of both SO2 and NOx. Coupled with this shared objective is the
integration of little or no waste product generated. These attributes signify
an economical emission control process when compared to commercially available
scrubbers in combination with a NOx control technology plus the potential for
a marketable by-product. Although the programs in progress are too premature
to quantify the removal costs, the predictions of removal efficiencies and
costs of equipment in place indicate viable alternatives to scrubbing systems
and offer uniquely attractive processes for compliance with the Clean Air Act
Amendments of 1990.
NOXSO
The NOXSO process is a dry flue gas treatment system that employs a
reusable sorbent. The sorbent consists of sodium impregnated on a high-
surface -area alumina. Flue gas exiting the ESP or baghouse is directed
through a fluidized bed of sorbent which simultaneously removes SO2 and NOx
from flue gas. The spent sorbent is regenerated for reuse by treatment at
high temperature with a reducing gas. This regeneration reduces sorbed sulfur
compounds to SO2. H2S and elemental sulfur. The SO2 and H2S are then
converted to elemental sulfur in a Claus-type reactor. The sulfur produced is
a marketable by-product of the process. Adsorbed NOx is decomposed and
evolved on heating the sorbent to regeneration temperature. Regeneration of
active NOx sorption sites is accomplished simply by heating the sorbent. The
concentrated stream of NOx produced is returned to the boiler with the
6B-61
-------
combustion air. Figure 5 represents a diagram of the project.
Construction of a 5 MW Proof-of-Concept project began in July 1990
at Ohio Edison's Toronto Plant. The unit is designed to treat approximately
12,000 SCJM from either of two boilers burning approximately 3.5 percent
sulfur coal. The process is designed to simultaneously remove 90 percent of
the SO2 and NOx in the flue gas. The regeneration process to reuse the
sorbent material results in no waste products being formed, and the potential
of improving the thermal efficiency of the boiler operation by returning hot
gases laden with NOx as combustion air, bypassing the air preheater.
The $9.5 million project is sponsored by U.S. DOE, OCDO, NOXSO,
MK-Ferguson, W. R. Grace and Ohio Edison. Start-up activities have commenced
with operating and design data being utilized to design and construct a full-
scale demonstration of the NOXSO process.
The full-scale demonstration was selected in Round III of the U.S.
DOE's Clean Coal Program and is currently being designed for installation at
Ohio Edison's Niles Plant.
This project will provide the technical and economic data for a
commercial installation along with availability and reliability information.
The full-scale demonstration will be fully integrated into a cyclone-fired 108
MW boiler and will include two additional features that the pilot project will
not be testing. These include the Claus Plant for sulfur recovery and the NOx
recycle to the boiler furnace.
This $66 million project includes all of the pilot plant co-sponsors and
additional participation by EPRI, the Gas Research Institute (GRI) and East
Ohio Gas. Engineering and procurement are scheduled for 1992 with
construction to commence in 1993, followed by two years of testing and
operation.
The attractive advantages to the utility operator in addition to the
high removal efficiencies and lower costs than current commercial applications
with similar emission reduction capabilities are the small footprint due to
the tower design and the installation downstream of the utility equipment
which minimizes changes to the current boiler operation. The other
significant advantage is the fact that no waste products are produced. Due to
the current and future water and waste regulations, the economic impact of
eliminating all disposal costs while producing a saleable product results in a
win-win situation.
SNRB - SOx-NOx-ROx-BOx
The SOx-NOx-ROx-BOx (SNRB) process, developed by B&W, is an advanced
emission control process for the combined removal of SO2, NOx and particulates
from coal-fired boilers or processes.
The key to the SfiRB process is a high-temperature baghouse in which
simultaneous SOx, NOx and particulate removal occurs. SO2 removal is
accomplished by injecting a dry sorbent such as hydrated lime or sodium
bicarbonate into the flue gas. NOx removal is accomplished in part by ammonia
injection with a selective NOx reduction catalyst. Finally, the particulates
and spent SOx sorbent are collected in a high-temperature baghouse. Figure 6
provides a flow diagram of the project.
6B-62
-------
Construction of the 5 MW pilot project began in March 1991 at Ohio
Edison's R. E. Burger Plant. The unit is designed to treat a slipstream from
a coal-fired PC boiler burning a nominal 2.5 percent sulfur coal. Earlier
tests have provided the project goals of 70-75 percent SO2 removal at
stoichiometric ratios of 2.0 to 2.5 and 90 percent NOx reduction at cost-
effective NH3/NOX stoichiometric ratios. Greater than 99.9 percent
particulate removal is expected in the pulse-jet baghouse which utilizes a
unique design of bag, bag cage and catalyst retainer. Removal efficiencies
will be optimized at a baghouse operating temperature range of 700*F to
850"F.
The $11.4 million project was selected in Round II of the U.S. DOE's
Clean Coal Program. Sponsors of the project include B&W, U.S. DOE, OCDO,
EPRI, Norton, 3M and Ohio Edison. Start-up and shakedown activities have
commenced and parametric testing and operation is scheduled for 1992.
Besides providing the technical and operating data for a full-scale
application, this project will provide commercial readiness of the technology
because the baghouse utilizes commercial size bags (approximately 6-1/2"
diameter, 20' length).
The economic attractiveness of this project is exemplified by the
use of a single vessel for removals of all three pollutants. Additional
advantages include simplification of operation due to minimal equipment
requirement (low man-hours required for operation) and minimal dry waste
product that is easily transported and has potential for by-product
utilization. An additional potential benefit is the ability to operate the
air preheater at a lower flue gas outlet temperature, thus improving heat
recovery and boiler thermal efficiency.
SNOX - Wet Sulfur Acid from SO2 and NOx Reduction
The SNOX technology is a catalytic removal process capable of
removing 95 percent of the SO2 and 90 percent of the NOx from a coal-fired
boiler. Flue gas upstream of the ESP is first processed through a bag filter
for removal of the fly ash, then heated in an exchanger by the exiting gas
stream to the required reaction temperature. A small amount of ammonia is
added and the mixture is then processed through a NOx SCR for conversion of
nitrogen oxides to nitrogen and water vapor.
Gas exiting the NOx reactor is heated further and then processed
through the SO2 reactor, in which SO2 is converted to SO3. The gas exiting
the SO2 converter is heat exchanged and then passed to a condensing tower, the
key component of this technology. In the condensing tower, the gases are
cooled to produce a high-concentration, commercial grade, sulfuric acid.
Figure 7 provides a flow diagram of the SNOX process.
The SNOX project was selected in Round II of the U.S. DOE's Clean
Coal Program. This $31.5 million project is the first domestic installation
of a technology developed "in Denmark and currently being tested in a project
there and on an additional installation in Italy. Project sponsors in
addition to the U.S. DOE include OCDO, Combustion Engineering, Haldor Topsoe,
Snamprogetti and Ohio Edison.
6B-63
-------
Construction on this project began In March of this year with start-
up activities commencing in December 1991 followed by an 18-month testing and
operation phase. The SNOX project is treating approximately one-third of the
flue gases from one of Ohio Edison's Niles Plant 108 MU cyclone-fired boilers,
or Che equivalent of a 35 MW demonstration.
This demonstration will validate Che removal efficiencies and
economic advantages of a domestic installation utilizing high sulfur coal.
The advantages include high removal efficiencies without producing any waste
products. Because of the exothermic properties and potential integration of
recovered heat to the utility boiler, the process offers significantly lower
O&M costs than commercially available technologies. The economics of the
process are also enhanced with the production of a saleable by-product, highly
concentrated sulfuric acid. Another distinct advantage is the location of the
SC>2 converter downstream of the SCR which eliminates ammonia slip
considerations allowing for maximum NOx removal capabilities. In addition,
the process reduces CO and hydrocarbon emissions and actually improves
efficiency with increasing sulfur content of the coal.
REBURN
An additional project at Ohio Edison's Niles Plant is the REBURN
project, which is a NOx reducing, in-furnace technology utilizing natural gas
as the rebum fuel.
REBURN technology involves creating a second combustion or "reburn"
zone downstream from the main burners in a boiler. Combustion gases that
result from burning a fossil fuel in the main combustion zone, move to the
"reburn" zone where additional fuel, in this case natural gas, is injected.
The injection of additional fuel creates a fuel-rich zone in which the NOx
formed in the main combustion zone are converted to molecular nitrogen and
water vapor which occur naturally in the atmosphere. Any unburned fuel
leaving the reburn zone is subsequently burned to completion in a downstream
burnout zone where additional air is injected. Rebuming is especially
attractive for cyclone-fired boilers and other wet-bottom boilers since low-
NOx burners and most other low-NOx combustion technologies used on
conventional boilers are not applicable to cyclone-fired and wet-bottom
boilers. The overall goal of the program is to successfully demonstrate a 50
percent reduction in NOx emissions from a cyclone-fired boiler employing
rebuming technology. Figure 8 shows a schematic of the REBURN project.
Project participants include Combustion Engineering as the project
manager and main contractor, along with U.S. DOE, U.S. EPA, GR1, EPRI, OCDO,
Ohio Edison and the East Ohio Gas Company.
Mobilization for this $10.3 million project occurred in March 1990,
with boiler modifications completed during a planned boiler maintenance outage
in June. Parametric testing began in late 1990 where testing results
indicated a 60 percent reduction in NOx emissions.
Although the initial data was extremely promising, an unexpected
phenomena was occurring in the boiler. Ash deposition had increased
significantly along the rear wall starting at the gas injection location and
continuing to higher elevations up to the slope wall of the boiler. This was
theorized as a result of the recirculated flue gas which was used as a mixing
6B-64
-------
medium for the natural gas and to keep the upper fuel injectors (UFI's) cool.
The cooling effect on the rear wall was allowing the normal ash thickness to
increase from 4-6" to 12-18".
Combustion Engineering redesigned the natural gas injectors by
fabricating a water-cooled, injection nozzle, which eliminates the
recirculation of flue gas and reduces the boiler penetrations to approximately
6" diameter circles. This application reduces the number of boiler tubes
requiring modification and eliminates the operation of a flue gas
recirculating fan.
This modification was completed during a scheduled boiler
maintenance outage starting in September of this year. Parametric testing
has commenced to examine the NOx reduction utilizing the new natural gas
injectors.
Advantages to this technology include a design that is the only
feasible means for in-furnace NOx reduction for a cyclone boiler. However,
the process is applicable for NOx reduction in a PC-fired boiler also.
The other striking advantages to the utility operator is the ease of
installation, the ability for quick on/off operation and the proven advantage
to operate the system without additional manpower requirements.
SORBTECH
Another combined removal technology at Ohio Edison is the
Mag*Sorbent Technology developed by SORBTECH (formerly known as Sanitech).
Mag*Sorbent is a dry, regenerable process utilizing a sorbent material capable
of removing 90 percent of the SO2 and 40 percent of the NOx from flue gas
emissions. The sorbent is comprised of two inexpensive industrial minerals,
magnesia and vermiculite.
The process involves a radial panel-bed adsorber which is
retrofitted to the utility downstream of the particulate control device. The
gases pass through a bed of approximately 12" of the sorbent material where
the NOx and SO2 are simultaneously removed. The sorbent is regenerated in a
heating process where the SO2 and NOx are driven off, and the sorbent is
screened to remove the fine particles and returned as make-up to the adsorber.
The Hag*Sorbents exhibit very high utilizations and need only to be
regenerated 10 times for attractive process economics. There are no waste
products generated by the process, and the spent sorbent has potential for
by-product utilization such as soil enhancement or developments are occurring
for possible regeneration of the sorbent constituents.
A 2.5 MW pilot demonstration was constructed at Ohio Edison's
Edgevater Plant this summer. This is a scale-up from previous testing that
was performed at Ohio Edison's Gorge Plant. This $700,000 project is co-
sponsored by OCDO, SORBTECH and Ohio Edison. Start-up and shakedown of the
equipment is scheduled to'begin before the end of the year followed by a six-
month testing program. Figure 9 shows a flow diagram of the project.
Advantages include the simultaneous high removal efficiency of the
SO2 combined with NOx removal, without creating a waste product. The results
of this project will validate the process economics which appear to be about
one-half the cost of commercially available equipment with comparable removal
6B-65
-------
races. The simplistic process and equipment Integrated downs cream of the
utility's equipment provide the potential for low capital costs, low operating
costs and ease of retrofit installation.
PROJECTS ON THE HORIZON
Land Application for Dry FGD Bv-Product
Air quality control regulations for sulfur dioxide have spurred the
development of several dry FGD processes suitable for retrofit on moderate-to-
high sulfur coal burning utility generating units. These processes may
generate enormous quantities of dry solid by-product material consisting
primarily of calcium-based excess sorbent, reaction products containing
sulfates/sulfites and fly ash. The major
-------
second-stage reactor to produce a regenerated solid calcium oxide and a gas
stream rich in sulfur dioxide. This calcium oxide can be collected in a
cyclone, while the product gas is used to produce sulfuric acid. Elemental
sulfur can be produced by use of the Claus process (or variations thereof).
A program has been funded by OCDO, GRI, Ohio Edison, Columbia Gas
and Consolidated Natural Gas to advance this technology toward
commercialization. Battelle will be a subcontractor and will perform the
majority of the experimental effort. A vital element in this commercial step
is an A&E organization that can contribute to the transition from laboratory
to pilot plant demonstration and ultimately design the commercial plant.
The overall project objective is to advance this technology toward
commercialization. The specific project objective is to generate the data
necessary to design a demonstration plant (50 to 100 tons/day) that will
establish commercial feasibility of the FGD gypsum recovery process and be
used as a large-scale development process. This large-scale developmental
facility may also be used for expanded applications of the technology.
The process being developed at Battelle for the recovery of chemical
values from by-product gypsum consists of two separate reactors used first to
reduce the gypsum to calcium sulfide (CaS) and then in the second reactor, to
roast the resulting sulfide with air to produce a gas stream rich in sulfur
dioxide (SO2) which can be used to generate elemental sulfur or sulfuric acid
and regenerated solid calcium oxide (CaO). The potential application of the
Battelle FGD lime recovery process eliminates some of the problems inherent
with other processes to convert gypsum.
The process is shown schematically in Figure 10. Wet gypsum from
FGD systems, waste acid neutralization, or storage is calcined using waste
heat from incineration of the Stage 1 reactor off gas and fed into the
fluidized bed Stage 1 reactor.
In Stage 2, the concentrations are suitable for conversion to
elemental sulfur or for conversion to SO3 as the first step in sulfuric acid
manufacture.
Summary
All of these projects share the goal of bringing promising new
technologies closer to commercialization.
There are only two options, today, in reducing emissions of sulfur
dioxide from coal-fired power plants -- expensive, complex scrubbers, and
switching to lower sulfur fuels.
Managers of businesses don't like constraints on their choices when
it comes to serving their customers. Electric companies are no exception, and
clean coal technologies allow flexibility in selecting options for the future.
Ohio Edison wants to help realize the potential of these new technologies,
help the Ohio coal industry and provide a medium to bring premier technologies
closer to commercialization.
(jB-6'7
-------
ACKNOPT-FTViEMEHTS
Rather Chan list specific individuals, I have alphabetically listed
all the organizations that have contributed or are contributing to any of the
clean coal projects/programs that involve Ohio Edison. These companies have
an integral role in executing the projects and have, therefore, contributed Co
Che development of this paper.
ABB Combustion Engineering
ABB Environmental Syscems
Babcock & Wilcox
Battelle
Bechtel Group
Columbia Gas
Consolidated Natural Gas
Consolidation Coal
Dravo Lime
Ease Ohio Gas
Electric Power Research Institute
Gas Research Institute
Haldor Topsoe
Minnesota Mining & Manufacturing
MK-Ferguson
Norton
NOXSO
Ohio Coal Development Office
Ohio State University
Radian
Snamprogetti
SORBTECH
Southern Research Institute
Stone & Webster
U.S. Department of Energy
U.S. ^PA
W. R. Grace
6B-68
-------
halt project
Hydrate Utilization
04
Average Inlet Conditions
15227 ACFM
333 OEG F
2167 PPM SO.
Oa
10
3D
00
(«0
O Ga/Zi 1 JO - 2.19
Removal Efficiency
a•
as
OA
to
30
80
O I inm Ga/ii 2JOZ
Figure 1
6B-69
-------
ORIGINAL
ESP
( 1 1
SPRAY MULTISTAGE
CHAMBER ESP r-ol p nPE
wonn |—1 if
SPRAYER
-1 _ PRECHARGERS
CONVENTIONAL
PULVERIZED
coai. kmh»
FLUE OAS
TO STACK-
LJME-
IRES AND PLATES
REMOVED
SLURRY
UME STORAOE.
HANDLING.
ANDSLURRY
PREPARATION
RECYCLE-
OPTIONAL
TO LANDFILL
Schematic of E-SO* process.
Figure 2
CLEAN CUAL FUELS HANAGEICNT
COAL CLEANING PROCESS
Figure 3
6B-70
-------
Hum
-------
OAS
m
OAS
FUlE
FAN
FLUE OAS HEAT
HWVW UMT
^Yr^Jmnj
9LO
TO
Schematic of tfc* 5-IMa SKKB ri«ld D«on»fcg»tioii Facility
Figure 6
The SNOX Project
ELECTROSTATIC
precipitator
COOLING
AIR
ASH
FLUE
GAS
CLEAN FLUE GAS
AIR
CONDENSER
CASTAS
HEAT
exchanger
SULFURIC ACID
COAL
FABRIC FILTER
uJ
SCR
REACTOR
BOOSTER
FAN
SOa
CON-
VERTER
ACIO
STORAGE
TANK
AMMONIA
ASH
SUPPORT
BURNER
ASH
Figure 7
6B-72
-------
UF1-UPPER FUEL
INJECTOR
un'
LOCATION
CAS / t—
RECIRCULATION
FAN
SCHEMATIC OF REBURN PROCESS
Figure 8
6B-73
-------
The SORBTECH Project
AIR
rf
COAL
BOILER
ELECTROSTATIC
PRECIPITATOR
FLUE GAS
SULFUR
CONDENSER
ASH
ELEMENTAL
SULFUR
AFTER-
BURNER
O
CLEAN
l^gas
STACK
HUMIDIFlCATlON
MAKE-UP
SORBENT
SPENT
SORBENT
__J-EaT)
Y REGENERATOR
-Or-
SO' -| ^ H | RADIAL
p°££-
BOOSTER FAN Pllj SORBER
SORBENT
BOOSTER FAN
SORBENT
SCREEN SIZING
FRESH SORBENT
Figure 9
6B-74
-------
Battelle's Gypsum Recovery Process Schematic
Heat
Recovery
Gypsum
|{||| Cyclone r
Cyclone
Figure 10
6B-75
-------
Intentionally Blank Page
6B-76
-------
Sanitech's 2.5-MWe Magnesia Dry-Scrubbing
Demonstration Project
6B-77
-------
Intentionally Blank Page
6B-78
-------
S. G. Nelson
Sanitech Inc.
1935 E. Aurora Road
Twinsburg, Ohio 44087
ABSTRACT
The performance of a new regenerable sorbent, developed by Sanitech
Inc., is being demonstrated in a 2.5-MWe pilot plant installation at
Ohio Edison's Edgewater power plant in Lorain, Ohio. The granular
sorbents, called "Mag*Sorbents," are 50 weight-percent magnesia and
50 weight-percent exfoliated vermiculite. A special heat-treatment
step endows the sorbents with high sorption capacities. In earlier,
smaller, pilot facilities, the new dry sorbents demonstrated
excellent combined SOz, NOx, and residua1-particulate removal rates.
During thermal regeneration with natural gas in the atmosphere,
captured NOx is converted to nitrogen and water, and the sorbed SOz
comes off as a concentrated stream of elemental sulfur, S02, and
H2S, for by-product production.
The objective of the present 2.5-MWe pilot plant program is to
duplicate the previous high S02 removal rates and high sorbent
utilizations at a larger scale, while demonstrating life-cycle
process operation, with sorbent cycling continuously between the
sorption said regeneration steps. The project is designed to
accumulate the data necessary for a full 100-MW facility
installation. To date, the Mag*Sorbent pilot plant has been
designed said installed. It is currently undergoing shakedown
testing.
6B-79
-------
INTRODUCTION
Sanitech, Inc. is scaling-up a unique, dry, regenerable acid-rain
control technology for low-cost retrofit utility applications. This
"Mag*Sorbentn process utilizes granular sorbents made from two
inexpensive industrial minerals, magnesia (MgO) and vermiculite. In
Sanitech's process, MgO is coated onto individual vermiculite
exfoliates, which have been thermally expanded into small accordion-
like structures. The Hag*Sorbents cam be loaded to about 50 percent
MgO by weight, so that process components are small and only low
quantities of materials are circulated through the system. A
special heat treatment before their use that makes the sorbents very
reactive toward S02 and NOx.
A special radial panel-bed filter is employed in the new process.
In this filter, which is retrofitted before a utility's smokestack,
dry sorbent in the form of a slowly-moving panel-bed removes greater
than 90 percent of the flue gas S02, compliance levels of NOx, and
much of the fine, residual particulates that pass through an
existing electrostatic precipitator, all in one unit. The
Mag*Sorbents exhibit very high utilizations and are thermally
regenerated typically five to ten times, which results in very
attractive process economics.1 Because the system is regenerable,
there are no wastes to dispose of, only marketable by-products.
PROCESS DEVELOPMENT
Sanitech began its work with the new sorbents in 1985. Since that
time, it has carried on research and development programs with the
assistance of the U.S. Environmental Protection Agency, the U.S.
Department of Energy, and the Ohio Coal Development Office,
advancing the technology.
6B-80
-------
A major development occurred in 1987 when Sanitech researchers
discovered that the new sorbents, after becoming saturated with SOz
and NOx, could be thoroughly regenerated by simply heating the
materials in a slightly-reducing environment. During regeneration,
the sorbed NOx is converted to nitrogen said water, and much of the
released S02 comes off as elemental sulfur.
Since 1987, Sanitech has been scaling up the Mag*Sorbent technology
in steps. During 1988, it demonstrated combined SOz and NOx removal
in a small panel-bed filter on a 0.02-MWe slipstream of flue gas at
Ohio Edison's Gorge power plant. During 1989, high SOz and NOx
removals were achieved on a 0.1-MWe slipstream.2 In early 1990, the
process was demonstrated at the 0.5-MWe level, with sorbent cycled
between multiple sorption and regeneration steps.3
In recent months the process has been scaled up one step further.
Equipment has been designed, constructed, and installed at Ohio
Edison's Edgewater power station in Lorain, Ohio to treat a 2.5-MWe
slipstream of flue gas.
SPECIAL SORBENT CHARACTERISTICS
Magnesia in the presence of moisture readily reacts with S02 to form
MgSOj and MgS04- The MgO Wet Scrubbing Process, employed, for
example, at Philadelphia Electric's Eddystone power plant, employs
magnesia in the form of a wet Mg(OH)2 slurry.4 Magnesia and
magnesium hydroxide can also sorb S02 in a more convenient dry form,
but they react at a slower rate.
Sanitech engineers discovered a way to increase the reaction
rate of dry magnesia, while at the same time increasing the
SOz-sorption capacity of the material. They achieved this by
(1) coating the magnesia onto individual expanded vermiculite
granules, and (2) heating the combinations to 550°C in air. These
procedures result in a sorbent with a large MgO surface area.
Figure 1 shows electron photomicrographs of the surface of a typical
Mag*Sorbent granule. At about 500°C# the already small magnesium
hydroxide crystals on the vermiculite surfaces were found to
6B-81
-------
recrystallize to magnesium oxide structures with larger active
surface areas and advantageous pore size distributions. These
recrystallized structures can be seen in the photomicrographs.
Kent State University surface chemists identified two mechanisms as
responsible for the higher-than-expected utilizations of the
sorbents.5 The two mechanisms are:
1. The normally-expected chemical reaction between MgO
and SOz, which principally forms MgSOs;
2. A physical phenomenon known as "capillary micropore
condensation," where SOz complexes are physically
captured and held by capillary forces within the
micropores of the sorbent structure.
This second mechanism boosts sorbent utilization each cycle,
resulting in lower process costs.
THE 2.5-MWe EDGEWATER PILOT PLANT PROJECT
A goal of the 2.5-MWe pilot-plant program currently underway is to
demonsrrate that the high S02-removal efficiencies and the high
sorbent utilizations observed in earlier small pilot-plant
facilities can be achieved at a large scale. A further goal is to
demonstrate the new technology as a continuous process, where
sorbent is continually cycled between sorption and regeneration
steps. The project is designed to accumulate the data necessary for
a full 100—MW facility.
The regenerable Mag*Sorbent technology is a straight-forward
process. The flue gas simply passes through a thin panel of
sorbent, then exits to the stack. The pilot plant consists of two
principal circuits, a gas sorption circuit and a sorbent
regeneration circuit, as shown in Figure 2.
6B-82
-------
Three major components make up the sorption circuit: the sorber, a
humidification spray, and a fan. The sorber, shown in Figure 3
before its installation, is a vertical cylindrical vessel through
which a radial panel of sorbent flows slowly and continuously as a
12-inch-thick bed. The bed is supported between two stainless steel
screens or louvered faces. Fresh sorbent is fed to the top of the
sorber from hoppers above the unit. Saturated sorbent is removed
continuously from the sorber discharge. The flue gas enters through
the top-center of the vessel, radiates out through the radial panel
of sorbent, and is channelled in the outer chamber to an exit duct
and fan. Because the sorbents are granular and the sorbent panel is
thin, pressure drops are low. This general sorber design has been
used at the boiler-scale for years in the pulp and paper industry,
among others.6 There, limestone chips are used instead of granular
sorbents, because the objective is simply to take out particulates.
The radial panel-bed design is very space-efficient, with a small
retrofit footprint. A scaled-up design is shown in Figure 4.
The low-temperature chemistry of the process is advantageous for
ease-of-retrofit. Humidification of the flue gas is accomplished by
simply spraying water into the flue gas approximately 20 feet in
front of the sorber. The added moisture also decreases the gas
temperature, which improves SOz removal. However, an approach to
saturation of at least 50 F degrees is maintained, to avoid any
corrosion problems. The fan in the sorption circuit is employed
principally to regulate the gas flow through the system.
The regeneration circuit contains three principal components:
the regenerator, a screening unit, and a condenser-burner system.
The regenerator design selected for the Edgewater pilot plant has
performed well in the past, although more efficient heat-transfer
designs will probably be used in larger, commercial plants. The
regenerator at Edgewater consists of two parts, a enclosed,
rectangular, horizontal electric kiln and a continuous belt conveyor
that passes through this kiln and carries the sorbent. The kiln is
maintained at about 600°C. Regeneration can be carried out in air
or in a controlled, reducing environment.
6B-83
-------
A screening station is employed after the regenerator to remove
spent sorbent from the system during each regeneration cycle.
Typically, 10 percent of the sorbent is expected to be removed with
each pass. The removed materials include the smallest granules in
the sorbent stream, including any fines that are produced through
attrition. An addition of fresh make-up Mag*Sorbent is made to
replace the removed material.
The condenser-burner system treats the off-gas from the regenerator
when a reducing atmosphere is employed in regeneration. When air is
employed in the regenerator, the
off-gas typically consists of S02, NOx, nitrogen, oxygen, and HzO.
In the pilot plant, this off-gas is simply reintroduced to
Edgewater's main flue-gas duct. In a full-scale plant, the
concentrated off-gas would be processed into sulfur products. When
a reducing-gas environment is employed in the pilot regenerator, the
off-gas includes copious amounts of elemental sulfur and H2S. The
condenser is used to collect the sulfur in solid form and the burner
is used to convert the H2S back to S02 before the off-gas is
released to the main flue-gas duct.
DEMONSTRATION TEST PLAN
The tests at Edgewater are being performed in four phases:
Phase 1. The first phase consists of equipment shakedown tests
involving trials with individual equipment pieces making up the two
circuits. Included in the shakedown tests are runs with both dry
flue gas and humidified gas and regeneration trials with both an air
atmosphere and a reducing atmosphere in the regenerator.
Phase 2. Parametric studies are being carried out in Phase 2. The
effects of changes in the following variables on sorption
performance are being evaluated:
1. Flue-gas face and space velocities
2. Sorbent flow rate
3. Approach to adiabatic saturation
6B-84
-------
4. Flue-gas temperature
5. S02 concentration in the flue gas
6. Sorbent composition
7. Sorber design.
Concurrent with the sorption performance studies, evaluations are
being performed on how changes in the degree of sorbent saturation,
sorbent processing rate, regeneration atmosphere, and regeneration
temperature affect regeneration performance.
Phase 3. Upon completion of the parametric studies, a series of
cycling runs will be performed. These runs are designed to
demonstrate short-term, integrated sorption-regeneration operation
of the system. The conditions and procedures that are found most
favorable in the parametric studies will be employed.
Phase 4. A number of longer-term continuous runs, covering several
days to several weeks of continuous operation, will also be
performed. These runs will be operated at steady-state conditions
to collect the operating data needed to scale-up the technology to
the 50 to 100-MWe utility level.
Once a significant amount of spent so^rbents have been generated,
Premier Services Corp., the leading U.S. magnesia producer, will
assist Sanitech in evaluating the potential of these materials as
commercial by-products. The spent magnesia and vermiculite hold
promise in soil conditioning and fertilizer markets, acid drainage
neutralization, and as well as other value-added uses.
Based on the experience and data generated at the Edgewater
facility, Sanitech will complete a full economic evaluation of the
Mag*Sorbent technology at the end of the project.
6B-85
-------
ACKNOWLEDGMENT
Sanitech wishes to thank the Ohio Coal Development Office for
co-funding this pilot-plant demonstration and the Ohio Edison
Company for providing the host site.
REFERENCES
1. S. G. Nelson. "The Economics of a New Regenerable Sorbent
Process," Comparative Economics of Clean Coal Technologies
Conference, Washington, D.C., March, 1990.
2. Sanitech. "Development of a New Sorbent," Ohio Coal
Development Office Final Report, September 1989.
3. Sanitech. "SOz/NOx Control System for Coal-Fired Boilers,"
Final Report, DOE Contract DE-AC02-89ER80689, March, 1990.
4. R. Bitsko, et al. "Regenerative FGD System Proves Viable,"
Electric World, p.44, December, 1985.
5. R. J. Ruch, et al. "The Physical Nature and Chemical
Reactivity of a Heterogeneous MgO/Vermiculite Flue-Gas
Sorbent," American Chemical Society Annual Meeting, Washington,
D.C., August, 1990.
6. D. Parquet. "The Electroscrubber Filter: Applications and
Particulate Collection Performance," Third Symposium on the
Transfer and Utilization of Particulate Control Technology,
Orlando, Florida, March, 1981.
6B-86
-------
Figure 1. Electron Photomicrographs of a Typical Sorbent Surface
at 10,000X and 40,000X Magnifications
6B-87
-------
0\
f
T1
/ AFTER
BURNER
CONCENTRATED SO?
ROTERY
FEEDER SATURATED SORBENT
REGENERATED
SORBENT
STORAGE
CONDENSER
SURGE BIN
STORAGE
REGENERATION
ROTERY
FEEDER
GAS OUT
SCREEN
OTERt
EEDER
REGENERATOR
F NES
H ROTERY
vY feeder
MAKE UP
SORBENT
OHIO EDISON
FLUE GAS DUCT
Figure 2. Sanitech/OCDO Pilot Plant
-------
Figure 3. Photograph of Panel-Bed Sorber
6B-89
-------
A-A '
GAS OUT
SORBENT
GAS ] N
flue r—s^.
GAS IN
RADIAL PANEL-BED
SORBER
PLANT SIZE 100 rw,
FACE VELOCITY 2.4 Fpa
SPACE VELOCITY 2.4 bv/.cc
PLANT SIZE 250 nv,
FACE VELOCITY 6.0 Fps
SPACE VELOCITY 6.0 bv/wc
CLEAN
GAS TO
STACK
Figure 4. Commercial Design for Larger Facility
6B-90
-------
Application of Dow Chemicals Regenerable Rue Gas
Desulfurization Technology to Coal-Fired Power Plants
By: Larry H. Kirby
The Dow Chemical Company
Freeport, Texas
Reiner W. Kuhr, Colin Sims, David Gullett
Stone & Webster Engineering Corporation
Boston, Massachusetts
6B-91
-------
Intentionally Blank Page
6B-92
-------
Abstract
The Dow Rue Gas DesuKurization Process is an advanced regenerate system wfth high S02 removal
efficiency, small fuutpilm. and the ability to recover S02-based by-products. The process uses a unique
absorbent fonrniatkxi developed by Dow that reacts reversfcty and selectively with S02, and has very
favorable physical and chemical properties.
This paper characterizes the Dow Hue Gas DesuKurization (FGD) Process in terms of Its developmental
status, anticipated design, and overall economics relative to a number of competing FGD technologies. The
design and cost information reported herein was completed by Stone & Webster Engineering Corporation
under contract to Dow Chemical U.S.A. Dow prepared system design parameters, flow schematics, and
a preliminary P&ID based on test results from the 1 MW pfot plant. Stone & Webster prepared a detailed
material and energy balance and developed preliminary designs using power plant flue gas parameters
identified in the EPRI report entitled Economic Evaluation of Rue Gas DestMurizaSon Systems.1 Process
equipment was sized based on the material balance; equipment costs were developed using budget
quotations from equipment suppliers and in-bouse estimating data and methods. Economic analysis
techniques applied to the competitive economic evaluation are consistent with EPRI assumptions and
approaches used to evaluate the other FGD systems reported. The restAs of this analysis suggest the
process compares favorably with commercial FGD systems.
Background
Dow has been a leader in acid gas treating since the early 1950*s and currently has about 150 commercial
units operating in the field that use a recirculating absorbent to remove an acid gas from a gas stream,
recover the gas as a product, and recycle the absorbent to repeat the process. The Dow S02 Removal
process has similar unit operations in the same configuration. However, this sorbent has very different
properties. In particular, this same absorbent molecule was designed and synthesized to react reversibly
with S02 and not react with other acid gases that might be present
Currently, these research activities have progressed to the piot plant stage and a 1 MW sized unit has been
running since June. The data to date has validated the laboratory findings. The work remaining is to
optimize operating parameters and costs, and to demonstrate the system on large scale gas streams. The
process has shown sufficient potential such that Dow is preparing to begin engineering on multiple large
scale (100 MW) demonstrations.
Process Design
Figure 1. Process How Schematic, illustrates the unit operations and major equipment required for DoWs
regenerate FGD process. The process system design is simBar to conventional gas sweetening processes
used for H2S and C02 removal in the gas and refining industries.
'Economic Evaluation ot Rue Gas OesuHurizafon Svsiems EPRI GS-7193. Volume 1. February, 1991.
6B-93
-------
Rue gas (stream 1) exiting the electrostatic precipitator (ESP) or fabric fitter passes through an existing ID
fan (R-101) upgraded to provide the additional draft requirements of the retrofitted FGD system or the
existing ID fan in series with a new booster fan (R-101). It is then quenched and scrubbed with water in the
prescmbber (A-103) to remove most of the hydrochloric acid (HCl). some of the remaining fly ash. and a
small amount of S02 and S03, which is present mainly as sulfuric acid mist Makeup water (stream 2) is
provided to replace evaporation and blowdown losses.
Blowdown from the prescrubber (stream 3) is neutralized with caustic or lime and could normally be
incorporated with coal pfle runoff treatment or sent to an ash pond.
The prescrubbed flue gas (stream 4) passes through a high efficiency mist elimination system before
entering the S02 Absorber (A-101) where it is contacted with DoWs absorbent for S02 removal. Scrubbed
flue gas passes through another high efficiency mist elimination system before proceeding to the stack
(stream 5). S02 rich absorbent from the absorber proceeds to the Rich/Lean Absorbent Exchanger (E-103)
where it is heated by hot lean absorbent
The heated rich absorbent flows to the S02 Stripper (A-102) where S02 is thermally desorbed. The vapor
phase of the S02 stripper overhead (stream 10) proceeds to by-product recovery. Lean absorbent (stream
13) is cooled in the Rich/Lean Absorber Exchanger (E-103) and In the Lean Absorbent Cooler (E-102) with
cooling water, and stored in the Lean Absorbent Surge Tank (T-102) prior to reentering the absorber. The
Absorbent Surge Tank (T-102) may not be required if sufficient hold-up for the system can be provided in
the bottom of the stripper.
The recovered S02 stripper discharge gas flows through a Water Condenser (E-104) where it is cooled with
cooling water, condensing most of the water vapor. The resulting sulfur dioxide rich gas is separated in the
Water Recycle Drum (D-105) as recovered sulfur dioxide by-product (stream 11). The recycled water
(stream 12). with an equSibrium amount of S02, is pumped back to be added to the rich absorbent (stream
6) leaving the absorber (A-101).
As the absorbent recirculates between the S02 absorption (A-101) and steam stripping (A-102) operations,
it accumulates impurities that need to be removed. These impurities include fine ash particles, heat stable
salts, and other soluble compounds. Filters win be used to remove ash particles from the absorbent. Heat
stable salts are removed in a slipstream of absorbent using a proprietary Dow technology.
Process Development Status and Schedule
The purpose of this paper is primarily to report the comparison of this developmental technology, in its
current to other technologies using recently published EPRI FGD Economics. It is felt that as the
research continues, a comparison to existing technologies must be made and regularly updated to assure
that a reasonable and viable process is emerging. An initial report on the process was presented at the
AICHE conference in Houston earlier in this year.
The process chemistry can be generally characterized by the following reactions.
SO2 + H2O — H2SO3
H2SO3 — H*+HS03"
R3N + H* .— R3N+H
6B-94
-------
The absorbent is a new compound and has been through the EPA evaluation procedures. The toxicology
testing, both for the new compound as it would be delivered to a plant and as It exists in the process, has
Indicated that the material has the EPA designation of "essentially non-toxic*. It has a very high boBing point
and therefore almost no material is vaporized and carried out with the scrubbed gas. The material is very
stable so that virtually no therma' degradation occurs in the process. Reactivity with S02 is very fast and
complete such that removal to very low levels can be achieved with low scrubber drctiation rates.
An aggressive schedule of development is being planned based on the favorable data collected to data
A 100 MW demonstration unit is anticipated to begin engineering in January, 199? A second 100 MW
project at a different site would begin in mid 1992. A 300 MW unit is planned beginning in 1993. The
purpose of these first units would be to verify the process on a variety of coals at commercial scale, and
to assure an understanding of the costs and operating parameters. This accelerated development program
is intended to travel the learning curve associated with a new technology in a timeframe that would allow
the technology to be considered commercially for plants to be built in the latter part of the decade.
Design Basis
The designs presented In the EPRI study, and for the Dow FGD systems evaluated in this paper, are based
on a 2.6% S Appalachian coal with a heating value of 13,100 BTU/lb, a chlorine content of .12% and 9.1%
ash. This coal is consistent with the base coal specified in the 1989 EPRI Technology Assessment Guide?
The 300 MW power plant has a net heat rate of 9722 BTU/kWh, with a coal bum rate of 111.3 tons/hr. The
raw flue gas stream from the ID fan is 3.19 million Ib/hr, or 1.03 million acfm at 28? F.
The process design for this Dow FGD system application is based on an S02 removal efficiency of 98%,
and a prescrubber system designed to saturate the flue gas and remove 90% of the chlorides arid 25% of
the fine fly ash that survive a high efficiency electrostatic precipitator. Booster fan capacity of about 10
inches water is required for this configuration.
By-product Recovery Options
Three by-product recovery options were evaluated, including the production of sulfuric acid, elemental sulfur,
and liquid S02. Process designs, capital and operating costs were developed for each recovery option.
Sulfuric acid production results in the highest capital cost; elemental sulfur production has the highest
operating cost; while liquid S02 has the lowest capital and operating cost However, the relatively small
market for liquid S02 will limit the applicability of this option. Most utility applications are expected to
require elemental sulfur production, primarily because of the abPity to store by-product indefinitely should
byproduct marketing and/or transportation difficulties interrupt the shipment of by-product from the plant
By-product systems to recover sulfuric acid and elemental sulfur are briefly described in the following
sections.
Sulfuric Add Recovery
Figure 2 presents a preliminary flow diagram of a sulfuric acid plant based on Information provided by
Monsanto.
JEPRI P-6587-L, 1989 EPRI Technology Assessment Guide
6B-95
-------
The absorbent regeneration plant by-product, a water saturated S02 gas, is the feed to a sulfuric add plant
The feed stream at approximately 18 psia and 217 F is first cooled to recover water that is recycled back
to the regeneration plant The S02 gas is dried by contacting with sulfuric add. The dried S02 gas is
heated and contacted with air In several vanadium pentcodde catalytic reactions. The cooled converter
effluent is contacted with an aqueous acid stream to absorb S03 from the gas stream and form sulfuric add.
This absorber overhead gas, consisting of a mixture of S02 and air, is preheated against the hot feed
stream. To reduce (recover) the remaining S02 content of this overhead gas, the gas is passed through
a secondary converter followed by a secondary absorber. The tail gas from the secondary absorber has
approximately 45 Ib/h of S02 and must be returned to the flue gas upstream of the scrubber. The
exothermic conversion of S02 to S03 provides the necessary feed preheat. At startup the catalyst bed must
be preheated to the initial reaction temperature of 95CT F to initiate the reaction. To achieve this initial
catalyst condition, air is heated by means of a natural gas or fuel 03 fired heater for approximately 24 hours
to soak the catalyst bed. in normal operation, the S02 Is bypassed around the heater. The air preheat
requirement applies to both the primary and secondary converters.
Elemental Sulfur By-product Recovery
Figure 3 is a flow schematic for a sulfur recovery process provided by Allied Chemical.
Natural gas is heated by low pressure steam in a preheater and then mixed with the S02 stripper overheads
stream to raise the fluid temperature above the dew point of sulfur before entering the reduction process.
The sulfur dioxide, S02, is reduced, in part, to form hydrogen sulfide. H2S. to a required ration of H2S/S02
of 2/1 with some formation of sulfur.
The S02 reduction is achieved in the catalytic reduction system. The reaction Is exothermic and sustains
the required reaction temperature. Elemental sulfur that is formed in the reactor system is condensed in an
inclined shell and tube exchanger by generating low pressure steam. Sulfur is condensed and flows to the
sulfur pit
The residual process gas stream flows to the first stage of a two-stage Clans reactor system where an
exothermic reaction occurs between the H2S and S02 to form sulfur and water. The partially converted hot
process gas is then cooled in a vertical steam generator to condense the sulfur which flows to the sulfur
pit
Unreacted process gas then flows to the second Oaus reactor where H2S and S02 forms additional sulfur.
The sulfur is condensed in a second vertical steam generator and flows to the sulfur pit The residual gas
passes through a demister to recover entrained liquid from the tail gas stream. This tail gas stream can then
be incinerated to oxidize any H2S to S02. and recycled back to the water recycle drum. D-105. Some of
the 50 psig steam generated is used to maintain liquid sulfur in the sulfur pit by means of a submerged con.
There is a net export of steam that can be used to supplement the S02 stripper reboDer steam requirements,
or the heating co3 steam to the liquid sulfur storage tanks.
6B-96
-------
Competitive Analysis
Levelized costs were calculated using EPRI methodology and assumptions based on the EPRI GS-7193
report. The levelized ($/ton) costs for the Dow FGD and several competing systems are summarized in
Figure 4 for the two by-product cases shown, alongside several of the conventional and advanced FGD
systems evaluated by EPRI. These represent values for a new 300 MW plant and do not Include allowances
for additional retrofit costs for adapting existing plants.
Conclusions
Based on the results of the technical and economic analyses completed for the 300 MW Dow FGD System,
the following conclusions summarize the implications of the information summarized in this paper.
1. The overall economics of the Dew FGD Process appear favorable to the other FGD
technologies evaluated to date by EPRI, primarily due to the value of by-products and the
elimination of the solid waste disposal costs associated with throwaway systems. Capital
and operating costs for the Daw FGD system with sulfuric add or elemental sulfur recovery
are lower than those for the limestone systems. These results are very sensitive to
variations in the assumptions and methodology derived from the EPRI study.
2. The Daw FGD system has many technical benefits compared to conventional and
advanced limestone systems. Most dramatic Is the fact that It is almost free of solids
handling systems, which are more expensive to operate arid maintain, and are less reliable
than liquid systems.
3. Of the three by-product recovery systems evaluated, liquid S02 is the most economic,
followed by sulfuric acid and elemental sulfur. The economics of each by-product system
are very sensitive to the capacity factor of the unit and byproduct unit value.
4. The results presented in this paper can be compared directly with the economics
published recently by EPRI for 15 other FGD technologies, as well as additional cases EPRI
will publish early next year.
6B-97
-------
FIGURE 1. 300 MW FGD PROCESS
o\
T
VO
00
A-103
PRESCRUBBER
A-101
802 ABSORBER
A-102
S02 STRIPPER
E-104
WATER
CONDENSER
BOOSTER
FAN
PRODUCT
CAUSTIC
ABSORBENT
TREATMENT
9YSTEM
DISPOSAL
D-109
WATER
RECYCLE
DRUM
A-102
STACK
LEAN
ABSORBENT
SURGE TANK
.vyyywM
MAKE-UP
ABSORBENT
FLUE OAS
FROM
ESP
A-103
STEAM
E-101
REBOILER
MAKE-UP
V/ATER
MAKE-UP
WATER
DI9POSAL
-------
FIGURE 2.1 SULFURIC ACID PLANT PROCESS
1
\
boostcn pan
aw
(JZ
PMIOHfATtn
•WUMOACIO
moouqt
tTOAAOl TANK
-------
FIGURE
3. ALLIED CHEMICAL S02 REDUCTION PROCESS
CONTROL BYPASS CLAUS
REACTORS
2ND SULFUR
CONDENSER
REACTOR
TRAIN
COMBUSTION
AIR BLOWER
NATURAL OAS
TAIL OAS
REOYCLE
STEAM
INCINERATOR
TAIL OAS MIST
ELIMINATOR
3RD 8ULFUR
CONDENSER
S02FEEO
DILUTION AIR
BLOWER
TO 8ULFUR
STORAGE
1ST SULFUR
CONDENSER
SULFUR PIT
-------
500
FIGURE 4. LEVELIZED COST COMPARISON
$/Ton S02 Removed
400
300
200
100
mm
lillli
tssii
LSFO LS/Wall
MgLime S Acid
DOW
I I Fixed O&M
HH Variable O&M
I I Fixed Charges
• 300 MW Unit
• Constant Dollars
• 2.6% sulfur coal
6B-101
-------
Intentionally Blank Page
6B-102
-------
PILOT TESTING OF THE
CANSOLV® SYSTEM
FGD PROCESS
L. E. Hakka
Union Carbide Canada Ltd.
Montreal, Quebec, Canada
R. W. Bimbaum
Union Carbide Chemicals and Plastics Company Inc.
Danbuiy, Connecticut
M. Singleton
Suncor Inc., Oil Sands Group
Fort McMunay, Alberta, Canada
Copyright© 1991 Union Carbide Chemicals & Plastics Technology Corporation
(Reproduced with permission from L. E. (Leo) Hakka)
CANSOLV is a registered trademark of Union Carbide Chemicals & Plastics Technology Corporation
Union Carbide tad Plaoics Company Inc. • SPECIALTY CHEMICALS DIVISION • 39 Old Ridgebvy Road. Daabury, CT 06817-0001
6B-103
-------
Intentionally Blank Page
6B-104
-------
ABSTRACT
Proof-of-concept pilot plant results are reported for a novel regenerable, retrofittable FGD
technology. The process utilizes an amine based absorbent and in-duct scrubbing. The absorbent
is regenerated by steam stripping to recover the S02 as a pure gas. The pilot plant processes
10,000 m3/hour (6,000 ACFM) of flue gas from utility boilers at Suncors Oil Sands plant, which
fires 7% sulfur petroleum coke. The pilot plant is a highly instrumented and versatile research
unit that is skid mounted for relocatability. Statistically designed experiments were run over a
wide range of independent variables, including 1,000-5,000 ppmv SOj, L/G of 0.03 - 0.26 1/m3
(0.25 - 2.0 gal/MACF) and scrubbing temperatures of 20° - 60°C (70° - 140°F). The CANSOLV
System achieved >95% SOj removal at low L/G ratios and scrubber residence times of less than
1 second and at a pressure drop of about 15 mm Hg (8"W.G). The results confirm that the
CANSOLV* System is economically superior to the advanced wet limestone FGD processes,
while delivering other benefits, such as small footprint, higher SOj removal and energy
efficiency.
6B-105
-------
INTRODUCTION
The 1990 U.S. Clean Air Act is a recent event in the continuing world wide trend towards
improving the environment of "Mother Earth". It is becoming widely recognized that mankind's
ability to pollute is starting to exceed the capacity of natural processes to cleanse and remedy the
air, water and soil contamination. The last 20 years have been a time of substantial progress in
the control of sulfur oxide emissions resulting from the burning of sulfur bearing fuels and from
industrial processes. Very substantial expenditures have been made in the development of FGD
technology, but considerable room for improvement remains. The dominant DeS0x technology
today is limestone or lime based processes in various forms. While in general reliable and, in
some forms, capable of high SOj removal efficiency, they produce large quantities of low value
waste products, are expensive to build and operate and are difficult to retrofit in constrained sites
due to the large equipment size.
Regenerable processes such as the Wellman - Lord or MgO process, avoid the waste product
problem of calcium based chemistry, but they are significantly more expensive in terms of both
capital cost and operating cost.
Research and development work by Union Carbide on a novel, regenerable, amine-based S02
scrubbing process has now progressed to proof-of-concept testing in a 2MW size field pilot plant.
This paper reports initial results from that work.
PROCESS CHEMISTRY
Due to technical simplicity, aqueous scrubbing/regeneration cycles have been the basis of the
most successful regenerable FGD processes such as the Wellman - Lord. In aqueous media,
dissolved SOz undergoes reversible hydration and ionization reactions that can be summarized
as:
S02 + HjO - H2S03 (1)
H2S03 - H+ + HSCTj (2)
Hsoj - ir + so3= (3)
The dissolution or equilibrium constants for steps (2) and (3) are reported as 1.54 x 10"2 and 1.02
x 10"7 at 18°C in dilute solution1. The scrubbing capacity of water can be increased by adding
a buffer or base to the absorbent which consumes hydrogen ions and causes reactions (1) - (3)
to shift to the right.
Steam stripping regenerative processes in which the bases used are stronger than sulfite,
degenerate to a sulfite/bisulfite scrubbing cycle, i.e. the effective base is sulfite.
2 NaOH + S02 -» Na2S03 + HjO (4)
Na^SOj + S02 + H20 - 2 NaHSOj (5)
6B-106
-------
Reaction (4) occurs in the initial contact of the base with S02- Reaction (S) is the basis for
scrubbing, being shifted towards the right in the absorber and being reversed by high temperature
in the regenerator.
The sodium ion does not participate in the reaction, its role being to provide electrical neutrality
to the solution. Reaction (S) can then be restated as (6), in order to highlight the essential
process.
SO=3 + S02 + H20 - 2 HSOj (6)
Any soluble cation can be used, as is sodium in the Wellman - Lord process, or a protonated
amine (the triethanolammonium ion) in the UCAP process2. The exact structure of the
CANSOLV® FGD Process amine absorbent is proprietary.
The scrubbing/stripping reaction can be represented as:
Ri R,
1 1
Rj — N + S02 + HjO - Rj - N* - H + HSO; (7)
I I
R, R3
The amine absorbent of the CANSOLV® FGD Process combines a low molecular weight and
high concentration, resulting in a net removal of 25-100 g SOj/1 (0.2 -0.8 lb. SO2/USG),
depending on the inlet S02 concentration, scrubbing temperature and % S02 removal desired.
The rate of S02 absorption into limestone slurries is limited by the slow dissolution of limestone,
which can only be partly controlled by limestone grind fineness and slurry pH.
Since the CANSOLV FGD Process, as represented by Equation (7), is essentially an acid base
reaction in a concentrated, homogeneous medium, its rate is very high. The limitation to mass
transfer then becomes the gas side resistance, which can be minimized by proper scrubber design.
The high S02 capacity of the CANSOLV FGD Process absorbent and its high reactivity eliminate
the need for absorbent recycle in the scrubber and permit operation at very low L/C ratios.
Practical S02 absorbents must be non-volatile in order to prevent equilibrium vapor phase losses
with the flue gas. The aromatic amines of the Sulphidine and Asarco processes exhibit
significant volatility and require removal from the treated flue gas stream by washing with dilute
sulfuric acid3. This is both costly and complicated. The absorbent of the CANSOLV FGD
Process is essentially nonvolatile (vapor pressure < 25 ppb).
Due to the special nature of the CANSOLV FGD absorbent, strong acids which either form in
or are captured by the amine as Heat Stable Salts (HSS), may be present at high concentrations
without limiting the amine solution's scrubbing capacity.
6B-107
-------
Heat stable salts form by reaction of the amine absorbent with acids that are either nonvolatile
or too strong to be driven off in the steam regeneration step. These acids are introduced into the
absorbent from the following sources:
1. Flue gas - may contain SOs (produces HjSOJ, HC1, HF, and N02.
2. SOz oxidation to S03 by oxygen.
3. Disproportionation of sulfite to sulfate and other sulfur species:
SO5 —» SOj + other sulfur species (8)
Many other reactions that produce strong acids are described in Reference 4.
PROCESS DESCRIPTION
The CAN SOL V® Flue Gas Desulfurization process flow diagram is similar to the well known
alkanolamine H2S-C02 removal process and is depicted in Figure 1. Countercurrent multi-stage
"in duct" scrubbing, utilizing air atomizing nozzles takes advantage of the absorbent's high
reactivity and S02 capacity to effect up to 99% removal in a very compact and energy efficient
manner.
The CANSOLV FGD absorbent is a homogeneous liquid throughout the process cycle and
exhibits no tendency to precipitate solids. This results in several benefits:
• There is no equipment erosion, as with slurry processes;
The S02 - amine reaction that occurs in the homogeneous solution is fast. It
therefore allows small contacting devices to be used in comparison to those
needed for limestone systems;
There is no scaling in the absorber or gas ducts;
• There are no significant solids handling problems.
The absorbent is non-volatile, stable both thermally and oxidatively and has good health and
safety characteristics.
The process consists of a gas cooling and prescrubbing section, an S02 scrubbing section and a
regeneration and solvent purification section.
The flue gas cooling and prescrubbing equipment reduces the temperature of the flue gas and
removes most of the particulatematter and strong acids (S03, N02, HQ, HF etc.) The flue gas
also leaves the prescrubber fully saturated with water.
Rue gas scrubbing is effected in-duct at flue gas velocities of up to 30-40 ft/sec. In the pilot
plant, the high reactivity of the absorbent has allowed each of the three mass transfer stages to
be only 8 feet long.
6B-108
-------
High speed interstage solvent collectors are used to recover the absorbent between stages and a
final mist eliminator downstream of the scrubber removes amine to an insignificant level from
the flue gas before it is sent to stack.
The regenerator is similar in design to regenerators in ethanolamine gas sweetening service. It
is equipped with a steam heated reboiler to regenerate the amine and a vacuum pump to ensure
that regeneration occurs at low enough temperatures to suppress the disproportionation of
rcgenerable SOz into non regenerable S03.
Strong acids in the flue gas, such as HQ and H2S04, which are not removed in the prescmbber,
react with the absorbent to form heat stable salts. Additional salts are also formed in the amine
solution when some of the dissolved SOz converts to S03. The concentration of heat stable salts
in the amine solution is controlled by taking a small, continuous purge stream of amine from the
unit, purifying it chemically and returning it to the unit.
The only waste streams generated in the CANSOLV® FGD Process are:
1) the blowdown from the prescrubber water loop, and;
2) the sodium salt purge stream from the absorbent purification unit.
Effluent treatment techniques are site specific and depend on the composition of the flue gas and
local environmental regulations that apply to the site. The small quantities of waste produced by
the CANSOLV FGD Process, however, allow zero effluent discharge processes to be considered.
PILOT PLANT TESTING
Laboratory testing proved that the absorption and regeneration concepts of the process were
sound. The commercial economics suggested by the lab data were sufficiently attractive to
convince Union Carbide that further testing of the CANSOLV® FGD Process in a larger facility
was justified. In May 1990, the Suncor Oil Sands Group Inc. in Fl McMurray, Alberta agreed
to work with Union Carbide to demonstrate the process at their plant and funds were secured to
build it. The pilot plant design was heavily impacted by the need to: a) extract the data that will
be needed to design and operate a larger facility and; b) to prove the viability and operability of
the process in an industrial setting.
The pilot plant was designed to treat 6000 ACFM of flue gas emanating from the Suncor utilities
boilers, which burn 7% sulfur petroleum coke as fuel. Three 70 MWe boilers fire 2,300 tons per
day of petroleum coke that is produced on site by the bitumen upgrading process. About 65 MW
of electricity is produced by each boiler, while the balance of the steam is used for the extraction
and upgrading of the bitumen to synthetic crude oil.
Properties of the coke are listed in Table I. Average flue gas conditions are given in Table H.
The volume of flue gas treated in the pilot plant is about 3% of one boiler's output, roughly
equivalent to about 2MWe. The pilot plant is of modular design and was shipped to Fort
McMurray in December 1990. It was started up on February 25, 1991.
6B-109
-------
OPERATIONAL RESULTS
From startup through to June 22, 1991, a total of 2832 hours of operating time were available.
The unit performed as follows:
Flue Gas/Conditionine
Amine Regeneration
Operating Time
2476 hrs
87%
2184 hrs
77%
Planned Shutdown
168 hrs
6%
168 hrs
6%
Adjust for Exps
0 hrs
0%
240 hrs
9%
Mechanical S/D
188 hrs
7%
240 hrs
9%
Total
2832 hrs
100%
2832 hrs
100%
The pilot plant operated very stably. When experimental plans allowed the operating conditions
to be left unchanged for extended periods, there was very little work to be done to supervise the
unit In addition, when changes were required, the pilot plant achieved steady state operating
conditions within about two hours of the change.
The pilot plant also permitted the reagent side to be isolated from the flue gas side so that minor
maintenance could be performed with minimal disruption to the overall unit. The ability to
"uncouple" the two systems allowed some maintenance of each system to be performed while
the other was still on line.
INSPECTION FOLLOWING NINETY DAYS OF OPERATION
A seven day planned shutdown was taken to perform a general inspection of the unit after ninety
days of operation. During this period, all corrosion coupons were pulled and several critical
areas of the unit were inspected in detail. Both the coupons and the general inspection of the unit
indicated that 316 stainless steel performed acceptably in amine and dilute acid service.
FLUE GAS SYSTEM
The flue gas cooling and conditioning system has proven to be very reliable. Outages of the flue
gas system were caused by two failures of the flue gas emergency shutdown valve. The heat
exchangers and the rotating equipment in the flue gas cooling and conditioning system have
operated acceptably. Additional outages of the flue gas side occurred because the supply of flue
gas quench water was unreliable. A change was made in the supply system and service was
upgraded.
6B-110
-------
AMINE SYSTEM
The operation of the SOj absorption and regeneration sections was intenupted several times to
repair seal and gear problems in the reflux pump. Most of the rotating equipment, including the
reflux pump, was not spared in the pilot plant. This decision contributed to nearly 200 hours of
amine side outage during the early operating period that could have been avoided. All other
rotating and heat exchange equipment in the SOz absorption and regeneration system have
operated acceptably.
PILOT PLANT EXPERIMENTATION - RESULTS
In excess of 15 independent variables characterize the operation of the CANSOLV® System.
Statistically designed experiments were run and the results were analyzed by regression analysis
to generate a statistical model of the process. Of these variables, the inlet S02 concentration, the
flue gas scrubber velocity and the inlet gas temperature served most to characterize the process's
performance. Figure 2 illustrates the "experimental universe" that established the parameters for
the experimentation work at the pilot plant.
The pilot plant was run at conditions within the experimental universe and SOz removals of
between 90% and 99% were demonstrated. The results were proved and are reproducible. S02
removals up to 99% at any inlet SOz concentration can be obtained. The variability in the
removal rates was impacted not only by the flow, inlet SOz concentration and temperature, but
also by other variables such as L/G ratios, stripping rate etc. The SOz removal rate can thus be
adjusted to almost any value through the adjustment of only a few of the critical values that
impact the operation of the unit.
PROCESS ECONOMICS
Union Carbide commissioned a study by an independent engineering firm to compare the
economics of several commercial processes to the CANSOLV® FGD Process. Four
lime/limestone processes and one regenerable process were selected for comparison. A power
plant consisting of 2 x ISO MW units was chosen as the basis for the study. The FDG capital
costs were based on coal containing 4.1%, while the operating costs were generated on the
assumption that 3.3% sulfur coal was normally used. The results are shown in Figure 3.
The study proved that when CANSOLV is compared with limestone processes in a high sulfur
application, the results can be very favourable.
Generally, the CANSOLV® System also is quite insensitive to variable coal sulfur compositions.
The graph in Figure 4 shows how the cost for CANSOLV would vary with sulfur content.
Furthermore, considering that the CANSOLV FGD Process is still at a relatively early stage of
optimization, it is not unreasonable to expect further improvement in process economics.
6B-111
-------
CONCLUSIONS
The early operation of the pilot plant has clearly shown that:
1) S02 can be removed from flue gas to any desired level up to 99% by varying the
IVG ratio of the solvent fed to the unit without exceeding an L/G ratio of about
0.26 1/m (2 USG/1,000 ACF) of gas.
2) The CANSOLV* System reagent is stable.
3) The CANSOLV System reagent absorbs S02 at high scrubbing velocities. S02
can be removed from the amine by steam stripping and the amine may be reused
in the absorber.
4) The interaction of sulfur species in the system is easily controllable and does not
impact significantly on the operability of the process.
5) The rate of heat stable salt formation in the system requires less than 1% of the
circulated solution to be removed for treatment and subsequent reuse.
Pilot plant results have clearly indicated that CANSOLV technology represents a viable, low cost
system for S02 emission control for coal fired power facilities. It has been demonstrated that the
CANSOLV System can remove up to 99% of the SO2 in the flue gas and that it represents a
trouble free system for S02 removal.
6B-112
-------
ACKNOWLEDGEMENTS
Financial support for this program was provided by the Department of Western Economic
Development of Canada. Suncor Inc. Oil Sand Group and Union Carbide Canada Limited/Union
Carbide Chemicals and Plastics Company Inc.
The dedicated efforts of all employees of I.P. Constructors, Partec Lavalin, Suncor Inc., Union
Carbide and numerous other contractors and suppliers is gratefully acknowledged.
REFERENCES
1. Handbook of Chemistry and Physics. 71" Edition, D.R. Lide, Ed., CRC Press.
2. U.S. Patent 3. 904. 735. September 9. 1975. G.R. Atwood, RJ. Blake, K.F. Butwell, and
D.A. Dunnery, assigned to Union Carbide Corporation.
3. "Gas Purification" A. L. Kohl and F. C. Riesenfeld, 4th Edition, Gulf Publishing.
4. "Fundamental Chemistry of Sulfur Dioxide Removal and Subsequent Recovery via
Aqueous Scrubbing", M. Schmidt, Int. J. Sulfur Chem.. Pan B, Vol. 7, Number 1 pp.
11-19.
6B-113
-------
Table I
SUNCOR Petroleum Coke Analysis
Moisture 8% - 10%
Volatiles 10%
Fixed Carbon 84% - 87%
Sulfur 7%
Ash 3%
Ash Analysis (as oxides):
Si
50.5%
A1
25.5%
Fe
75%
Ca
1.9%
V
5.5%
Others
9.1%
Table H
Average Flue Gas Conditions
Composition:
N2 81%
'2
o
co2 11%
S02 3600 ppm
CI, F Present
NO, 175 - 375 ppm
Particulate 0.06 - 0.11 kg/10? kg flue
(0.03 - 0.06 gr/SCF)
Temperature 475° - 550°F
Pressure -2 mm Hg (-1" HjO)
6B-114
-------
1
VLD1 US POD
TO STICK BBOOVBir AND
FIB
XKH SOEVTNT
1
80S TO
cnzmmoK
r
scsobbxb I 1-T-J amonm
I I SUF9BUK
ntfiiKnas
RD COOLING 1 ^ TBMIMBU
(OPTIONAL) T r BUIMMT
pcBiFiCAinm
PUBGX 10
V1SZX
Figure 1. CAN SOLV® System Schematic
Temp (Deg.
20
5,000
t
PPM Inlet S02
1,000
5 Velocity (M/Sec) 12
>
Figure 2. Flue Gas Conditions
6B-115
-------
1W///SA
Figure 3. Capital and Operating Costs CAN SOLV vs Other FGD Technologies
is.
03
03
o.-t
0.2
o *—
2.1
2.4 2.7 3 33
X Sulfur la Coal
3.6
3.9
Figure 4. CAN SOLV® Cost Sensitivity To % Sulfur in Coal
6B-116
-------
DRY DESULPHURI2ATION
TECHNOLOGIES INVOLVING HEMIDIFICATION FOR
ENHANCED S02 REMOVAL
D. PAUL SINGH
PROCEDAIR INDUSTRIES INC.
625, PRESIDENT KENNEDY, 14th FLOOR
MONTREAL, QUEBEC.*, H3A 1X2
DR. M.Y CHUGTAI
GEA LUFTKUHLER GmbH
DORSTENER STR. 484
D—4630 BOCHUM 1
6B-1I7
-------
Intentionally Blank Page
6B-X18
-------
1. ABSTRACT
The paper is in two (2) parts. The first part describes an In-Furnace Dry
SO; Scrubbing Technology while the second part describes a Post Combustion
Dry SOj Scrubbing Technology. Both technologies utilize enhancement
features to improve overall sulphur removal efficiencies.
The first part of the paper describes the key design features and the
results of lime injection into a furnace plus gas humidification upstream
of the electrostatic precipitator. The desulphurization plant has been
installed on a 65 MW (thermal capacity) stoker-fired boiler. The results
cover desulphurization efficiency, the impact of the process on the boiler
and electrostatic precipitator, ash utilization, and overall reliability.
The flue gas humidification, in the conditioning reactor, result in an
overall sulphur capture of better than 90% at high boiler load and a Ca/S
stoichiometric ratio around 2.0. A desulphurization efficiency of
approximately 65% was obtained with lime injection alone.
(The paper further describes the pertinent design features and the results
of dry humidified hydrated lime in a fluidized bed reactor after the air
heater and upstream of the fabric filter dust collector, or electrostatic
precipitator). The desulphurization plant is installed on a district
heating plant in the South of France with a capacity of 3 MW (thermal).
The results cover the desulphurization efficiency at various Ca/S
stoichiometric ratios and humidification levels. Results show that with
an acceptable level of humidity in the reactor, an overall sulphur capture
of better than 90% by weight, at a Ca/S stoichiometric ratio around 1.5
can be achieved.
2. INTRODUCTION
An "In-Furnace" process utilizes limestone injection directly into the
furnace, with a conditioning-tower installed downstream. In contrast, a
"Post-Combustion" system employs the injection of dry hydrated lime or
lime powder into the flue gas stream downstream of the furnace and
economizer. The "Enhanced-Dry" process activates the powder reagent
reactivity by increasing its humidity or chemical reactivity prior to
injecting the powder into the flue gas stream, downstream of the furnace
and economizer.
The method of direct desulphurization discussed here is the absorption of
sulphur by the basic sorbents CaCO, (powdered limestone) and Ca(OH):
(slaked lime). Fig.1 shows the qualitative process of sulphur capture,
plus that of any halogens present.
When CaCO, or Ca(OH): is employed as the additive for pollutant capture, an
endothermic decarbonation or dehydration reaction (calcination) takes
place in the furnace in accordance with the following equations:
CaCO, > CaO + COi (1)
Ca(OH)2 > CaO + Hp (2)
The CaO particles react with oxides of sulphur and other pollutants like
halogens according to the following exothermic reactions.
CaO + SO; + 1/2 Oj > CaSO, (3)
CaO + HC1 > CaCl- + Hp (4)
CaO + 2 HF > CaF2" + H: (5)
CaO + SOj CaSO* (6)
6B-119
-------
^~\^alOH)2 or CaC03
¦ i X ¦
<0
/J\ / \ CatOHJj^H CaOHjO
ho/ X S XCO, CaCOj *^H — Ca0»C02
S02_ {Sfe\-55L
. <%>
- / ^T~v • Ca0*S02*I0j—CaS0t*^H
°r 1 \$ CaO* S03 — CaSOt *M
I ^
incL gaseous pollutant reactions
CaO* 2 HCl CaCl2 ~HjO* M
CaO*2HF CaFz ~»^0* M
Polishing Reactor
CaCUJ^O Ca(OH32 **H
CaOSOj CaSOj *aH
Ca(0H)j«S0j*Hj0*l Oj-CaSO^H^H
CaSOt
CatOHIj
CaSOj
Disposal
and re-use
Fig. 1: Schematic representation of direct Desulpburization
process with Conditioning reaccor
6B-120
-------
During direct desulphurization in the furnace the additive particles become
covered with a compact layer of sulphate. Migration of the SOj deep into
the pores and adequate diffusion velocity in the boundary layer can only be
attained with very long residence times (minuteB or hours), therefore, the
core of the additive particle is not available for reaction with S02-
In order to make this free CaO available for further sulphur capture, a
conditioning reactor is employed downstream) of the furnace. The
prerequisite for applying this additional process for better sorbent
utilization is the presence of free active CaO in the flue gas exiting from
the boiler.
For the conditioning process, it is necessary to further humidify the flue
gas. In order to accomplish this, water is injected into the flue gas at
the reactor inlet. Considerable care has to be taken to achieve good
atomization in order to ensure complete evaporation inside the reactor. As
a result of the hygroscopic properties of the free CaO, the water vapour
diffuses within the pores of the CaO core which is surrounded by a layer of
CaSO*. The free CaO in the pores then combines with the H-O to form Ca(OH)2.
This exothermic reaction tends to progress outwards from the pores on the
exterior of the particle, and is also encouraged by discontinuities and
substitution molecules.
Since Ca(OH); has a lower bulk density and a greater specific surface area
than CaO, the resulting expansion of the hydroxide and the heat produced
during slaking tends to split the material along the particle boundaries.
At these break points a new surface capable of further reaction with the
pollutants, is produced. During the first reaction there is slaking in
order to form Ca(OH):, followed by the capture of sulphur dioxide.
CaO + H-O Ca(OH); + 1.952 KJ/Kg CaO (7)
Ca(OH);*+ SO- + H-O + 1/2 Oi CaSO„ . 2H2O (8)
Ca(OH); + ScC + H2O CaS03 . 2H-0 (9)
At the fracture zones of the particle the free CaO takes part
preferentially, in the sulphur capture reactions according to the following
equations:
CaO + S02 + 2 H2O + 1/2 O, CaS04 . 2 Hp (10)
CaO + SO, " CaS03 (11)
CaO + SoI + 2 Hp CaSOj . 2 Hp (12)
A comparison of the reaction equilibrium constants show that for sulphur
capture the conversation to calcium sulphate is the most favoured reaction.
Simultaneous with the absorption of sulphur dioxide, the HC1 and HF
concentrations in the uncleaned gas are also reduced. The chemical
reactions involved are described by the following equations:
Ca(OH)- + 2 HC1 CaCl, + 2HjO (13)
CaO + 2 HC1 CaClj + Hp (14)
Ca(OH): + 2 HF CaF, " * (15)
CaO + 2HF CaF; + H;0 (16)
The achievable desulphurization efficiencies depend heavily upon the
approach to the adiabatic saturation temperature of the flue gas.
The technical components of the conditioning process are a conditioning
reactor, a flue gas side bypass and (possibly) a flue gas reheat system as
shown in Fio. 2.
Preliminary experiments have shown that the degree of desulphurization
achievable in the reactor substantially increases with the amount of water
evaporated there. Thus, the aim is to run the conditioning reactor as close
to the dew point as possible, ensuring that neither incomplete evaporation
nor condensation on the reactor walls will occur.
6B-121
-------
Stack
Bypass
Boiler
v Fly-ash
Reheat system
Polishing
Reactor
Additive
v Fly-osh
pjq. 2: Schematic flow diagram ot direct desulphuriiatior. „ith conditioning reactor.
-------
A retrofit with the direct desulphurization process in combination with a
conditioning reactor, is recommended, especially for boilers equipped with
electrostatic precipitators, is quenching the flue gas with water helps to
upgrade the performance of the electrostatic precipitator.
3. TEST BOILER
The boiler shown in Fio. 3 is a stoker-fired unit with a thermal capacity
of 65 mw. It fires a bituminous coal containing roughly 1 % sulphur with
a net calorific value of approx. 29 MJ/kg. The unit is a 3-pass boiler,
the furnace forming the 1st pass, with the 2nd pass containing
superheaters and the 3rd pass the economizer. The superheater section and
the economizer are equipped with sootblowers. Spray attemperation is
included in the steam circuitry between the superheater section 1 and 2 as
well as between 2 and 3 to bring the steam temperature down to 525" C.
The unit is equipped with a cold-side ESP mounted on top of the boiler
with a nominal design specific capture area of 81.8 nr / (nr'/s). The ESP
consists of two separately energized electrical fieldB in series. The
particulate emissions are less than 50 mg/tsP.
Bottom ash from the travelling grate is extracted dry and stored
separately in a hopper.
The flyash from the ESP and the bend between the 2nd and 3rd pass is
pneumatically conveyed to a storage silo from where it can be discharged
either into trucks or railcars.
4. CONCEPT OF SORBENT INJECTION
Extensive investigations have shown that the following four key parameters
mainly determine direct desulphurization:
1. Flue gas temperature at the point of injection of the additive.
2. Residence time of the flue gases in a definite temperature range.
3. Dispersion of the additive in the furnace.
4. Reactivity of the additive.
Previous investigations have shown that additive injection should not take
place at a temperature higher than 1200° C and that the period required to
cool the flue gases down to approx. 750° C should not be shorter than
around 1 to 2 seconds. The shorter residence time applies to additives
with a smaller particle size or larger specific surface area, e.g. Ca(OH)2.
In order to determine the optimum sorbent injection level, the mean axial
temperature profile in the furnace was computed with help of field
measurements and heat transfer modelling, as shown in Fio. 4. With the
aid of a sulphation model, in which the temperature at the point of
injection and the residence time down to 750 ° C play a major role, two
injection levels at 10.8 m and 18.5 m were established. The Injection
nozzle spray direction can be adjusted over a range of + 30° to - 30° along
the vertical axis of the furnace.
Fig. 5 shows the residence time as a function of boiler load for the
higher and the lower level and for three different angles of sorbent
injection.
6B-123
-------
Flue gas
outlet
Superheaters
21.850m
20.750m
Coal
storage
Sorbent
injection
Furnace
10.8 m
Sorbent
injection
Coal
feeder
~ 5.675 m
Travelling
grate
To fly ash
storage
Ash from grate
Pig- 3: Schematic diagram of test unit at Boehringer
Ingelheim.
6B-124
-------
Httsurtd Points
¦ 60% Load
• 40% load
CL
to
o
or
3
750
500'
m
•s
Grata Lrrt(1 ltvct2
Elevations
Flq. 4: Flue gas temperature profile along the axis
of boiler furnace.
c
«
W
ee
Botltr Load
Fla. 5: Residence time of flue gas as a function of boiler
load for higher and lower level of additive injection
in the sulfation window.
6B-125
-------
Vent Filter
o\
V
N)
OS
Distributor 4m
Pill Line
•VI Feeder
Two-Way
Junction V
Air Compressor
O
Injector Conveyor
Pig. 6: Schematic Clow of direct desulphuriaation system.
-------
The additive flow rate can be set manually to produce a constant molar
ratio Ca/S at a particular load e.g. for test purposes. In the normal
case when the boiler is operated according to the load demand, the
additive flow rate is adjusted automatically with the help of a controller
according to a function m = f (load) to achieve a required percentage of
desulphurization. In addition to the input curve the additive flow can be
automatically corrected so as not to exceed a given SO2 emission level.
5. Conditioning REACTOR DESIGN
Fig. 7 shows in schematic form the arrangement of the conditioning reactor
in the flue gas duct system. The dsimper is the bypass and can be set at
intermediate positions between"open" and "closed", thus making it possible
to route any desired flow through the conditioning reactor right up to the
full flue gas flow.
The conditioning reactor is cylindrical and located in a vertical
position. At full load, the flue gases sure at a temperature of about 150°
C and the residence time in the cylindrical section of the reactor is
approx. 8 seconds. Flow through the conditioning reactor is from top to
bottom.
The head of the reactor consists of a diffusor which diverges the flow
cross-section to the diameter of the reactor. At the inlet to the
diffuser, nine external twin-fluid nozzles, manufactured by Lechler, for
the injection of water, are located in the flue gas duct cross-section
(Fig. 7). The external mixture of water and atomization fluid allows
either steam or compressed air to be used. The liquid (water) is fed in
the middle of the sprayer and leaves the nozzle in the form of a hollow
cone. The compressed air is fed through a co-axial pipe to the chamber,
which is equipped with swirl inserts.
The atomizing nozzle is a prefilm type; that is the liquid is formed into
a liquid sheet before it is- hit by the high velocity swirled airstream.
The prefilming of the water provides fine atomization due to the very thin
water sheet.
The droplet size distribution (Sauter Mean Diameter) depends on the air to
water ratio and on the water pressure. With higher water pressures it is
possible to achieve finer droplets for a relatively low air consumption.
This allows selection of the optimum design parameters for atomization.
The main advantages of a nozzle with external mixing and prefilming of the
liquid are the higher amount of atomized water and the possibility of
utilising steam instead of air. Because of the separate feeds of air and
water flow to nozzle, they can be very simply controlled, by setting the
required pressure for each medium.
The length and angle of divergence of the diffusor approximately match the
range and spray angle of the nozzles, so that the water droplets are
dispersed in the flue gas, at the outlet of the diffusor section. In the
cylindrical part of the reactor the residual water is evaporated, and the
secondary desulphurization described above, is effected.
The flue gas can be cooled down to within 10°C of the adiabetic saturation
temperature at the conditioning reactor outlet. The amount of water
sprayed is controlled, in order to set up a predefined temperature at the
reactor outlet. It is important that the conditioning reactor is kept
dry; that is, the water sprayed in must be fully evaporated. This can be
achieved by the choice of a suitable length for the reactor and by
limiting the droplet size of the water which is injected. In the present
case, the objective is attained by making the cylindrical part of the
reactor around 20 m long and limiting the maximum size of the droplets
produced during atomization to 130 micron.
6B-127
-------
Flue Gas
to ESP
Flue Gas
from Boiler
Water
Atomizing
Air
Seal Air
3*3
Atomizing Guns
Ol
Ol
13
conditioning
Reactor
Fiq- 7' Contitloning reactor arrangement.
6B-128
-------
This last measure involves high opera-ting costs which might be reduced, in
future, by using steam as the evaporating medium. One possible method of
shortening the evaporating section for the droplets would be to use water
heated to a of 85°C for spraying purposes.
When the flue gases emerge from the bottom of the conditioning reactor,
they are sharply deflected and at least part of the entrained particulate
matter is ejected. The bottom of the reactor is conical to allow the
solid material thus collected, to be drawn off to the fly ash storage
silo.
In the rising clean gas duct, the flue gas is heated, if necessary, to be
at a temperature of at least 88°C, upstream of the stack. Heating is
provided by a tubular heat exchanger operated with saturated steam at a
temperature of 193°C sr:d a pressure of 13.5 bar. The amount of steam is
controlled on the condensate side Ln accordance with the flue gas
temperature downstream of the heat exchanger.
6. RESULTS
6.1 Direct Desulphurization
Desulphurization efficiency N, is defined as
N. = SO.,^ ~ SO-. (17)
sew.
where SO, is the maximum theoretical value calculated from the sulphur
content of the fuel and SO, is the measured value during sorbent injection.
The molar ratio Ca/S is given by the amount of calcium injected into the
furnace and the sulphur content of the fuel.
Fie.10 shows the desulphurization efficiency as a function of the molar
ratio Ca/S for boiler full load and a part load of 43%. These sulphur
capture rates are obtained in the optimum operation mode. At a molar
ratio Ca/S - 2 with the sorbent Ca(OH)-, the sulphur capture achieved at
full load is about 65% and at low load about 80%.
The sulphur capture curve in Fig.8 demonstrates the expected shape, which
has been always round in the investigations. Mathematically such curves
can be described with help of exponential functions. Theoretical
considerations /1/ show that theBe Bulphur capture curves can be closely
represented by the mathematical function:
N. = 1 — e" A
The coefficient A represents the maximum achievable efficiency N„, which
is attained when the ratio Ca/S tends towards zero.
A = Na, when Ca/S > 0
since A is the slope of the tangent to the curve described by equation
(18) when Ca/S = 0. The coefficient A can be seen as the characteristic
of the direct desulphurization Bystem and hence represents a Bimple way of
quantifying the potential for sulphur capture in the system.
The optimal mode of operation for full and part load can be derived from
Fig.9 with the aid of the coefficient.
6B-129
-------
Fig -9 summarized the desulphurization characteristic in relation to
residence time. The effect of employing alternately the upper or the
lower level and likewise the angle of injection (-30„, 0„ and +30o) can be
clearly seen. It is also apparent that increasing the residence time will
only improve desulphurization if the sorbent is reliably prevented from
encountering excessively high temperatures. Increasing the residence time
in the upper part of the permissible temperature range always has a very
favourable effect on sulphur capture, leading to a steep rise in the
value with increasing residence time. For example, at full load an
increase in residence time of 0.3 sec., achieved by directing the nozzles
downwards, produces A-value of 55% instead of 36%.
A drop in the value of the coefficient is observed when the residence time
is increased, which indicates that the sorbent is being injected at too
high temperature. This could, for example, be caused by injection in the
tower plane at full boiler load.
6.2 Conditioning Reactor
An initial test series has been already run in order to establish the
optimum conditions for sulphur capture while keeping the reactor dry.
For this purpose, the amount of water sprayed into the reactor has been
increased in steps, with the Ca/S stoichiometry being maintained constant
at Ca/S = 1.5. Since the evaporation heat for the water added was
completely taken from the flue gas, the amount of water can be expressed
as the difference between flue gas temperature at reactor outlet and the
adiabatic saturation temperature for water in the flue gas. oT. The
maximum water flow to be evaporated by a given flue gas flow is given by
esT. - 0. At low values of &T, the Ca/S ratio was varied between 1.5 and
2.5. The result is shown in Fig.10.
For all stoichiometric Ca/S ratios, the desulphurization efficiency in the
conditioning reactor Na4. increases as the temperature difference £>T between
the outlet and the adiabatic saturation temperature of the flue gas
decreases. On the other hand, the Ca/S ratio strongly affects sulphur
capture, at least at low values of oT. With Ca(OH)2 at a Ca/S ratio — 2,
65% sulphur capture inside the conditioning reactor has been attained at
85% boiler load, resulting in an overall capture N,^. of more than 90%. At
the same stoichiometric ratio but 70% boiler load, Nli4. = 84% (N.j,. = 98%)
were measured. 43% load resulted in near 100% capture as the stack SO-
monitor could no longer detect any SCK.
During experiments with partial bypassing of the conditioning reactor and
increase of residence time sulphur capture was not improved.
However, a certain influence of boiler load on the sulphur capture in the
conditioning reactor was recorded: desulphurization improves at lower
boiler loads. This is rather unexpected as the flue gas temperature at
the reactor inlet is then also lower and the amount of water to be
evaporated for a given oT is also lower. Consequently, the water burden
in the flue gas decreases. Further experiments must be carried out to
prove whether higher temperatures in the radiant section of the boiler at
higher loads also make the sorbent less reactive, as far as
desulphurization in the conditioning reactor is concerned.
Up to Ca/S = 2.5, increasing the stoichiometry significantly improves
sulphur capture in the reactor. As desulphurization efficiency better
than 90% is achieved at this ratio, a further increase of the Ca/S is not
necessary.
Going down to less than about 15°C above the saturation point was found to
cause condensation on the reactor walls. Thus,
-------
100
•0'
*k
CMl
5-U1%
Addirnr*- CjCOHJj
a <3% uqc
» x% iMd
Ca/S
Fig. ft Deaulohurixacion efficiency -jjs as a function
of molar ratio Ca/S.
1
InjttfiOf
» An# *
f
/
L
£
>
LM0
opp*rf low*f
l««»l
o
o
9
100%
©
•
s s s w
Residence Tine
Pig. 9": Desulphurization. characteristics for different
operating modes of the additive injection
system.
6B-131
-------
100
%
90 A
80
70
60
'S.Q
50 -
UD
30
20
V,
/S =2.5
Lood = 85 %
V
\
/S=2.0
v \
\ ^
*\ <
\ 4
\
\
: \
4
Vs^
Co/s=n.5
10 20 30 50 °C 60
AT
y<<7 lQ:Pesulphuritatlon. efficiency ijs ^ In the conditioning
reactor as a function of approach to adiabatic
saturation temperature A T with the molar ratio
Ca/S as parameter.
6B-132
-------
So far, only Ca(OB)2 has been tested as a sorbent. However, it is also
planned to conduct experiments with CaCo,.
6.3 Operating Experience with Downstream Plant Components
fonJLug of beating Borfaceaz
It has been observed that the deposits on conveetive heating surfaces are
very loose and light and are easy to clean off with the installed
sootblowers. As a result, the fouling remaining layer on the tubes after
sootblowing is thinner than in the case of boiler operation without direct
desulphurization.
No detrimental effects have been observed on the flue gas ducts, ID fan or
stack.
Slectrostatic precipitator:
No unusual fouling or caking has been observed on the electrostatic
precipitator during DDP operation.
The prescribed limiting value for dust emissions of 50 mg/m3 STP was a rule
maintained during DDP operation as well.
Electrostatic precipitators which have a specific collecting area larger
than 75 in; / (n? /s) (384 ft3/l,000 acfm) can comply with a limiting value
for emissions of 50 mg/m1 STP (0,04 11/ 10* BTU) under normal boiler and
fly ash conditions.
CONCLUSIONS
Although long-term tests are on-going, the suitability of the combined
direct desulphurization process with the conditioning reactor has been
proven on an industrial scale.
The capture efficiency provided by a low-cost process involving relatively
low sorbent consumption and by-product volume, noticeably exceeds that
produced by the other direct desulphurization methods used so far.
7. PRINCIPLE OF THE PROCEDAIR ENHANCED DRY
The principle used by Procedair in its "dry" systems is the absorption of
the gaseous pollutant by a powdery material uniformly dispersed within the
flue gas stream. The efficiency of the process is a function of the true
contact between reagent and the polluted gas, and this has been mastered
over many years with the Procedair Venturi Reactor Tower. This however, is
not the only factor involved, and for FGD the efficiency has been
"enhanced" by the humidification or conditioning of the reagent with a
piece of equipment called a conditioning drum patented by Procedair in
1985.
The conditioning of the reagent, hydrated lime in this case, has two
favourable effects:
Cracking of the lime particles with a resultant increase in the
solid-gas contact surface,
- Reduction in the temperature of the flue gas being treated.
6B-133
-------
The pilot/experimental plant which was installed during 1985 was on a
municipal heating system boiler plant at Gardanne (France) which
incorporated 3 MW thermal power. This plant was operated until 1988 when
it was judged that sufficient data and experience had been accumulated to
develop and standardise the equipment and to be able to predict and
guarantee performance levels.
The original objectives of the program were set to achieve a 90% SO,
reductioln with a Ca/S stoichiometric ratio of around 2. (equivalent to
levels achieved with Spray Dryer Technology). In fact the test
demonstrated a superior performance with 95% SO- reduction at Ca/S ratios
of 1.3.
8. SYSTEM DESCRIPTION (Fig.11)
8.1 Possible Gas Pre-conditioning
The flue gas may be conditioned prior to the FGD stage, if necessary, to
reduce its temperature, in order to achieve with humidified reagent the
optimum temperature in the filter stage.
In accordance with the degree of cooling required, one of the following
methods could be used :
- Slight dilution with ambient air
- Evaporative cooling
- Heat exchanger
- Heat recovery.
8.2. FGD Stage
The FGD stage is a compact vertical up flow reactor tower which is
comprised of a venturi throat at its base followed by a diffuser section
and reactor column, with an external return section.
The reagent, after being humidified in the conditioning drum, ie injected
into the throat of the venturi. The reagent injected is a mixture of
fresh material and the partly used reagent collected in the filter. The
velocity of the flue gases in the throat and the reactor column are
critical design prints as is the overall residence time in the reactor
column and external return section.
It is essential that the reagent is injected into the flue gas stream in
a manner which ensures a dispersion which is as homogenous as possible.
Its quality is a function of the quantity of recycled product and the
injection characteristics. The resultant concentration in the stream is an
important parameter.
8.3 Filtration Stage
The loaded flue gas stream passes into a modern generation pulse Jet
filter unit which separates the reaction products from the gas stream.
The scrubbed gas is exhausted via the main draught fan to the stack and
the reaction products are evacuated from the filter hopper by a live
bottom screw conveyor.
This partly used reagent 'and ash are conveyed, either mechanically or
pneumatically, to an intermediate holding hopper fitted with product
agitation and twin discharge screws. These screws, in conjunction with the
level probes in the holding hopper, control the flow of product to the
conditioning drum and/or to discharge.
6B-134
-------
gly. 11; Gacdanne prooass
-------
The product: directed to discharge ia automatically replaced by fresh
reagent and dust.
The use of a fabric filter enhances the efficiency of the system as the
reaction of the SO, with the powdery reagent continues as the gas passes
through the dust cake built up on the filter surface. It also ensures that
very low particulate emission levels can be achieved, certainly superior
to those required by legislation.
8.4 Conditioning Drum
The original design of the conditioning drum was extensive and required
considerable detailed thought, in order to ensure that:
- The fresh and recycled reagent were homogenously mixed,
- The water was uniformly atomised and dispersed.
- The product was uniformly humidified.
- The action of the drum did not agglomerate the particles.
Whilst the original design proved to be correct, there was continual
development for the first three years, which contributed to the improved
operational results In year four.
9. GARDANNE FGD PILOT PLANT CHARACTERISTICS
The plant handles a fixed volume of 6000 NMJ:/H or lO 000 AM'/H at 180°C and
is capable of taking the flue gases from 2 or 3 boilers, due to the
recirculation system.
The coal burnt at Gardanne contains 6% sulphur and 8% ash.
There is a 100% variation in this figure due to the operational variations
of the boilers and it is relatively low due to high excess combustion air
amounts and high recirculation rates at low duty.
10. EXPERIMENTAL RESEARCH PROGRAMME RESULTS
The programme consisted basically of :
Defining the operational ranges of various parameters compatible
with a reduction efficiency equal to or greater than 90%:
Gas temperature
Reagent volume and Ca/S ratio
Reagent recycling rate
Reagent humidification rate
Pressure drop across filter
Testing 4 different reagents.
Verification of the test results by an official organisation IRCHA.
average weight of sulphur
sulphur content in the gas
dust content in the gas
reagent used
12 kg/h
1.2 to 2.5 g/NM>
average fresh lime consumption
averge filter pressure drop
fresh lime silo
discharge skip
200 mg/NM3
Commercial extra fine
hydrated lime 90 to 95%,
purity. Specific surface
(BET) 15 m/g.
30 to 45 kg/h
130 - 140 mm WG
30 m3
6 m3
6B-136
-------
10.1 Reduction Efficiency Against Ca/S Ratio
A series of curves (Fig.12) summarise the percentage reduction in S02
against the stoichiometric ratio for various operating conditions.
The S02 reduction efficiency can be read directly as a function of the
stoichiometric ratio Ca/S expressing the reagent consumption, e.g.:
For an inlet concentration of 2.0 g/NM3 of sulphur and in optimum FGD
conditions.
Ca/S ratio
1
1.2
1.5
Efficiency 4
85
96
99
Sulphur Emission mg/NM3
Net at 13.5% Q2 ICorrected to 6%
300
80
20
160
40
The average temperature at the inlet to the FGD stage was 180°C.
These figures have good reliability factor; as a dual continuous S02
measurement instrument was installed upstream and downstream of the
equipment; this allowed for simultaneous measurements and monitoring in
real time of the variations observed, as a function of the operating
parameters.
10.2 Influence of Reagent Recycling
The recycling of used reagent in the system plays an important role in the
reduction efficiency. The rate of recycling originally envisaged based on
previous dry scrubbing experiences has been slightly increased.
10.3 Influence of Reagent Humidification
The addition of water is essential for high S0: reduction. This is a fact
which was quickly established but the quality of spray atomization and
distribution also play a role. Modifications to the drum design resulted
in increased reduction efficiencies.
10.4 Influence of Gas Temperature
Tests have shown that the operation should be close to the minimum
temperature compatible with no condensation talcing place in the filter.
An increase in this temperature relates to a slight decrease in reduction
efficiency. This iB, however, of lower influence than a decrease in the
resultant moisture content.
For example, and in broad terms, a 20° C rise in gas temperature or a 14
decrease in moisture content will both translate to a 5 to 7% efficiency
loss.
6B-137
-------
»swlts — €pcir«l«ann<
S02 CAPTURE EFFICIENCY
|-«-MHEK -—law/. — io-sj:
Co/S Ratio
PROCSDAIR INDUSTRIES
gip- »¦' Percentage reduction in so? against the stoichiometric
ratio for various operating conditions
-------
10.5 Alternative Reaoenta
A week of tests was set for each of the following alternative reagents:
- Calcium
- Coal soot from Gardanne laden with calcium
- Quick lime
- Magnesian lime (CaO : MgO)
The first two can not be adapted but the last two are possibilities since
the results were only slightly less than the stoichiometric equivalent.
However, quick lime is more difficult to adapt : (mechanical handling and
operator protection).
The magnesian lime requires special grinding and probably transport from
considerable distance which affects its economic viability.
11. INDUSTRIAL APPLICATIONS
The system, as developed is judged to have considerable potential or. all
industrial and smaller sized utility installations. These are not limited
to coal fired or incineration plants. One particular plant which has
recently been brought onto line, is the treatment of the flue gases coming
from Borosilicate glass furnace.
For this application, the operating parameters and gas composition axe as
given in the table in Fig. 13. The gas composition relates to using
natural gas as the fuel, but the use of No. 2 fuel oil as an alternative
was to be considered and accounted for in the design.
Apart from a guaranteed reduction of 90% In S0Z, the system was also
required to attain a 90% reduction in Fluorides and a reduction in all
chemically formed condensables (borates) or matter issuing from the stack,
to achieve an opacity of less than 6%.
The complete installation (Fig.14) comprised:
- A' stainless steel evaporative quench tower to reduce the gas
temperature from 870„C to 250°C, with possibility to supplement with
ambient air dilution before and after the quench tower
- A Procedair vertical venturi reactor tower
- A Procedair off-line cleaned pulse jet filter with on maintenance
facility.
- a used reagent recycling system to the conditioning drum and a waste
reagent pneumatic conveying system.
- a fresh reagent silo and handling system feeding the conditioning
drum.
The main I.D. fan and exhaust stack.
The installation has been operating since May 1991 with EPA proving trials
being carried out in October 1991, which demonstrated compliance with the
contract guarantees.
The visible emissions from the furnace have been eliminated and with
comparison is possible with the plume being emitted from the second
untreated furnace.
12. CONCLUSION
The Gardanne pilot plant enabled valuable experience to be gained on the
influence and interaction of several operational parameters. It
demonstrated that the high SO* reduction efficiency normally associated
with wet type systems, can be achieved with a basically dry system, with
its operational and reliability advantages.
6B-139
-------
The following basis was used for design of the APC system:
Operating Conditions Maximum Minimur
• Gas Flow to System, NM3/Hr 21,000 14,000
• System Inlet Temperature, °C 880 880
• Furnace Pressure, in. MM W.G. 50 6.0
FLUE GAS ANALYSIS (at inlet of system)
•
Air, % mol
25.0
•
N2, % mol
50.0
•
C02, % mol
7.5
•
H20, % mol
17.5
•
F2, ppmwv
280
•
S02, ppmwv
700
•
BjO,, ppmwv
300
•
NOx, ppmwv
650
•
Particulate, kg/hr
10
DESIGN PERFORMANCE LIMITS
• Particulates, mg/NM3 30
• HF, ppmdv @ 7% 02 45
• S02, ppmdv @ 7% 02 200
• Opacity < 6% *
• Includes all chemically-formed condensibles (Borates) or matter issuing from
the stack.
Furnace pressure will be controlled within _±£5" W.G. under normal operating
conditions.
iMp. 13: Process dasign conditions
6B-140
-------
iMp. 14: Manville Installation
-------
The reagent used by preference in the system is commercial hydr&ted lime
which is relatively easy and safe to handle, and needs no treatment or
preparation on site. Since 95% efficiency is obtainable, at Ca/S ratio
close to 1.2:1 stoichiometry reagent consumption is minimised which makes
for economic acceptability, this is unusual for a completely dry system.
Due to the geographical location of Gardanne in the south of France, the
municipal heating system was not run in the summer months, which extended
the time of running the pilot plant and obtaining all the data necessary.
This, however, had the effect of worsening potential corrosion and
operational problems as the plant was shut down and restarted each season,
without any particular or special procedures.
We have toi report that surface corrosion on the interior walls of the
reactor tower and filter unit, which were unprotected, was evident, but
the amount of corrosion was judged to be less than that which would have
been expected for a filter system on untreated flue gas.
The system does not create an effluent problem, and very serious
opportunities exist for using the waste product in other processes.
6B-142
------- |